Development of a Bacteriophage-based Biopesticide for Fire Blight by Susan M. Lehman, B.Sc. Submitted to the Department of Biological Sciences in partial fulfillment of the requirements for the degree of Doctor of Philosophy September, 2007 Faculty of Science, Brock University St Catharines, Ontario © Susan M. Lehman, 2007 2 Abstract Fire blight is an economically important disease of apples and pears that is caused by the bacterium Erwinia amylovora. Control of the disease depends on limiting primaly blosson1 infection in the spring, and rapidly removing infected tissue. The possibility of using phages to control E.amylovora populations has been suggested, but previous studies have. failed to show high treatment efficacies. This work describes the development of a phage-based biopesticide that controls E. amylovora populations under field conditions, and significantly reduces the incidence of fire blight. This work reports the first use of Pantoea agglomerans, a non-pathogenic relative of E. amylovora, as a carrier for E. amylovora.phages. Its role is to support a replicating population of these phages on blossom surfaces during the period when the flowers are most susceptible to infection. Seven phages and one carrier isolate were selected for field trials from existing collections of 56 E. amylovora phages and 249 epiphytic orchard bacteria. Selection of the . /' phages and carrier was based on characteristics relevant to the production and field perfonnance of a biopesticide: host range, genetic diversity, growth under the conditions of large-scale production, and the ability to prevent E. amylovora from infecting pear blossoms. In planta assays showed that both the phages and the carrier make significant contributions to reducirig the development of fire blight symptoms in pear blossoms. Field-scale phage production and purification methods were developed based on the growth characteristics of the phages and bacteria in liquid culture, and on the survival of phages in various liquid media. Six of twelve phage-carrier biopesticide treatments caused statistically signiflca11t 3 reductions in disease incidence during orchard trials. Multiplex real-time PCR was used to simultaneously monitor the phage, carrier, and pathogen populations over the course of selected treatments. In all cases. the observed population dynamics of the biocontrol agents and the pathogen were consistent with the success or failure of each treatment to control disease incidence. In treatments exhibiting a significantly reduced incidel1ce of fire blight, the average blossom population of E.amylovora had been reduced to pre-experiment epiphytic levels. In successful treatments the phages grew on the P. agglomerans carrier for 2 to 3 d after treatment application. The phages then grew preferentially on the pathogen, once it was introduced into this blossom ecosystem. The efficacy of the successful phage-based treatnlents was statistically similar to that of streptomycin, which is the most effective bactericide currently available for fire blight prevention. The in planta behaviour of E. amylovora was compared to that of Erwinia pyrifoliae, a closely related species that causes fire blight-like synlptoms on pears in southeast Asia. Duplex real-time PCR was used to monitor the population dynamics of both species on single blossonls. E. amylovora exhibited a greater competitive fitness on Bartlett pear blossoms than E. pyrifoliae. The genome of Erwinia phage <l>Ea21-4 was sequenced and annotated. Most of the 8-4.7 kB genome is substantially different from previously described sequences, though some regions are notably similar to Salmonella phage Felix 01 . Putative functions were assigned to approximately 30% of the predicted open reading frames based on amino acid sequence comparisons and N-terminal sequencing of structural proteins. 4 Acknowledgements This research was supported by Agriculture and Agri-Food Canada (AAFC) and an AAFC Improved Farming Systems and Practices grant. The author was supported by an NSERC Canada Graduate Scholarship, an NSERC Doctoral Post-graduate Scholarship, and Brock University. I could not have completed this work without the advice, assistance and support of several people: My supervisors, Dr. Antonet Svircev and Dr. Alan Castle, have been extraordinary mentors throughout this process. I can not thank you enough for your support and advice, or for the opportunities you have given me to shape the direction of this research project, and to work with a diverse and talented team of people. I am also grateful to my committee members, Dr. YousefHaj-Ahmad, Dr. Andrew Reynolds, Dr. Glenn Tattersall, and Dr. Michael Bidochka, for their advice and input on the direction of this research. Ron Smith (University of Westem Ontario) graciously gave me his tinle, expel1ise, and resources so that I could obtain the transmission electron micrographs needed for this study. Dr. Peter MacDonald (McMaster University) provided invaluable statistical advice regarding the design and analysis of blossom assays and field trials. Many thanks are also due to the Svircev laboratory group at Vineland: Ed Barszcz, for photographic and mechanical wizardry, and for making sure I always received the nlaterials I needed; Karin Schneider, for dedicated technical assistal1ce and moral support. Dr. Won-Sik Kim, for generous advice an-d for developil1g the real-time PCR primers and probes that I used to 5 monitor the·progress of my field trials. Dwayne Roach, for insightful commentary and debate. I can not thank you all enough for your assistance, and for the exciting challenge of working with such a dynamic research team. Thank you to the Agriculture and Agri-Food Canada staff at Vineland Station, Jordan Research Farm, and Delhi Research Farol, particularly Peter Raakman, Gord Rattray, Tom Roberts, Kathy Jensen, Barry Kemp, Brad Arbon, Albert Azstalos, Robert Wismer, Jam,es Piluke, and Mary Soen. Their contributions to field trials, and to equipment acquisition, maintenance, and troubleshooting have been invaluable. Dormant budwood for blossom assays was generously provided by Jordan farm, and by Cherry Lane Orchards (Vineland, ON). Several undergraduate students provided invaluable assistance. Ashleigh McCarl, Francesca Tumini, and Naiya Patel helped carry out certain field trials and host-range studies. Heidi Bench assisted with field trials, growth experiments, phage purification, and several other tests. Her organization and independence over three years have forever spoiled me for future summer students that I may have the privilege to supervise. Last, but not least, unending thanks to the family, friends and teachers who have supported me and nurtured my love of science. To Bill, for the incomparable friendship, love, and home cooking that have sustained me through frustration and success alike. And to nlY family, for unending love and support; for raising me in the belief that I could do whatever I put my mind to; and for handing mel my first Arthur C. Clarke and Isaac Asinl0v books as a child, thereby instilling a deep sense of wonder about the world, and about the possibilities that await a curious and dedicated mind. 6 Table of Contents Abstract 2 Acknowledgements 4 List of Tables 10 List of Figures 12 Abbreviations 14 General Introduction 15 Part I: A Review of Fire Blight Chapter 1: Fire blight: The disease and current management strategies ~ Introduction Erwinia amylovora, the Fire Blight Pathogen The Disease Cycle Pathogenesis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Disease Management Strategies Orchard Management Genetic Resistance Chemical and Biological Control Agents Antibiotics Plant Growth Regulators Copper Biological Control Summary 28 29 32 38 41 42 43 46 46 48 49 51 '55 Part II: Characterization of Biopesticide Components and Pathogen Grolvth Chapter 2: Selection of Erwinia amylovora phages and a Pantoea agglomerans carrier Abstract Introduction Methods Phage Isolates / , Phage Culture Media and Conditions Phage Storage Media and Conditions Determination of Phage Titre Revival of Vineland Phage Collection Isolation of Phage DNA '59 60 66 66 69 69 70 71 7 Restriction Fragment Length Polymorphisms (RFLPs) .. ~ Endpoint PCR for Group 3 Phages Real-time PCR Transmission Electron Microscopy (TEM) Bacterial Isolates Bacterial Culture Media and Conditions Bacterial Storage Medium and Conditions Antibiotic Production by P. agglomerans Amplification of a Gene from the Pantocin A Biosynthetic Cluster Host Range of the Vineland Phage Collection on E. amylovora and P. agglomerans Isolates Efficacy of Carrier Candidates and Phages in Blossom Assays 73 74 74 75 75 76 76 79 79 80 81 Results Revival of Vineland Phage Collection Molecular Characterization of Phage Collection Transmission Electron Microscopy Antibiotic Production by P. agglomerans Host Range of the Vineland Phage Collection Susceptibility of Carrier Candidates to Phage Infection Susceptibility of Common Orchard Bacteria to Infection by E. amylovora Phages Efficacy of Carrier Candidates and Phages in Blossom Assays Discussion Chapter 3: The complete genome sequence of Erwinia phage <l>Ea21-4 Abstract Introduction Methods Strains and Growth Conditions Transmission Electron Microscopy DNA Isolation and RFLP Genome Sequencing Sequence Analysis Identification of Major Structural Proteins Results Morphological Features of <l>Ea21-4 DNA Extraction Genome Sequencing General Features olf the <I> Ea21-4 Genome Gene Annotation Structural Proteins Discussion 84 84 85 90 93 93 95 95 99 106 107 112 112 112 112 113 114 117 117 120 120 126 136 140 8 Chapter 4: Real-time PCR reveals competition between Erwinia amylovora and Erwinia pyrifoliae on pear blossoms Research Contributions Disclaimer Abstract Introduction Methods Bacterial Strains and Media Real-time PCR Optimization of Real-time PCR fro Quantitative Analysis E. amylovora and E. pyrifoliae Competition Assays Assessment of Bacterial Numbers on Blossoms Statistical Analysis Results Specificity and Sensitivity of Duplex Real-time PCR Virulence of E. amylovora and E. pyrifoliae Effect of Initial Pathogen Ratio of Disease Severity Relative Population Sizes of E. amylovora and E. pyrifoliae Discussion Acknowledgements 148 149 15'0 153 153 154 157 158 158 160 161 162 164 168 172 Part III: Development of the Phage-Carrier Biopesticide Chapter 5: Evaluation of the interactions between Erwinia amylovora, Pantoea agglomerans Eh21-5, and Erwinia phages Abstract 174 Introduction 175 Methods Bacteria, Phages, and Growth Media 177 Bacterial Growth Curves 177 Phage Growth Curves 179 Phage Survival in Liquid Media 179 180 Optimization of Phage-Carrier Application to Blossoms' 181 Statistical Analysis Results Bacterial Growth Curves 182 Phage Growth Curves 182 Phage Survival in Liquid Media 186 187 Optimization of Phage-Carrier Application to Blossoms Discussion ' 191 9 Chapter 6: Detection of Erwinia phages in orchard soil Abstract .. ". . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 Introduction 197 Methods Growth Media and Strains 203 Soil Sample Preparation and Analysis 203 Phage Elution 204 Real-time PCR 205 Removal of PCR Inhibitors 206 Endpoint PCR 208 208 Optimization of EDTA Treatment Results Chemical Characteristics of Orchard Soils 210 Elution of Phages and the Presence of Soil-derived PCR Inhibitors 210 Removal of PCR Inhibitors 214 Discussion 221 Chapter 7: Control of Erwinia amylovora with a phage-based biopesticide under field conditions Abstract 227 Introduction 228 Methods Phages and Bacterial Strains 233 Production of Bacteria for Field Trials ~ 233 Production of Phages for Field Trials a) Growth 233 b) Removal of Cells and Large Debris 235 c) Concentration and Buffer Replacement 235 Field Trials 237 Population Monitoring 242 Real-time PCR 2·44 Statistical Analysis 2·44 Results 246 Sensitivity and Specificity of Real-time PCR 249 Production of Phages for Field Trials Disease Incidence and Microbial Population Dynamics - 2004 251 Disease Incidence and Microbial Population Dynamics - 2005 254 Disease Incidence and Microbial Population Dynamics - 2006 257 Weather Conditions 259 Discussion 262 General Summary 2·68 Literature Cited 277 10 List of Tables Chapter 1: Fire blight: The disease and current management strategies 1-1 Relative sus'ceptibility of apple and pear scion cultivars to fire blight ... 00... 45 Chapter 2: Selection of Erwinia amylovora phages and a Pantoea agglomerans carrier 2-1 Phage isolates used in this study . 0.. 0. 0. 0. 00. 0.. 0..... 0000. 0000000000067 2-2 Bacterial isolates used in this study . 000000.. 0000. 0. 00. 0. 0000.' 000000. 0077 2-3 Molecular characterization of the revived phage collection 0 0 86 2-4 Family-level characterization of selected E. amylovora phages . 0. 0.. 0000. 0. 88 2-5 Growth inhibition of Erwinia sppo by P. agglomerans antibiotics 0.. 00. 0.. 0091 2-6 Host range of the Vinelalld phage collection on Erwinia and Pantoea . 0. 000.. 94 2-7 Host range of Erwinia phages on Po agglomerans carrier candidates ... 0.... 097 Chapter 3: The complete genome sequence of Erwinia phage <l>Ea21-4 3-1 Characteristics of previously described E. amylovora phages . 0. 00000. 00... 109 3-2 Effect of extraction method on the A260/A280 ratio of genomic <l>Ea21-4DNA .000000 .. 0.00.0000000000.0 .. 0.0000.0.00 ... 00. 0~ 00.119 3-3 General features of the <l>Ea21-4 genolne compared to two other E. amylovora phages, <l>Eal(h) and Eral03, and Salmonella phage, Felix. 0... 0. 00.. 00. 0 0.. 0.. 0. 0. 00. 0. 00. 0000. 000.. 0... 0000000127 3-4 Predicted genes of <l>Ea21-4 00. 0 0000. 00 0000. 00.. 000. 00128 3-5 Locations and features oftRNAs encoded by the <l>Ea21-4 genome 00. 000. 00137 3-6 N-tem1inal sequence and identification of <l>Ea21-4 structural proteins . 00. 00139 Chapter 4: Real-time PCR reveals competition between Erwinia amylovora and Erwinia pyrifoliae Bacterial strains used in this study .. 0. 0.. 00. 0000000000. 0. 000.. 0. 0.. 0. 155 4-1 Sequences, product sizes, and targets of primers and TaqMan probes 4-2 used for duplex detection and quantification of E. amylovora and' E. pyrifoliae ..... 00000. 000. 00.. 0... 0.... 00. 000000000. 0. 0.... 00. 0156 The effect of initial inoculum size on disease severity in pear blossoms . 00.. 165 4-3 Chapter 5: Evaluation of the interactions between Erwinia amylovora, Pantoea agglomerans Eh21-5, and Erwinia amylovora phages 5-1 Bacteria and phages used in this study 000. 00. 00. 0 0000000 00000178 5-2 Survival of <l>Ea46-1A2 in various liquid media 000 00. 00. 000.0 .. 0. 000188 5-3 Effect of ionic str~ngh on the survival of<l>Ea31-3 and <l>Ea46-1A2 00.000 .. 189 Chapter 6: Detection of Erwinia phages in orchard soil 6-1 PCR primers and probes for Erwinia phages, Eo amylovora, and P. agglomerans 0 0 00 209 11 6-2 6-3 6-4 6-5 6-6 Chemical characteristics of orchard soil at the sampling sites on the AAFC-Delhi Research Farm Chemical characteristics of typical orchard soil on the AAFC-Jordan Research Farm Effect of elution medium on phage recovery Recovery of viable phages by centrifugation and diafiltration Relative contributions of chelation and dilution to the removal of soil-derived PCR inhibitors 211 211 212 217 220 Chapter 7: Control of Erwinia amylovora with a phage-based biopesticide under field conditions Experimental design of 2004 field trials 239 7 -1 7-2 Experimental design of 2005 field trials 240 Experimental design of 2006 field trials 241 7-3 7-4 Specificity of real-time PCR primers and probes used in tl1is study 247 7-5 Concentration and loss of phages throughout production and purification 250 12 List of Figures Chapter 1: Fire blight: The disease and cun"ent managenlellt strategies 1-1 The fire blight disease cycle and the structure of pear blossoms 1-2 The structural features of a dissected Bartlett pear blossom 1-3 Factors affecting the development of fire blight Chapter 2: Selection of Erwinia amylovora phages and a Pantoea agglomerans can"ier 2-1 The lytic phage replication cycle 2-2 Rating scale describing the severity of fire blight symptoms in the in planta pear blossom assay 2-3 TEM of five phages of E. amylovora 2-4 Growth inhibition of E. amylovora by P. agglomerans antibiotics 2-5 Screening bacterial epiphyte isolates for susceptibility to infection by Erwinia phages Chapter 3: The complete genome sequence of <I>Ea21-4 3-1 Transmission election microscopy of <l>Ea21-4 3-2 RFLP patterns resulting from the digestion of genolnic <l>Ea21-4 DNA each of MvnI, BgIII, BamHI, and EcoRI 3-3 Similarity between <l>Ea21-4 endolysin and the <l>Ea1 lysozyme 3-4 Genome-wide similarity between the <l>Ea21-4 and <I> Era103 genomes 3-5 Genome-wide similarity between the <l>Ea21-4 genome and the Salmonella phage Felix genollle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-6 Electrophoretic separation of <l>Ea21-4 structural proteins Chapter 4: Real-time PCR reveals competition between Erwinia alnylovora and Erwinia pyrifoliae on pear blossoms 4-1 Duplex standard curves for the quantitative detection of E. amylovora and E. pyrifoliae 4-2 Effect of initial pathogen ratio on disease severity 4-3 Population sizes and relative growth performance of E. amylovora and E. pyrifoliae, alone and in mixtures 34 35 57 61 83 89 92 96 118 122 123 124 125 138 163 166 167 Chapter 5: Evaluation of the interactions between Erwinia amylovora, Pantoea agglomerans Eh21-5, and Erwinia phages 5-1 Calibration curves for spectrophotometric detennination of' E. amylovora Ea6~4, and P. agglomerans Eh21-5 cell concelltration 183 in liquid culture 5-2 Growth of E. amylovora Ea6-4 and P. agglomerans Eh21-5 in liquid culture at 27°C 184 5-3 Growth of E. amylovora Ea29-7 and <l>Ea31-3 in synchronously infected and uninfected cultures. . 185 e •••••••••••••••••••••••••••••••• 13 5-4 Effect of phage-carrier cOlnbinations on disease severity in the in planta pear blossom assay Chapter 6: Detection of Erwinia phages in orchard soil 6-1 Effect of elution medium and passage through soil on the amplification of phage DNA 6-2 Amplification of phage DNA following treatment of soil eluate 6-3 Standard curve for the real-tinle PCR amplification of <I>Ea45-1B using the dpo 1 prilners and probe 190 213 216 219 Chapter 7: Control ofErwinia amylovora with a phage-based biopesticide under field conditions Aeration of bioreactors for field-scale phage production 23-4 7-1 7-2 Standard curves for·the multiplex detection of E. amylovora, P. agglomerans, and phage during orchard trials 248 7-3 Incidence of fire blight in 4-year old Bartlett pear (2004) 252 7-4 P. agglomerans populations on pear and apple blossom hypanthia during 2004 field trials 253 7-5 Incidence affire blight in 12-year old Bartlett pear and 4-year old Gala apple orchards (2005) 255 7-6 Incidence of fire blight in 8-year old Gala apple treated with varying 256 amounts of <j>Ea46-1A2 in different phage:carrier ratios 7-7 Population dynamics of E. amylovora, P. agglomerans, and Erwinia 758 phage <j>Ea46-1A2 during orchard trials 7-8 Orchard weather conditions ill 2004 and 2005 261 General Summary D-1 Integrated development of an effective and practical phage biopesticide ..... 273 14 Abbreviations CFU - colony forming units EPS - exopolysaccharide HR - hypersensitive response Mal - multiplicity of infection; the ratio of phages to host 'cells (eg. when the concentration of phages is ten times that of host cells, Mal = 10) OD600 - ORF - open reading frame PB - sodium phosphate buffer PB salt - PCR - polymerase chain reaction PFU - plaque forming units optical density to light with a wavelength of 600 nm sodium phosphate buffer amended with 100 mM NaCI and 2 mM MgCl2 RFLP - restriction fragment length polymorphism TEM - transmission electron tnicroscopy 15 General Introduction This work describes the development of a phage-based biopesticide for tire blight. Fire blight is a necrotic disease of rosaceous plants that is. caused by the bacterial species, Erwinia amylovora. The pathogen is present in most apple and pear growing regions of the world, and causes serious economic losses to commercial growers in n10st of them (Bonn & van del" Zwet, 2000). The disease cycle begins anew each spring when warm weather favours the growth and dispersal of E. amylovora. The pathogen colonizes open blossoms, multiplies, and infects the tree through natural openings in the blossoms (Wilson, Epton & Sigee, 1989). Most control strategies are tllerefore focused on suppressing the growth of E. amylovora so as to prevent blossom infection. Currently, the most effective of these treatments is the timed application of streptomycin to open blossoms. However, specific concerns about streptomycin-resistant E. amylovora, along with the general trend to avoid antibiotic use in agriculture, are drivil1g the development of alternative control strategies. In Canada, the registration ofstreptomycil1 for fire blight prevention is under constant scrutiny, and the registration of bacterial biopesticides has begun. Bacteriophages, or simply "phages", are viruses that il1fect bacterial cells. They consist of a nucleic acid genome contained within a protein or lipoprotein capsid. Upon infection, the phage-encoded gene products re-direct the host cell metabolisn1 to cease norn1al function and instead devote its resources to replicating the phage. Once mature phage virions have accumulated, the bacterial cell is lysed, releasing a new generation of dozens or hundreds of infectious phages from a single initial infection. 16 The use of bacteriophages to C'ontrol bacterial pathogen.s is an idea that dates back to the discovery of phages in the early 20 th century. That discovery is formally attributed to illdependent investigations by Frederick Twort in 1915 and Felix d'Herrelle in 1917. D'Herrelle was the first to advocate the potential therapeutic uses of phages, perhaps because he first encountered them while working on another biological control problem (Sumlners, 1999). D'Herrelle's first published report on phages described a filterable agent that killed the dysentery bacillus that is now known as Shigella (D'Herrelle, .1917). He noted that the agent was abundant in the stools of recovering patients, but rare or absent in those of acute patients, and suggested that the agent \vas an exogenous means of immunity. He concluded that the agent present in recovering dysentery patients was a particulate microbe infecting bacteria. Drawing on the Greek word phagein (eater), d'Herrelle named these invisible agents "bacteriophages", quite literally, "eaters of bacteria" (D, Herrelle, 1917). After his research with dysentery, he n10ved to France, where he studied avian typhosis. He demonstrated that phage-treated flocks had fewer deaths, shorter epidemics, and fewer recurrences of disease (d'Herelle, 1921). In the early 1920s he treated bovine hemorrhagic septicemia in Indochina (d'Herelle, 1926). In that case, parenteral inoculation ofsp'ecific phage protected water buffaloes from experimental infections. D'Herrelle also experilnented with phage therapy of human diseases, treating bacillary dysentery and four cases of bubonic plague, and also initiating a large-scale trial to reduce cholera infectiol1s in India (Summers, 1999). I The success of d'Herelle's work garnered much attentioll, and sparked a great deal of interest in phage therapy. However, not all phage treatments were successful, and the ilnage of phage therapy was being tarnished by the sale of unproven remedies ill the United "States. When 17 antibiotics were introduced just before World War II, phages were relegated to the sidelines as a therapeutic tool. Penicillin and other antibiotics offered an effective broad-spectrum therapy that was easy to mass-produce, whereas phages were pathogen-specific. Instead, phages became the star tool of early molecular biology and genetics research since their small genonles facilitated the study of basic cellular processes (Sunlmers, 1999). Limited work on phage therapy continued in North America and Western Europe, including two large cholera studies in Pakistan that were sponsored by the World Health Organization in the 1960s (Mansur et aI., 1970; Marcuk et aI., 1971). However, these studies are notable because.oftheir rarity, and research interest in phage therapy waned as pronlising studies were counteracted by reports of failed experiments and by general skepticism about t]le practical applicability of the idea. In contrast, research and active use of phage therapy continued in a few parts of Europe during World War II, and in several Eastern Bloc nations afterwards. In the 1920s, Josef Stalill had enthusiastically allowed Georgyi Eliava and Felix d'Herrelle to found an institute of phage research in Tblisi, in the Soviet Republic of Georgia (Summers, 1999). Much of the work conducted in humans has occurred there, but most has not involved double-blind, placebocontrolled trials, or has only been published in Russian or Polish. However, most of the patients who are referred to the phage therapy clinicians at Eliava have chronic or antibiotic-resistant infections that have proven untreatable by every other method available, and have been told that I they have less than a 5% chance of recovery. Yet the recovery rate among these patiellts is a stunning 80% or better if they receive phage treatnlent at Eliava (reviewed by Sulakvelidze & Kutter, 2005). Clearly, phage therapies can be effective, and yet they have rarely been developed 18 in other parts of the world, and are virtually unknown in agriculture. Phages have several properties that make them valuable as therapeutic ag·ents. The mechanisnls by which they infect and kill bacteria are unrelated to the lllechanisms by which antibiotics act, which means that phages can be used successfully against multi-drug resistallt bacteria (Chanishvili et aI., 2001). They do not have toxic side-effects in the organiSlTI suffering from bacterial infection. Under certain circumstance phages can elicit an immune respOllse in treated individuals, but the response is ,usually subclinical, and at least'in humans, does not oftell alter the outcome of treatment (reviewed by Sulakvelidze & Kutter, 2005). Pllages are specific to one or a few related target bacterial species, and therefore do not affect either eukaryotic cells or the normal bacterial ecology of the surrounding environment. They are both self-replicating alld self-limiting; a single application can initiate a population that will flourish so long as a susceptible target is present, but they tend to be rapidly inactivated in the absence of a susceptible host and so do not persist in a non-therapeutic environment (Incilley, 1969; Geier, Trigg, & Merril, 1973). Because of this close association with bacterial host cells, they are often able to penetrate tissues that are not accessible to chemical or other treatlllents (Bogovazova, Voroshilova, & Bondarenko, 1991; Bogovazova et aI., 1992, both as cited in Sulakvelidze & Kutter, 2005). Finally, their production is inexpensive and does not require complex technologies. Unfortunately, some of these same characteristics can complicate the practical applications of phage therapy and have been presented as reasons why phages should not be used to control phytopathogenic bacteria. A review by Vidaver (1976) presents the most comll10n of these arguments. She notes that bacteria can become resistant to pllages, whether by lysogeny or 19 .mutation, and that phages can act as genetic vectors, facilitating the spread of virulence factors or other genes among a bacterial populatioll by generalized transduction or lysogenic conversion. She also points to the fact that phage replication is highly sensitive to environlnental conditiollS. The transfer of virulence genes or other undesirable traits between bacteria can be minimized by only using phages that are obligately lytic, and by screening for generalized transduction of molecular markers by those phages. While it is impossible to prove that transduction will never occur, it is possible to quantify a maximum risk. The use of phage mixtures should also help minimize the likelihood of transduction, since it is unlikely that all phages in a mixture would be capable of transduction following careful screening. If transduction did occur, the hypothetical transductant cell should still be lysed by at least one of the other phages in the mixture, preventing its propagation. Phage mixtures are recommended for any phage therapy because the combination of many carefully selected phages drastically increases the likelihood that the pathogen strains endemic to a particular region will be susceptible to the biopesticide. hl fact, Stewart (2001) suggests that the tendency towards single-strain biopesticides is one of the most inlportant and most easily rectified barriers standing in the way of consistent perfonnance by biological control agents in general. The use of phage mixtures also substantially reduces the likelihood of bacteria becoming phage-resistant in the first place (Tanji et aI., 2004; 2005). Finally, while it is true that phages are highly sensitive to environmental conditions, so are their host bacteria. Phages that are effective in the environmental conditions ill which the target bacterial species is pathogenic can be selected during phage enrichmellt and efficacy screening, and their survival in the environment can be enhanced by the addition 'of protective 20 chenlicals during the production of the comnlercial product. These issues should not stand in the way of commercially feasible phage-mediated control of phytopathogens, so long as they are taken into consideration during the process of biopesticide development. Nevertheless, only a few phage therapies have been successfully brought to market. Phages formulated in protective media have been successfully used to control bacterial spot of pepper and tomato caused by Xanthomonas campestris (Balogh, 2002; Obradovic & Jones, 2004). This biopesticide is available on a small scale from OmniLytics (Salt Lake City, UT). In addition, a six-phage suspension that reduces contatnination by Listeria monocytogenes on raw meat products (Carlton et aI., 2005; Wagenaar et aI., 2005) has been approved by the United States Food and Drug Administration, and is produced by a cotnpany called Intralytix (Baltimore, MD). The key to creating an effective phage-based biopesticide is the philosophy on which the development program is based. The successful development of microbial biopesticides requires that several factors be considered at every stage of research: efficacy, economics, and regulatory constraints. Regulatory requirements include safety/toxicity studies, environmental persistance and impact studies, and aesthetic acceptability of the basic idea to the human beings in charge of the regulatory process. Economic considerations include the ease and cost of production, storage, and use. Finally, and most importantly, the treatment must work. It must predictably control an appropriately broad spectrum of bacterial strains in the field, either reducing the patho~gen population to subeconomic levels or removing the pathogen's ability to cause disease. (Murdoch & Briggs, 1996; Schisler & Slininger, 1997; Wilson, 1997b; Stewart, 2001; Ojiambo & 'Schernl, 2006). 21 Schisler and Slininger (1997) identified three aspects of initial research that are critical for the development of a commercially feasible biopesticide: "( 1) choosing an appropriate pathosystem to investigate; (2) choosing a appropriate method of microbe isolation; (3) conducting an appropriate isolate characterization and performance evaluation." The selection of an appropriate disease target is the first step. There must be a bottleneck in the disease cycle, at which point the prevention of transmission or infection will substantially limit further disease development. The factors that limit or permit transmission or infection at this point must be vulnerable to a phage-based treatment in a manner that is consistent with existing treatment infrastructure. In the case of fire blight, open apple and pear blossoms are the primary site of infection. They are present for a finite period oftinle, and are easily accessible using existing pesticide application technology. The second step in the development of a biopesticide is the isolation of Inicrobial antagonists. In the case of phages, the isolation of a large number of diverse phages with broad host ranges is facilitated by the use of a multiple host enrichnlent system (Jensen et aI., 1998). This was done by Gill (2000) during the collection of the Vineland phages, and resulted ill a collection of E. amylovora phages that is much more diverse than any previously reported (Erskine, 1973; Vandenburgh et aI., 1985; Ritchie & Klos, 1979; Schnabel & Jones, 2001). Thirdly, the criteria by which isolates are selected must approximate the criteria which will determine success or failure in the field. A candidate that is highly effective against a laboratory strain in an in vitro assay may not be effective against a natural infection. In the case of fire blight research, screening processes are oftell media-based or use fruit tissue or seedling shoots. Of these, only the seedling assay reflects part of the natural infection cycle, and in none of 22 these assays do the results reliably correlate with perfonnance on blossoms (Wilson, Epton, & Sigee, 1990; Thibault, & Le Lezec, 1990, as cited by Lespinasse & Aldwinckle, 2000; \Vilson, Epton, & Sigee, 1992). Blossom infection is the primary mode of seasonal disease initiation, alld so assays based on apple or pear blossoms, such as the one used in this research, are a much more accurate assessment of tIle efficacy of a biopesticide in controlling E. amylovora populations. The final and most critical step is proving that the biopesticide has high efficacy under field conditions. The complex interaction of biotic and abiotic factors in the orchard can also alter the physiological and biochemical properties of trees in ways that can not be replicated elsewhere. As a result, green-house and growth chamber plants sometimes support higher epiphytic microbial populations (O'Brien & Lindow, 1989; Beattie & Lindow, 1994). The pathogen pressure against which biopesticides are tested must also reflect the natural infectioll process as much as possible. In many cases, applying a pathogen concentration comparable to indigenous populations will not produce enough disease to obtain statistically significant data unless a impractically large number of sample plants is used (Johnson & Stockwell, 2000). If a higher concentration of pathogen is necessary, care should be taken to ensure that the pathogen population must still increase on the plant surface before infection can take place, otherwise there will be no opportunity for the biopesticide to antagonize the pathogen, as it would in a natural infection setting. Without a doubt, the greatest hindrance to the development of successful phage therapies has been the absence of an adequate knowledge of the phage-host ecology in the therapeutic environment (Goodridge, 2004; Summers, 2005). The consequence of this has been an historical illability to explain the success or failure of a given treatmellt. Addressing this issue absolutely 23 requires that the population dynamics of the biopesticide components and the pathogen be monitored throughout field trials. The effects of environmental stresses on the culturability of bacteria (Wilson & Lindow, 2000) requires that this be done using a culture-independent method. This research constitutes the first report of an effective, practically feasible phage-based biopesticide for fire blight. The novel strategy developed here uses a non-pathogenic bacterium that is also susceptible to infection by the same phages as a "carrier". Pantoea agglomerans (Ewing & Fife) Gavini et al (formerly Erwinia herbicola) is an orchard epiphyte that is closely related to E. amylovora. The underlying hypothesis of this work is that the P. agglomerans carrier will playa dual role in control of E. amylovora, acting directly via competitive exclusion, and indirectly by supporting a replicating population of phages in the hostile blossom environment over the period of time at which the blossoms are at risk for infection. If E. amylovora colonizes the blossom, the phages will infect and kill the pathogen such that its population never grows large enough to cause a blossom infection. To accomplish this goal, phages and a carrier will be selected from highly diverse collections of unique E. amylovora phages and P. agglomerans isolates. These phages are the most morphologically and .genetically diverse, and the promiscuous E. amylovora phages that have yet been described. The iIlteractions of E. amylovora phages, E. amylovora, and P. agglomerans are extensively studied in both excised blossoms and in the orchard. Multiplex real-time peR is used here, for the first time, as a means of simultaneously monitoring the population dynamics of the three organisms releval1t to the success of the phage-carrier biopesticide - the E. amylovora pathogen, E. amylovora phages, and the P. agglomerans carrier - and thereby make the critical correlation between treatment outcome and microbial ecology. 24 The research encompassed by this strategy is divided into three parts, outlined below. Part I: A Review of Fire Blight Chapter 1: Fire blight: The disease and current management strategies. A thorough appreciation for the biological context to which a disease control strate,gy will be applied is fundamental to the development of that strategy. The disease alld its causative agent are reviewed here, including distribution, epidemiology, and pathogenesis. Current disease management strategies are en1phasized, as this research describes the ,development of a novel means of fire blight control. Part II: Characterization of Biopesticide Components and Pathogen Growth Chapter 2: Selection of Erwinia amylovora phages and a Pantoea agglomerans carrier This section describes the performance-based selection and preliminary characterization, of a carrier and the phages that will be the focus of field trials. In planta pear blossom assays were used to select carrier and phage isolates that are most likely to have high efficacy under field conditions. Chapter 3: The complete genome sequence of <f>Ea21-4 The Vineland collection of E. amylovora phages displays a tremendous anl0ullt of genetic diversity. This diversity has challenged efforts to monitor the population dynamics of these phages. The first complete genome sequence of an E. an1ylovora phage belongillg to the Myoviridae, <f>Ea21-4, was sequenced in order to ilnprove understanding of E. 25 amylovora phages, and to assist in the continued development of molecular monitoring techniques. Chapter 4: Real-time PCR reveals competition between Erwiniaamylovora and Erwinia pyrifoliaeon pear blossoms. This work on the Asian pear blight pathogen, E. pyrifoliae, and its relationship to E. amylovora provided an opportunity to study the growth of E. amylovora on blossoms, and to validate some of the protocols being developed to monitor the population dynamics of biopesticide components in the orchard. Part III: Development and Evaluation of the Phage-Carrier Biopesticide Chapter 5: Evaluation of the interactions between Erwinia amylovora, Pantoea agglomerans Eh21-5, and Erwinia amylovora phages An in planta assay is used to determine the optimum parameters for field applications of the phage-carrier biopesticide. In vitro growth characteristics of E. amylovora, P. agglomerans, and Erwinia phages relevant to the efficient production of the biopesticide components are also studied. Chapter 6: Detection of Erwinia amylovora phages in orchard soil This chapter describes the development of a method to isolate phages from soil, and to remove soil-derived PCR inhibitors, without the need for a DNA extraction step. This work was undertaken in order to develop a simple and rapid means of trackin.g the 26 environmental fate of the phages being used in these experiments. Chapter 7: Control of Erwinia amylovora with a phage-based biopesticide under field conditiol1.s Phage-carrier combinations are shown to reduce disease incidence by controlling pathogen populations in field trials. Multiplex real-time PCR is used to cOITelate disease outcome with biopesticide and pathogen population dynamics. Phage production and purification techniques are also described. 27 Part I: A Review of Fire Blight 28 Chapter 1: Fire blight: The disease and current management strategies Introduction Fire blight is a necrotic disease caused by the bacterial phytopathogen Erwinia amylovora (Burrill) Winslow et al. The disease affects roseaceous plant species, including Malus spp. (apple), Pyrus spp. (pear), Cydonia (quince), Cotoneaster (cotoneaster), Sorbus (mountain ash) , Rubus spp. (raspberry, blackberry), and Crataegus (hawthorn). Fire blight h.as the distinction of being the first disease proven to be caused by a bacterium (Burrill, 1878), and E. amylovora was the first phytopathogenic bacterium shown to have insect vectors (Baker, 1971). Fire blight was first described on apple trees in 1780 in the Hudson Valley of the USA (Denning, 1794), where it probably evolved on local rosaceous species such as hawthorn, mountain ash, and crab apple. The disease spread throughout the United States and Canada during the 19th century, and reached New Zealand and England by 1960 (Bonn & van der Zwet, 2000). Fire blight then appeared in the Eastern Mediterranean, and spread quickly through Europe. It is also present in Mexico (Lopez & Fucikovsky, 1990), and unconfim1ed observations have been reported in Bermuda, Guatemala, China, Vietnam, and Russia (Bonn & van der Zwet, 2000). The dissemination of E. amylovora has been largely facilitated by hun1an activity. The westward spread of fire blight across North America generally coincided with the n10velnent of European settlers and their establishment of fruit orchards (Bonn & van der Zwet, 2000). Transoceanic introductions into New Zealand and England seem to have resulted from the importation of infected nursery stock (Bonn & van der Zwet, 2000). The source of the il1itial Egyptian 29 infections is not clear, but the fact that the first repoli was made near Alexandria, a port city on the Nile delta, makes it likely that this introduction was also related to an agricultural import. The timeline of disease spread suggests that once present in Englalld and Egypt, E. amylovora quickly spread to the other reporting European and Mediterranean countries via natural means (Bonn & van der Zwet, 2000). In total, the disease has been reported in at least 40 countries, and is economically important in many of them. The greatest economic impact of fire blight results from the loss of commercial fruit crops, particularly apple and pear. Because outbreaks occur sporadically and with variable severity, it is difficult to estimate the annual impact of fire blight. Nor are the effects limited to a single year, sipce shoot blight can kill wood that would bear fruit spurs in subsequent years. A 1998 outbreak in Washington and Oregon caused an estimated $68 million loss in apple and pear crops (Smith, 1997). The 2000 outbreak of fire blight in southern Michigan caused an immediate crop loss of $9.7 million, a tree loss of $9 million, and an estimated total loss of $42 million over five years until damaged trees recovered their full fruit-bearing capacity (Lollgstroth, 20'00). Outbreaks tend to be more frequent and more severe in warm, humid regions (Bonn & van der Zwet, 2000) where the temperature and moisture conditions that favour pathogell dissemination and infection coincide with the most susceptible host phenology. Erwinia amylovora, the Fire Blight Pathogen E. amylovora is a member of the Enterobacteriaceae, the Gram-negative, motile, facultatively anaerobic, non-sporulating bacilli. The organism was originally named Bacillus amylovorus (Burrill, 1878). Ninety years ago, all of the phytopatllogenic Enterobacteriaceae 30 were classified in the genus Erwinia (Winslow et aI., 1917). That genus was subsequently divided into the "amylovora", "carotovora", "herbicola", and "atypical" groups based on metabolic characteristics and the type of disease they cause (Dye, 1968; Dye, 1969a; Dye, 1969b; Dye, 1969c). DNA:DNA hybridization studies and rDNA sequence allalyses led to further taxonomic reorganization of these organisms, with most of Dye's herbicola group and SOlne of the atypical group being assigned to the genus Pantoea (Ewing & Fife, 1972), most of the carotovora group being assigned to the genus Pectobacterium (Lelliot & Dickey, 1984; Hauben et aI., 1998), and an additional monophyletic group of species being placed in the genus Brennaria (Hauben et aI., 1998). For all the phylogenetic turbulence suffered by current and former members of the Erwinia genus, E. amylovora is, itself, a remarkably homogeneous species. There is little variability in metabolism or serology, and what variation has been described has not been liIlked to differences in virulence (Billing et aI., 1961; Vantomme et aI., 1982; Slade & Tiffin, 1984; Vantomme et aI., 1986; Verdonck et aI., 1987). Traits of note include a strict requirement for nicotinic acid, which is not common among the Enterobacteriaceae, and the ability to use both sucrose and sorbitol as carbon sources, since these are the main forms of carbon reserves in apple and pear (Bieleski, 1969). There are no recognized pathovars or biovars of E. amylovora, and most strains are not species-specific. Those isolated from Maloidae (ie. apples, pears, hawthorn, and quince) are pathogenic on Amygdaloidae (ie. Japanese plum), Rosoidae (ie. raspberry), and other Maloidae. The one exception seems to be isolates from Rubus spp., which are not patho.genic on apple or pear (Starr, Cardona, & Folsom, 1951; Ries & Otterbacher, 1977; Heimann & Worf, 1985). 31 Some genetic markers distinguish these isolates, but the biological basis of their differellt host specificity is unknown (Laby & Beer, 1992; McManus & Jones, 1995b; Jock & Geider, 2004; Giorgi & Schotichini,- 2005). In 1903 Uyeda reported fire blight on apple trees in Japan. He identified the pathogen as Bacillus amylovorus, and believed that it been introduced on infected nursery stock from the United States (Uyeda, 1903, as cited by Bonn & van der Zwet, 2000). Sin1ilar reports of outbreaks on apple and pear were made in the 1920s, and in 1955, a report on bacterial phytopathogens present in Japall included a pear disease caused by an unidentified species of Erwinia (Okabe & Goto, 1955, as cited by Bonn & van der Zwet, 2000). In 1992, a report on "bacterial shoot blight of pear" (BSBP) described symptoms very similar to fire blight and a causative agent that was almost, but not quite, identical to E. amylovora. The precise identity of this organism became the subject of some debate. Recent studies have presented evidence that the Japanese Erwinia isolates are much more closely related to, and may in fact be, Erwinia pyrifoliae (Kim et aI., 2001a; Matsuura et aI., 2007; Chapter 4). Given the historical reports of fire blight-like disease in Japan, and the pattelTI of similarities between E. amylovora, the Japanese Erwinia isolates, and E. p.,vrifoliae, one possibility is that E. amylovora was imported into Japan on nursery stock in the 'early 1900s, and then evolved through a combination of genetic drift and adaptation to local host species, particularly Pyrus pyrifolia, as it spread through Japan, eventually reaching Korea. But where t gene flow between E. amylovora populations within North America and Europe likely continues to this day, the population imported into Japan 100 years ago would have been geographically isolated, setting the stage for speciation. As a result, the "newest" species is E. pyrifoliae, found 32 in Korea, and the Japanese Erwinia isolates are much more similar to that species than they are to E. amylovora. The constant occurrence of local adaption is supported by tIle gradual spread of E. amylovora from highly susceptible pears to more resistant apples over the first 10 years in which fire blight is reported in a new region (van der Zwet & Keil, 1979), and by regional and stressrelated microdiversity within the short-sequence repeat (SSR) region of plasmid pEa29 (Jock et aI., 2003a). Certain similarities between the SSRregions of E. amylovora and the Japanese Erwinia strains also suggest that these two organisms shared a single, common ancestor. It has been argued that "bacterial shoot blight of pear" (BSBP) is identical to fire blight, regardless of the host specificity or bacterial strain involved (Beer et aL, 1996). There is substantial merit to this argument. In medicine, the term "disease" refers to pathology, the set of symptoms that a patient presents. Frequently, that disease may be caused by one of several different etiologic agents. In plant pathology, however, the general practice is to define a disease by the specific identity of its causative agent. If this is the standard to be used then it must be conceded that Asian pear blight, and most likely BSBP, are not, strictly speaking, fire blight. However, this perspective must not be allowed to overshadow the practical implications of disease management, or the potential insight into phytopathogen evolution that could be provided by more detailed phylogenetic studies of these Erwinia species. The Disease Cycle The nature of a disease cycle is that it has no objectively definable beginning or ending. However, from the perspective of the commercial grower, the fire blight disease cycle begins anew each spring. There are five generally accepted types of fire blight: blossom blight, shoot 33 blight, canker blight, rootstock blight, and trauma blight (Steiner, 2000). Each is characterized by a distinct set of symptoms, and the location of the E. amylovora population that is responsible. Figure 1-1 shows the progressions of infection and symptom development that have been observed. Primary infection generally takes the form of blossom blight. E. amylovora cells that have survived the winter in canker margins replicate and form a bacteria-rich "ooze" that is disseminated by wind, rain, and pollinating insects. The bacteria will then colonize the blossoms that are just beginning to open (Miller, 1929). The young blossom (Figure 1-2), and the stigtna in particular, offers a very favourable environment for the growth of E. amylovora (Thonlson, 1986; Wilson, Epton, & Sigee, 1989). It is a moist, sugar-rich environment amid the comparatively dry orchard canopy, and there is little conlpetition from other epiphytic bacteria when the blossom first opens (Stockwell et aI., 1999). Bacteria can then easily spread from blossom to blossom by foraging pollinators (van Leare, de Greef, & de Wael, 1982; Thonlson et aI., 1992; Johnson et aI., 1993a; Thomson, Wagner, & Gouk, 1998). The actual infection of the blossom occurs through natural openings in the hypanthium, when bacterial cells that have accumulated in the stigmatic secretions are washed into the hypanthium by rain or dew (Thomson, 1986; Wilson, EptOtl & Sigee, 1989; Pusey, 1997). The susceptibility of individual blossoms seems to be greatest within a few days of opening, and then declines, becoming resistant once petal fall commences (Hildebrand & Heinicke, 1937; Gouk, Bedford, & Hutshins, 1996; Thomson & Gouk, 20,03). It should be noted however, that most of these assessments are based on studies of bacterial growth, rather thatl on direct assessments of the susceptibility of hypanthial tissues. 34 Spring B) Blossom Blight Dissemination of spring inoculuin froin overwintered cankers C) Shoot Blight ,\ E) Internal Movement Overwinter Summer D) Canker Blight Figure 1-1. The fire blight disease cycle. A) Young blossolns are colonized with E. alnylovora; B) Blossoms become necrotic and release ooze; C) Shoots become necrotic as E.amylovora grows intercellularly froll1 the point of primary blossom infection, or froll1 secondary infection due to insects feeding on new succulent tissue, or fronl direct intemaiinoveinent of bacteria from overwintered cankers; D) Cankers develop frOITI bacteria that have Inigrated frOITI priinary or secondary infection sites; E) Necrosis extends through the interior of an infected shoot due to internal movelnent of E. an1.ylovora (right). Trauma blight not shown in this figure. (Photos by S. M. Lehnlan and E. S. Barszcz) 35 SepalHypanthium Nectaries/Ovary Figure 1-2. Structure ofaBartlettpearblossorn. [Photo: S.M. Lehman] 36 The pathogen continues to Inultiply as it invades the intercellular spaces of the host. The infected blossom appears water-soaked, then will discolour and become necrotic as cells undergo plasmolysis (van der Zwet & Keil, 1979). The bacteria continue to travel through intercellular spaces, invading the peduncle, the spur, and then moving into the steIn. Infected tissues will first appear water-soaked, then wilt and tum brown-black in colour. Infected shoots take on a characteristic "Shepherd's crook" shape. As the bacteria invade the host tissue they tend to enter xylem vessels (Suhayda & Goodman, 1981; Bogs et aI., 1998). Exopolysaccharide (EPS) accumulation is believed to disrupt water flow in the xylem and lead to the formation of bacterial aggregates (Sjulin & Beer, 1978; van Alfen & Allard-Turner, 1979). It is not known whether these aggregates are assemblages of planktonic cells or if E. amylovora forms biofilms in a manner comparable to the grape pathogen, Xylella fastidiosa (Marques et aI., 2002; Newman et aI., 2003). E. amylovora biofilms have not been observed, but the res genes that are implicated in the transition between the planktonic and biofilm lifestyles are only beginning to be studied in this organism (Pristovsek et aI., 2003; PriiB et aI, 2006). Regardless of their fOIm, these aggregates cause the xylem vessels to leak, and an exudate of bacterial cells and EP'S is forced to the surface, producing the ooze that is characteristic of fire blight. This ooze contains vilulent E. amylovora cells that are easily spread by rain and insects (Miller, 1929; Eden-Green, 1972). The ooze can also dry into long, thin strands that can be dispersed by wind (Bauske, 1971), and easily rehydrated to release virulent bacteria (Eden-Green & Billing, 1972; Keil & van der Zwet, 1972a). Shoot blight refers to the necrosis of vegetative shoots due to secondary infection following blossom blight or canker blight. Most vegetative tissue infections result fronl th'e 37 spread of endophytic populations (ie. due to prior blossom blight), or from infections through wounds created by insects feedil1g on succulent shoot tips (Steiner, 2000). The enlergence of susceptible succulent shoots generally coincides with the developnlent of large amounts of bacterial ooze from blossom infections, and with the warm, wet weather that favours its dissemination. In susceptible shoot tissues, fire blight symptonls may progress at rates >2.5 cm/day (Blachinsky et aI., 2003), spreading rapidly to woody tissues and killing large parts of mature trees. Orange-brown cankers develop on infected leaders and trunks due to the collapse of the cortex parenchyma (van der Zwet & Keil, 1979). In both pear and apple, cankers that form late in the season are more likely to have indeterminate margins that harbour overwintering populations of E. amylovora, and become major sources of inoculum for the following season (Beer & Norelli, 1977; Biggs, 1994). Cankers on other rosaceous hosts nearby can also provide inoculum for the following season (Billing, 1980). Rootstock blight occurs in apples when E. amylovora travels through healthy scion tissue into the rootstock, where it causes cankers that can girdle and kill the tree. It is most common and most lethal when susceptible scions are grafted on to· highly susceptible M.9 and M.26 rootstocks (van der Zwet & Beer, 1995; Steiner, 2000), but resistant scions do not necessarily protect susceptible rootstocks (Suleman & Steil1er, 1994). Rootstocks can also become infected through suckers or water sprouts. It is not known why some trees develop rootstock blight while others do not, but the condition causes an average of 25% of apple trees in the Appalacian region of the USA to die within five years of planting (Steiner, 2000). Loss estimates suggest that this could total almost $9000 per acre over that five year period (Norelli et aI, 2000). 38 Trauma blight refers to necrosis that results from direct infection of wounds caused by external factors such as hail, frost, and severe winds. Even normally resistant cultivars can become infected when the bacteria have direct access to the xylem through the wound site (EdenGreen, 1972). The occurrence of trauma blight is largely independent of disease incidence at other points in the growing season since inoculun1 may be present in the orchard without causing disease through any of the previously described mechanisms. The optimum growth temperature of E. amylovora is 25 to 27°C (Billing E. et aI., 1961), but it is able to grow between about 5 to 37°C. A substantial change in the relationship betw·een doubling time and temperature has been reported for temperatures above (versus below) 18°C, both in pure culture and in apple shoots, suggesting that temperatures above 18°C are significant to disease development in the orchard (Billing, 1974). Several studies describe the detection of E. amylovora, and even the isolation of virulent cells, in asymptomatic tissue (Baldwin & Goodman, 1963; Keil & van del' Zwet, 1972b; McManus & Jones, 1995a). Asyn1ptomatic infections may play an important role in the lon~g­ term ecology of E. amylovora, and in the spread of the pathogen via nursery stock (Calzolari et aI., 1982; van del' Zwet & Walter, 1996), but the mechanisms that trigger this switch frOlTI asymptomatic to an active infections are unknown. Pathogenesis The pathogenicity of E. amylovora arises fronl five factors: bacterial EPS, the hypersensitive response (HR) and pathogenicity gene cluster (hrp), the disease-specific gene cluster (dsp), and the capacity to thrive in the host environment. Unlike the closely related 39 Pectobacterium species, formerly the soft-rot Erwinias, E. amylovora does not degrade the cell walls of the plant host, as it does not produce any pectinolytic or cellulolytic ellzymes (Seemuller & Beer, 1976). E. amylovora produces two exopolysaccharides, amylovoran and levan, that form a loose capsule around the bacterial cells. This composite EPS capsule is a main component of the ooze that exudes from diseased plant tissue (Bennett & Billing, 1978). EPS encapsulation also prevents agglutination by the apple factor that interacts with E. amylovora LPS (Romeiro, Kan", & Goodman, 1981 a; 1981 b), and helps cells retain water and nutrients in dry environnlents (Jock, Langlotz, & Geider, 2005). Amylovoran is the principle component of the EPS capsule. It is a heteropolymer that consists mostly of galactose and glucaronic acid residues (Nimtz et aI., 1996). Its synthesis requires 12 genes on the ams operon and is influenced by temperature, pH, and nutrient availability (Geider, 2000). Levan is produced in the presence of sucrose. The secreted levansucrase enzyme (lse) cleaves sucrose molecules, releasing glucose and creating the P-2,6-fructofuranan homopolymer known as levan (Gross et aI., 1992). Levallsucrase secretion is constitutive in some E. amylovora isolates, and temperature-dependent ill others (Geier & Geider, 1993; Bereswill et aI., 1997). Amylovoran is absolutely required for pathogenicity, while levan deficiency only results in reduced virulence (Steinberger & Beer, 1988; Geider et aI., 1993; Bernhard et aI., 1996; Tharaud et aI., 1997). However, the means by which EPS contributes to disease development is not clear. Restriction of water movement and changes in membrane permeability have b,een implicated in wilt induction and tissue collapse, but the results are not definitive (Sjulin et aI., 1978; Brisset & Paulin, 1992). 