<%BANNER%>

Plant-Associated Bacteria

Permanent Link: http://ufdc.ufl.edu/UFE0041068/00001

Material Information

Title: Plant-Associated Bacteria Biological, Genomic, and Metagenomic Studies
Physical Description: 1 online resource (161 p.)
Language: english
Creator: Tyler, Heather
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: citrus, endophytes, gluconacetobacter, klebsiella, liberibacter, metagenomics
Microbiology and Cell Science -- Dissertations, Academic -- UF
Genre: Microbiology and Cell Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Bacteria can interact with plants in a variety of ways, as pathogens causing disease, as epiphytes living on plant surfaces, or they can live within plant tissue without causing disease as symbionts or endophytes. Therefore, a range of different plant bacterial interactions was studied in this work. A metagenomic study utilizing high-throughput sequencing data from citrus tissue with the disease, Huanglongbing (HLB), was conducted to assess the microbial community in infected tissue. The only bacterium identified in infected tissue was the proposed causative agent of HLB, ?Candidatus Liberibacter asiaticus.? As this bacterium has yet to be cultured, this work substantiates its proposed role in disease development. Plant growth promoting endophytes were also examined, looking at the genomic level as well as at specific plant growth promoting responses. Using optical mapping, two contradictory genome sequences of Gluconacetobacter diazotrophicus PAl 5 were distinguished, identifying which was the best representation of the actual PAl 5 chromosome. The results of this study are important since the accuracy of the genome sequence could have significant impacts on comparative genomic analyses with this strain. The plant growth promoting mechanisms used by two endophytic enteric bacteria, Klebsiella pneumoniae 342 and Enterobacter cloacae P101, were also investigated, utilizing their genome sequences to guide experiments. Both bacteria increased lateral root numbers on Arabidopsis plants and genome annotation indicated each posses the auxin synthesis gene, ipdC. As auxin is known to play a role in lateral root development, subsequent work on lateral root promotion focused on this phytohormone. As part of this work, a plasmid was constructed that expressed the ipdC gene and could be stably maintained without selection pressure. While these enteric endophytes are related to human pathogens and Kp342 was found to have pathogenic potential, the information gained by studying these bacteria could yield insight into enhancing the growth-promoting effects of other endophytes. Thus, the work described here exemplifies how metagenomic, genomic, and biological experiments can be used to gain both an overview and in depth view of plant bacterial associations.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Heather Tyler.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Triplett, Eric W.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0041068:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041068/00001

Material Information

Title: Plant-Associated Bacteria Biological, Genomic, and Metagenomic Studies
Physical Description: 1 online resource (161 p.)
Language: english
Creator: Tyler, Heather
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: citrus, endophytes, gluconacetobacter, klebsiella, liberibacter, metagenomics
Microbiology and Cell Science -- Dissertations, Academic -- UF
Genre: Microbiology and Cell Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Bacteria can interact with plants in a variety of ways, as pathogens causing disease, as epiphytes living on plant surfaces, or they can live within plant tissue without causing disease as symbionts or endophytes. Therefore, a range of different plant bacterial interactions was studied in this work. A metagenomic study utilizing high-throughput sequencing data from citrus tissue with the disease, Huanglongbing (HLB), was conducted to assess the microbial community in infected tissue. The only bacterium identified in infected tissue was the proposed causative agent of HLB, ?Candidatus Liberibacter asiaticus.? As this bacterium has yet to be cultured, this work substantiates its proposed role in disease development. Plant growth promoting endophytes were also examined, looking at the genomic level as well as at specific plant growth promoting responses. Using optical mapping, two contradictory genome sequences of Gluconacetobacter diazotrophicus PAl 5 were distinguished, identifying which was the best representation of the actual PAl 5 chromosome. The results of this study are important since the accuracy of the genome sequence could have significant impacts on comparative genomic analyses with this strain. The plant growth promoting mechanisms used by two endophytic enteric bacteria, Klebsiella pneumoniae 342 and Enterobacter cloacae P101, were also investigated, utilizing their genome sequences to guide experiments. Both bacteria increased lateral root numbers on Arabidopsis plants and genome annotation indicated each posses the auxin synthesis gene, ipdC. As auxin is known to play a role in lateral root development, subsequent work on lateral root promotion focused on this phytohormone. As part of this work, a plasmid was constructed that expressed the ipdC gene and could be stably maintained without selection pressure. While these enteric endophytes are related to human pathogens and Kp342 was found to have pathogenic potential, the information gained by studying these bacteria could yield insight into enhancing the growth-promoting effects of other endophytes. Thus, the work described here exemplifies how metagenomic, genomic, and biological experiments can be used to gain both an overview and in depth view of plant bacterial associations.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Heather Tyler.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Triplett, Eric W.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0041068:00001


This item has the following downloads:


Full Text

PAGE 1

PLANT-AS SOCIATED BACTERIA: BIOLOGICAL, GENOMIC, AND METAGENOMIC STUDIES By HEATHER LEE TYLER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009 1

PAGE 2

2009 Heather Lee Tyler 2

PAGE 3

To Mom and Dad 3

PAGE 4

ACKNOWLEDGMENTS I thank my parents, who have always supported me in my endeavors. I would also like to thank my adviser, Eric W. Triplett. Lastly, I would like to thank all the members of the Triplett Lab, both past and present, for their advice a nd support. Without them, I would not have survived the rough times when nothing seemed to work and I thought Id never make it through. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES.......................................................................................................................10 ABSTRACT...................................................................................................................................11 CHAPTER 1 INTRODUCTION................................................................................................................. .13 2 PLANTS AS A HABITAT FOR BENEFI CIAL AND/OR HUMAN PATHOGENIC BACTERIA............................................................................................................................16 Introduction................................................................................................................... ..........16 Cycle of Human Pathogens in the Environment.....................................................................17 Observations of Pathogens on Plant Surfaces........................................................................23 Endophytic Colonization........................................................................................................ 24 Bacterial Competition.......................................................................................................... ...29 Role in Plant Growth Promotion............................................................................................30 Bacterial Mechanisms of Colonization...................................................................................32 Common Virulence Factors in Plant and Animal Pathogens.................................................37 Genomes of Endophytic Bacteria...........................................................................................41 Future Directions....................................................................................................................44 3 CONFIRMATION OF THE SEQUENCE OF CANDIDATUS LIBERIBACTER ASIATICUS AND ASSESSMENT OF MICROBIAL DIVERSITY IN HUANGLONGBING-INFECTED CITRUS PHLOEM USING A METAGENOMIC APPROACH...........................................................................................................................45 Introduction................................................................................................................... ..........45 Results.....................................................................................................................................47 Verification of Ca. L. asiaticus str. psy62 Contigs.......................................................47 Confirmation of Ca. L. asiaticus Contig Sequences.....................................................47 Comparison to Alphaprot eobacterial Relatives...............................................................49 Assessment of Bacterial Diversity...................................................................................50 Other Reported HLB Associated Bacteria Not Present...................................................52 DNA Viruses and Viroids Not Present............................................................................55 High Coverage of Citrus..................................................................................................56 Discussion...............................................................................................................................56 Materials and Methods...........................................................................................................61 Sample Preparation and DNA Extraction........................................................................61 PCR to Confirm Validity of Ca L. asiaticus Con tigs in GenBank..............................62 5

PAGE 6

ARISA Analysis, Cloning and Sequencing f or Initial Assessment of Bacterial Diversity and to Confirm Presence of Ca L. asiaticus.............................................62 ARISA Sequence Analysis and Classification................................................................64 High-Throughput Sequencing.........................................................................................64 Sequence Analysis...........................................................................................................65 4 OPTICAL MAPPING OF GLUCONACETOBACTER DIAZOTROPHICUS PAL 5 REVEALS CHROMOSOMAL REARRANGEMENTS IN COMPLETED GENOME SEQUENCE...........................................................................................................................75 Background.............................................................................................................................75 Results.....................................................................................................................................76 Optical Map of G. diazotrophicus PAl 5.........................................................................76 Identification of Sequence Rearrangements Using Optical Mapping.............................77 Differences in Annotation Between Genome Sequences................................................78 Discussion...............................................................................................................................79 Conclusions.............................................................................................................................83 Methods..................................................................................................................................84 Bacterial Strain............................................................................................................... .84 Preparation of Cells for Optical Mapping.......................................................................84 Optical Mapping and Analysis........................................................................................84 Comparison of Annotation..............................................................................................85 5 ENDOPHYTE MEDIATED PLANT GROWTH PROMOTION.........................................90 Introduction................................................................................................................... ..........90 Lateral Root Development..............................................................................................91 Modes of Action of Plant Growth-Promoting Bacteria...................................................92 Secretion of plant hormones.....................................................................................92 Production of volatile compounds............................................................................96 Making mineral nutrients more available to plants..................................................96 Secretion of other compounds..................................................................................97 Results.....................................................................................................................................98 Increased Lateral Root Number in Arabidopsis thaliana ................................................98 Lateral Root Increase Not Due to Nitrogen Fixation......................................................98 Lateral Root Increasing Phe notype is Strain Specific.....................................................99 Lateral Root Promotion Due to Secreted Diffusible Product........................................100 Secretion of Lateral Root Promo ting Compound is Plant Inducible.............................101 Lateral Root Promotion May Involve Secretion of a Plant Hormone...........................102 Exogenous Application of Auxin Partially Mimics Lateral Root Promotion by Kp342.........................................................................................................................103 Role of ipdC in Auxin Production.................................................................................103 ipdC Mutants on Plants ...............................................................................................105 Discussion.............................................................................................................................106 Materials and Methods.........................................................................................................111 Plant Culturing...............................................................................................................111 Bacterial Strains and Inoculum Preparation..................................................................111 6

PAGE 7

Extraction of Compounds from Culture Supernatant....................................................112 Comparison of Bacterial Inoculation to Kp342 Supernatant Extracts and Exogenous IAA............................................................................................................................112 Construction of ipdC In-Frame Deletion in E. cloacae P101.......................................113 Complementation of P101 ipdC ..................................................................................114 Determination of Plasmid Stability...............................................................................115 Auxin Assay..................................................................................................................116 Insertion of ACC Deaminase into pHLT14...................................................................116 6 CONCLUSION................................................................................................................... ..127 APPENDIX A GENOME OF KLEBSIELLA PNEUMONIAE 342..............................................................132 LITERATURE CITED................................................................................................................139 BIOGRAPHICAL SKETCH.......................................................................................................161 7