40 The hrp gene cluster contributes to pathogenesis in host plants, and induces HR in non.host plants (Steinberger et aI., 1988). The cluster contains two independent regulatory pathways, hrpXY and hrpS, each inducible by specific carbon and nitrogell sources, temperature, water potential, and low pH (Wei, Sneath, & Beer, 1992; Wei, I\.im, & Beer, 2000). Both pathways act through theHrpL sigma factor, which activates the secretory and secreted hrp genes (Wei & Beer, 1995). At least 12 proteins are secreted into the host plant cells through a type III secretion system (Nissinen et aI., 2007), including HrpN and DspA/E. HrpN is the most abundant of the secreted elicitors of HR and SAR, called harpins, but is not clear whether it is absolutely required for pathogenicity (Wei et aI., 1992; Barny, 1995). The dsp genes are essential for pathogenesis (Gaudriault et aI., 1997; Bogdanove et aI., 1998), but their mechanism of action is not understood (Meng et aI., 2006; Bonasera et aI., 2006). The development of fire blight is also contingent on the ability of the pathogell to thrive within the host. In the case of E. amylovora, this means overcoming iron limitation and metabolizing the available carbon sources. In iron-limited environments such as the blossonl and the plant intercellular spaces, E. amylovora produces iron-chelating ·desferrioxamine siderophores and cognate receptors associated with the outer membrane (Kachadourian et aI., 1996; Dellagi, Reis, Vian, & Expert, 1999). Receptor expression is required for full virulence, but desferrioxamine E deficiency only impairs growth and virulence on blossoms (Dellagi et aI., 1998; Dellagi et aI., 1999), suggesting that the actual scavenging of iron for later uptake is less problematic within shoot tissue than it is on blossoms. The sri regulon an'd the scr operon are responsible for sorbitol and sucrose metabolism, respectively. Sorbitol is the nlain nl0lecule used for carbohydrate transport (Zimmermann & Ziegler, 1975; Raese, WillialTIs, & Billingsley, 1978; 41 Grant & Rees, 1981), and sucrose is most abundant in the nectaries of host blossoms (Bieleski, 1969). Since the blossom is an important infection site for E. amylovora, and the pathogen multiplies and moves intercellularly, E. amylovora must be able to use both of these carbon sources in order to thrive. Little is known about the determinants of host specificity in E. alnylovora. The restriction fragment length polymorphisms (RFLPs) that are characteristic of the limited host-range isolates from Rubus spp. are not known to be directly related to the causes of host specificity. Photosynthetic products do not appear to be involved (Braun & Hildebrand, 2005). The only possibility currently suggested relates to the dsp gene cluster. It was noted that the four apple genes whose products interact with DspA/E are conserved the genomes of host plants, but not in several non-host plants (Borejsza-Wysocka et aI., 2006). The interactions between DspE and these four proteins seem to be involved in disease development, and this is tIle only such difference between host and non-host plants that has been reported. Current Disease Management Strategies Plant disease is frequently discussed in terms of the disease triangle. Factors associated with the pathogen, the host and the environment all combine to determine the total amount of disease that develops. Environmental considerations include the general climate of the growing region, any microclimate effects produced by local geography, and the day to day weather. : Several factors can affect overall host susceptibility, including the intrinsic genetic susceptibility of a particular variety, the nutritional and water status of the planting, planting density, and other cultural practices. The infectiousness of the pathogen is affected by its intrinsic virulence, total 42 population size, and the presence of suitable vectors. Fire blight is a prime example of an agricultural disease that can not be controlled by any single method. The complexity of the disease cycle, and the virtual impossibility of eradicating E. amylovora once it is present in an orchard force the commercial grower to manage the economic dainage from fire blight by suppressing the pathogen population and maximizing host resistance, in the context of the local environment. This is achieved through a combination of orchard management and the timed application of biological or chemical pesticides. Orchard Management The risk of fire blight can be substantially reduced by minimizing the amount of E. amylovora present in an orchard. Previously infected tissue, including branches, cankers, or whole trees, can be removed during winter dormancy as part of the usual efforts to optimize tree shape and fruiting capacity (Steiner, 2000; Celetti & Carter, 2004; OMAFRA, 2006). This removes a major source of spring inoculum without spreading active populations of E. amylovora. Mid-season removal of active infections is sometinles conducted as well, but because bacterial invasion of healthy tissue outpaces symptom development, and because summer pruning can encourage a flush of susceptible growth, this practice can actually increase diseas·e severity (Suleman et aI., 1994; Wilson, 2000; Shtienberg et aI., 2003). To a degree, growers can create nutritional conditions that are less favourable for the infection and intercellular spread of E. amylovora. As a general rule, cultural conditions that stimulate vigourous growth will increase the incidence and severity of fire blight (van der Zwet & Keil, 1979). The biological basis of this relationship is poorly ullderstood, but Inay steIn frOln 43 the pathogen's response to either high sorbitol levels or high rates of flux in sorbitol levels that are observed in highly susceptible, rapidly growing tissues (Bieleski, 1969; Suleman et aI., 1994; Li & Li, 2005; Blachinsky et aI., 2006). Regardless of the underlying mechanism, well-drained soils facilitate the development of extensive root systems that provide adequate hydration without supporting excessive vegetative growth, and are generally associated with reduced fire bight susceptibility (van der Zwet & Keil, 1979; Toselli et aI., 2002). In addition, extreme nitrogen levels have long been associated with fire blight susceptibility because it favours succulent growth (Nightingale, 1932; Hildebrand et aI., 1937; Lewis & Kenworthy, 1962; Keil & Shear, 1972; Aldwinckle & Beer, 1976; van der Zwet & Keil, 1979). Since growers are far more likely to over-fertilize orchards in an attempt to increase fertility than they are to under-fertilize them, orchard management guidelines tend to emphasize the risk of excessive nitrogen application (Bonn & Carter, 2002; OMAFRA, 2006). Mineral nutrition can affect fire blight susceptibility independently of tree vigour. Mineral deficiencies in general have been shown to increase fire blight susceptibility (Aldwinckle et aI., 1976), while high foliar levels of calcium, potassium, and some other specific nutrients are associated with a protective effect (Lewis & Kenworthy, 1962; Koseoglu et aI., 1996). In general, growers should strive for uniform, moderate soil fertility, balanced mil1eral nutrition, and moderate seasonal vegetative growth, while making minin1al external inputs. At the very least, sudden, large changes in these factors should be avoided. Genetic Resistance Nearly all Maloidae fruit and ornamental cultivars are clonal, and are vegetatively 44 propagated by grafting desirable cultivar scions onto rootstocks that are selected for the attributes they confer upon the grafted tree. The intrinsic fire blight resistance of a given planting is determined by the particular combination of rootstock and scion cultivar. However, those genetic factors interact heavily with the cultural and environmental influences discussed above. Tissues of different ages and types often show different degrees of susceptibility, and the mode and site of inoculation can also affect disease outcome (Quamme, van der Zwet, & Dirks, 1976). Resistance classifications are therefore relative, fairly general, and subject to SOUle debate, since they are necessarily a compilation of many studies and years of field observation. Table 1-1 shows the relative susceptibilities of celiain apple and pear scion cultivars. Physiology affects these classifications to some extent, with the more vigourous apple cultivars being more susceptible to fire blight, but the underlying genetic basis of cultivar resistance is poorly understood. No nl0nogenic resistance has been characterized in either apple or pear. Breeding experiments and molecular data indicate that the genetic determinants of tire blight resistance are mostly additive, due to quantitative trait loci (Quamme, Kappel, & Hall, 1990; Calenge et aI., 2005; Khan et aI., 2006). Unfortunately for pesticide reduction initiativ:es, lnost popular commercial varieties are moderately or highly susceptible to fire blight. This is particularly true of the fresh market varieties, which are most subject to aesthetic restrictions and changing consumer preferences. The rootstock onto which a scion cultivar is grafted also affects its growth and disease resistance. The most commonly used apple rootstocks in Canada are M.9 and M.26. They generate productive dwarf trees that thrive in high-density plantings, but are very susceptible to fire blight. The M.7 rootstock confers increased fire blight resistance to grafted scions, but is not 45 Table 1-1. Relative susceptibilities of commercial apple and pear cultivars to fire blight. Data are compiled from van der Zwet & Beer, 1995, Babadoost, 2005 and Beckerman, 2007 Apples Pears Highly Susceptible Ambrosia, Braeburn, Cortland, Fuji, Gala, Idared, Jonathon, Liberty, Mutsu (Crispin), Paulared, Pink Lady, Rome Beauty, Russet Highly Susceptible Anjou, Bartlett, Bose, Cornice, Flemish Beauty Moderately Susceptible Empire, Golden Delicious, Granny Smith, Honeycrisp, Jonagold, McIntosh, Northern Spy, Spartan Moderately Susceptible Asian pears Hosui and Shinseiki Moderately Resistant Red Delicious Moderately Resistant Keiffer, Old Home ~46 commonly used because it does not have a dwarfing effect (Cline et aI., 2001; Cline, Byl, & Rain.stock, 2003). Pear scions are usually grafted onto pear or quince seedling rootstocks, 11either of which confers any fire blight resistance on the scion. The creation of transgenic scion and rootstock tissues that carry n10nogenic resistance factors has also been investigated, but regulatory and public resistance to genetically modified organisms has hindered further development. Chemical and Biological Control Agents Several different classes of chemicals are used to reduce fire blight-related crop losses. Copper compounds are used prior to bloom, antibiotics and bacterial antagonists are used to reduce blossom populations of E. amylovora, plant growth regulators are used to minimize Sl100t blight, and inducers of systemic acquired resistance are intended to elicit a general protective effect. Of these, copper, streptomycin, bacterial antagonists, and the plal1t growth regulator prohexadione-calcium, are registered in Canada, and only streptomycin and bacterial antagol1ists can directly prevent blossom blight (OMAFRA, 2007). Antibiotics Streptomycin is the bactericide of choice for controlling E. amylovora populations. It is an aminoglycoside antibiotic, and acts by binding to the 30S ribosomal subunit, leading to translation errors and an overall inhibition of protein synthesis (Chemotherapeutic agents, 1994). It is highly effective against blossom blight, and is sometimes also used late in the season to prevent trauma blight following severe weather damage. 47 Mounting concerns about antibiotic resistant bacteria in general, and streptomycinresistant E. amylovora in particular, have led to substantial restrictions on streptomycin use by Health Canada and the regulatory bodies of several other nations. In general, antibiotics are not registered for regular use in the European Union, though they may be used in some countries under special circumstances. Streptomycin resistance has been most problematic in the United States, particularly in Michigan and the northwest coast. Resistance problems in Israel have led to a drastic reduction in streptomycin use in that region, and increased interest in alternative bactericides (Manulis et aI., 1998; Shtienberg et aI., 2001). Streptomycin-resistant isolates ofE. -, amylovora have been found in British Columbia (Sholberg et aI., 2001) and New Zealand (Thomson et aI., 1993; Vanneste & Voyle, 1998), but the existence of these isolates has not resulted in streptomycin resistant outbreaks of fire blight. In the United States, epidemic outbreaks of streptomycin resistant E. amylovora have generally been associated with excessive use of the antibiotic (Johnson et aI., 1998). Drastic reductions in streptomycin use can substantially reduce the proportion of resistant E. amylovora in the local population (Schroth, Thomson, & Moller, 1979; McManus & Jones, 1994; Manulis et aI., 2003), but the resistant population persists in the absence of selective pressure, and can rebound.rapidly if streptomycin use is resumed (Loper et aI., 1991; McManus & Jones, 1994). The risk of resistant E. amylovora outbreaks can be substantially tninimized through the use of infection forecasting models. MaryBlyt (Steiner & Lightner, 1996) and Cougarblight (Smith, 1996) both use information about wetting events, temperature, and bloom stage to predict when an infection is likely to occur, assuming the presence of E. amylovora. In Canada, up to three applications of streptomycin are permitted each year. With proper use of these programs, growets 48 become confident that they are optimizing their allowed streptomycin applications. As a result, they are less likely to place gratuitous selective pressure on the balance betw'een susceptible and resistant E. amylovora populations. Other antibiotics have been registered or tested for fire blight control outside of Canada. Oxytetracycline, formulated as a calcium or hydrochloride complex, is registered in parts of the United States for use on pear and can also be used on apple in Mexico and some US states where streptomycin resistance has become a problem (McManus et aI., 2002). Two quinolone antibiotics, flumequine (marketed as Firestop®) and oxolinic acid, can prevent blossom blight as effectively as streptomycin (Dimova, 1990; Brisset et aI., 1990; Tsiantos J & Psallidas P, 1993; Aldwinckle, Bhaskara, & Norelli, 2002; Manulis et aI., 2003). They are only registered ill parts of Europe and Israel, respectively, but are not widely used in these regions because of the prohibitive cost (Hartman et aI., 2000; Manulis et aI., 2003). Kasugamycin, or kasumin, is another aminoglycoside antibiotic, but its efficacy against E. amylovora is variable and has not been widely tested (Aldwinckle & Norelli, 1990; Shtienberg et aI., 2001). Strepton1ycin ren1ains the only chemical currently registered for blossom blight prevention in Canada. This registration is under constant review and may be withdrawn with little notice, leaving Canadian growers without a reliable method of controlling E. amylovora. Plant Growth Regulators Prohexadione-calcium (marketed as Apogee®), is available in Canada and the 'United States for the prevention of fire blight in apples. Apogee is not useful for preventing blossom blight as it has no bactericidal effect. Instead, it is applied during late bloom or ,early petal fall in ·49 order to limit the growth rate, and thus the susceptibility, of the current year's vegetative growth (OMAFRA, 2006). The dosage of Apogee must be adjusted according to the vigour of a particular cultivar, so as not to excessively suppress vegetative growth in a moderately or minimally vigourous graft-scion combination. In some cultivars, chemical growth regulation, much like mid-season pruning, can induce secondary or "late-season" bloom, creating a second risk period for blossom blight (Deckers, Faust, & Miller, 1992). However, when applied properly, Apogee can be effective against shoot blight in mature apple trees (Aldwinckle et aI., 2002). Copper The use of copper as a prophylactic treatment· for fire blight began in the early twentieth century, following the development of Bordeaux mixture in France (Sutton, 1996). More recently, "fixed copper" formulations have been developed that contain complex copper sulphates, oxychlorides, and oxides (OMAFRA, 2006). The activity of copper compounds is largely attributed to free copper ions, which disrupt cell membranes and protein function by inactivating enzymes and structural proteins (Disinfectants and antiseptics, 1994). Unfortunately, it is not only orchard microorganisms that are affected; vegetative and reproductive plant tissues are also susceptible. The phytotoxic effects of copper compounds generally limit their application to early spring when only woody tissue is present. Copper compounds appear to be an economically attractive treatment, since they are a relatively inexpensive pesticide. However, the efficacy of copper as a Ineans of preventing fire blight is debatable. When Bordeaux mixture was first introduced, two to four applications per 50 season were common. The true impact of these treatments on fire blight is not clear, but apple growers still experienced significant disease-related losses (Sutton, 1996). Controlled studies of modem copper-based products are highly variable. Some have reported disease control comparable to that afforded by streptomycin (Aldwinckle et aI., 2002), but most reveal little to no efficacy (EI Nasr, Hamdy, & Ali, 1990; Dimova, 1990; Tsiantos et aI., 2003) or efficacy only when a high concentration is applied (Tsiantos et aI., 1993) or when inoculum pressure is low (Tsiantos et aI., 2003). Advocates for copper use stress the importance of thoroughly covering exposed bark and buds in order to create an inhibitory barrier that will prevent E. amylovora from colonizing those surfaces (Steiner, 1992; Steiner, 2000; OMAFRA, 2006). Bacteria residing in canker margins will not be eradicated, and ooze from cankers will penetrate the residue to release live bacteria. It is just that any dispersed bacteria should not be able to colonize the rest of the orchard while the copper residue persists. With this in mind, the lack of definitive results fronl efficacy studies nlay be attributable to two factors: the timing of copper application relative to pathogen activity, al1d the choice ofwllere to apply it. Advocacy of green-tip copper applications is predicated on the belief that substantial movement of E. amylovora through the orchard begins at this stage of growth (Steiner, 2000). In cool climates such as southern Ontario, the temperatures that favour ooze production and active insect vectors are not gel1erally achieved until trees begin to bloom. At this point, residual copper levels may be insufficient to affect bacterial colonization, al1d i would still be irrelevant insofar as blossom colonization is concerned. Even if copper is present in sufficient quantities, it can only have an effect if all potential colonization sites in the vicil1ity of the orchard are treated. This includes non-susceptible cultivars that could otherwise harbour 51 viable bacteria even while remaining unaffected (Steiner, 2000; OMAFRA, 2006). Biological Control The purpose of biological control is to reduce the abundance of a particular pest or pathogen by exploiting the ecological interactions between that organism and its enemies or competitors. Ideally, the active agent in a biopesticide is a species that is endemic to the region in which it will be used. This removes the issue of exotic species introductions from the regulatory process and simplifies environmental impact assessments. From a practical standpoint, however, this criterion limits the development of commercial biopesticides. Geographic differences in species diversity make it difficult to find a species that is endemic to enough areas to make it profitable to develop. To date, five commercial biopesticides are available for fire blight management, all of which consist of a single, lyophilized bacterial species. NufallTI Agricultural Inc. produces two products under the BlightBan™ name. BlightBan A506 contains Pseudomonas fluorescens A506, and BlightBan C9-1 contains Pantoea agglomerans C9-1. Bloomtin1e Biological FD (Northwest Agricultural Products) and BlosS0111 Bless (Gro-Chen1 New Zealand Limited; marketed as "PomaVita" in Italy) contain P. agglomerans E325 a11d PI Oc, .ylevitcps~l Bacillus subtilis is the active ingredient in Serenade® Max (AgraQuest, Davis, CA), which was recently registered in Canada. All of these products function to suppress E. amylovora populations on susceptible blossolTIs. Pre-emptive exclusion is the process by which one organism establishes itself in a particular niche, thereby preventing an organism that arrives later from flourishing. In this case, that involves colonizing blossom surfaces and beginning to utilize the available nutritional 52 resources prior to the.arrival of E. amylovora. Applied alone, the efficacies of BlightBan A506 and Serenade depend entirely upon this process (Wilson & Lindow, 1993), whereas P. agglomerans can also suppress E. amylovora populations by antibiosis, the antagonism of one organism by the metabolites of another (Stockwell et aI., 2002; Giddens, Houliston, & Mahanty, 2003). A diverse range of antibiotics are produced by many strains of P. agglomerans (Ishinlaru, Klos, & Brubaker, 1988; Wodzinski & Paulin, 1994; Keams & Mahanty, 1998; Jin et aI., 2003), the best described of which are herbicolin 0 and I (or pantocin A and B, respectively), the two antibiotics produced by P. agglomerans C9-1. The Bloomtime Biological FD strain repoliedly produces antibiotics, but it is not known whether they are active on blossoms (Pusey, 2002). It now appears that P. fluorescens A506 is capable of antibiosis on blossoms, but this activity, alld the concomitant increase in disease control, requires the co-application of biologically available iron (Temple et aI., 2006). The efficacy of bacterial antagonists is inherently more variable than that of streptolnycin. BlightBan A506 reduces the incidence of fire blight symptoms by 30% to 70%, and BlightBan C9-1 by 50% to 80% (Johnson et aI., 1998). Little data is available for either Bloomtime Biological FD or Blossom Bless, but their efficacy appears to be similarly v~riable (Pusey, 2002; Vanneste, Cornish, Yu, & Voyle, 2002; Werner, Heidenreich, & Aldwillckle, 2004). Some of this variability is attributable to inconsistent blossom colonization. The use of lyophilized cells, as in commercial product formulations, results in more reliable establishment of bacterial populations greater than the 1 x 104 to 1 x 106 CFUlblossom that is required for successful inhibition of E. amylovora (Johnson et aI., 1993a; Stockwell et aI., 1998). The generally greater efficacy of P. agglomerans relative to other species is attributed to th'eir capacity for antibiosis, and their 53 superior growth and survival on blossom hypanthia (Pusey, 1997; Pusey, 2002). The Inajority of antagonism between E. amylovora and bacterial biopesticides is thought to occur on the stignla, but in order to infect a blossom, the pathogen must ultimately have access to the nectaries via the hypanthium. Combinations of BlightBan A506 and C9-1 have also been tested. It \vas thought that the juxtaposition of two organisms with overlapping, but not identical, growth characteristics and mechanisms of action might increase the reliability of fire blight control by creating a more robust bacterial community on the blossom surfaces. Unfortunately, while co-inoculation increased the colonization success of both strains, there was no additive or synergistic effect on disease suppression (Stockwell, Loper, & Johnson, 1992; Nuclo, 1997). It was eventually determined that P. fluorescens A506 produces an extracellular protease that inactivates the antibiotics produced by P. agglomerans C9-1 (Anderson, Stockwell, & Loper, 2004). The efficacy of these biocontrol agents depends first upon their ability to colonize open blossoms, and then on their ability to inhibit the growth of E. amylovora. Apple and pear blossoms do not usually support detectable populations of bacteria prior to petal expansion, but are then rapidly colonized by a diverse range of species (Stockwell et aI., 1999). Young blossoms are also more easily colonized than older ones (Pusey & Curry, 2004). It is therefore recommended that these products be prepared as a suspension of 1 x 10 8 CFU/mL, and applied when orchards are at 15-200/0 bloom, and again at 75-100% bloom (Johnson & Stockwell, 2000; Agriculture and Agri-Food Canada, 2007). Initially, antagonist populations are about 1 x 102 to 1 X 104 CFU/blossom (Stockwell, Johnson, & Loper, 1998; Lindow & Suslow, 2003). Populations then increase gradually and spread to blossoms that have opened after the initial treatnlent 54 application, aided by warm moderate daytime temperatures and pollinator activity (Johnson et aI., 1993a; Nuclo et aI., 1998; Johnson et aI., 2000; Pusey, 2002; Lindow & SllSlow, 2003). There has been some resistance to the registration of these products in the European Union because of the association between organisms in the Enterobacter agglomerans cOlnplex and human opportunistic infections. The Enterobacter agglomerans complex is an extremely diverse collection of strains alld species that were previously classified in the genus Enterobacter based on biochemical phenotypes. DNA analysis has since revealed that some of these species do not truly belong to the genus Pantoea. Of those that do, there is little data available on the differences between clinical and plant-associated strains, though some phenotypic differences have been reported (Lindh et aI., 1991). Plant-associated P. agglomerans, which is ubiquitous in the environment (Riggle & Klos, 1972; Ishimaru, Klos, & Brubaker, 1988; Grimont & Grilnont, 2005), has occasionally been implicated in human infections (Kratz et aI., 2003). However, such reports are extremely rare, generally involve immunocompromised or seriously wounded patients, and do not conclusively delnonstrate that P. agglomerans is the causative agellt rather than a benign bystander. 55 Summary Figure 1-4 summarizes the multitude of factors that can affect the fire blight disease cycle. The integration of multiple control products into a pest management program that takes advantage of these factors in a commercial orcllard requires some adjustment to usual practices. Forecasting programs exist that can accurately predict peak infection risk periods (Smith, 1996; Steiner & Lightner, 1996), but they must be used differently for biopesticides than for streptomycin. Streptomycin is immediately effective upon application, and active residues will persist for a few days (Smith, Wiens, & Svircev, 2002). In contrast, bacterial antagonists are most effective when applied 48 to 72 h prior to the establishment or growth of significant E. amylovora populations (Wilson and Lindow, 1993; Nucla et aI., 1998). As the attempts to combine BlightBan A506 and C9-1 clearly show, the compatibility of different treatments must also be considered. Each of the described biopesticides is naturally resistant to streptomycin, allowing them to be used as part of an integrated pest management program designed to reduce overall antibiotic use without relying completely on less predictable biopesticides. Other pesticides can also have indirect effects on each other. Prohexadione-calcium, which is applied to apples during the latter half of the bloom period to reduce the incidence of shoot blight, alters the composition of nectar, which can in tum affect the growth of bacteria in the blossolTI (Pusey, 1999; Spinelli et aI., 2005). The prevention of plant disease accounts for a small fraction of total antibiotic use in the f developed world, less than 0.5% in the United States (McManus et aI., 2002). Nevertheless, the desire to reduce agricultural antibiotic use has become a commOll theme in regulatory arenas around the world, and is driving determined searches for more ecologically sound alternativ,es. 56 However, in the case of fire blight, this research continues to emphasize the superior efficacy of antibiotics and the importance of preventing blossom blight. The efficacy of heavy metals is questionable, the commercial release of transgenic tissue is not yet widely accepted by the public, and the individual efficacy of plant growth regulators, and biological controls is generally less than that of streptomycin. This should not necessarily be interpreted as a failure to produce alternatives to antibiotics. Rather, it is a reflection of the nature of n10dern integrated pest management, where dependence on single, potent, pesticides is being replaced by a more diversified set of compatible "reduced risk" treatments. 57 Factors Counteri ng Disease Factors Contri buti ng to Di sease +-- Severe weather. Animal damage Streptomycin ~ Dormant Pru ni ng Ove rvvi nteri ng Cankers Inoculum Copper Sprays ~ Streptomyci n Bacteri al Antago ni sts I Growth reg ulators SAR inducers. Geneti c factors I Blossom Infection I ~ 1. Intercellular grovvth and tissue invasion 2. Secondary infection of vegetati ve ti ssue Alternate Hosts (orname ntals. hawthorn .etc) Vectors (polli nators. rai n wi nd) I Weather +-- Nutritional status and tree vigour I +-- Geneti c factors (scion & rootstck) Figure 1-3. F actors affecting the development of fire blight in the orchard. Both Trauma blight and shoot infections can lead to the formation of overwintering cank.ers~ which provide inoculum for the following season. I Part II: Characterization of Biopesticide Components and Pathogen Growth 59 Chapter 2. Characterization and screening of bacteriopha,ges and Pantoea agglomerans isolates with the potential to inhibit Erwinia amylovora. Abstract The components of a phage-based biopesticide for fire blight were select'ed and characterized in terms of their individual genetic and phenotypic traits, and their interactiol1S. Fifty-six phage isolates and 249 bacterial epiphyte isolates were screen,ed. The genetic diversity of phages was assessed using RFLPs of genomic digests, and by PCR. The in vitro susceptibility of E. amylovora strains and orchard epiphyte isolates to infection by E. amylovora phages was assessed. Biocontrol efficacy was assessed using an in planta assay. Pear shoots bearing dormant buds were harvested in late winter, placed in water at 20°C, and allowed to flower. Individual opened blossoms were inoculated with a phage or carrier candidate, alld then challenged with the pathogen. Disease symptoms were evaluated after 4 d. Ten phages were selected for ful1her testing based on host ranges, growth characteristics, and genetic div,ersity. P. agglomerans Eh215 was selected as the carrier based on in planta biocontrol assays, susceptibility to phage infection, species identity and growth characteristics. P. agglomerans Eh21-5 produces an antibiotic different from pantocin A that inhibits multiple strains of E. am.,vlovora. 60 Introduction The possibility of using Erwinia phages to manage fire blight by controlling orchard populations of control E. amylovora has been suggested several times (Erskille, 1973; Ritchie, 1978; Schnabel et aI., 1998; Schnabel & Jones, 2001; Gill, 2000; Gill et aI., 2003). Phages are obligate parasites, completely dependent on the metabolism of the host cell for their replication. Figure 2-1 depicts the two possible modes of phage replication. In the case of tailed phages, the lytic replication process consists of five basic stages: adsorption, penetration, transitioll to phagedirected metabolism, morphogenesis, and lysis (Guttman, Raya, & Kutter, 2005). The tail fibres or similar structures interact specifically with host cell surface molecules, leading to irreversible adsorption. The cell wall and inner melnbrane are penetrated, alld the phage genome is transferred from the phage head to the host cell, via the tail. The first phage genes are expressed very quickly, and cause the cell's metabolism to switch [roln host-directed to phage-directed processes. Many copies of the phage genome are thell made. The last genes to be expvessed are the ones that encode the structural components of the intact phage pal1icle and the proteins that direct virion assembly. After a certain period of time the phage holin protein illteracts with the ilmer membrane to permit the passage of the endolysin. The endolysin is thell able to digest the 'c'ell wall, causing cell lysis and the release of a new generation of lnature phages troln one initial infected cell (Young & Wang, 2005). Each of these progeny can then illfect the next susceptible cell it encounters, allowing the phage population to increase exponentially at the expel1se of the bacterial host population. Some phages are also capable of lysogenic replication. Under ,certain ,conditions, rapid 61 A D B c Figure 2-1. The lytic phage replication cycle. A) a single, infective Podoviridae; B) irreversible adsorption to an E. amylovora cell; C) progeny phage being assembled within the host cell; D) upon lysis, multiple progeny phage are released from a single infected cell; E) a cell in the lysogenic state carries a copy of the phage genome integrated into its own chromosome. [Micrographs A, B, and Dare <pEa9-5 and E. amylovora, taken by Ronald Smith; C, from Weinbauer & Peduzzi, 1994] 62 expression of a phage-encoded repressor immediately after genome transfer prevents expression of most phage genes, and promotes recombination between the phage and hostgenonles. The phage genome becomes integrated into the bacterial chromosome and is passed to all daughter cells in this prophage form. These cells, called lysogens, are also resistant to super-infection by certain other phages, by virtue of the active repressor. Excision of the prophage genome, DNA replication, morphogenesis, and lysis are only triggered when environmental factors interfere with the continued production or activity of the repressor protein. Depending on the precision of the excision process, bacterial genes may be carried along with the phage genome to its next host, where it can become part of that cell's genome. The consequences of this lysogenic replicatiol1 cycle can include the transfer of pathogenicity or resistance genes throughout a bacterial population. Bacteria can become resistant to a particular phage independent of lysogeny, usually by altering the surface receptors to which phages adsorb. Since nlany of these receptors also have important functional roles, that resistance can have adaptive consequences for the bacterial lineage. The development of resistance can be mitigated by using cocktails of phages that interact with different receptors to infect the same cell (Tanji et aI., 2004; Tanji et aI., 2005). This creates a situation in which the cell could only become resistant in the unlikely event that it acquired mutation(s) that simultaneously conferred resistance to all of the phages in the cocktail. From this overview of phage biology, it is evident that phages must possess certail) distinct traits in order to be useful as a biopesticide. Multiple effective phages are l1eeded, so that they can be applied as a mixture. The selected phages should be exclusively lytic, and 010st should have broad, but overlapping host ranges. 63 In order to maintain replicatillg populations of E. amylovora phages on blossonl surfaces, those pllages must be protected from destruction by ultraviolet light and dessication (Balogh, 2002; Guttman, Raya, & Kutter, 2005; Iria11e et aI., 2007). This can be accolnplished by formulating a protective suspension medium containing optical brighteners or colloidal suspensions of water-soluble protein (Balogh, 2002). The alternative strate,gy developed here uses P. agglomerans, a non-pathogenic bacterial epiph)rt:e that is also susceptible to illfection by the same phages, as a "carrier". Schnabel et al (1998) reported that E. amylovora phage populations declined in the field when applied alone to apple blossoms, but relnain,ed high on clusters inoculated with E. amylovora, where they reduced fire blight illcidence by 26% to 37%. Since virulent E. amylovora would obviously not be applied as pa11 of a treatillent, the authors suggested that phage survival, and thus disease control, might be ellhanced by co-inoculation with avirulent E. amylovora mutants. In this, they neglected to consider the potential for reversion of avirulent mutants, and were limited by only considering phages that infected E. amylovora rather than those with a slightly broader host rallge. The superior exploration of this idea was actually published 25 years earlier when Erskine (1973) reported the discovery ofa lysogenic phage capable of infecting both E. amylovora and "a yellow, amylovora-like saprophyte" fitting the description of P. agglomerans. She noted that none of the disease symptoms associated with E. amylovora inoculation appeared wh'ell pear fruit slices were co-inoculated with E. amylovora and the lysogenized saprophyte, whereas coill0culation with the unlysogenized saprophyte had only delayed and reduced those SynlptOll1S. Erskine ultimately advocated the use of the "phage-infected saprophyte" as a llTeans ofbiolo,gical control of fire blight, correctly noting that phages applied by themselves would be rapidly 64 inactivated by the ambient conditions, and that the saprophyte would provide some 111easure of pathogen inhibition on its own. Erskine did not distinguish between the use of bacterial lyso,gens such as the ones she found, and the use of exclusively lytic phages that happen to infect both bacterial species, as is being proposed in this study. She also supposed that the yellow saprophyte could enter plant tissues to release pha:ges systemically alld halt an existing infection, which the non-pathogenic P. agglomerans has not been seen to do. However, Erskine did articulate one of the most complete visions of how to put phages to practical use in an agricultural setting. P. agglomerans is a non-pathogenic epiphyte that has long been found in association with E. amylovora (Farabee & Lockwood, 1958; Smith & Powell D., 1968; Riggle & Klos, 1972a; Riggle & Klos, 1972b; Erskine & Lopatecki, 1975). Its potential to inhibit E. amylovora has also been well-studied. In culture media, some strains of P.. agglomerans release highly acidic metabolites as they grow, inhibiting E. amylovora growth as a result (Wodzinski, Ulnholtz, Rundle, & Beer, 1994). However, these same strains did not cause a substantial pH chan,ge when growing on ilnmature pear fruit tissue, nor was mediulll acidification in any way associated with a strain's ability to protect pear tissue or apple blossoms from E. amylovora infectioll. Clearly, the in vitro behaviour of P. agglomerans does not necessarily reflect its interactions with E. amylovora in natural infection courts. This is true of antibiotic production as well. Of the 90 antibiotic-producing P. agglomerans isolates described by Wodzinski and Paulin (1994), 84 produced antibiotics that were at least partially inactivated by certain anlino acids. 'Strains whose antibiotics are inactivated by the amino acids that are abundant in young pear shoots alld in apple and pear fruit did not effectively protect pear tissue from E. amylovora infection. In contrast, the antibiotics produced by effective comm'ercial antagonists such as the BlightBan® and 65 Bloorntime® strains were either unaffected by anlino acids or were only affected by those not present in pear tissue. The selection of suitable phages and a P. agglomerans carrier is described here. In 1998, aerial tissue and samples of the surrounding soil were collected fronl rosaceous plants with activ·e fire blight infections (Gill, 2000). Phages infecting E. amylovora were enriched using a nlixed culture of six E. amylovora strains; 44 isolates were collected. Most of these could be classified into six genetic groups, some of which were also able to infect a few strains of P. agglon1erans. This phage collection was revived for the current work. The phages were characteriz'ed in terlTIS of host range, genetic groupings, and detectability by a real-time peR. A collection of orchard epiphytes from southern Ontario, consisting largely of P. agglomerans isolates, was sc:r.eened for susceptibility to the Vineland phages and for antibiotic production. The interactions of the phages with bacteria, and of the carrier candidates with E. amylovora, were then assessed. The'efficacy of the Vineland phages and certain P. agglomerans isolates in reducing fire blight Syt11ptoms was screened using an in planta pear blossom assay. The lnost pronlising phages and carrier candidates were selected for use in the biopesticide development program. ,66 Methods Phage Isolates All phage isolates are shown in Table 2-1. Phages were stored in llutriellt broth at 4°C unless otherwise indicated. Phage Culture Media and Conditions Phages were prepared using either the liquid culture nlethod or the confluent plate lysis method. Unless otherwise indicated, phages were grown on the E. amylovora host indicated in Table 2-1. Liquid cultures were prepared using 8 giL nutrient broth (Difco Laboratories, Sparks, MD). A 250 mL capped flask containing 50 mL of sterile llutrient broth was inoculated with 1 x 109 CFU of the bacterial isolation host, and incubated at 25°C on an orbital shaker at 100 rpm. After 1 h, 1 InL of phage suspension was added to the flask. Flasks were returned to the orbital shaker and incubated for 16-20 h. Chloroform was added to each flask (2%, v/v) and 'returned to the shaker for 20-60 min. The crude lysate was decanted into 50 mL roulld-bottom, FEP centrifuge tubes (fluorinated ethylene propylene, Oakridge), leaving th,e chlorofolID behind, alld centrifuged at 8 000 xg for 25 lllin. The supernatant was syrin,ge-filtered into Isterile 50 lllL polypropylene tubes using 0.2 Ilm surfactant-free cellulose acetate filters (Nalgene, Rochester, NY). Larger liquid cultures were prepared using 500 lllL of nutliellt broth in 1 L flasks, with proportionately larger bacterial and phage inocula. These lysates were centrifuged in 250 mI.; polypropylene bottles, and filtered using 0.2 Ilm GP Express Plus Steri-top filters (Millipore, Billerica, MA ). 68 Isolate a Isolation Host Plant source b Source cPEa31-3 Ea29-7 soil beneath symptomatic Malus X domestica AAFC-Vineland cPEa31-4 Ea29-7 soil beneath symptomatic Malus X domestica AAFC-Vineland cPEa35-2 Ea 17-1-1 soil beneath symptomatic Pyrus communis AAFC-Vineland cPEa35-3 Ea 110 soil beneath symptomatic Pyrus communis AAFC-Vineland cPEa35-4 Eal10 soil beneath symptomatic Pyrus communis AAFC-Vineland cPEa35-5 EaD-7 soil beneath symptomatic Pyrus communis AAFC-Vineland cPEa35-6 EaD-7 soil beneath symptomatic Pyrus comn1unis AAFC-Vineland cPEa35-7A cPEa35-7B cPEa35-7C cPEa35-7D Ea29-7 original <pEa35-7 was isolated from soil beneath symptomatic Pyrus con1munis; source of phages in mixed stock is unknown AAFC-Vineland cPEa45-1 A cPEa45-1 B Ea29-7 original <pEa45-1 was isolated from symptomatic Pyrus communis tissue; source of phages in mixed stock is unknown AAFC-Vineland cPEa45-3 EaG-5 soil beneath symptomatic Pyrus communis AAFC-Vineland cPEa46-1 A 1 cPEa46-1 A2 cPEa46-1 A3 cPEa46-1 C cPEa46-1 E EaD-7 original <pEa46-1 was isolated from symptomatic Malus sylvestris tissue; source of phages in mixed stock is unknown AAFC-Vineland cPEa46-2 EaD-7 symptomatic Malus sylvestris tissue AAFC-Vineland cPEa50-3 Ea17-1-1 unknown AAFC-Vineland cPEa51-1 Ea17-1-1 soil beneath symptomatic Magnolia sp. AAFC-Vineland cPEa51-2 Ea110 soil beneath symptomatic Magnolia sp. AAFC-Vineland cPEa51-3 Ea6-4 soil beneath symptomatic Magnolia sp. AAFC-Vineland cPEa51-4 Ea29-7 soil beneath symptomatic Magnolia sp. AAFC-Vineland cPEa51-6 Ea6-4 soil beneath symptomatic Magnolia sp. AAFC-Vineland cPEa51-7 Ea29-7 soil beneath symptomatic Magnolia sp. AAFC-Vineland cPEa51-8A cPEa51-8B <PEa51-8C Ea 17 -1-1 Ea 17 -1-1 Ea 17 -1-1 original <pEa51-8 was isolated from soil beneath symptomatic Magnolia sp.; source of phages in mixed stock is unknown AAFC-Vineland a The naming conventions within this phage collection were changed during this work. Previously, all numerical phage designations were prefixed with "PEa". This prefix has been replac'ed with "<pEa". b The isolation host is the E. amylovora strain upon which the phage was originally isolated (Gill, 2,000). cThe phages in the Vineland collection were collected from orchards in southern Ontario by J. J. Gill (2000) between 1998 and 2000. 69 Confluent plate lysates were prepared using the molten agar overlay technique (Adams, 1959). Top agar consisted of 8 ,giL nutrient agar (Difco), 2.5 giL yeast extract (Difco), and 5 giL food-grade sucrose. Bacteria were suspended to 1 x 109 CFU/lnL in 0.01 M PB (pH 6.8). One hundred microlitres of the bacterial isolation host suspension was mixed with 100 ~L of phage suspension in a test tube for 10 min. Three millilitres of molten top agar at 55°C was added, and the mixture was poured over a 90 mm petri plate containing solidified nutrient agar. The plate was swirled to evenly distribute the top agar, and then incubated overnight at 37°C. Following incubation, the plate was flooded with 3 mL of nutrient broth and allowed to sit at rOOITI temperature for approximately 15 min. The nutriellt broth and top agar were aseptically scraped into a 50 mL round-bottom FEP centrifuge tube and placed on a gyrorotary shaker for 30 min. Cell debris and top agar were pelleted by centrifuging the mixtures at 8 000 xg for 25 nlin. The supernatant was decanted and syringe-filtered as previously described. Phage Storage Media and Conditions Syringe-filtered lysates were stored in nutrient broth at 4°C. Alternatively, the syringefiltered lysates were centrifuged at 16 000 xg for 45 min. The supernatant was decanted and discarded. The pellet was resuspended in the medium of choice, usually 0.01 M sodiunl phosphate buffer (pH 6.8) amended with 100 mM NaCI and 2 mM MgCl 2 (PB salt). Suspensions were stored in a sterile tube at 4°C. Determination ofPhage Titre Titres of phage stoc](S were detelmined using the nl0lten agar overlay lnethod (AdaITIS, 70 1959). Ten-fold serial dilutions of filtered lysate were prepared in nutrient broth. One hundred microlitres. of each dilution was combined with 100 JlL of expone11tial phase host culture collected from overnight growth on a plate and suspended in 0.01 M sterile sodium phosphate buffer at pH 6.8 (PB) at 109 CFU/mL. The mixture of phage and host bacteria were incubated for 10-15 min at room temperature. Three millilitres of molten top agar at 55°C was added and the mixture of phage, bacteria and top agar was then poured into a petri plate containing nutrient agar. The plate was swirled to evenly distribute the top agar and incubated over11ight at 37°C. Following incubation, plaques were counted 011 any plates where plaques were visible as discrete spots. Alternatively, the spot lysis technique was used, wherein the agar overlay was seeded with only the bacterial host. Once the poured overlay solidified, 5 or 10 JlL of each phage dilution was dropped onto the agar overlay and incubated overnight. Spots contai11ing discrete plaques were used to estimate the titre of the original phage suspension. Revival ofthe Vineland Phage Collection The Vineland phage collection included both liquid lysates a11d lyophilized lysates that had been prepared according to the ATCC skim milk powder fOffilulation. Lyophilized lysates were rehydrated by the addition of 0.5 JlL of nutrient broth. Ten microlitres of the nlix'ed nlaterial was dropped onto a solidified soft agar lawn seeded with lxl0 8 CFU of the isolation host and f incubated overnight. Liquid cultures were tested for the presence of viable phage by dropping 10 JlL of lysate onto a seeded molten agar lawn. If no plaques developed, the entire volume of lysate was transferred to a 50 filL round- 71 bottom centrifuge tube and centrifuged at 16 000 xg for 45 min in order to concentrate any viable phages. The supernatant was decanted and the interior surfaces of the centrifuge tube were washed with 1 mL of nutrient broth. The resulting suspension was then tested for the presence of viable phages. If plaques developed from either the original or the concentrated lysate, a 1DO-fold dilution series of the lysate was prepared in 0.01 M PB, and plated using the soft agar overlay method. A single, isolated plaque was removed from the top agar using a 1 mL aerosol baiTier pipet tip, and placed in 1 mL of nutrient broth. The mixture was vortexed to release phages fronl the agar matrix. A dilution series was prepared from this sample and the process was repeated for at least two more rounds of single plaque isolation, until a stable plaque morphology was observed. When multiple plaque morphologies were observed, three rounds of siIlgle plaque isolation were conducted for each distinct type. If this process resulted in a single, stable plaque morphology, the newly purified lysates were named by adding a letter suffix to their previous designation (ie. phages isolated from the stored <t>Ea45-1 lysate were named <l>Ea45-lA and <t>Ea45-1 B). Isolation ofPhage DNA Two methods were used to isolate phage DNA: organic extraction, and a method using cetyltrimethyl ammonium bromide (CTAB) that was nl0dified froin Manfioletti alld Schneider (1988). Phages were grown in liquid culture on E. amylovora Eal1 0, regardless of which nlethod was used. When DNA was isolated using the CTAB method, bacterial nucleic acids were digested 72 by incubating 100 ng of RNase A and 100 U of DNase I with 10 lnL of syringe-filtered phage suspension at room temperature for 15 min. Nucleases were inhibited and phage pal1icles "lysed by adding 0.8 mL of 0.5 M EDTA (pH 8.0), 0.5 mg of proteinase K, and incubating the mixture at 45°C for 15 min. 440 JlL of 5% CTAB in 0.5 M NaCI was added, and the CTAB:DNA complex was precipitated by cooling the solution on ice for 15 lnin, followed by centlifugation at 8 000 xg for 10 min. The resulting pellet was resuspended in 1.2 M NaCI. DNA was precipitated by adding 2 mL 95% ethanol, mixing by inversion, and centrifugation at 8 000 xg for 10 min. The DNA pellet was washed with 70% ethanol, allowed to ait-dry, and resuspelld-ed in 0.5 mL of I 10 mM Tris-HCI (pH 8.0), and stored at -20o e. The organic extraction method was nl0dified from the New England BioLabs lambda DNA isolation protocol l . Phage lysates were prepared in liquid ·culture and syringe-filtered. Ten millilitres of the lysate was concentrated by centrifugation at 16 000 xg for 45 min at 4°C, resuspended in 700 JlL SM buffer (50 lnM Tris-CI, pH 7.5; 0.1 M NaCI; 8 lnM MgS0 4), and transferred to a 2 mL microcentrifuge tube. Bacterial nucleic acids were digested by adding 1 JlL each of 1 mg/mL DNase I and RNase A and incubating the mixture at 37°C for 30 lnin. An equal volume of 20% (w/v) PEG 8000 in 2.5 M sodium acetate was added, and the tube was vortexed and incubated on ice for 2 h. The precipitated phages were collected by centlifuging the tubes at 15 000 xg for 10 min at 4°C. The supernatant was aspirated and the phage pellet was resuspended in 500 JlL SM buffer. The phage particles were lysed by addillg 5 JlL of 10% SDS, 500 JlLO.5 M EDTA (pH 8.0), and incubating at 65°C for 15 min. Phage nucleic acids were then puritl'ed in a I Available online: tools.neb.com/wolbachia/labsite/protocol<s/lanlbdayhageyreps.htu1 [Accessed S'eptember 2005] 73 three-stage organic extraction. An equal volume of buffer-equilibrated phenol was added alld mixed gently by inversion for 3 min. Tubes were centlifuged at 13 500 xg for 5 min at rOOlll temperature and the upper aqueous phase was renl0ved to a new microcentlifuge tube. The extraction was repeated with an equal volume of 1: 1 phenol:chloroform, and then again with an equal volume of chloroform. DNA was precipitated from the aqueous phase of the final extraction by adding sodium acetate to a final concentration of 0.3 M alld adding 100% ethanol 100 L~ at a time, until the phage DNA had just precipitated (about a IX volume). The precipitated DNA was collected by centrifugation at 15 000 xg for 15 min at 4°C. The supernatant was decanted, the pellet was washed with 70% ethanol and recelltrifuged briefly. The ethanol was aspirated and the DNA pellet was allowed to air-dry before being resuspellded in 10 mM Tris-HCI (pH 7.5) and stored at -20°C. Restriction Fragment Length Polymorphisms (RFLP~) Restriction endonuclease digestions were conducted using MvnI (Roche Diagnostics, Laval, QC), EcoRI (Invitrogen Canada, Burlington, ON), BamHI (New England Biolabs, Ipswich, MA) and BgIII (MBI Ferlllentas, Hanover, MD). MvnI digests were conducted at 37°C in a total volume of25 ~L. Each reaction contained IX Buffer M, 5 U Mvnl, 2-3 DNA, and sterile distilled water. If the DNA concentration was between 50 ng/~L ,L~/gn the reaction was scaled up to a 50 ~L ~g of phage and 100 volullle to accommodate the 'additional volume of DNA required. The reaction products were visualized using using agarose gel electrophoresis. Reaction products were resolved on a 1% (w/v) agarose gel at 150 V for approximately 75 min. The gel 74 was stained in a 0.5 Lm/g~ ethidium bromide solution for 45 min, destained in water for 25 min, and visualized using the GelDoc system (BioRad Laboratories, Hercules, CA). Endpoint PCRfor Group 3 Phages Each phage was tested with polymerase-chain reaction primers, called PEal-A and PEalB (Schnabel and Jones, 2001), designed to amplify a 304 bp fragment of <l>Eal(h). Alnplification L~ reactions were conducted in 25 200 ~M volumes. Each reaction contained 200 ~M of each primer, each of dATP, dCTP, dGTP, and dTTP, 1.5 mM MgCI 2 , 1.5 U Taq (MBI Fermentas), IX polymerase buffer, and 1.5 L~ of phage suspension in nutrient broth. Reactions were run in a GeneAmp 9700 thermal cycler (Applied Biosystelns, Foster City, CA) under the following conditions: 95°C for 2 min; and 30 cycles of 95°C for 30 s, 53°C for 30 s, and 72°C for 30 s. Reaction products were visualized using agarose gel electrophoresis, as previously described. Real-time PCR TaqMan-style real-time PCR reactions were conducted using two sets of probe and primers, called <t>-dpol and <t>-dp02, that were developed by Dr. W. -8. Kim (unpublished) based on the depolymerase (dpo) gene of <t>Eal (NCBI Accessioll no. AJ278614). Detection is based on amplification of a 171 bp ( <t>-dpo 1) or 72 bp ( <t>-dp02) region of the ,gene. Probes were synthesized by Integrated DNA Technologies (Coralville lA, USA), and were labelled with 6carboxyfluorescein (FAM) at the 5' end and either Black Hole Quencher 1 (BHQ-I) or Iowa Black (lAbRQ) at the 3' end.. Each reaction was conducted in a total volume of25 ~L, and contained IX Brilliant 7S QPCR Master Mix (Stratagene, La Jolla, CA), 200 ~M of each primer, 100 M~ prob'e. Reactions were run in a Stratagene Mx4000 Multiplex Quantitative PCR system (Stratagene) under the following conditions: 95°C for 10 min; 40 cycles of 95°C for 30 s and 60°C for 60 s, with three endpoint fluorescence readings during each amplification segment. Transmission Electron Microscopy (TEM) At least 10 mL of each phage suspension was centrifuged at 16 000 xg for 45 min and the resulting pellet was resuspended in 0.5 mL TE buffer (10 mM Tris-HCI, ImM EDTA, pH 8.0). A drop of the phage suspension was placed on ~ 400 mesh copper fonnvar grid. After lInin the excess sample was drawn off by capillary action using a kimwipe drawn against the edge of the grid. A drop of 2% uranyl acetate was immediately placed on the grid. After I min the excess stain was drawn off in the same manner. The sample was allowed to air dry before being examined using a Philips CMI0 transmission electron microscope at an accelerating voltage of 80 kV. The microscope was equipped with a digital imaging system from Anlerican Microscopy Techniques Corp. Micrographs were taken at the University of Westem Ontario, by Ronald Smith. Phage dimensions were automatically calculated from the digital ima,ges based on the number of pixels per micron in the field of view at 72,000 or 105,000-fold magilification. Bacterial Isolates All bacterial isolates used are shown in Table 2-2. Strains plated fronl frozen stocks w,ere assessed for unifonnity of colony morphology, then subcultured from a single colOIlY for each experiment. 76 Bacterial Culture Media and Conditions Bacteria were cultured in 90 mm Petri plates on semi-solid media containing 11.5 giL nutrient agar (Difco Laboratories, Sparks, MD). Erwinia spp. alld Pantoea spp. were incubated at 28°C for 16-20 h. E. coli strains were incubated at 37°C for 16-20 h. Unless otherwise indicated, bacteria were aseptically scraped from the agar surface, suspended in 0.01 M PB pH 6.8, adjusted to 1 x 109 CFU/mL (OD 600 = 0.6, Beckman DU640 spectrophotometer), and stored 011 ice until use. The lOX stock ofPB (pH 6.8) was prepared as described by GOlnori (1955), combining 49 mL of 0.2 M Na2 HP0 4 , 51 mL of 0.2 M NaH2 P04 , and 100 mL of distilled water. Antibiotic production tests were conducted using GA minimal media (Vanneste & Beer, 1992), which contained 20 giL M D-(+)-glucose (Sigma Chemical Co., St. Louis, MO), 0.3 giL L-asparagine (Sigma-Aldrich Co., St. Louis, MO), 11.5 giL K2HP0 4 (J. T. Baker Chemical Co., Phillipsburg,.NJ), 4.5 giL KH 2 P04 (Fisher Scientific, Fair Lawn, NJ), 0.12 giL MgS04 ·7H2 0 (J. T. Baker Chemical Co.), and 50 mglL nicotinic acid (Siglna Chemical Co.). The lnedium was solidified with Noble Agar (Difco Laboratories, Detroit, MI), 12 giL for plates, alld 7 giL for overlays. A 2X stock of GA was prepared and sterilized by filtration, and added ill a 1: 1 ratio to a 2X preparation of melted Noble Agar that had been sterilized by autoclaving. Bacterial Storage Medium and Conditions Bacteria were stored at -80°C in a nlediuln contailling4 giL nutrient brotll, 1 g/L.yeast extract, 2.5 giL glucose, 5.7 mM K2 HP04 , 1.8 mM KH 2P04 , 0.5 mM MgS04 ·7H2 0, and 50% (v/v) glycerol. 77 Table 2-2. Bacterial isolates used in this study Host Plant Use Reference or Source Ea6-4 Pyrus communis Phage Growth; Host Range; Blossom assays (Jeng et aI., 2001) EaI7-1-1 Pyrus communis Phage Growth; Host Range (Jeng et aI., 2001) Ea29-7 Malus X domestica Phage Growth; Host Range (Gill et aI., 2003) Eall0 Malus X domestica Phage Growth; Host Range (Ritchie & Klos, 1977) EaD-7 Pyrus communis Phage Growth; Host Range (Jeng et aI., 2001) EaG-5 Pyrus communis Phage Growth; Host Range (Jeng et aI., 2001) Ea273 Malus X domestica Antibiotic testing Beer, S.Y. Ea 1/79 Cotoneaster sp. Host Range (Falkenstein et aI., 1988) Eh252 Malus X domestica Antibiotic testing (Yanneste, Yu, & Beer, 1992) C9-1 Malus X domestica Antibiotic testing (Ishimaru, Klos, & Brubaker, 1988) E325 Malus X domestica Host Range (Pusey, 1997) Eh21-5 Pyrus communis Phase 2 carrier candidate testing A. M. Svircev 1-28b Malus X domestica Phase 2 carrier candidate testing A. M. Svircev (1997) 13P-4 (2) II Pyrus communis Phase 2 carrier candidate testing A. M. Svircev 13 P-5 (1) II Pyrus comn1unis Phase 2 carrier candidate testing A. M. Svircev 17P-3 (2) II Pyrus communis Phase 2 carrier candidate testing A. M. Svircev (1997) 21-1-1-2 Pyrus comlnunis Phase 2 carrier candidate testing A. M. Svircev (1998) 21-1-5-1 Pyrus communis Phase 2 carrier candidate testing A. M. Svircev(1998) 39C II S. aucuparia (European Mountain Ash) Phase 2 carrier candidate testing A. M. Svircev (Royal Botanical -Gard·ens, Burlington, ON) Isolate Erwinia amylovora a Pantoea agglomerans b (1998) 7'8 Strain Host Plant Use Reference or Source 39L II 2 Dorothea Crab Phase 2 carrier candidate testing A. M. Svircev (1998, Royal Botanical Gardens, Burlington, ON) 39U II Cotoneaster spp. Phase 2 carrier candidate testing A. M. Svircev (1998, Royal Botanical Gardens, Burlington, ON) 39V II Cotoneaster amigus Phase 2 carrier candidate testing A. M. Svircev (1998, Royal Botanical Gardens, Burlington, ON) 39W II Cotoneaster (Lucida) Phase 2 carrier candidate testing A. M. Svircev (1998, Royal Botanical Gardens, Burlington, ON) generally, Pyrus communis or Malus X domestica Phase 1 carrier candidate testing A. M. Svircev; 1997 and 1998 Other 244 isolates collected from blossoms in southern Ontario D. Cuppels Pectobacterium carotovora, Ecc26 C Pseudomonas jluorescens A506 field testing (Lindemann & Suslow, 1987) Pseudomonas syringae morsprunorum, psm7 Environmental Impact Testing Teresa Ainsworth P. syringae morsprunorum, psm6 I Environmental Impact Testing Teresa Ainsworth P. syringae morsprunorum, psm37 Environmental Impact Testing Teresa Ainsworth P. syringae papulans, psp mutsu Malus X don1estica, mutsu Environmental Impact Testing D. Hunter Xanthomonas campestris pv. pruni, Xc69 Prunus sp. Environmental Impact Testing Teresa Ainsworth Xanthomonas campestris pv. pruni, Afl Prunus sp. Environmental Impact Testing Teresa Ainsworth b b CUCPB, Cornell University Collection of Phytopathogenic Bacteria Agriculture & Agri-Food Canada, Sou~hern Crop Protection and Food Research Centre, Vineland, ON C Agriculture & Agri-Food Canada, Southern Crop Protection and Food Research Centre, London, ON a b 79 Antibiotic Production by P. agglomerans P. agglomerans Eh21-5 was harvested froln a fresh, overnight plate culture and washed once with GA medium to remove trace nutrients from the rich medium. Cells were suspel1ded in 1 mL 0.01 M PB, centrifuged at 13 000 xg for 5 min, and resuspended in 1 mL of fresh PB. A 125 mL Erlenmeyer flask containing 25 mL of liquid GA medium was inoculated with the resuspended cells. P. agglomerans C9-1 and E. amylovora Eall 0 were used as positive and negative controls, respectively. After 48 h, cells were removed from the culture media by centrifuging the suspension at 8 000 xg for 25 min and syringe-filtering the supernatant through a surfactant-free cellulose acetate filter with 0.2 ~m-dia et r pores. Filter-paper discs, 6 mm in diameter, were soaked in the filtered supernatant for 2-3 min. Excess liquid was blotted from the disc, and it was placed on a GA plate that had been overlaid with GA top agar seeded with lxl0 9 CFU/lnL of the indicator strain. Alternatively, 30 ~L of filtered growth media was placed on the disc and allowed to soak through it. Plates were incubated at 28°C, and checked after 24 and 48 h. A zone of inhibited growth on the bacterial lawn around the location of the infiltrated disc indicated production of an antibiotic compound by the test strain. Amplification oja Gene Jrom the Pantocin A Biosynthetic Cluster P. agglomerans carrier candidates were tested for the presence of the paaB gene USil1g i primers designed to amplify an 813 bp developed by Jin et al (2003). Reactions were conducted in 50 ~L volumes. Each reaction contained 200 ~M of each primer, 1.5 mM MgCI2 , 200 each of dATP, dCTP, dGTP, and dTTP, 1 U of Taq polymerase (MBI Fem1entas), IX ~M 80 ThermoPol buffer, and 5 ~L of template. Reactions were carried out in a GeneAmp 9700 themlal cycler (Applied Biosystems) under the following conditions: 95°C for 2 min; 30 cycles of 95°C for 15 s, 53°C for 15 s, and 72°C for 30 s. Reaction products were visualized on a 0.8% agarose gel, as previously described. Templates were prepared by scraping a small amount of bacterial -cells from a nutrient agar plate and suspending them in 0.5 mL of sterile, distilled water. P. agglomerans Eh252 and C9-1 were used as positive controls. E. amylovora Ea6-4 and Ea273, E. pyrifoliae 1/96, P. fluorescens A506, water, and master mix were used as negative controls. Host Range ofthe Vineland Phage Collection on E. amylovora and P. agglomerans Isolates Bacterial isolates were tested for in vitro susceptibility to phage infection usin,g the spot method, a modification of the soft agar overlay method. A bacterial lawn was created by seeding 3 mL of molten top agar with 100 L~ of lxl0 9 CFU/mL of the test isolate and pouring the mixture over a nutrient agar plate. Ten microliters of a 1 x 10 7 PFU/mL phage suspension were dropped onto the overlay and the plate was incubated overnight. Two hundred and fifty-six bacterial isolates that had previously been collected from the aerial tissue of rosaceous hosts in southern Ontario were considered as part of the carrier selection process. Ninety-eight of these had previously been identified as Pantoea agglomerans based on PCR amplification of a 16S-23 S intergenic region, as described by Jeng et al (2001) (A.M. Svircev, unpublished data). The host range of the Vineland phage collection was also tested on the six E. amylovora isolation hosts, several strains of P. agglomerans from other sources, and on bacterial strains 81 isolated from local orchards: Pseudomonas syringae morsprunorum Psm7, P. syringae morsprunorum Psm61, P.syringae morsprunorum Psm3 7, P. syringae papulans psp mutsu, Xanthomonas campestris pv. pruni Xc69, and X campestris pv. pruni Aft. Efficacy ofCarrier Candidates and Phages in Blossom Assays The biocontrol activity of individual bacteriophages, carrier bacteria, and 'con1binations of the two was evaluated using a pear blossom bioassay. Pear shoots beari11g dormant buds were harvested in later winter. Budwood was bundled, loosely wrapped in clear plastic, and stored at I°C until use. The lower 15 to 20 cm of each branch was surface disinfected by dipping theln in 70% ethanol and then cutting 2 to 5 cm off of the bottom with clean pruning shears. Budwood was then forced to form blossoms by placing them in tap water at 20°C. Branches were surface disinfected, re-cut, and placed into fresh water every 3 to 4 d. Individual newly opened blossoms were collected by hand and placed into sterilized glass scintillation vials containing sterile tap water, such that the peduncle extended through a hole drilled in the lid of each vial. To determine the minimum c011centration of E. amylovora Ea6-4 11eeded to produce full 1 disease in the untreated control, I0 ~L of E. amylovora Ea6-4 at 1x I0 8 CFU/n1L or Ix 106 CFU/mL or Ixl04 CFU/n1L was applied to each of five blossoms. PB was used as a control. This entire experiment was repeated at a later date with the following changes: sets of lOblossoms were used for each treatment; treatments were repeated in two more sets of 10 blosSOll1S, using independently prepared bacterial cultures for each set and suspensions of E. amylovora Ea6-4 prepared to Ixl0 8 CFU/mL, Ixl0 7 CFU/lnL, Ixl0 6 CFU/mL, Ixl0 5 CFU/lnL, and PB. '82 To screen for ptotective effects of previous treatment with a phage or candidate carri·er bacterium, 10 individual blossoms were treated with 10 L~ L~ of phage at lxl0 8 PFU/111L or with 10 of carrier bacteria at lxl0 8 CFU/mL, and then challenged by applying 10 L~ Ea6-4 at lxl0 8 CFU/mL. Controls consisted of treatment with 10 L~ E. amylovora Ea6-4, or 20 ~L of E. amylovora ofPB followed by 10 L~ of of PB with no subsequent application of E. amylovora. This inoculation procedure was repeated using independently prepared bacterial cultures. D,ep'ending on the number of phages or bacteria being screened, the interval between the treatment and pathogen applications was approximately 10 to 20 Inin. Vials were held in plastic racks, and loosely sealed inside large plastic bins that had been flooded with 500 mL of tap water in order to maintain a high relative humidity. Blosson1s were incubated at room temperature, about 22 to 25°C. After 4 d, disease symptoms were evaluated according to the rating scale in Figure 2-2. Results were analyzed in SAS (Statistical Analysis Systems 8.2; 'SAS Institute, Cary, NC) using the general linear model (PROC GLM). Disease severity index ratin.gs from the competition assay were converted to a percent scale based on the total hypanthiunl and ovalY surface area showing visible necrosis (0 == 0%, 1 == 20%, 1.5 == 38%, 2 == 48%, 2.5 == 55%, 3 == 61 %, 3.5 == 68%, 4 == 82%, 5 == 100%) . The differences alnong different levels of pathogen inoculum were determined using Duncan's multiple range test. The differences in disease severity between blossoms treated with buffer and blossOlTIS treated with an individual phage or carrier candidate were analyzed ising a one-sided Dunnett's test for multiple 'comparisons to the specified PB control. 83 Top o r;) V o 1 Side 2 1.5 4 3 2.5 5 3.5 Figure 2-2. Rating scale describing the severity of fire blight syn1ptolTIS in the pear blosson1 bioassay. Integer values are based on the vertical progression of necrosis through the blossom head to the peduncle. Half-scale ratings are based on the furthest vertical progression of symptoiTIS, less a half point if symptom progression extends less than half the way around the nectary at that distance. [Black and white graphic is taken frOITI Gill (2000); Blossom photos were taken by S. M. Lehman] 84 Results Revival ofthe Vineland Phage Collection When this work commenced, three stored copies of the Vineland collection existed: filtered liquid lysates prepared by J. J. Gill ca. 2000, stored at 4°C ill nutrient broth over a drop of chloroform (Gill, 2000); filtered lysates prepared by M.Schmuck ca. 2003 fronl the Gililysates, stored at 4°C in nutrient broth; and lyophilized lysates prepared by J. J. Gill ca. 2000 in skilllmilk medium. No viable phages were recovered froln the lyophilized lysate. .l~ of the phages collected by J. J. Gill (Gill, 2000), except for <t>Ea31-1, <t>Ea45-2 and <t>Ea50-2, were recovered from at least one of the liquid lysates. A stable plaque morphology was observed for most phage isolates. However, certain isolates, most notably the group 1 phages, have a variable plaque morphology. Su'cc'essive rounds of single plaque isolation, in which the different morphologies were carefully noted, always yielded the same mixture of plaque appearances. Generally, plaques noted to be of one appearance would resemble the other type if the plate was allowed to incubate lon.ger. When multiple plaque morphologies were observed, comparative RFLP allalysis was COlldu'cted on the genomic DNA of each isolate. If the same RFLP pattern was observed for 'each one, th,ey were assumed to be the same phage and the letter suffixes were dropped. Molecular Characterization ofPhage Collection RFLPs were used to compare the recovered phages to the original collectioll, as described by Gill (2000). In that work, BamHI, EcoRI, BglII, and ThaI were used to digest DNA. Only ThaI digested all phage genolnes, and gave a ullique pattern for all the described :groupings. Therefore MvnI, which is an isoschizomer of ThaI, was used in the present study. The phages were also classified based on two PCR assays. The endpoint PCR assay was taken from Schllabel and Jones (2001), who cloned and sequenced a 1.8 kB BglII fraglnent of <l>Ea1, and designed prim-ers to amplify a 304 bp region of it. The real-time PCR assays use two primer and probe sets designed by Dr. W. -So Kim, based on the sequence of the <l>Ea1 depolynlerase gene. Table 2-3 shows the results of these assays. Transmission Electron Microscopy Some of the phages selected for phage-carrier biopesticide development were examined by TEM. The sizes and family assignlnents of these phages are described in Table 2-4. Phages with short tails are members of the Podoviridae (Figure 2-3B). Of the relnaining, long-tailed, phages, some were observed in both the contracted and uncontracted state, and have therefore been classified as Myoviridae (Figure 2-3 ACD). Curved tails with no apparent narrowing of the neck (the proximal end of the tail), such as those observed in the preparations of <l>Ea35-4 and <l>Ea35-5, are characteristic of the non-contractile Siphoviridae (Figure 2-3E). The morphologies of <l>Ea9-5, <l>Ea 21-3, and <l>Ea31-3 that were detemlined ill this study are consistent with those reported by Gill et al (2003). The morphologies of the relnaining phages were not previously reported. However, the morphology of <l>Ea35-4 is consistent with the morphologies reported for other group 2 phages (Gill, 2000; Gill et aI., 2003), while the morphology of <l>Ea35-5 is not the Podoviridae morphology that was reported for other group 4 phages (Gill, 2000; Gill et aI., 2003). 86 Table 2-3. Molecular characterization of the revived phage collection. RFLP confitmation of the group assignment made by Gill (2000) is given, as are the results of three peR assays. Gill G roup 3 Phage Name RFLP b Confirmation PCR Assay pEa 1 p-dpo 1 C p-d,po2 Ea6-4 <pEalO-2 + <pEa 10-3 <pEa 10-4 + <pEa21-1 2 <pEa21-2 1 <pEa21 -3 1 <pEa21-4 1 + EaD-7 (+) EaG-5 EallO + + (+) EaD-7 (+) Ea6-4 <pEa 31-1 + + EallO <pEa35-2 + + Ea17-1-1 <pEa 10-5 + + EallO + Ea29-7 <pEa31-2 + <pEa31-4 <pEa35-4 2 (+) EaD-7 + Ea29-7 (+) + <pEa35-6 <pEa35-7 A + EallO + EaD-7 (+) Ea29-7 <pEa35-7B Ea29-7 <pEa35-7C Ea29-7 <pEa35-7D Ea29-7 <pEalO-7 3a <pEa 10-8 + + Ea29-7 + + Ea29-7 + + EallO <pEa 10-10 + + EallO <pEa 10-11 + + Ea 17-1-1 + + Ea6-4 + + EallO EallO <pEa 10-9 <pEa10-13 3a 3a <pEa10-14 <pEa 10-15 3a + + <pEa31-3 3a + + + + EaD-7 + + EallO 3b <pEa46-2 3c <pEal 4 + EaG-5 <pEalO-6 3a Isolation Host 3c + Ea29-7 <pEalO-12 (+) EaG-5 <pEalO-16 (+) Ea 17-1-1 + + EallO + + EaD-7 <pEa3 5-3 <pEa35-5 4 Gill G roup 3 Phage Name 5 <pEa9 -2 RFLP b Confirmation PCR Assay PEal p-dpo2 + EaG-5 <pEa9 -5 5 + + <pEa5l-l 6 + + Ea 17-1-1 <pEa5l-2 3a (+) EallO 6 (+) Ea29-7 Ea6-4 (+) <pE a5l-3 <p Ea5l-4 Ea6-4 Ea6-4 <pE a5l-6 6 (+) <pE a5l-7 6 (+) Ea29-7 <pEa5l-8A Ea17-l-l <pEa5l -8B Ea17-l-1 <pEa5l-8C Ungrouped Isolation Host Ea17-l-l <pEa9-4 6 p-dpol C Ea17-l-l + <pEa9 -3 + (+) <pEa 10-1 <pEa45-lA Ea29-7 ungrouped (+) <pEa45-l B + <pEa45-3 + <pEa46-l A 1 EaI7-1-1 Ea29-7 + + NA EaG-5 + + + + + + EaD-7 <pEa46-l C + + EaD-7 <pE a46-lE + + EaD-7 <pEa 50-3 + (+) Ea17-l-l (+) Ea29-7 <pEa46-l A2 Not digested <pEa46-l A3 pEalO-17 3b EaD-7 (+) EaD-7 The RFLP groupings assigned by Gill (2000), based on ThaI, BgIII, BamHI, EcoRI. b RFLPs are based on MvnI (an isoschizomer of ThaI), BgIII, BamHI, EcoRI. 6 C Amplification results from stocks containing at least 1x 10 PFU/mL are shown as: +, positive; (+) weak positive; -, negative. A blank cell indicates that the phage was not tested in that assay. a 88 Table 2-4. Family-level characterization of selected E. amylovora phages based on TEM. Phage Family Approximate Head Width (om) <pEa9 -S Podoviridae 59-65 <pEal0-l Myoviridae 70-81 <pEa21-3 Myoviridae 60-80 <pEa31-3 Podoviridae 58-66 <pEa3S-4 Siphoviridae 113-129 <pEa3 S- S Siphoviridae 53-61 <pEa4S-1B Myoviridae 60 <pEa46-1 A2 Myoviridae 105-113 <pEa51-1 Myoviridae 113 89 B c Figure 2-3. TEM of five phages of E. amylovora. A) <l>EalO-l, B) <l>Ea31-3, C) <l>Ea45-1B, D) <l>Ea46-1A2, E) <l>Ea35-4. [micron marker = 50 nm] 90 Antibiotic Production by P. agglomerans Several of the candidate carrier bacteria were tested for antibiotic production. Table 2-5 summarizes the results of those tests. The weaker ZOlles of inhibition on E. amylovora Ea6-4 alld Ea110 were produced by filter discs that had been inoculated with 30 ~L of the respective aseptic growth medium, instead of having been soaked in it. The antibiotic activity of n1edia in which P. agglomerans Eh21-5 and C9-1 had been grown was still present at a similar level after one week of storage at 4°C. Figure 2-4 shows the zones of inhibition produced by ditTerent antibiotics. The clear, inner zone of inhibition produced by 39CII and C9-1 is similar to that repolied for pantoeill A (Ishimaru et aI., 1988; Vanneste et aI., 1990). The larger, more diffuse zone of inhibition produced by C9-1 pantocin B appears similar to the 12 mm zones of inhibition produced by Eh21-5 and many of the other orchard P. agglomerans isolates. The clear zone of inhibition produced by 13P-5(2)II is different from either pantocin A or B. The paaB primers amplified a DNA fraglnent just larger than 750 bp from Eh252, PaC91, 39CII, and 1-28b (data not shown). 91 Table 2-5. Growth inhibition of Erwinia spp. by P. agglomerans antibiotics. The radii of growth inhibition (mm) is given for each producer-indicator combinatioll. Indicator Producer P. agglomerans C9-1 E. amylovora Ea273 E. amylovora Ea6-4 E. amylovora Ea110 3; (12) 5 3; (18) a c NT 3 Eh21-5 (12) (12) ( 12) 3 4 3 NT 39L11(2) (12) (12) (12) NT 39U11 (12) (12) ( 12) NT 39W11 (12) (12) ( 12) NT 13P-4( 1)11 (12) (12) ( 12) NT 13P-5(2)11 8 13 11 NT (12) (12) (12) NT 21-1-1-2 NT 21-1-5-1 1-28b E.an1ylovora (12) (12) Ea273 uninoculated GA media numbers in brackets indicate weak inhibition in a zone of the given radius. indicates that no growth inhibition was observed C "NT" indicates that the combination was not tested. a b "_" NT Eh252 39C11 NT E. pyrifoliae Ep 1/96 (12) NT NT 92 Figure 2-4. Growth inhibition of E. amylovora by P. agglomerans antibiotics. Hatched lines indicate the dimensions of the zones of inhibition produced by extracellular metabolites of A) strain C9-1, B) strain Eh21-5, C) strain 13P-5(2)II, and D) negative controls: uninoculated GA (top) media and E. amylovora Ea273 (bottom). 93 Host Range ofthe Vineland Phage Collection Table 2-6 summarizes the host ranges of the Vineland collection on E. amylovora and P. agglomerans. Among the E. amylovora strains, the group 3 <pEal-like phages and the group 6 phages have different host range pattelTIS than the other groups. Many of them only infected EaIIO, EaD-7, Ea29-7, and Eal/79, strains that produce noticeable amounts ofEPS e\'en wh·en grown on nutrient agar without additional sucrose. The host range patterns of these phages in P. agglomerans were much more variable than in E. amylovora. Only the group 3 phages showed noticeable trends, with most P. agglomerans strains being infected by <pEal (group 3C), and few strains being infected by the local group 3 isolates. Susceptibility ofCarrier Candidates to Phage Infection Two hundred and fifty-six isolates of P. agglomerans and u11identified orchard epiphytic bacteria were considered. The results are shown in Figure 2-5. The 42 isolates that were eliminated from further testing generally did not flourish on nutrient agar or formed colonies that were not easily dispersed in liquid media. Therefore these isolates would not be cOlnpatible with the methods currently employed for large-scale biopesticide production. In Phase 1 of the screening process, the remaining 214 isolates were tested for susceptibility to infection by 10 of the Vineland Erwinia phages: <pEa9-4, <pEa9-5, <pEal 0-1, <pEa21-3, <pEa31-3, <pEa35-4, <pEa35-5, <pEa45-lB, <pEa46-lA2, <pEa51-1. These phages had been chosen based on their perfom1ance in the in planta pear blossom assay, alld for their diversity in terms of original isolation site and original RFLP grouping (Gill, 2000). Of the 109 isolates that were infected by 5 to 10 phages, 12 were selected for th·e phase 2 94 Table 2-6. Host range of the Vineland phage collection on Erwinia and Pantoea. For each combination of phage group and host strain is indicated the nun1ber of phages in that group that "+" infected the host, "(+)" weakly infected the host, or "-" did not infect the host. Group 1 Host Group 2 a Group 3C Group 3B Group 3A + (+) - + (+) - + (+) - + (+) - + (+) - Ea6-4 7 0 0 6 1 3 2 2 4 0 0 1 0 0 1 EaI7-1-1 7 0 0 4 1 5 2 0 6 0 0 1 0 0 1 Eall0 7 0 0 10 0 0 8 0 0 1 0 0 1 0 0 EaD-7 7 0 0 9 1 0 8 0 0 1 0 0 1 0 0 Ea29-7 7 0 0 7 3 0 8 0 0 1 0 0 1 0 0 EaG-5 7 0 0 2 3 5 2 0 6 0 0 1 0 0 1 Eal/79 7 0 0 10 0 0 8 0 0 1 0 0 1 0 0 E325 0 5 2 3 5 2 0 1 7 0 1 0 0 0 1 1-28b 5 0 2 7 0 3 2 1 5 0 0 1 1 0 0 13P-5(2)II 4 1 2 7 0 3 2 1 5 0 0 1 1 0 0 Eh21-5 4 0 3 4 1 5 0 3 5 0 0 1 1 0 0 39CII 0 5 1 3 1 6 0 0 8 0 0 1 1 0 0 39LII 5 1 0 3 4 3 1 I 6 0 I 0 I 0 0 E. amylovora P. agglomerans GroupS Group 4 Host Ungrouped Group 6 + (+) - + (+) - + (+) - + (+) - Ea6-4 3 0 1 2 1 0 I 0 5 11 0 6 EaI7-I-1 3 0 1 3 0 0 1 1 3 II 0 6 EaIIO 4 0 0 3 0 0 5 0 0 15 2 0 0 0 E. amylovora EaD-7 3 1 0 0 3 0 5 0 0 17 Ea29-7 3 I 0 0 3 0 5 0 0 16 I 0 EaG-5 3 0 1 3 0 0 I 0 4 9 2 6 Eal/79 4 0 0 3 0 0 5 0 1 17 0 0 Pa E325 0 0 4 0 0 3 4 0 1 0 4 13 1-28b 2 I 0 2 0 1 0 0 6 5 1 II 13P-5(2)II 1 0 3 .0 0 3 1 1 3 5 3 9 1 0 3 0 2 6 4 7 P. agglomerans a Eh21-5 1 3 0 1 39CII 1 1 1 2 0 1 0 I 4 7 3 6 39LII 2 0 1 3 0 0 1 3 1 7 4 5 RFLP group assignments are those of Gill (2000). 9S screening. They were selected to represent a range of geographical locations and host plant species, as well as varied perfornlanoe in Phase 1 to ,colnpensate for the possibility that the 10 phages used in that screen were less diverse, and therefore a more biased selection tool, than originally thought. Table 2-7 shows the results of the phase 2 s<creening. Four cani'er 'candidates, 21-5, 39LII2, 39VII, and 39WII, were infected by more than 30 of the 54 phages. Susceptibility ofCommon Orchard Bacteria to Injection by E. amylovora Phages Gram-negative bacteria that had previously been isolated from southelTI Ontario orchards were screened for susceptibility to E. amylovora phages in order to assess the potential impact of a phage-based biopesticide on orchard microbial ecology. None of the Pseudomonas syringae pv. morsprunorum isolates or Xanthomonas campestris pruni isolates were infected by <pEal or allY of the E. amylovora phages in the Vineland collection. Some of the group 1,2, 3a, and ungrouped phages infected P. syringae pv. papulans, Psp mutsu, weakly for the most part, and ill a pattern similar to that on P. agglomerans E325. Efficacy oJCarrier Candidates and Phages in Blossom Assays In the first set of experiments designed to optimize the pathogen inoculunl, a 1 x 108 CFU/mL, suspension of E. amylovora Ea6-4 was required ill order to cause an average synlptom severity of 80% or greater (P < 0.05). In the more detailed experinlent COllducted later, treatnJent with 1 x 106 CFU/mL was the minimum inoculum needed. Treatm'ent with 1 x 105 CFU/ll1L of E. amylovora caused a mean of 55% disease, which was greater than the Ullill0culated blossoms, 96 256 Bacterial isolates I I I 42 214 Elilninated because of undesirable growth characteristics Tested for susceptibility to infection by 10 phages I J 138 76 "Hosts" "Non-hosts" I I 109 29 Infected by 5 or more of the tested phages Infected by ] to 4 of the tested phages 12 Tested for susceptibility to infection by all 54 phages Eh21-5 Figure 2-5. Screening bacterial epiphyte isolates for susceptibility to infection by E. amylovora phages. 97 Table 2-7. Host range of Erwinia phages on P. agglolnerans carrier candidates. Phage groupings are those assigned by J. J. Gill (2000), including the ungrouped ("NA") phages. Shaded cells indicate that the bacterial isolate was susceptible to infection by the phage. White cells indicate that the isolate was not susceptible. "NT" indicates that th·e combination was 110t tested. P ha e etlEa 9-3 Grou ·(6"E·a···i:i:'·1 v •••• ' ••.••• etlEa '0-5 etlEa '0-14 .-= l~= t= t=C· · · · , ·, ·, ·, ., etlEa '0-17 ................................ . . . ." ., "., ., " · A1· : §"i~ a /'·)j( .......................................... 1------10 etlEa 45-18 etlEa 45-3 _ .........................".,.,.,.,.".,.,.,.,.".. li li ~ f- - I.· ." ,. "., ., ., ". ··a;·§··a···46:···t:\···1···.. .· · · NA === ................................... 1 - - - - - + - - etlEa 46- tA.2 ·'·a;·'Eta·'·4'i3':·1A:·s·,··,·,·• • • _ _......-_ _+ - _ ~ +- _ J· . : · Gj; § .4· : ? ~ .PC·: etlEa 46-1E etlEa 50-3 · A8 ~ 5 a § ; _ _........._ _-+-_ _ . ····· ........................................ etlEa 51-88 etlEa 51-8e etlEa '0-2 ·(j)"E~ i l = 3· ' , ·(i) '[~ a 15:· r · ................................. etlEa 21-1 '<D§'a"'2'1-2 etlEa 21-3 4·~f: a· E )j( etlEa '0-6 etlEa 31-2 · ·4~ 3 · a §;· 2 etlEa 35-2 ·(j)·Ea···§..§·=4..· · · · · · . "'6 ~5 3 a" E'D< (i)E'a35~7" etlEa '0-7 etlEa '0-8 etlEa '0-9 . '<D Ei"15~ b ' 3a ~~~~~ $[:·a··. .15=·13·····1 etlEa'O-" . ··· 51·~ i · a E)i(· etlEa 31-3 etlEa 46-2 etlEa1 etlEa '0-12 3b 3c 4 . (D·E·a·..·61~i ' 3~5' a E )i(' (DE'a'S'5=g' etlEa etlEa etlEa etlEa etlEa 5 6 .. "3~5aED< 9-2 9-4 9-5 51:-1 51-2 <DEa..4tf~· . ·· $'E"a"'§"t6 NT NT 27 28 etlEa 51-7 Total 24 27 32 26 34 29 32 35 98 but less than the more concentrated treatments, all of which caused at least 90% disease (P < 0.05). Results similar to this second experiment were also obtained when the E. amylovora suspensions were applied using a custom-made air pressure-driven atomizer at a spray rate of 15 to 70 ~L/cm2. The carrier candidates that were screened for susceptibility to phage infection in the Phase 2 screening were also tested in blossom assays. Very few of thenl had any etTect on th,e development of fire blight symptoms. Only blossoms treated with 39CII or 39LII2 showed a significant reduction in disease severity (P < 0.05) 4 d after inoculation with E. amylovora, ranging from 34 to 45% less diseased surface area than the untreated controls. BlOSSOlTIS treated with Eh21-5 and 39UII showed 21 and 13% reductions in symptom severity, resp'ectively, though these were not statistically significant. Very few phages caused a noticeable reduction in symptom severity, and statistically significant reductions in disease severity were only observed for <pEa10-1, <pEa10-6, <f>EalO-7, <pEa10-8 (P < 0.05). Other phage treatments resulted in a mean symptom severity less than that of buffer treated blossoms, even though the difference was not statistically sig11ificant: from group 1, <pEa2l-3, <pEa2l-4; from group 2, <pEal 0-6 and <pEa35-4; of the group 3 phages, <pEa10-15 and <pEa3l-3; from group 4, <pEa35-5; from group 5, <pEa9-5; from group 6, <pEa51-1, and <pEa5l-2; of the ungrouped phages, <pEalO-l, <pEa45-1B, and <pEa46-lA2. There was a general trend towards less severe symptom development in blossoms treated with an ungrouped phage. 99 Discussion A selection of molecular and morphological, and infection characteristics of the Vineland collection are presented here. Molecular analyses indicate that the revived phage collection is not identical to the original collection (Gill, 2000; Gill et aI., 2003). The RFLP patterns of some isolates indicate that the stocks became mixed, and the original phage was not re-isolated during revival. Some of the stocks exhibit plaque morphologies, and even RFLP patterns, that are consistent with the isolate description, even though an PCR amplicon was produced based on the <pEal primers designed by Schnabel and Jones (2001). This latter case suggests that the stocks may be mixed, with the group 3 phage present in sufficient concentration to be detected by PCR. COlllparisons between the current collection and previous descriptions (Gill, 2000; Gill et aI., 2003) should be drawll with care. There were some differences among the results of the three PCR assays (Table 2-3), but most strong real-time PCR signals were obtained froln phage stocks that also gave a positive result with the <pEal primers. The differential success of the two real-time PCR assays indicates variation in the depolymerase gene sequence between phages that apparently possess the gelle. The fact that some phages could not be amplified by any of the three PCR assays (Table 2-3) is consistent with other indicators of the diversity of the Vineland collection. All are based on sequence data from <pEal, and while nlany E. amylovora phages are Podoviridae similar to <pEal, there is no reason to expect that all of thelTI share substantially similar .genom'es. The EP'Sdegrading enzyme on which the real-time PCR detection was based is not even common to all Podoviridae of Enterobacteriaceae (Geller et aI., 1998). Real-tinlePCR will eventually be used lOO to quantify the populations of phage, carrier, and pathogen during field trials, alld to track the environmental fate of the phages after their application. Not all of the phages selected for use in field trials need to be detectable by these real-time peR prinlers and probes, but some of them do so that the population dynamics of the phages can be related to those of the carrier and pathogen, as well as to the overall efficacy of a given treatment. The host range of the Vineland collection on E. amylovora straills was slightly different from that reported by Gill (2000), which is consistent with the results of the collection revival and the molecular characterization data. However, the characteristic host range patterns of group 3 and group 6 phages were similar to those described previously. Group 3 phages belollg to Podoviridae, and produce depolymerase, as evidenced by the expanding halo around each plaque. Most of these phages only infected the E. amylovora strains that produce noticeable amounts of EPS even on media that did not contain any added sucrose. The group 4, group 6, and some of the ungrouped phages are also nlembers of the Podoviridae (Gill et aI., 2003). 'Ofthese, only the group 6 phages showed this same strong bias towards infecting strains with abundant EPS. However, none of the group 4 or ungrouped phages failed to infect those sa111e straillS, whereas some phages in the other groups did. With the exception of phage S1 (Erskine, 1973), none of the previously described E. amylovora phages were reported to infect other bacterial species, eve11 the closely related P. agglomerans (Ritchie & Klos, 1979; Schnabel & Jones, 2001). V-ery different results were obtained in the Phase 2 carrier screens conducted here, where nlany P. agglomerans isolates were infected by multiple phages (Table 2-7). Previous studies have used on a single bacterial host strain in their phage isolation protocols and have mostly recovered <pEal-like phages, which did 1{)1 exhibit narrower host ranges in this study as well. Previous studies also tested only 1 to 3 strains of the species in question. Even within E. amylovora there are strains which are resistant to infection by many or most of the phages that will infect other strains. Most phages are highly specific to one or a few bacterial species, with a few notable exceptions. 0-81 infects multiple strains of at least 17 different Pseuodomonas species and biotypes (Kelln & Warren, 1971). PI infects multiple enteric species (Yarmolinsky & Steinberg, 1988). Phage mu also has a broad host range, in this case as the result of a genetic switch that introduces variation to the structure of its tail fibres (van de Putte, Cranler, & Giphart-Gassler, 1980; Grundy & Howe, 1984; Plasterk, Kanaar, & van de Putte, 1984). The host range data collected in this study show the Vineland phages to be more promiscuous than Inost phag·es are thought to be, but still generally limited to closely related species and genera. The phages used in this work were isolated from orchards and gardens and so would not be expected to greatly disrupt the microbial ecology of an orchard to which they were exogenously applied. Nevertheless, it is important to know what other orchard bacteria Illight be affected by their presence. The specificity of phages is an advantage in this respect, though th'e Vineland collection exhibits a broader host range than many phages, infecting both E. amylovora and P. agglomerans. The orchard bacteria tested here are by no means an exhaustive sample of orchard microflora, but the resistance of the tested isolates to phage infection illdicates that the effect of these E. amylovora phages on other orchard bacteria would be millinlal to nOll-existent. The susceptibility of P. syringae pv. papulans psp mutsu to E. amylovora phages is suspicious, given the phylogenetic distance between the Enterbacteriaceae and the Pseudonl0nadaceae. However, there was a striking similarity in the identity of phages that infected this isolate and P. 102 agglomerans E325, which suggests that the original identification of the P. syringae pv. papulans psp mutsu isolate as Pseudomonas may be inaccurate. Antibiotic production was observed in all of the phase 2 ,can"ier candidates. Several of the carrier candidates produced an antibiotic that may be similar to pantocill B, based on the large, diffuse zone of inhibition produced by the metabolites of all of these strains. Unfortunately the biosynthetic pathway responsible for the production of pantocin B has not beell characterized, and so no peR screen is available to confirm the identity of this antibiotic. Inactivation tests could be used to support or refute this identification, since palltocin B is illactivated by histidine and by an extracellular protease made by P. fluorescens AS06. Mass spectrolnetry 'could also ,be used to compare these antibiotics to pantocin B, following purificatioll of the active 'soluble species by liquid chromatography. Part of the pantocin A biosynthetic gene was present in Eh252, C9-1, 39CII, and 1-28b. The first three of these strains produced zones of inhibition characteristic of panto'Cill A (Ishimaru et aI., 1988). Strain 1-28b did not produce this zone of inhibition, even though th'e paaB gene fragment was amplified. Since pantocin A is produced by a complex biosynthetic process (Jin, Wright, Beer, & Clardy, 2003), it is possible that the paaB gene is present in the 128b genome, but that the biosynthetic pathway is not complete. The antibiotic produced by 13P-4-5(2)II strongly inhibits E. amylovora strains, and is not similar to a previously described antibiotic. If this antibiotic is not inactivated by any of the amino acids that are usually found in pear or apple nectar (Lewis, Tolbert, & Kenworthy, 1964) then it would likely be active on the blossom surface, and lnay allow straill 13P-4-5(2)II to inhibit E. amylovora very effectively. 103 Because of the non-quantitative nature of the filter disc assay, conclusions can not be drawn regarding the relative susceptibilities of different E. amylovora strains to these antibiotics. Comparative susceptibility tests could be done using volulnes of growth nledium that are standardized based on the final cell density in the culture. Rather than reducing the volunles of culture filtrate from lower density cultures, the total amount of the antibiotic present in the plugs can also be increased by allowing the disc to dry slightly before applying all additional volunle of the culture filtrate. This is more likely to allow detection of the weaker antibiotics such as pantocin B. The in planta assay used to screen the phages and carrier candidates for their ability to inhibit E. amylovora was a slightly modified version of the blossom assay described by Gill (2000). Assays based on apple and pear blossoms have been used previously to assess E. amylovora pathogenicity, test the efficacy of biological control agents, and study the effects of blossom nutrition on blossom chemistry and microbial growth (Pusey, 1997; Gill, 2000; Pusey & Curry, 2004; Johnson, Stockwell, & Sawyer, 2004). Blossom assays are based on the fact that most fire blight outbreaks begin with blossom infection, and thus biopesticide efficacy depends on the microbial ecology of the blossom. Other commonly used bioassays for the patho,genicity and biological control of E. amylovora are the pear plug bioassay and illfection of seedling shoot tips with scissors dipped in bacterial suspensions. The pear plug assay is based on the growth of E. amylovora and the development of a characteristic ooze on the surface of immature pear fruit tissue. The seedling assay is meant to nlimic the infection of a succulent shoot during in'sect feeding. While these are both convenient screening methods, and E. amylovora does infect 'both fruit tissue and shoots, neither of these assays reflect the plimary nl0de of host tissue infection, 104 and certainly does not reflect the characteristics under which the phage-carrier biopesticide is expected to function. The selection of phages and the carrier was not entirely based on their efficacy in blossonl assays. Since the carrier must be infected by the chosen phages, the selection of each is dependent on the selection of the other. Very few phages consistently caused a 110ticeable reduction in symptom severity. Itl retrospect, this may be largely attributable to an unnecessarily high pathogen pressure. Initial tests had indicated that a 1 x 108 CFU/mL suspension of E. amylovora was needed to cause full disease in at least 80% ofblosson1s, but all inoculun1 tests conducted later have shown that the san1e level of disease is possible with a 1 x 106 CFU/mL suspension. The latter result is considered to be more accurate given the larger sample size used, the finer gradations of pathogen inocula tested, and the greater repeatability of the results. Therefore the phages and carrier were selected from among those candidates that gave notable protection from fire blight symptoms in the blossom assays, even if that difference was 110t statistically significant. From the phages within each RFLP group, one or two were selected that appeared to have been the most effective in the blossom assays, and that infected at least six of the canier candidates during the Phase 2 screening. Very few of the effective group 3 phages had a ·broad enough host range in P. agglomerans isolates to be useful. The Phase 2 carrier candidates w'ere then considered in terms of how many of these phages il1fected it, and it's perfon11ance in the blossom bioassay. This should have balanced phage selection so that a genetically diverse group of effective phages is tested, which can later be formulated in to diverse cocktails to reduce the risk of selecting for phage-resistant E. amylovora. 1'05 P. agglomerans isolate Eh2l-5 was selected as the bacterial canier for thos·e field trials based on its biocontrol ability in blossom assays, and its susceptibility to a wide range of bacteriophages. This strain also produces a weak antibiotic, but it is not known whether this antibiotic is active on the blossom surface. P. agglomerans l3P-4-5(2)II is a promising backup candidate, since it is infected by many of the same phages as Eh2l-5 and produces a strong, novel antibiotic. Ten phages, <f>Ea9-5, <f>Eal 0-1, <pEalO-6, <f>Ea 21-4, <f>Ea3l-3, <f>Ea35-4, <f>Ea35-5, <f>Ea45-lB, <f>Ea46-1A2, and <f>Ea5l-l, were chosen for further development and for use in field trials based on their performance in blossom assays, genetic diversity as indicated by RFLP groupings and host range in potential P. agglomerans carriers, al1d detectability by the real-time peR primers designed for determining phage numbers in the orchard. Three of these, <f>Ea9-5, <f>Ea35-5 and <f>Ea5l-l were later dropped from the development program. Seven phages and P. agglomeransEh2l-5 were selected for further study toward the development of the phage-carrier biopesticide for fire blight. 106 Chapter 3: The Complete Genome Sequence of Erwinia pha:ge <l>Ea21-4 Abstract Very little genome sequence information is available from which to develop PCR-based detection tools for Erwinia phages. In order to address this shortage of information, the conlplete genome of <l>Ea21-4was sequenced to an average PHRAP 40 quality, from a shotgun gell0nlic library, followed by primer walking. The single-copy genonle is 84.7 kb in length, with a GC content of 43.8%. The packaged genome is terminally redundant. Eighty-two ORFs were predicted using GeneMark.hmm, a trained Markov-chain model. Since all phage ORFs are expected to be translated, preliminary annotation was conducted by comparing predicted protein sequences to all possible translations of the global nucleotide database. The cI>Ea21-4 genome appears to be substantially different, both globally and locally, from previously reported sequences. About 31 % of predicted ORFs were assigned a putative function based on alllino acid sequence comparisons. Four more ORFs were identified as stluctural proteins based OllSDSPAGE of denatured phage particles and N-terminal sequencing of the isolated proteills. III each case, the complete protein sequence was highly similar to a previously unidentified protein fronl Salmonella phage Felix 01. No notable similarity exists between the cI>Ea21--4 genome and previous sequences from the cI>Eal(h) and Eral03 genonles. Only the Felix .genome shows any broader similarity to cI>Ea21-4, and even this consists of only 40% to 50% sequence idelltity across less than 20% of the two ,genomes. No significant similarity to available sequences was found for 23% of the predicted ORFs. 107 Introduction About 5300 different phages have been described (Ackermann, 2006). They are classified based on the type of nucleic acid genome (single- or double-stranded RNA or DNA), the nunlber of nucleic acid segments per genome, morphology (binary, cubic, or helical symmetry; pleiomorphy), the presence and nature of a lipid membrane, whether they are capable of lysogeny, and the mechanism by which mature progeny are released from the host cell. Approximately 96% of known phages belong to the order Caudovirales, the tailed phages (Ackermann, 2001). This does not necessarily reflect the relative frequency of Caudovirales species in nature. It is estimated that less than 0.0002% of Earth's phage metagenol11e has been described (Rohwer, 2003; Guttman, Raya, & Kutter, 2005), and most detailed work has been conducted on phages of Gram-negative bacteria, particularly phages of the Enterobacteriaceae since these bacteria frequently cause disease in humans and other animals. As a result, the list of described phages is likely biased in favour of those types that infect the most commonly studied and easily cultured host organisms. The Caudovirales order is divided into three families based on shared morphological features: Siphoviridae, Myoviridae, and Podoviridae. The Siphoviridae have long, flexible, noncontractile tails, and comprise about 61 % of observed tailed phages. The Myoviridae have contractile tails made up of a sheath and central tube, and account for about 25% of observed tailed phages. The remaining 14% of Caudovirales phages are identifiable by their extrel11ely short tails, and belong to the Podoviridae family. All of these phages have double-strand'ed DNA genomes, and are composed of about 50% protein and 50% nucleic acid (Ackermann, 2001). Phages that infect E. amylovora have been isolated from soil since the 1950s (Okabe & 108 Goto, 1963), but detailed characterization only began in the 197{)s. 'Sele'cted characteristics of these described phages are compiled in Table 3-1. Phages Sl (Erskille, 1973) and EraI03 (Vandenbergh, Wright, & Vidaver, 1985) appear to be unique, but were the sole phages discussed in their respective publications. More useful infoffilation as to the diversity of E. a·mylovora phages can be culled from larger studies in which many phage isolates were collected. Schnabel and Jones (2001) collected 50 phage isolates frOln apple and pear orchards in Michigan and California, and a raspberry farm in Michigan. Forty-two of the isolates from Michigall and California tree fruit orchards could not be distinguished froln <pEal. Four isolates fronl raspberry and Michigan apple orchards were indistinguishable from <pEa7. Only four isolates appeared to be novel phages, and two of these, <pEa100 and <pEal 04 were quite similar. Gill et al (2003) reported a 111ore diverse collection of isolates. These were classified by RFLP patterns and <pEa1based PCR. Ten phages, groups 3a and 3b, were related, but not identical, to <PEal. Ten isolates were not placed into the reported groups, either because they produced a conlpletely unique RFLP pattern or could not be digested by any of the endonucleases used. The remaining 30-plus isolates were distributed nluch more evenly alnong the different RFLP groups than the isolaftes collected by Schnabel and Jones (2001), and it would be prelnature to state that ,each RFLP group constitutes a single phage strain, since electron microscopy and fUl1her molecular analysis have revealed some differences between isolates within a group (see Tables 2-3 alld 2-4). The greater diversity of the Vineland collection may be due to use of six E. amylovora'straills during the enrichment and isolation process. Other studies have only used a single host 'strain, which would bias the enrichment process in favour of those phages that infect that straill lnostefliciently, or that produce the largest burst size from each infected cell (Jensen et aI., 1998). 109 Table 3-1. Characteristics of previously described E. amylovora phages. Except for EraI03, genome size estimates are minimums based on PFGE and RFLP data. Phage SI Isolated from soil beneath infected pear Host Range (other than E. amylovora) P. agglomerans Family Approximate Genome Size Reference a (kb) b (Erskine, 1973) Caudovirales, icosahedral head Eral03 infected apple and pear tissue not P. agglomerans Podoviridae 45 (Vandenbergh et aI., 1985; Summer et aI., 2007) GenBank: EF160123 <pEal d infected apple tissue not Pseudomonas or P. agglomerans Caudovirales, polyhedral head 46 (Ritchie & Klos, 1979; Schnabel & Jones, 2001) <pEa7 infected apple tissue not Pseudomonas or P. agglomerans Caudovirales, octahedral head 35 (Ritchie & Klos, 1979; Schnabel & Jones, 2001) <pEal00 soil beneath infected apple NT 35 (Schnabel & Jones, 2001 ) <pEa 104 soil beneath infected apple NT 35 (Schnabel & Jones, 2001) <pEa 125 soil beneath infected apple NT 35 (Schnabel & Jones, 2001) <pEa 116C infected apple tissue not Pseudomonas or P. agglomerans 75 (Schnabel & Jones, 2001 ) Group 1 (9 isolates) infected apple tissue and soil beneath infected pear P. agglomerans; not Pseudomonas or E. coli Myoviridae 75 (Gill, 2000; Gill et aI.,2003) Group 2 (4 isolates) various NT Myoviridae 75 (Gill, 2000; Gill et aI., 2003) Group 3a (9 isolates) soil beneath infected apple not Pseudomonas, E. coli, or P. agglomerans Podoviridae 36 (Gill, 2000; Gill et aI., 2003) Group 3b (1 isolate) infected crabapple tissue P. agglomerans; not Pseudomonas or E. coli Podoviridae 45 (Gill, 2000; Gill et aI., 2003) e 110 Phage Isolated from Host Range (other than E. Family Approximate Genome Size (kb) amylovora) Reference Group 4 (2 isolates) soil beneath infected apple P. agglomerans; not Pseudonlonas or E. coli Podoviridae 67 (Gill, 2000; Gill et aI., 2003) Group 5 (3 isolates) soil beneath in fected pear NT Podoviridae 45 (Gill, 2000; Gill et aI., 2003) Group 6 (6 isolates) soil beneath infected Rosaceae P. agglomerans; not Pseudomonas or E. coli Podoviridae 62 mostly soil beneath infected apple or pear NT various Various (10 isolates) (Gill, 2000; Gill et aI., 2003) (Gill, 2000) a References are to the publication(s) from which the phage's description is taken, which is not necessarily the original report of its isolation. b In several of these studies, P. agglomerans is referred to by its former name, Erwinia herbicola. C "_" indicates that no information about the trait was reported. d Also referred to as <pEa 1(h) in Ritchie and Klos (1979). The phage designated <pEa 1(nh) was a derivative of <pEa 1(h) that was never isolated directly from plant material. e "NT" indicates that the host range was not tested with any species other than E. amylovora. 111 PCR primers designed based on the sequenced region of <f>Eal failed to anlplify about 50% of the phages in Vineland collection (Chapter 2). This, conlbined with the data fronl RFLP analysis and transmission election miscroscopy (Chapter 2; Gill, 2000), indicate that there is a substantial amount of genetic diversity among E. amylovora phages that is not reflected by the currently available sequence data. In order to take full advantage of this diversity during biopesticide development, more genomic information is needed in ord,er to und-erstand it. To that end, the complete genome of <f>Ea21-4 was sequellced and annotated. <f>Ea21-4 is a tailed, contractile E. amylovora phage that was isolated in 1998 from a pear orchard in southern Ontario, Canada. It performed well in preliminary biological control assays, but was not detected by primer and probe sets based on the <f>Eal depolymerase gene (see Table 2-3). The National Center for Biotechnology Information (NCBI) genome database cun"ently contains complete genome sequences for almost 400 bacteriophages, and yet, at the time this work was undertaken, the only sequence data available was a 3.3 kb region of <f>Eal (Kinl, 'Sahn, & Geider, 2004). Early in 2007, the complete genome sequence of E. amylovora phage Eral03 was deposited (Summer et aI., 2007). 112 Methods Strains and Growth Conditions E. amylovora phage <l>Ea21-4 was grown on E. amylovora Ea6-4 in liquid culture, as described in Chapter 2. Transmission Electron Microscopy A concentrated suspension of E. amylovora phage <l>Ea21-4 was prepared by centrifugillg 10 mL of filtered lysate at 16 000 xg for 75 min, decanting the supernatant and resuspending the phage pellet in 5 mM Tris-HCI, pH 8.0, containillg 0.1 mM EDTA. TEM work was conducted as described in Chapter 2. DNA isolation and RFLP Three methods of DNA isolation were compared: the CTAB and organic extraction methods previously described (see Chapter 2), and the Qiagen lanlbda Inini kit (Qiagen, Mississauga, ON). DNA extraction using the Qiagell kit was conducted accordillg to the manufacturer's protocol, except that the phage precipitation step was doubled to 2 h. An RFLP was conducted using each of MvnI, BgIII, BamHI, and EcoRI, as described in Chapter 2. Genome Sequencing The DNA ,from multiple organic extractions was cOlnbined and the 'conc'entration was determined based on the absorbance of ultraviolet light (A == 260 11111, 50 Jlg/mL == aD of 1). A composite sample was prepared by combining extraction products with all absorbance ratio (260 113 nm vs. 280 nm) between 1.7 and 1.9 and adjusting the concentration to approximately 100 J.1g/mL. Approximately 100 ~g of DNA was sent to Agencourt Bioscience (Beverly, MA) for shotgun cloning and sequencing. The shotgun genome library was cOllstructed by mechallically shearing the <t>Ea21-4 DNA. The resulting 3 to 4 kb fragments were cloned into the proprietary pAGEN vector using the BstXI adaptor. Escherichia coli colonies carlying the cloned vectors were selected and inserts sequenced for lOX coverage. Sequencing reactions were conducted using an ABI PRISM 3730xl DNA Analyzer and BigDye Terminator v3.1 reagents. Sequence Analysis Putative open reading frames (ORFs) were identified using GeneMark.hmm, version 2, and Fickett's TESTCODE, and the ORFs predicted by both models were accepted. GeneMark is a Markov chain model trained on completed microbial genomes (Beselller, LOlllsadze, & Borodovsky, 2001). Fickett's TESTCODE method is based on asymllletlical base usage in coding regions and is built into the Clone Manager Professional Suite program (v. 7.11, SciEd Central). Similarity to described genes in the global database (available through NCBI at www.ncbi.nlm.nih.gov) was tested using the tblastn algorithm, which compares protein translations of the predicted ORFs to all possible translations of the nucleotide database. Two programs were used to search for phage-encoded tR ~As: tRNAscanner (Lowe & Eddy, 1997), and FAStRNA (El-Mabrouk, & Lisacek, 1996). Loci predicted by both programs were accepted. Genome-wide comparisons were made using the GenOllle Shov'el dot-plot pro:granl, available from the Japan Science and Technology Agency (www-btls.jst.go.jp). Th·e genoll1es 114 were aligned using the tblastx algorithm, and are displayed as DNA. Two-way protein alignments were constructed using ALIGN Query, available froln the Genestream Search network server IGH in Montpellier, France (xyliall.igh.cnrs.fr/bin/alignguess.cgi). Protein domain searches were conducted using the Pfam 21.0, available from the Sanger-Wellcome Trust (Finn et aI., 2006) The 3.3 kb fragment of <pEal and the complete .genolnes of <pEral 03 and Felix were downloaded from the NCBI GenBank database (AJ2786 14, NC_009014, and NC_005282, respectively). Identification ojMajor Structural Proteins Intact phage particles were purified using cesium-chloride gradient centrifugation, as previously described (Sambrook & Fritsch, 1989). Syringe-filtered <pEa21-·4 lysate was concentrated by centrifugation at 16 000 xg for 1 h alld resuspended in 1/1 Oth volume of SM buffer. Bacterial nucleic acids were digested by adding 1 JlL each of 10 mg/mL DNase I and RNase A, and incubating the reactioll at 37°C for 30 min. Approxinlately4 nlL of this phage suspension (2.1 x 1OJO PFU/mL) was dissolved in sterile distilled water with CsCI to a final density of 1.48 g/mL in a final volume of approximately 23 mL. The suspension was divided evenly between two Beckman OptiSeal 11.2 InL polyallomer ultracelltrifuge tubes (Becktnan Coulter Canada, Mississauga, ON). The suspension was centrifu.g·ed at 30,000 rplll for 24 h using the NVT 65 rotor on a Beckman Optima CL-I00K ultracentrifuge. Following centrifugation, the phage particles were visible as an opaque band. The celltrifuge tube was pierced just below the band and the phages were drawn offusing a 26 lh-gauge hypodemlic needle with a IlnL 'syringe. 115 Cesiunl chloride was renl0ved by dialysis uSillg four chan.ges of buffer (10 mM Tris-HCI, 10 mM MgCI 2 , 1 mM EDTA, pH 8.0), lasting 15 min each. A second batch ofpuritied phage was prepared in the same way, with the following changes: the syringe-filtered phage lysate contained 3 x lOll PFU/mL, and the gradient-purified phage were drawn from the CsCI using an I8-gauge needle. The phage suspension was concentrated by freeze-drying. A 2 InL microcentrifuge tube containing 0.5 mL phage suspension was capped with a 1" square of kimwipe and imnlersed in liquid nitrogen for approximately 30 s, until frozen. Samples were vacuum-dried oveluight using a ThermoSavant MicroModulyo (E-C Apparatus, Holbrook, NY). The sample was resuspended in 100 ~L of2X gel loading buffer (62.5 mM Tris-Hel, pH 6.8, 2% (w/v) SDS, 5% (v/v) 2-mercaptoethanol, 10% (v/v) glycerol, 0.1 nlg/L brolnphenol blue). Proteins were denatured by immersing the sample in a boiling water bath for 5 min. Proteins were separated by SDS-PAGE. Twenty-five microliters of the concelltrated sample, and 10 L~ of BenchMark pre-stained protein ladder (Invitrogen) was loaded on each of two 12% SDS-PAGE gels, prepared according to published protocols (Sambrook & Fritsch, 1989) alld run at 150V for 70 min. One gel was stained overnight in Coomassie blue (0.1 % (w/v) Coonlassie blue R-250, 40% (v/v) methanol, 10% (v/v) glacial acetic acid), and destained in 40% (v/v) methanol with 10% (v/v) glacial acetic acid, and observed at various times during d:estaillin.g. The gel was photographed with a Kodak Easy-Share CX7430 digital calnera, through a'SYBR Safe photographic filter (Wratten filter No.9, Molecular Probes, Eugene, OR) filter. The proteins were transferred from the second gel to a polyvinyl diflouride (PVDF) membrane using the mini-Protean II transfer cell (BioRad Laboratories, Hercules, CA). The 116 PVDF membrane was prepared by brief inlmersion in lnethanol, and the gel, 111elnbrane, alld filter paper were equilibrated in Towbin transfer buffer (3.03 giL Tris base, 14.41 giL glycille, 20% (v/v) methanol) for 15 min. The electroblot sandwich was assenlbled as described by Sambrook and Fritsch (1989) and protein transfer was conducted overnight at 30 V. The separated proteins were visualized by immersing the membrane in Coomassie blue for 10 lnin, and destaining in several rinses of 50% (v/v) methanol for a total of approxinlately 15 min. Stained bands were excised from the membrane using a razor blade, placed in lnicrocelltrifuge tubes, and stored at 4°C. N-terminal Edman lnicrosequencing was conducted by the Advanced Protein Technology Centre at the Hospital for Sick Children. N-terminal amino acid sequences were compared to the hypothetical translated sequences of all predicted ORFs in the <pEa21-4 genome. In order to separate the two proteins with an apparellt molecular size of 42 kDa, a 1 111L volume of the CsCI-purified phage suspension was separated by isoelectric focusing. The Rotofor isoelectric focusing system (BioRad) was assembled, loaded, and run according to the manufacturer's instructions, using the standard focusing chamber. The sample was prepared by combining 58 mL distilled water, 1 mL of purified phage suspension that had been denatured at 100°C for 5 min, and 3 mL of Bio-Lyte ampholytes (pH range 3/10). The fractionation run was conducted at 15 W for approximately 2 h, and the resulting fractions were collected in 20 12 x 75 mm culture tubes. The fractions were transferred into 2 mL microcentrifuge tubes and free2edried as previously described. The lyophilized fractions were resuspended in IX gel-loading buffer and run on a 12% SDS-PAGE ·gel as previously described. 117 Results Morphological Features of<l>Ea21-4 <l>Ea21-4 is a tailed, contractile phage belollging to the Myoviridae fanlily, and as such has a double-stranded DNA genome. Figure 3-1 shows the contracted and uncontracted states of the <l>Ea21-4 virion. The head is icosahedral, and approximately 60 nm across, with a 90 nm tail. In the uncontracted state, a slight narrowing of the proximal end of the tail, called the neck, is visible above the tail sheath where the central tube connects to the head. In the cOlltracted state, the central tube can be seen extending past the base of the contracted sheath, and the neck and base plates are visible above and below the sheath, respectively. DNA Extraction The organic solvent extraction method was the best of the three DNA purification methods tested. Table 3-2 shows the relative purity of DNA obtain,ed from each nlethod. DNA extracted by the CTAB method had the lowest and most variable purity. The organic and kit extraction methods yielded DNA of similar purity, but results were more consistent with the organic extraction method. In additioll, the kit nlethod rarely yielded detectable anl0ullts of DNA, whereas DNA was consistently recovered using the organic extraction method. During later organic extractions it was noted that the additioll of the usual 2.0 to 2.5 volumes of 100% ethanol during DNA precipitation resulted in the precipitation of a dense, viscous material, in which flecks of precipitated DNA were ell1bedded. This substance was not precipitated from the aqueous phase of back-extracted DNA, in which fresh Tlis-HCI was added to the used phenol phase, extracted as usual, added to the used phenol-chlorofonn phase, 118 Figure 3-1. TEM of <pEa21-4. The phage is shown here in the uncontracted (left) and contracted states (right). [micron marker == 50 nm] 119 Table 3-2. Effect of extraction method on the A260lA280 ratio of genomic <l>Ea21-4 DNA. A bsorbance(260nm)/A bsorbance(280nm) Method mean Organic Extraction (n Qiagen kit (n CTAB (n a = = 6) 3) = 3) a SD range 1.94 A 0.02 1.92-1.95 1.91 A 0.17 1.66 - 2.13 1.52 B 0.32 Means with the same letter are not significantly different (Tukey test for mean separation, ex 1.00 - 1.82 = 0.05) 120 extracted, etc. to recover DNA lost to the organic phases in the original extractiol1. The use of fresh extraction reagents did 110t eliminate the problem. The ultraviolet absorbanee spectrun1 (200 to 320 nm) of this viscous substance was characteristic of carbohydrate, with a p,eak at 225 nrn, and very little absorbance above 240 nm. fu an attempt to precipitate the carbohydrate fraction separately from the dissolved DNA, ethanol was added to the final aqueous phase 100 J.1L at a time, mixing by inversion after each addition. After an equal volunle of ethanol had been added to the final aqueous phase, the DNA precipitated, but the carbohydrate did not. The precipitated DNA could be sedimented, dried, and resuspend,ed as usual from this point, and the purity of the resulting sample was conlparable to that indicated in Table 3-2. Genome Sequencing Shotgun sequencing yielded a 30 kb contig and nine nOll-overlapping illserts, for a total of approximately 39 kb of sequence data. Since the total length of these sequences was only half the 75 kb genome size estimated from RFLP analysis (Gill, 2000), the sequencin,g c011lpany was requested to begin primer walking from the ends of each contig. After nine rounds of primer walking, a single contig of 80 kb was obtained. After two more rounds of tel11linal primer walks, the same 182 bp sequence was obtained fronl both reactions and prinler walkin,g was 11alted. The final sequence was 84.7 kb for the single-copy genome, which Ineans that slightly nlore than 50% of the genome was not represented in the random clolle library. General Features oJthe <l>Ea2J-4 Genome The genome sequence obtained fronl Agencou111llatches the lalown sequence 121 characteristics of the DNA isolated from Q>Ea21-4 lysate. Figure 3-2 shows the RFLP pattell1s that were obtained by digesting genomic <j>Ea21-4 DNA with each of ThaI, BamHI, BgIII, and EcoRI. The fragment sizes predicted by virtual restriction digests of the sequence were consistent with these RFLP patterns, cOllfirming that the sequenced genome matched the known ll101ecular characteristics of <j>Ea21-4. Also, the sequences of the <j>-dpo I primers and probe (see Chapter 6), which did not amplify DNA from the <j>Ea21-4 lysate, were not found in the genollle sequence. The <j>Ea21-4 genome is quite different from <j>Eal and <j>Eral03. The region sun"ounding the <pEa21-4 endolysin gene was compared to the 3.3 kb sequence from <pEal. The only re-gion that could be meaningfully aligned was that of the endolysin gene itself, and then with only 32% sequence identity at the amino acid level (Figure 3-3). In addition, none of the plimers or probes designed from the dpo gene of <j>Eal by Dr. Kim (see Chapter 2) match regions of the <pEa21-4 genome with more than 50% identity, and none of the partial matches could reasonably permit PCR amplification. The only other complete genome sequence available for a E. am.,vlovora phage is <pEral03. A dot-plot comparing the <pEa21-4 and <pEral03 genomes was constlucted (Figure 3-4). Even using low identity thresholds, there was very little similarity betweell the two genomes. The only phage genome with notable similarity to <pEa21-4 ·genollle is that of Salmonella phage Felix 01. Figure 3-5 is a dot-plot comparing the genonles of <pEa2l-4 and Felix. Using a 100 bp sliding window, no regions of 65% or greater identity were observed. Regions totalillg perhaps 15-20% of the genomes were 40-65 % identical. These regiolls fell mostly between bases 12,000 and 40,000, or between 63,000 and 70,000. 