PAGE 8

LIST OF TABLES Table page 3-1 Overview of metagenom ic sequence statistics..................................................................69 3-2 Fold coverage of each Candidatus Liberibacter asiaticus contig determined by reference assembly with 454, Solexa, and SOLiD metagenomic data from Huanglongbing-infected citrus phloem..............................................................................69 3-3 Percent length of each Candidatus Liberibacter asiaticus contig covered by 454, Solexa, and SOLiD metagenomic data fr om Huanglongbing-infected citrus phloem as determined by reference assembly................................................................................70 3-4 All metagenomic phloem sequences compared to the contigs of Candidatus Liberibacter asiaticus or th e fully sequenced genomes of other alphaproteobacteria by reference assembly........................................................................................................71 3-5 Fold coverage of the top 16S rRNA RDP database matches with 454, Solexa, and SOLiD metagenomic data from Huan glongbing-infected citrus phloem..........................71 3-6 Comparison of Candidatus Liberibacter 16S rRNA to 454, Solexa, and SOLiD metagenomic sequences from Huangl ongbing-infected phloem by reference assembly....................................................................................................................... ......72 3-7 Comparison of Candidatus Liberibacter asiatic us and other reported Huanglongbing-associated bacteria to all metage nomic phloem sequences.....................72 3-8 Comparison of metagenomic assemblies with total metagenomic data to metagenomic data with chloroplast sequences removed...................................................72 3-9 Candidatus Liberibacter asiaticus contig primers...........................................................73 3-9 Continued.................................................................................................................. .........74 4-1 Optical and in silico BglII restriction maps for G. diazotrophicus PAl 5..........................87 4-2 Rearrangement positions in G. diazotrophicus PAl 5 genome sequence from RioGene.............................................................................................................................87 4-3 Regions of in silico maps not aligned to the G. diazotrophicus PAl 5 optical map..........87 4-4 Comparison of coding sequences between two genomic sequences of G. diazotrophicus PAl 5 based on percent identity................................................................87 4-5 Unique functional roles between G. diazotrophicus PAl 5 genome sequences.................88 4-6 Transposases in G. diazotrophicus PAl 5 genome sequences...........................................89 8

PAGE 9

5-1 Examples of plant hormones secreted by bacteria...........................................................125 5-2 Comparison of growth-promoting mechanisms in K. pneumoniae 342, E. cloacae P101, and E. coli K12......................................................................................................125 5-3 Hormone insensitive Arabidopsis mutants used to examine lateral root promotion.......125 5-4 Gene deletion and plasmid construction primers.............................................................126 A-1 Kp342 Antibiotic Resistance Profile...............................................................................138 A-2 Infection of K. pneumoniae 342 and C3091 in a Mouse Infection Model......................138 9

PAGE 10

LIST OF FIGURES Figure page 3-1 Automated ribosomal intergenic spac er analysis (ARISA) of healthy and Huanglongbing infected Citrus sinensis phloem...............................................................68 4-1 Alignment of G. diazotrophicus PAl 5 optical map with in silico maps of genome sequences...................................................................................................................... .....86 5-1 Lateral root promotion by Kp342....................................................................................118 5-2 Effect of nitrogen fixati on on lateral root promotion.......................................................118 5-3 Bacterial strain specificit y of lateral root promotion.......................................................119 5-4 Lateral root promotion due to a secreted product............................................................120 5-5 Lateral root promotion in re sponse to culture supernatants.............................................121 5-6 Response of hormone insensitive Arabidopsis mutants to Kp342...................................121 5-7 Comparison of Kp342 inoculation to exogenous IAA treatment....................................122 5-8 Auxin production in P101 strains....................................................................................123 5-9 Plasmid stability in P101 ipdC ......................................................................................123 5-10 Diagrams of selected plasmid constructs.........................................................................124 10

PAGE 11

Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PLANT-ASSOCIATED BACTERIA: BIOLOGICAL, GENOMIC, AND METAGENOMIC STUDIES By Heather Lee Tyler December 2009 Chair: Eric W. Triplett Major: Microbiology and Cell Science Bacteria can interact with plants in a variety of ways, as pathogens causing disease, as epiphytes living on plant surfaces, or they can live within plant tissue without causing disease as symbionts or endophytes. Therefore, a range of different plant-bact erial interactions was studied in this work. A metagenomic study utilizing hi gh-throughput sequencing data from citrus tissue with the disease, Huanglongbing (HLB), was c onducted to assess the microbial community in infected tissue. The only bacterium identified in infected tissue was the proposed causative agent of HLB, Candidatus Liberibacter asiaticus. As this bacterium has yet to be cultured, this work substantiates its proposed role in di sease development. Plant growth-promoting endophytes were also examined, l ooking at the genomic level as well as at specific plant growth-promoting responses. Using optical mapping, two contradictory genome sequences of Gluconacetobacter diazotrophicus PAl 5 were distinguished, identifying which was the best representation of the actua l PAl 5 chromosome. The results of this study are important since the accuracy of the genome sequence could have significant impacts on comparative genomic analyses with this strain. The plant growth-promoting mechanisms used by two endophytic enteric bacteria, Klebsiella pneumoniae 342 and Enterobacter cloacae P101, were also investigated, utilizing their genome sequences to guide experiments. Both bacteria increased 11

PAGE 12

12 lateral root numbers on Arabidopsis plants and genome annota tion indicated each posses the auxin synthesis gene, ipdC As auxin is known to play a ro le in lateral root development, subsequent work on lateral root promotion focused on this phytohormone. As part of this work, a plasmid was constructed that expressed the ipdC gene and could be stably maintained without selection pressure. While these enteric endophyt es are related to human pathogens and Kp342 was found to have pathogenic potential, the info rmation gained by studying these bacteria could yield insight into enhancing the growth-promotin g effects of other endophytes. Thus, the work described here exemplifies how metagenomic, geno mic, and biological experiments can be used to gain both an overview and in depth view of plant-bacterial associations.

PAGE 13

CHAP TER 1 INTRODUCTION Bacteria can interact with plants in a variety of different ways. Many plant-associated bacteria are pathogens that cause disease in plants. One plant di sease studied in this work is known as Huanglongbing (HLB), and the bacter ium proposed to cause this disease, Candidatus Liberibacter spp., has yet to be cultured (Bov 2006). There are also many other bacteria that live in association with plants without causi ng disease symptoms. These bacteria can be epiphytes living on plant surfaces, symbionts living within plant tissue and creating symbiotic structures, or endophytes living in plant tissue without forming symbiotic structures. Such growth-promoting bacteria examined in this work include Gluconacetobacter diazotrophicus PAl 5, Klebsiella pneumoniae 342, and Enterobacter cloacae P101 (Chelius and Triplett 2000; Riggs et al. 2001; Se villa et al. 2001). Alarmingly, some plant-associated bacteria ar e also related to huma n enteric pathogens. Colonization of plants with such bacteria can lead to disease when they inhabit produce crops that are eaten raw. As a result, considerable effort has gone into studying how these human pathogens colonize and interact with plants. The ultimate goal of such research has been to gain enough insight into the colonization of plants by enteric human pathogens in order to limit it, thus making food supplies safe for human consumption. Alternatively, plant-associated bacteria are also studied extensively due to their positive effects on plant growth. In contrast to the work done on human pathogens, these plant growth-promoting strains have been studied in the hopes of enhanci ng the interactio n rather than limiting it. In this way, knowledge of plant gr owth-promoting bacteria can be extended toward increasing production in plant crops Interestingly, some of the plant-associated bacteria found to enhance plant growth turn out to be rela ted to human pathogens, as was the case with K. 13

PAGE 14

pneumoniae 342 used in the current study. As a resu lt, it is important to check even plant growth-promoting isolates for pathogenic potential before co nsidering their use in agricultural settings. Regardless of whether plant growth-promoti ng bacteria are huma n pathogens, study of such bacteria can still provide valuable inform ation. For example, the genome sequences of these bacteria can be used in comparative anal yses with the genomes of non-plant associated strains as well as the genomes of other plant-associated bacteria to identify genes unique to plant colonization and growth promotion. Such ge nes could be engineered into nonpathogenic bacteria to enhance their coloni zation and growth-promoting effects on plants. Therefore, even if a bacterial species cannot be used in agricultural settings due to pathogenicity, knowledge of the plant colonization and growth-promoting mechanisms these ba cteria utilize may be applied to enhance other, nonpathogenic pl ant growth-promoting bacteria. Given the breadth of interactions seen betw een plants and bacteria, and the increasing emphasis on genomic analysis of plant-associated bacteria, a range of interactions between plant-associated bacteria and their hosts were st udied in this work using a variety of different approaches. The literature on plant-associated ba cteria was reviewed and has been published in Annual Review of Phytopathology (Tyler and Triplett 2008). Though the emphasis of this literature review was on the pres ence of human pathogens in plants, plant growth-promoting bacteria and the prospects of using genomics in the study of these plant-as sociated bacteria are also discussed. In regards to examining plan t pathogens, research using metagenomics to identify bacteria within citrus tissue infected with the disease, HLB, is also presented. This work has been accepted for publication at Molecular Plan t-Microbe Interactions (T yler et al. in press), and is significant not only for th e knowledge it contribute s to the study of this devastating citrus 14

PAGE 15

15 disease, but also as an exampl e of how metagenomics can be u tilized to study other uncultured plant pathogenic bacteria. When examining plant growth-promoting bacteria work was also performed to distinguish between contradictory genome sequences of G. diazotrophicus PAl 5 and has been submitted for publication to BMC Genomics. Such work is important as the accuracy of genomic sequences could significantly impact the results of comparative genomic analyses planned for plant growth-promoting bacteria in the future. Las tly, while genomic and metagenomic-based studies are useful in providing global views of plant-bacterial interactions, it is still important to examine specific mechanisms in these interactions to confirm the observations seen in these studies, for sequence analysis can only provide so much information without veri fication in the actual bacterium. Therefore, the pl ant growth-promoting effects of K. pneumoniae 342 and E. cloacae P101 were also examined using a variety of a ssays, ultimately utilizing genomic sequences to guide experiments on the growth-promoting phenotype Thus, the work presented here aims to examine several aspects of plant-bacterial intera ctions, ranging in depth from examining entire genomes to studying specific mechanisms utilized by plant growth-promoting bacteria.