122 L 1 2 3 4 5 10 kb 6 kb 4 kb 3 kb 2.5 kb 2 kb 1.5 kb 1 kb 0.75 kb 0.5 kb 0.25 kb Figure 3-2. RFLP patterns resulting from the digestion of genomic cPEa21-4 DNA each of MvnI (lane 1), BglII (lane 2), Bamffi (lane 3), and EeoR! (lane 4). Undigested genomic DNA was loaded in lane 5. 123 10 20 30 40 50 60 <pEal lys MSVKKALAGGACSLALVTASFFGIVTDKVRISQEGLEHLIDCEGCKRQAYKDGAGVPTAG <pEa21-4 endo MEV----------------------------SQKGQQALEVMEGFSAKAYLDVAGVPTIG ... . . .. . .. . .. . . .. . . . . . . . .. . 10 70 80 20 90 30 100 110 <pEal lys VGSTI----GIVMGRLYTDGEVAKM-LAKDVMI----AEQCLNRNVKVDLNQGEWDAYVS <pEa21-4 endo FGDTSVRARKVKMGDT-TTLEAAKAELALDLHDFKSGVEKYLAKAVK-GTTQNQFDALVI . .. . .. .. . .. 40 120 <pEal lys <pEa21-4 endo 130 lys <pEa21-4 endo 60 70 140 150 . . .. 80 160 90 170 FVFNVGCSAFVSSTTYRILNGVKPGTRIQACEAMGMWNKITVNGVKVFSQGVYNRRIKDM .. . . .. .. . . . .. . . .. . . . . .. . .. .. .. . . . FAYNVGLTNFASSSVLR--NHLA-GDFEAAAKSFALWNKITVKGKKVVSKGLVNRRAKEI 100 <pEal 50 .. 110 120 130 140 ALCVKY---M EIYLHSNYGV 150 Figure 3-3. Similarity between <pEa21-4 endolysill and the <pEal lysozYlne. (32% identity) 124 1 10Kb 20Kb 30Kb 50Kb 60Kb 70Kb fo 'C""'I1------t-------+------t------+------+-------4----+------t------i § o Nt-------t------+------t------+------+-------4----+------t------i .D :x: c· ('i""Jt------t------i------+------+------+-------4-----t-----+------t Figure 3-4. Genome-wide similarity between the cPEa21-4 genome (horizontal axis) and the cPEra103 (vertical axis) genome. The alignment was constructed using the tblastx engine with a sliding window of 50 bp. Regions with at least 50% identity are indicated in red, and regions with at least 20% identity are indicated in orange. 125 "I'"'ir-l b~K0-8;,_ - b,; . KO,; . 6T_ - ~b; , .;KO . : 4,_ - bK;, ";,O~2 . .0 ::::.::: o Nt----------iI----------4--------+-------4----I .0 :::L o "d"t---------II-----------4--------+-"""IIf------4----I .0 :::L o 1"Dt-I4+~ .# .0 :::L o OO.------------f--------I--------+-------+-----i Figure 3-5. Genome-wide similarity between the <j)Ea21-4 genome (horizontal axis) and the Salmonella phage Felix genome (vertical axis). The alignment was constructed using the tblastx engine with a sliding window of 100 bp. Regions with at least 65% identity are indicated in red, and regions with at least 40% identity are indicated in orange. 126 Table 3-3 summarizes the general features of the <l>Ea21-4 genon1e, con1pared to the two other E. amylovora phage genomes from which sequence il1fonnation is available, al1d to the Felix genome. The single-copy genome of <l>Ea21-4 is 84,676 bp with a GC content of 43.8%. The <l>Ea21-4 genome is similar in size to the Felix genonle, and about twice the size of the two E. amylovora Podoviridae genomes. Both the <l>Ea1 and Era103 genolnes are only slightly less GC-rich than E. amylovora, whereas the <l>Ea21-4 and Felix genolnes are substantially less GC-rich than their respective host genera. Gene Annotation A total of 82 ORFs were predicted by both the GeneMark.hlnm and Fickett's testcode algorithms (Table 3-4). Since all ORFs in a phage genome are predicted to be trallslated, functional predictions were based on similarities between the translated sequence of each predicted gene, and all possible translations of the global nucleotide database. There were 110 significant similarities found between 23% of the translated ORFs and any previously described gene sequences. These ORFs were classified as "predicted proteins". For 16% of the predicted ORFs, there was not enough similarity to describ'ed sequences to assigJl a putative function, but there were regions within the ORF that were sufficiently sin1ilar to described functiol1al don1ains or phage-related proteins to suppo11 their identification as a gene. A fU11her 29% of the predicted ORFs showed no significant similarity to any sequences except a predicted ORF in the Salmonella phage Felix genome. These ORFs, for which at least one significant Inatch was obtained, were classified as unknown. The ren1aining 31 % of the predicted ORFs were assign'ed a putative identification based on silnilaritie'S to known 'proteins. 127 Table 3-3. General features of the cPEa2I-4 genome compared to two other E. amylovora phages, <l>Eal(h) and EraI03, and Salmonella phage, Felix. Feature <pEa21-4 <pEa 1(h) Size 84.7 kb approx. 37 kb G+C content 43.8% G+C content of host Genus Total number of predicted ORFs Average ORF length Percentage of genome constituting coding regions Morphology 50-540/0 c Era 103 Felix 45.4 kb 86.2 kb 50.40/0 49.8% 39.0% 50-54% 50-54% a b 82 53 800 bp 785 bp 78% 91.5% Myoviridae Podoviridae e Podoviridae based on published 3.3 kb sequence only (Kim, Salm, & Geider, 2004) estimate based on RFLP (Gill, 2000) C (Hauben & Swings, 2001) d (Le Minor, 1984) e (Ritchie & Klos, 1979) f probable morphology, based on Vandenburgh er al (1985), and presence of depolymerase g (Lindberg & Holme, 1969) a b 50-53% d 910/0 f Myoviridae g 128 Table 3-4. Predicted genes of <l>Ea21-4. ORF Start End Protein Size (a a) 79 4 201 80 269 81 82 2 Putative Function or Conserved Domain Selected best tblastn matches a 66 Predicted protein NS 505 79 Pred icted protein NS 577 825 83 Predicted protein NS 2461 2273 * 63 Predicted protein NS 2694 2464 * 77 Pred icted protein NS 3657 3184 * 157 Endolysin Felix lysis prot. (AF320576), phage Gifsy-2 lysozyme (NP_460003) Pfam: phage lysozymel 0glycosyl hydrolase domain (residues 28-145) 3 4321 3707 * 204 Structural (21 kDa); %Identity (%Similarity) 1 # residues b 44 (62) 1 156 40(54)/130 Ig-like fold domains of bacterial proteins Pfam: Big-2 family Ig-like domain (residues 123-301) 4 4942 4325 5 5738 6 9703 * 205 Unknown Hyp. Felix prot. (AAQ14772) 28 (45) 1 205 6103 121 Unknown Hyp. Felix prot. (AAQ14773) 58 (77) 1 118 10437 244 A Red-like protein Hyp. Felix prot (AAQ14777) 43 (67) 1 240 Pfam-B: Red-like domain families 66931 (residues: 3149) and 94232 (residues: 194-242) 7 10963 12567 534 Packaging terminase, large subunit Pfam: Terminase-6 domain (residues: 90-505) 8 9 12573 14028 14021 14522 482 164 Unknown Unknown Hyp. prots: Felix (AAQ14779), Mesorhizobium loti (BAB53654), phage psiM2 terminase large subunit (NP_ 046964) Hyp. prots: F-elix (AAQ 14780), Corynebacterium efficiens (BAC17652) Hyp. Felix prot (A.AQI4781) 69 (82) 1 532 40 (57) 1 445 33 (47) 1420 66 (82) 1 482 25 (45) 1 263 34 (58) 1 ] 65 129 ORF Start End 10 14522 14854 11 14862 16232 Putative Function or Conserved Domain Selected best tblastn matches a % Identity/ % Similarity (# residues) 110 Unknown Hyp. Felix prot. (AAQ14782) 43(71)/107 456 Prohead protease Hyp. Felix prot. (AAQ 14669), phage BcepNazgul ClpP protease (NP_918994) 47 (64) / 429 42 (58) / 124 Protein Size (aa) Pfam: peptidase S49 family (residues: 133-285) Pfam-B: family 3464 of capsid proteases (residues31130) 12 16244 16615 123 Structural (11 kDa) Hyp. Felix prot. (AAQ14784) 13 18232 18717 161 Unknown Hyp. prots: Felix (AAQI4787), Vibrio cholerase (ZP_00755065) 33 (56) / 298 52(72)/161 30 (48) / 143 14 18714 19100 128 Unknown Hyp. Felix (AAC24149) 54 (70) / 131 15 19100 19708 202 Unknown Hyp. F'elix (AAQI4789) 47 (64) / 192 16 19713 21071 452 Structural (47 kDa) Hyp. prots: Felix (AAQ 14790), H. In.fluenzae (AAX8834-6) 59 (75) /450 30 (48) / 280 Hyp. prots: Felix (AAQ14791), H. lnfluenzae (AAT40857) 83 (89) / 148 28 (52) / 125 Pfam-B: phage-related falnily 12740 (residues 190-397) 17 21087 21542 151 Structural (14 kDa) Pfam-B: phage-related fan1ily 28462 (residues: 7-140) 18 21600 22010 136 Unknown Unk. Felix (AAQ14792) 36(61)/132 19 22247 24469 740 Possible tail tape measure Unk. Felix (AAQ14794) phage tail tape measure prots: BcepNazgul (NP_918983), B3 (AAQ13967) 39 (57) / 751 Pfam: DNA-binding family 130203 (residues 1-306), phage tail tape measure falnily 2595 (residues 362-404), membrane protein family 117417 (residues 421-551) 23 (40) / 412 28 (45) / 283 130 ORF Start End 20 24473 25300 Protein Size (aa) 275 Putative Function or Conserved Domain Selected best tblastn matches a ~;O Identityl % Similarity (# residues) Unknown Hyp. prots: Felix (AAQI4795) Erwinia carotovora (CAG75399) 39 (55) 1 246 21 25300 25623 107 Unknown Hyp. Felix (AAQ 14796) 22 25633 26718 361 Unknown Hyp. prots: Felix (AAQ 14797), Haemophilus influenzae (AAX88354) 23 26718 27362 214 Baseplate assembly Pfam: family 16888, phage baseplate-related prots 24 27362 27778 138 Unknown Pfam: family 18442 of phagerelated proteins 25 27778 29262 494 Baseplate component Pfam-B: fanlily 5270, phage P2 baseplate J-like prot. 35 (50) 1 251 55 (75) 1 106 50 (69) 1 355 22 (45) 1 185 Hyp. Felix prot. (AAQI4798), P2 baseplate assembly (ZP_00708498) 48 (68) 1 211 Hyp. prots: Felix (AAQ 14799), Burkholderia sp. (ZP_00426838; ZP_00986347), S. enterica phagerelated (NP_455547) 54(72)/140 Hyp. prots: Felix (AAQI4800) Haemophilus influenzae (AAX88357) 32 (47) 1 192 37 (55) 1 85 32 (53) 1 83 51 (67) 1486 26 (46) 1 436 26 29272 30120 282 Unknown Hyp. Felix (AAQI4801) 39 (62) 1 278 27 30123 30476 117 Unknown Hyp. Felix prot (AAQ14802) 52 (65) 1 110 28 30480 31553 357 Tail fibre put. tail fibre: Felix (AAQ 14803) 44 (58) 1 395 Pfam: contains five repeat units COlTIlTIOn to phage tail fibres (residues: i) 227-240, ii) 241-254), iii) 255-268, iv) 293-306, v) 307-320 131 Putative Function or Conserved Domain Selected best tblastn matches a % Identity/ 0/0 Similarity (# residues) 502 Possibly structural (54 kDa) Chlorobium phaeobacteroides putative Ig (ZP_00533375) 28(44)/173 33893 244 Pre.d icted protein NS 37366 1157 Structural, possible tail fibre Hyp. Zymomonas mobilis (AA V89024) Unk. <pK02 (AAR83037) phage OP 1 put. tail fibre component (YP_ 453579) ORF Start End 29 31641 33149 30 33159 31 33893 Protein Size (aa) Pfam-B: undescribed 114474 family (residues 231-372) 32 37686 38021 33 38997 38061 34 39554 38994 * * III Unknown Unk. Felix prot. (AAQI4806) 311 Thymidylate synthase phage dTMP synthase Felix (AAQI4807) T5 (AAU05229) 186 Dihydrofolate reductase phage DHFRs: Felix (AAQ 14808), Aeh 1 (AAQ 17977) 35 40102 39554 * 182 Predicted protein NS 36 41198 40095 * 367 DNA ligase Unk. Felix prot. (AAQI4813), phage Xp15 DNA ligase (YP_239300) Pfam: ATP-dependent DNA ligase domain (residues 92282) 37 42715 41447 * 422 Nucleic acid-independent RNA polymerase Pfam: 1743 family (residues 3-127) 38 43073 42723 * 116 Unknown Polynucleoti de adenylyltransferase: Nitrosococcus oceani (ABA56915), Poly-A polymerase: Yersinia sp. (ZP_00829667) Unk. Felix prot. (AAQ14810) 21 (36) / 733 30 (48) /181 29 (48) / 164 25 (SO) / 108 43 (59) / 314 46 (59) / 265 36 (55) / 80 28(51)/152 63 (74) / 367 24(41)/374 36(48)/419 35 (51) /415 36 (52) / 119 132 ORF Start End 39 44058 43180 Protein Size (aa) * 292 Putative Function or Conserved Domain Selected best tblastn matches a Unknown Pfam: Band_7 integral membrane proteins, inc HflK & HflC-like (residues 23-239) put. serine protease: phage P27 (CAC83523), Shigella flexneri (AEOI4073) 40 44985 44668 * 105 Predicted protein NS 41 48525 45835 * 896 DNA polymerase DNA polY111erase: Felix (AAQ 14817), PaP3 (AAL85539) 42 49040 48564 * 158 H-N-H endonuclease PfamB-25515 135 Unknown Hyp. Felix (AAQ14818) 44 49729 50292 187 Predicted protein NS 45 50285 51136 283 Pred icted protein NS 46 51196 52182 328 D eoxyribonuc leoside monophosphate kinase dNMP kinase: Felix (AAQ 14820), KVP40 (NP_899571) DN A primase-helicase primase/helicase: Felix {AQ 14821) Q>gh-l (AA073154) 637 62(74)/894 24 (43) 1 413 49 (70) / 131 49685 54088 37 (58) 1 265 43 (60) / 139 41 (57) / 140 49278 52175 37 (58) 1 256 phage endonucluease: RB16 (AAY44388), Xpl0 (AAP58717) 43 47 % Identityl % Similarity (# residues) 31 (46) /235 30 (46) / 220 70 (83) / 609 27 (45) / 590 48 54191 55054 287 Unknown Unk. Felix prot. (AAQ14822) 42 (58) / 267 49 55078 56196 372 DNA-directed DNA polymerase Felix (AAQI4823) Thermus sp. polA (CAA46900) 53 (69) / 347 28 (42) / ] 94 Pfam: 5' - 3' exodeoxyribonuclease (residues 268-360) 50 51 56]89 57000 56704 57746 171 248 Unknown Unknown Pfam: glutaredoxin family (residues 162-205) Hyp. phage prots: Felix (AAQ 14824), Cyanophage P-SSM2 (YP_214449) Unk. Felix prot. (AAQ14826) 60 (77) / 156 32 (52) / 11 ] 37 (57) / 245 133 ORF Start End 52 58405 60639 53 60694 61791 Protein Size (aa) 744 365 % Identityl % Similarity (# residues) Putative Function or Conserved Donlain Selected best tblastn matches a ribonucleotide triphosphate reductase (ex) RN TPRase (ex): Felix (AAQ14827), Q>KZ (NP_B03872) 57 (75) 1 746 39 (57) 1 738 ribonucleotide triphosphate reductase (p) RN TPRase (p): Felix (AAQ 14828), Q>KZ (NP_803871) 50 (68) 1 348 31 (54) 1369 54 62718 63827 369 Pred icted protein NS 55 64309 64821 170 Unknown Unk. Felix prot (AAQ14737) 44 (62) 1 156 Unk. Felix prot. (AAQ14838) Cytophaga hutchinsonii PRPP synthetase (ZP_00310994) Unk. phage K, G 1prot 40 (59) 1 249 Hyp. Felix prot. AAQ14840) C. violaceufn N a PRTase (NP_899704) phage PI 00 (YP_ 406440) 46(61)/591 Hyp. Reinekea sp. Prot. (ZP_01114443), Pseudoalteromonas tun icata putative solute/DNA competence effector (ZP_01134797) 27 (48) 1 95 Pfam-B: undescribed family 115568 (residues 46-165) 56 65094 65939 281 Phosphoribosyl pyrophosphate synthetase PfatTI: Phosphoribosylpyrophosphate kinase domain (residues 142-250) 57 65953 67782 609 NMN phosphoribosyl transferase Pfam: nicotinamide phosphoribosyl-transferase domain (residues 185-574) 58 67942 68355 137 Unknown 59 68358 68684 108 Pred icted protein NS 60 68934 71384 816 rIIA rIIA frotTI phages: Felix (AAQ14746), RB43 (YP_238977) 32 (54) 1 198 25 (43) 1 201 38 (51) 1 593 35 (53) 1 634 29 (52) 1 101 33 (49) 1 787 38 (53) 1 296 134 ORF Start End 61 71436 72563 62 72889 73452 Protein Size (aa) 375 187 Putative Function or Conserved Domain Selected best tblastn matches a rIIB rIIB from phages: Felix (AAQ1474), RB69 (NP_861963), 1'4 (NP_049889) Unknown Pfam-B: N rdA -like protein family 63 73693 74205 170 Unknown Pfam-B: undescribed family 14879 from certain Myoviridae and E proteobacteria (residues 20-164) Pfam-A: low significance Inatch to SL l' transglycosylase family 64 74207 74857 216 Unknown Hyp. Streptomyces (NP_828254) and M esorh izobium (EAN04910) prots. Burkholderia phage 1'4 NrdA-like prot. (NP_919003) ~ Identity/ % Sin1ilarity (# residues) 57 (72) / 376 43 (60) / 230 41 (61)/240 35 (49) / 103 40 (51) / 82 34 (43) / 79 Unknown Felix prot. (AAQ14595) Conserved hypo prot. in Aeromonas phages Aehl (AAQ17934) 31 (YP_238834), Enterobacteria phages 44RR2 (AAQ81426), RB49 (AAL15146) 45 (59) / 169 Unknown Felix prot. (AAQ14597) 34 (55) / 149 Enterobacteria phage RB69 (NP_861919) 25 (46) / 149 31 (51) / 170 32 (47) / 184 32(47)/184 39 (53) / 130 65 74940 75407 155 Unknown Unknown Felix prot. (AAQ14749) 63 (75) / 120 66 75937 76302 121 Unknown Hypothetical Felix prot. (AAQ14765) 27 (48) / 120 67 76304 76870 188 Pred icted protein NS 68 76912 77268 118 Unknown Hypothetical E.coli (ZP_ 00730579) 34(53)/82 Hypothetical Felix (AQ14750) 30 (50) / 86 Unknown Felix prot. (AAQ14754) 44 (60) / 153 Pfam: A I pp ADP-ribose binding dOlnain (residue 30156) Acinetobacter phage 133 (AA1'J'8492) 38 (49) / 154 Pred icted protein NS 69 70 77415 78657 78197 79004 260 115 Unknown 135 ORF Start End 71 79043 79708 72 79798 73 % Identityl % Similarity (# residues) Putative Function or Conserved Domain Selected best tblastn matches a 221 Predicted protein NS 80295 165 Unknown Unknown Felix prot. (AAQI4755) 80734 81150 138 Predicted protein NS 74 81399 81734 111 Pred icted protein NS 75 82195 82782 195 Predicted protein NS 76 82861 83184 108 Pred icted protein NS 77 83576 83851 92 Unknown Unknown Felix prot. (AAQI4618) 47 (57) 1 88 78 84208 23 164 Unknown Unknown Felix prot. (AAQI4759) 32 (49) 1 126 * ORFs Protein Size (aa) 46 (65)/166 marked with an asterisk are transcribed from the antiparallel strand The listed hits represent the best local alignment, and the most informative of the next best local alignments. These are not necessarily the most closely related sequenes. In most cases these alignments have an expect value greater than e-4, though in cases where the only match was a hypothetical Felix protein, the expect value was between 0.01 and e-4• b "NS" indicates that no significant local alignment was found between any portion of the indicated ORF and any portion of sequences a 136 A single cluster of 19 tRNA genes was predicted in the <J>Ea21-4 genome. Table 3-5 summarizes these predictions. The predicted secondary structure of each putative tRNA gene is consistent with a functional tRNA. This cluster lies between orf-5 and orf-6. Structural Proteins Phage genomes include a limited number of genes encoding the structural proteins that make up the free virion. These genes were identified by denaturing purified <J>Ea21-4 virions, separating the component proteins, and matching the N-terminal sequence of each to the predicted protein sequences in the <J>Ea21-4 genome. Figure 3-6 shows the outcome of electrophoretic separation of the phage structural proteins. Ten bands were visible to the naked eye after about 1 h of destaining, though 110t all of these are visible in the gel photo. All of these proteins are listed in Table 3-6. There was an insufficient quantity of the 23 kDa, 54 kDa, and 120 kDa proteins for Nterminal sequencing, and the 36 kDa protein appeared to be N-tenninally blocked, since no sequence was obtained from more than 32 ng of protein. Partial sequence was obtained fronl the 38 kDa protein, but the encoding orf could 110t be identified. The four proteins for which a cleal1 sequence was produced could all be matched to predicted ORFs in one region of the ·genoUle. None of these ORFs had been identified based on sequence sinlilarities, but all were Sigtlificalltly similar to predicted proteins in the Felix genome. The 42 kDa band contained two proteills of similar size, so a se'quencecould not be determined for either. An attempt was made to resolve these ptoteins by isoelectric fo'cusing, but no protein bands were visible following :gel electrophol:esis of the concel1trated {i"actions. 137 Table3-5. Locations and features oftRNAs encoded by the <f>Ea21-4 genon1e. tRNA Type Anti-codon tRNA start tRNAend Pro TGC 6394 6471 Met CAT 6847 6921 Tyr GTA 7012 7094 Met CAT 7101 7174 Asp GTC 7254 7330 Lys TTT 7732 7807 Lys CTT 7815 7891 Met CAT 7903 7978 Ile GAT 7990 8064 GIn TTG 8071 8146 Arg TCT 8153 8229 Leu CAA 8647 8723 Trp CCA 8935 9010 Val TAC 9102 9177 Leu AAG 9184 9261 Pseudo-tRNA ACG 9268 9344 GIn CTG 9347 9423 9430 9506 95]0 9586 Undetennined His GTG 138 190 kDa 120 kDa ~- 85 kDa 60 kDa ~- , - 50 kDa ~- ~-, • -~ ., •.• 0." '. ~ ~ ~-' - --- 40 kDa 25 kDa 20 kDa 15 kDa 10 kDa Figure 3-6. Electrophoretic separation of <l>Ea21-4 structural proteins. Apparent sizes of the prestained ladder are indicated on the right. Arrows on the left point to the locations of each of the visible phage structural proteins. 139 Table 3-6. N-terminal sequence and identification of <l>Ea21-4 structural proteins. Approximate Size (kDat N-tenninal Sequence Encoding ORF Predicted Protein Characteristics b Length Size pI analogous Felix gene C (Accession No.) 11 AYAGYI orf-12 123 aa 12.8 kDa 5.] 8 orf-]35 (AAQ14784) ]4 SLFQQY orf-17 ] 51 aa 16.3 kDa 4.]9 orf-]55 (AAQI479]) 21 ARETFN orf-3 204 aa 21.9 kDa 4.99 orf-79 (NP_94484"8) 23 NEd 36 N -terminally blocked 38 H/A/V,G, C or modified, LNL 42 mixed 47 AYTPIV orf-] 6 452 aa 48.7 kDa 4.79 orf-150 (AAQ ]4790) 54 NE 120 NE As observed on SDS-PAGE, estimated from stained ladder As predicted by Clone Manager (SciEd Central), from the sequence data C Based on Bacteriophage Felix complete genome sequence: NC_005282.1 d "NE" indicates that insufficient protein was extracted from the PVDF blot to allow N-terminal sequencing; e "_" indicates information that is unknown as a result. a b 140 Discussion The sequencing process revealed several characteristics of the <l>Ea21-4 genolne. More than half of the genome was not represented in the random shotgun library, despite the presence of terminator sequences designed to prevent transcription of genes cloned into the vector. At least one of the unrepresented regions contained the endolysin gene, which is to be expected since this gene product is toxic to bacterial cells. For the same reason, the holin gene almost certainly lies in one of the regions that was not contained in the shotgun library. However, it is not knOWll why so much of the genome was not represented. By comparison, the authors responsible for sequencing the genome of the KVP40 vibriophage reported that 24 regions comprising only 25% of that 245 kb genome were not represented in the shotgun library (Miller et aI., 2003a). The use of insert sizes smaller than the 2-4 kb fragments used in these two studies would be nl0re likely to produce shotgun libraries with greater genome coverage. Information about the packaged form of the genome can also be inferred fronl the sequencing process. The replication strategy of linear double-stranded DNA phages is related to the form of the packaged virion DNA, in that there nlust be a means of preserving the genetic information upstream of the initial replication primers (Jardine et aI., 2006). Phages such as lambda have 5' cohesive ends that permit circularization of the phage ,gell0nle upon infection. The replication of the <1>29 genolne is prilned by glycoproteillS that are covalently linked to the 5' termini. The T3 and T7 phages have terminal direct repeats that allow recombination. Each of these replication strategies results in packaged gell0nles with a defined Ullit length and identical terminal sequences for all virions in the population. Alternatively, the genolnes of a populatioll may also be circularly pernluted with respect to each other, as 'seen in T4, SPP1, and P22. This is 141 the result of less specific terminal redundancy, in which phage ,genomes are replicated as a concatemer, and a length of DNA greater than 100% of the single-copy genome is packaged into each phage head (termed "headful packaging"). Direct, custom-prillled sequencing frolll the termini of the final contig was halted when the same 182 bp sequence was obtained from both ends. <f>Ea21-4 clearly uses direct terminal repeats to prevellt sequence loss during replication, but these repeats could be defined, as in T7, or circularly permuted, as in T4-like phages. The 82 ORFs listed in Table 3-4 are almost cel1ainly an underestimate of the total number of <f>Ea21-4 genes. Most phage genomes contain very densely packed genes, yet these 82 ORFs comprise only 78% of the single-copy genolne. In comparison to SOlne other phages of the Enterobacteriaceae, coding sequences comprise 91.5% of the Era103 genome, 92.1 % of the KVP40 genome, 93% of the T4 genome, and at least 87% of the lalllbda genollle (Sullllner et aI., 2007; Miller et aI., 2003a; Miller et aI., 2003b; Sallger et aI., 1982). There are four places ill the <f>Ea21-4 genome where two adjacent ORFs are separated by more than 1000 bp: between orf-81 and orf-82, between orf-4 and orf-5, between orf-5 and orf-6, and between orf-12 al1d orf-13. The region between orf-5 and orf-6 encodes a tRNA cluster. The other three regions are excellent candidates to harbour ORFs that were not identified by the conservative methods employed here. Nineteen of the predicted ORFs, or 23%, are completely unique to <f>Ea21-4, containing no recognizable conserved domains and having no significant similarity to the products of known or predicted genes. An additional 34 ORFs, or 42% appear to be conserved hypothetical proteins, with similarities to unidentified predicted gene products in at least one other phage genollle. The discovery of unique genes is a common theme in phage genome sequencillg (Kapfualnlner et aI., 2002; Miller et aI., 2003a; Pedulla et aI., 2003; Rohwer, 2003), alld is indicative of a substalltial 142 pool of information about phage ecology that has yet to be tapped. Nineteen phage-encoded tRNAs were predicted, including one pseudo-tRNA. Of the 18 predicted to be functional, one is suspicious since the anti-codoll could not be predicted. The purpose of phage-encoded tRNAs can be to facilitate the incorporation of those amillo acids for which the phage and host exhibit substantially different codon usage patterns. This appears to be the case for a cluster of eight tRNAs in T4 (Kunisawa, 1992). However, the vibriophage KVP40 genome contains a cluster of 25 apparently functional tRNAs, and yet significant codon usage differences between host and phage were not observed for many of the corresponding amino acids (Miller et aI., 2003a). The authors suggest that there may be different codon usage pattenls within KVP40 gene clusters, rather than a genome-wide trend. Both possibilities should be considered for <l>Ea21-4. Few protein-encoding genes involved in transcription or translation were identified in the <l>Ea21-4 genome. Orf-37 encodes a nucleic acid-independent RNA polymerase. This protein family includes the poly(A)-polymerase enzymes that are responsible for generating the poly(A) tail of nascent mRNAs, as well as the tRNA nucleotidyltransferases that attach the CCA tliplet to the 3' end oftRNA. The latter function would be consistent with the presence of phage-en-coded tRNA genes, two of which are not predicted to have .genomically encoded 3' CCA temlilli. The <l>Ea21-4 genome includes genes for many aspects of nucleic acid metabolislTI. Dihydrofolate reductase and thymidylate synthase both participate in the cycling of5,6,7,8tetrahydrofolate to produce deoxythymidine phosphate. The two proteins are encoded by adjacent ORFs, and are likely transcribed as part of a single operon since the gelles are separated by 3 nucleotides. The ribonucleotide triphosphate reductase holoenzynle 'ellcoded by orf-52 and orf-53 is responsible for the conversion of ribonucleotides into their deoxyribonucleotide forms. The product of orf-51, which immediately precedes these two genes, contains a glutaredoxin donlain, and may therefore function as an electron carrier for ribonucleotide reductase holoenzynle. DeoxYfibonucleoside monophosphate kinase also plays a role in nucleotide tumov·er, catalyzing the phosphorylation of dNMP, at the expense of ATP. <l>Ea21-4 also appears to encode a NadV homolog. NadV is a principle component ofa pYfidine nucleotide salvage pathway that has been identified ill a few bacterial genomes alld, recently, in vibriophage KVP40 (Martin, Shea, & Mulks, 2001; Miller et aI., 2003a). The NadV protein, a nicotinamide phosphoribosyltransferase, converts nicotinamide to nicotinanlide mononucleotide. The KVP40 genome also appears to encode proteins that catalyze the active uptake of nicotinamide mononucleotide, its conversion to NAD+, and its regeneration from NAD+. Homologues of these other genes are not found in <l>Ea21-4, none of the described genes are found in E. amylovora. Interestingly, E. amylovora displays a strict requirement for nicotinic acid (Hauben & Swings, 2005), which suggests that the <l>Ea21-4-encoded NadV may playa role in phage adaptation to or interference with host cell metabolism. It is not surprising that the snlall number of genes identified by sequence similarity alone included those for thymidylate synthase, DNA polytnerase, ribollucleotide reductases, and dihydrofolate reductase. Families of these genes appear to have diverged prior to the divergence of eukaryotes and prokaryotes, and therefore lnembers of one fanlily tend to be silnilar across different branches of the phylogenetic tree, with less similarity between different families in closely related organisms (Briissow & Kutter, 2005). Escherichia coli phage T4, the best characterized of the Myoviridae, consists of 35 144 different structural proteins (Mosig & Eiserling, 2006). The <l>Ea21-4 genome cOlltains genes for at least 10, including the four that were identified by N-tenninal sequencing. Orf-3 encodes a 21 kDa structural protein that has an Ig-like domain from the Big-2 family (aa region). The immunoglobulin fold is very common and very widely dispersed (Halaby, Poupon, & Momon, 1999). The Big-2 type consists of seven p-strallds arranged in two parallel sheets, and is most commonly found in proteins involved in bacterial cell adhesion (Fraser et aI., 2006). At least one protein with an Ig-like domain is found in the proteome of about 25% of fully sequen'ced Caudovirales phages, usually within tail fiber, baseplate wedge initiator, major tail, lnajor head, and highly immunogenic outer capsid (HOC) proteins (Fraser et aI., 2006). In this case, orf-3 is lnore likely to be one of the head proteins, since the genes for tail components and tail assembly seem to be clustered on the other side of a putative prohead protease, in the vicinity of orf-31 through orf-19. The predicted orf-3gene product did align with part of a Felix protein annotated as "HK97 nlajor tail protein". This Felix protein contaiIls an Ig-like donlain from the I-set family, and its annotation has been questioned because it is based on similarity to a Siphoviridae tail protein (Felix belongs to Myoviridae) and thus probably reflects the sinlilarity of their Ig-like domains rather than true functional homology of the whole proteins (Fraser et aI., 2006). Similarly, the alignment between these <l>Ea21-4 and Felix gene products is almost celiainly based on the common features of the Big-2 and I-set Ig-like domains. Orf-12 encodes an 11 kDa structural protein that shares 40% identity with a predicted Felix protein of unknown function. Since all adjacent gene, orf-ll, has a conserved ClpP serine protease donlain, which is commOll to prohead protease enzymes, it is possible that orf-12 encodes one of the capsid conlponents. Orf-16 alld orf-17 encode 47 kDa and 21 kDa stluctural 145 proteins that fall between the region associated with head morphogenesis and the region associated with tail morphogenesis. The proteins encoded by orf-16 and orf-17 are 58% and 82% identical to adjacent predicted Felix proteins of unknown function. The ~ 120 kDa structural protein seen on the SDS-PAGE gel is probably encoded by orf-31, the largest gene in the <j>Ea214 genome. The predicted product of this gene is 1157 amino acids, has a predicted l11ass of 130 kDa, and it is similar to the putative tail fibre protein of Xanthomonas oryzae phage OP 1, another tailed, dsDNA phage. The nl0rphogenesis of mature <j>Ea21-4 virions also illvolves DNA packaging. Orf-7 contains a conserved phage terminase domain. Terminaseholoenzynles consist of two proteins, a large and a small subunit(Jardine & Anderson, 2006). They target the DNA to the proh,ead, initiate packaging, power DNA translocation by ATP hydrolysis, and terminate packaging. Assuming that orf-7 does, in fact, encode one of the terminase subunits, presunlably either orf-'6 or orf-8 encodes the second subunit of the holoenzyme. The final stage of lytic phage replication is host cell lysis. As with all known doublestranded nucleic acid phages, <j>Ea21-4 appears to use a holin-endolysin lytic strategy. The <j>Ea21-4 endolysin bears little similarity to the <j>Eal or T4 lysozyme, and is therefore not likely to be a muramidase, or "true lysozyme". No putative holin gene was identified itl the <j>Ea21-4 genome, but this is perhaps not surprising given the incredible degree of holin diversity. There are currently more than 250 putative phage holins in over 50 families, and the lethality of the gene product has prevented functional analysis of all but a few of them (Young & Wang,·2006). The key feature is the presence of transmembrane domains, usually two or three, with cytoplasmic C-termini (classes II and I, respectively), or two trallsmelnbrane domains with a 146 large C-terminal periplasmic domain (class III) (Young & Wang, 2006). The holin is responsible for the precise timing of endolysin-mediated host cell lysis. On a whole-genome scale, cPEa21-4 is not silnilar to th·e cPEral03 genome or the sequenced portion of cPEal. The cPEal fragment only contains three genes: a depolymerase gene, and the lysozyme and holin genes just discussed. No depolytnerase gene was identified within the cPEa21-4 genome. Depolymerase is an virion-bound ·exopolysaccharide-degrading enzyme. It is most commonly isolated from lytic Podoviridae (Scholl & MeiTil, 2005), which is consistel1t with its presence in cPEal and its absence in cPEa21-4. The morphology of cPEral03 has not beel1 described in detail (Vandenbergh, Wright, & Vidaver, 1985; Vandenbergh & Cole, 1986), but the production of a depolymerase and the similarity of its genome size to that of cPEa1 strongly suggest that it, like cPEal, is a member of the Podoviridae. The only significant similarity between cPEa21-4 and a previously described phage is with Salmonella phage Felix 01. Felix is a ubiquitous vilulent phage that infects al1110st all Salmonellae and is therefore used as a diagnostic and typin,g phage (Cherry et aI., 1954). Like cPEa21-4, Felix belongs to the Myoviridae. Its head has been repolted to be either 72 11111 (Ackermann & Nguyen, 1983) or 60 nm across (Lindberg & Holme, 1969), the latter of which is the same size as cPEa21-4. The Felix genome is 86.2 kb, which is only 1.5 kb larger than the cPEa21-4 genome, al1d has a lower GC content, about 39% (Sriranganathan ·et aI., 2006). The similarity of the cPEa21-4 and Felix genomes appears to be largely 'correlated with their shared morphology. When the aligned genome sequences were cOll1pared, the regions that share 40-65% identity mostly involve the regions in cPEa21-4 that appear to encode tail fibre stlucture al1d assembly proteins. 147 There is currently no publication associated with the complete sequence of Felix, but the authors of a study dealing with an 11.5 kb fragment declared its sequence sufficiently different from known phages as to represent "the prototype of a new phage family" (Kuhn et aI., 2002). E. amylovora phage <l>Ea21-4 would appear, then, to be the second m·ember of this "family". These two phages both encode 21 predicted ORFs that are not significantly similar to any known gene. <l>Ea21-4 orf-3, orf-12, orf-16, and orf-17, idelltified as ellcoding structural proteins in this study, are significantly similar to predicted Felix genes orf-79, orf-135, orf-150, and orf-155, respectively. The organization of the <l>Ea21-4 genome is consistent with the functional clustering that is usually seen in phage genomes (Calendar, 2006). The "immediate early", or simply "early", genes are responsible for protecting the phage genome and mediating the transition from cellular to phage-directed metabolism. They are transcribed by host RNA polymerase, and so their promotors may resemble those of the host genome more closely than do other prOlTIotors in the phage genome. The middle genes are responsible for replicating the phage genOlTIe, and for the transcription and translation of most of the phage genome. These genes appear to enCOlllpass at least orf-33 through orf-57. Finally, the late genes are responsible for morphog·enesis alld lysis. (Guttman, Raya, and Kutter, 2005) Within the <l>Ea21-4 genollle, the structural, assell1bly and lysis genes appear to encompass orf-31 through orf-2, and perhaps some of orf-70 through orf82. No genes associated with a lysogenic life cycle were predicted. In summary, the genome sequence of E. amylovora phage <l>Ea21-4 rev·eals this phage to be substantially different from previously characterized phages, including the extrelllely limited number of genetically characterized E. amylovora phages. 148 Chapter 4: Real-time PCR reveals competition between Erwinia amylovora and Erwinia pyrifoliae on pear blossoms The following chapter is a manuscript that has been prepared for submission to Phytopathology. Submission has been delayed untillicensiIlg negotiations regarding one of the reagents are concluded. Authorship of the manuscript is as follows: Susan M. Lehman, Won-Sik Kim, Alan J. Castle, and Antonet M. Svircev, with equal contributions by the first and second authors. First and third authors: Department of Biological Science, Brock University, 500 Glenridge Avenue, S1. Catharines, ON, Canada L2S 3A 1; first, secolld and fourth authors: Agriculture and AgriFood Canada, Southern Crop Protection & Food Research Centre, 4902 Victoria Ave. North, P.O. Box 6000, Vineland 'Station, ON, Callada LOR 2EO. Experimental contributions to this work were as follows: Design and optinlization of the described prilners and probes: Dr. Kim Specificity testing: Dr. Kim, with some contributions by S. M. Lehlnan Sensitivity testing: both S. M. Lehman and Dr. Kim Competition experiments and population nl0nitoring: S. M. Lehlnan Data analysis: S. M. Lehman The manuscript was written by S. M. Lehman, with input froln the other authors. 149 Chapter 4: Real-time peR reveals competition between Erwinia amylovora and Erwinia pyrifoliae on pear blossoms Abstract E. amylovora and E. pyrifoliae are the causative agents of fire blight alld Asian pear blight, respectively. The pathogens are closely related, with overlapping host ranges. Data are unavailable on the current distribution of E. pyrifoliae, and on the interaction between the two species when they are present together on the same host. In this study, a duplex real-time peR protocol was developed to monitor the population dynamics of E. amylovora and E. pyrifoliae "on the surface of Bartlett pear blossoms. Bacterial cells washed from blossoms were used directly as the peR template without DNA extraction. Primers and a probe based on the E. amylovora levansucrase gene detected all E. amylovora isolates. All E. pyrifoliae isolates, including the Japanese Erwinia strains previously described as E. amylovora, were detected with a plimer alld probe combination based on the E. pyrifoliae hrpW gene. Disease appearance and severity were not significantly different in blossoms inoculated with individual Erwinia species or with a mixture of the two species. However, E. amylovora grew to higher population sizes thall did E. pyrifoliae in both single species inoculations and in mixtures, 'suggesting that E. amylovora has a greater competitive fitness on Bartlett pear blossoms than E. pyrifoliae. 150 Introduction E. amylovora is well known as the causative agent of fire blight, a n'ecrotic disease affecting species in the Maloidae subfamily. Fire blight was first described on apple trees in 1780 in the northeastern United States. The disease has since spread throughout lTIuch of North America, to New Zealand, England, most of Europe, and the Middle East (Bonn & van der Zwet, 2000). E. amylovora has a significant economic impact on comlTIercial apple and pear crops. In the United States alone, fire blight is estimated to cost about $100 million annually in crop loss and management costs (Norelli, Jones, & Aldwinckle, 2003). The annual disease 'cycle begins when E. amylovora present in the orchard, usually in the margins of overwintering cankers from previous year's infections, colonize open blossoms in the spring (van der Zwet & Keil, 1979). Under favourable temperature and moisture conditions, the E. amylvora population will grow and infect via natural openings in the blossom. Subsequent intercellular and intravascular growth of the pathogen leads to necrosis which can spread rapidly throughout the tree (van der Zwet & Keil, 1979). Chemical and biological control methods are available to prevent initial infections, but once a tree is infected, the spread of the bacteria can Oilly be stopped by cutting out infected tissue. Erwinia pyrifoliae Kim et aI., was isolated in the 1990s from Asian pear trees in South Korea that showed symptoms very similar to fire blight. The isolated organism had several morphological and biochemical characteristics in conlmon with E. amylovora (Rbim et aI., 1999), and its 16S rRNA gene shared 99% homology with the E. amw'vlovora 16S rRt"\J"A gelle (Kim et aI., 1999). However, DNA:DNA hybridization data and the sequenc·es of the 16S-23S intergenic transcribed spacer regions revealed the E. pyrifoliae isolates to be a single species, 151 distinct from E. amylovora (Kim, Gardan, Rhim, & Geider, 1999). Additiollallll0lecular evidence for the distinction between E. pyrifoliae and E. amylovora has also been presented (McGhee et aI., 2002; Maxson-Stein et aI., 2003; Jock & Geider, 2004). To date, E. pyrifoliae has only been isolated from Asian pear trees, and has not been reported since the localized outbreaks in the 1990s (Kim et aI., 2001a). Extensive work has been done to genetically characterize E. pyrifoliae, but except for some small-scale pathogenicity tests on apple and pear seedlings and ilnmature pear fruit, the in vivo behaviour of this organism has not been studied. Despite the genetic ditTerences on which their classification is based, and difTerences in celiain pathogellicity factors (Kim et aI., 2002), both E. an1ylovora and E. pyrifoliae are pathogenic on both Asian and westelTI pear varieties (Kim et aI., 1999; Kim et aI., 2001b). An understandiIlg of how the two species interact when present together on open blossom may shed light on their current distribution, and on the potential for future spread of each organism. Here, we describe the development of a duplex real-time PCR protocol that p,eImits simultaneous quantification of E. pyrifoliae and E. amylovora directly from plant samples. This approach is rapid and quantitative, and thus has advantages over previously described nlethods for differentiating E. amylovora and E. pyrifoliae using conventional PCR and gel electrophoresis (Kim et aI., 2001b). We then used this system to study the population dyilainics of these two pathogens when present individually or together on the surface of Bartlett pear blossoms. A second purpose of this study was to use the two primer alld probe combinatiolls to investigate relationships between E. amy'lovora, E. pyrifoliae and Erwinia isolates froin Japan. 152 The latter were isolated frOITI Asian pear trees on Hokkaido. hlitial studies (Be'eret aI., 1996) suggested that these isolates were very similar to E. amylovora, which agreed with a 19-81 Japanese report (cited by Beer et aI., 1996). However, the subsequent identification of E. pyrifoliae as a distinct species from E. amylovora, despite many shared morpholo.gical, metabolic, and genetic similarities, prompted a reassessment of that conclusion. 153 Methods Bacterial Strains and Media Isolates of E. amylovora, E. pyrifoliae, Pantoea agglomerans (foffi1erly Erwinia herbicola), Erwinia carotovora, and Escherichia coli were grown on Difco nutrient agar (Bectoll Dickinson and Company, Sparks, MD). Pseuodomonas isolates were grown on NBY agar (Schaad, Jones, & Chun, 2001) containing 8 giL Difco nutrient broth (BD), 2 giL Bacto yeast extract (BD), 2.5 giL glucose, 2 giL K 2HP0 4 , 0.5 giL KH2P04 , 15 giL Bacto agar (BD), and 1.,0 mL of a sterile solution of 1.0 M MgS0 4 .7H2 0 which was added after autoclaving . All strains were incubated at 28°C, except for E. coli, which was incubated at 37°C. Bacteria used in th'ese experiments are described in Table 1. All listed isolates were used for specificity testing of realtime PCR probes and primers. E. amylovora Ea6-4 and E. pyrifoliae Ep1/96 were used for competition assays. Real-time peR E. amylovora and E. pyrifoliae -specific primers and TaqMan probes were designed based on the levansucrase gene (GenBank: X75079) alld the hrpW gene (G,enBank: AY237642) respectively by using web-based software provided by Integrated DNA Technologies (Coralville, IA). Candidate oligonucleotide designs were analyzed to ensure con1patibility in a duplex reaction. Probe and primer sequences are listed in Table 4-2. All primers and probes w'ere synthesized by Integrated DNA Technologies. Reactions were run in 25 ilL volumes using IX ThelIDoPol PCR buffer (New Englalld Biolabs, Ipswich, MA), 0.2 mM each dTTP, dCTP, dATP, and clGTP (lnvitro·gen CDrporation, 154 Carlsbad, CA), 2 mM MgCI 2, 0.1 nlM each probe, 0.2 mM each primer, and 1.5 U of Taq polymerase (NEB). Two ~L of the processed 'Sample was us'ed as the teillplate ill each reactioll. Reactions were lun in a Stratagene Mx4000 Multiplex Quantitative PCR systeln «Stratagene, La Jolla, CA) under the following cOllditions: 95°C for 5 nlin; 40 cycles of 95°C for lOs alld 6'0°C for 16s, with two endpoint fluorescence readings during ,each alnplification segment. Positive, suspension buffer, and master lnix-only controls were also included in each reaction cycle. Specificity tests were conducted ill duplex reactions USillg the bacteria listed ill Table 4-1, Sillgly, and in mixtures. Optimization ofReal-time peRfor Quantitative Analysis Singleplex and duplex standard curves were constructed to allow quantificatioll of each target organism. Fresh, ovelnight plate cultures of E. amylovora Ea6-4 and E. pyrifoliae Epl/96 were aseptically scraped from the plate surface and suspen'ded ill 0:01 M PB, pH 6.8 to a concentration of at least 1 x 109 CFU/mL (OD 600 = 0.6) using a Becknlan DU 640 spectrophotometer. The concentration of each suspension was -calibrated by plating 100 ~L of serial dilutions in PB and counting single colonies after 2 d. Telnplates for real-tinle PCR standard curves were prepared by serially diluting cell suspensions in clean blossom wash, either individually or as a mixture of both species. Standard curve COllstluction was repeated using new bacterial cultures. PCR reactions were run in triplicate. 155 Table 4-1. Bacterial strains used in this study Host Origin Source or Reference Ea6-4 Pyrus sp. Canada Gill et aI., 2003 Ea1/79 Cotoneaster sp. Germany Falkenstein et aI., 1988 Ea321 Crataegus oxyacantha France CFBP 1367 IH3-1 Raphiolepis indica USA Holcomb, 1998 IL5 Rubus sp. USA McManus & Jones, 1995b CAll Malus sp. USA Chiou & Jones, 1991 UTRJ2* Malus sp. USA Thomson & Ockey, 2001 Leb B66* Malus sp. Lebanon Saad et aI., 2000 NA1614-2a* Crataegus sp. Spain Llop et aI., 2006 Ep1/96 Pyrus pyrifolia Korea Kim et aI., 1999 Ep2/97 Pyrus pyrifolia Korea Jock et aI., 2003b Ejp556 Pyrus pyrifolia Japan Kim et aI., 2001 a Ejp557 Pyrus pyrifolia Japan Kim et aI., 2001 a Ejp617 Pyrus pyrifolia Japan Kim et aI., 2001 a Eh21-5 Pyrus sp. Canada A. M. Svircev C9-1 Malus sp. USA Ishimaru et aI., 1988 E325 Malus sp. USA Pusey, 1997 Isolate E. amylovora a E. pyrifoliae Japanese Erwinia strains P. agglomerans e b b E. carotovora Ecc26 D. Cupples d E. coli DH-5a A. J. Castle e P. jluorescens A506 Lindemann and Suslow, 1987 P. syringae pv. papulans 4404 A. M. Svircev b e * strains marked with an asterisk do not contain the pEA29 plasmid Collection Franyaise de Bacteries Phytopathogenes b P. agglomerans C9-1 and P. jluorescens A506 are the active ingredients in commercial bacterial antagonists produced under the BlightBan® label by Plant Health Technologies, for the control of fire blight. P. agglon1erans E325 is the active ingredient in a commercial bacterial antagonist produced by Northwest Agricultural Products for similar use. e Agriculture and Agri-Food Canada, Southern Crop Protection and Food Research Centre, Vineland Station, Canada d Agriculture and Agri-Food Canada, Southern Crop Protection and Food R-esearch Centre, London, Canada e Department of Biological Sciences, Brock University, S1. Catharines, Canada a 156 Table 4-2. Sequences, product sizes, and targets of the forward prill1er (F), the reverse prinler (R), and the TaqMan® probe (P) used for duplex detection and quantification of E. amylovora (Ea) and E. pyrifoliae (Ep). o ligonuc leotide a b Position Sequence (5' to 3 ') Ea-IscF CGCTAACAGCAGATCGCA 345-362 Ea-lscR AAATACGCGCACGACCAT 449-432 Ea-IscP (CY5)CTGA TAATCCGCAA TTCCAGGATG(IAbRQ) a Product Size 105 bp 366-389 b Ep-hrpwF CGCTAACCCGACTGTGCT 756-773 Ep-hrpwR TGAAGGTTTGCCCTTTGC 832-815 Ep-hrpwP (FAM)ATGACACCA TCATCGT AAAGGCGG(BHQ-l) b Binding positions are based on GenBank files X75079 (lse) and AY237642 (hrp W). IAbRQ and BHQ-l are the manufacturer's proprietary quencher molecules. 776-799 77 bp 157 E. amylovora and E. pyrifoliae Competition Assays Bartlett pear shoots bearing dOffilant buds were harvested in late winter. Shoots w,ere forced to bloom by placing in water at 23°C, in a well-lit room. Newly opened blossolns with yellow, undehisced anthers were harvested by hand and individually placed into scintillation vials such that the peduncle extended through a hole drilled in the lid and into the sterile tap water contained in the vial. Suspensions of E. amylovora and E. pyrifoliae were prepared by aseptically scraping cells from fresh, overnight cultures into 0.01 M PB, pH 6.8. Suspensions were adjusted to 1 x 109 CFU/mL (OD 600 = 0.6) using a Beckman DU 640 spectrophotometer, stored on ice, and further diluted in PB as needed immediately prior to application. Ten J.lL of a bacterial suspension was inoculated directly onto the hypanthium. Treatments were: 0.01 M PB, pH 6.8 (negative control); 1 x 10 6 CFU/blossolTI of either E. amylovora or E. pyrifoliae (positive control for virulence); 1 x 104 CFU/blossom of E. amylovora, or E. pyrifoliae, or a 1: 1 mixture of both for a total bacterial count "of 1 x 1.04 CFU/blossom; 1 x 102 CFU/blossom of E. amylovora, or E. pyrifoliae, or a 1: 1 lllixture ofb'oth for a total bacterial count of 1 x 10 2 CFU/blossom. Each treatment was applied to 20 blossollls. Following inoculation the blossoms were loosely sealed in large plastic bins containing about 500 mL of water to maintain a high relative humidity and incubated at 23°C. Ten blossoms from each treatment were assessed after 3 d, and another 10 blossonls after 5 d. Each blossom was removed from the scintillation vial using sterile forceps. Disease symptoms were scored USillg a severity index (0, no necrosis; 1, necrosis on th'e stigilla and hypanthium; 2, necrosis visible on the ilnmediate underside of the blossOlll; 3, 11ectosis 'extellds 158 into the ovary, no farther than the widest point; 4, necrosis extends to the base of the ovary; 5, necrosis extends into the peduncle). Every second blossom was further processed in order to quantify surface bacterial populations (see below). The experil11ent was conducted three tilnes using suspensions prepared from independent bacterial cultures. Assessment ofBacterial Numbers on Blossoms After symptom severity was assessed, the petals and peduncle were removed and the remaining tissue (ovary, hypathium, stamens, stigma) was plaoed in a sterile 1.5 mL microcentrifuge tube. One mL of Direct Plant Extraction Buffer (DiPEB, Cat. No. 00690, Agdia Inc., Elkhart, IN) was added to each tube. Tubes were mixed by vortexing briefly, placed in a water bath sonicator such that only the cap was not imlnersed, and sonicated for 3 min to dislodge bacterial cells. Plant tissue was removed using a sterile toothpick al1d discarded. Bacterial cells were concentrated 10-fold by centrifuging the blossom wash at 13 000 xg for 5 min, decanting the supernatant, and resuspending the pellet in 100 JlL of DiPEB. This processed sample was used directly as the template for real-time PCR. This procedure was also conducted using uninoculated blossoms, and the resulting clean blosson1 wash was used as a negative control. Statistical Analysis Regression analysis of standard curve data was conducted in Sigt11aPlot, version 8 (SPS'S, Inc., Chicago, IL). All other data were analyzed using SAS (Statistical Analysis Systenls '8.2; SAS Institute, Cary, NC). Disease severity index ratings fron1 the con1petitiol1 assay wete 159 converted to a percent scale based on the total hypanthium and ovary surface area showing visible necrosis. Population data were logarithmically transfonned (base 10) in order to produce normally distributed data suitable for further analysis. If a given pathogen species was not detected on a blossom, the population size was adjusted froln 0 to 2 CFU in order to avoid undefined log functions or ratios of logs. Disease severity data and trallSfolTIled population data were analyzed with the general linear model (PROC GLM), using the Tukey-Kramer lnean separation test for multiple comparisons. To assess whether competition occurred between E. amylovora and E. pyrifoliae, their growth on independent blossoms was compared to their growth on the sanle blossom. Each blossom inoculated with only E. amylovora was paired randomly with a blossom from the same replicate that had been inoculated with only E. pyrifoliae. The "relative growth performance" of each species was expressed as: logespecies 1 alone) / [logespecies 1 alone) + logespecies 2 alone)], or log(species 1 in mixture) / [log(species 1 in mixture) + log(species 2 in mixture)]. For each species, relative growth performallce alone was conlpared to relative growth performance in a mixture using a t-test with a Satterthwaite adjustment for ullequal variallces. There was no significant difference between replicated experiments (P > 0.05), therefore data from multiple replications were pooled. Where data were not normally distributed (P < 0.05 ill a Kolmogorov-Smimov test), the results of parametric tests and non-parametric tests were compared. In no case was the departure from normality sufficient to invalidate the parametric tests. All meall values are given as mean ± 95% confidellce linlits. 160 Results Specificity and Sensitivity ofDuplex Real-time peR The specificity of levansucrase (lsc) probe and primers was tested using all bacterial isolates and species listed in Table 4-1. All E. amylovora isolates w,ere positive regardless of the presence or absence of the pEA29 plasn1id. No cross reactions were observed with E. pyrifoliae strains or any other species. The hrpW probe and pri111ers detected all E. pyrifoliae strai11s as well as the Japanese Erwinia strains. This c01nbination did not detect any E. amylovora isolate or any other species. No change in threshold cycle (CT ) values were observed in the presn~ of non-target species. Fluorescence signals were regularly obtained from standardized sa1nples contai11ing as few as 20 CFU of either E. amylovora or E. pyrifoliae in a 25 L~ reaction. This detection threshold occurred at an approximate CT value of 33. Reaction efficiency, as calculated by the Stratagene Mx4000 software, was greater than 95 %. Duplex standard curves for E. amylovora and E. pyrifoliae are shown in Figure ·4-1. Concentrations higher than 5 x 106 CFU per 25 L~ reactio11 were 110t tes-t,ed. For E. amylovora, the relationship between (C T ) and initial quantity was linear across a 4.S-log range. For E. pyrifoliae, the relationship between CT and initial quantity was linear across a 5-log range. While detection, indicated by a fluorescence signal by 33 cycles, was reliable and for i11itial pathogen quantities above 20 CFU per 25 L~ reaction, quantification of both pathogens was highly repeatable only for initial quantities greater than approxi1nately 100 CFU per reaction. The standard curves for both species are . essentially identical; the slight differences between slope an-d intercept values were within the range of variation·between replicate real-time PCR luns. 161 Standard curves were prepared from eellsdiluted in buffer that had been sonicated with uninoculated blossoms according to the sanlpling procedure used for the competition assays. The curves therefore correspond directly to initial cell numbers, and account for the presence of any PCR inhibitors that may be presellt in the experilnental samples. Since blossom infection occurs through the stigma and hypanthium, petals and peduncles were removed from the blossom, and bacteria were collected from the surfaces of the remaining structures. Relnoval of petals and peduncles improved the sensitivity of bacterial quantification (data not shown), presumably because it reduced the total volume of plant tissue, and thus the concentration of plant-derived PCR inhibitors. DNA extraction was not performed as part of sample preparation. The components of the wash buffer, along with the 5 min initial denaturation cycle at 95°C, were sufficient to make the bacterial target DNA accessible for amplification. There is no substantial difference betweell the standard curves produced by singleplex and duplex reactions (data not shown). For a given target species, duplex detection of a given eell concentration occurred about 0.5 CT later than if the same sample was tested in a singleplex reaction. However this difference is within the range of variation between replicate luns. Sensitivity did not differ between duplex and singleplex reactions. Virulence ofE. amylovora and E. pyrifoliae Disease severity in blossoms treated with either E. amylovora or E. p.,vrifoliae alOll'e was modelled in terms of pathogen identity, initial inoculunl size, and time elapsed sillce ill0culation. Across the experiment as a whole, initial inoculunl size and tinle eJapsed since il10culation were significant predictors of disease severity (P < 0.0001 for both). There was also a signiflCant 162 interaction between these two factors (P = 0.0075). Specifically, initial inoculum size show,ed a significant (P < 0.0001), but qualitatively different, effect on disease severity at 'each of the two sampling times. Table 4-3 shows the results of a virulence assessment of E. amylovora and E. pyrifoliae on Bartlett pear blossoms. Blossoms showed no disease symptonls at the titl1e of inoculation. In general, mean disease severity increased with increasing inoculunl. Three days after inoculation, blossoms inoculated with 1 x 106 CFU of either pathogen showed sigllificantly more severe symptoms than blossonls that had received a smaller inoculull1 (P < 0.01). The differences between blossoms treated with 1 x 104 CFU, 1 x 102 CFU, and 0 CFU are less clear, as the difference between the 0 CFU and 1 x 104 CFU treatments is just significant at th'e 5% level (P = 0.0496). By day 5, however, the only significant difference in disease severity was between inoculated and uninoculated blossonls (P < 0.0001). There was no significant effect of pathogen identity on disease severity across the experiment as a whole (P = 0.5182), or within any level of initial ill0culum size at either sampling time (all P > 0.4100). Effect ojInitial Pathogen Ratio on Disease Severity Figure 4-2 shows the overall effect of initial pathogen ratio on disease developmellt, both 3 and 5 d after inoculation. Regardless of the time of observation, within each inoculunl level, there is no significant difference between the severity of symptonls in blossoms treated with a 1: 1 mixture of both pathogens and the severity of symptoms in blossoms treated with the sallle total amount of either species alone. 163 40 ....... IU ....... cv 30 I .....•~ ..~ . ... •. ~ -c 20 a .c en E. pyrifoliae • E. amylovora ~ () ~ .s:: I- m 10 0+- T . r 1~- T ' . - , T'" - r T~ " -r 'T ! 100 101 102 10 3 104 105 106 107 Initial quantity (CFU) Figure 4-1. Duplex standard curves for the quantitative detection of E. amylovora and E. pyrifoliae. Known amounts of each pathogen were diluted in clean blosson1 wash. Quantification of E. amylovora (solid line) is given by y = 38.4 - 3.48 log(x), R 2 = 0.975. Quantificatio11 of E. pyrifoliae (dotted line) is given byy= 37.0 - 3.481og(x), R 2 = 0.973. 164 Relative Population Sizes ofE. amylovora and E. pyrifoliae Bartlett pear blossoms were inoculated with E. amylovora, E. pyrifoliae, or a 1: 1 lllixture of the two. Real-time peR was used to deteffi1ine the population sizes of both species after 3 d and again after 5 d. The logarithmically transfolmed population sizes were IDodelled in teffi1S of pathogen identity, whether the pathogen was inoculated alone or as a nlixture, initial inoculum size, and time elapsed since inoculation. There was no difference in population sizes between days 3 and 5 (P = 0.5491), or between different levels of initial illocululll (P = 0.3414). Patho,gen identity was significant (P < 0.0001), as was nlixture (P = 0.0003), and there ",'as a significant interaction between these two factors (P = 0.0021). Figure 4-3A shows the mean population sizes of E. amylovora and E. pyrifoliae on the pear blossoms, when present alone or together. In blossoms inoculated with only one of the two pathogen species, E. amylovora populations grew to a ·greater size than did E. pyrifoliae populations (P = 0.0002). E. amylovora populations were not sigllificantly smaller 011 blossonls inoculated with both species than they were when E. pyrifoliae was absent (P = 0.6688). However, E. pyrifoliae population sizes were significantly slllaller in the pl~senc of E. amylovora than when they grew alone on the blossoms. This di11epence is lllore clearly delll0nstrated by the relative growth performance index (Figure 4-3B), where the .growth of a species in either situation was expressed in relation to the total growth of both species. The relative contribution of E. amylovora to total bacterial growth was significantly greater (P = 0.0001), and the relative contribution of E. pvvrifoliae was significantly less (P = 0.0001), 011 blossoms inoculated with both species than on bloSSOlTIS inoculated with a single species. 