PAGE 16

CHAP TER 2 PLANTS AS A HABITAT FOR BENEFI CIAL AND/OR HUMAN P ATHOGENIC BACTERIA Introduction Non-plant pathogenic bacteria are common inhabitants of the interior of plants. These bacteria can prom ote plant growt h, improve the nitrogen nutrition of plants, and be pathogenic to humans. Early work in bacterial endophytes was encouraged by their abili ty to improve plant growth. As endophytic bacteria we re discovered in a variety of host plants, it became clear that enteric bacteria were frequent inhabitants of the plant cortex. These observations coupled with some well-publicized disease outbr eaks from the consumption of ra w alfalfa sprouts in the late 1990s led a number of groups to study the ability of enteric pathogens to co lonize the interior of plants. Human pathogens in raw produce took on national prominence in the United States between August and October of 2006 when 205 illn esses in 26 states were reported as a consequence of the consumption of spinach contaminated with Escherichia coli O157:H7. This outbreak resulted in 105 hospitaliz ations, 31 hemolytic uremic syndrome cases, and three deaths (Calvin 2007). The geographic scale of this epid emic startled the public and fostered a renewed interest in research in this ar ea. However, these outbreaks ar e not uncommon. According to the Centers for Disease Control and Preventions data (Lynch et al. 2006), from 1998 to 2002, an average of 25,674 food borne illn esses and 18 deaths in the United States were caused by 1329 outbreaks annually. The number of outbreaks traced to plant consumption (vegetables, fruits, and nuts) during that same period aver aged 56 and caused 2,109 illnesses each year. Although the proportion of outbreak s attributed to plant consumption is just 4.2%, outbreaks from plants appear to be more serious for human health as they cause 8.2% of the illnesses and 8% of the deaths. Assuming that 750 million meal s are consumed each day in the United States 16

PAGE 17

(roughly 2.5 m eals per day per person), food bor ne illnesses caused by plant consumption are very rare and our food supply is remarkably safe. Nevertheless, these outbreaks are well publicized and cause considerable alarm. Any outbreak stemming from raw produc e can cause far more harm if not reported quickly and followed by rapid source identification. Plant-deri ved outbreaks can be caused by many factors: Pathogens may be located on the plant surface or within plant tissues; pathogens may be unwittingly added to raw produce during food preparation. In this review, we discuss one source of th ese pathogens in plants. What is known about the ability of enteric bacteria to enter plants? To what extent do plant hosts regulate their entry? How do clinical enteric pathogens differ from plant-derived enteric pathogens? What do we still need to know to address this problem? How can improved knowledge lead to a reduction in outbreaks? For more information about the ep idemiology and fitness of human pathogens on plant surfaces, the readers are refe rred to the recent review by Bra ndl (2006). For a review of the diversity of bacteria in plan ts, the readers are referred to Rosenblueth and Martinez-Romera (2006). Cycle of Human Pathogens in the Environment Hum an pathogens have been found in plants in the environment as well as in crop plants. In nature, a high number of facultative human pathogens have been associated with the bryophyte Sphagnum including Staphylococcus, Hafnia Yersinia and Pantoea (Opelt et al. 2007). In addition, Pantoea, Salmonella Enterobacter, Citrobacter, and Klebsiella have been found to endophytically colonize the leguminous tree, Conzattia multiflora in Mexico (Wang et al. 2006). The presence of Salmonella in this plant indicates that plants in natural settings may harbor human pathogens (Wang et al. 2006). Seve ral members of the Enterobacteriaceae have been isolated from both wild and cultivated pl ants, including dandelions plantains, tomatoes, 17

PAGE 18

and potatoes (Markova et al. 2005). Though none of the specie s isolated in this work were known pathogens, m any were resistant to multiple antibiotics and capable of adhering to human erythrocytes (Markova et al. 2005). More widely known and studied is the occurr ence of human pathogens in crop plants due to fertilizing and watering with contaminated waste materials. High incidences of microbial contamination, including species of E. coli, Enterobacter cloacae and Klebsiella pneumoniae, have been reported in several ve getable species irriga ted with untreated wa stewater (Ibenyassine et al. 2007). Other cases of potential pathogens in crops include the isolation of Listeria monocytogenes and E. coli from organically grown lettuce and Enterobacter spp., Klebsiella and E. coli on mung bean, though the enterohemorragic E. coli O157:H7 strain was not identified in either of these cases (Loncarevic et al. 2005; Robert son et al. 2002). One explanation for the occurren ce of human pathogens in plants is that they cycle through the environment, using plants as an alternative host to surviv e in the environment and as a vehicle to recolonize animal hosts once ingested. To accomplish this process, enteric pathogens shed in animal feces must be transported to and come in contact with plants eaten by the host. One possible mechanism for this transfer is through nematodes, such as Caenorhabditis elegans which are capable of transporting Salmonella newport to fruits and vegetables through soil (Kenney et al. 2006). When soil inoculated with C. elegans was placed on top of contaminated manure, the bacteria were found on produce above the soil surface, whereas the bacteria were generally not found on the vegetables when C. elegans was absent from soil (Kenney et al. 2006). Other possible mechanisms of pathogen transp ort to plants include irrigation water and runoff from livestock pastures. Numerous inst ances have been reported whereby contaminated 18

PAGE 19

irrigation water led to contam inated crops. In a field study, a wide ar ray of produce plants, including lettuce, parsley, carrots, and tomato, had elevated concentrations of fecal indicator organisms when irrigated with highly polluted e ffluents, whereas irrigation with only slightly polluted water resulted in lower numbers of indi cator organisms (Armon et al. 1994). In another study, the same Salmonella strains present in ed ible parts of vegetables were found in the contaminated raw wastewater used to irrigate the plants (Melloul et al. 2001). Vegetables that grow on the surface of the ground, such as lettuce or parsley, were more contaminated than those that develop above the surface, such as tomato (Melloul et al. 2001). Salmonella can also contaminate alfalfa sprouts produced with cont aminated water. A strong correlation was observed between the numbers of Salmonella enterica cells present in alfalfa sprouts and in the irrigation wastewater (Howard and Hutcheson 2003). The population of S. enterica cells was able to grow in the wastewater, using com pounds in it as a sole car bon and nitrogen source (Howard and Hutcheson 2003). In addition, the b acteria multiplied rapidly on alfalfa during the first 24 hours of the germination process (Howard and Hutcheson 2003). The ability to grow on alfalfa was independent of serovar, source of isolation, or virulence, with isolates from meat and stool reaching densities similar to those of plan t isolates (Howard and Hutcheson 2003). This rapid establishment on alfalfa sprouts and similar colonization by all strains tested demonstrates how Salmonella could pose a significant problem in contaminated plants meant for human consumption. In addition to culturing bacteria, polymerase chain reaction (PCR) of repetitive deoxyribonucleic acid (DNA) sequences has also been used to study the role of irrigation water in contaminating plants. Enteropathogenic E. coli strains present in irrigation water were also detected on vegetables and in soil by this method, demonstrating that bacteria can be transported 19

PAGE 20

by irrigation water and result in contam ina tion of produce (Ibenyassine et al. 2006). The irrigation method employed also app ears to play a role in contam ination: Greater contamination of plants was found with furrow irrigation than w ith subsurface drip irrigation (Song et al. 2006). Human pathogens contaminate crops from both contaminated manure and irrigation water. When exposed through either contaminated manure or irrigation water, E. coli O157:H7 persisted in soil for 154 to196 days after treatment and were detected on onion and carrots for 74 and 168 days, respectively (Islam et al. 2005). In the case of S. enterica serovar Typhimurium, bacteria survived in soil 203 to 231 days and were found on radishes and carrots 84 and 203 days after sowing in soil with contaminated manure or irrigation water (Islam et al. 2004). Thus, both contaminated soil and tainted irrigation water can contribute to coloniza tion of crops by human pathogens. Another mechanism for the transfer of pathogens is runoff from contaminated areas such as cow pastures that spreads bacteria to land where crops are cultivated. For example, isolates of E. coli can be transported rapidly in runoff, with attachment to soil part icles affecting how far they travel (Muirhead et al 2006). When not preattached to larger particles (>45 m), E. coli in runoff attach to much smaller fragments (<2 m) and are not slowed or deposited during runoff transport (Muirhead et al. 2006). Cattle and other livestock animals kept in close proximity to croplands are a potential source fo r pathogenic bacteria in runoff and manure. Ruminant diet appears to play a role in survival of certain pathogens in cow feces: E. coli O157:H7 lived longer in manure from cows eating grass and mai ze silage than in manure from cows on a pure straw diet (Franz et al. 2 005). In addition, the pathogen Campylobacter jejuni was found in the feces of feedlot cattle (Besser et al. 2005). Th e presence of these bact eria in the feces of 20

PAGE 21

livestock could possibly lead to their spread to crop plants th rough runoff or from the use of manure from these cattle as soil amendments. Another major factor important to colonization of plants is how long these bacteria persist in the soil before dying. E. coli O157:H7 present in animal waste was capable of surviving in soil, although the source of the waste did have an effect on survival (Williams et al. 2007). Bacteria from ovine stomach content waste survived in higher numbers than bacteria from cattle slurry (Williams et al. 2007). E. coli O157:H7 was also found to pe rsist in soil for 8 weeks after treatment with contaminated manure, though it wa s not found in or on lettuce leaves or roots (Johannessen et al. 2005). The lack of E. coli O157:H7 on any part of lettuce may be due to the presence of Pseudomonas flourescens in the rhizosphere, which was found to inhibit O157:H7 in culture (Johannessen et al. 2005). The use of Ecos an sludge in agriculture has also been linked to bacterial contamination of produce. The application of sludge containing Salmonella fecal coliforms, and fecal streptococci to soil increa sed the bacterial counts in spinach and carrot plants, with increasing amounts of sludge result ing in higher bacteria l numbers in plants (Jimenez et al. 2006). Listeria innocua and Clostridium sporogenes present in contaminated soil amendments have also been found to persist in the soil and were both found on the leaves of parsley grown in the contaminated soil, though they were not observed within the plants (Girardin et al. 2005). While Clostridium spores persisted in soil for 16 months, Listeria decreased by 10,000,000-fold over 90 da ys (Girardin et al. 2005). Because plants can serve as hosts for enteric human pathogens in the environment, these enterics may be adapted to surviving longer and in higher numbers in soil when the plants are present. Gagliardi and Karns (2002) found that the presence of certain crops increased the persistence of E. coli O157:H7 in soil. In unplanted, fallow soils, E. coli O157:H7 persisted only 21

PAGE 22

for 25 to 41 days, but was found up to 92 and 96 days if alfalfa or rye plants were present, respectively (Gagliardi and Karns 2002). Ibekwe and associates (2004) also found that the presence of roots in contam inated soils increased concentrations of E. coli O157:H7. In this study, E. coli introduced through irrigation water was found to reach higher densities in rhizosphere soils than in non-rhizosphere soils (Ib ekwe et al. 2004). In both studies, clay was found to increase th e persistence of E. coli O157:H7 and other coliforms (Gagliardi and Karns 2002; Ibekwe et al. 2004). In addition, bacter ial populations in soil may increase after the addition of plants to soil. For example, the ba cterial population in soil spiked after the addition of plant material and then fluctu ated in a wave-like fashion that was not found to be the result of nitrogen shortages or pH (Zelenev et al. 2005). In contrast to th ese results, the presence of maize roots did not effect survival of E. coli in soil (Williams et al. 2007). Likewise, the presence of legume crops other than alfalfa did not increase persistence of E. coli more than in fallow soils (Gagliardi and Karns 2002). In addition to direct colonization from expos ure to contaminated soil and water, some pathogens that establish themselves endophytically in plants may be seedborne and inherited from generation to generation. In Conzattia trees, nonpathogenic bacteria that were isolated as endophytes were also found in all seeds that were tested, indicating that these endophytes may be seedborne and inherited from the previ ous generation (Wang et al. 2006). In S. enterica and E. coli O157:H7 on Arabidopsis the bacteria were occasionally found to contaminate seeds, and this contamination was highly correlated with contaminated chaff (Cooley et al. 2003). Salmonella survived both in and on tomato throu ghout plant growth, fr om inoculation to flowering and fruit ripening (Guo et al. 2001). Salmonella were found in tomato fruit from plants inoculated on the stem bot h before and after flower set, with 43% and 40% of tomatoes 22