1.65 Table 4-3. The effect of initial inoculum size on disease severity in pear blossoms. Time Frame Initial Inoculum Size (CFU/blossom) Percent Disease (mean ± SD) 0 12.5 A 1 x 10 2 38.8 B 1 x 10 4 48.2 B 1 x 10 6 64.4 C 0 5.0 a Across Entire Experiment A 10 2 13.4 1 x 10 4 23.3 B 1 x 10 6 46.0 C 0 20.0 A 10 2 64.2 B 1 x 10 4 73.1 B 1 x 10 6 82.8 B 1 X A,B 3 days Elapsed Since Inoculation 1 X 5 days Elapsed Since Inoculation Disease severity is expressed as percent of surface area that is visibly necrotic. Values are the means of two experiments, each with 10 blossoms per treatment. Within each time frame, means with the same letter are not significantly different (P < 0.05, Tukey-Kramer test for multiple comparisons). a 166 100 80 I: w ~ : · ; l E. amy/ovora alone E. pyrifoliae alone IZ5Z5Zi5Z5Z5I mixture (1:1 ratio) m C 100 (1) OJ tn tn A c 60 rJe. 40 a aba 102 eu is abc abc a 104 ~ bc 20 Buffer Control Positive 6 Control (10 ) Inoculum (CFU/blossom) 80 In ~ 60 a :: -"------,n---J'--_ '---~ Buffer Control 102 104 Positive 6 Control (10 ) Inoculum (CFU/blossom) Figure 4-2. Effect of initial pathogen ratio on disease severity in blossoms inoculated with E. amylovora only, E. pyrifoliae only, or a 1: 1 mixture of both. Percent Disease is shown as mean + 95% confidence limit for A: 3 d after inoculation, B: 5 d after inoculation. On each day, l11eans with the same letter are not significantly different (P < 0.05). 167 Q;') "4) a 41) c o 8, y,:::::;,::::::::::::::::::::::::':'; . A. II1 EO'.. am. .·:/ovora ··r Eo< pyrifoliae «S E ~ 6 ;: 1.. 0 .,:., :.:.:.: . :.; U' c: , 0.8: ~ :..:.:,. :, :..:.:.:.: .:,..: :, :.;.;.: . . . . . . . . . P ..... :.: >, ••;••.;.;••.•• : ••••• : •.•.•,: ;.; ;.: ••••>.•••.•,: : :.: : ;.• ;.: Alone Mixture :.• ;••.; ., ••:.: •.;•• ~; B & 0..6: ie 0.4 £ 4'G 'S c. ... o en a. C;; 2 4). ~: ..2 0.2 «I ¢; .t:t: 0.0: ,;, o Alone Mixture Gr:owth Environment. ;j;••:...............•.•••••.;s..:.:. E. amyJoJ/ora E..pyrifoliae Pa:thogen Figure 4-3. A: Population sizes of E. amylovora and E. pyrifoliae on pear blossoms inoculated with each species alone, or with a 1: 1 mixture of both species. Populations are given as mean + 95% confidence limits. Means with the same letter are not significantly different (P < 0.01). B: Relative growth performance of E. amylovora and E. pyrifoliae when growing on Batilett pear blossoms, alone or in a mixture. Values are mean + 95% confidence litnits. 168 Discussion Fire blight causes economically significant losses of rosaceous fruit crops, particularly apple and pear, in many parts of the world. The causative agent, E. amylovora, is the focus of intensive research aimed at understanding and inhibiting its epiphytic and endophytic growth phases. E. pyrifoliae is very closely related to E. amylovora, and causes a silnilar disease on Asian pear trees. A rapid, sensitive, molecular method of detectillg and quantifying these phytopathogens will allow more convenient detection, and more accurate nl0nitoring of their in vivo growth characteristics. In this study, a duplex real-time PCR method was developed that allows simultaneous identification and quantification of E. amylovora and E. pyrifoliae directly from blossom surfaces, independent of their cultivability. Previous PCR methods for E. amylovora detection have been based on the pEA29 plasmid (McManus, Jones, & Bonn, 1996; Llop -et aI, 2000; Salm & Geider, 2004). A chromosomal target was chosen for this study because of the emergence of E. amylovora isolates in nature that lack pEA29 (Llop et aI., 2006). Specifically, the levallsucrase gelle (lse) was selected as the target region since levan is commonly produced by E. amylovora strains as a second exopolysaccharide and plays significant role ill virulence (Gross, Geier, Rudolph, & Geider, 1992; Geier & Geider, 1993). E. pyrifoliae and Japall'e'se Erwiniastrains fail to produce levan on sucrose-amended media (Kim et aI., 2001a). Therefore, ill order to differentiate E.amylovora from E. pyrifoliae, lse was selected as an amplification target. Specific quantification of E. pyrifoliae was achieved with primers alld a probe that amplify and detect a 77 bp region of the hrpW ,gelle. This gene is pal1 of tIle hrp clust,er, which is present on the chromosome of both E. amylovora and E. pyrifoliae, but which sho\vs 'Sufficient 169 interspecies variation among other hrp genes to permit reliable differentiation of E. pyrifoliae. The detection of the Japanese Erwinia isolates with a system based on differences between E. amylovora and E. pyrifoliae is further evidence that these isolates should not be considered variants of E. amylovora, but may well belong to E. pyrifoliae. This is supported by data from other studies, which indicate that these isolates are not identical to E. amylovora, and, in fact, resemble E. pyrifoliae more closely than E. amylovora (Kim et aI., 2001a; Jock et aI., 2003a; Matsuura et aI., 2007). Fluorescence signals were regularly obtained from standards containing as few as 20 CFU per 25 ilL reaction (C T = 33). Quantification was very reliable for cell concentrations as low as 100 CFU per 25 ilL reaction (C T = 30). For the purposes of qualitatitive detection, samples yielding a CT value of 30 or less would be considered positive for the respective pathogell, while samples yielding a CT value between 30 and 35 should be considered suspect and re-tested. During the population monitoring experiments described in this study, CT values less than 26 were very rarely obtained. Despite the fact that the primers alld probes described here target single-copy chromosomal genes, the sensitivity of the method developed ill this study is comparable to that achieved in a pEA29-based real-time PCR assay for E. amylovora alone (Salm & Geidel", 2004). One of the major advantages of the plimers and probes developed in this study is that they are compatible in a duplex reaction, allowing Sil11ultaneous, independent qualltification of both E. amylovora and E. pyrifoliae. The primers and probes developed in this study were used to investigate the population dynamics of E. pyrifoliae on th'e blossom surface, and to study the interactions of E. pyrifoliae and E. amylovora 011 a comn1011 110St plant. The bloSSOll) bioassay 170 used for these experiments was based on the fact that prinlary infection by E. am.ylovora occurs through the reproductive organs (van der Zwet & Keil, 1979). Therefore each treatnlent group consisted of a series of isolated blossonls that were individually inoculated and assessed. In blossoms treated with a single pathogen species, there was no apparent difference between the virulence of E. amylovora and E. pyrifoliae on Bartlett pear blossoms. No nlore thatl 100 cells of either pathogen were needed to cause fire blight symptonls in a sillgle blos-solu, and there was no difference in the progression of symptoms resulting from inoculation with a ,given amount of either pathogen. However, the mean population size of E. pyrifoliae on the blossom surfaces was significantly smaller than the mean E. amylovora population size, which suggests that E. pyrifoliae does not grow as well as E. amylovora on Bartlett pear blossolus. It 111ay be that this difference was not sufficient to hinder the infection process. This suggestion of a threshold population size for infection is consistent with a report fronl Johnson et al (1993b) that incidelloe of fire blight in Bartlett pear orchards was associated with the nUluber of blossoms that suppol1ed E. amylovora populations greater than 1 x 10 5 CFU/blossom. In blossoms treated with a 1: 1 luixture of both E. amylovora alld E. p}'rifoliae, clisease severity was not significantly different than in blossoms inoculated with either species alone. Clearly the presence of both species together neither amplifies nor inhibits their patho~g,elicty. What is not clear from these results is whether the two species are contributing equally to the production of disease, or whether one species outcompetes the other to become dominant. Realtime PCR data revealed that the growth of E. amylovora on Balilett pear blosSOll1S was unaffected by the presence of an equal nUluber of E. pyrifoliae ·cells. In COlltrast, the growth of E. pyrifoliae was 11egatively affected by E. amylovora and the populatioll size of E. pyrifoliae 171 relative to E. amylovora was significantly smaller when both species were present together than when they grew alone. This observation suggests that E. amylovora has greater competitive fitness on Bartlett pear blossoms. It is not known whether this simply represents differential success in the utilization of available nutrients and other resources, or whether any specific antagonism is involved. It would be interesting to test whether the relative fitness of these pathogens is the saIne on blossoms of Pyrus pyrifolia, or whether E. pyrifoliae is better adapted to Asian pears thall E. amylovora. E. amylovora and E. pyrifoliae are not currently known to co-exist in nature, but they are very closely related and have partially overlapping host ranges. It is generally accepted that E. amylovora is native to North America (Bonn & van der Zwet, 2000), but nothing is knOWll about the evolutionary relationship between E. amylovora and E. pyrifoliae beyond the substantial genomic similarity. An understanding of how they behave when they are present together on different host species could shed light on their current distribution or potential distribution in the future. In light of the rapid spread of modem plant pathogens, it is certainly possible that th,ese species could be introduced to the same regions in the future. Their relativ,e fitness will influence the outcome of such introductions. The methods developed in this study are transferable to a field situation, and have been successfully used to monitor E. amylovora populations in experimelltal orchards as pali of a fire blight biological control study (Kim, Lehman, Castle, & Svircev, unpublished data). The omission of a DNA extraction step makes this technique attractiv,e in terlllS of tinle- and costsavings, especially for field studies, which tend to rapidly generate a large nunlber of sanlples. The potential for simultaneous quantification of several target species is not linlited to E. 172 amylovora and E. pyrifoliae. The methods describ·ed here should be easily applicable to simultaneous monitoring of a pathogen and biocontrol agent. The primers and probes developed in this study are species-specific. Strain-specific real-time PCR has been used to mOllitor the population of an applied fire blight biocontrol agent, Pseudomonas jluorescens EPS62e (Pujol, Badosa, Manceau, & Montesinos, 2006). Thecombillation of species-specific and straill-specific primers and probes. could be used to study the relative contribution of an applied biocolltrol agent to the total population of that species. Ultimately, since all prinlers and probes nlust fullCtioll at the same reaction temperatute, the applicability of multiplex real-tinle PCR to phytopathology, in general, and biocontrol, in particular, will depend on the availability of a large amount of sequence data. Acknowledgements This work was supported by an Improved Famling Systenls and Practices Initiative grant from Agriculture and Agri-Food Canada. S. M. Lehman is suppolied by a doctoral Post-graduate Scholarship from the Natural Sciences and Engineering Research Council of Canada. The authors thank K. Geider for providing the Japanese E. pyrifoliae strains, Ejp556 and Ejp557, and for generously assisting with the specificity testing involving those strains. S. V. Beer (Cornell University), G. W. Sundin (Michigan State University), M. M. Lopez (IVIA, Spain), L. Pusey (United States Department of Agriculture), and D. Cupples (Agriculture alld Agri-Food Canada) each provided certain bacterial strains for use during specificity testing. We also thank Ed Barszcz, Barry Kemp, Brad Arbon, and ChelTY Lane Orcharcls (Vineland, ON) for assistance in acquiring the budwood used for the competition assays. 173 Part III: Development and Evaluation of the Pha·ge-Carrier Biopesticide 174 Chapter '5: Evaluation of the interactions between Erwinia amyiovora, Pantoea aggiomerans Eh21-5, and Erwinia pha,ges Abstract The in vitro and in planta growth characteristics and interactions of E. amylovora, P. agglomerans, and E. amylovora phages were studied in order to establish parameters for treatment preparation and application during field trials. In vitro growth curves were constructed for P. agglomerans Eh21-5, E. amylovora Ea6-4 and Ea29-7, and two Erwinia phages. \Vith adequate aeration, the doubling time of exponentially growing E. amylovora and P. agglomerans is approximately 45 min in nutrient broth at 27°C. The latency period of <t>Ea31-3 in Eall0 is approximately 1.5 h, but the initial rate of adsorption is slow. The in planta interactions among E. amylovora, the P. agglomerans carrier, and four Erwinia phages were also studied. Four phages were used to test the effects of treatment timing and multiplicity of infection on biocontrol efficacy using the pear blossom assay. Phage-carrier combinations were nl0re effective at reducing fire blight symptom severity than the carrier alone if 3 h (vs. 0 h) was allowed to elapse between treatment application and pathogen application. 175 Introduction The replication of phages is very dependent upon the nletabolic condition of the host cell, and by extension the environment in which the cell exists. The environment 011 a pear or apple hypanthium is very different fronl in vitro conditions, but both are relevant to the dev·elopnlellt of a phage-based biopesticide. The biopesticide components lllUSt be effective in the field, but they must also be prepared in a laboratory, be it on a research or all industrial scale. Optimizing the propagation of phages in liquid culture requires all understanding of a phage's growth characteristics. These are different for each phage-host COlllbination alld set of incubation conditions, but even a general idea of the growth cycle can be helpful. Adsorption of tailed phages to Gram-negative bacteria is frequently based on conlponents of the outer membrane, and may require the presence of certain cofactors in the surroundillg environll1ent (Kutter, Raya, & Carlson, 2005). Adsorption can be a multi-stage process in which reversible illteractionB involving host recognition and virion positioning are followed by irreversible bindillg of phage proteins to receptors on the host cell, penetration of the cell wall and cell membrane, and transfer of the phage genome into the host cell (Kutter, Raya, & Carlson, 2005). Irreversible binding marks the beginning of the "eclipse" period, durillg which artificial lysis of the host cell will not result ill the release of infective phages, since the infe-ctin;g phage can not free itself to infect another cell, and no new virions have been produced. The eclipse period lasts until the first progeny phages have been assembled withill the 'c'ell. After this point, artificial lysis of the cell will cause the release of an in.creasing nunlber of intact and infective progeny virions. The time between irreversible phage adsorption and pha,ge-directed lysis of the 176 host is called the latent period. The average nunlber of progeny phages released by a single infected cell due to phage-directed lysis is known as the burst size, and can vary substantially among specific phage-host combinations. Since phage DNA replication and gene expression depend 011 host conlponents such as ribosomes and raw materials, the length of the eclipse and latency periods call be affected by th,e nutritional status and metabolic activity of the cell (Kutter, Raya, & Carlson, 2005). The efficiellt propagation of phages therefore requires all understanding of the optill1umgrowth conditions of both phages and their host bacteria. Once propagated, phages that are to be used therapeutically must then be stored ill a manner that is consistent with the conditions of their eventual use, and that l11aintaills a stable population of viable virions. To address the issues ofpropagatioll and preservation, the in vitro growth and ·survival characteristics of some of the bacteria and phages to be us,ed in field trials were exanlined. Growth curves were constructed for the carrier, P. agglomerans, and for two of the strains ofE. amylovora on which phages will be propagated. The lel1gth of the latent period in phage-il1fected E. amylovora liquid cultures was studied. In additioll, the lon.g-telTIl survival ofm rph l gica ~y different phages was monitored in liquid media in which high-titre pllage stocks might be stored, and in ionic solutions that may be used to encapsulate phages as part of a field delivery systeln. To facilitate the transfer of the phage-carrier biopesticide into the field, the pear blosSOlll bioassay was used to evaluate the efficacy of phage-carrier conlbinations in reducing the severity of E. amylovora-induced necrosis, and to evaluate how MOl and prior establishlnent ofnlicrobe populations influence that efficacy. 177 Methods Bacteria, Phages, and Growth Media Media and growth conditions w,ere as described in Chapter 2, ullless otherwise indicated. Table 5-1 lists the bacteria and phages used in this experiment. Isolate origins are described in Table 2-1 and Table 2-2. Bacterial Growth Curves The in vitro growth of E. amylovora Ea6-4, E. amylovora Ea29-7, and P. agglom,erans Eh21-5 was studied by nlonitoring their populations in pure liquid ,.culture. One litre flasks containing 500 mL of nutrient broth were inoculated with 10 mL of a bacterial suspellsion prepared to 1 x 109 CFU/mL in PB. 1~he cell 'Concentration was lTIonitofed by spectrophotometry, specifically, the absorbance of light at 600 nnl. Growth curves were repeated at least twi-e,e, using independent bacterial cultures. Two different experiments were conducted. In the first, E. a~ov lyma Ea6-4 alld P. agglomerans Eh21-5 were studied, and the flasks were incubated 011 all orbital shaker at 15{) rpm. In the second E. amylovora Ea6-4 and E. amylovora Ea29-7 were studied, and the flasks were incubated in a water bath. In this second experinlent, there was very little constant motioll, but the flasks were vigourously shaken when sanlples were taken every hour. Within leach experiment, at least 3 replicate flasks per species weregro\vn, each on 'dilfetent days. Spectrophotometric readings of cell cOllcentration were ~alibrtedy plating serial dilutions of Ea6-4 and Eh21-5 on nutrient agar. sinlultalTeously 178 Table 5-1. Bacteria and phages used in this study. Strain Bacterial Host Used For Ea6-4 NA Bacterial growth curves; Blossom assays Ea29-7 NA Bacterial growth curves EallO NA Phage growth curves NA Bacterial growth curves; Blossonl assays <pE a9-5 Ea-4 Blossom assays <pE a21-4 Ea6-4 Blossom assays <pE a31-3 Ea29-7 Phage growth curves <pE a35-4 EallO Phage growth curves <pEa35-5 EaD-7 Blossom assays <pEa46-1 A2 EaD-7 Blossom assays Erwinia amylovora Pantoea agglomerans Eh21-5 Phages 179 Phage Growth Curves Phage growth curves were constructed using a modification of the method d,esclibed by Carlson (2005). Ten millilitres of a 1 x 109 CFU/mL suspension of the bacterial host was added to each of two 1 L flasks containing 480 mL of nutrient broth. The flasks were incubated in a water batl1 at 27°C, and shaken vigourously at least every 30 Inin to ellsure adequate aeration. When the bacterial culture reached a concentration of approxilnately 1 x 10 8 CFU/I11L, $Ea31-3 was added to one flask for a final concentration of 1 x 109 PFU/mL, and an equivalent volume of nutrient broth was added to the uninfected control culture. Inl1nediately, and every 30 min thereafter, each flask was shaken well, and a 100 J.lL salnple was taken. These samples were serially diluted in nutrient broth. The concentration of bacterial survivors at 'each time point was determined by spreading a 100 J.lL aliquot from the 10-5 , 10-6 , and 10-7 dilutions over a 11utrient agar plate, and counting the nunlber of colonies that fOffiled after 2 d. The totalcOllcentration of infective centres was determined by mixing a 100 J.lL aliquot frolll the sall1e dilutions with 100 J.lL of 1 x 109 CFU/mL Eall0 and 3 mL of molten top agar to fom1 an agarov1erlay. eh~1 nUlnber of plaques was counted after 1 d of growth. Phage Survival in Liquid Media The long-term survival of <J>Ea46-1A2 in liquid Inedia stored at 4°C was tested. Six nledia were used: nutrient broth, 10 mM PB (pH 6.8) amended with 100 lllM NaCI al1d 2 nlMM:gCI 2 , reverse osmosis water, autoclaved reverse osmosis water, lllunicipal tap water, autoclaved municipal tap water. Filtered <J>Ea46-lA2 lysates were mixed with the desired suspensioll medium in a 3: 1 ratio and subjected to normal flow diafiltration at 4°C usin:g the Amicon 180 apparatus described in Chapter 7. The volume of each suspension was reduced from 2 L to 300 mL, an additional 600 mL of the diluent was added to the stirred oell, and the retentate was reduced to a final volume of300 mL. The result was a 12-fold dilution of dissolved substances, with a 30% increase in phage concentration. The concentration of viable phage in each nledium was estimated by serially diluting an aliquot of each in nutrient broth, and plating 100 J..LL of each dilution according to the soft agar overlay technique described in Chapter 2. Ail of the suspensions still had a yellow tinge following diafiltratioll, and so they w'ere diluted 10-fold ill their respective diluents immediately before being plated at the 13-week point. The survival of<l>Ea31-3 al1d <l>Ea46-1A2 was also tested in 0.015 M, 0.15 M, and 0.5 M NaCl. A 3.0 x 109 PFU/mL suspension of <l>Ea46-1A2 in nutrient broth was diluted 100-fold in each saline solution. The same was done with a 2.1 x 109 PFU/mL suspension of <l>Ea31-3. The saline suspensions were stored at 4°C. The concelltration of viable phage in each Inedium was determined after 24 h and after 11 weeks. Optimization ofPhage-Carrier Application to Blossoms The efficacy of phage-carrier mixtures was tested using the blossom assay described in Chapter 2. Treatments were pipetted directly on to the hypanthiunl ill 10 J..LL volumes. Treatnlents consisted of the bufTer control, the pathogen-free control, call·ier alone, or carrier-phage combinations. Each carrier-phage combination was prepared at two ditTerent nlultiplicities of infection, 0.5 or 0.05, with 5 x 10 7 CFU/mL of carrier. Three hours after treatment applicatioll, the blossoms were challenged by applying 10 J..LL of a 1 x 10 8 CFU/mL suspension of E. amylovora Ea6-4. The entire experimellt was repeated with no elapsed tinle betweell treatnlellt 181 application and pathogen application. Blossoms were incubated at room telnperature with high relative humidity. Disease symptoms were evaluated after 4 d. Statistical Analysis Regression analysis of exponential bacterial growth was conducted in Excel 2003 (Microsoft Corporation, Redmond, WA). All other data were analyzed using SAS (Statistical Analysis Systems 8.2; SAS Institute, Cary, NC). The stability of viable phages in each mediun1 was analyzed by regression of the logarithmically transformed concentration data on elapsed till1e (PROC GLM). The distribution of residuals showed that the assun1ption ofhomoscedasticity was not violated. Disease severity data from the blossom assays were converted to a percent scale based on the total hypanthium and ovary surface area showing visible necrosis. The severity data were then analyzed with the general linear model (PROC GLM) using the Tukey mean s.eparation test for multiple comparisons. 182 Results Bacterial Growth Curves E. amylovora and P. agglomerans spectrophotometric -calibration 'curves are given in Figure 5-1. There is a slight difference between the two curves, but in both cases an OD600 of 0.6 corresponds to 1 x 109 CFU/mL. Figure 5-2 shows the growth of E. amylovora Ea6-4 and P. agglomerans Eh21-5. For both species, the exponential growth phase -commenced approxilnately 3 h after the media was inoculated, and lasted for approximately 4 h. During exponential growth, the number of cells present at time t is defined as Nt = Noekt, where doubling time is equal to In(2)/k. Based on regression analysis of the exponential growth phase, the doubling tinles of E. amylovora Ea6-4 and P. agglomerans Eh21-5 were 44 and 48 nlinutes, respectively. Sinlilar results wel~ obtained in the second experiment, with exponential growth beginning approximately 3 h after inoculation, and ceasing 7 or 8 h after inoculation. However, the doubling tim·es of E. alnylovora Ea6-4 and E. amylovora Ea29-7 were longer, 130 min and 109 nlin, respectively (data not shown). Phage Growth Curves The results of the phage growth experiments are shown in Figure 5-3. Very Sill1ilar results were obtained in both trials. Cell concentration in the control cultures continued to increase throughout the course of the experiment, with an initial doubling time of about 90 mill. Cell growth reached a plateau after about 3 h, or 9 h after the culture was first ill0culated. By comparison, cell concentrations in the infected cultures were about 5-fold lower than in the 1.00 ~ o g 0.80 ~ 0.60 o o 4) u c: (U .c 0.40 :a-. o .c . . :.....•.. (I) 0 20' « 0.00 7 8 9 10 Cell Conee ntration ,tlogCAJ/ml) 1.00 ~ o oCD 0.80 o o~ 0.60 Q) u c: m 0.40 .c Lo ! « 0.20 0.00 7 1D C·ell Concentration (log Cr-'1J/ml) Figure 5-1. Calibration curves fo~· spectrophoton1'etric deternlinatiol1 of A) E. amylovora Ea6-4, and B) P. agglomerans Eh21-5 cell concentratiol1 in liquid culture. Regressiol1 of absorbance on cell concentration is described by y = 0.115 In(x) - 1.9D, R2 = '0.967 lor E. amylovora, al1d y = 0.197In(x) - 3.36, R 2 =0.974 for P. agglomerans. 184 ~o ~ 10 - r - - - - - - - - - - - - - - - - - - - - , A LL (.1 m o 9 ~. I:: ~ o r.s .j:: Iiu 8 I: o (.') "iii (.) 7 -+----....--,---.....----.....-,U - -__ ,--or------t o 2 4 6 8 12 10 Time (h) =:. 10 - - - - - - - - - - - - - - - - - - - . .~ ~ B ~ ~ U C') o ........, 9 ~ 1 -+~_r.,t o 2 4 6 8 10 12 Tilne (h) Figure 5-2. Growth of A) E. amylovora Ea6-4 and B) P. agglomerans Eh21-5 in liquid culture at 27°C. The exponential growth phase is described by y = (8.4 x 105)eo.938x, R 2 = 0.951 for E. amylovora and y = (3 x 106)eo.864, R 2 = 0.936 for P. agglomerans. 185 ~ ... ::>= u.. u 10 - r - - - - - - - - - - - - - - - - - - - - - - - - - - - - , • IJni nfected (1) A - - - 0- - - Infe cted t~1) IJni nfected (2) - - - tJ.- - - Infe cled • 9 C') o ,~) ~ ~ •• _ •• D _- -!:J.-_ ....•. . t,. .. _ •... • ···n···_·t,.· o u - -!:J.--- _-11- - - - -!::J.- ~ _- -- rl:!. £1 1 6-+-----------......----......----..-----..---------t o 2 3 5 6 1 4 Ti 111e Ela()sed Since Ph age Add ition f:l1) 10 - r - - - - - - - - - - - - - - - - - - - - - - - - - - - - , ,¢ - - - - _<>- - - - -<>- - - - -0 - - - - - <>- - - - -<>- - - __ <> - -<> B • • Infe c1ed (1) - - -<>- - - Infe cted (2) 6-+r,~_t o 1 2 .3 4 5 6 Tilne E1allsed Since Phage Addition (h) Figure 5-3. Growth of A) E. amylovora Ea29-7 and B) <j>Ea31-3 in synchronously infected and uninfected control cultures. Data from two independent trials are ShOWll (opell vs. closed POilltS). 186 uninfected culture. In trial 1, the overall :growth of the infected culture essentially Inin"ored that of the uninfected one, although at a lower concentration. In trial 2, the cell 'concentration in the infected culture did not change substantially ov,er the course of the exp,eriment, until showing a sharp rise and drop at the 4.5 - 5.0 h point. The total nunlber of infective centres, meaning free phages and infected cells, declined over the first 30 min after phage were added to the culture, and then rose sharply at 1.5 h. Another drop and spike in the concentration of infective oentres was observed after 4 h, and may indicate a second lysis event. Experiments conducted with <l>Ea35-4 and Eall0 at 23°C showed similar trends, with lysis beginning about 1.5 h after phage inoculation, and phage concentration reaching a maximum 2 h after phage inoculation (data not shown). Phage Survival in Liquid Media Phages may be exposed to different liquid media during the production and processing of lysates for field trials. Table 5-2 gives the rates of change in the ,concentration of viable <l>Ea46lA2 in six of these media over a 38 wk period. During the first 11 wk, the concelltration of viable phages declined significantly in allinedia except autoclaved reverse osmo"sis water. III nutrient broth, this initial decline appears to have OCCUlTed in the first 4 wk. Betweell 13 and 38 wk, there was no sigllificant decline in the concelltration of viable phages in either nutrient broth (P == 0.252) or autoclaved reverse osmosis water (P == 0.074). Phage concentration 'CoIltinued to decline in all other media. The total decline in absolute phage concentration, adjusted for the 10fold dilution at week 13, was no greater than 40-fold ill any nlediunl, and was within IO-fold tor all media except the two involving tap water. 1'87 The survival of cPEa31-3, a small, short-tailed phage and cPEa46-IA2, a large, long-tailed phage were also tested in solutions of different iOllic strengths. Table 5-3 gives the av·erage rates at which the viable concentration of these phages changed in each solution over an II week period. cPEa31-1 was stable in sodium chloride solutions ranging fronl 0.015 to 0.5 M. l~he concentration of viable cPEa46-IA2 was only stable in 0.015 M NaCI, and declined significantly in the two solutions with greater iOllic strength. Optimization ofPhage-Carrier Application to Blossoms Four phages were used in blossom assays to test the etTects of treatment tinlillg and phage:carrier ratio (Figure 5-4). Thi11een phage-carrier combinations significantly redu'ced the severity of fire blight symptoms (P < 0.05). When pathogen was applied immediately after treatment, all treatments significantly reduced disease severity relative to the control (P < 0.05), but the efficacy of phage-carrier combinatiolls was not significantly different from the ,efficacy of the carrier alone (P > 0.05). When 3 h elapsed b'etweell treatn1ent and pathogen application, the efficacy of phage-based treatments was greater than that of the carrier alolle (P < 0.05). 1-'he ratio of phage: carrier had no effect on disease severity (P > 0.05). 188 Table 5-2. Survival of <l>Ea46-1A2 in various liquid lnedia. Rate of Chan,ge in Viable Titre ( lo,glo(PFU/mL)/wk ) a Solution First 11 Weeks 13 to 38 Weeks Autoclaved Nutrient Broth - 0.033 * PB with salts - 0.048 * - 0.014 * Reverse Osmosis Water - 0.072 * - O.OID * Tap Water - 0.044 * - 0.038 * - 0.002 N on-autoclaved Reverse Osmosis Water - 0.00079 Tap Water - 0.038 * Rates of concentration change marked with an asterisk are significantly greater than zero (P an asterisk are not significantly greater than zero (P > 0.05). a - 0.004 - 0.005 * ~ 0.010). Rates without 189 Table 5-3. Effect of ionic strength on the survival ofQ>Ea31-3 and <l>Ea46-1A2. Rate of Change in Viable Titre ( loglO(PFU/mL)/d ) a Solution <pEa31-3 <pEa46-1A2 0.5 M NaCI 0.0028 -0.0042 * 0.15 M NaCI -0.0001 -0.0033 * 0.015 M NaCI 0.0042 0.0036 Rates of concentration change marked with an asterisk are significantly different from zero (P < 0.05). Rates of change without an asterisk are not significantly different from zero. a 190 (3 hr wait before challenge) ... 100 90 ...Vt 80 ...m 70 It' 60 C ..., 50 s:: G.t 40 e 30 fL. 20 10 0 (no wait before cha.llenge) t + + + f IlI I f * I f ~ . .. 1 ... ~ I * I" I * I. * I f f f I f f f c Q) 0') 0 :5 ro 0- I::' c OJ 0- L- Q) "E ..--.. ~ ..--.. ..--.. ~ ~ ~ ~L.l ro u ~ ~t ro W e m ro W e l.J.( ..--.. ~ ~ l.L'1 (T) ro W e- ..--.. ".,.-.., a C:!- o C:!- J~ ~ 1 to ~ ro w €t- ~ ~t ro W e 0 l.L'1 I Ci) ro W e ".-.. a ~ ~ ~ e ~ Q) C) 1 w ".,.-.., L- "E ~ ro l.L'1 LD 1:'"1'1 ro OJ 0- ~ I tD ~ ro W e 00:-- ~ ro W e- ".,.-.., .----. ~ l.L( ~ l.L( I:J') l.L-J ro W (I) ro W ".,.-.., ~ l'J ".,.-.., ~" ~ L1l l.L'1 m l:ti l.L'1 1:'"1"') ~ 1 00:-- ~ ro W ..--.. 0 I:J C:!- Ci Ci <l .:!J ".-.. a ~ ro W I W 0 I ro W 1 tr:, ~ ro W Treatm ent (MOl) Figure 5-4. Effect of phage-carrier combinations on disease severity in blossom assays, mean ± 95% confidence limits for three replications (10 blossoms per replication). Four phages were applied with the carrier using an MOl of 2, and then again using an MOl of 2·0. These eight phage treatments, along with the carrier and all untreated control, were repeated twice: Ollce, where 3 h elapsed between treatnlent and pathogen application (left-hand side), and once, where the pathogen was applied immediately after treatmellt (right-hand side). All phage and can"ier treatments except those marked with t resulted in significantly less severe disease than the PB control. Phage-carrier treatments marked with an asterisk were significantly better than the carrier alone. 191 Discussion Efficient propagation of phages and bacteria for in planta and field trials depends OIl their growth characteristics in liquid culture, at the approximate optimum temperature for E. amylovora growth. At 27°C, the growth of E. amylovora and P. agglomerans was quite silnilar. The exponential doubling time of E. amylovora Ea6-·4 was three times longer when the cultures were incubated in the water bath than when they were grown on the orbital shaker. Also, the doubling times of the two different E. amylovora strains ,grown usin,g the water bath incubation were much less similar to each other than were the doubling times of the two different species studied using the orbital shaker. These differences were likely the consequence of differential oxygen availability, since the agitation of the orbital shaker was constant and sufficient to break the surface tension of the culture, whereas the flasks in the water bath were only shaken vigorously when samples were taken. Since equipment availability dictated the use of the water bath shaker for the subsequent phage growth experiments, that procedure was modified to include regular manual agitation of the culture flasks. Regardless of the different rates of bacterial growth in each experilnent, exponential growth had begun within 4 h of inoculation. Many phages require expollentiallygrowin:g cells for their replication (Carlson, 2005), therefore phages being grown in large-scale liquid cultures for field use will be added to host cultures 4 h after the media is inoculated with the bacterial host. The rate and efficiency of phage adsorption can also be affected by changes in the expression of surface molecules due to the physiological state of the host (Guttman, Raya, & Kutter, 2005). Therefore aeration is likely to be critical for the efficient propagation of phages in liquid culture. The latent period of <l>Ea31-3 in Eall 0 is between 60 and 90 111in, which is lon:ger than 192 the 45 min latent period reported for the related phage, <pEal (Ritchie & Klos, 1979). The rang'e in the estimated duration of latency is due to the apparently slow rate of phage adsorption. Phage concentration peaked 1.5 h after the addition of phage to the bacterial culture, but in"eversible adsorption does not appear to have been complete until 30 min after phage were added to the culture. At time 0, immediately after addition of the phag'e, the concentration of bacteria in the infected culture was about 5-fold lower than the uninfected ,control. This suggests that approximately 80% of the cells were rapidly infected by at least 011e phage, and thus did 110t form colonies when plated on nutrient agar. Traditional Poisson models and empirical infection models would have predicted that an MOl of 10, with 1 x l08CFU/lnL of susceptible host 'cells, would result in infection of99.99% of those cells (Kaslnan et aI., 2002). However, some culture conditions al1d certain phages are prone to lower than expected infection efficiencies (Carlson, 2005), and certain phage-host combination,s'can sometimes yield few productive initial infections while producing a first generation of progeny phage with very high infectivity (Evans, 1940; Wollman & Stent, 1952). The additional manual agitation of these cultures compared to the previous cultures grown in the water bath did have an effect on bacterial growth. Though a differel1t strain 'of E. amylovora was used in the phage growth experiments, the initial doubling time ofcells in the uninfected Eal10 culture was shorter than it was for either of the stl"ail1S in the second bacterial growth curve experiment. As in the earlier experinlents, the growth rate of the ullinfected bacterial culture slowed within 3 h of phage addition, or about 9 h after the ,culture \vas initiated. Encapsulation within a polysaccharide matrix nlay be one way to preserv'e the phages for commercial.use. These matrices are gelled at a moderate salt concel1tration and dissolved at a 193 higher one. Therefore the survival of two phages with different morphologies was l110nitored ill solutions of varying ionic strength. The most dilute solution, 0.015 M, is similar to the ionic strength of the PB used throughout this research. The 0.15 M solution is standard physiological saline, and is similar to the concentration of sodiull1 chloride that is used in PB salt for Erwinia phages (Gill, 2000; Ravensdale, 2004; this work). The 0.5 M solutioll is the highest concentration that is expected to be used to dissolve a polysaccharide encapsulation matrix. The viability of <l>Ea31-3 was ullchan.ged after 11 wk in any of these 'Solutions, but the viable titre of <l>Ea46-1A2 declined significantly in 0.15 M and 0.5 M solutiollS. <l>Ea31-3 is a ll1ember of the Podoviridae, and <l>Ea46-1A2 is a member of the Myoviridae. Based on only two phages, it is llot clear whether their differential survival in saline is correlated with their morpl101ogies, but clearly some phages are affected by storage at high salt concentrations. Phages produced for use in the field will be grown in nutrie11t broth in order to nourish the host cells. However, the majority of the nutrient broth must be replaced with a nOll-nutritive buffer before use so as not to alnend the nutritional status of the blossom in a way that would favour pathogen growth. Diafiltration will be used to conc-entrate th'e phage sus,pen"sions and filter out small dissolved substances, but the replacen1ent buffer must be one in which the phages remain stable for an extended period of time. Other than nutrient broth, the only dilution n1edium in which the concentration of viable phage did not decline significantly was the autoclaved RO water. Inorganic chlorine in municipal tap water is known to decrease viral survival (Engelbrecht et aI., 1980; Berg, Sanjaghsaz, & Wangwongwatana, 1989), as is the presence of microflora (Ward, Knowlton, & Winston, 1986). The reason for phage inactivatioll in PB salt may be the concentration of sodium chloride. This buffer contained 0.1 M sodiun1chloride, which is only 194 slightly less concentrated than the 0.15 M solution in which the viable titre of <f>Ea46-IA2 also declined. Both solutions were prepared with sterile reverse osmosis water. These results are similar to results obtained for phages of Erwinia carotovora subsp. carotovora, where the concentration of viable phages was stable in fertilizer solutions made with autoclaved reverse osmosis water or PB saJt ' but was unstable in solutions nlade with non-sterile water (tap or reverse osmosis) or with sterile tap water (Rav,ensdale, 2004). Tail,ed E. ·coli phage f2 also does not survive well when suspended in non-autoclaved distilled water (Letler & Katt, 1974). However, the stability of <f>Ea46-1A2 in autoclaved reverse osmosis water and of <t>Ea313 in sodium chloride solutions was unexpected, since divalent metal cations are often requil~d to maintain viability (Adams, 1959; Gill, 2000; Ravensdale, 2004), and 110ne were added to these media. Either these phages are stable in their absence, or the ions wet:e present ill trace amOullts in the reverse osmosis water, as the lab supply of this water is not conlpletely deionized. The stability of <f>Ea46-1A2 in dilute sodium chloride also appeats to contradict th~e results of the tap water tests, in which the d.ecline of the <f>Ea46-1A2 populatioll was attributed to the presence of chlorine species. However, where the dissolution of sodium chloride salt releases chloride iOIIS, the dissolution of highly acidic chlorine gas or highly basic hypochlorite in nlunicipal drinking water produces mostly hypochlorous acid (HOCI) and hypochlorite (OCI-) at pH values greater than 3.0 (Edstrom, 2003). HOCI, in paliicular, is a strong oxidizing disinfectant. Once phages are produced and delivered to the blossonl, their efficacy depends on their interactions with the P. agglomerans carrier alld the E. amylovora pathogen. 1'he results of the blossom assay show that the carrier can reduce sylnptom severity on its own, but that the addition of certain phages can increase that etlicacy. The greatest efficacy was achieved when the phage- 195 carrier combination had time to become established on the blossom 'Surface prior to the arrival of the pathogen. This is consistent with field studies in which sup l~e ion on E. am}Jlovora populations was directly associated with early establishment of BlightBan A506 alld 'C9-1 (Johnson et aI., 1993b; Nuclo et aI., 1998). If early phage-host interactions 011 the blossom, while both are still suspended in the applied liquid droplets, resenlble phage growth in liquid culture, then the elapsed 3 h would have easily allowed one generatioll of progeny phage to be released from the initial carrier population. The uninfected carrier cells would have had a 3 h advantage over the arriving pathogen population in terms of adapting to the new nutritional environmellt and beginning to replicate. There was no significallt effect of MOl in this experinlent, so this should be <tested in field trials. Some effect of Mal is to be expected, since a very slllall population of phages lnay not replicate quickly enough to significantly impact pathogen growth, but the carrier populatioll lllay be rapidly lysed if the Mal is too high. A dynalllic equilibrium b-etween phage and -carrier is necessary in order to maintain the biopesticide on the blossoln sutface for several 'days. With this in mind, and since there was no clear effect of Mal in the blosSOlll assays, fi.eld trials will begin with an Mal of 1. Based on these experiments, the following parallleters will be used for field trials: wh,en phages are grown in large-scale liquid culture, phages will be added to the bacterialcultut::es 4 h after their inoculation, at the start of exponential growth; liquid cultures of phages must be v,elY well aerated; phage-carrier treatments will be applied well before the pathogen, so that they can become established on the blossoms; phage-carri'er cOlnbinatiollS will be lllixed at least 30 min prior to application, in order to ensure that infected earlier cells are applied to the blosSOlllS. 1'96 Chapter 6: Monitoring Erwinia phage populations in orchard soil Abstract In order to develop a biopesticide for practical use, the environmelltal fate of its components must be monitored in the orchard soil. However, the direct quantification of microorganisms that have been washed from soil is complicated by the presence of soil-derived PCR inhibitors. Five elution media were tested, all at pH 8.0: nutrient broth, nutrient broth with 0.1 % egg albumin, 250 mM glycine, 10 mM tetrasodium pyrophosphate, and 10 InM PB salt • All media were equally efficient at eluting phages from saluples of sandy soil, but glycine and nutrient broth permitted the most sensitive detection of phages by real-time PCR. In all cases, PCR detection was not possible immediately following the elution of phages from soil. The addition of 0.1 M EDTA, followed by diafiltration to remove the EDTA, was the most effective means of removing soil-derived PCR inhibitors from the phage-containing eluate. Even 600-fold dilution of soluble substances in the eluate was less effective than the combination of-chelation and diafiltration. This suggests that luetals were the principle nleans ofPCR inhibition. This method of removing peR inhibitors without DNA extraction is best suited for proeessin,g a small to moderate number of large samples. 197 Introduction A critical component ofbiopesticide registration requires accounting for the environmental fate of its components, ideally in a quantitative fashion. Traditiollally, biolo;gical agents are enriched prior to detection. If even one viable microbial cell or infective phage is present in a sample, it'should be detectable followin.g appropriate enrichnlent. In reality, the efficiency of enrichment can be greatly affected by the enrichment conditions. Viable bacterial cells can be rendered non-cultivable by certain environmental stresses (Wilson & Lindow, 2000), infective phages may be temporarily unavailable to host bacteria as a result of interactions with charged substances, and even the recovelY of available phages is affected by the type alld number of their specific hosts (Jensen et aI., 1998). However, the sensitivity afforded by enrichlnent comes at the cost of being able to quantify the initial population. Quantification requires that the target be detected without altering the populatioll size, either directly within the sample environment or after extracting the target organi'sm frOlTI that environnlent. DNA-based methods are generally the most sellsitive, but are often inhibited by substances in the sample environment. Chapters 2 and 7 of this work describe the 1110l1itoring of bacteria and phage on aerial plant tissues, but the phage-calTier biopesticide is also expected to be washed into the soil surrounding the treated trees as a result of rain. New techniques are therefore required in order to easily track the persistence of Erwinia phages in soil. The survival of Erwinia phages in soil has not been studied extensively. E. amylovora phages are easily isolated from soil beneath rosaceous hosts exhibitin:g signs of an active fire blight infection (Erskine, 1973; Schnabel & Jones, 2001; Gill et aI., 2003), but have not b'e-en recovered from soil beneath healthy trees, even using enrichment techniques (Ritchie & KIos, 198 1977; Schnabel & Jones, 2001; A. M. Svircev, personalconlmul1ication). Sitnilarly, free Bacillus phages were recovered from environmental soil salnples in very low nUlnbers until the population of endogenous bacterial hosts was il1creased by adding rich nledia and incubating the resultant slurry at 37°C (Tan & Reanney, 1976). In the presence of actively growing host eells, high concentrations of multiple phage strains were recovered. These results suggest that 1011gterm phage survival in soil is limited, but that a small number of phages may often persist, constituting a parent population that can quickly ilIcrease in nunlber under favourable conditiollS. Direct quantification of phages fi·om soil generally depends on microscopy or 011 specific serological detection such as ELISA (Shigeharu et aI., 2000; Willianlson, Radosevich, & Wommack, 2005). Unfortunately, microscopic nlethods are time-consunling, and serological methods are semi-quantitative and are less sensitive than DNA-based techl1iques. Ashelford et al (2003) used TEM to count phages in filtered soil suspensions, alld repo11ed populations at least 350-fold higher than those estimated from viable plaqu·e COUlltS. However, the ;ecological significance of these extra phage is questiollable since sonle of these samples did 110t yield allY viable phages even after enrichment on nlultiple hosts (Ashelford, Day, & Fry, 2003). Recovery of those limited numbers of surviving phages fronl soil is also hindered by adsorption of phages to soil particles (Bitton, 1975; Burge & Enkiri, 1978). Adsorptioll is mediated by pH-dependent electrostatic interactions (Burge & Enkiri, 1978; Taylor, Moore, & Sturman, 1981; Dowd et aI., 1998) alld seems to be greater for phages with longer tails (Ashelford, Day, & Fry, 2003; Williamson, Wommack, & Radosevich, 2003). Sev,eral detailed studies of elution techniques have been published (Lanning & Willialns, 19,82; Hu, 1998; Williamson, Wommack, & Radosevich, 2003), but such studies often lack the type of truly 199 systematic approach that is l1eeded to ascertain which factors influence the success or failure of a particular method. Hu (1998) recovered more coliphages with glycine buffer than with beef extract, but the difference may have due to differences in the sanlple treatment associated with each medium, rather than the innate characteristics of the two nledia. He did find that phage recovery improved with increasing COlltact time between 30 and 90 min. WillialTISOll, Womnlack, and Radosevich (2003) tested the recovery of five different phages frol11 two soil types using 10% beef extract, 1% potassium citrate, 10 lllM sodium pyrophosphate, and 250 InM glycine. Phage recovery was greatest with beef extract and glycine, but recovery varied sigllificantly between phage and soil types. Additionally, the beef extract and glycine were the only llledia prepared to a non-neutral pH (pH 9.0 and 8.0, respectively, cOlllpared to pH 7.0), \vhich could be the real reason for their success. The most useful study of phage elution was conducted by Lanning and Williallls (1982). In order to recover actinophages frOln multiple soil types they compared the efficacy of nutrient broth, 3% beef extract, 1 M glycine, 1 M sodiulll nitrate, peptone/yeast/calcium broth, and glucose/Casamino acid/proline broth, all at the sanle pH. After obtaining th,e best results with nutrient broth they then compared recovery by 11utrient broth at different pH levels, and with different types of exogenous protein. Recovery was n10st efficient using nutrient broth at a pH of 8.0, and only egg albumin improved phage recovery. Despite the variation in l11ethodology among these studies, three common themes emerge: the efficacy of any given treatnlel1t varies with soil type, is lower for phages with long tails, and tends to be better for ,eluants at slightly basic pHs. 200 Successful elution methods do not necessarily equate to successful qual1tification since soil contains many substances that are known to inhibit PCR, and that tend to be washed from the soil along with the recovered phages. As little as lOng humic acid can inhibit a conventional, endpoint PCR reaction (Tsai & Olsen, 1992). It has been postulated that phellolic nloities in humic substances react with, and covalently bind to, DNA alld protein, preventillg the neoessary interactions between the polymerase and the target DNA, or between primers and target DNA (Young et aI., 1993). Humic acids nlay also interfere with the fluorescence processes upon which real-time PCR depends by quenching fluorescence of SYBR Green, Hoescht 33258, and PicoGreen COlllplexed to DNA (Bachoon, Otero, & Hodson, 2001; Zipper ,et aI., 2003). In the case of SYBR Green, which has excitation and emissioll wavelengths in the same ran,ge as the dyes used for probe-based real-tilne peR, this is due to multiple factors: absorption of radiation at the fluorophore excitation and emission wavelengths by hUlllic acid; diffusive collisions between humic acid and the fluorophore; and the formation of stable cOlnplexes between certain humic acids and both bound and unbound dye (Zipper et aI., 2003). A variety of metal ions can inhibit PCR, apparently by interfering with the binding and activity of the polymerase enzyme. Calcium ions in milk can interfere with PCR anlplification (Bickley et aI., 1996). Iron and other heavy metals are ,gen·erally present in soil, wheth·er in high levels as pollutants, or levels appropriate for lllicronutrition of plants, and are known PCR inhibitors (Wilson, 1997a; Ogram, 1998; Hao, Dick, & Tuovinen, 2002). Any substances that sequester Mg2+ will also inhibit the polYlnerase enzyme, which requires the iO]l as a cofactor and is sensitive to changes in its concentration (Satsangi et aI., 1994; Wilson, 1997a). Polyalllines (Ahokas & Erkkila, 1993), phenol (Katcher & Schwartz, 1994), and plallt polysarecharides 201 (Demeke & Adanls, 1992) can also inhibit amplification by directly affecting the DNA polymerase. Foulds et al (2002) were able to remove PCR inhibitors by washing E. coli cells collected from environmental water samples with EDTA, a lnetal iOll chelator, prior to DNA extraction. Extensive work has also been done to develop nlethods of removing these inhibitors in the course of extracting total community DNA from soil or soil eluates (Zhou, Bruns, & Tiedje, 1996; Sj6stedt et aI., 1997; Miller et aI., 1999; Desai & Madamwar, 2006). The ainl of this work is to develop a rapid, simple method of monitorillg the persistence of Erwinia phages in the soils of orchards where the phage-based biopesticide has been used. Unlike most other studies, soil samples were not homogenized, dried, alld sieved prior to use. It is unlikely that actual field samples would receive this kind of treatnlent, since it nlight damage the phages trying to be recovered. Instead, large clumps of soil were broken up when the soil samples were weighed out, and when the flasks were shaken to lllix the soil with the phage inoculum. Five elution media were tested in an attelnpt to optimize the recovery of phage particles from soil samples. These media were chosen from those that produced the b'est results in previous studies (Lanning and Williams, 1982; Hu, 1998; Willialnson, WOln111ack, & Radosevich 2003). All elution lnedia were adjusted to a pH of 8.0, since those studies also indicate that buffers having a higher pH are more likely to disrupt the electrostatic interactions between phage and soil particles. The elution methodology is based 011 the results of Lanning and Williams (1982) as they conducted the most extensive assessnlent of the effects of contact tilne and motion on phage recovery. Several methods of removing soil-derived PCR inhibitors from the eluate were tested, in particular the use of a pre-amplification EDTA treatm·ent.Since EDTA also chelates Mg2+, it is, 202 itself, a peR inhibitor. EDTA was therefore removed by diafiltration, a process that allows undesirable small solutes to be removed from a suspension nledium during pressure-driven filtration, without diluting the larger substance of interest (in this case, the phage). Durin.g diatiltration, the dilution effects of successive additions of clean media are multiplied, allowing high dilution rates to be achieved with comparatively small volumes of fresh nledia. 2'03 Methods Growth Media and Strains Two bacteria and one bacteriophage were used in this work: E. amylovora Ea6-4, P.. agglomerans Eh21-5, and E. amylovora phage <I>Ea45-1 B. All strain origins are described in Tables 2-1 and 2-2. <I>Ea45-1B was grown on E. amylovora Ea29-7 in overnight liquid 'cultures using 8 giL Difco nutrient broth. Lysates were treated with 2% (v/v) chloroforn1 for 30 11lin, centrifuged at 8 500 xg for 25 nlin, and syrillge- filtered USillg 25 mm diameter, 0.2 ~ln, surfactant-free cellulose acetate filters (Nalgene, Rochester, NY). Filtered lysates were stored in nutrient broth at 4°C. Bacteria were grown overnight at 28°C on 23 giL Difco nutrient agar. Soil Sample Preparation and Analyses Several 30 cm x 2 cm soil cores were collected from the root zones of Bartlett pear trees in experimental orchards at the Agriculture and Agri-Food Canada (AAFC) research farm in Delhi, Ontario. Pooled soil cores were stored at 4°C between collection and use. Soil analyses had been previously conducted by A & L Laboratories (London, ON) in 2004 and 2005. The results of these analyses were obtained from Barry Kemp, the Farm Manager for AAFC-Vinelalld (encompassing both the Jordan and Delhi famls). Experimental soil samples consisted of 10 g (wet weight) of soil in 250 mL capped Erlenmeyer flasks. Samples were sterilized by autoclaving the flasks for 20 mill at 121°C, under steam pressure. Soil samples wer~ spiked with 1 mL of 1 x 10 8 PFU/tnL of <I>Ea4S-1 B, and mixed by shaking. Negative controls were inoculated with I mL of sterile nutrient broth. Inoculated soil samples were stored overnight at 4°C. 204 Phage Elution Five solutions were used to elute the phages: Ilutrient broth (pH 8.0), nutrient broth (pH 8.0) containing 0.1 % (w/v) grade II egg albumin, 250 mM glycine (pH 8.0), 10 mM tetrasodium pyrophosphate (pH 8.0), and 10 mM Pbsalt (pH 8.0). To elute the phages from the soil sample, 5 InL of solution was added to the flask and incubated for 25 min at 4°C. The flask was then placed on an orbital shaker at 200 rpm for 30 min at 4°C. The entire contents of the flask was transferred to 50 mL round-bottom centrifug·e tube and centrifuged at 8 000 xg for 10 min at 4°C. The supernatant was decanted into a sterile tube for storage at 4°C. The sedimented soil was resuspended in 10 mL of the sanle solution al1d returned to the orbital shaker for another 30 min, then decanted and centrifuged as before. The first and second supenlatants were combilled and were stored at 4°C. This procedure was conducted with each solution, for both an inoculated soil sample and an uninoculated, phagefree, control sample. The titre of each eluate was determined using the soft agar overlay nlethod described in Chapter 2. The entire process was repeated twice at a later date, with the following change: only the supernatant from the first elution was transferred from the Erlellmeyer flask to the celltrifug'e tubes. The remaining solids were returned to the shaker with the second volulne of elution nledia while the first batch was being centrifuged. Again, the first and second eluates from each flask were combined and stored at 4°C. Data were analyzed with SASv8.2 (SAS Institute, Cary, NC) using the general linear model (PROC GLM), with the Tukey adjustnlent for nlultiple comparIsons. 