PAGE 23

from these treatments testing positive for the b acteria, respectively (G uo et al. 2001). Although most inoculated flowers were reported to abort, 25% of the tomatoes fr om inoculated flowers harbored Salmonella (Guo et al. 2001). In apples, E. coli O157:H7 were observed attached to seed integuments, with infilt ration occurring through the blo ssoms calyx and traveling up the floral tube to the internal parts of the apple (B urnett et al. 2000). These observations support the hypothesis that the enteric pat hogens colonizing produce can sp read to the seeds and be transmitted to the next generation. Observations of Pathogens on Plant Surfaces Strains of enteric pathogens differ in their ab ility to colonize the surface of plants. For example, cells of E. coli O157:H7 were rinsed off af ter repeated washes, whereas E. coli serotypes isolated from cabbage roots that had b een exposed to sewage adhered to sprouts (Barak et al. 2002). In contrast, all S. enterica serovars tested did not differ among themselves in binding to alfalfa, and they all bound significantly better than did E. coli O157:H7 (Barak et al. 2002). Surface moieties on cells are not n eeded for the initial attachment of E. coli O157:H7 to lettuce, since nonbiological FluoSph eres, as well as live and dead E. coli cells were retained on plants at similar densities (Solomon and Ma tthews 2006). This, and the observation that E. coli did not bind as strongly to plants as Salmonella demonstrate that not all pathogens are as disposed to colonize plants as others. Given that strains vary in their ability to bind to plant tissue (Barak et al. 2002), the E. coli strain used in this study may be one that is not efficient at attachment. E. coli did not grow on plants as well as Salmonella with E. coli O157:H7 growing significantly less than S. enterica on sprouting alfalfa (Cha rkowski et al. 2002). Salmonella also colonized both the seed coat and sprout root s, whereas O157:H7 only colonized alfalfa roots (Charkowski et al. 2002). One commonality between colonization by these two bacteria is that 23

PAGE 24

both tem perature and inoculum density affected initial colonization by both strains, with both higher temperatures and greater inoculum densiti es resulting in higher colonization (Charkowski et al. 2002). Temperature is also important to the colonization of the cilantro phyllosphere by S. enterica serovar Thompson, which was isolated from a cilantro-related outbreak (Brandl and Mandrell 2002). The pathogen did not colonize as well as other plant surface bacteria at 22C, but reached higher numbers when grown at higher temperatures, most likely by outcompeting the other epiphytes due to its higher grow th rate (Brandl and Mandrell 2002). Endophytic Colonization Endophytic colonization of hum a n pathogens into plants ha s been observed in several studies, but how the bacter ia enter these plants and where they are localized within plants are not completely understood. The intern alization of human pathogens in to produce is of particular interest since it may prot ect these bacteria from sanitizing tr eatments meant to make fruits and vegetables safe for consumption. Additionall y, the amount of endophytic colonization varies according to the bacterium and plant species examined. From numerous observations of bacterial colonization of plant root s, it is widely assumed that bacteria can colonize and enter roots at sites of lateral root emergence. Both Salmonella and E. coli have been observed by confocal microscopy to invade plant roots through latera l root cracks of Arabidopsis thaliana (Cooley et al. 2003). E. coli O157:H7 was also observed colo nizing preferentially at root junctions on lettuce (Jablasone et al. 2005). Confocal microscopy of the endophyte K. pneumoniae 342 on plants revealed si gnificant colonization around lateral root cracks of Medicago sativa M. truncatula A. thaliana Triticum aestivum and Oryza sativa, suggesting a possible entry site for the bacteria into th e plant (Dong et al. 2003a and b). In addition, Serratia cells were seen at higher c oncentrations around lateral root emergence sites of rice, again 24

PAGE 25

indicating that these sites m ay be potential entry points for bacteria into plants (Gyaneshwar et al. 2001). Several studies have focused on inte rnalization of the enteric pathogen E. coli O157:H7, related species of Salmonella and Listeria E. coli has been observed to be internalized through the root system of maize grow n in contaminated hydroponic media and was found in the shoot of plants 48 hours after inoculati on (Bernstein et al. 2007). Dama ging the root system by cutting off root tips or removing root s at the root shoot juncti on increased the number of E. coli internalized in shoots 27.8 and 23.9 times greater than in undama ged roots (Bernstein et al. 2007). Though capable of colonizing maize roots, E. coli was observed to decline over time, but a sharp reduction in population was not observed until 4 days post inoculation, and bacteria were still present at approximately 102 colony forming units (CFU)/g fresh weight after 7 days (Bernstein et al. 2007). E. coli O157:H7 has also been found to in ternalize within radish sprouts and carrots (Auty et al. 2005; Itoh et al. 1998). E. coli found in root-inoculated radish sprouts was observed within stomata of cotyledons a nd the inner tissues of edible hypocotyls and cotyledons (Itoh et al. 1998). In carrots, E. coli was found at cell juncti ons and intercellular spaces up to 50 m within carrot tissue, but did not pene trate the carrot cells (Auty et al. 2005). Several studies have observe d the internalization of Escherichia coli O157:H7 within lettuce plants (Franz et al. 2007; Solo mon et al. 2002; Wachtel et al. 2002). E. coli O157:H7 has been found within the tissue of thoroughly surface-ster ilized lettuce leaves at a density of 3.95 log CFU/g (Franz et al. 2007 ). Two studies found that E. coli O157:H7 can be internalized into lettuce when exposed to contaminated irri gation water and manure-tr eated soil (Solomon et al. 2002; Wachtel et al. 2002). Wachte l and associates (2002) observed E. coli O157:H7 binding preferentially to the roots of lettuce from contam inated irrigation water, and also observed strains 25

PAGE 26

in the edible portion of the plan t. In confirm ation of this observation, fluorescently labeled bacteria could be seen moving in the vasculature of a hypocotyl, most likely in the xylem (Wachtel et al. 2002). When grown in soil, E. coli O157:H7 was associated with the plants 3 days after inoculation, even with low inoculum doses, indica ting that the bacteria could multiply in the plant (Wachte l et al. 2002). Solomon and associates (2002) observed E. coli O157:H7 up to a depth of 45 m within lettuce leaf tissue in aggr egates in intercellular spaces and found that direct contact with contaminated ma terial was not required for internalization of the bacteria within edible leafy parts of the plant. Strains of Salmonella have also been obser ved within lettuce. S. enterica serovar Typhimurium was found in high numbers of 2.57 log/g in surface-sterilized lettuce (Franz et al. 2007). The ability to internalize and endophyti cally colonize lettuce is not common to all Salmonella strains. In a study of three S. enterica serovars, all strains we re capable of colonizing the plant epiphytically, but only the Dublin sero var could colonize endophy tically (Klerks et al. 2007). Differences between Salmonella strains have also been observed in tomato, with Montevideo being the most persistent and Poona the most dominant (Guo et al. 2001). S. enterica serovar Typhimurium colonizes barley ( Hordeum vulgare ) roots endophytically up to 106 CFU/g fresh weight (Kutter et al. 2006). Fluorescence in s itu hybridization and confocal laser microscopy confirmed endophytic colonization, and revealed that Salmonella had colonized root cells and was present in the inner co rtex of the roots (Kutter et al. 2006). Endophytic colonization by human pa thogens has been studied in several other plants. In A. thaliana, S. enterica and E. coli O157:H7 colonize the entire plan t following root inoculation, indicating that the bacteria can mi grate on or in plants and are dete ctable in the plants for up to 21 days after inoculation (Cooley et al. 2003). Internaliz ation of enteric pat hogens has also been 26

PAGE 27

investigated in legum e species. Both E. coli and Salmonella Montevideo were found to internalize into mung bean sprouts from inoculated seeds (Warriner et al. 2003). These bacteria were observed in roots and between epidermal cells on hypocotyls and were isolated from surface-sterilized tissue and apoplastic fluid, indicating the bacteria were internalized and protected from sterilization treatment s (Warriner et al. 2003). Also, E. coli was visualized in the vascular system of mung bean sprout hypocotyls using -glucuronidase (GUS) staining (Warriner et al. 2003). In testi ng several pathogenic strains of E. coli and S. enterica for endophytic colonization of alfalfa ( M. sativa ) and its relative, M. truncatula Dong and associates (2003a) found that the strains varied greatly in their ability to coloni ze plants. Although endophytic colonization was observed by all strains tested, significant variability in the extent of endophytic colonization between strains was obser ved (Dong et al. 2003a). The pathogenic strains were compared to a plant-isolated endophyte, Klebsiella pneumoniae 342 and a lab strain of E. coli K12 and were found to have colonized the pl ants at a level betw een these two strains (Dong et al. 2003a). Endophytic colonization of E. coli O157:H7 within fruits has also been studied. In apple, scanning confocal laser microscopy (S CLM) has been used to observe E. coli O157:H7 attached preferentially to discontinuiti es in the cuticle of apples and at puncture wounds down to 70 m in the tissue (Burnett et al. 2000). E. coli O157:H7 has also been observed colonizing apple lenticels, in some cases to a depth of 40 m (Burnett et al. 2000). Bruising and rubbing of apples appears to increase the depth to which bacteria can internalize within lenticels (Kenney et al. 2001). Beyond infiltration from the surface, E. coli O157:H7 can colonize the interior of apples in the core and subsurface st ructures, attaching to the cart ilaginous pericarp and seed integuments with infiltration o ccurring through the blossoms cal yx and traveling up the floral 27

PAGE 28

tube to the internal parts of the apple (Burnett et al. 2000). In wounded apples, E. coli O157:H7 was observed to colonize and survive by causing cell m embranes to degrade and release their contents (Janes et al. 2005). E. coli within apple tissue formed gra nules and vesicles that were not present in bacterial growth in brain heart infusion broth, possibly due to osmotic conditions in the apple tissue (J anes et al. 2005). In oranges, E. coli O157:H7 and Salmonella colonize the interior of fruit at the same rate as the entry of a dye, suggesting a passive mechanism of entry (Eblen et al. 2004). Oranges with large puncture wounds were more likely than on es with smaller wounds to internalize the bacteria (Eblen et al. 2004). Both E. coli and Salmonella were able to grow in oranges at 24C (Eblen et al. 2004). The roots and xylem of c itrus plants can also be colonized by enteric bacteria. K. pneumoniae 342 (Kp342) was able to e ndophytically colonize both Citrus sinensis and Catharanthus roseus though it was present in higher numbers in C. roseus plants (Lacava et al. 2007). The bacteria that were inoculated onto root s were subsequently found in the xylem of roots and branches of both species tested, with Kp342 colonies observed in the xylem by fluorescent microscopy of seedling cr oss-sections (Lacava et al. 2007). Kp342 has also been found to colonize the interiors of M. sativa (alfalfa), M. truncatula A. thaliana, T. aestivum (wheat), and O. sativa (rice) (Dong et al. 2003b). Kp342 was more efficient at colonizing the plant apopla st compared to the type strain of K. pneumoniae (Dong et al. 2003b). Differences in coloniza tion levels between plants were also observed, with monocots being colonized in higher numbers than dicots (Dong et al. 2003b). Ente ric bacteria isolated from plant tissues colonized the interior of plants in higher nu mbers than those isolated from clinical or other settings. For example, of several Salmonella strains tested for endophytic colonization of alfalfa seedlings, strain SCH7976 colonized the interior more than the other 28