2DS Real-time peR TaqMan-style real-time PCR reactions were conducted using prilners and probes developed by Dr. W. -S. Kim (unpublished). Each reaction was conducted in a total volunle of25 ~L, and contained IX Brilliant QPCR Master Mix (Stratagene, La Jolla, CA), 200 primer, 100 M~ ~M of each probe. Reactions were run in a Stratagene Mx-4000 Multiplex Quantitative PCR system (Stratagene) under the following conditions: 95°C for 10 mill; 40 cycles of 95°C for 30 s and 60°C for 60 s, with three endpoint fluorescellce readings during each amplificatioll segnlent. A standard curve was constructed to allow quantification of <I>Ea45-1B in soil eluates. Templates for the standard curve were prepared by serially diluting the same phage suspension used for soil inoculations in nutrient broth. Five microliters of each dilution was used as the reaction template. PCR reactions were run in duplicate The efficiency of phage elution from inoculated soil samples was tested usin:g 5 ~L of each eluate as a template. Fresh media of each type, containing 3 x 108 PFU/mL of phage, were used as positive controls. Sterile reverse osmosis water was used as a negative control. To confirm the presence of soil-derived PCR inhibitors in the eluted phage suspensiolls, reactions were conducted using the following as telnplates: 1 ~L nutrient broth (pH 8.0), 1 ~L through sterile soil, 5 ~L of a phage suspensioll in fresh of a phage suspension in nutrient broth that had been passed of a phage suspension in nutrient broth passed through soil, or sterile water. Phage suspensions contained 4.4 x 10 7 PFU/mL. Identical reactions were also run using E. amylovora Ea6-4 and P. agglomerans Eh21-5 ill the saille nledia. All amplifications were conducted in 25 L~ volulnes and were run in duplicate. Information on the primers and probes is given in Table 6-1. Phage detection was based on 206 release of the 6-carboxyfluorescein (FAM) fluorescent reporter fronl the 5' end of the hydrolized probe, P. agglomerans detection was based on release of hexachlorofluorescein (HEX), alld E. amylovora detection was based on release ofCy5. Probes were synthesized by Integrated DNA Technologies (Coralville lA, USA), and were labelled at the 3' elld with -either Black Hole Quencher 1 (BHQ-l) or Iowa Black (IAbRQ). Removal ofpeR Inhibitors Three treatments were tested in an effort to renlove soil-derived PCR inhibitors fi·onl the phage eluted with nutrient broth at pH 8.0: centrifugation, DNA extra'etion, and chelatioll. The centrifugation and chelation treatnlents were conducted using the same initial numbers of phage. For centrifugation,S mL of eluate was added to 25 nlL of nutriellt broth ill a 50 mL FEP centrifugation tube, and centrifuged at 16 000 xg for 45 min, at 4°C. eh~1 additional volume of nutrient broth was required to meet the manufacturer's guidelines for high-sp'eed use of the centrifuge tubes. Following centrifugation, the supernatant was itnnlediately decanted, alld the sedimented material was resuspended in 500 nlL of clean nutrient broth, pH 8.0. Both fractions were stored at 4°C. The concentration of viable phages in each fraction was dete11llilled using the soft agar overlay method described in Chapter 2. For chelation, EDTA was added to 5 InL of eluate for a final EDTA concentration of 0.1 M, and the mixture was incubated at roonl temperature for 20 min. One hundred fifty milliliters of clean nutrient broth, pH 8.0, was added in order to dilute the EDTA. The Volullle of the suspension was reduced to approximately 20 tnL by by nortnal flow diafiltration using an Amicon Model 8400 stirred cell apparatus (Millipore, Billerica, MA) with a 1,0,0 kDa YM-type 207 regenerated cellulose membrane filters. In order to further reduce the concentration of EDTA in the retained phage suspension, an additional 100 mL of clean nutrient broth was add,ed to the phage suspension, and the retentate volume was again reduced to approxinlately 20 InL. The Amicon apparatus was assembled according to manufacturer's instructions. Preservatives were removed from new filters by floating them skin-side down for at least one hour, with three changes of water. Approxinlately 400 mL of distilled water was flushed through the systenl prior to each run. Phage suspensions were cOllcentrated under 50 psi of nitrogen pressure until the retentate was reduced to the desired volume. Residual phages adhering to the filtration menlbralle were collected by floating the membrane upside down in a sterile petri dish containing a small volume of nutrient broth, and shakillg the dish gently. Two such washes were added to the retained phage suspension. Both the retentate and filtrate were stored at 4°C. The concelltration of viable phages in each fraction was determined using the soft agar overlay 111eth·od d:escribed ill Chapter 2. Filters were used no more than three times each. In between uses, filters w·ere sterilized by floating them in 70% ethanol for 20 min, then in 0.1 % TergAZyme for 30 lnin. Sterilized filters were flushed with distilled water and stored in 10% ethanol at 4°C. A one-step DNA extraction was also tested. Five hundred microliters of 24: 1 chloroform:isoamyl alcohol was added to an equal volume of eluate, and lnixed by ~gentl inversion for 5 min. The emulsion was centrifuged at 13 000 xg for 5 nlin. The upper, aqueous layer was removed with a pipet and transferred to a sterile microcentrifuge tube. DNA was precipitated from a 200 J.1L aliquot of the crude extract by adding sodium ac'etate ,to a final concentration of 0.3 M, and an equal volume of 95% ethanol. The precipitated DNA \vas collected by centrifugation at 13 000 xg for 5 min, washed with 70% ethanol, and allowed to 208 airdry before being reuspended in 10 mM Tris-HCl. Both the pre'cipitated DNA and the remaining crude extract were stored at 4°C. Endpoint peR Amplification of phage DNA from the treated ·eluates was atteillpted using the Q>-dpo 1 primers (see Table 6-1). Amplification reactions were conducted in 25 flL volumes. Each reaction contained 400 flM of each primer, 200 flM each of dATP, dCTP, dGTP, and dTTP, 1.5 mM MgCI 2 , 1.25 U Taq (MBI Fermentas), IX polymerase buffer, and 2 flL of phage suspension in nutrient broth. Reactions were run in an Applied Biosystenls Genei\mp 9700 thermal cycler under the following conditions: 94°C for 5 min; 30 cycles of 94°C for 15 s, 60°C for 15 s, and 72°C for 30 s; 94°C for 10 min. Reaction products were \lisualized using agarose gel electrophoresis, as described in Chapter 2. Optimization ofEDTA Treatment The relative contributions of chelation and solute dilution to the success of the EDTAbased treatment were determined by repeating that treatlnent with -celiain modifications. First, the process was repeated exactly, but using a final concentration of 0.01 M EDTA instead of 0.1 M. Second, the original treatment was repeated using sterile water in lieu of the EDTA solutioll. Third, five times the initial amount of eluate was used, again with no addition of EDTA. Finally, the original treatment was repeated using water in lieu of EDTA, but this tilne using twice the volume of diluent at each stage of diafiltration, such that 300 nlL was reduced to 20 nlL, alld then diluted to 200 mL and again reduced to 20 mL. 209 Table 6-1. peR primers (F or R) and probes (P) for Erwinia phages, E. amylovora, and P. agglomerans. Oligonucleotide Target Species Target Gene Amplicon Size (bp) <l>Ea45-1 B depolymerase (dpo) 171 E. an1ylovora levansucrase (lse) 105 P. agglomerans gluconate-6-dehydrogenase (gnd) 73 a <I>-dpo 1F <I>-dpo 1R <I>-dpo I P Ea-IscF Ea-lscR Ea-IscP Pa-gndF Pa-gndR Pa-gndP Design was based on the following sequences deposited in the NCB I nucleotide database: AJ278164 (dpo), X75079 (lse), and AF208633 (gnd). a 210 Results Chemical Characteristics ofOrchard Soils Tables 6-2 and 6-3 show the results of soil analyses conducted on theexperinlental apple and pear orchards at the AAFC-Delhi site, and the AAFC-Jordan site. The Jordan site is located in the Niagara fruit growing region and is much more typical of commercial orchards than the Delhi site. The Jordan soil contains nlore organic matter and therefore has a higher pH and cation exchange capacity (CEC). This greater ability to hold and supply cations is reflected in the higher levels of magnesium and calcium in the Jordan soil. The Delhi alld Jordall soils appear to COlltain similar concentrations of iron and bicarbonate phosphorus, but the Delhi soilcolltains less ZillC and copper, and more manganese. Elution ojPhages and the Presence ojSoil-derived PCR Inhibitors The recovery of phage from soil was assessed in two ways: viable titre, and real-time PCR. Table 6-4 shows the percentage of the original phage inoculum that was recov'ered USillg each type of nledia. The relative efficiency of these media differed in each trial, but there was no significant difference in average recovery among the five media (P = 0.850). Figure 6-1 is the amplification plot from the real-time PCR assessnlent of phage recovery. Positive controls consisted of phage suspended in elution media that had not come in to ,contact with soil. Fluorescence signals w'ere obtained fronl four out of five positive 'controls, but 110t froln any of the soil eluates, suggesting the presence of a soil-derived PCR il1hibitor ill the experimental 211 Table 6-2. Chemical characteristics of orchard soil at the sampling sites on the AAFC-Delhi Research Farm. Data are from 2004. Plot # pH CEC a (meq/ 100 g) Elemental Concentration (ppm) Mg Ca AI Fe P (bicarb) b Mn Cu 80 Zn 66 VH 937 64 VH 846 58 VH 9 6.9 3.8 60 M 350 L 1080 11 6.8 5.2 90 M 590 M 22 6.6 4.6 75 M 500 L 98 VH 0.8 M VH 3.8 M Cation exchange capacity Superscripts indicate whether the micronutrient level is low (L), medium (M), high (H), or very high (VH), from the perspective of soil fertility. a b Table 6-3. Chemical characteristics of typical orchard soil on the AAFC-Jordan Research Farm. Data are from 2005. Plot # pH CEC a (meq/ 100 g) Elemental Concentration (ppm) Mg Ca AI Fe Cu P (bicarb) b Mn Zn 2B 7.1 10.3 230 H 1370 H 88 VH 3.9 VH 47 M 10.4 13 7.2 7.8 195 H 1020 H 99 VH 6.6 VH 30 M 5.6 M 21 7.3 8.9 140 H 1420 H 107 3.8 VH 33 M 8.8 H VH H 56 40 59 VH H VH Cation exchange capacity Superscripts indicate whether the micronutrient level is low (L), medium (M), high (H), or very high (VH), froITI the perspective of soil fertility. a b 212 Table 6-4. Effect of elution medium on recovery of the original 1 x 10 8 PFU. Data for trials 2 and 3 are the mean of three subsatnples taketl fi·om each eluate. Percent Recovery of Original Phage Inoculum Medium Trial 1 Trial 2 Trial 3 Mean ± SD Nutrient broth (pH 8.0) 31.5 37.1 43.4 37.3 ± 5.9 Nutrient broth (pH 8.0) + 0.1 % (w/v) egg albumin 23.9 42.1 44.7 36.9 ± 11.4 A 10 mM PB salt (pH 8.0) 16.4 46.2 30.9 31.1 ± 14.9 A 250 mM glycine (pH 8.0) 24.8 54.5 38.6 39.3 ± 14.9 A 24 39.5 30.1 31.2 ± 7.8 10 mM tetrasodium pyrophosphate (pH 8.0) a Means with the same letter are not significantly different. a A A 213 2200 2000 1800 1600 1400 C2 1:1 ~ 1200 c ~ ~ o 1000 :J Li: 800 600 400 200 4 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 Cycles Figure 6-1. Effect of elution medium and passage through soil on the amplification of phage. Positive signals are from glycine (blue), nutrient broth (green), PB salt (gold), and nutrient broth with albumin (grey) positive controls. 214 samples. FUliheml0re, each positive control contain,ed 3 x 105 PFU per 25 L~ reaction, yet detection occurred earlier in some media than with others. Glycine and nutrient broth permitted the most sensitive detection, followed by PB salt and albumin-alnellded nutriel1t broth after four or more additional amplification cycles. Phages suspended in tetrasodium pyrophosphate w'ere not detected at all. Nutrient broth (pH 8.0) was chosen as the elution medium of choice for the remainder of this work. Glycine also performed very well in terms of elution efficiency alld compatibility witll PCR detection, but nutrient broth gave slightly more efficient elution ill two of the three tlials (Figure 6-1) and previous work has demonstrated that this phage remains viable in nutriellt broth for long periods of time (see Chapter 2 and Chapter 5). The presence of soil-derived PCR inhibitors was ,confinlled by attemptillg to anlplify the phage and two bacterial species, E. amylovora and P. agglomerans, from nutrient broth that had been passed through sterile soil according to the saIne procedure used to elute the phages. No fluorescence signals were obtained from any of the targets when th·ey were susp,end,ed in the buffer that had been passed through soil, regardless of whether 1 ~L or 5 ~L of the teIllplate suspension was used. In contrast, signals were obtained from each target when it was suspended in clean nutrient broth. Removal ofpeR Inhibitors Three methods of removing the soil-derived PCR inhibitors from the eluate were teste,d. The eluted phage suspension was centrifuged to determine whether the inhibitory substan,c-es were associated with particulate matter in the suspension. A single-step organic extraction \vas 215 also performed in an attelllpt to obtain inhibitor-free DNA, and both the crude extract and the precipitated DNA were tested. In the third treatment, EDTA was added to tIle phage suspension in order to chelate metal ions. Since EDTA 'can also interfere with PCR, the EDTA was relTIoved by diafiltration. The distribution of viable phages in anlong the resultillg fractiollS is shown in Table 6-5. Contrifugation allowed the collection of a more concentrated phage suspensioll, but only half of the phages present in the original salllple remained viable, in either the supenlatant or the resuspended pellet. Of the remaining viable phages, only slightly more than halfwere present in the sedimented fraction. Better recovery of viable phages was obtained with chelation and diafiltration. Fewer than 18% of viable phages were lost over the course of the treatment, with only 2.8% of the remaining phage being lost to the filtrate. The efficacy of each treatment was evaluated by attempting to alllplify a 171 bp segtl1ent of the phage depolymerase gene. Figure 6-2 shows the results of these aillplifi·cation reactions. As expected, phage DNA was amplified from clean nutrient broth spiked with 3.7 x 10 7 PFU/lllL of phage, but not frolll nutrient broth that had been passed through soil. Substantial specific amplification was also achieved frolll the EDTA-treated salllple, which contained 1.9 x 1'0 5 PFU/mL. Very little amplification was obtained from the centrifugation supernatant, even though that fraction contained 8.8 x 104 PFU/mL, which was only slightly less than the EDTl\-treated sample. No phage DNA was amplified frolll any of the other treated f1 actions. 4 The relative contributions of chelation and dilution to the removal of PCR inhibitors were assessed by altering the amount of EDTA added and the degree of dilution, measuring the apparent phage concelltration using real-time PCR, and cOlllparing the efficiency of each 216 g - - en u<D C u .Q ~ Q. .0 o o u ~ C Q -0 cO _Q 5Q~ Q. ~ ';=::J C(/) c (]) u ~ o ::J gu .Q ~ xw c 8 QZ ~ c c -- Oen C :t-Q.~ -E ngo..~ + +=0 0 (]) 0-- X w 0 ~ w u~ ~ Ow (J)o.. Z <D Q co o o :I u u .Q Q. .0 o o 1 KB"" 800bp 600bp-" 400bp 300bp 200bp-lOO/80bp" Figure 6-2. Amplification of phage DNA following treatment of soil eluate. Far left and right lanes contain GeneMark 100 bp DNA ladder (MBI Fermentas). The number of target gene copies present in each 25 L~ reaction was as follows: 1.8 x 102, centrifugation supernatant; 1.3 x 104, centrifugation pellet; 4.2 x 105, crude organic extract; 4.2 x 105, precipitated, extracted DNA; 3.9 x 102, diafiltration retentate; 2, diafiltration filtrate; 7.4 x 104, positive control; 1.1 x 105 , negative (inhibition) control; 0, water control; these values are based on the viable titre of each fraction (Table 6-5), or, in the case of the organic extraction fractions, are inferred from the viable titre of the original soil eluate. 217 Table 6-5. Recovery of viable phages by centrifugation and diafiltration. Treatment Centrifugation Chelation & Diafiltration ~tO of Original Phages Viable Titre (PFU/mL) Fraction Volulne (mL) Total PFU Present Supernatant 8.77 x 10 4 30 2.63xl0 6 25.1 Pellet 6.36 x 10 6 0.5 3.18 x 10 6 30.3 Retentate 1.94 x 10 5 43 8.36 x 10 6 79.6 Filtrate 1.02x 10 3 238 2.40 x 10 5 2.8 Fraction Present Lost 44.6 17.6 218 treatment. The quantification of <t>Ea45-1 Busing real-ti1ne peR was 'calibrated using serial dilutions containing a known concentration of viable phages. The standard curve is ShOW11 ill Figure 6-3. The relationship between threshold cycle (C T) alld initial quantity was 'linear across the tested 4-log range. The lower limit of phage detection was not explored in detail, but would not be expected to be much lower than the 80 PFU per 25 J..LL reaction that was observed here, since this is on the same order of magnitude as the 20 CFU per reaction threshold d,escribed in Chapter 4. Table 6-6 shows the relative efficiency of the different EDTA and diafiltration treat1nents. In the original treatment (No.1), EDTA was added to an aliquot of the soil eluate to a final concentration of 0.1 M. By diluting that initial 5 mL san1ple to 150 n1I.J and reducing the volu1ne of retentate to 20 nlL, the total concentration of dissolved EDTA would have bee1l Feduced by 3D-fold. The subsequent dilution of that retentate to 100 nlL resulted ill a further 5-fold dilution, for a 150-fold dilution of solutes in total. The concentration of viable phage ill the filtrate revealed a 3% loss across the 100 kD membrane. The expected concentrations of phage were therefore based on the concentration of the original sample and the VO]U1ne used, less 3%. U1lder this expectation, treatment 1 removed 1110St of the soil-deriv'ed PCR i11hibitors, such that th'e estimate of phage concentration based on real-time PCR was 8,4% of the theoretical total. 1~h'e same dilution rate, with a lower concelltration of EDTA (treatment 2) was not nearly as effective. In the absence of EDTA, even very high dilution rates were less eftective ·(treatnlents 3 and 5). Dilution alone was c01npletely ineffective when the dilution rate was only a fifth ofth,e original treatment (treatment 4). 21'9 40 ~ ~ 30 Q) C3 ~ (J "'C 20 (5 .J: en Q) L. .J: t- 10 O 10° 101 102 10 3 104 105 106 Initial Quantity (PFU) Figure 6-3. Standard curve for the real-time PCR amplification of <t>Ea45-1B using the <t>-dpol primers and probe. Initial quantities refer to the number of phages present in the 25 IlI-, PCR reaction. Quantification is given by the equation y == 34.7 - 3.11 log(x); r 2 == 0.986. 220 Table 6-6. Relative contributions of chelation and dilution to the removal ofsoil-'derived peR inhibitors. Volume of Eluate EDTA Dilution of Solutes Phage Concentration According to realtime peR (PFU/mL) 5 mL 0.1 M 150-fold 1.99 x 10 5 2.37x 10 5 2 5 mL 0.01 M 150-fold 2.70 x 10 4 1.16 x 10 5 3 5 mL 150-fold 1.93 x 10 4 1.33 x 10 5 4 25 mL 30-fold 0 6.98 x ] 0 5 5 5 mL 600-fold 4.01xI0 4 1.18 x 10 5 Treatment No. Expected Concentration (PFU/mL) Efficiency of Treatment 84.00/0 23.~ 15.00/0 O~ 35.20/0 22] Discussion There are many published studies describing DNAextractiol1 techniqu,es that can be used, with varying success, to remove soil-derived PCR inhibitors. The SDS-based nlethod of Zhou, Bruns, & Tiedje (1996) was used successfully with several soil types, including sandy loams. The Ultraclean Soil DNA Isolation kit from MO BIO Laboratories has been used to extract total bacterial DNA for real-time peR detection of E. coli 0157:H7 (Ibekw,e et aI., 2002), though Desai and Madamwar (2006) describe a protocol that removes metallic and organic inhibitors more efficiently. Few studies bother to mention attempts to anlplify microbial DN~t\ directly froll1 soil eluate, and those that do report consistent failure (Sjostedt et aI., 1997). Here, a Inethod of removing soil-derived PCR inhibitors from soil eluates without DNA extraction is described. There was no significant difference in the efficacy of the five tested eluellts. Proteinaceous substances, and egg albumin in particular, have been found to aid the release of phages from soil (Lanning & Williams, 1982; Hurst et aI., 1991), but no such ilnprovenlent was observed here. Lanning & Williams (1982) did observe dit1erences between the perfolmance of different eluents at a given pH, but that was at pH 7.0, which was not the optimuln value, at least for nutrient broth. The performance of the different eluents were not compal~ed at the higher pH. Since adsorption is mediated by electrostatic processes (Taylor, M:oore, & Stuffilan, 19'81; Dowd et aI., 1998), it is possible that there is no detectable diffeFence between ,elution Inedia at the optimum pH, but that at non-optinlal pH levels, the specific types of charged sp'ecies in each medium are more or less successful in disrupting the interactions between phages and soil particles. Since the primary goal of this work was to find a way to deal with PCR inhibitors, rather than to fully optimize the elution process, the elution nledia were only cDll1pared at pH 8.0. 222 The most impoliant difference between the different eluents was their influence on the sensitivity of real-time PCR. When using fresh media, glycine offefed the least illterferen'oe with amplification and fluorescence detection. Phage detection in nutriellt broth was only slightly less sensitive than in glycine. In a rich, undetilled medium such as nutrient broth, any number of inhibitory substances could be present in trace amounts that would account for this difference. The PCR reaction will tolerate the presence of a certain amount of protein, as evidellced by the sensitivity of methods that use whole or boiled cells as telnplates (Stal11bach, Falkow, & Tomkins, 1989; Salm & Geider, 2004; Chapter 4). Nev,ertheless, abundallt proteill can substantially reduce the sensitivity of PCR, presumably by non-specific blocking of DNA molecules (Rijpens et aI., 1996). Thus it is not surprising that the presence of albumin reduced the sensitivity of real-time PCR. However, this inhibitioll is not a concen1 sillce albulllin did not enhance the efficiency of phage elution. Presumably, the reduced sensitivity of phage detectioll in PB saJt was due to the high salt concentration, even though the addition of 5 ~L of the buffer would only have increased the magnesium concentration in the reaction by 0.2 mM. The tetrasodium pyrophosphate buffer contained much less sodium than the PB saJt ' but this compound is comlll0nly used as a water softener, and likely inhibits PCR by sequestering Mg2+ ions. However, this ilnplies that tetrasodium pyrophosphate would also help chelate soil-derived PCR inhibitors. It has been used by soil scientists to dissolve organic matter and extract metals bound to hUlllic substallces (McKeague, Brydon, & Miles, 1971; Manninen et aI., 199,6). It would be interestil1g to see whether phage elution with tetrasodium pyrophosphate, followed directly by diafiltratioll with a medium like nutrient broth, would renlove soil-derived peR inhibitors as etTectiv,ely as the 223 EDTA-based treatment described here. Regardless of the eluent used, soil-derived inhibitors prevented the direct quantificatioll of both phages and bacteria in the soil eluate using real-tilDe PCR. Several methods of eliminating these inhibitors were tested. The most successful treatment was clearly the combination of chelation and diafiltration, which suggests that metal ions are the 111ain cause of PCR inhibition in these eluates. Calcium, iron, nlagnesiull1, copper, nlangan'ese, and ,zinc are present in appreciable amounts in the soil.from which these samples were takell, as they are in most soils. All of these metals can exist as divalent cations, which are known to affect the efficiency ofPCR (Satsangi et aI., 1994; Bickley et aI., 1996; Wilson, 1997a). Ev,en elenlental concentrations that are low from a nutritional standpoint can inhibit enzymatic reactions. PCR amplificatioll was not possible from most of the centrifugation, extraction, and chelation/diafiltration treatments. There was only a 2.8% loss of viable phages across the 100 kDa diafiltration membrane, which could be reduced by using a nlembrane with a smaller pore size. Given the small volume used for each PCR reaction, the nunlber of phages in the filtrate was too low to be detected by PCR, regardless of whether inhibitors were present. of inhibitors, amplification should have been possible from all other fractiollS. eh~1 III the absence n,egative results from the organic extraction are consistellt with other studies in which nl0ditied extraction protocols were needed to eliminate soil-derived PCR inhibitors. Single-step chlorofornl-isoanlyl alcohol extraction has been as successful as multi-step phenol and chlorofoffil-isoamyl alcohol extractions in removing PCR inhibitors from intraocular fluids (Wiedbrauk, WeIl1er, & Drevon, 1995). However, the inhibitors present in the aqueous alld vitr,eous fluids w,ere not definitively identified, and may differ sufficiently in character from those present ill soil. 224 Amplification from the centrifugation supernatant was much lower than from the EDTAtreated retentate, despite a very small differellce b·etween the phage concentrations ill those two fractions. This indicates that a substantial amount of the illhibitor was still present in the supernatant, either conlplexed with the phage particles, or freely dissolved in the nutrient broth. Amplification was not possible fronl the sedimented fraction, which suggests that the inhibitor is mostly complexed with the sedimented particulate nlatter, though not necessarily to the phage particles themselves, as the soil eluate contained some small soil particles. Large losses of phages by adsorption to 0.2 flm filtration membranes have generally not been observ;ed during this work (see Chapter 7), but the soil eluate was not filtered ill this experilnellt in order to guard against the potential effect of soil-derived substances on the electrostatic interactions between phages and the membrane. Both centrifugation and diafiltration resulted in overall losses of viable phages. Presumably these phages were inactivated as a result of physical stress, or were bound to the membrane during diafiltration. Defolmation of phages durillg high-speed ,centrifugation in the presence of particulate nlatter has been reported previously (Wommack et aI., 1992; A'shelford, Day, & Fry, 2003), as has the adsorption of viIuses to diafiltration nlenlbralles (Winona et aI., 2001). In fact, tailed phages were particularly likely to be lost to lnembrane adsorption. Since the quantity of phages estimated by real-tilne PCR estimate was 84% of the expected value, the differences in band illtensity between the positive control and the EDTAtreated sample (Figure 6-2) is lnostly attributable to the relative concentrations of phage ill each of the tested fractions, rather than differential alnplification efficiency. It should be note>d that the actual concentration of viable phages in the retentate of treatnlent 1 was 1.94 x l 0 5 PFU/mL, r 225 which is essentially almost identical to the concentration measured by real-time PCR. Viable phages in the original eluate sanlple were lost in the course of treatment, probably because of physical damage from the pressure in the filtration cell or by adsorption to the diafiltration membrane. However, since the exact route of loss is not known, and since physically danlaged phages should still be detectable by PCR, the viable titre of the retentate does not necessarily reflect the true expected quantity, and treatment 1 can not conclusively be said to have completely eliminated PCR inhibitors. This question might be partially addressed by repeating the experiment with a phage from the Podoviridae fanlily, since the cOlnpact morphology of Podoviridae may make them less susceptible than long-tailed Myoviridae like <l>Ea45-IB to inactivation by the physical forces imposed during diafiltration. With a mean elution efficiency of 37% and a m'ean detection efficiellcy of 82%, the overall efficiency of this process was about 30%, which is well within an order of ma.gnitude of the actual number of viable phages originally added to the soil samples. None of the other combinations of EDTA addition and diafiltration were as successful as the application ofD.I M EDTA and I 50-fold dilution (Table 6-6). Dilution alone resulted in the removal of sonle inhibitors, but was much less effective than the combination of chelation and filtration, even at a very high dilution rate. This suggests that divalent metal ions in the soil were a substantial cause of PCR inhibitioll. The Delhi orchard soil is quite sandy, and so these results may not be directly transferable to richer soils that have greater cation exchange capacities and contain higher levels of humic acids. In addition, the method was developed USillg autoclaved soil so that the efficiency of phage elution and detection could be measured. This heat treatmellt lnay affect the presence of soil- 226 derived PCR inhibitors. Skipper and Westermann (1973) reported that autoclavin.g for 1.5 'or 3.0 h increased soil pH by 0.2 units, and postulated that this was due to the breakdown of organic acids (Skipper & Westermann, 1973). A pH change that small should not illlpact tIle buffered PCR reaction mixture, but even though the sterilizing heat treatment used here was only 20 min, it is possible that certain PCR inhibitors could have been removed by the process. The efficacy of the chelation and diafiltration treatlllent should therefore be rep,eated on inoculated and uninoculated fresh soil. The chelation and diafiltration method described here should be suitable for proc-essing a small number of large samples, as in a metagenomic study of microbial COlll11lunities in a particular environment. Unfortunately, the volumes and handling steps involved in diafiltration make it,less appropriate for processing many small salnples. If the phages applied as part of this biopesticide are rapidly inactivated in the soil, then large samples will have to be analyzed in order to recover any phages at all. In that case, this method has the advantage of pellnitting large volumes of soil eluate to be substantially cOllcentrated during diafiltration. If the use of many small samples is necessary, then a possible alternative option for removing the EDTA and chelated metals froln the treated soil eluate might be to chemically precipitate the phages using polyethylene glycol. Prolonged incubation of phages with PEG 8000 can result ill the precipitation of salts, but it may be possible to balance recovery of the phages with the elimination of EDTA. The use of Tth or Tfl polYlllerases should also be explored, as these have been shown to be much more resistant than Taq polynlerase to inhibitors that directly affect the DNA polymerase (Katcher & Schwatrz, 1994; Wiedbrauk, Werner, & Drevon, 1995). Failing that, DNA extraction from the soil itself should be attempted, rather tllan phage elution. 227 Chapter 7: Successful field application of a phage-based biopesticide for fire blight Abstract The efficacy of biopesticides for the control of E. amylovora, the causative agent of fire blight, were tested in field trials. In 2005, six of twelve treatments consisting of E. amylovora phages and a P. agglomerans "caITier" significantly reduced the incidence of bIosso111 blight when tested using a randomized complete block design. The c'ontrol afforded by these treatmellts was not statistically different from that afforded by streptonlycin, which is the most effective treatment for the prevention of blossom blight. The population dynamics of the phage, calTier, and pathogen were monitored over the course of selected treatments. In treatmellts exhibitin,g a significantly reduced incidence of fire blight, the average blossom population of E. amylovora had been reduced to pre-experiment epiphytic levels. An average phage population greater thallI x 105 PFU/blossom at the time of pathogen arrival was required to significantly reduce the chance of E. amylovora infection. 228 Introduction "Unreliable" is one of the most damning criticisn1s that can be leveled at any therapeutic treatment, coming in right behind "the cure is worse than the disease". Unreliability is the curse that has dogged phage therapy for almost 100 years. Leaving aside work done in the first half of the 20th century, before the biological and molecular nature of phages or genes were understood, the results are better, but there is still some ammunition for this charge. One issue is that the success of phage-mediated treatments depends on the survival a11d activity of a biological agent. Since biological control agents generally require a nan'"ower range of environn1e11tal conditions than do chemical agents, their performance tends to be much more variable (Johnson & Stockwell, 2000; Johnson et aI., 2000). However, the etiologic agents of disease also have specific environmental requirements, and since phage prey upon 'bacteria in natural ecosystems, their ecological niches must necessarily be similar. In this case, that ecological niche is th,e apple or pear blossom, and the prey is E. amylovora, the fire blight pathogen. Assuming that virulent phages have been chosen from a diverse collection, using screening methods that reflect the conditions under which they are expected to perfonn, the ultimate test ofbiopesticide efficacy is the field trial, where the biopesticide is also ,exposed to variable environmental conditions, and the existing microbial ecology of the or-chard. It is only in field trials that the efficacy and reliability of treatment can be accurately ass'essed, but few agricultural phage therapy studies reach this stage. Most efficacy studies are YC011ducted on small numbers of lab-grown seedlings or young plants (Kuo et aI., 1971), small cultivation chan1bers isolated from the applicable industrial environn1ent (Boyd, Hild'ebrandt, & Allen, 1971; M-uns'cl1 & Olivier, 1995), or greenhouses (Civerolo & Keil, 1969). Notable exceptions are the use of 229 phages to control soft-rot of calla tubers by Pectobacterium (form'erly Erwinia)carotovora subsp, carotovora (Ravensdale, 2004; Ravensdale et aI, 2007), bacterial spot of tomato by Xanthomonas campestris pv. vesicatoria (Balogh, 2002; Obradovic & Jones, 20'04), and citrus canker caused by Xanthomonas axonopodis (Balogh, 2006). Ravensdale et al (2007) des{;ribe greenhouse trials, but in this case the commercial production of the crop is greenhouse-based, at least in more northern climates such as Ontario. The Xanthomonas campestris worl( is the first agricultural application of phage therapy to pass through a regulatory process alld reach the commercial market, through an American company called OlnniLytics (Salt Lake City, Ul~). Reliability of a phage-based biopesticide is partly dependent on survival in the field environment, where viruses are subject to rapid inactivation by environlnental factors (Zacchardelli et aI., 1992; Schnabel & Jones, 2001; McGuire et aI., 2001; Balogh, 2Q02; Balogh, 2006). Balogh et al (2003) significalltly and substantially enhanced the long-tenn survival of the Xanthomonas campestris phages on field-grown tonlato foliage by fOlTI1UIating the phages with skim milk and sucrose. Suspension in this colloidal medium nlinimized phage inactivation ,due to ultraviolet light and dessication. The alternative strategy described in this study uses P. agglomerans, a non-pathogenic epiphyte that is also susceptible to infectioll by E. amylovora phages, as a "carrier". When both phages and carrier are ap li~d together, the callier should support continual phage replication, thereby limiting the amount of tilne that the phages are exposed to harmful environmental conditions, and increasing their total populatioll frOlTI the time of application until the pathogen is present. The lytic activity of the phages, COll1bine,d with the antagonistic effects of the P. agglomerans carrier, should then be able to "suppress the p'opulation of and reduce the incidence of disease. 230 The efficacy and reliability of the biopesticide is not assured simply because of the natural setting. The variability of weather conditions greatly complicates field trials. Conditions must favour disease development at the same time that the experimental orchards are in bloonl, and therefore susceptible to infection. The disease pressure, the 111eans by which performall.ce is evaluated, must also be reasonable. In most biopesticide field trials disease is assessed SOlne tinle after a concentrated suspension of the pathogen is applied. This type of artificial inoculation is often criticized as unrealistic since the initial pathogen pressure is higher thall what would generally occur in a commercial orchard. This may not be a fair test of the biopesticide, since significant effects that might be seen in a natural orchard infection nlay not be detected (Lindow, McGourty, & Elkins, 1996; Schnabel & Jones, 1999; Malnoy et aI., 2005). However, in trials where exogenous E. amylovora is not applied and infection is solely dependent on 11atural orchard ecology, the amount of disease in the untreated controls may not be high 'enough to detect even a 90% reduction in disease incidence (Werner, Heidellreich, & Aldwinckle, 2004). It has been suggested that honey bees be used as a semi-natural dispersal mechanism by forcillg them to traverse a passage lined with lyophilized pathogen cells as they exit the hive, but evell this method produces variable levels of disease (Johnson et aI., 1993b). While this may be the most realistic test, artificial inoculation produces statistically meaningful results with more consistency. Given the expense and effort of conducting field trials, the moest frequently -chosen compromise is to artificially inoculate trees with a moderate pathogen pressure, and to evaluate success by comparison to the standard treatment rather than to expect 100% control. Based on the pathogenicity tests in Chapters 2 and 5 of this work, the initial E. amylovora population ill field trials should probably b'e less than 1 x 105 CFU/blossom. 231 After all of these factors have been considered, at the root of this questioll of reliability is the need to understand why a particular treatment succeeds or fails at a cel1ain time. This means monitoring the population dynamics of the biopesticide componellts and the pathogen over the course of field trials. In the case of the phage-carrier biopesticide for fire blight, the populations of phage, P. agglomerans carrier, and E. amylovora lllust be tracked. Did the biop'esticide components become established in the target environnlent? Did they continue to flourish until the pathogen arrived? What minimum phage population was required to control the pathogen population? If the treatment fails to reduce disease, knowing the answers to these questions directs the next stage of inquiry. If the biopesticide cOlnponents did not b"eCOnle established, the means or timing of application may need to be altered, or the conditions of·cultivation nlay not be selecting for environmental tolerance. If they became established but were thell illactivated, they may need to be applied more frequently, additives may be needed to better protect them t}om the environment, or the ratio of phage: carrier may need to be adjusted to optimize th,e dynanlic predator-prey equilibrium. If the phage and carrier were well-established but still failed to control the pathogen population, there is a more fundamental problem. TaqMan-style multiplex real-time peR allows the populations of l1lultiple targets to b·e quantified simultaneously. By using primers and probes that all have the same an_nealing temperature, and by using different fluorophores for each speci-es-specfic prob·e, we can simultaneously quantify the phage, the carrier, and the pathogen in a single reaction. This molecular method of population monitoring is faster, more specific, and nl0re sensitiv·e thall traditional culture-based methods. To date, application of this techl1ique to the biological ~oltr of fire blight has been limited to tracking single bacterial speci,es, an-d has not been used to 232 explain biopesticide performance in field trials (Salm & Geidel", 2004; Pujol et aI., 2006). Here, the efficacy of the phage-carrier biopesticide is tested in orchard trials and compared to the efficacy of commercially available biopesticides and to streptomycin, which most effective and most reliable means of preventing fire blight that is currently available. The population dynamics of the phages, the P. agglomerans can'"ier, and E. amylovora on the treated blossoms are monitored using real-time peR, and correlated with the disease outcome of trees treated with the biopesticide. 233 Methods Phages and Bacterial Strains Refer to Chapter 2 for descriptions of all strains and isolates. Tables 7-1 through 7-3 outlines the strains used in each experiment. E. amylovora Ea6-4 was used to atiificially itlfect trees in all field trials. Production ofBacteria for Field Trials P. agglomerans Eh21-5 and E. amylovora Ea6-4 were grown and harvested as described in Chapter 2, except that 150 mm diameter Petri plates were used. Plates were flooded with 0.01 M PB before scraping cells from the surface, and the resulting suspension \vas decanted froIn t·he plate. Each plate produced approximately 100 mL of a 1 x 109 CFU/mL suspension. Fresh cell suspensions were harvested each morning before use, and stored on i'ce. Production ofPhages for Field Trials a) Growth Phages were grown in liquid culture on the E. amylovora host indicated in Table 2-1. Four-port bioreators with a 12 L capacity (Nalgene), containin:g 7 L of nutrient broth were sterilized by autoclaving in pairs for 80 min at 121°C. The bioreactors were left to sit at room temperature for at least two days to ensure the sterility of the nledia. Just before use, bioreactors were connected to a filter-sterilized air supply as shown in Figure 7-1. The bubbled air supplied both agitation and aeration. Each bioreactor was inoculated with 140 mL of E. amylovora at 1 x 109 CFU/mL in 10 n1M PB. After 4 h of growth, 1.8 x lOll PFU of phage 234 Air flow meter: 10L/min Millipore Milled-FG PTFE vent filter A Assemble and sterilize all tubing downstream of the filter (inc. filter). Connect this airflow apparatus to bioreactors just before inoculating the bioreactors. Tubing, stiffened by tying a long, thin scupula tool to the hose bioreactor bioreactor B Figure 7-1. Aeration ofbioreactors. A) Schematic representation of air supply setup. B) Phage cultures after 18 h growth. Note that the air flow meter and vent filter are hidden behind the first bioreactor. 235 were added, for a multiplicity of infection of 1. Cultures were incubated overnight, about 18 h at a room temperature of 22 to 25°C. b) Removal of Cells and Large Debris Bioreactors were disconnected from the air supply and processed one at a time. In general, phages were grown in pairs made up of one phage that produced large plaques with an expanding halo, and one that produced small plaques. 1~he phage that produced large plaques was processed first to make it easier to detect cross-contamination of the two cultures. Bacterial cells and large cell debris were mostly removed by continuous flow centrifugation at 8 000 xg, with an average outflow of about 45 nlL/nlin. Outflow rate was controlled by clamping the outflow tubing. The bioreactor "containing the clude lysate alld the collection vessel were suppolied on ice. To remove all remaining viable cells, the outflow was processed in batches of2 to 30 L, by vacuum-driven filtration through a low protein-binding polyethersulfone membran.e with a 0.2 J..Lm pore size (Millipore Sterivac QP20 cartridges). The filtrate was collected ill sterile nledia bottles with 10 mL of chloroform, and stored at 4°C for an extended period of tim,e. c) Concentration and Buffer Replacement In 2005, phages were concentrated by normal flow diafiltration usin.g all i\nlicon Model 8400 stirred cell apparatus (Millipore) with YM-type regenerated cellulose nletnbrane filters. Preservatives were removed from new filters by floating theln skin-side down for at least Olle hour, with three changes of water. Filters were us-ed no more than 3 times. In between uses, filters were sterilized by floating them in 70% ethanol for 20 lnin, then in TergAZylne 'for 30 min. Sterilized filters were flushed with distilled water and stored in 10% ethanol at 4°C. The Amicon apparatus was assembled according to manufacturer's instructions, with the one gallon pressure vessel reservoir. Approximately 400 mL of distilled water was flush.ed through the system prior to each run. One litre each of phage lysate and distilled water were added to the reservoir. The lysate was concentrated under 50 psi of nitrogen pressure until th·e retentate volume was approxilnately 150 mL. Dissolved nutrients were partially removed by adding 600 mL of 0.01 M PB salt ' and again reducing retentate volume to 110 more than 250 mL. The ultrafiltration membrane was replaced whenever the filtrate flow rate slowed to 5 filL/mill. Membranes were rinsed in a minimal volume of 0.01 M PB salt to recover adsorbed phage, and the rinsate added to the ultrafiltration retentate. In order to determine the optilnal mel11brane size, 10kDa and 100kDa membrane filters were each tested with aliquots of two phage lysates: <t>Ea31-3, a small, short-tailed phage; and <t>Ea46-IA2, a large, long-tailed phage. Flow rates were fastest across the 100 kDa membrane, with less than 0.04% loss in the filtrate. Therefore the 100 kDa membrane was used for all subsequent Amicon phage filtratiolls. In 2006 and 2007, phages were concentrated by tangential flow ultrafiltration using a 30kDa cartridge, with 6 ft 2 regenerated cellulose membralle. The Prep/Scale-TFF apparatu'S (Millipore) was assembled and the cartridge was prepared according to the manufacturer's instructions. Phages were concentrated by combining 3L of the 0.2 Ilm filtrate with an equal volume of sterile distilled water or 10 mM PB, reducing the total retentate volume to about SOD mL at 0.75 mbar internal pressure. Another 2 L ofPB was added to the retentate, which was again reduced to about 500 mL. The system was backflushed once with PB to renlove phages 237 adsorbed to the filtration membrane. Before processing a new phage, the cartridge was cleaned with sodium hydroxide according to the manufacturer's instructions. Total proc'essillg time for 3 L of phage suspension was about 2 hours, including filter preparation. Field Trials Experimental orchards of Malus X domestica and Pyrus communis each COllsisted of 150 (Golden Delicious) or 100 (all others) trees at the Agriculture and Agri-Food Canada Research Farm in Delhi, ON. All apple cultivars were grafted on to M.9 rootstock. Pears were grafted onto seedling quince rootstock. As of2004, Bartlett pear orchards were 4 and 12 y old respectively, Golden Delicious apple orchards were new plantings, and the east and west Fulford Gala apple orchards were 3 and 7 y old, respectively. The Idared orchard was planted in 2006, after spellding 2005 in pots on Jordan Farm. Most trees were purchased from the nurs,ery as 3 y old wood. Each Gala orchard is 0.1 ha, 4 m row spacing and 2.5 m tree spacing within each row. Each Golden Delicious orchard is 0.2 ha, with 4.5 m row spacing and 2.5 m tree spacing. Orchards were assigned to each experiment based on the bloom stag:e, weather forecast, and the number of blossom clusters available. Trees were organized in a ralldolnized complete block design, with one tree per treatment per block. Treatnlents were as described in Tables 7-1 through 7-3 . Treatments were prepared in 1 L hand-held spray bottles, and phage-carrier treatements were left to incubate for 45 min after mixing, in order to allow phage adsorption. Application schedules were as indicated in Tables 7-1 through 7-3. Treatlnents were applied at noon, to near-runoff. E. amylovora Ea6-4 was applied to near runoff with a backpack sprayer. 2004, a 1 x 107 CFU/mL suspension was applied by E. Barszcz. In all other y'ears, a 1 x 1{)6 111 238 CFU/mL suspension was applied by S. M. Lehman. The incidence of diseased blossom clusters was assessed after fire blight symptoms appeared in the controls. Incidence was assessed on a per cluster basis. If necrosis extended to th'e base of the peduncle in at least one blossom in a given cluster, that cluster was considered diseased. If all blossoms had fallen off, the remaining stump and surroundillg tissue at the cluster base was examined for necrosis. The percent of treated clusters that were diseased was recorded. Temperature and relative humidity within the canopy were monitored using HOBO environmental monitors (Onset Computer Corporation) positioned throughout the orchard, protected by weather shields. Data were recorded every 15 min. Data from differellt nlonitors were pooled. There was little variation between distantly spaced orchards (ie. Idared and Gala) therefore data from a given time period was considered valid for all orchards ill Delhi. Two trials were conducted in 2004, one in 4 y old Bal1lett pear and one ill newly planted Golden Delicious (Table 7-1). In the pear trial there were 7 blocks, with 10 treatlnellts per block. In the apple trial there were 8 blocks, each with 12 treatments. BlightBan A506, a commercially available P. fluorescens-based bacterial antagonist, was provided by Nufanll Agricultural Products (Calgary, AL). Three field trials were conducted in 2005 (Table 7-2). hse~F samples of BlightBall C9-1, a commercially available P. agglomerans-based bacterial antagonist, and BlightBan A506 were provided by Plant Health Technologies (Boise, ID). Green Julius Plant Wash and instructions for its use were provided by Lorne Allin, of LifeTinle Solutions (Newcastle, ON). 239 Table 7-1. Experimental design of 2004 field trials. Treatment Type a Orchard Bartlett Pear (4 yold) Golden Delicious (new) Tap water; 100 ppm a.i. Streptomycin 17 Tap water; 100 ppm a.i. Streptomycin 17 BlightBan A506; P. agglom erans 21-5 BlightBan A506; P. agglomerans 21-5 Eh21-5 + cPEa9-5; Eh21-5 + cPEa21-4; Eh21-5 + cPEa35-4; Eh21-5 + <l>Ea46-1 A2; Eh21-5 + cPEa9-5 (I 120th volume); Eh21-5 + cPEa46-IA2 (I/20th vol.) Eh21-5 + 3xI0 2 PFU/mL cPEa9-5; Eh21-5 + 3x 10 2 PFU/mL <l>Ea21-4; Eh21-5 + 3xI0 2 PFU/nlL cPEa45-1B; Eh21-5 + 3xl0 2 PFU/mL <pEa46-1A2; Eh21-5 + 5x 10 5 PFU/mL <l>Ea51-2; Eh21-5 + 5xl0 5 PFU/mL cPEa31-3; Eh21-5 + 5xI 0 7 PFU/mL cPEa35-4; Eh21-5 + 5xI 0 6 PFU/mL cPEa35-4; Treatment Application 80-100% bloom (II May) 50-70% bloom (19 May) Pathogen Application 100% bloom (12 May) 80-1000/0 bloOlTI (20 May) Controls Bacterial Antagonists Phage-Carrier b b All treatments were prepared with tap water. 8 b All bacterial suspensions were prepared to 1 x 10 CFU ImL. The concentrations of the phage suspensions used in the pear orchard were unknown. a 240 Table 7-2. Experimental design of 2005 field trials. Treatment Type Controls Bacterial Antagonists Phage-Carrier Orchard a b b Bartlett Pear (12 y) Fulford Gala (4 y old) Fulford Gala (8 y old) 10 mM PB; 100 ppm Streptomycin 10 mM PB; 100 ppm Streptomycin 10 mM PB; 100 pplTI Streptomycin BlightBan A506; BlightBan C9-1; P. agglomerans Eh21-5 BlightBan A506; B lightB an C9-1 ; P. agglomerans Eh21-5 P. agglomerans Eh21-5 Eh21-5 + <pEa2l-4; Eh21-5 + <pEa45-1 B; Eh21-5 + <pEa46-1 A2; Eh21-5 + <pEa45-1 B + <pEa46-1 A2 Eh21-5 + <pEa2l-4; Eh21-5 + <pEa45-1B; Eh21-5 + <pEa46-1A2 Eh21-5 + <pEa45-1 B + <pEa46-1 A2 10 6 Eh2l-5 + <pEa46-1 A2 (MOI=O.l); 10 6 Eh21-5 + <pEa46-1 A2 (MOI=I); 10 6 Eh21-5 + <pEa46-1 A2 (MOI=10); 10 8 Eh21-5 + Q>Ea46-1 A2 (MOI=O.l); 10 8 Eh21-5 + <pEa46-1 A2 (MOI=l) Other Green Julius Plant Wash (1 :100 dilution)* Treatment Application 90-100% bloom (19 May) 30-400/0 bloom (20 May) 40-500/0 bloonl (21 May) 30-40% bloom (20 May) 40-50% bloom (21 May) Pathogen Application 100% bloom (21 May) 50-700/0 bloom (22 May) 50-70% bloom (22 May) All treatments were prepared in PB except for streptomycin, Green Julius, and BlightBan, which were prepared in tap water according to the manufacturer's instructions. 8 b Bacteria and phage were prepared to final concentrations of 1 x 10 CFU/mL, unless otherwise indicated. a 241 Table 7-3. Experimental design of 2006 field trials. Treatment Type Orchard a Controls Bacterial Antagonists Phage-C arrier b b Treatment Application Pathogen Application Treatment Type C C Phage-Carrier Idared (new) 10 mM PB; 100 ppm Streptomycin 10 mM PB; 100 ppm Streptomycin 10 mM PB B lightB an C9-I ; BlightBan A506; P. agglomerans Eh21-5 BlightBan C9-I; BlightBan C9-1; P. agglomerans Eh21-5 P. agglomerans Eh2I-5 Eh21-5 + <pEa21-4; Eh2I-5 + <pEa45-1B; Eh2 1-5 + <pEa46-1 A2; Eh21-5 + <pEa45-1 B + <pEa46-I A2 Eh21-5 + <pEalO-l; Eh2I-5 + <pEa 10-6; Eh21-5 + <pEa3I-3; Eh21-5 + <pEa35-4; Eh2I-5 + <pEa31-3 with Physpe overspray Eh21-5 + <pEa45-1 B; Eh2I-5+ Q>Ea46-1 A2 30-400/0 bloom (13 May) 40-50% blooln (14 May) 30-40% bloom (13 May) 40-50% bloom (14 May) 80-1000/0 bloom* (10 May) 50-70% bloom (15 May) 50-700/0 bloom (15 May) NA Orchard b b Other Golden Delicious (new - north) Golden Delicious (new - south) 10 mM PB; 100 ppm Streptomycin 10 mM PB; 100 ppm Streptomycin P. agglomerans Eh21-5 P. agglomerans Eh21-5 Eh2I-5 + <pEa46-I A2 + <pEa21-4 + <pE a3l-3; Eh2I-5 + <pEa2I-4 + <pEa31-3 + <pEa35-4; Eh2I-5 + <pEa3I-3 + <PEa35-4 + <pEa 10-1; Eh21-5 + <pEa35-4 + <pEa 10-1 + <pEa46-1 A2; Eh21-5 + <pEa21-4 (30 nlin); Eh21-5 + <pEa21-4 (3 :30 min); Eh21-5 + <pEa31-3 (30 min); Eh21-5 + <pEa31-3 (3 :30 nlin); Eh21-5 + q,Ea46-1 A2 (30 min); Eh21-5 + <pEa46-1 A2 (3:30 nlin) Inventek soap (1: 124 v/v); kasumin (84 ppm) Treatment Application Pathogen Application Fulford Gala (9 y old) a Controls Bacterial Antagonists Fulford Gala (5 y old) C C 30-40% bloom (13 May) 40-500/0 bloom (14 May) 30-40°A> bloom (13 May) 40-500/0 bloom (14 May) 50-70% bloom (15 May) 50-700/0 bloom (15 May) a All treatments were prepared in PB (pH 6.8) except for streptomycin, Green Julius, and BhghtBan, which were prepared in tap water according to the manufacturer's instructions. 8 b Bacterial and phage were prepared to final concentrations of 1 x 10 CFU or PFU/mL, unless otherwise indicated. If more than one phage was used, the summed concentration of virions was 1 x 10 8 PFUhnL. C Individual clusters were flagged for treatment and followup asse ssnlent if most blossoms in the cluster were fully open. Therefore, this note reflects the bloom stage of the treated clusters, not the orchard as a whole. c 242 Five field trials were conducted in 2006 (Table 7-3). Sufficient amounts of the BlightBan products remained in storage from 2005. Their viability was tested by suspending 1 g of each formulation in 10 mL of PB for 15 min, and then plating serial dilutions of each suspellsion on nutrient agar. The culturable concentration of BlightBan C9-1 was unchanged from the time of packaging. The culturable concentration of BlightBan A506 was 10-fold lower than the 1 x lOll CFU/g stated on the label. The amount of BlightBan A506 used was adjusted accordingly, in order to achieve a final concentration of 1 x 10 8 CFU/mL. A new formulation of bactericidal soap, called Inventek, and instructions for its use were provided by Lome Allin, of LifeTime Solutions Inc. Physpe is an SAR inducer produced by Goemar Laboratories (Saint-Malo, FR). Population Monitoring During 2004, blossom populations of P. agglomerans were estimated by plate COUlltS. Three blossoms from different locations on each pear tree in block 2. Each set of three blosSOll1S was sonicated for 2 min in 10 mL of 10 mM PB. The blossom wash was serially diluted in PB and 100 ilL were plated on both nutrient agar and modified Miller-Schroth medium (Brulez & Zeller, 1981). Total P. agglomerans populations were estilnated based on the number of yellow bacterial colonies on nutrient agar, which were the dominant colony type. These -counts were similar to the total number of orange, EPS-producing colonies on MMS. MM'S is a senliselective media, on which P. agglomerans and Erwinia amylovora produ·ce characteristic COl011Y types. Blossoms were sampled 20 h after treatment application, alld 4 dafter treatlnel1t application. Very few blossoms were open in the Golden Delicious orchard, therefore one blossom 243 was sampled from each tree in blocks 1,3,5, and 7,20 h after treatnlent application. Bacterial populations were collected in 1 mL PB. Total P. agglomerans populations were estimated based on the number of yellow colonies on nutrient agar. The identity of these colonies was confirmed by replica plating on CCT, a semi-selective medium on which P. agglomerans has a characteristic appearance. Pre-treatment populations in both orchards were estimated frOlTI 8 trees distributed evenly through the orchard. During 2005 and 2006, multiplex real-time peR was used to monitor the populations of P. agglomerans, bacteriophages, and E. amylovora on the blossom surfaces at three tilne points during the experiment: before BCA application, immediately following first BCA application, and immediately before application of the pathogen. In 2005, three blossolTIS were sal1lpled fronl each tree, in each of three blocks. The petals were removed from each blossom, the renlaining blossom heads were plucked from the petiole and placed ill a sterile, capped plastic culture tube. Samples were transported on ice, alld stored overnight at 4°C before analysis. Direct Plant Extraction Buffer (DiPEB, Cat. No. 00690, Agdia Inc., Elkhart, IN) was added to each sample in a ratio of 1 mL buffer per blossolTI. Samples were sonicated for 2 min, and a 1 mL aliquot of each was centrifuged at 10 000 xg for 10 min at 4°C. The resultillg p,ellet was resuspended in 100 JlL of wash buffer and used 3 JlL as the template for real-time PCR reactions. 2·44 Real-time peR Primers and TaqMan-style probes were designed by Dr. W. -S. Kim (unpublished data) and were as described in Table 6-1, except that the <I>-dpo2 probe and plim·ers described in Chapter 2 were used for phage detection. All reactions were conducted in a 25 L~ total volume. In 2005, reactions were calTied out under the following conditions: 95°C for 10 min for initial denaturation, followed by 40 cycles of 95°C for 15 s and 60°C for 30 s, with 3 endpoint fluorescence readings taken durin,g the amplification step. Each reaction contained IX Brilliant QPCR Multiplex Master Mix (Stratagene), 200 ~M each primer, 100 ~M each probe, 3 ~L tClnplate. In 2006, reactions were carried out under the following conditions: 95°C for 5 Inin, followed by 40 cycles of 95°C for lOs and 60°C for 16 s, with 2 endpoint tluorescenoe readin,gs taken during the amplification step. Each reaction contained 200 nM each prinler, 100 nM 'each probe, 3 ~L template. Standard curves were constructed using serially dilutillg cell or phage suspensiolls in clean blossom wash, either individually or as nlixtures. Standard curve construction was repeated using new cultures. PCR reactions were run in triplicate. Statistical Analysis Regression analysis of standard curve data was conducted in SigmaPlot, version 8 {SPSS, Inc., Chicago, IL). All other data were analyzed using SAS (Statistical Analysis Systell1s 8.2; SAS Institute, Cary, NC). Disease incidence was analyzed using the ,gellerallinear nlodel (PROC GLM). A one-sided Dunnett-Hsu test (ex = 0.01) was used to·evaluate the reduction in disease 245 relative to the water or buffer control. Differences between treatnlents w'ere assessed using a twosided Tukey-Kramer test (ex = 0.05). Very little disease was observed in the 2004 Golden Delicious trial, and it developed too late in the season to directly relate sytnptoms to specific blossom clusters. Therefore the presence or absence of disease was modeled using PROC GENMOD, with a binomial distribution. To compensate for overdispersion, the scale paral11eter was estimated from the square root of the ratio of Pearson's X2 statistic to degrees ofl~edn. 246 Results Sensitivity and Specificity ofReal-time peR The specificity of the lTIultiplex real-tinle PCR primers and probes is shown in Table 7-4. The lsc gene is present on the ChrOlTIOSOme of E. amylovora, therefore all strains were detected regardless of whether they carry the pEA29 plasnlid. Multiplex standard curves for E. amylovora and E. pyrifoliae are shown in Figure 7-2. Concentrations higher than 3 x 106 CFU per 25 L~ reaction were not tested. The relationships between threshold cycle (C T ) and initial quantity were linear across a 4-10.g range. The standard curves for both species are essentially identical; the slight differen'ces between slope and intercept values were within the range ofvariatioll between replicate real-time PCR runs. As in Chapter 4, there is no significant difference between the standard curv'es produced in singleplex vs. multiplex reactions. Standard curves were prepared from cells diluted in buffer that had b'een sonicated \vith uninoculated blossoms according to the sampling procedure used for the field trials. The curves therefore correspond directly to illitial cell numbers, and account for the pl~esn·c of any P'CR inhibitors that may be present in the experimental samples. Since blossom infection occurs through the stigma and hypanthiulTI, petals and peduncles were rell10v,ed fi·om each blosso111, and bacteria were collected from the surfaces of the renlaining stluctures. RetTIoval of petals and peduncles improved the sensitivity of bacterial quantification (data not shown), presull1ably because it reduced the total volume of plant tissue, and thus the cOll'oentration ofplant-d·erived PCR inhibitors. DNA extraction was not performed as part of sanlple preparation. The components of the wash buffer, alollgwith th-e 5 lTIin initial dellaturation cycle at 95°C, w,ere Table7- 4. Specificity of the real-tirne peR prinlers and probes used Species Tested Strain flost Plant Origin ill this stlldy. All strains \vere tested Signal frolu Priluers and Probe designed for: E. anlylovora Erwinia olnylovora with pEA29 plaSll1id \vithout pEi\29 pIa slni d Panloea agglo/nerans Bacteriophages of E. an1.l'!ovora ~h1liv,: E pyr(fbliae E'nvin ia car%vora .P agglonlerans ill a rnultiplex reaction.. Original Strain R.eferencc or Source Phage Canada + Ai\.FC-Vincland .i\4alus .)1~ Nlichigan, UBA -+ Ritchie and Ea273 Alalu.. .