PAGE 29

strains. This strain was derived from an alfa lfa sprout salmonellosis outbreak in California in 1998. Kp342, which was originally isolated from a nitrogen-efficient line of m aize (Chelius and Triplett 2000), colonizes plants in far higher numbe rs than any other enteri c bacterium tested to date (Dong et al. 2003a and b). Damage to plant tissue can provide points of entry for bacteria into plant tissue. For example, bacteria were not detected on the su rface of unbroken spinach leaves using scanning electron microscopy, but they were present on those leaves where th e cuticle was broken, spreading to the internal palisade parenchyma (Babic et al. 1996). In this case, strains of the Pseudomonadaceae, Enterobacteriaceae (including K. pneumoniae ), and Micrococcaceae, and lactic acid bacteria were found (Babic et al. 1996). Bacterial Competition When colonizing plants, hum an pathogens undoubtedly come into contact with native bacteria and must overcome microbial competition to establish themselves. These interactions vary according to which pathogens and host plants are being studied and wh ich bacteria are used as antagonists. When looking at Salmonella growth on alfalfa sprout s, coinoculation with bacteria isolated from market sprouts reduced Salmonella numbers more than coinoculation with P. fluorescens 2-79 seven days after inoculation, whereas microbial communities from lab-grown sprouts showed the least reduction on Salmonella growth (Matos and Garland 2005). The observation that microbial commun ities from market sprouts reduced Salmonella numbers more than those from lab-grown sprouts suggest s that bacteria undergo selective conditions in the field and/or during handling on the way to market that result in a more robust native community for competition against contaminatin g pathogens (Matos and Garland 2005). This conclusion is supported by the observation that market sprout microbial communities could grow faster on a wide array of substrates (Matos and Garland 2005). 29

PAGE 30

In m ost cases, rhizobacteria did not reduce the persistence or number of enteric pathogens on plants, with the exception of E. cloacae, which reduced the numbers of E. coli O157:H7 and L. monocytogenes on lettuce (Jablasone et al. 2005). Another Enterobacter, E. asburiae was also observed to be antagonistic toward E. coli O157:H7 on plants, lowering colonization on lettuce by 20to 30-fold (Cooley et al. 2006). This effect is most likely due to competition for carbon and nitrogen sources, since E. asburiae and E. coli both utilize ov er 20 of the carbon sources present in plant exudates (Cooley et al. 2006). The conditions on the plant where bacteria initially land may also play a role in their survival. Immigrant bacteria of both P. agglomerans and P. flourescens were twice as likely to survive on leaf surfaces when they landed on aggregates of previously established bacteria than when they landed on uncolonized areas (Monier and Lindow 2005). In addition, th e plant structures on which these bacteria landed also influenced their survival, with b acteria landing on veins and granular trichomes having higher survival rates th an those landing on hooked trichomes (Monier and Lindow 2005). In addition, bacteria landing on epidermal cells altered by P. agglomerans (which were not necessarily coincident) had higher survival rates than if they had landed on normal, unaltered epidermal cells (Monier and Lindow 2005). It would be interesting to see if similar relationships are seen in human pathogens colonizing plant surfaces. Role in Plant Growth Promotion Many plan t-associated bacteria are known to have beneficial effects on plant growth, and the capability of some animal pathogens to enha nce plant growth may i ndicate an evolutionary reason why these pathogens evolved to colonize plant hosts, with the plant benefiting from the bacteria, and the bacteria gaining nutrients from the plant and protection from the environment. There are numerous instances in the literature of bacterial plant growth pr omotion, so this review gives only a brief summary of the topic as it relates to potential human pathogens. Several 30

PAGE 31

genera of endophytes isolated from the leguminous tree, C. multiflora were identified as members of the Enterobact eriaceae family, including Enterobacter Klebsiella Citrobacter, and Pantoea and were shown to increase the height of s eedlings after inoculation (Wang et al. 2006). Plant-associated bacteria can increase plant gr owth by several mechanisms, the most commonly studied of which involve producti on or lowering of plant hormones important to development. Several plant hormones have been identif ied in the culture supernatants of K. pneumoniae E. coli Proteus vulgaris and Bacillus cereus including auxin, gibberell in, and cytokinin, with Klebsiella producing significantly more auxin than th e other strains (Karadeniz et al. 2006). El-Khawas and Adachi (1999) also identified auxin in K. pneumoniae supernatants and found that application of the supernat ants to rice roots resulted in increased root length, surface area, and fresh weight. The role of bacterial auxin production has been further confirmed in studies of other plant-associated ba cteria. Knocking out the ipdC auxin synthesis gene in Azospirillum brasilense lowered auxin production by 90% and result ed in reduction of the growth-promoting effect this bacteria has on the root system (Barbieri and Galli 1993; Dobbelaere et al. 1999). Also, in Pseudomonas putida GR12-2, the wild type strain in creased root length 35% to 50% more than an ipdC mutant in mung bean (Patten and Glick 2002). In addition to secreting plant hormones, some plant growth-promoting bacteria produce enzymes that interfere with and break down horm ones that inhibit plant growth. For example, 1-aminocyclopropane-1-carboxylate (ACC) deaminase, an enzyme that lowers the concentration of ethylene in plants by cleaving ACC, the precurs or of this hormone, is found in several plant growth-promoting bacteria (Glick 2005). Inoculation with an ACC deaminase producing strain, E. cloacae UW4, resulted in increased root le ngth on canola seedlings, whereas an acdS mutant in this strain that does not possess ACC deaminase activity is diminished in its ability to enhance 31

PAGE 32

root length (Li et al. 2000). In addition, transferring the ACC deam inase gene from E. cloacae UW4 into a P. fluorescens strain lacking this enzyme confers the ability to increase root length in canola, suggesting that bacterial ACC deaminase lowers ethylene concentrati ons inside the plant, resulting in an increase in root length (Wang et al. 2000). ACC deaminases ability to enhance root growth has also been linke d to changes in gene expressio n. For example, inoculation of Arabidopsis with E. cloacae UW4 increased the transcription of genes involved in cell division and proliferation while down-re gulating genes involved in ethylene-induced stress and plant defense signaling (Hontzeas et al 2004). These results make sens e since bacteria would need to increase cell growth a nd proliferation in order to increase root length. Bacterial Mechanisms of Colonization The m olecular mechanisms used by bacteria to colonize the interior of plants are still largely unknown. Observations th at different bacteria colonize different plants to varying degrees indicate that colonization of the plant interior is an activ e process that is controlled by genetic determinants on both sides of the intera ction (Dong et al. 2003b). Most studies focus on reporting the internalization of bacteria and their general loca tion on and within plants, though some have linked bacterial colonization to specific genes. While many studies have been done on plant growth-promoting bacteria or plant path ogens, similar colonization mechanisms may be utilized by human pathogens. Evidence for the host regulation of endophytic co lonization by enteric bacteria came from the observation that the interior of an ethylene-insensitive mutant of M. truncatula was colonized in much higher numbers than the wild type pl ant (Iniguez et al. 2005). Subsequently, ethylene addition to plants greatly d ecreased endophytic colonization by Salmonella and K. pneumoniae 342, which could be reversed by the addition of an ethylene action inhibitor (Iniguez et al. 2005). As ethylene triggers induced syst emic resistance (ISR) in plants, Arabidopsis lines that vary in 32

PAGE 33

both the ISR and system ic acquired resistance (SAR) defense responses were tested for endophytic colonization. The results showed that ISR and SAR reduce plant invasion of Salmonella strains but only ISR affects interior colonization by Kp342 (Iniguez et al. 2005). Salmonella mutants deficient in the pr oduction of flagella or the type III secretion system showed increased endophytic col onization (Iniguez et al. 2005). These systems appeared to elicit SAR but not ISR. As Kp342 lacks both flage lla and type III secretion systems (Fouts et al. 2008), it is not surprising that this strains ability to colonize the in terior of plants is influenced only by ISR, not by SAR. These results suggest why Klebsiella strains are so commonly found in the interior of plants (Chelius and Triplett 2000; Chelius and Triplett 2001; Engelhard et al. 2000; Kuklinsky-Sobral et al. 2004 ; Reiter et al. 2003; Rosenblueth et al. 2004; Surette et al. 2003). One feature of the Klebsiella genus is that they lack flagella and, as in Kp342, may commonly lack some secretion systems as well. As a result, Klebsiella probably elicit a lower response from plant defenses than do most other bacteria. Microarray analysis revealed that several pa thogenicity genes, incl uding type II secretion genes and a gene involved in capsular polysaccharide synthesis, were induced in S. enterica serovar Typhimurium during coloni zation in response to lettuce r oot exudates (Klerks et al. 2007). These genes are believed to aid in att achment to the plant (K lerks et al. 2007). Differential expression of genes rela ted to carbon utilization, including OtsA (trehalose-6-phosphate synthase) and UhpC (hexose phosphate utilization protein), was also observed (Klerks et al. 2007). These results, in combination with chemotaxis experiments showing that the bacteria move toward root exudates, suggest that root exudates may serve as carbon sources for Salmonella and may condition S. enterica for root attachment by triggering chemotaxis and turning on genes that could play a role in adherence (Klerks et al. 2007). 33

PAGE 34

In the nitrogen-fixing endophyte, Azoarcus sp. strain BH72, an endoglucanase gene is im portant for bacterial colonization of rice roots (Reinhold-Hurek et al. 2006). This endoglucanase, encoded by eglA shows high homology to endoglucanases from the phytopathogenic bacteria, Xanthamonas campestris and Ralstonia solanacearum (Reinhold-Hurek et al. 2006). This gene is transcriptionally activ ated in the presence of rice roots, at lateral root emergence sites and lateral root tips (Reinhold-Hurek et al. 2006). A mutant in this gene is unable to spread into rice shoo ts and exhibits reduced intracellular colonization (Reinhold-Hurek et al. 2006). Motility is also known to play a role in endophytic colonization. For example, root colonization and migration of the bacteria to other parts of the plant were decreased in nonmotile Salmonella enterica mutants compared to the wild type st rain (Cooley et al. 2003). In addition, twitching motility is important for endophytic colonization by Azoarcus (Bohm et al. 2007). The structural component genes of type IV pili, pilT and pilA are responsible for twitching motility, and mutants in these genes abolished the twitching phenotype (Bohm et al. 2007). The pilA mutant was strongly reduced in both rhiz osphere and endophytic colonization. The pilT mutant was also greatly reduced in endophytic colonizati on compared to wild type but was only 50% reduced in rhizosphere colonization. Thus th e twitching phenotype is pivotal in endophytic colonization but less important in root surface atta chment (Bohm et al. 2007). The type IV pili locus, pilAB is also important to plant colonization, w ith both genes required for attachment to rice roots by Azoarcus (Dorr et al. 1998). The pilA gene encodes a short pilin precursor whereas pilB shares similarity with FimF, a type I fimbria from E. coli (Dorr et al. 1998). The similarity of plant colonization genes to pathogenicity factors in human pat hogens indicates these bacteria may be capable of colonization of their hosts by similar mechanisms. 34