· sp, ·Nc'JI.' York, US/\. + Beer (1971) Ea1/79 Cotoneaster "'p. Gennany + Falkenstein et at, 1988 1113-1 Raphiolepi,s' indica Louisiana. lISA + £--lOkOD)b. 1998 LebB66 Alalus sp. Lebanon + 8aad,2000 lJTR.T2 A1alus .\"p. lJtah, lJSA + 1'ho111s0n and ()ckey, 200 I 1614-2a Crataegus sp. Spain + Llop et aI., 2004 Eh21-5 Pyrus .sp. Canada Ehl-28b i\1a '~,ul Canada + AA.FC-Vineland 33243 not specified Canada .+- Alnerican Type Culture Collection 49018 Ala/us sp. not specified + Arnerican Type Culture Collection 2696 Prunus J\1'ichigan, USA. C9-1 Al(llu.\· sp, l\t1khigan, USA + Ishirnaru et al~ E325 Ala/us .\1). vVashington, tJSA + Pusey, 1997 NEal !\lalus sp. ~/lich gan, NEa31-3 soil Canada NEa46-1 soil Canada NEa45-J soil Canada Epl/96 p"vrus pvr~j()liJe Korea Ea6-4 P.rru..v sp. EallO ."1). arn1e hl(~a DH-5" P flu 0 rescen:\' 1\506 1977 this \vork International Collection of 1\1 icroorganislns tion) Plants USA -+- Ritchie, 1978 + Gill, 2000 1988 (,ill., 2000 -t· Gill, 2000 Kiln et aL. 1999 I). Cupples Ecc26 ]:':.\'cherichia coli Klos~ A. J. Castle .. Lindenlann and Suslovv, 1987 1'0 ~ '-J 248 45 , - - - - - - - - . - - - - - - . . , . - - - - - - - - . - - - - - - . . . - - - - - - . , 40 . 1:::. 35 ---- .. T' - -.:, ..:. :. ..:n: -~ 30 : " " " .' ":'"~' --. 25 . 20 . .. 15 10 ·1·········· _ '.' .---- ..:~ .. ~ ~ _ - . . ~- - -r - -~ - ~- - r- - - i 102 103 104 105 106 genome copies present in reaction Figure 7-2. Standard curves for the detection of E. amylovora (open boxes, solid line), P. agglomerans (open circles, dashed line), and phages (open triangles, dotted lille) ill a 111ultiplex reaction. Quantification is given by the equations y = 47.57 - 3.11 log(x); R 2 = 0.987 (E. amylovora), y = 46.98 - 3.17 log(x); R 2 = 0.992 (P. agglomerans), alld y = 68.01 - 5.821og(x), R 2 = 0.962 (phage). 249 sufficient to make the bacterial target DNA accessible for amplification. Production ofPhages for Field Trials The concentration of two phages was monitored throu,gh each stage of production in 2005. <l>Ea46-1A2 is a long-tailed phage with a large head, and <l>Ea31-3 is a short-tailed phage with a small head. The concentrations of these phages in two independent trials, conducted with different initial concentrations of phage, are given in Table 7-5. Very slight losses were observed across the 0.2 of the <l>Ea31-3 0.2 ~m ~m membrane. 111 trial 1, the net volume filtrate was reduced 10-fold across the 100 kDa Inelnbrane. In all other cases the net volume of the 0.2 ~m filtrate was reduced 5-fold across the 100 kDa nlembrane. The relative concentrations of the two fractions are generally consistellt with th,eir relative volumes. No net changes in retentate volume was made using the 50 kDa nlembrane. In trial], there was less than a 0.04% loss of total viable phages to the 100 kDa filtrate, even for the smaller phage. Larger losses were observed across the 100 kDa mell1brane in trial 2, 7% for <l>Ea46-1A2 and 0.2% for <l>Ea31-3. Additional losses of phage, apparently by adsorption to the membrane or physical damage, were observed durillg ,certain 'diafiltration steps. There was a substantial loss of <l>Ea31-3 during the 50kDa filtration in trial 1, and a 25% loss of <j>Ea31-3 to the 100 kDa membrane. In 2006, tangential flow filtration was used to concentrate the phages illstead of normal flow diafiltration. With a 100 kDa membrane, up to 10% of <l>Ea31-3 was lost to the filtrate (data not shown). Switching to a 30kDa membrane reduced this loss to less than 0.1 %. 25'0 Table 7-5. Concentration and loss of phages throughout production and purification. Concentration (PFU/mI-J) Phage Fraction Trial 1 <pEa46-1 A2 After continuous flow centrifugation 8 x 10 7 1.7 X 10 9 100 kDa diafiltration retentate 6 x 10 8 9.0 X 10 9 1.2 10 4 X 6.5x10 7 50 kDa diafiltration retentate NT 6.0 x 10 9 50 kDa diafiltration filtrate NT 1.8 x 10 7 After continuous flow centrifugation NT 7.3 x 10 9 0.2 Ilm Millipore filrate 100 kDa diafiltration retentate 6 x 10 3 6.7 100 kDa diafiltration filtrate 50 kDa diafiltration retentate 50 kDa diafiltration filtrate a 2.0 x 10 9 0.2 Ilm M illipore filtrate 100 kDa diafiltration filtrate <pEa 31-3 a NT Trial 2 "NT" indicates that the fraction was not tested. X 10 4 X 0 X 10 9 2.5 x 10 10 6.5 x 10 6 0 2.0 6.6 10 1 2.5 X 10 10 3.8xl0 6 251 Disease Incidence and Microbial Population Dynamics - 2004 Disease incidence in the pear orchard is shown in Figure 7-3; the results were highly variable among different blocks. Disease incidence in the control treatment was 31.4% of clusters. None of the treatments were significantly different fronl the water control (P > 0.3, Tukey-Kramer). Disease symptoms progressed rapidly through these trees, and the orchard had to be removed. Very little disease developed in the Golden Delicious orchard, even in the buffer control, and very few blossom clusters were present in these )loung trees. Therefore disease was recorded as presence or absence in the tree as a whole. There was no significallt difference between the likelihood of disease in any of the treatments alld likelihood of disease in the water control (P > 0.09, PROC GENMOD). Despite the lack of treatment efficacy in 2004, there does appear to have been a sho,rtterm effect of some treatments on the blossom populations of P. agglomerans (Figure 7-4A). Twenty hours after treatments were applied to the pear orchard, the nlean P. agglomerans populations on trees treated with streptomycin, BlightBan A506, or calTier with <l>Ea21-4 were lower than the water-treated trees by a factor of 102 to 103, and were at least 10-fold lower than the pre-treatment populations the day before. Twenty hours after treatnlents were applied to the apple orchard, P. agglomerans was not detected on BlightBan A506-treated trees (Figure 7-43B. The mean P. agglomerans populations on trees treated with streptonlycin orcanier with <t>Ea45IB were at least 102-fold lower than any other treatments, includin,g the control, but still within an order of magnitude of the pre-treatment populations. 252 100 _ Control c::=J Bacterial Antagonist '§' 0.> _ _ 80 1i) Phage-Carrier Antibiotic :J U 0 ;:R e- 60 o.> <.) C 0.> U '0 40 E 0.> CJ) co 0.> CJ) (5 20 0 gc 0 () ~ Q) ~ ~ co 0 I,() « c a> 'C ro () ..,. N co W v J> «C'\I co cD ('t) W v co e+ e .2> a> a> e+ co 'C 23 .c: in + "E 'E () () co co w a> ro () I,() a> enco ~ e+ ..,.~ w a> 'C ro () ~ ~ C'\I N co w ~ cD a> e+ e+ 'C ro () Treatment :g v co w a> 'E co () Figure 7-3. Incidence of fire blight in 4 y old Bartlett pear (2004). ~. c E 0 a. ~ Ci5 253 :u..-s ~ u A 8 6 C") Q .......,- 4 ~ ..Q E ... 2 z o "l- i 1:Qt.D '..... 0 -=wn <~- co Treatlnent ~ :: 10 ~ B Q (fj fA 0 ~ 8 :::; u.. u 6 C') Q ---' Ii- 4 4> ..Q = z ~ .... -a; U 2 0 - .(J >. ~ ...;... C 0 ~ i "l- as ... 0 -=... <t U CQ '<.D ~ CO> CQ Il"'> (C Lt( ~ UJ e ., ~ ~ 'P.""" ~ M e e LU ,tt:l LU ., LFf' e ...,. Lon M '(c a:l J.r) \~ ('0 ~ Ln' 7- LU Q ......- e ~ ~ UJ e UJ li- e tJ'J. Treatnlent Figure 7-4. P. agglomerans populations on A) 4 y old pear, and B) new Golden Delicious apple blossom hypanthia during 2004 field trials. In panel A, the illitial MOl was 1 for all phage-carrier treatments except those marked "(dilute)". The MOl for treatlllents ill pallel B was variable (see Table 7-1), and is only stated here to distinguish betw·een the different applications of<pEa35-4. The mean population immediately before treatment was 1.5 x 104CFU/blosSOlll in the pear orchard, and 8.6 x 104 CFU/blossonl. All phages were mixed with the call·ier prior to application. 254 Disease Incidence and Microbial Population Dynamics - 2005 In 2005, three field experiments were conducted in apple and pear orchards (100 trees/orchard, randomized complete block design). Disease incidence in the bufTer-treated controls was approximately 50% in the two apple trials, and 72% in the pear trial (Figure 7-5 and 7-6). Streptomycin, which is considered the gold standard for tire blight prevention, significantly reduced the incidence of disease in all three experiments. The efficacy of BlightBan products was variable. In the 12 y old pears that were treated once, neither BlightBan A506 not BlightBan C9-1 caused a significant reduction in disease incidence (Figure 7-5A). In the apple orchard, only BlightBan C9-1 was effective (Figure 7-5B). The Green Julius bactericidal soap was highly effective (Figure 7-6). Disease incidence ill Green Julius-treated apple trees was 27.2%, which was significantly less than in the control trees (p=0.0048, Dunnett-Hsu test). Three phages were applied with the P. agglomerans Eh21-5 calTier, in various combinations. Across all experiments, 6 out of 12 phage-based treatments significantly reduced disease incidence relative to the water-treated control (P < 0.01). In the 12 y old pear and east Gala orchards, the carrier alone did not cause a significant reduction in disease incittence, while certain phage-carrier combinations did (Figure 7-5). However, different phage-carrier combinations were effective in each orchard. In the west 'Gala orchard, the effects of the total and relative populations of phage and carrier were tested. Only two phage-can'"ier treatments in this trial were effective, the treatment consisting of 1 x 10 6 CFU/mL 'Carrier with 1 x 1-0 5 PFUlnlL phage, and the treatment consisting of 1 x 10 8 CFU/mL carrier with 1 x 10 8 PFU/mL of phage (Figure 7-6). 255 100 100 _ Buffer _Buffer c=J Bacterial Antagonist [=:::::J Bacterial Antagonist '§' Q.) '§' lI!!!!!!!Il Phage-Carrier 80 _ (i5 Antibiotic Q) ::J <3 '0 60 ~ Q.) t> C '0 t> Q.) 40 "0 '0 .E Q.) en Q.) en 40 .E Q.) en Ctl (5 60 Q.) C Q.) "0 Phage-Carrier Antibiotic ::J <3 '0 ~ _ _ 80 (i5 co Q.) 20 en (5 20 ~o (J Q.) ~ + ~. Treatment li> Treatment () Figure 7-5. Incidence of fire blight in A) 12 y old Bartlett pear and B) 4 y old Gala apple orchards (2005). Asterisks indicate significantly less disease than in the control. 256 a ~ ~m U) c=J Bacterial Antagonists a m _ _ co en::3 Bactericidal Soap Buffer _ Phage-Carrier Antibiotics U I+- 0 a co ~ ~ Q) c..> c Q) "0 '(3 a ~ ~ Q) en ro Q) en 0 a N a e C 0 <-> ~ :::J CO Q3 "E co <-> 0 II <5 6 c.b ~ co UJ Ie ~ S <5 6 eO ~ 0 II <5 0 6 6 ~ co c.b ~ c.b ~ e UJ UJ UJ co e 106 CFU/ml carrier co II e .r:. ~ <5 6 .~ cco a. (J) Ci5 ~ 0= ~ ~. eO co c (J) co ~ E 0 ~ ::; ..., UJ e I~ c: (9 108 CFU/ml carrier Treatment Figure 7-6. Incidence of fire blight in 8 y old Gala apple treated with varying amounts of <l>Ea46lA2 in different carrier:phage ratios (2005). Asterisks indicate significantly less disease than in the control. 257 In the 2005 field trials, multiplex Real-time peR was used to monitor the populatiolls of the carrier, two phages, and the pathogen in two experimental orchards (Figure 7-7). Sanlples were collected immediately before and after initial treatment application, ilnmediately after pathogen application, and 3 days after pathogen application. In all cases, disease out'COlne was correlated with an E. amylovora population of less than 1 x 104 CFUlblossonl 3 -days after its application. In all cases where significant control was not achieved, the total P. agglomerans population was lower than in the one experiment where the carrier-alone treatmel1t was successful; and/or if phage had been applied, the phage populatioll showed a net decline by the time pathogen was applied. In all cases where significant cOlltrol was achieved, if pha,ge had been applied, there had been a net increase in the phage population by the time pathogen "vas applied (at the expense of the P. agglomerans population). The phage population thell continued to increase, but now growing preferentially on the pathogen. If phage had not been applied, the P. agglomerans population was larger than in Ullsuccessful treatnlents. In all experimellts, the efficacy of successful treatments was not statistically ditlerent from the efficacy of streptonlycin (P < 0.05, Tukey-Kralner test), which is the most effectiv·e bactericide cun"ently available. This amounted to a 50% reduction in disease incidence in apple trees, and a 33% reduction in pear trees. Disease Incidence and Microbial Population Dynamics - 2006 A severe, unforecasted frost four days after treatnlent resulted in the dest1uction of all open blossoms in the experimental orchards. As a result, no data were collected, and these experiments are scheduled to be repeated il12007. Howev'er, within one day oftreatl11ent, 258 --------------r s- 107 o 106 10 A 1()8 20 ~ o 10 o 106 ~ % ~ enc 10 ~ 103 f- o CO Q.. 102 101 Treatment Applications: s- May 20 May 21 May 22 May 23 May 24 10 SGA c t SGA t May 20 May 21 May 23 May 24 May 25 SGA t SGA t Pathogen I .., 25 30 25 S- 107 '5. 106 20 o 0- N ~ 15 4 .Q 103 10 102 ~ 20 0- CL 't3 ~ -105 <D ~ ~ N <D cE 106 ::J CL ~ o CO Q.. ro 15 10 103 10 ::J ~ 4 ~ E 10 2 10' 10 1 May 19 May 20 May 21 May 22 May 23 May 24 May 19 May 25 May 20 May 21 t SGA t SGA t Pathogen May 22 May 23 May 24 May 25 Date Date Treatment Applications: May 22 ::ID------------. -----------~30 105 CO Q.. t Treatment Applications: Pathogen % ~ E ~ 102 May 19 107 10 ::J ~ Date ::J ~ ro ~ May 25 '5. o 15 4 Date 1()8 E 10 1 May 19 109 20 105 N ~ 10 7 :[ ::J ~ 15 104 103 CO Q.. ~ 5 N ~ 30 25 25 ::J 10 B 108 '5. % ----------------r 9 30 Treatment Applications: t SGA t SGA t Pathogen Figure 7-7. Population dynamics of E. amylovora (red bars), P. agglomerans (yellow bars), and Erwinia phage <pEa46-1A2 (blue bars) on 8 y old Gala blossoms (2005). Population sizes are the average numbers of each species present on the stigma, anthers, and hypanthium of each blossom, based on composite samples from each of 3 experimental blocks. Timing ofbiocontrol agent (BCA) and pathogen applications are indicated below the x-axis. Temperature (solid black line) was recorded throughout all experiments. (A) Buffer-treated trees; (B) Trees treated with the P. agglomerans carrier alone; (C) Trees treated with a mixture of 1 x 106 CFU/mL P. agglomerans and 1 x 10 5 PFU/mL <pEa46-1A2; (D) Trees treated with a mixture of 1 x 10 8 CFU/mLP. agglomerans and 1 x 10 8 PFU/mL <pEa46-1A2. 25'9 moderate to severe phytotoxicity was observed on blossonls treated with the bactericidal soap. Weather Conditions Figure 7-8 shows the weather conditions that were monitored in the experilnental orchards during the 2004 and 2005 field trials. High relative humidities were recorded throughout all experiments, and were generally associated with nightly dew. However, the recorded temperatures varied substantially anl0ng these three experilllents. The daytime temperatures in early May of 2004 (panel A) were well withil1 the optimum growth range for E. amylovora. The experinlental orchard, 100 trees of 4 y old Baltlett p'ear, was so severely diseased that it was destroyed. These sanle conditions also led to the eruption of shoot blight in an adjacent orchard of 11 y old Bartlett pear trees that had been infected with E. amylovora in previous years, but which had been asynlptomatic at the time. Thiliy percent of that orchard was destroyed, and the relnaining trees were heavily prulled in order to remov,e diseased tissue. Later in the same month, a similar experiment was conducted in the 1 y old Golden Delicious orchard. The average daily temperatures were low·er than they were during the pear experiment, but were still well within the optimum growth range for E. amylovora 011 most days. In this case, however, the temperature dropped sharply, to less than SoC, within 7 h of pathog'en application. For the rest of the season, only one or two infected blossoln clusters \vere 'observed among the 150 trees in this orchard, but evidence of latent illfection was observed in 50% of the trees at the start of the following spring. The average daily temperatures in 2005 (panel B) appear nlargil1al when ·considering only 260 the optimum pathogen growth range, but were really quite moderate when the total growth range is considered, particularly on the day of pathogen application and th'e day imn1ediately after it. A 43% incidence of disease was observed in the control trees in this orchard (Figure 7-4). Moderate pruning was required to remove the diseased tissue from these 3 to 7 y old Gala trees, and 110ne had to be destroyed. The temperatures during the 2006 trials (panel D) were similar to those in 2005 (panel B), but a frost on May 23 killed all of the open blossoms. The dead blossoms fell off of the trees and no disease data could be collected. 261 A B 35 35 100 100 80 25 70 10 50 50 40 40 10 30 30 . - cold snap overnight 20 20 + - - - , - - - - - , - - - - - , - - - - r - - - - - - , - - - , - - - - - r - - - _ + _ 10 o N >. ~ v C") N ~ Cll N >. ~ Date Date c >. E c ~ ..c Q.co roQ..«S a. 0 E ~ ..c Q.co ro a. 0 Q..«S C 35 D 35 100 100 30 30 80 80 25 E ~ 25 ~ 60 20 ~ Q) ~ E ~ ::c ro ~. .2 0- E ~ i5 Q) 15 40 > :w 10 CD 0- Cll E Q) a::: I- Q) 60 20 :::J 15 40 10 20 20 0 N >. ~ Cll N ~ N N >. C") ~ ~ N >. >. Cll ~ Date Cll ~ L{) 0 N >. Cll ~ >. ~ >. N ~ >. <D V ~ >. ~ >. co >. :E Date Figure 7-8. Orchard weather conditions during the A) 2004 4 y old pear trial, B) 2004 Golden Delicious apple trial, C) 2005 Gala trial, and D) 2006 trials. The date of pathogen application, and the concentration of the E. amylovora suspension that was applied are indicated. Note that a higher pathogen pressure was applied during 2004. The optimum growth temperature range of E. amylovora is indicated for all years. The broader growth range of E. amylovora is shown in panels Band D. Discussion The phage production methods developed during the course of these fi.eld trials demonstrate that larger-scale ·batch culture and processing is possible. Phage concentrations in the partially clarified crude lysate (following continuous flow centrifuguation) were on the order of 1 x 109 PFU/mL. Ideally, this should be about 10-fold higher, but it was sufficient for the needs of these field trials, and is about the average for many types of phages cultured in volunles of 0.5 to 10 L (Munsch & Olivier, 1995; Civerolo & Keil, 1969; Civerolo, 1970). The greater loss of phages to filtrate using a tangential vs. normal (ie. Amicon) flow systenl with the 'same nlembrane pore size suggests that phage retention by the 100 kDa membrane is pal11y dependent on electrostatic phage-phage and phage-membrane interactions. If retelltion was solely due to size exclusion, the greater surface area of the tangential flow membrane should hav·e had a substantial increase in filtration rate without a substantial illcrease in phage permeability. The key to efficient filtration, at any stage of processing, is the use of a nlelnbral}e with a large-surface area. However, this can also increase the amount of phage lost to the filtrate, making it inlpOl1ant to balance phage retention with filtration speed wh'en choosing the filter size. The weather and disease data from 2004, 2005, and 2006 paint a fairly complete pictUI:e of the disease triangle, as it applies to fire blight in Canada. In general, cool temperatures result in lower disease incidence. However, moderate temperatures do no guarantee illfection, since extremely low temperatures immediately following pathogen application can severely limit E. amylovora activity, and frost events can cause the death of infected blossoms before E. amylovora reaches the base of the peduncle. It is not surprising that the conlbination of high temperatures and a high inoculum pressure resulted in the c'olnplete destruction of a young p,ear 2·63 orchard. The loss of this pear orchard in 2004 prompted a 10-fold reduction in pathogen pressure for the 2005 and 2006 trials. High pathogen pressure may also have prevented the detection of significant treatment effects. Low concentrations of phages were applied in this experinlent, but the lack of streptomycin efficacy indicates that an additional factor, comlnon to all treatm'ents, might have masked treatment effects. Some of the effects of weather can be seen by examining a situatioll in which a higher pathogen pressure did not lead to a higher incidence of disease. During the 2005 Gala apple trial, about 50% disease developed in the untreated control trees as a result oftelnp ra u ~s that were cool to moderate from the perspective of pathogen growth. By comparisoll, the daily temperatures in the 2004 Golden Delicious trial were higher, yet virtually no dis'ease developed even though the pathogen pressure was IO-fold hi,gher. These were -both apple orchards illvolving susceptible scion cultivars grafted on to highly susceptible rootstocks. The most notable difference between these two trials was the overnight cold snap that followed patho,gen application. This seems to have suppressed the E. amylovora population enou.gh to prevent the development of disease during the 2004 season. And yet brown, sh'epherd's crook-shapedsh-oot tips were observed the next spring on half of the treated trees. The shepherd's crook is a characteristic sign of necrosis due to prior E. amylovora infection, but, unlike tlue cankers, does not generally harbour living bacteria that can serve as future inoculum. This sugg'ests that these trees were asymptomatically infected, probably on the saIne day that pathogen was applied. The effect of host susceptibility can be seen by comparing the apple and pear trials ill 2005. Data from 2005 suggest that even the mature 12 y old pear trees were inherently 1110re susceptible to infection than the younger apple trees. Ail 3 orchards w'ere inoculated with E. 264 amylovora in the same manner, alld experienced identi'cal weather conditions following inoculation. Despite this, a higher incidence of disease was observed among the buffer-treated controls in the pear orchard as compared to the apple orchard. The only difference between the two orchards, other than species, was the bloom stage. However, since blossonls become less susceptible to infection as they age (Hildebrand & Heinicke, 1937; Gouk, Bedford, & Hutshins, 1996; Thomson & Gouk, 2003), this supports the conclusion that the Bartlett pear trees are more susceptible to fire blight than Fulford Gala. In light of this, futureexperilnents in Bartlett pear should be conducted by inoculating treated trees with an even lower concentratioll of E. amylovora, such as a 1 x 105 CFU/mL suspension. The 2005 field trials demonstrate that a phage-mediated biopesticide for fire blight is a practical possibility. Successful phage-carrier treatments reduced th'e incid'ence of blossom blight by reducing the population of E. amylovora on orchard blossoms to the epiphytic level that was present in the orchard before the start of the experiment. In previous efficacy screens usin:g some of these phages, synlptom control on an 11 mm diameter pear fruit plugs was correlated with as nluch as a 97% reduction in surface E. amylovora populations (Gill, 2000). However, control was not always associated with significant population reductions and even those reductions 'still left as much as 1 x 106 to 1 x 107 CFU per plug. In the orchard trials described -here, successful phage treatment caused a 99% reduction in the E. amylovora population, and this reduction was correlated with disease control. Overall, phage-carrier treatments were no less reliable than the conl1nercially available fire blight biopesticides, and phage-carrier combinations were sometim,es eflective when the carrier alone or the commercial biopesticides were not. eh~1 et1icacy of successful treatm,ents was 265 not statistically distinguishable from the efficacy of streptolnycin. Finally, molecular nlonitoring of the phage, carrier, and pathogen populations allowed the success or failure of selected treatments to be correlated with logical chang"es in biopesticide alld pathogen population sizes, and showed that at least one phage preferentially reproduces at the expense of the pathogen when it is present. In addition, the use of a 1 x 10 6 CFU/nlL suspension of E. amylovora in 2005 appears to be an appropriate test of fire blight biopesticides, since it produced 50% to 70% disease in the control trees and permitted detection of significant treatment effects. An average phage population greater than 1 x 105 PFU/blossonl at the time of pathogen arrival was required to significantly reduce th'e chance of E. amylovora infe'ction. The nlean phage population observed in successful phage-based treatlnents \vas approximately 1 x 106 PFU/blossom, in comparison to 1 x l0 5 PFU/ blossom or less in Ullsuccessful tl~eamns. Balogh (2002) observed no significant difference in the efficacy of 1 x 106 PFU/nlL v'S. 1 x 108 PFU/nlL X campestris phages in controlling tonlato leafblight, but these values refer to the initial concentration of phages applied. The actual concentration of active pha.ges on the leaf surfaces was not monitored over the course of the experiments. In vitro replication experiments show that phage-host 'encounters are a stochastic process dependent on Brownian or externally applied motion. As such, they are d-ep'endent upon the presence of a critical density of susceptible host cells in order to teplicate, 'So 111uch so that the traditional calculations of MOl are inappropriate for host concentrations below 1 x 107 CFU/mL (Kasman et aI., 2002). This may have been a contributil1gfactor in the greater efficacy of phagecarrier treatments that were prepared to 1 x 108 rather than 1 x 106 CFU or PFU/nlL. Illitial phage adsorption in the more concentrated preparations would have beell COllsistent with a tlue M~OI of 2~6 1, ensuring that enough infected carrier cells reached the blossom surface to initiate a sustainable phage population. The model presented by Kasman et al (2002), which is based on liquid culture, does not necessarily reflect phage growth patterns once the biopesticide has been applied to the blossom. Interactions among the phages, carrier, and pathogen are likely occurring within the small volumes of aqueous secretions present on the stigma and hypanthium. However, the scale on which this occurs is small enough that growth within an extremely thin agar overlay nlay be the better analogy: replication within microscopic regions resembles liquid culture, but expansioll to adjacent areas is limited the way plaque expansion is limited. It is also possible that neither analogy is appropriate, in which case the population monitoring techniqu'es used ill this study may be useful in modeling the population dynamics of the phages and their hosts on plant surfaces. Several aspects of phage-carrier perfonnance must still be tested ill a field situation before this system could be confidently used on a larger scale. All of the phage-carrier biopesticide components were prepared fresh before field use. This strategy is not practical for large-scale production and use. Lyophilization nlay be an altelnative option. Lyophilized bacterial antagonists become established on pear and apple blossom more consistently, and at higher levels, than fresh cultures of the same strains (Stockwell, Johnsoll, & Loper, 1998). The effect of lyophilizing phages is less certain. E. amylovora phages do not seem to survive well when lyophilized in the skim milk fonnulation, as evidenced by the poor recovelY of the Vineland collection (Chapter 2), and the inconsistent recovery of PEal and PEa7 fronl ATCC cultures (A. M. Svircev and J. J. Gill, personal communication). In contrast, many pha,ges tend to 267 survive well when stored at 4°C in liquid suspension, with little or no loss of viable titre in a year's time (Munsch & Olivier, 1995; Chapter 2). Since the field trials involving phage cocktails were lost to frost in 2006, these experiments must be repeated at a future date. The phage-carrier biopesticide system lnust ultimately involve mixtures of several different phage types, which should ideally be tested against a mixed suspension of multiple E. amylovora strains. This is partly because cocktails have been shown to reduce the emergence of phage-resistant bacteria (Tanji et aI., 2004; Tanji et aI., 2005), and partly because phages can be selected based on their ability to broaden the host range of the treatment. However, the success of different phages in experiments inv'olving different host plants at different stages of development suggests that cocktails may improve the reliability of a phage-carrier biopesticide. The rate and efficiency of phage adsorption can be affected by the presence of external cofactors, and by the physiological state of the bacterial host (Guttman, Raya, & Kutter, 2005). Since blossom nectar composition is known to vary with scion cultivar and with blossom age (Paulin, 1987; Pusey, 1999; Pusey & Curry, 2004), it is entifely possible that these factors could influence the efficacy of anyone phage by af1e-ctin_g i·ts ability to replicate on the carrier or to attack the pathogen. 268 General Summary The presentation of this research began by emphasizing the impol1ance of integrating multiple performance-based goals into every stage of the biopesticide developnlent process. My application of these principles is to the developnlent of a biopesticide that cOlltrols orchard populations of E. amylovora through the combined activities of Erwinia phages and P. agglomerans is sUllllllarized here and in Figure D-I. First, fire blight is an appropriate target for biopesticides in general, alld for phage therapy in particular (Chapter 1). The primary POillt of infection is the open blossoln. This is not the Ollly infection court that is important for disease developll1ent, but the 1110St comlnon disease stages do follow directly from the blossom blight phase. The open blossom is a surface that is easily accessible using ~xist ng pesticide application technology, and its life span represents a tinite period of susceptibility. Therefore, phages applied to the blossonls can infect and lyse the pathogen as during the early stages of blossom colonization and growth, thereby preventillgboth primary infection and the expansion of the total available inoculum in the orchard. Once an appropriate disease target is selected, the biopesticide n1ust be designed such that it remains active at the necessary site. The blossom surface is, however, a ho'stile environn1ent ~for phages. The P. agglomerans carrier was used to prolong the perio-d over whi'Ch infective phages were present by supporting their replication in the absence of the pathogen. New generations of phage were produced at the expense of the can'"ier, increasing the phage population and lillliting the all10unt of time that free phage were exposed to damaging ultraviolet light and dessication. The selection of phages and a carrier was based on the sinlultaneous consideration of factors that would affect all stages ofbiopesticide production and use (Chapter 2). A P. 269 agglomerans carrier was selected that was susceptible to a large number of phages with differellt host ranges, thereby ensuring that it could ev'entually suppol1 the replication of a mixture of phages that will infect a broad range of E. amylovora strains. The callier also produces an antibiotic against E. amylovora which, if active on the blossom, would contribute to pathogen control by antibiosis. Seven phages with broad and differing host ranges were also selected frolll the Vineland collection. The novelty of the <l>Ea21-4 genollle (Chapter 3) underscores the diversity of both the Vineland phage collection and the global phage metagenome as a whole. The ability of these phages and carrier candidates to reduce fire blight synlptom severity in isolated pear blossoms was also factored into the selection process, since that scenario is representative of the environment in which the biopesticide lnust be active (Chapter 2). Significant differences in the efficacy of different isolates was rarely obtained in this illitial screening process. This was probably the result of treatment tillling. Subsequent experilllents (Chapter 5) demonstrated that preventative treatment is more likely to be effective wh'en the treatment is applied several hours before inoculation with the pathogen. These later experilllents also confirmed that the chosen phages and carrier, when applied in combination, could effectively reduce symptom development in pear blossoms. The significant, though lower, efficacy of the carrier alone showed that it also contributes directly to the control of E. arov l ~ma populations on the blosSOlll. The in planta blossom assay was also used to develop the initial paranleters for field experiments (Chapter 5). The superior performance of established phage-can·ier populations, as well as established protocols for comlllercial bacterial biopesticides, dictated that treatlnents should be applied at least one day before the pathogen. The optinlulll E. amylovora inoculunl 270 required for the blossom assays (1 x 106 CFU/ll1L) was increased ten-fold for the initial field trials based on the disease incidence rates reported for similar pathogen pressures. The destruction of the pear orchard in 2004 showed this to have been all overestilTIate, and pathogen suspensions were prepared to 1 x 106 CFU/mL for subsequent trials (Chapter 7). To ensure that phage-carrier biopesticide components could be easily produced, the in vitro growth characteristics of bacteria and phage were studied, and the results were used to optimize the production of phages for field trials (Chapter 5). Phages that could not be easily grown to high concentrations were removed from consideration. Purification and storage protocols were also developed (Chapter 7). Finally, the phage-carrier biopesticide was tested in the field (Chapter 7). The results of field trials demonstate that the tested phage-carrier combinatiolls and application protocols can significantly reduce the incidence of blossom blight in the orchard. Moreover, that reduction is biologically meaningful, having been indistinguishable frOITI the efficacy ofstreptonlycill. These results were achieved by applying a carrier suspension of 1 x 108 cfu/mL, with an MOl of 1. This is consistent with previous studies in which phage:bacteria ratios greater than or equal to 10 were required in order to protect plant tissues from disease using pre-treatnlent with phages (reviewed by Vidaver, 1976). This suggests that a 10: 1 ratio of phage:host is a killillg ratio, and that a low'er ratio would facilitate the long-term persistence of both phage and carrier. Key to the success of the field trials was the use of multiplex real-time PCR to simultaneously monitor the phage, carrier, and pathogen populatiolls over the cours,e of the experiment. Sampling protocols were developed from in planta blossonl assays conducted with E. an1ylovora and E. pyrifoliae (Chapter 4). With this method it was possible to d-evelop an 27] understanding of microbial ecology that determines the success or failure of the biopesticide, an understanding which has historically been Olle of the greatest stumbling blocks in the development of phage-mediated therapies (Goodridge, 2004). The population dynanlics elucidated by this method confirmed the hypothesis underlying the design of this biopesticide. Specifically, it was shown that the P. agglomerans can·ier can support the replication of phages on the blossom surface, and that the phage replicated preferentially on the E. amylovora pathogen when it was present. This thesis is the first reported use of an exogenously applied catTier bacterium to directly increase the efficacy of a phage-based therapy. The sequence data obtained from the genonle of <pEa21-4 (Chapter 3) will facilitate the development of phage-specific I~eal­ time PCR detection so that the relative performance of phages in mixtures can be mOllitored ill future field trials. The data collected during field trials also demonstrate the complex way in which \veather, pathogen pressure, and host susceptibility interact to determine disease outcomes (Chapter 7). One consequence of this variability is that effective fire blight nlanageIl1ent in Canada may always require that growers have a range of pesticide types available to thenl: biopesti-cid;es for general E. amylovora population suppression, and the option of occasional streptoInycin applications in those marginal scenarios when the efficacy ofbiopesticides nlay be low, but the risk of fire blight is real. This critical demonstration of efficacy, along with the 1110lecular tools to explain it, are only directly useful for disease control if growers are eventually allowed to use it. Regulatory factors ultimately determine whether a biopesticide is practically useful, re,gardless of its efficacy. The development of a practical biopesticide therefore requires that the enVirOl1l11ental fate of its 272 components be accoullted for, for reasons of human safety and environmental protection. The safety of the other active component of the phage-canier biopesticide, the P. agglomerans can ier, is not likely to be an issue with respect to either hUlnan exposure or 4 potential impacts on orchard ecology. The organism is applied before fruiting, is surface-limited, is naturally present in the orchard at low levels, and has not been shown to be toxic or pathogenic to humans. It is not expected to present any non-occupational exposure risk, and even the occupational risk is minimal. In fact, the US Environmental Protectioll Ag,el1cy has gral1ted exemptions to the usual requirelnent for a maximum permissible residue level of the two commercial biopesticides containillg P. agglomerans, BlightBan and Bloolntime, stating that "Pantoea agglomerans is ubiquitous in the environnlent, and is recogIlized as an epiphyte of a wide variety of plants", and citing extensive evidence that the species is not toxic, pathogenic, or an irritant to animals or hunlans (Environnlental Protection Agency, 2006a; Environm,ental Protection Agency, 2006b). These products have also been successfully registered in Canada. Human exposure to phages is not a particular concern. Their specificity for bacterial cells precludes the possibility of human infections, and there is extensive evidence that phages pose no other risk to humans. Phages are the most abundant entities 011 Earth, and have been found as contaminants in vaccines and sera (Merril et aI., 1973; Milch & Fornsosi, 1975; Moody, Trousdale, Jorgenson, & Shelokov, 1975). Perhaps most tellingly, the United States Food and Drug Administration recently approved a phage 111ixture for use on ready-to-eat foods, placil1g it the "Generally Recognized as Safe" category (EBI Food Safety, 2006). Still, the fate of exogenously applied E. arov l ~ma phages will likely require 1110re study than the fate of P. agglomerans.The main 'concell1 would be the effect of a high concel1tration of 273 Phage Characterization Broad Host-range .;...-""'....----------1 .. Phages (Genetic and Biological) / In .lJlanta Assays to Deternline Application Paralneters P. agglonlerans Carrier , r Appropriate Disease Target Maintenace ofActive Biopesticide at Active Site Real-til11e peR to Monitor PhageCarrier-Pathogen Dynatnics High Efficacy , Practical Production & Purification Methods ~ - ~ .. A Useful Biopesticide Regulatory Acceptability H_ost Range Studies to Ensure Minimal Off-target Inlpact COll1ponents that Have/Can be Demonstrated to be Safe Protocols to Trace lEnvironmental Fate 'I Figure D-I. Integrated developnlent of an effective and practical phage biopesticide. 274 phages on the overall microbial ecology of the orchard. This was addressed in two ways, by developing a method to track the persistance of phages in soil '(Chapter 6), and by cOllfinning that the Vineland phages do not infect other comnl0n orchard bacterial genera (Chapter 2). The research presented in this thesis has implications for the broader study of nlicrobial ecology in soil and in aerial plant tissues and soil. The nlultiplex real-time PCR protocols that were developed to monitor bacteria and phages in soil salnples and on blossolns, in both the ti'eld and in planta assays, should be applicable to a range of other projects. The ability to 1110nitor natural microbial communities in a completely culture-independent manner is inlportant for our understanding of the role of phages in lnicrobial ecology. Reanney and Marsh (1973) pointed out that if phages occur in soil at even 0.1 % of the levels observed in laboratory culture, they must be the most abundant genomes in that environment. Viruses also contribute substalltially to biogeochemical cycling (Bratbak, Thingstad, & Heldal, 1994; Middleboe, Jorgensell, & Kroer, 1996), and can facilitate genetic change in their hosts (Holland & Donlingo, 1998; Boyd, Davis, & Hochbut, 2001). Enlichment methods not only obscure vital infonnation about population sizes, they restrict studies to phages infecting culturable hosts when only 1% to 50/0 of endogenous soil bacteria are thought to be culturable at all (Torsvik, Goks0yr, & Daae, ] 990). One response to the unique challenges posed by phage therapy has beell to adopt a lTIOre reductionist approach and use phage components such as purified or transgenically expressed lethal enzymes (Gaeng et aI., 2000; Ko et aI., 2002; Salm et aI., 2006), lysis-deficient phag,es which kill bacteria without a sudden increase in endotoxin dispersal (Matsuda et al., 2005), or to simply use inactivated phages as a convenient means of displaying immuno·genic epitopes independently of their pathogenic owners (Cao et aI., 2000; Solonl0n, 2007). A ullifying thenle 275 among these strategies is that the kinetics of a self-replicating therapeutic agent have been removed fronl the scenario, retumillg the treatment to the dose-response relationships associated with classical chemotherapeutic methods. This silnplifies many things, but for the nlo'St part it does so by sacrificing one of the unique strengths of phages as therapeutic agents. The data presented in this work shows that a phage-based biopesticide is a practical possibility for the control of fire blight, and demonstrates that it is possible to develop a phage-Inediated therapy can be developed which capitalizes on, rather than circumvents, the unique propel1ies of phages. Phage therapy is not suitable for all bacterial diseases, whether in plants or anilnals. Certain infection courts may not be accessible to phages, as in the case of intracellular mycobacteria, or may not allow them to persist long enough to be effective, as in the bovine udder (Gill et aI., 2006a; 2006b). Biofilms also present a particular challellge to phage therapy because of their heterogeneous physical, chemical and biological structure (reviewed by Briissow & Kutter, 2005). There may be creative solutions to these problems in certain cases, but probably not all of them. At the same time, not all phages are suitable for phage therapy. Phages that carry toxin genes, or that facilitate high rates of transduction, will not be suitable for use in their native form. The success of phage therapy against any particular disease also depends on the availability of many phages strains that, together, are effective against nlost or all strains of the causative agent. Assuming those factors are favourable, the initial developlnent of the elelnents of a phagebased biopesticide must be always be conducted with the eventual requirelnents of the fi·eld situation in mind. The population dynanlics of the relevant species must also be monitored so that it is possible to develop an understanding of the microbial ecology that detemlines the success or failure of a treatment in a true in situ trial. Finally, the results of those field trials nlust 276 be used to identify elements of the biopesticide-pathogen-host plant interactions that 'Can be optimized by further research. As this research has demollstrated, the lik,elihood of developing practically feasible phage therapies is substantially increased by satisfying these cliteria. 277 Literature Cited Ackermann, H.-W. (2001). Frequency ofnl0rphological phage descriptions in th·e year 2000. Archives of Virology, 146, 843-857. Ackermann, H.W. (2006). Classification of bacteriophages. ]n R. Calendar (Ed.), The Bacteriophages. (pp. 8-16). New York, NY: Oxford University Press. Ackermann, H.W., & Nguyen, T.M. (1983). Sewage coliphages studied by electron lnicroscopy. Applied and Environmental Microbiologv, 45,1049-1059. Adams, M.R. (1959). Bacteriophages. New York: Interscience Publishers, Inc. Agriculture and Agri-Food Canada. (2007). New biopesticides help manage tire blight. Pesticide Risk Reduction Program Pest Management Centre. Ahokas, H., & Erkkila, M.J. (1993). Interference ofPCR amplification by the polyamines, spermine and spermidine. PCR Methods and Applications, 3, 65-68. Aldridge, P., Metzger, M., & Geider, K. (1997). Genetics of sorbitol metabolisln in Erwinia amylovora and its influence on bacterial virulence. Molecular and General Genetics.. 256,611-619. Aldwinckle, H.S., & Beer, S.V. (1976). Nutrient status of apple blossoms and their susceptibility to fire blight. Annals of Applied Biolgy, 82, 159-163. Aldwinckle, H.S., Bhaskara, R.M.V., & Norelli, J.L. (2002). Evaluation of control of fire blight infection of apple blossoms and shoots with SAR inducers, biological agents, a growth regulator, copper compounds, and other materials. Acta Horticulturae., 590, 325-331. Aldwinckle, H.S., & Norelli, J.L. (1990). Evaluation ofkasumin for control of fire blight. Acta Horticulturae, 273,391 Anderson, L.M., Stockwell, V.O., & Loper, J.E. (2004). An extracellular protease of Pseudomonas jluorescens inactivates antibiotics of Pantoea agglomerans. Phytopathology, 94,1228-1234. Ark, P.A. (1937). Variability in the fire-blight organism, Erwinia amylovora. Phytopatholo.gy., 27, 1-28. Ashelford, K.E., Day, M.J., & Fry, J.C. (2003). Elevated abulldance of bacteriophage infectin.g bacteria in soil. Applied and Environmental Microbiology, 69, 25-289. 278 Babadoost, M. (2005). Fire blight of apple. Report on Plant Disease, 801, University of Illinois Extension, Department of Crop Sciences, Urbana-ChalTIpaign, Illinois. Bachoon, D.S., Otero, E., & Hodson, R.E. (2001). Effects of humic substances on iluorOll1etric DNA quantification and DNA hybridization. Journal of Microbiological Methods, 47, 73-82. Baker, K.F. (1971). Fire blight of pome fruits: The genesis of the concept that bacteria can be pathogenic to plants. Hilgardia, 40, 603-633. Baldwin, C.H.Jr., & Goodman, R.N. (1963). Prevalence of Erwinia amylovora in apple buds as detected by phage typing. Phytopathology, 53,1299-1303. Balogh, B. (2002). Strategies for improving the efficacy of bacteriophages for <controlling bacterial spot of tomato. University of Florida: Masters thesis. Balogh, B., Jones, J.B., Momol, M.T., Olson, S.M., Obradovic, A., King, P., & Jackson, L.E. (2003). Improved efficacy of newly fomlulated bacteriophages for managenlent of bacterial spot on tomato. Plant Disease, 87, 949-954. Balogh, B. (2006). Characterization and use of bacteriophages associated with citrus bacterial pathogens for disease control. University of Florida: PhD thesis. Barny, M.A. (1995). Erwinia amylovora hrpN mutants, blocked in harpin synthesis, express a reduced virulence on host plants and elicit variable hypersensitive reactions on tobacco. European Journal of Plant Patholgy~ 101, 333-340. Bauske, R.J. (1971). Wind dissemination ofwaterbolne Erwinia amylovora fronl Pyrus to Pyracantha and Cotoneaster. Phytopathology, 61,741-742. Beattie, G.A. & Lindow, S.E. (1994). Comparison of the behavior of epiphytic fitn,ess lTIutants of Pseudomonas syringae under controlled and field conditions. At1plied and Environll1ental Microbiology, 60, 3799-3808. Beckennan, J. (2007). Fire blight on fruit trees in the home orchard. Fruit Disea ~ Purdue Extension, Purdue University, West Lafayette, Indiana. BP-30-W, Beer, S.V., & Norelli, J.L. (1977). Fire blight epidemiology: factors atTectillg release of Erwinia amylovora by cankers. Phytopathology, 67, 1119-1125. Beer, S.V., Kim, J.H., Zumoff, C.H., Bogdanove, A.J., Laby, R.J., Tanii, A., Tanlura, 0., Gustafson, H.L., Momol, T., & Aldwil1ckle, H.S. (1996). Characterization of bacteria that cause "bacterial shoot blight of pear" in Japan. Acta Horticulturae, 411, 179-182. 279 Beji, A., Mergaert, J., Gavini, F., Izard, D., Kersters, K., Leclerc, H., & De Ley, J. (1988). Subjective synonymy of Erwinia herbicola, Erwinia milletiae, and Enterobacter agglomerans and redefinition of the taxon by genotypic and phenotypic data. International Journal of Systematic Bacteriologv.. 38, 77-88. Bennett, R.A., & Billing, E. (1978). Capsulation and virulence in Erwinia amylovora. An.nals of Applied Biology, 89,41-45. Bereswill, S., Jock, S., Aldridge, P., Jallse, J.D., & Geider, K. (1997). Molecular characterization of natural Erwinia amylovora strains deficient in levall synthesis. Physiological and Molecular Plant Pathology, 51, 215-225. Berg, G., Sanjaghsaz, H., & Wangwongwatana, S. (1989). Potentiation of the virucidal effectiveness of free chlorine by substances in drinking water. 55, 390-393. Bernhard, F., Schullerus, D., Bellenlann, P., Nimtz, M., Coplin, D.L., & Geider, K. (1996). Genetic transfer of amylovoran and stewa11an synthesis between Erwinia amylovora and Erwinia stewartii. Microbiology, 142, 10'87-1096. Besemer, J., Lomsadze, A., & Borodovsky, M. (2001). GeneMarkS: a self-training nletho'd for predicition of gelle starts in nlicrobial genomes. Ilnplications for finding sequen·ce nlotifs in regulatory regions. Nucleic Acids Resarch~ 29,2607-2618. Bickley, J., Short, S.K., McDowell, D.G., & Parkes, H.C. (1996). Polymerase chaill reaction (PCR) detection of Listeria monocytogenes in diluted milk and reversal ofPCR inhibition caused by calcium ions. Letters in Applied Microbiology, 22, 153-158. Bieleski, R.L. (1969). Accumulation and trallslocatioll of sorbitol in apple phlo,eln. Australian Journal of Biological Science, 22, 611-620. Biggs, A.R. (1994). Characteristics of fire blight cankers following shoot inoculations of three apple cultivars. HortScience, 29, 795-797. Billing, E. (1974). The effect of temperature on the growth of the fireblight pathogen, Envinia amylovora. Jounlal of Applied Bacteriology, 37, 643-648. Billing, E. (1980). Hawthorn as a source of the fireblight bacteliulll for pear, apple and ornamental hosts. Boston, MA: Pitnlan Advanced Publishillg Progranl. Billing E., Baker, L.A.E., Crosse, J.E., & Garrett, C.M.E. (1961). Characteristi'cs of En,glish isolates of Erwinia amylovora (Burrill) Winslow et al. Journal of Ap,plied Bacterio)o,gv, 24,195-211. 280 Biswas, B., Adhya, S., Washart, P., Paul, B., Trostel, A.N., Powell, P., Carlton, R. & Mell·il, C.R. (2002). Bacteriophage therapy rescues mice bacteremic from a clinical isolate of vancomycin-resistant Enterococcus faecium. Infection and ImlTIUnity'l 70, 204 - 21-0. Bitton, G. (1975). Adsorption of viruses onto surfaces in soil and water. Water Research" 2, 473-484. Blachinsky, D., Shtienberg, D., Oppenheinl, D., Zilberstaine, M., Levi, S., Zanlski, E., & Shoseyov, O. (2003). The role of autunln infections in the progression of fire blight synlptoms in perennial pear branches. Plant Disease, 87, 1077-1082. Blachinsky, D., Shtienberg, D., Zamski, E., Weinthal, D., & Manulis, S. (2006). Effects of pear tree physiology on fire blight progression in perennial branches and on expression of pathogenicity genes in Erwinia amy'lovora. Europeall Journal of Flant Patholo,gy, 116, 315-324. Bogdanove, A.J., Kim, J.F., Wei, Z.M., Kolchinsky, P., Charkowski, A.O., Conlin, A.K., Collmer, A., & Beer, S.V. (1998). Homology and functional similarity of an hrp-linked pathogenicity locus, dspEF, of Erwinia amylovora and the avirulen·ce locus avrE of Pseudomonas syringae pathovar tonlato. Proceedings of the National Acadenlyof Sciences, 95, 1325-1330. Bogovazova, G.G, Voroshilova, N.N., & Bondarenko, V.M. (1991). The efficacy of Klebsiella pneumoniae bacteriophage in the therapy of experinlental Klebsiella il1fection. Zhurnal Mikrobiologii, Epidemiologii" i Imnlunobiologii'l 4, 5-8. Bogovazova, G.G, Voroshilova, N.N., Bondarenko, V.M., Gorbatkova, G.A., Afanas'eva, E.V., Kazakova, T.B., and SlTIirnov, V.D. (1992). IUlmunobiological properties and therapeutic effectiveness ofpreparatiolls from Klebsiella bacteriophag·es. Zhurnal Mikrobiologii" Epidemiologii, i Immunobiologii, 30-33. Bogs, J., Bruchmuller, I., Erbar, C., & Geider, K. (1998). Colonization of host plants by the fire blight pathogen Erwinia amylovora marked with genes for biolull1inescence and fluorescence. Phytopathology, 88, 416-421. Bogs, J., & Geider, K. (2000). Molecular analysis of sucrose metabolism of Erwinia am)Jlovora and illfluence on bacterial virulence. Journal of Bacteriologv, 182,5351-5358. Bonasera, J.M., Meng, X., Beer, S.V., Owens, T., & Kinl, W.S. (2006). lnteraction of DspE/A, a pathogenicity/avirulence protein of Erwinia amylovora, with pre-ferredoxin froll1 apple and its relationship to photosynthetic efficiency. Acta Horticulturae" 704,473-478. Bonn, G. & Carter, N. (2002). Fire Blight of Apple and Pear in Ontario. OMAF Factsheet Order 281 No. 02-011. Ontario Ministry of Agriculture and Food. Bonn, W.G., & van der Zwet, T. (2000). Distribution and economic impol1ance of fire blight. In J. L. Vanneste (Ed.), Fire Blight: The Disease and its Causative Agent, Erwinia amvlovora. (pp. 37-53). Wallingford, UK: CABI Publishing. Borejsza-Wysocka, E.E., Malnoy, M., Aldwinckle, H.S., Meng, X., Bonasera, J.M., Nissinen, R.M., Kim, J.F., & Beer, S.V. (2006). The tire blight resistance of apple clones ill which DspE-interacting proteins are silenced. Acta Horticulturae, 704, 509-514. Boyd, E.F., Davis, B.M., & Hochbut, B. (2001). Bacteriophage-bacteriophage interactions in th·e evolution of pathogenic bacteria. Trends in Microbiology, 9, 137-144. Boyd, R.J., Hildebrandt, A.C., & Allen, O.N. (1971). Retardation of crown gall enlargenlent after bacteriophage treatment. Plant Disease Reporter, 55, 145-148. Bratbak, G., Thingstad, F., & Heldal, M. (1994). Viruses and the microbial loop. Microbial Ecology, 28,209-211. Braun, P.G., & Hildebrand, P.D. (2005). Infection, carbohydrate utilization, and profein profiles of apple, pear, and raspberry isolates of Erwinia amylovora. Canadian Journal of Plant Pathology, 27, 338-346. Brisset, M.N., Chartier, R., Paulin, J.P., & Chevalier, R. (1990). Experimentations with Firestop 3M in the chelnical COlltrol of fire blight. Acta Horticulturae, No. 273, 413-418. Brisset, M.N., & Paulin, J.P. (1992). A reliable strategy for the study of disease alld hypersensitive reactions induced by Erwinia amylovora. Plant Science, 85, 171-177. Brulez, W. & Zeller, W. (1981). Seasonal changes of epiphytic Erwinia an1ylovora 011 ornamentals in relation to weather COllditions and the course of infection. Acta Horticulturae, 117, 37-43. Briissow, H., & Kutter, E. (2005). Genomics and evolution of tailed phages. ]11 E. Kutter & A. Sulakvelidze (Eds.), Bacteriophages: Biology and Applications. (pp. 91-128). Boca Raton, FL: CRC Press. Bugert, P., & Geider, K. (1995). Molecular analysis of the an1S operon re-quired for exopolysaccharide synthesis ofErwinia amylovora. Molecular Microbiolo-gv, 15,917933. Burge, W.D., & Enkiri, N.K. (1978). Virus adsorption by five soils. Journal ,of Environmental Quality, 7,73-76. 282 Burrill, T.J. (1878). Pear blight. Transactions of the Illinois 'State Horticultural Society, 114116. Calendar, R. (Ed.) Individual Tailed Phages. In The Bacteriophages. (pp. 209-496). New York, NY: Oxford University Press. Calenge, F., Drouet, D., Denance, C., van de Weg, W.E., Brisset, M.N., Paulin, J.P., & Durel, C.E. (2005). Identification of a major QTL together with several minor additive or epistatic QTLs for resistance to fire blight in apple in two related progenies. Theoretical and Applied Genetics, Ill, 128-135. Calzolari, A., Peddes, P., Mazzucchi, D., Mori, P., & Garzeno, C. (1982). Occurrence of Erwinia amylovora in buds of asymptomatic apple plants in COlnn1erce. Phytopathology, 103, 156-162. Cao, J., Sun, Y., Berglindh, T., Mellgard, B., Li, Z., Mardh, B., Mardh,S. (2000). Helicobacter pylori-antigen-binding fragments expressed on the filan1entous M13 phage prevent bacterial growth. Biochin1ica et Biophysica Acta, 1474, 107-113. Carlson, K. (2005). Working with bacteriophages: Con1mon techniques and methodolo.gical approaches. In E. Kutter & A. Sulakvelidze (Eds.), Bacteriophages: Biolo:gy and Application. (pp. 437-487). Boca Raton, FL: CRC Press. Carlton, R.M., Noordman, W.H., Biswas, B.,de Meester E.D., & Loessner, M.J. (2005). Bacteriophage PI 00 for control of Listeria monocytogenes in foods: Genon1e sequence, bioinformatic analyses, oral toxicity study, and application. Regulatory Toxicology and Pharmacology, 43,301-312. Celetti, M. & Carter, N. (2004). Remove fire blight cankers in the winter. Ontario M.inistry of Agriculture and Food. Chanishvili, N., Chanishvili, T., Tediashvili, M., Barrow, P. A. (2001). Phages and th·eir application against drug-resistant bacteria. Journal of Chemical Technolo.gv and Biotechnology, 76, 689-699. Chemotherapeutic agents. (1994). In T. D. Brock, M. T. Madigan, J. M. Martinko, & J. Parl<er (Eds.), Biology of Microorganisms. (pp. 352-356). Prentice-Hall, Inc. Cherry, W.B., Davis, B.R., Edwards, P.R., & Hogan, R.B. (1954). A sin1ple procedure for the identification of the genus Salmonella by means of a specific bacteriophage. Journal of Laboratory and Clinical Medicine, 44,51-55. Chiou, C.-S. & Jones, A. L. (1991). The analysis of plasmid-mediated streptolnycin resistance in 283 Erwinia amylovora. Phytopathology, 81, 710-714. Chiou, C.-S. & Jones, A. L. (1993). Nucleotide sequence analysis of a transposon (Tn5393) carrying streptomycin resistance genes in Erwinia amylovora and other gram-negative bacteria. Journal of Bacteriology, 175,732-740. Chiou, C.-S. & Jones, A. L. (1995). Molecular analysis of high-level streptomycin resistance in Erwinia amylovora. Phytopathology, 85, 710-714. Civerolo, E.L., & Keil, H.L. (1969). Inhibition of bacterial spot of peach foliage by Xanthomonas pruni bacteriophage. Phytopathology, 59, 1966-1967. Civerolo, E.L. (1970). Comparative relationships between two Xanthomonas pruni bacteriophages and their bacterial host. Phytopathology, 60, 1285-1288. Cline, J.A., Hunter, D.M., Bonn, W.G., & Bijl, M. (2001). Resistance of the Vineland series of apple rootstocks to fire blight caused by Erwinia amylovora. Journal of the American Pomological Society, 55, 218-221. Cline, J.A., Byl, M., & Hainstock, L. (2003). Rootstock performance in Ontario: III. Apple rootstock. Ontario Ministry of Agriculture and Food. d ,Herrelle, F. (1917). Sur un microbe invisible antagonistic des bacilles dysenteriques. Con1ptes Rendus de l'Academie des Sciences, 165,373-375. d 'Herrelle, F. (1921). Le Bacteriophage: Son Role dans l'Immunite, part II, chapter III. Paris, France: Masson. d 'Herrelle, F. (1926). The Bacteriophage and its Behaviour, 497 pp. Baltimore, MD: Williams and Wilkins. de Wet, A.F. (1998). United States Department of Agriculture. Pear Roostock BPI. PPI0231. Deckers, T., Faust, M., & Miller, S.S. (1992). Influence of chemical growth regulation on secondary blossom fonnation on young pear trees 'Conference' and 'Doyenne du C0111ice'. Acta Horticulturae, 322,293-297. DeIcher, A.L., Harmon, D., & Salzberg, S.L. (1999). Improved microbial gene identification with GLIMMER. Nucleic Acids Research, 27, 4636-4641. Dellagi, A., Brisset, M.N., Paulin, J.P., & Expert, D. (1998). Dual role of desferrioxalnine in Erwinia an1ylovora pathogenicity. Molecular Plant-Mjcrobe Interactions, 11,734-742. 284 Dellagi, A., Reis, D., Vian, B., & Expert, D. (1999). Expression of the fen·ioxamine receptor gene of Erwinia amylovora CFBP 1430 during patho,gellesis. Molecular Plant-Microbe Interactions, 12, 463-466. Demeke, T., & Adams, R. (1992). The effects of plant polysaccharides and buffer additives on PCR. BioTechniques, 12,333-334. Denning, W. (1794). On the decay of apple trees. New York 'Society for the Pronl0tion of Agricultural Arts and Manufacturers Transaction, 2, 219-222. Desai, C., & Madanlwar, D. (2006). Extraction of inhibitor-free Inetagell0nlic DNA fronl polluted sedinlents, cOlnpatible with nl0lecular diversity analysis using adsorption and ion-exchange treatments. Bioresource Technology, 98,761-768. Dimova, A.M. (1990). Chemical control of fire blight blosSOll1 infection under field conditions in Cyprus. Acta Horticul rae~ 273, 377-382. Disinfectants and antiseptics. (1994). In T. D. Brock, M. T. Madigan, J. M. Martinko, & J. Parker (Eds.), Biologv of Microorgallisms. (pp. 350-352). Prentice-Hall, Inc. Dondini, L., Pierantoni, L., Gaiotti, F., Chiodini, R., Tartarini, S., Bazzi, C., & Sansavilli, S. (2004). Identifying QTLs for fire-blight resistance via a 'european pear (Pyrus comlnunis L.) genetic linkage map. Molecular Breeding, 14,407-418. Dowd, S.E., Pillai, S.D., Wang, S., & Corapcioglu, M.Y. (1998). Delineating the specific influence of virus isoelectric point and size on virus adsorptioll and transport through sandy soils. Applied and Ellvironmental Microbiology, 64,405-'410. Dye, D.W. (1968). A taxonomic study of the genus Erwinia. I. The "amylovora" group. New Zealand Journal of Science, 11,590-607. Dye, D.W. (1969a). A taxonomic study of the genus Erwinia. II. The "carotovora" group. New Zealand Journal of Science, 12, 81-97. Dye, D.W. (1969b). A taxonolnic study of the genus Erwinia. III. The "herbicola" group. New Zealand Journal of Science, 12(223-236) Dye, D.W. (1969c). A taxonomic ,study of the genus Erwinia. IV. "Atypical" Erwinias. New Zealand Journal of Science, 12(833-839) EBI Food Safety. LISTEXTM PI 00 phages for control of Listeria approved by U'S FDA as GRi\S. The Hague, Netherlands, October 24th, 2006. 285 Eden-Green, S.J. (1972). Studies in fire blight disease of apple, pear and hawthorn (Erwinia amylovora (Burrill) Winslow et al). PhD thesis. University of London; Eden-Green, S.J., & Billing, E. (1972). Fireblight: occurrence of bacterial strands on various hosts under glasshouse conditions. Plant Pathology, 12, 121-123. Edstrom. (2003). Forms of chlorine in water. Edstrom Illdustries, Inc. Waterford, WI. EI-Mabrouk, N. & Lisacek, F. (1996). Very fast identificatiol1 of RNA nl0tifs in -genonlic DNA. Application to tRNA search in the yeast genolne. Journal ofM:olecular Biology., 264, 46-55. El Nasr, S., Hamdy, I., & Ali, M.H. (1990). Fire blight incidence in pear orchards and its control in Egypt. Acta Horticulturae, 273, 392 Engelbrecht, R.S., Weber, M.J., Salter, B.L., & Schmidt, C.A. (1980). COll1parative inactivation of viruses by chorine. Applied and Environmental Microbiology., 40,249-256. Environmental Protection Agency (2006a). Pantoea agglomerans Strain C9-1; Exemption froln the Requirement of a Tolerance. Federal Register: 71 (80): 24590-24596. Environmental Protection Agency (2006b). Pantoea agglomerans Strain £325; Exemption from the Requirement ofa Tolerance. Federal Register: 71(182): 54928-54933. Erskine, J.M. (1973). Characteristics of Erwinia amylovora bacteriophage and its possible role in the epidemiology of fire blight. Canadian Journal of Microbiology, 19, 837-845. Erskine, J.M., & Lopatecki, L.E. (1975). In vitro and in vivo interactions between Elwinia amylovora and related saprophytic bacteria. Canadian Journal 'of Microbiolo:gv, 21, 3541. Evans, A.C. (1940). The potency of nascent Streptococcus bacteriophage B. Journal of Bacteriology, 39, 597-604. Ewing, W.H., & Fife, M.A. (1972). Enterobacter agglomerans (Beijerinck) cOlnb. nov. (the herbicola-Iathyri bacteria). International Journal of Systenlatic Bacteriology., 22, 4-11. Falkenstein, H., Bellemann, P., Walter, S., Zeller, W., & Geider, K. (1988). Identification of Erwinia amylovora, the fireblight pathogen, be colony hybridization with DNA fl·Ol11 plasnlid pEA29. Applied and Environlnental Microbiology'l 54, 2798-2802. Farabee, G.J., & Lockwood, J.L. (1958). Inhibition of Erwinia amylovora by Bacterium sp. isolated from fire blight cankers. Phytopathology., 48, 2,09-211. 286 Finn,R. D., Mistry, J., Schuster-Bockler, G., Griffiths-Jones, S., Hollich, V., Lassmann, T., Moxon, S., Marshall, M., Khanna,A., Durbin, R., Eddy, S.R., Sonnhamnler, E.L.L., & Bateman, A. (2006).Pfam: clans, web tools and services. Nucleic Acids Research, 34, D247-D251. Foulds, LV., Granacki, A., Xiao, C., Krull, U.J., Castle, A.J., & Horton, P.J. (2002). Quantification of microcystin-producing cyanobacteria and E. coli in water by 5'-nuclease PCR. Journal of Applied Microbiologv, 93, 825-834. Fraser, J.S., Yu, Z., Maxwell, K.L., & Davidson, A.R. (2006). Ig-like dOlnains on bacteriophages: a tale of promiscuity and deceit. Journal of Molecular Biology, 359, 496-507. Gaeng, S., Scherer, S., Neve, H., and Loessner, M.J. (2000). Gene cloning and expressiol1 alld secretion of Listeria monocytogenes bacteriophage-lytic enzymes ill Lactococcus lactis. Applied and Environmental Microbiology, 66, 2951-2958. Gaudriault, S., Malandrin, L., Paulin, J.P., & BanlY, M.A. (1997). DspA, an essential pathogenicity factor of Erwinia amylovora showing homology with AvrE of Pseudomonas syringae, is secreted via the Hrp secretion pathway in a DspB-dependent way. Molecular Microbiology, 26, 1057-1069. Geider K, Geier G, Bellemann P, Bernhard F, Bugert P, & Metzger M. (1993). Exopolysaccharides in pathogenicity of Erwinia amylovora. Acta Horticulturae'l 33'8, 255-262. Geider, K. (2000). Exopolysaccharides of Erwinia amylovora: Structure, biosynthesis, regulation, role in pathogenicity of amylovoran and levan. In J. L. Vann.este (Ed.), Fire Blight: the Disease and its Causative Agent, Erwinia amylovora. (pp. 117-140). Wallingford, UK: CAB International. Geier, G., & Geider, K. (1993). Characterization and influence on virulence of the levansucrase gene from the fire blight pathogen Erwinia amylovora. Physiological and Molecular Plant Pathology, 42, 387-404. Geier, M.R., Trigg, M.E., & Merril, C.R. (1973). The fate of bacteriophage lambda in non-immune germfree mice. Nature, 246, 221-222. Geller, B.L., Kraus, J., Schell, M.D., Hornsby, M.J., Neal, J.J., & Ruch, F.E. (1998). High titer, phage-neutralizing antibodies in bovine colostrunl that prevent lytic infection of Lactococcus lactis in fermentatiol1s of phage-contanlinated milk. Journal of Dairy Science, 81, 895-900. 287 Giddens, S.R., Houliston, G.J., & Mahanty, H.K. (2003). The influence of antibiotic production and pre-emptive colonization on the population dynamics of Pantoea agglomerans (Erwinia herbicola) Ehl087 and Erwinia am~vlo ra in plantae Environmental Microbiology, 5, 1016-1021. Gill, J.J. (2000). Biological control of Erwinia an1vlovora usin.g bacteriophages. University of Guelph: Masters Thesis. Gill, J.J., Svircev, A.M., Smith, R., & A. J. Castle. (2003). Bacteriophages of Erwinia amylovora. Applied Environmental Microbiology, 69,2133-2138. Gill, J.J., Pacan, J.C., Carson, M.E., Leslie, K.E., Griffiths, M.\V., & Sabour, P.M. (2006a). Efficacy and pharmacokinetics of bacteriophage therapy in treatnlent of subclini'cal Staphylococcus aureus mastitis in lactating dairy cattle. Antilnicrobial Agents and Chemotherapy, 50,2912-2918. Gill, J.J., Sabour, P.M., Leslie, K.E., and Griffiths, M.W. (2006b). Bovine why proteins inhibit the interaction of Staphylococcus aureus and bacteriophage K. Journal of Applied Microbiology, 101,377-386. Giorgi, S., & Scortichini, M. (2005). Molecular characterization of Erwinia amylovora strains from different host plants through RFLP analysis and sequencing of hrpN and dspA/E genes. Plant Pathology, 54, 789-798. Gomori, G. (1955). Preparation of buffers for use in enzynle studies. Methods in ~vg,olmyznE 138-146. 1, Goodridge, L. D. (2004). Bacteriophage biocontrol of plant pathogens: fact or fiction? Trends in Biotechnology, 22, 384-385 Gouk, S.C., Bedford, R.J., & Hutshins, S.Q. (1996). Effect of apple flower phenology on growth of Erwinia amylovora. Phytopathology, 86, S42 Grant, C.R., & Rees, T. (1981). Sorbitol metabolism by apple seedlings. Phytochemistry, 20, 1505-1511. Grimont, P.A.D. & Grimont, F. (2005). Genus XXIII. Pantoea. In D. J. Brenner, N. R. Krieg, & J. R. Staley (Eds.), Bergey's Manual of Systenlatic Bacteriology. (pp. 713-721). Gross, M., Geier, G., Rudolp, K., & Geider K. (1992). Levan and levan'sucrase -synthesized by the fireblight pathogen Erwinia am)Jlovora. Physiological and M·olecular Plant Pathology, 40,371-381. 288 Grundy, F.J. & Howe, M.M. (1984). Involvement of the invertible G segnlent in bacteriophage mu tail fiber biosynthesis. Virology, 134, 296-317. Guttman, B., Raya, R., & Kutter, E. (2005). Basic phage biology. In E. Kutter & A.Sulakvelidze (Eds.), Bacteriophages: Biology and Applications. (pp. 29-66). Boca Raton, FL: CRC Press. Halaby, D.M., Poupon, A., & Momon, J. (1999). The immunoglobulin fold fanlily: 'Sequence 12, 563-571. analysis and 3D structure cOlllparisons. Protein Engier ~ Hao, Y., Dick, W.A., & Tuovinen, O.H. (2002). PCR amplification of 16S rDNA sequenees in Fe-rich sediment of coal refuse drainage. Biotechnology Letters, 24, 1049-1053. Hartman, J.R., Paulin, J.P., Parisi, L., & Thomson, S. (2000). INRA and apple disease research in the Loire Valley region of France. Plant Disease, 84,928-936. Halwng, J.S., Fulbright, D.W., & Klos, E.J. (1988). Cloning ofa bacteriophage polysaccharide depolymerase gene and its expression in Erwinia amylovora. Molecular Plant-Microbe Interactions, I, 87-89. Hauben, L., Moore, E.R.B., Vauterin, L., Steenackers, M., Mergaeli, J., Verdonck, L., & Swings, J. (1998). Phylogenetic position of phytopathogells within the Enterobacteriaceae. Systematic and Applied Microbiology, 21, 384-397. Hauben, L., & Swings, J. (2005). Genus XIII. Erwinia. In D. J. Brenner, N. R. Krieg, & J. R. Staley (Eds.), Bergey's Manual of Systematic Bacteriologv. (pp. 670-679). Heimann, M.F., & Worf, G.L. (1985). Fire blight of raspberry caused by Erwinia am}'lovora in Wisconsin. Plant Disease, 69, 360 Hildebrand, E.M., & Heinicke, A.J. (1937). Incidence of fire blight in young apple trees in relation to orchard practices. New Yock Agricultural Station Memo 203, Holcomb, G. E. (1998). First report of fire blight on Indian Hawtholll cultivar Olivia in 82,1402. Louisiana. Plant Disea ~ Holland, J. & DOlllingo, E. (1998). Origin and evolution of viruses. Virus Genes, 16,13-21. Hu, T.L. (1998). A comparison of two methods to recover phages fronl soil salllples. Bioresource Technology, 65, 167-169. Huang, P.Y., & Goodman, R.N. (1970). Morphology and ultrastructure ofnonnal rod-shaped and filalllentous forms of Erwinia amylovora. Journal of Bacteriology, 102, 862-866. 289 Hurst, C.J., Schaub, S.A., Sobsey, M.D., Farrah, S.R., Gerba, C.P., Rose, J.B., Goyal, S.M., Larkin, E.P., Sullivan, R., Tienley, J.T., O'Brien, R.T., Safferman, R.S., Morris, 1\1.E., Welling, F.M., Lewis, A.L., Berg, G., Britton, P.W., & Winter, J.A. (1991). Multilaboratory evaluation of methods for detecting enteric viruses in soils. Applied and Environmental Microbiology, 57, 395-401. Ibekwe, A.M., Watt, P.M., Grieve, C.M., Sharma, V.K., & Lyons, S.R. (2002). Multiplex fluorogenic real-time PCR for detection and quantification of Escherichia coli 0157:H7 in dairy wastewater wetlands. Applied and Environnlental Microbiology, 68, 4853-4862. Inchley, C.J. (1969). The activity OflTIOUSe Kupffer cells following intravenous injection ofT4 bacteriophage. Clinical and Experimental ImmunoloQY, 5, 173-187. Iriarte, F.B., Balogh, B., Momol, M.T., Smith, L.M., WilsOll, M., & Jones, J.B. (2007). Factors affecting the survival of bacteriophage on tomato leaf surfaces. Applied and Environmental Microbiology, 73, 1704-1711. Ishimaru, C.A., Klos, E.J. (1984). A new medium for detecting Erwinia am)Jlovora and its use in epidemiological studies. Phytopathology, 74, 1342-1345. Ishimaru, C.A., Klos, E.J., & Brubaker, R.R. (1988). Multiple antibiotic production by Erwinia herbicola. Phytopathology, 78, 746-750. Jardine, P.J., & Anderson, D.L. (2006). DNA packaging in double-stranded DNA phages. In R. Calendar (Ed.), The Bacteriophages. (pp. 49-65). New York, NY: Oxford University Press. Jeng, R.S., Svircev, A.M., Myers, A.L., Beliaeva, Hunter, D.M. and Hubbes, M. (2001). The use of 16S and 16S-23S rDNA to easily detect and differentiate common 'GraIn-negative orchard epiphytes. Journal of Microbiological Methods, 44, 69-77. Jensen, E.C., Schrader, H.S., Rieland, B., Thonlpson, T.L., Lee, K.W., N"ickerson, K.W. & Kokjohn, T.A. (1998). Prevalence of broad-host-range lytic bacteriophages of Sphaerotilus natans, Escherichia coli, and Pseudomonas aeruginosa. Applied and Environmental Microbiology, 64,575-580. Jin, M., Wright, S.A.I., Beer, S.V., & Clardy, J. (2003). The biosynthetic gene cluster of pantocin A provides insights into biosynthesis and a tool for screening. Angewandte Chemie, 115,3008-3011. Jock, S., Donat, V., Lopez, M.M., Bazzi, C., & Geidel", K. (2002). Following spread of fire blight in Western, Central and Southern Europe by Il101ecular differentiation 'of Erwinia amylovora strains with PFGE analysis. Environnlental Microbiology, -4, 106-114. 290 Jock, S., Jacob, T., Kim, W.-S., Hildebrand, M., Vosberg, H.P., & Geidel", K. (2003a). Instability of short-sequence DNl.t\ repeats of pear pathogenic Erwinia strains from Japan alld Erwinia amylovora fruit tree and raspberry strains. Molecular Genetics and Genomics, 268, 739-49. Jock, S., Kim, W.-S., Barny, M.-A., & Geidel", K. (2003b). Molecular characterization of natural Erwinia pyrifoliae strains deficient in hypersellsitive response. Applied and Environmental Microbiology, 69, 679-682. Jock, S., & Geidel", K. (2004). Molecular differentiation of Erwinia amylovora strains froll1 North America and of two Asian pear pathogens by analyses ofPFGE patterns and hrpN genes. Environmental Microbiology, 6, 480-490. Jock, S., Langlotz, C., & Geidel", K. (2005). Survival and possible spread of Erwinia anlylovora and related plant-pathogenic bacteria exposed to environnlental stress conditions. Journal of Phytopathology, 153, 87-93. Johnson, K.B., Stockwell, V.O., Burgett, D.M., Sugar, D., & Loper, J.E. (1993a). Dispersal of Erwinia amylovora and Pseudomonas fluorescens by honey bees from hives to apple and pear blossoms. Phytopathology, 83, 478-484. Johnson, K.B., Stockwell, V.O., McLaughlin, R.J., Sugar, D., Loper, J.E., & Roberts, R.G. (1993b). Effect of antagonistic bacteria on establishlnent of hOlley bee-dispersed Erwinia amylovora in pear blossoms and on fire blight control. Phvtopa lgy~ 83, 995-1002. Johnson, K.B., & Stockwell, V.O. (1998). Management of fire blight: A case study in microbial ecology. Annual Review of Phytopathology, 36,227-248. Johnson, K.B., & Stockwell, V.O. (2000). Biological control of fire blight. In J. L. Vanlleste (Ed.); Fire Blight. (pp. 319-337). Wallingford, UK: CAB Interllational. Johnson, K.B., Stockwell, V.O., Sawyer, T.L., & Sugar, D. (2000). AsseSSlllent of environmental factors influencing growth and spread of Pantoea agglomerans on and among blossoms of pear and apple. Phytopathology, 90, 1285-1294. Johnson, K.B., Stockwell, V.O., & Sawyer, T.L. (2004). Adaptation of fire blight forecasting to optimize the use of biological controls. Plant Disease, 88,41-48. Jumel, K., Geidel", K., & Harding, S.E. (1997). The solutioll molecular weight and shape of the bacterial exopolysaccharides amylovoran and stewartan. International Journal of Biological Macromolecules, 20,251-258. Kachadourian, R., Dellagi, A., Laurent, J., Bricard, L., Kunesch, G., & Expert, D. (1996). 291 Desferrioxamine-dependent iron transpolt in Erwinia amylovora CFBPI430: cloning of the gene encoding the ferrioxamil1e teceptor FoxR. Biometals, '9, 143-150. Kapfbammer D, Blass, J., Evers, S., & Reidl, J. (2002). Vibrio cholerae phage K139: ~ol1pet genome sequence and comparative genol11ics of related phages. Journal of Bacteriologv, 184, 6592-6601. Kasman, L.M., Kasman, A., Westwater, C., Dolan, J., Schmidt, M.G., & Norris, J.S. (2002). Overcoming the phage replication threshold: a nlathematical nl0del with inlplications for phage therapy. Journal ofVirologv, 76, 5557-5564. Katcher, H.L. & Schwartz, I. (1994). A distinctive property of Tth DNA polymerase: enzymatic amplification in the presence of phenol. BioTechniques, 16, 84-92. Keams, L.P., & Mahanty, H.K. (1998). Antibiotic -production by Erwinia herbicola Ehl 087: Its role in inhibition of Erwinia amylovora and partial characterization of antibiotic biosynthesis genes. Applied and Environmental Microbiology, 64, 1837-1844. Keil, H.L., & Shear, C.B. (1972). Influence of nitrogen and potassium on fire blight severity in Bartlett pear trees in sand culture. Phytopahlo_gy~ 62, 768-796. Keil, H.L., & van del" Zwet, T. (1972a). Aerial strands of Erwinia am.ylovora: 8tm'cture and enhanced production by pesticide oil. Phytopalgy~ 62, 355-361. Keil, H.L., & van del" Zwet, T. (1972b). Recovety of Erwinia amylovora fron1 synlptomless stems and shoots of Jonathan apple and Baltlett pear trees. Phytopatholo.gy, 62, 39-42. Kelln, R.A., & Warren, A.J. (1971). Isolation and propelties of a bacteriophage lytic for a wide range of pseudomonads. Canadian Journal of Microbiology, 17,677-682. Khan, M.A., Duffy, B., Gessler, C., & Patocchi, A. (2006). QTL l11apping of fire .blight resistance in apple. Molecular Breeding, 17, 299-306. hrpA and hrpC operons ofErwinia amylovora Kim, J.F., Wei, Z.M., & Beer, S.V. (1997). l~he encode cOlnponents of a type III pathway that secretes harpin. Journal of Bacteriologv, 179, 1690-1697. Kim, W.S., Gardan, L., Rhim, S.L., & Geidel", K. (1999). Erwinia pyrifoliae sp. nov., a novel pathogen that affects Asian pear trees (Pyrus pyrifolia Nakai). International Journal of Systematic Bacteriology, 49, 899-905. Kim, W.-S., Hildebrand, M., Jock, S., & Geidel", K. (2001 a). Molecular comparison of pathogenic bacteria from pear trees in Japan and the fire blight pathogen Erwinia 292 amylovora. Microbiology, 147,2951-2959. Kim, W.-S., Jock, S., Paulin, J.P., Rhiln, S.L. & Geidel", K. (200]b). Molecular detection and differentiation of Erwinia pyrifoliae and host range analysis of the Asian pear pathogen. Plant Disease, 85,1183-1188. Kim, W.-S., Salm, G., and Geidel", K. 2004. Expression of bacteriophage phiEa]h lysozynle in Escherichia coli and its activity in growth illhibition of Erwinia amylovora. Microb l gy~ 150,2707-2714. Kim, W.-S., Schollmeyer, M., Nimtz, M., Wray, V. & Geidel", K. (2002). Genetics of biosynthesis and structure of the capsular exopolysaccharide from the Asian pear pathogen Erwinia pyrifoliae. Microbiology, 148, 4015-4024. Ko, K. , Norelli, J.L., Reynoird, J.P., Aldwinckle, H.S., & Brown, S.K. (2002). T4 lysozylne and attacin genes enhance resistance of transgellic 'Galaxy' apple against Erwinia amylovora. Journal of the American Society for HOliicultural Science, 127,515-519. Koseoglu, A.T., Tokmak, S., & Momol, M.T. (1996). Relationships between the incidence of fire blight and nutritional status of pear trees. Journal of Plant Nutrition, ] 9, 5] -61. Kratz, A., Greenberg, D., Barki, Y., Cohen, E., & Lifshitz, M:. (2003). Pantoea agglomerans as a cause of septic arthritis after palnl tree thorn injury; case report and literature review. Archives of Disease in Childhood, 88, 542-544. Kuhn, J., Suissa, M., Chiswell, D., Azriel, A., Bernlan, B., Shahar, D., Rezllick, S., Sharf, R., Wyse, J., Bar-on, T., Cohen, I., Giles, R., Weiser, I., Lubinsky-Mink, 'S., & Ulitzur, 'So (2002). A bacteriophage reagent for Salmonella: molecular studies on :Felix 0]. International Journal of Food Microbiology, 74,217-227. Kunisawa, T. (1992). Synonymous codon preferences in bacteriophage T4: a distinctive use of transfer RNAs from T4 and froln its host Escherichia coli. Journal of Theoretical Biology, 159, 287-98. Kuo, T., Chang, L., Yang, C., & Yang, S. (197]). Bacterialleafblight ofric'e plant. IV. Effect of bacteriophage on the infectivity of Xanthomonas oryzae. Botanical Bulletin of Acadelnia Sinica, 12, 1-9. Kutter, E., Raya, R., & Carlson, K. (2005). Molecular nlechanislns of phage infection. In E. Kutter & A. Sulakvelidze (Eds.), Bacteriophages: biologv and applications. (pp. 165222). Boca Raton, FL: CRC Press. Laby, R.J., & Beer, S.V. (1992). Hybridization and functional conlplenlentation ofth'e hrpgene 293 cluster from Erwinia amylovora strain Ea321 with DNA of the other bacteria. Molecular Plant-Microbe Interactions, 5,412-419. Lanning, S., & Williams, S.T. (1982). Methods for the direct isolation and ellunleration of actinophages in soil. Journal of General Microbiology, 128, 2063-2071. Le Minor, L. (1984). Salmonella. In N. R. Krieg & J.G. Holt (Eds). (pp 427-458) Bergey's Manual of Systematic Bacteriology (vol. 1). Baltimore, MD: Willianls and Wilkills. Lefler, E., & Kott, Y. (1974). Virus retention and survival in sand. In J. F. Malina & B. P. Sagik (Eds.), Virus survival in water and wastewater systems. (pp. 84-91). Center for Research in Water Resources: University of Texas, Austin. Lelliot, R.A., & Dickey, R.S. (1984). Genus VII. Erwinia. In N. R. Krieg & J. G. Holt (Eds.), Bergey's Manual of Systematic Bacteriology. (pp. 469-476). Baltinl0re, MD: WillialTIs and Wilkins. Lespinasse, Y., & Aldwinckle, H.S. (2000). Breeding for resistance to tire blight. In J. J...I. Vanlleste (Ed.), Fire Blight: the Disease and its Causative ~Agent'l Erwinia amvlovora. (pp.253-273). Wallingford, UK: CAB International. Lewis, L.N., & Kenworthy, A.L. (1962). Nutritional balallce as related to leaf conlposition and fire blight susceptibility in the Bartlett pear. American Society of Horticultural 'Science Proceedillgs, 81, 108-115. Lewis, L.N., Tolbert, N.E., & Kenworthy, A.L. (1964). Influence ofmilleral nutrition on the anlino acid composition of Bartlett pear trees. Proceedings of the American Society for Horticultural Science, 83, 185-192. Li, T.H., & Li, S.H. (2005). Leaf responses of micropropagated apple plants to water "stress: nonstructural carbohydrate composition and regulatory role of 11letabolic enzynles. Tree Physiology, 25, 495-504. Lindberg, A.A., & Holme, T. (1969). Influence of 0 side chains on the attachment of the Felix 01 bacteriophage to Salmonella bacteria. Journal of Bacteriology, 99, 513-519. Lindemann, J., & Suslow, T.V. (1987). Competition between ice nu'cleation-active wild type and ice nucleation-deficient deJection mutant strains of Pseudomonas syringae and P. jluorescens biovar I and biological control of frost injury on strawberry blossolTIs. Phytopathology, 77, 882-886. Lindh, E., Kjaeldgaard, P., Frederiksen, W., & Ursing, J. (1991). Phell0typical properties of Enterobacter agglomerans (Pantoea agglomerans) froll1 hUl11an, anilnal and plant 294 sources. Acta Pathologica, Microbiolo.gica et Imunolgica~ 99, 347-352. Lindow, S.E., McGourty, G., & Elkins, R. (1996). Interactions of antibiotics with Pseudomonas jluorescens strain A506 in the control of fire blight and frosty illjUry to pear. Phytopathology, 86, 841-848. Lindow, S.E., & Suslow, T.V. (2003). Temporal dynamics of the biocontrol agent Pseudomonas jluorescens strain A506 in flowers in inoculated pear trees. Phytopathology, 93, 727737. Llop, P., Bonaterra, A., Penalver, J., and Lopez, M.M. (2000). Development of a highly sel1sitive nested-PCR procedure using a single closed tube for detection of Erwinia amylovora in asymptomatic plant material. Applied and Environmental Microbiology, 66, 2071-2078. Llop, P., Donat, V., Lopez, M.M., Cabrefiga, J., Ruz, L., Montesinos, E., Palomo, J.L., & Rodriguez, M. (2006a). An Erwinia amylovora strain without plasmid pEA29 isolated in a nursery from hawthorn. Acta HOliiculturae, 704, 431-438. Llop, P., Donat, V., Rodriguez, M., Cabrefiga, J., Ruz, L., Palolno, J.L., Montesinos, E., & Lopez, M.M. (2006b). All indigenous virulent strain of Erwinia amylovora lacking the ubiquitous plasmid pEA29. Phytopathology, 96,900-907. Loeffler, J.M., Nelson, D., & Fischetti, V.A. (2001). Rapid killing of Streptococcus pneumoniae with a bacteriophage cell wall hydrolase. Science, 294, 2170-2172. Longstroth, M. (2000). The 2000 fire blight epidemic in southwest Michigan. Available online: http://www.canr.msu.edu/vanburenlfb2000.htm {Accessed January 2007]. Loper, J.E., Henkels, M.D., Roberts, R.G., Grove, G.G., Willet, M.J., & Snlith, T.J. (1991). Evaluation of streptomycin, oxytetracycline, and copper resistance of Erwinia amylovora isolated from pear orchards in Washington State. Plant Disease, 75,287-290. Lopez, C., & Fucikovsky, L. (1990). Distribution of fire blight in Mexico and the identification of the bacterium on pear and pyracantha. Acta Horticulturae, 273,33-36. Lowe, T.M. & Eddy, S.R. (1997). tRNAscan-SE: a program for inlproved detection of transfer RNA genes in genomic sequence. Nucleic Acids Research, 25, 955-964. Malnoy, M., Faize, M., Venisse, J.S., Geidel", K., & Chevreau, E. (2005). Expression of viral EPS-depolymerase reduces fire blight susceptibility in transgenic pear. Plant Cell Reports, 23 , 632-638. Manfioletti, G., & Schneider, C. (1988). A new and fast tn'ethod for preparing high quality 295 lambda DNA suitable for sequencing. Nucleic Acids Research, 16, 2873-2884. Manninen, A., Hall, G.E.M., Vaive, J.E., & MacLaurin, A.1. (1996). Analytical aspects of the application of sodium pyrophosphate reagent in the specific extractio11 of the labile organic component of humus and soils. Journal of Geochemical Exploration, 56, 23-36. Manulis, S., Zutra, D., Kleitman, F., Dror, 0., David, I., Zilberstaine, M., & Shabi, E. (1998). Distribution of streptomycin-resistant strains of Erwinia amylovora in Israel and occurrence of blossom blight in the autumn. Phytoparasitica, 26, 223-230. Manulis, S., Kleitman, F., Shtienberg, D., Shwartz, H., Oppenhein, D., Zilberstaine, M., & Shabi, E. (2003). Changes in sensitivity of Erwinia amylovora populations to streptomycin and oxolinic acid in Isreal. Plant Disease, 87, 650-654. Marcuk, L.M., Nikiforov, V.N., Scerbak, J.F., Levitov, T.A., Kotljarova, R.I., Mamnlsina, M.S., Dovydov, S.U., Monsur, K.A., Rahman, M.A., Latif, M.A., Northrup, R.S., Cash, R.A., Huq, I.., Dey, C., & Phillips RA. (1971). Clinical studies of the use of bacteriophage in the treatment of cholera. Bulletin of the World Health Organization, 45,77. Marques, L. L. R., Ceri, H., Manfio, G. P., Reid, D. M., & Olson, M. E. (2002). Characterization ofbiofilm fonnation by Xylellafastidiosa in vitro. Plant Disease, 86,633-638. Martin, P.R., Shea, R.J., & Mulks, M.H. (2001). Identification of a plasmid-encodied gene fro111 Haemophilus ducreyi which confers NAD independence. Joulual ofBacteriolo,gv, 183, 1168-1174. Matsuda, T., Freeman, T., Hilbert, D., Duff, M., Fuortes, M., Stapleton, P., & Daly, J. (2005). Lysis-deficient bacteriophage therapy decreases endotoxin and inflanlmatory mediator release and improves survival in a murine peritonitis 111odel. Surgery, 137, ,639-646. Matsuura, T., Shinohara, H., Inoue, Y., Azegami, K., Tsushima, S., Tsukanloto, T., & Mizuno, A. (2007). Erwinia isolates from the bacterial shoot blight of pear in Japan are closely related to Erwinia pyrifoliae based on phylogenetic analyses of gyrB and rpoD g·enes. Journal of General Plant Pathology, 73, 53-58. Maxson-Stein, K., McGhee, G. C., Smith, J. J., Jones, A. L., and Sundin, G. W. (2003). Genetic analysis of a pathogenic Erwinia sp isolated from pear in Japan. Phvtopatholo.gy, 93, 1393-1399. McGhee, G. C., Schnabel, E. L., Maxson-Stein, K., Jones, B., Strol11berg, V. K., Lacy, G. H., and Jones, A. L. (2002). Relatedness of chromosonlal and plasmid DNAs of Erwinia pyrifoliae and Erwinia amylovora. Applied and Environmental Microbiology, 68, 61826192. 296 McGuire, M.R., Tamez-Guerra, P., Behle, R.W., & Streett, D. (2001). Conlparati¥e field stability of selected entonl0pathogenic baculovirus formulations. Journal of Economic Entomology, 94, 1037-1044. McKeague, J.A., Brydon, J.E., & Miles, N.M. (1971). Soil Science Society of Amelica Proceedings, 35, 33-38. McManus, P.S., & Jones, A.L. (1994). Epidemiology and genetic analysis of streptomycinresistant Erwinia amylovora from Michigan and evaluation of oxytetracycline for control. Phytopathology, 84, 627-633 McManus, P.S., & Jones, A.L. (1995a). Detection of Erwinia amylovora by nested PCR and PCR-dot-blot and reverse-blot hybridizations. Phytopathology, 85,618-623. McManus, P.S., & Jones, A.L. (1995b). Genetic fingerprinting of Erwinia amo/vlovora strains isolated from tree-fruit crops and Rubus spp. Phytopathology, 85, 1547-1553. McManus, P.S., Jones, A.L. & Bonn, W.G. (1996). Detection of Erwinia amylovora by llested PCR and PCR-dot-blot and reverse blot hybridizations. Acta Horticulturae'l 411,87-90. McManus, P.S., Stockwell, V.O., Sundin, G.W., & Jones, A.L. (2002). Antibiotic use in plant agriculture. Annual Review ofPhvtopathology, 40, 443-465. Meng, X.D., Bonasera, J.M., Kim, J.F., Nissinen, R.M., & Beer, S.V. (2006). Apple proteins that interact with DspA/E, a pathogenicity effector of Erwinia amylovora, the fire blight pathogen. Molecular Plant-Microbe Interactions'l 19, 53-61. Merril, C.R., Fliedlnan, T.B., Attallah, A.F., Geier, M.R., Krell, K., & Yarkin, R. (1973). Isolation of bacteriophages from commercial sera. In Vitro'l 8,91-93. Middleboe, M., Jorgensen, N.O.G. & Kroer, N. (1996). EtTects of viruses on nutriellt tunlover and growth efficiency of noninfected marine bacterioplankton. Applied and Environmental Microbiology, 62, 1991-1997. Milch, H. & Fomosi, F. (1975). Bacteriophage contamination of live poliovirus vac'cine. Journal of Biological Stalldardization, 3, 307-310. Miller, D.N., Bryant, J.E., Madsen, E.L., & Ghiorse, W.C. (1999). Evaluation and optill1ization of DNA extraction and purification procedures for soil and sedill1ent sanlp]es. Applied and Environmental Microbiology, 65, 4715-4724. Miller, E.S., Heidelberg, J.F., Eisen, J.A., Nelson, W.C., Durkin, A.S., Ciecko, A., Feldblyull1, T.V., White, 0., Paulson, I.T., Nierman, W.C., Lee, J., Szczypinski, B., & Fraser, C.M:. (2003a). Complete genome sequence of the broad-host-range vibriophage KVP40: Comparative genomics of a T4-related bacteriophage. Journal of Bacteriolo,gy, 185, 5220-5233. Miller, E.S., Kutter, E., Mosig, G., Arisaka, F., Kunisawa, T., & Riiger, W. (2003b). Bacteriophage T4 genome. Microbiology and M-olecular Biologv Reviews, 67, 86-156. Miller, P.W. (1929). Studies of fire blight of apple in Wisconsin. Journal of A,gricultural Research, 39, 579-621. Monsur, K.A., Rahman, M.A., Huq, F., Islam, M.N., Northrup, R.S., & Hirschhorn, N. (1970). Effect of massive doses of bacteriophage on excretion of Vibrio, duration of dialThoea and output of stools in acute cases of cholera. Bulletin of the World Health Organizatiol1, 42, 723, 1970. Moody, E.E, Trousdale, M.D., Jorgenson, J.H., & Shelokov, A. (1975). Bacteriophages and endotoxin in licensed live-virus vaccines. Joulllal of Infectious Disease, 131, 588-591. Mosig, G., & Eiserling, F. (2006). T4 and related phages: structure and developmellt. In R. Calendar (Ed.), The Bacteriophages. (pp. 225-267). New York, NY: Oxford University Press. Munsch, P., & Olivier, J.M. (1995). Biocontrol of bacterial blotch of the cultivated 111ushroonl with lytic phages: some practical considerations. In T. J. Elliot (Ed.), Sciellce and cultivation of edible fungi, Vol. II: Proceedings of the 14th International COl1gress. (pp. 595-602). Murdoch, W.W. & Briggs, C. J.(1996). Theory for biological cOlltrol: Re-cent developnlents. Ecology, 77,2001-2013. Newman, K.L., Almeida, R.P.P., Purcell, A.H., & Lindow, S. E. (2003). Use of a Green Fluorescent Strain for Analysis of Xylella fastidiosa Colonization of Vilis vinifera. Applied and Environmental Microbiology, 69, 7319-7327. Nightingale, A.A. (1932). Some chemical constituents of apple associated \vith susceptibility to fire blight. New Jersey Agricultural Experiment Station Bulletil1 613., Nimtz, M., Mort, A., Domke, T., Wray, V., Zhang, Y.X., Qiu, F., Coplin, D., Geider, K., & Zhang, Y.X. (1996). Structure of amylovoran, the capsular exopolysa-ccharide fron1 the fire blight pathogen Ervvinia amylovora. Carbohydrate Research, 287, 59-76. Nissinen, R.M., Ytterberg, J., Bogdanove, A.J., van Wijk, K.J., & Beer, S.V. (2007). Allal)Tses of the secretomes of Erwinia amylovora al1d selected hrp ll1utants reveal novel type III 298 secreted proteins and an effect of HrpJ on extracellular harpin levels. Molecular Plant Pathology, 8,55-67. Norelli, J.L., Aldwinckle, H.S., Momol, M.T., Johnson, B., Demaree, A., & Bhaskara, R. (2000). Fire blight of apple rootstocks. New York Fruit Quarterly, 8, 5-8. Norelli, J.L., Jones, A.L., & Aldwinckle, H.S .. (2003). Fire blight managenlent in the twenty-tirst century. Plant Disease, 87,756-765. Nuclo, R.L. (1997). Natural spread of and cOlnpetition between two bacterial antagonists of the fire blight pathogen, Erwinia amylovora, on blossonls of Bartlett pear. Oregon State University, Covallis, OR, Masters thesis, 67 pp. Nuclo, R.L., Johnson, K.B., Stockwell, V.O., & Sugar, D. (1998). Secondary colonization of pear blossons by two bacterial antagonists of the fire blight pathogen. Plant Disease, '82, 661-668. O'Brien, R.D. & Lindow, S.E. (1989). Effect of plant species and environnlental conditions on epiphytic population sizes of Pseudomonas ~yringae and other bacteria. Phytopathology, 79,619-627. Obradovic, A., & Jones, J.B. (2004). Management of tomato bacterial spot in the tieldby foliar applications ofbacteliophages and SAR inducers. Plant Disease, 88, 736-740. Oh, C.S., Kim, J.F., & Beer, S.V. (2005). The hlP pathogenicity island of Erwinia amylovora and identification of three novel genes required for systenlic infection. M·olecular Plant Pathology, 6, 125-138. Ogram, A. (1998). Isolation of nucleic acids from environmental samples. In R.S. Burlage, R. Atlas, D. Stahl, G. Geesey, & G. Sayler (Eds.) Techniques in Microbial Ecology. (pp. 273-288). Oxford, UK: Oxford University Press. Ojiambo, P.S. & Scherm, H. (2006). Biological and application-oriented factors illfluencing plant disease suppression by biological control: A meta-analytical review. Phytopathology, 96, 1168-1174 Okabe, N., & Goto, M. (1955). Bacterial plant diseases in Japan. 1. A list ofbactelial diseases and their pathogens. Shizuoka University Faculty of Agriculture Report, 5, 63-71. Okabe, N., & Goto, M. (1963). Bacteriophages of plant pathogens. Anllual Review of Phytopathology, 1,397-418. OMAFRA. (2006). Publication 360: Fruit Production Recomlnendations 2006-2·007. Toronto, 299 ON: Ontario Ministry of Agriculture, Food and Rural Affairs. OMAFRA. (2007). Publication 360: Fruit Production Recommendations 2007-2008. Toronto, ON: Ontario Ministry of Agriculture, Food and Rural Affairs. Palmer, E.L., Teviotdale, B.L., & Jones, A.L. (1997). A relative of the broad-host-rang plasmid RSF1010 detected in Erwinia amylovora. Applied and Environnlental Microbiolgoy, 63, 4604-4607. Paulin, J.P. (1987). Susceptibility to Erwinia amylovora of flower buds of pear trees under natural conditions. (pp. 623-627). In E.L. Civerolo, A. CollIner, R.E. Davis, & A.G. Gillespie (Eds.). Plant Pathogenic Bacteria. Boston: Ma11inus-NijhotT. Paulin, J.P., Chartier, R., Lecomte, P., LaRue, P., Brisset, M.N., & I.Jachaud, G. (1990). Experiments with Aliette (phosetyl-aluminium) in fire blight control. Acta H011iculturae. 273, 383-390. Paulin, J.P. (2000). Erwinia amylovora : General characteristics, biochemistry and serology. In J. L. Vanneste (Ed.), Fire Blight: The Disease and its Causativ,e Agent, Erwinia amylovora. (pp. 87-115). Wallingford, UK: CAB International. Pedulla, M.L., Ford, M.E., Houtz, J.M., Karthikeyan, T., Wadswo11h, C., Lewis, J.A., JacobsSera, D., Falbo, J., Gross, J., Pannunzio, N.R., Brucker, W., Kumar, V., Kandasamy, J., Keenan, L., Bardarov, S., Kriakov, J., Lawrence, J.G., Jacobs, W.R., Hendrix, R.W., & Hatfull, G.F. Origins ofhighlYlnosaic mycobacteriophage genomes. (2003). Cell, 113, 171-182. Plasterk, R.H., Kanaar, R., & van de Putte, P. (1984). A genetic switch in vitro: DNA inversion by Gin protein of phage Mu. Proceedings of the National Academy of Sciences.. 81, 2689-92. Pristovsek, P., Sengupta, K., Lohr, F., Schafer, B., von Trebra, M.W., Riiterjans, H., & Bernhard, F. Structural analysis of the DNA-binding dOlnain of the Erwinia amylovora RcsB protein and its interaction with the RcsAB box. Journal of Biological Chemistry, 278, 17752-17759. PriiB, B.M., Besemann, C., Denton, A., Wolfe, A. J. (2006). A conlplex transcription network controls the early stages ofbiofilm developlnent by Escherichia coli. Journal of Bacteriology, 188,3731-3739. Pujol, M., Badosa, E., Manceau, C., & Montesinos, E. (2006). Assessment of the envirOll111ental fate of the biological control agent of fire blight, Pseudo1110nasfluorescens Ep'S62e, on apple by culture and real-time PCR methods. ApIJlied and EnvironnTental Microbiologv, 300 72,2421-2427. Pusey, P.L. (1997). Crab apple blossoms as a nl0del for research on biological control of fire blight. Phytopathology, 87 , 1096-1102. Pusey P.L. (1999). Effect of nectar on microbial antagonists evaluated for use in control of fire blight of pome fruits. The American Phytopathological Society, 89,39-46. Pusey, P.L. (2002). Biological control agents for fire blight of apple compared under conditions limiting natural dispersal. Plant Disease, 86, 639-644. Pusey, P.L., & Curry, E.A. (2004). Temperature and pomaceous flower age related to colonization by Erwinia amylovora and antagonists. Phytopathology, 94, 901-911. Quamme, H.A., van del" Zwet, T., & Dirks, V. (1976). Relationship of fire blight resistance of young pear seedlings inoculated in the greenhouse to mature seedling trees naturally infected in the field. Plant Disease Reporter, 60, 660-664. Quamme, H.A., Kappel, F., & Hall, J.W. (1990). Efficacy of early selection for fire blight resistance and the analysis of cOlnbining ability for fire blight resistance in several pear progenies. Canadian Journal of Plant Science, 70,905-913. Raese, J.T., Williams, M.W., & Billingsley, H.D. (1978). Cold hardiness, sorbitol and sugar levels of apple shoots as influenced by controlled temperature and season. Journal of the American Society for Horticultural Science, 103, 796-801. Ravensdale, M. (2004). Bacteriophages and their use for biological control of Erwinia carotovora subsp. carotovora soft rot of calla lily. University of Guelph: Masters thesis. Ravensdale, M., Blom, T., Gracia-Garza, J.A., Slnith, R.J., & Svircev, A.M. (2006). Bacteriophages of and the control of Erwinia carotovora subsp. carotovora. Canadian Journal of Plant Pathology, 29, 121-130 . Reanny, D.C. & Marsh, S.C.N. (1973). The ecology of viruses attacking Bacillus stearothermophilus in soil. Soil Biology and Biochemistry, 5, 399-408. Reynoird, J.P., Mourgues, F., Norelli, J., Aldwinckle, H.S., Brisset, M.N., & Chevreau, E. (1999). First evidence for, improved resistance to fire blight in tra11sgenic pear ·expressing the attacin E gene from Hyalophora cecropia. Plant Science, 149, 23-31. Rhim, S.L., Volksch, B., Gardan, L., Paulin, J.P., Langlotz, C., Kim, W.S., & Geidel", K. (1999). Erwinia pyrifoliae, an Erwinia species different fi·onl Erwinia amylovora, cau"Ses a necrotic disease of Asian pear trees. Plant Pathology, 48, 514-520. 3{)1 Ries, S.M., & Otterbacher, A.G. (1977). OccuITence of fire blight on thornless blackberry in Illinois. Plant Disease Reporter, 61, 232-235. Riggle, J.H., & Klos, E.J. (1972). Relationship of Erwinia herbicola to Erwinia amylovora. Canadian Journal of Botany, 50, 1077-1083. Rijpens, N.P., Jannes, G., van Asbroeck, M., Rossau, R., & Herman, L.M.F. (1996). Direct detection of Brucella spp. in raw milk, by PCR and reverse hybridizatioll with 16S-23S rRNA spacer probes. Applied and Environmental Microbiology, 62, 1683-1688. Ritchie, D. (1978). Bacteriophages of Erwinia an1ylovora: their isolation, distribution, characterization, and possible involvelnent in the etiology and epidemiology of fire blight. Ph. D. thesis. Michigan State University, East Lansillg. Ritchie, D.F., & Klos, E.J. (1977). Isolation of Erwinia amylovorabacteriophage from aerial parts of apple trees. Phytopathology, 67, 101-104. Ritchie, D.F., & Klos, E.J. (1979). Some properties of Erwinia amylovora bacteriophages. Phytopathology, 69, 1078-1083. Rohwer, F. (2003). Global phage diversity. Cell, 113, 141. Romeiro, R.S., Karr, A.L., & Goodman, R.N. (1981a). Erwinia amylovora cell wall receptor for apple agglutinin. Physiological Plant Pathology, 19, 383-390. Romeiro, R.S., Karr, A.L., & Goodman, R.N. (1981b). Isolation ofa factor from apple that agglutinates Erwinia amylovora. Plant Physiology, 68, 772-777. Saad, A.T., Hanna, L., & Choueiri, E. (2000). Evaluation of streptomycin and oxytetracycline resistance of Erwinia amylovora populations in Lebanon. Ph}rtopathologv 90, S68. Salm, H., & Geider, K. (2004). Real-time PCR for detection and quantification of Erwinia amylovora, the causal agent of fire blight. Plant Pathology, 53,602-610. Salm, H., Kim, W.S., Geider, K., & Schneider, B. (2006). Control of fire blight with a lysozyl11e from the Erwinia alnylovora phage phiEalh. Acta Horticulturae. 704,371-374. Sambrook, J., & Fritsch, E.F.M.T,. (1989). Molecular Cloning, A Laboratory Manual. (211d ed.). Plainview, NY: Cold Spring Harbour Press. Sanger, F., Coulson, A.R., Hong, G.F., Hill, D.F., & Petersen, G.B. (1982). Nucleotide sequ,ence of bacteriophage lambda DNA. Journal of Molecular Biology, 162, 729-773. 302 Satsangi, J., Jewell, D.P., Welsh, K., Bunce, M., & Bell, J.I. (1994). Effect of heparin on polymerase chain reaction. Lancet, 343, 1509-1510. Schaad, N.W., Jones, J.B., & Chun, W. (2001). Laboratory guide for identification of plant pathogenic bacteria. Washington, DC: APS Press. Schisler, D. A. & Slininger, P. J. (1997). Microbial selection strategies that enhance the likelihood of developing cOll1nlercial biological control products. Journal of Industrial Microbiology and Biotechnolo.gy, 19, 172-179. Schnabel, E.L., Fernando, W.G.D., Meyer, M.P., JOlles, A.L., & Ja-ckson, L.E. (1998). Bacteriophage of Erwinia amylovora and their potential for biocontrol. Acta Horticulturae, 489, 649-654. Schnabel, E.L., & Jones, A.L. (1999). Distribution of tetracycline resistance genes and transposons among phylloplane bacteria in Michigall apple orchards. Applied and Envirollmental Microbiology, 65, 4898-4907. Schnabel, E.L., & Jones, A.L. (2001). Isolation and characterizatioll of five Erwinia amylovora bacteriophages and asseSSlllent of phage resistance in strains of Erwinia amylovora. Applied and Environmental Microbiology, 67,59-64. Scholl, D., & Merril, C. (2005). Polysaccharide-degrading ellzymes. In M. K. Waldor, D. I. Friedman, & S. L. Adhya (Eds.), Phages: Their role in bacterial patho,genesis alld biotechnology. (pp. 400-429). Washington, DC: ASM Press. Schroth, M.N., Thomson, S.V., & Moller, W.J. (1979). Streptomycill resistance in Erwinia amylovora. Phytopathology, 69, 565-568. Seemuller, E.A., & Beer, S.V. (1976). Absence of cell wall polysaccharide degradation by Erwinia amylovora. Phytopalg~ 66, 433-436. Seemuller, E.A., & Beer, S.V. (1977). Isolation and partial charactelization of two neutral proteases of Erwinia amylovora. Phvtopathologische Zeitschrift:, 90, 12-21. Shigeharu, T., Yasufumi, H., Yoichi, K., & Tetsuro, O. (2000). Detection oftoball10viluses frOID soils by non-precoated indirect ELISA. Journal of general plant patholo-gv, 66, 153-158. Sholberg, P.L., Bedford, K.E., Haag, P., & Randall P. (2001). Survey of Ervvinia amylovora isolates from British Colulnbia for resistallce to bactericides and vilulence on apple. Canadian Journal of Plant Patholo.gy, 23, 60-67. Shtienberg, D., Zilbersteine, M., Oppenheiln, D., Herzog, Z., Manulis,S., Shwaliz, H., & 303 Kritzman, G. (2001). Efficacy of oxolinic acid and other bactericides in suppression of Erwinia amylovora in pear orchards in Israel. Phytoparasitica, 29, 143-154. Shtienberg, D., Shwartz, H., Oppenheim, D., Zilberstaine, M., Herzog, Z., Mallulis, S., & KritzlTIan, G. (2003). Evaluation of local and imported fire blight warning systems in Israel. Phytopathology, 93, 356-363. Sj6stedt, A., Eriksson, D., Ramisse, V., & Garrigue, H. (1997). Detection of Bacillus anthracis spores in soil by PCR. FEMS Microbiologv Ecology, 23, 159-168. Sjulin, T.M., & Beer, S.V. (1977). Mechanism of wilt induction by amylovorill in cotoneaster shoots and its relation to wilting of shoots infected by Erwinia an1ylovora. Phytopathology, 68, 89-94. Skipper, H.D., & Westermann, D.T. (1973). Comparative effects of propylene oxide, sodiulll azide, and autoclaving on selected soil properties. Soil Biology alld Biochemistry, 5, 409-414. Slade, M.B., & Tiffin, A.I. (1984). Biochemical and serological characterization of Erwinia. III Bergon (Ed.), Methods in Microbiology. (pp. 228-293). London, UK: Acad,elllic Press. Slingerland, K., Lay, B., & Hunter, D. (2002). Pear Cultivars. Factsheet 02-039. Ontario Millistry of Agriculture and Food. Smith, H.W. & Huggins, M.B. (1982). Successful treatlnent of experinlental Escherichia coli infectiollS in mice using phage: its general superiority over antibiotics. Journal of General Microbiology, 128, 307-318. Smith, H.W. & Huggins, M.B. (1983). Effectiveness of phages in treating experinlental Escherichia coli diarrhoea in calves, piglets and lambs. Journal of Gen,eral Microbiology., 129, 2659-2675. Smith, J.H., & Powell D. (1968). A disc electrophoretic comparison of protein pattell1S of Erwinia amylovora with other bacteria, including associated yellow fornls. Phytopathology, 58, 972-975. Smith, D., Wiens, B., & Svircev, A.M. (2002). Streptomycin residues in orchard canopies and fruit. Agriculture and Agri-Food Canada. Smith, T.J. (1996). A risk assessment lTIodel for fire blight of apple and pear. Acta Horticulturae, 411,97-104. Smith, T. (1997). Fireblight control in the northwest is Inultifaceted. Good Fruit Grower, 31e- 3D4 35f. Solomon, B. (2007). Active immunization against Alzheimer's beta-anlyloid peptide using phage display technology. Vaccine, 25,3053-3056 Solymar, B. (2006.) Integrated management of fire blight on apple and pear in Canada. Agriculture alld Agri-Food Canada, Pesticide Risk Reduction Program Pest Managenlent Centre. Spinelli, F., Vanneste, J.L., Marcazzan, G.L., Sabatini, A.G., & Costa, G. (2005). Effect of prohexadione-calcium on nectar composition of pOlnaceous flowers and on bacterial growth. New Zealand Plant Protection, 58,106-111. Sriranganathan, N., Whichard, J.M., Pierson, F.W., & Kapur, V. Bacteriopha,ge Felix 01: Genetic Characterization (NC 005282). (unpublished) StalT, M., Cardona, C., & Folsonl, D. (1951). Bacterial fire blight of raspberry. Phvtopatholo:gv'l 41,915-919. Starnbach, M.N., Falkow, S., & Tonlkins, L.S. (1989). Species-specific detection of Legionella pneumophila in water by DNA amplification and hybridization. Journal of Clinical Microbiology, 27,1257-1261. Steinberger, E.M., & Beer, S.V. (1988). Creation and complementatioll ofpathogenicitYlllutants of Erwinia amylovora. Molecular Plant-Microbe Interactions, 1, 135-144. Steiner, P.W. (1992). Aggressive l11anagement progranl for fire blight. Transactions-of-theIllinois-State-Horticultural-Society, 125, 21-26. Steiner, P.W., & Lightner, G. (1996). Maryblyt, Version 4.3, A Predictive Progranl for Forecastillg Fire Blight Disease in Apples and Pears. {Conlputer Software]. College Park, MD: Office of Technology Liaison, University of Maryland. Available through Gempler's Inc. Belleville, WI. Steiner, P.W. (2000). Integrated orchard and nursery manageillellt for the cOl1trol of fire .blight. In J. L. Vanneste (Ed.), Fire Blight: the Disease and its Causative Agent~ Erwinia an1ylovora. (pp. 339-358). Wallingford, UK: CAB International. Stewart, A. (2001). Commercial biocontrol - reality or fantasy? Australasian Plant 127-131. Patho] gy~ 30, Stockwell, V.O., Loper, J.E., & Johnson, K.B. (1992). Establishnlent of bacterial antagonists on blossoms of pear. Phytopathology, 82, 1128 305 Stockwell, V.O., Johnson, K.B., & Loper, J.E. (1998). EstablishnJent of bacterial antagollists of Erwinia an1ylovora on pear and apple blossonls as influenced by iIloculuIll preparation. Phytopathology, 88, 506-513. Stockwell, Y.O., Mclaughlin, R.J., Henkels, M.D., Loper, J.E., Sugar, D., & Roberts, R.G. (1999). Epiphytic colonization of pear stigmas and hypanthia by bacteria -during prinlary bloom. Phytopathology, 89, 1162-1168. Stockwell, V.O., Johnson, K.B., Sugar, D., & Loper, J.E. (2002). Antibiosis cOl1tributes to biological control of fire blight by Pantoea agglomerans strain Eh252 in orchards. Phytopathology, 92, 1202-1209. Sugar D, Johnson KB, Lindow SE, & Stockwell YO. (1993). Effects of postharvest and bloOI11 applications of phosetyl-AI on fire blight and fruit quality in 'Ba11Iett' pear. Acta Horticulturae, 338, 289-296. Suhayda, C.G., & Goodman, R.N. (1981). Early proliferation and migration and subsequent xylem occlusion by Erwinia amylovora an_d the fate of its extracellular polysaccharide (EPS) in apple shoots. Phytopathology, 71, 697-707. Sulakvelidze, A. & Kutter, E. (2005). Bacteriophage therapy in humalls. In E. Kutter & A. Sulakvelidze (Eds.), Bacteriophages: Biology and applications. (pp.381-435), Boca Raton, FL: CRC Press. Suleman, P., & Steiner, P.W. (1994). Relationship between sorbitol and solute potential in apple shoots relative to fire blight symptom development after infection by Erwinia am~vloDra. Phytopathology, 84, 1244-1250. Sunlmer, E.l., Thilaganathan, N., Hall, J., Vidaver, A.K., King, M.D., Robinson, S., Young, R., & Gonzalez, C.F. Genomic analysis of Erwinia amylovora phage Era] 03, a new Inenlber of the T7 superfamily related to SP6~ Kl-5 and K1E. GenBank: NC_009014 (unpublished) Summers, W.C. (1999). Felix d'Herrelle and the Oligins of Molecular Biology. N-ew Haven, CT: Yale University Press. Summers, W.C. (2005). Bacteriophage research: Early history. In E. Kutter & A. Sulakvelidze (Eds.), Bacteriophages: B,iology and applications. (pp. 5-27). Boca Rato11, FL: CRC Press. Sutton, T.B. (1996). Changing options for the control of deciduous fruit tree diseases. Phytopathology, 34, 527-547. Tan, J.S.H., & Reanney, D.C. (1976). Interactions between bacteriophages and bactelia in soil. 306 Soil Biology and Biochemistry'l 8, 145-150. Tanji, Y., Shimada, T., Yoichi, M., Miyanaga, K., Hori, K., & Unno, H. (2004). Toward rational control of Escherichia coli 0157:H7 by a phage cocktail. Applied Microbiology alld Biotechnology, 64, 270-274. Tanji, Y., Shimada, T., Fukudomi, H., Miyanaga, K., Nakai, Y., & Unno, H. (2005). Therapeutic use of phage cocktail for controlling Escherichia coli 0157:H7 ill gastrointestinal tract of mice. Journal of Bioscience and Bioengineering, 100, 280-287. Taylor, D.H., Moore, R.S., & Sturman, L.S. (1981). Influence of pH and electrolyte composition on adsorption of poliovirus by soils and minerals. Applied and Environmental Microbiologv, 42, 976-984. Temple, T.N., Stockwell, V.O., Loper, J.E., & Johnson, K.B. (2006). Bioavailability of iron to Pseudomonas fluorescens strain A506 on flowers of pear and apple. Acta Horticulturae, 704,301-305. Tharaud, M., Laurent, J., Faize, M., & Paulin, J.P. (1997). Fire blight protection \vith avilulent lTIutants of Erwinia amylovora. Microbiology, 143, 625-632. Thibault, B. & Le Lezec, M. (1990). Sensibilite au feu bacteriell des principales varietes de pommier et de poirier utilisees en Europe. (pp. 96-109) In Fire Blight ojPomoi"deae (Erwinia amylovora, Burrill, Winslow et al.). Applied Research in Europe (1978-1988), Brussels-Luxembourg: EUR 12601, ECSC-EEC-EAEC. Thomson, S.V. (1986). The role of the stigma in fire blight illfections. Phytopatholo,gy, 76, 476-482. Thomson, S.V., Hansen, D.R., Flint, K.M., & Valldenberg, J.D. (1992). Disselllillation of bacteria antagonistic to Erwinia amylovora by hOlley bees. Plant Disease, 76, 10521056. Thomson, S.V., Gouk, S.C., Vanneste, J.L., Hale, C.N., & Clark, R.G. (1993). The presence of streptomycin resistant strains of Erwinia an1ylovora in New Zealand. Acta Horticulturae., 338, 223-230. Thomson, S.V., Wagner, A.C., &, Gouk, S.C. (1998). Studies of the epiphytic colonization of apple flowers and the role of insects and rain. Acta Horticulturae, 489, 459-462. Thomson, S.V. & Ockey, S.C. (2001). Sudden widespread appearance ofstreptonlycin resistallce in Erwinia amylovora in Utah. Phytopathology, 91, S88. 307 Thomson, S.V., & Gouk, S.C. (2003). Influence of age of apple flowers on growth of Erwinia amylovora and biological control agents. Plant Disease, 87,502-509. Toselli, M., Malaguti, D., Marangoni, B., Cabrini, L., Bazzi, C., Sponza, G., &Scudellari, D. (2002). Reduction of fire blight infection by manipulation of pear tree water status. Acta Horticulturae, 590, 193-199. Torsvik, V., GOks0yr, J., & Daae, F.L. (1990). High diversity in DNA of soil bacteria. Applied and Environmental Microbiology, 56, 782-787. Tsai, Y.L., & Olsen, B. (1992). Rapid method for separation of bacterial DNA from hunlic substances in sediments for polymerase chain reaction. Applied and Environmental Microbiology, 58, 2292-2295. Tsiantos J, & Psallidas P. (1993). Experiments on chemical control of fire blight (Elwinia amylovora) in Greece. Phvtopathologia Mediterranea, 32, 201-205. Tsiantos, J., Psallidas, P., & Chatzaki, A. (2003). Efficacy of alternatives to antibiotic chell1icals for the control of fire blight of pears. Annals of Applied Biology., 143, 319-323. Twort, F.W. (1915). An investigation on the nature of ultra-microscopic viruses. Lancet, 2, 1241-1243. Uyeda, E. (1903). The causal bacterium of apple blight [in Japanese]. Dai-Nippon No-Kaiho., 260, 1-3. van Alfen, N.K., & Allard-Turner, N.C. (1979). Susceptibility of plants to vascular disruption by macromolecules. Plant Physiology, 63, 1072-1075. van de Putte, P., Cramer, S., & Giphart-Gassler, M. (1980). Invertible specificity of bacteriophage mu. Nature, 286,218-22. DN~ deterl11ines host van der Zwet, T., & Keil, H.L. (1979). Fire Blight: A bacterial disease of rosaceous plants. Washington, DC: United States Departlnent of Agriculture. van der Zwet, T., & Beer, S.V. (1995). Fire Blight - Its Nature., Prevention, and Control. A Practical Guide to Integrated Disease Managemellt. Agriculture Infonl1ation Bulletin No. 631. Washington, DC: US Departnlent of Agriculture. van der Zwet, T., & Walter, J. (1996). Presence of Erwinia an1ylovora in apparently h,ealthy nurserypropagatingmaterial. Acta Horticulturae, 411,127-130. van Leare, 0., de Greef, M., & de Wael, L. (1982). 111fluence of the honey bee 011 fire blight 30E transmission. Acta Horticulturae, 117,131-141. Vandenbergh, P.A., Wright, A.M., & Vidaver, A.K. (1985). Partial purification and characterization of a polysaccharide depolymerase associated with phage-infected Erwinia amylovora. Applied and Environmental Microbiology, 49,994-996. Vandenbergh, P.A., & Cole, R.L. (1986). Cloning and expression in Escherichia coli of the polysaccharide depolymerase associated with bacteriophage-infected Erwinia amylovora. Applied and Environmental Microbiology, 51, 862-864. Vanneste J.L., Smart, L.B., Zumoff, C.H., Yu, J., & Beer, S.V. (1990). Control of fire blight by Erwinia herbicolaEh252: role of antibiotic production. Acta Horticulturae, No. 273, 393-394. Vanneste, J.L., Yu, J., & Beer, S.V. (1992). Role of antibiotic production by Erwinia herbicola Eh252 in biological control of Erwinia amylovora. Journal of Bacteriology, 174, 27852796. Vanneste, J.L., & Voyle, M.D. (1998). Genetic basis of streptomycin resistance in pathogenic and epiphytic bacteria isolated from apple orchards in New Zealand. Acta Horticult rae~ 489,671-672. Vanneste, J.L. (2000). What is fire blight? Who is Erwinia amylovora? How to control it? In J. L. Vanneste (Ed.), Fire Blight: the Disease and its Causative Agent, Erwinia amvlovora. (pp. 1-6). Wallingford, Uk: CAB Internatio11al. Vanneste, J.L., Cornish, D.C., Yu, J., & Voyle, M.D. (2002). PI0c: a new biological control agent for control of fire blight which can be sprayed or distributed using honey bees. Acta Horticulturae, 590, 231-235. Vantomme, R., Swings, J., Goor, M., Kersters, K., & de Ley, J. (1982). Phytopathological, serological, biochemical and protein electrophoretic characterization of Erwinia amylovora strains isolated in Belgium. Phytopathologische Zeitschrift, 103, 349-356. Vantomme, R., Rijckaert, C., Swings, J., & de Ley, J. (1986). Characterizatio11offurther Erwinia amylovora strains and the application of the API 20E systen1 in diagnosis. Journal of Phytopathology, 117,34-42. Venisse, J.S., Gullner, G., & Brisset, M.N. (2001). Evidence for the involvement of an oxidative stress in the initiation of infection of pear by Erwinia am~vlor. Plant Physiolgv~ 125, 2164-2172. Venisse, J.S., Barny, M.A., Paulin, J.P., & Brisset, M.N. (2003). Involvelnent of three 309 pathogenicity factors of Erwinia amylovora in the oxidative stress associated with compatible interaction in pear. FEBS Letters, 537, 198-202. Verdonck, L., Mergaert, J., Rijckaert, C., Swings, J., Kersters, K., & de Ley, J. (1987). Gellus Erwinia: numerical analysis of phenotypic features. International Journal of Svstelnatic Bacteriology, 37,4-18. Vidaver, A.K. (1976). Prospects for control of phytopathog'enic bactetia by bacteriophages alld bacteriocins. Annual Review of Phytopathology, 14, 451-465. von Bodman, S.B., Bauer, W.D., & Coplin, D.L. (2003). Quorunl sensing in plant-pathogenic bacteria. Annual Review of Phytopathology, 41, 455-482. Wagenaar, J., van Bergen, M., Mueller, M., Wassenaar, T., & Carlton, R. (2005). Phage therapy reduces Campylobacter jejuni colonization in broilers. Veterinaty Mjcrobiology, 109, 275-283. Ward, R.L., Knowlton, D.R., & Winston, P.E. (1986). Mechanism of inactivation of enteric viruses in fresh water. Applied and Environmental Microbiolo·Qv, 52, 450-459. Wei, Z.M., Laby, R.J., Zumoff, C.H., Bauer, D.W., He, S.Y., Collnler, A., & Beer, S.V. (1992). Harpin, elicitor of the hypersensitive response produced by the plant pathogen Erwinia amylovora. Science, 257, 85-88. Wei, Z.M., Sneath, B.J., & Beer, S.V. (1992). Expression of Erwinia amylovora hrp genes in response to environmental stimuli. Journal of Bacteriologv, 174, 1875-1882. Wei, Z.M., & Beer, S.V. (1995). HrpL activates Erwinia amylovora hrp gene transcriptioJl and is a member of the ECF subfamily of sigma factors. JounlalofBacteliology, 177, 6201-6210. Wei, Z.M., Kim, J.F., & Beer, S.V. (2000). Regulation of hrp genes and type III protein secretion in Erwinia anlylovora by HrpX/HrpY, a novel two-Colllponent systenl, and HrpS. Molecular Plant-Microbe Interactions, 13, 1251-1262. Weinbauer, M.G. & Peduzzi, P. (1994). Frequency, size and distlibution ofbacteriophag,es in different marine bacterial morphotypes. Marine Ecology Progress 'Series, 108, 11-20. Werner, N.A., Heidenreich, G.E., & Aldwinckle, H.S. (2004). Field evaluation ofnlaterials for control of fire blight infection of apple, 2004. Geneva, NY. Department of Plant Pathology, Cornell University, Geneva. Wiedbrauk, D.L., Werner, J.C., & Drevon, A.M. (1995). 111hibition ofPCR by aqueous and 310 vitreous fluids. Journal of Clinical Microbiology, 33, 2643-2646. Williamson, K.E., Wommack, K.E., & Radosevich, M. (2003). Sampling natural viral conlmunities from soil for culture-independent analyses. Applied and Environnlental Microbiology, 69, 6628-6633. Williamson, K.E., Radosevich, M., & Wommack, K.E. (2005). Abundance and diversity of viruses in six delaware soils. Applied alld Environmental Microbiology, 71, 3119-3125. Wilson, E., Epton, H.A.S., & Sigee, D.C. (1989). Erwinia amylovora infection of hawthorn blossom, II. The stigma. JournalofPhvtopathology, 127, 15-28. Wilson, E., Epton, H.A.S., & Sigee, D.C. (1990). Biological control of fire blight of hawthorn (Crataegus monogyna) with Erwinia herbicola under protected conditions. Plant Pathology, 39, 301-308. Wilson, E., Epton, H.A. S. & Sigee, D.C. (1992). Interactions between Erwinia herbicola and E. amylovora on the stigma of hawthorn blossoms. Phytopathology, 82,914-918. Wilson, M., & Lindow, S.E. (1993). Interactions between the biological control agent Pseudomonas jluorescens A506 and Erwinia amylovora in pear blossoms. Phytopathology, 83, 117-123. Wilson, I.G. (1997a). Inhibition and facilitation of nucleic acid amplificatiol1. Applied and Environmental Microbiology, 63,3741-3751. Wilson, M. (1997b). Biocontrol of aerial plant diseases in agriculture and horticulture: cun~et approaches and future prospects. Journal of Industrial Microbiology and Biotechnology, 12,188-191. Wilson, K. (2000). Pruning Fruit Trees. OMAF Factsheet 00-005. Olltario Ministry of Agriculture and Food. Wilson, M. & Lindow, S.E. (2000). Viable but nonculturable cells in plant-associated bacterial populations. In R.R. Colwell & D.J. Grimes (Eds.), Nonculturable microorganisms in the environment. (pp. 229-241). Washington, DC: ASM Press. Winona, L.J., Ommani, A.W., Olszewski, J., Nuzzo, J.B., & Oshilna, K.H. (2001). Efficient and predictable recovery of viruses from water by snlall scale ultrafiltration systenls. Canadian Journal of Microbiology, 47,1033-1041. Wodzinski, R.S., & Paulin, J.P. (1994). Frequency and diversity of antibiotic production by putative Erwinia herbicola strains. Journal of Applied Bacteriology, 76,603-<507.