PAGE 35

Two partner secretion system s (TPS) also play roles in plant coloniza tion by bacteria. In TPS, a TpsA family exoprotein with specific cons erved secretion signals is transported across the membrane by a TpsB family channel-forming tr ansporter that recognizes the secretion signal (Jacob-Dubuisson et al. 2001) The plant pathogen, Erwinia chrysanthemi possesses a TPS, hecAB in which hecA encodes an adhesin (Rojas et al. 2002). A mutant in this hecA gene had reduced attachment, cell aggregate formation, and virulence on Nicotinia clevelandii (Rojas et al. 2002). Homologs of this gene appear in both plant and animal pathogens and seem to be universal among necrotic plant pathogens (R ojas et al. 2002). In the mutualist P. putida KT2440, a TPS, hlpBA is involved in seed colonization and iron uptake (Molina et al. 2006). The hlpA gene encodes a secreted protein similar to iron-regulated hemolysins and HlpB is responsible for its transport acro ss the membrane (Molina et al. 2006). Mutations in this operon result in reduced colonization of corn seed a nd root attachment, demonstrating that the HlpA protein plays a role in plant at tachment (Molina et al. 2006). There are other factors involve d in bacterial attachment and plant colonization. In P. putida KT440, a cell density-regulated gene, ddcA, is involved in corn seed colonization (Espinosa-Urgel and Ramos 2004). The func tion of this gene is still unknown, but ddcA mutants are reduced in seed adhesion whereas co mplementing the gene restores adhesion (Espinosa-Urgel and Ramos 2004). ddcA expression is induced both by cell density and by seed exudates, indicating that both bacterial a nd plant signals can regulate colonization (Espinosa-Urgel and Ramos 2004). Of interest in regard to human pathogens in plants, ddcA is a member of a conserved protein family found in many prokaryotes, including a gene expressed during macrophage invasion by S. enterica serovar Typhimurium and a phosphate starvation gene, psiE in E. coli (Espinosa-Urgel and Ramos 2004). 35

PAGE 36

An argum ent has been made for passive entry of E. coli into lettuce tissues based on the observation that FluoSpheres enter these plants at a similar rate and in similar numbers to E. coli cells (Solomon and Matthews 2005). However, comp arison of living cells to inanimate objects can be difficult to interpret. Just because the rate and number of entry is the same between the two does not mean that the modes of entry are th e same. In addition, plant defenses could be repressing the number of viable E. coli in plants, which can result in similar numbers of FluoSpheres and cells within the plant. FluoSpheres could thus be entering passively at a lower rate while E. coli cells are battling against plant defenses to maintain a relatively low number of viable cells within the plant. The simplest first experiments in this area would be to determine the effects of plant defense elicitors su ch as ethylene or salicylic acid on E. coli endophytic colonization of lettuce plants. Surface adhesion plays a large role in pathoge n colonization of plants. Several genes and mechanisms have been identified as being involved in attachment of human pathogens to plants. These mechanisms include curli, fimbriae, a dhesins, and capsule production (Hassan and Frank 2004; Jeter and Matthysse 2005; Torres et al. 2005) When the curli biosynthesis regulatory gene, mlrA or the curli biosynthesis operon, csgA G were introduced into E. coli K12, it conferred the ability to attach to alfalfa, while deletion of the csgA gene in diarrheagenic E. coli O157:H7 strains did not reduce att achment compared to wild type Thus, the authors concluded that although curli may conf er plant attachment to E. coli K12, they are not necessary for attachment by pathogenic E. coli as it may have multiple mechanisms involved in attachment (Jeter and Matthysse 2005). Transferring the adhesin genes cah and aidA1 or the csg loci encoding fimbria formation into a nonbinding E. coli K12 also increased its ability to bind to alfalfa seeds and sprouts, though these genes di d not have an additive effect on increasing 36

PAGE 37

attachm ent when introduced together (Torres et al. 2005). As was observed with curli genes, deletion of these genes from a strain of E. coli O157:H7 did not alter binding to alfalfa; the only gene found to be required for attachment was ompA (Torres et al. 2005). These results suggest adhesins that mediate attachment in E. coli O157:H7 have overlappi ng and redundant functions (Torres et al. 2005). In addition, capsule production has also been a ssociated with the ability of E. coli O157:H7 to attach to le ttuce (Hassan and Frank 2004). Common Virulence Factors in Plant and Animal Pathogens Several studies have revealed the presence of similar virulenc eand pathogenicity-related genes within both plant and anim al pathogens. Pseudomonas aeruginosa PA14 has served as a model organism for this work as it is a br oad host range pathogen. This opportunistic mammalian pathogen has been demonstrated to be a facultative pathogen of the model plant, A. thaliana, causing local and systemic infection resulting in plant d eath (Plotnikova et al. 2000). The bacteria were able to enter through stom ata and wounds on leaves, colonize intercellular spaces, move basipetally along the vascular pa renchyma, and disrupt plant cell walls and membrane structures (P lotnikova et al. 2000). Several specific genes ha ve been identified in P. aeruginosa as common bacterial virulence factors. In the process of developi ng and testing plant models to help elucidate virulence mechanisms of PA14, 9 mutants isolated for reduced vi rulence to plants were also found to be required for complete virulence on a burned mouse model, several of which correspond to genes with no known f unction (Rahme et al. 2000). So me of the genes identified in this study include toxA plcS dbsA hrpM gacA and gacS (Rahme et al. 2000). Of these genes, toxA plcS and gacA were previously found to be common virulence factors necessary for causing disease in both Arabidopsis and a mouse model (Rahme et al. 1995). Two of these common virulence factors were the exported proteins, endotoxin A and phospholipase S. 37

PAGE 38

Endotoxin A, encoded by toxA inhibits protein synthesis (Ig lewski and Kabat 1975; Ohm an et al. 1980), and phospholipase S, encoded by plcS attacks eukaryotic membranes (Ostroff and Vasil 1987). The dsbA gene encodes a periplasmic disulfide bond-forming enzyme and may function to affect periplasmic virulence-rela ted proteins (Bardwell et al. 1991; Rahm et al. 2000). DsbA has been observed to be important to the pathogenici ty of both human and plant pathogens, including Shigella flexneri, Vibrio cholera and E. chrysanthemi (Peek and Taylor 1992; Shevchik et al. 1995; Watarai et al. 1995). The hrpM gene is homologous to the E. coli gene, mdoH involved in membrane-derived oligosaccharide synthesis (Loubens et al. 1993). Though its role in E. coli pathogenesis is not understood, deletions in hrpM in P. syringae pv. syringae reduces virulence, abolishing plant disease sympto ms (Anderson and Mills 1985). GacA and GacS encode proteins from a two-component family regulator, in which GacS is the sensor kinase and GacA is its response regulator (Hrabak and Willis 1992; Laville et al. 1992). The gacA gene functions as a transcriptional regulator of pathogenicity gene s encoding extracellular products (Rich et al. 1994). The observation that gacA functions at the regulatory le vel suggests that these common virulence factors share common re gulatory mechanisms (Rahme et al. 1995). Mutants in these genes in P. syringae are decreased in lesion formation on bean (Hrabak and Willis 1992; Hrabak and Willis 1993; Rich et al. 1994). Overall, this study found there are several universal bacterial virulence mechanisms that are highly conserved in P. aeruginosa and are used to infect evolutionarily divergent hos ts (Rahme et al. 2000). Two modular pathogenicity islands have been found in P. aeruginosa PA14, with genes from a wide array of bacterial species and mobile elements (He et al. 2004). Most of the genes in these islands are homologous to ones found in human and plant pathogens, including type IV 38

PAGE 39

group B pilus genes (He et al. 2004). Eleven of the genes in th ese islands are required for full virulence on both plants and anim als (He et al. 2004). A large portion of the PAPI-1 island was similar to open reading frame (O RF) clusters in the phytopathogen X. axonopodis pv. citri and the human pathogen S. enterica serovar Typhi and may have been acquired by PA14 from these bacteria (He et al. 2004). Inte restingly, 80% of the PAPI-1 island was unique, with ORFs that are not related to known proteins or functional domains, possi bly representing a toolbox of pathogenesis gene variants for colonizing multiple hosts (He et al. 2004). Another gene present in P. aeruginosa involved in virulence on both plant and animals is mucD which is orthologous to degP a gene encoding a periplasmic protease and chaperone in E. coli with homologs to known virulence factors involved in stress responses by various sp ecies (Yorgey et al. 2001). A mutant in mucD had a reduced ability to cause disease in Arabidopsis and mice, and significantly lower growth in Arabidopsis (Yorgey et al. 2001). MucD is required for protection against environmental stresses, including temperature and oxidative stress, and also appears to be necessary for production of an extracellu lar virulence factor (Yorgey et al. 2001). In addition to P. aeruginosa several other bacterial pathog ens colonize and cause disease in multiple hosts. Burkholderia cepacia an important opportunistic pathogen in immunocompromised people (Coenye et al. 2001; G ovan et al. 1996), can infect alfalfa (Bernier et al. 2003). In a study to develop a plant model for studying virulence of Burkholderia most strains of B. cepacia that were virulent in alfalfa were also found to cause disease in a lung infection model (Bernier et al. 2003). Although not all Burkholderia strains were capable of virulence in alfalfa, those that were reduced in alfalfa virulence also caused less severe symptoms in the lung model (Bernier et al. 2003). This correlation reinforces the idea that some virulence factors may play a common role in both plant and animal infection. Another 39

PAGE 40

opportunistic pathogen, Enterococcu s faecalis causes disease in Arabidopsis attaching to the leaf surface, entering through stomata and wounds colonizing the intercellular spaces, and causing rotting of cell walls and me mbrane structures (Jha et al 2005). Two of five mammalian virulence factors tested, fsrB a putative quorum sensing gene, and sprE a serine protease, were important for plant pathogenesis, with mutants in these genes bei ng strongly reduced in virulence and colonization on Arabidopsis (Jha et al. 2005). These obser vations suggest an evolutionary crossover of virulence factors between plants and animal pathogens (Jha et al. 2005). In addition, the plant pathogen, E. chrysanthemi adheres to human adenocarcinoma cells, and causes oxidative stress response followed by cell death (Duarte et al. 2000). A virulence factor important to E. chrysanthemi pathogenicity on human cells was a type III secretion system, since type III secretion mutants killed significantly fewer cells than wild type Erwinia (Duarte et al. 2000). In addition, E. chrysanthemi has a surface protein that shares immunological identity with the protein, intimin, which is required for virulence of pathogenic strains of E. coli (Duarte et al. 2000). Since adherence to the surface of an organism is an important step in initiating infection, it is not surprising that some virulence genes in common between plant a nd animal pathogens are involved in attachment. For example, XadA, an outer membrane protein in X. oryzae shares similarity to nonfimbrial adhesins from animal pathogens, including YadA from Yersinia and UspA1 from Moraxella (Ray et al. 2002). A region of si milarity between YadA and XadA was repeated six times throughout the entire Xa dA protein (Ray et al. 2002). Also, homology modeling of XadA suggests it has a -helix conformation, similar to the nonfimbrial adhesin pertactin, which is a virulence factor in the human pathogen Bordetella pertussis (Emsley et al. 1996; Ray et al. 2002). Infection of rice plants by xadA mutants was significantly reduced 40

PAGE 41

com pared to the wild type strai n, demonstrating this gene does pl ay a role in pl ant colonization (Ray et al. 2002). The xadA gene was induced on minimal me dia, indicating it is regulated by growth conditions (Ray et al. 2002). Regulati on by growth conditions is not surprising for virulence genes, because they are need ed only at the time of colonization. Human pathogens also possess genes capable of mediating attachment to plants. In Salmonella a screen for attachment mutant s identified 20 genes necessary for S. enterica adherence to plants, 65% of which had no reporte d function (Barak et al. 2005). Some of these genes were also required for animal virulence (B arak et al. 2005). Specif ic genes identified in this screen include agfB agfD and rpoS, which are involved in production and regulation of curli and cellulose (Barak et al. 2005). In addition, rpoS regulates other virulence and adhesion factors, such as pili (Barak et al. 2005). Taken together, this information suggests that plants can serve as a secondary host and vector for S. enterica between animals (Barak et al. 2005). Several other virulence genes in either plant or animal pathogens share similarity to each other. The AvrA gene in S. enterica which encodes a target of the centrisome 63 type II secretion system, shares sequence similarity with AvrRxv, an avirulence determinant from the plant pathogen X. campestris pv. vesicatoria (Hardt and Galan 1997). AvrA is an effector protein translocated into host cells and it is suggested that AvrRxv and AvrA may be part of a family of effector proteins related to host/pathogen crosstal k (Hardt and Galan 1997). The presence of such common virulence genes be tween plant and human pa thogens could explain how human pathogens are capable of colonizing plant hosts. Genomes of Endophytic Bacteria Genom es of endophytic bacteria are being sequenced, and analys is of this information is yielding insight into how these bacteria can inhabit the plant interior but not cause disease symptoms or cell death. The annotation of thes e genomes also provides targets for mutagenesis 41

PAGE 42

to identify genes that m ay be involved in endophytic colonization. The first fully sequenced genome of a bacterial endophyte was that of the diazotrophic strain, Azoarcus sp. BH72. In BH72, there is a lack of toxin and virulence ge nes common in other plan t pathogenic bacteria, including type III and IV secretions systems, pr esumably to avoid damaging the host tissue in the presence of high bacterial numbers within the pl ant (Krause et al. 2006). Also, these secretion systems can elicit plant defenses (Iniguez et al 2005). Their absence may allow BH72 to avoid plant defenses. The BH72 genome has few phage or transposable elements, suggesting a low rate of lateral gene transfer, which is in contrast to the soil-associated Azoarcus sp. strain EbN1. This suggests that BH72 is adapted to a low stress, stable environm ent such as that found in plant tissue (Krause et al. 2006). BH72s lipopolysacc haride, exopolysaccharide, and capsule genes are more closely related to those from plant symb ionts and pathogens than to its close relative, Azoarcus sp. EbN1 (Krause et al. 2006). In a ddition, the BH72 genome ha s a low degree of synteny with that of EbN1, while it possesses gene clusters that show sign ificant synteny to plant pathogens, such as P. aeruginosa (Krause et al. 2006). Also in accordance with BH72s role as an endophyte, its genome appears to be adapted to low-nitrogen and low-iron environments, with high-affinity ammonia assimilation and transporter proteins, a full set of n itrogen fixation genes, and twice as many iron transporters as in EbN1, demonstrating that it can fix nitrogen and uptake both nitrogen and iron from the environment (Krause et al. 2006). The genomes of other diazotrophic endophytes have either recently been completed or are nearly finished, including those of K. pneumoniae 342, Gluconacetobacter diazotrophicus PAl 5, and Herbaspirillum seropedicae Z67. Of these, the Kp342 genome has been published (Fouts et al. 2008). As expected, the full complement of nif genes was found in this genome. As in BH72, this genome has few mobile elements and many transport genes. 42

PAGE 43

Given the plant origin of Kp342, it was surp rising to find m any animal virulence determinants in Kp342 that are presen t in clinical strains. Due to the presence of these genes, the animal virulence of Kp342 was tested in a m ouse model and compared to a commonly studied clinical isolate. Kp342 did not col onize spleens or livers at a level si milar to the clinical isolate. Kp342 did colonize lungs and kidneys where colonizat ion is likely to be important in disease development. However, the level of Kp342 colonization of these two organs was 27 and 49 times lower in the kidneys and lungs, respectively, than with the clinical isolate. So although Kp342s pathogenicity in mice is less than that ob served in a clinical is olate, the level of lung and kidney colonization is sufficien tly high to suggest that Kp342 not be used in any agricultural application. A preliminary report on the genom e of the sugarcane endophyte, H. seropedicae indicates that this endophyte also possesse s nitrogen fixation genes, as we ll as 27 genes involved in iron transport (Pedrosa and Consortium 2005). In contrast to BH72, H. seropediae does have type III secretion genes, though both endophyt es lack a type IV secreti on system (Krause et al. 2006; Pedrosa and Consortium 2005). Importan t to plant microbe interactions, H. seropedicae possesses genes similar to hrp genes and ysc from Yersinia pestis (Pedrosa and Consortium 2005). In addition, H. seropedicae has genes similar to the ty pe III secretion genes found in Y. pestis (Pedrosa and Consortium 2005). The presen ce of genes similar to those found in an animal pathogen indicates that these pathogens could also possess the ability to colonize plant tissue. The G. diazotrophicus PAl 5 genome was recently deposited in GenBank. This bacterium was the first one definitively shown to provide fi xed nitrogen to a grass sp ecies (Sevilla et al. 2001). As of this writing, a manuscript describi ng the genome has not been published. This 43

PAGE 44

44 genome has two plasmids, 16.6 and 38.8 kilobases (kb) in size, and a 3.94 megabase (Mb) chromosome. Fully sequenced close relatives exist for all of the four endophytic genomes sequenced to date. These genomes should allow valuable comparative genomic analyses that should provide significant testable ideas on how plants interact with these bacteria. Future Directions Specific control m echanisms to reduce the num ber of human pathogens on or within raw produce have not been reviewed here. The objec tive here is to describe our understanding to date of the biology of endophytic colonization with a specia l emphasis on human bacterial pathogens. However, an understanding of endophytic colonization is useful not just in future control methods for human pathogens in edible plants but also to improve the growth and nutrition of plants through the use of diazotrophic endophytes. An understanding of the biology of endophytic colonization will result in a sa fer food supply and in more efficient food production. Control mechanisms are most valuable when they are coupled with a knowledge of the system. Fortunately, our knowledge of th e biology of endophytic ba cteria is increasing rapidly and should grow more give n the availability of complete genomes. Perhaps the largest impediment to future progress in this area is our lack of understa nding of plant defense mechanisms in food crops. Although the knowledge gained in recent years on plant defenses in Arabidopsis is very impressive, this knowledge needs to be translated to t hose plants that are consumed daily. Where that knowledge cannot be tr anslated, we need to do the hard work to understand defense mechanisms in lettuce tomato, spinach, and similar crops.

PAGE 45

CHAP TER 3 CONFIRMATION OF THE SEQUENCE OF CANDIDAT US LIBERIBACTER ASIATICUS AND ASSESSMENT OF MICROBIAL DIVERS ITY IN HUANGLONGBING-INFECTED CITRUS PHLOEM USING A METAGENOMIC APPROACH Reproduced with permission from Tyler, H. L ., Roesch, L. F. W., Go wda, S., Dawson, W. O., and Triplett, E. W. Conf irmation of the sequence of Candidatus Liberibacter asiaticus and assessment of microbial diversity in Huanglongbing-infected citr us phloem using a metagenomic approach. Mol. Plant-Microbe In teract. (In press). In this study, H. L. Tyler performed all metagenomic analyses, most confirmatory polym erase chain reactions (PCRs), and wrote the manuscript. L. F. W. Roesch extracted the citrus deoxyribonucleic acid (DNA) and performed the automated method of ribosomal intergenic spacer analysis (ARISA), S. Gowda performed some confirmatory PCRs, and E. W. Trip lett and W. O. Dawson provided input on the manuscript. Introduction Huanglongbing (HLB), also known as citrus gr eening, is a destruc tive and devastating disease of citrus causing great lo sses in citrus industries throu ghout the w orld (Bov 2006). This disease is characterized by yellow shoots, blotch y mottle on the leaves, and fruit that are small and lop-sided, with inverted coloring. HLB is transmitted between citrus trees by the psyllids, Diaphorina citri (in Asia and America) and Trioza erytreae (in Africa) (McClean and Oberholzer 1965; Capoor et al. 196 7; Bov 2006). Progress in the study of this disease, including detection, treatment, and control, has been hindered due, in part, to the lack of effective culturing. Through microscopic examination and sequence analysis of polymerase chain reaction (PCR)-amplified 16S ribosomal DNA of in fected tissue, it has be en proposed that the causal agent of citrus greening is a phloem sieve tube-limited gram-negative bacterium that is a member of the alphaproteobacter ia (Garnier et al. 1984; Jagoue ix et al. 1994). More recent 45

PAGE 46

studies have found HLB to be associated with three m embers of the genus Candidatus Liberibacter: Ca. Liberibacter asiaticus, Ca. L. americanus, and Ca. L. africanus (Jagoueix et al. 1997; Teixeira et al. 2005). In order to get around the obstacle of Ca. Liberibacter spp. being uncultured, attempts have been made to sequence the genome direc tly from infected tissue, in the hopes that knowledge of the genome will give better insight into how to culture this organism as well as how to combat it. Although Ca. L. asiaticus is at a low titer in citrus trees, it is present in much higher numbers when infected in periwinkle a nd Dodder (Ghosh et al. 19 78; Garnier and Bov 1983). Therefore, DNA from infected plants in th is system have been sequenced to obtain 8.56 and 14.7 kilobase (kb) segments of the Ca. Liberibacter genome using a genomic walking method (Doddapaneni et al. 2008; Lin et al. 2008). Recently, 34 contigs of Ca. L. asiaticus str. psy62 genomic sequence, ranging in size from 1.033 to 186.24 kb for a total of 1.2 megabases (Mb) of genomic sequence, has been de posited into GenBank (Accession number ABQW00000000). These genomic sequences were obtained from bacteria extracted from the psyllid vector (Duan et al. 2009). At the time of this writing, only 34 contigs were available for analysis, though the genome has been closed and is now published (Duan et al. 2009). Therefore, in this article, we se t out to confirm that the proposed Ca. L. asiaticus contigs are, indeed, associated with HLB symptomatic plants and not asymptomatic plants. Most HLB-infected trees in Florida are also infected with Citrus tristeza virus (CTV). We examined a CTV-free HLB isolate that was used to charact erize the host range and symptoms of Florida HLB (Folimonova et al. in press). In addi tion, a non-PCR based metagenomic approach was used to investigate the microbial diversity within HLB-infected citrus tissue and whether there are any other pathogens present. 46

PAGE 47

Results Verification of Ca. L. asiaticus str. psy62 Contigs In order to confirm the Ca. L. asiaticus contigs deposit ed in GenBank belong to the causative agent of HLB, it was necessary to verify that these contigs are present only in infected citrus tissue and not in healthy tissue. To that end, PCR reactions to assay for the presence of each contig were performed with geno mic DNA from healthy and infected Citrus sinensis tissue. Before PCR, the 34 putative Ca L. asiaticus contigs on GenBank were assembled and reduced to 32, with NZ_ABQW01000005 assemb ling with NZ_ABQW01000033 and NZ_ABQW01000022 assembling with NZ_ABQW01000031. Of these 32 contigs, 30 were found to have bands of the expected size range in infected tissue, but not healthy tissue (data not shown), confirming that these sequences are, indeed, associ ated with HLB symptoms and, therefore, from the likely causative agent of th e disease. Of the 2 contigs that were not confirmed, there were either positive bands in bo th infected and healthy reactions or no positive bands in either reaction. Thes e results could be explained by non-specific primer binding to regions in the C. sinensis genome. Confirmation of Ca. L. asiaticus Contig Sequences Because Ca. L. asiaticus contigs were detected only in infected tissue by PCR, a metagenomics approach was used to confirm that the entire seque nce of each contig was present in HLB-infected phloem, further verifying that th e sequences are from the causative agent of the disease. To do this, three ne xt-generation sequencing platform s, 454, SOLiD, and Solexa, were used to generate a total of 13.6 gigabases (Gb) of metagenomic sequencing reads from HLB-infected citrus phloem (Table 3-1). With th e large amount of sequence data obtained from HLB-infected phloem, high coverage of the Ca. L. asiaticus genome and, therefore, the Ca. L. asiaticus contigs, was expected. To determine the level of Ca L. asiaticus coverage, all 47

PAGE 48

m etagenomic sequencing data was run in a reference assembly against the Ca L. asiaticus contigs. Of all phloem sequences, 0.23% matched the Ca L. asiaticus contig s, with coverage ranging from 20 to 60-fold per contig (Table 32). In addition, sequences from HLB-infected phloem spanned the entire length of each cont ig, with 99.69 to 100 % of each contig matching the consensus sequence generated from th e reference assembly (Table 3-3). The dataset from each high-throughput seque ncing method was also run against the Ca. L. asiaticus contigs separately in order to determ ine if there were any biases due to sequencing method. One of the major differences between the results of the three sequencing methods appears to be due to number of reads, with the 58.8 Mb of 454 sequences only yielding 0 to 0.341-fold coverage of the Ca L. asiaticus contigs (Table 3-2). In contrast, reference assemblies with Solexa and SOLiD data, which provide significantly more sequences (1.78 and 11.8 Gb), yielded 10.78to 28.44-fold and 5.62to 31.29-fold coverage, respectively, of the Ca L. asiaticus contigs (Table 3-2). When looking at the consensus seque nces generated by these two reference assemblies, Solexa sequences covered 98.9 to 100 % of each Ca. L. asiaticus contig and SOLiD data covered 82.39 to 100 % (Table 3-3). Of the th ree sequencing methods, Solexa yielded slightly bette r fold coverage of the Ca L. asiaticus contigs overall. For example, 23 of the 32 Liberibacter contigs had higher fold coverage when assembled against Solexa data than with 454 or SOLiD (Table 3-2). Specifically, c ontig NZ_ABQW01000028 had 5.62-fold coverage with SOLiD sequences, but wh en assembled with Solexa sequences, fold coverage of this contig almost tripled, going up to 15.03 (Table 3-2). There were also cases where SOLiD data yielded higher fold coverage of a contig but the increase was not as much. In one such case, contig NZ_ABQW01000026 had 18.077-fold coverage with SOLiD data and 10.862 with Solexa (Table 3-2). Sequences from 454 did not approach th e level of coverage 48

PAGE 49

seen using S OLiD or Solexa for any of the c ontigs, mostly due to the reduced number of sequences from 454 compared to the other two. For exampl e, 454 yielded 0.25 million reads while SOLiD and Solexa yielded 337 and 49 m illion reads respectively. However, overall, sequences from each of the 3 methods compleme nted gaps missed by the other two because, when analyzed separately, only 4 of the 32 Ca L. asiaticus contigs had consensus sequences that covered 100% of their length but, when all data were assembled together against the Ca L. asiaticus contigs together, 11 of those contigs had a 100% match (Table 3-3). With such high fold coverage and consensus sequences covering the entirety of each contig, it was concluded that these contigs are present in the phloem. Comparison to Alphaproteobacterial Relatives In order to discount the possibility that the m atches seen when comparing citrus phloem sequences to Ca. L. asiaticus contigs are due to the presence of close alphaproteobacterial relatives, all infected phloem sequence da ta was compared to 5 fully sequenced alphaproteobacteria deemed to be closely related to Ca L. asiaticus after a BLAST of its contigs against an alphaproteobact erial database. The results of the reference assemblies with each of the 5 alphaproteobacteria showed that over 100-fold fewer sequencing reads match these alphaproteobacteria compared with Ca L. asiaticus contigs (Table 3-4). For example, the closest match, Rhizobium leguminosarum bv. viciae 3841, had 8,012 reads match its genome while 900,124 reads matched the Ca L. asiaticus contigs. In addition, sequencing reads from infected phloem yielded an average 26.34-fold coverage of the Ca. L. asiaticus contigs but none of the alphaproteobacteria ha d greater than 0.175-fold coverage (Table 3-4). Therefore, the bacterial DNA in phloem is specific to Ca. L. asiaticus and even close relatives do not match this DNA. This confirmation of Ca. L. asiaticus in symptomatic plant tissue further supports its proposed role as the causative agent of HLB. 49

PAGE 50

Assessment of Bacterial Diversity Automated ribosomal intergenic spacer anal ysis (ARISA) was performed for the initial assessment of bacterial diversity within HLB-infected phloem. Amplification with ARISA primers yielded 2 dominant peaks in healthy (740 and 1,330 basepairs (bp)) and 3 dominant peaks in infected (740, 1,330, and 1,600 bp) phloem (F igure 3-1). To determine the identity of these peaks, the ARISA PCR was run on a gel, a nd the bands extracted, cloned, and sequenced in a 96-well plate. The 740 and 1,330 bp bands were from C. sinensis chloroplast and mitochondria, respectively, while the 1,600 bp band unique to infected tissue was from Ca. Liberibacter 16S ribosomal ribonucleic acid (rRNA). Of all the sequences obtained from the 1,600 bp band, there was a 100-fold excess of Ca. Liberibacter sequences, indicating it as the most abundant bacterium present within the phloem. Once the Ca. L. asiaticus contigs were confirmed in infected citrus tissue, metagenomic sequencing data from the infected phloem was analyzed to further assess whether Ca. L. asiaticus was the only bacterium present in this sample. Bacterial divers ity in the phloem of C. sinensis plants infected with Ca. L. asiaticus was analyzed by searching total DNA sequences obtained from infected phloem for 16S rRNA gene sequences. Because plant chloroplasts are related to Cyanobacteria, plant ch loroplast sequences were filtered out of the phloem sequences. To do this, the entire dataset was assembled against the C. sinensis chloroplast genome and nonassembled reads were taken for further analysis. Approximately 5.7 million of the total sequencing reads matched the chloroplast genome and were removed. The remaining 381,056,442 sequences were run in a reference assemb ly against the Ribosomal Database Project (RDP) database. Sequences from the 454, Solexa, and SOLiD data sets were run both separately and together. Each reference assembly yielded an output with the number of sequencing reads matching each reference in the database, averag e coverage of each 16S rRNA match, and length 50

PAGE 51

of the consensus sequen ce matching each 16S rRNA gene. In each assembly, the 16S rRNA sequence from the database with the greatest number of matches to infected citrus phloem sequence was Ca. Liberibacter. In addition, the Ca. Liberibacter 16S rRNA gene had highest fold coverage and longest consensus sequence over the entirety of the gene compared with any other bacteria in the database (Table 3-5). In total, the phloem metagenome from all three sequencing platforms combined yielded a fold coverage of 23.17 of the Ca. Liberibacter 16S rRNA gene, with a consensus sequence of 1,212 bp in length (Table 3-6). The length of the Ca. Liberibacter 16S rRNA sequence in the database was 1,448 bp; therefore, 83.7% of the 16S rRNA gene was covered (Table 3-6). In contra st, the next highest match, the 16S rRNA gene from Nitrospina gracilis had a fold coverage of 1.45 a nd a consensus sequence that only covered 7.15% of the gene. All other matches to members of the RDP database covered less than 14% of the total 16S rRNA gene. Even co verage over the entire length of the 16S rRNA gene is expected for any bacterium present in the phloem; therefore, Ca. Liberibacter is likely the only bacterium present. In addition, the shor t regions within the 16S gene of the other RDP matches had 90 to 100% identity with sequences in Ca. L. asiaticus and citrus chloroplast and mitochondria when compared against these refere nces on GenBank. Therefore, it was concluded these matches were false positives. Given that it would be impossible eliminate all false-positive matches due to alignment and sequencing errors, a threshold fold coverage detection limit was determined. In order to calculate this detecti on limit, the mean and st andard deviation were calculated from the fold coverage of every RD P 16S rRNA match. Any RDP match with a fold coverage greater than the mean plus 3 standard deviations (1.89) was de emed to be significant and above the threshold fold coverage that w ould be seen with any 16S rRNA gene assembling to remaining citrus and Ca. Liberibacter sequences due to misa lignments or sequencing errors. 51

PAGE 52

Because Ca. Liberibacter was the only RDP database m ember to have coverage above 1.89, it was deemed to be the only bacterium likely present in citrus phloem. Similar results were seen when SOLiD and Sole xa data were run against the RDP database separately, yielding a fold coverage of the Ca. Liberibacter 16S rRNA of 17.37 and 5.23, respectively (Table 3-6). When 454 sequences we re run separately, fold coverage was 0.33, with a consensus sequence matching only 26.17% of the Ca. Liberibacter 16S rRNA gene (Table 3-6). This number is not entirely surprising because very few 454 reads matched any of the 16S rRNA sequences. Given that the average read length of the 454 sequences was 248 bp and only two of those reads matched the 1,448 bp Ca. Liberibacter 16S rRNA gene, a match to only 26% of the 16S rRNA is unsurprising. There was only one other species in the entire database that had a matching member when run against th e 454 data and it had 100-fold lower coverage than Ca. L. asiaticus (Table 3-5). As a result, the 454 data had li ttle effect on the results of citrus phloem community analysis using the RDP da tabase. It is likely this is due to the low amount of sequence generated by 454 compared to the other two methods. It is clear, however, that both SOLiD and Solexa contributed to the community analysis, since the overall fold coverage determined by the reference assembly using all three sets of sequencing data was higher than the fold coverage found when analyz ing these datasets separately (Table 3-6). Other Reported HLB Associat ed Bacteria Not Present In addition to Ca. Liberibacter spp., other bacteria have been reported to be associated with HLB infection in citrus, including Propionibacterium acnes Ca. Phytoplasm a asteri, and pigeon pea witches'-broom phytoplasma (Davis et al. 2008; Chen et al. 2 009; Teixeira et al. 2008). These reports are in contradiction to the results of the 16S rRNA analysis of the citrus metagenome, which yielded a fold coverage of 0.023 and 0.026 for Ca. P. asteri and P. acnes respectively, which corresponds to one 35 bp sequ encing read matching the 16S rRNA gene. To 52

PAGE 53

further confirm the presence or absence of th ese bacteria in the current study, reference assemblies of the metagenomic sequences were run against the genomes of these bacteria available on GenBank. In both cases, the fold coverage of these genomes was more than 100-fold lower than the coverage of Ca. L. asiaticus (Table 3-7). These results indicate that Ca. P. asteri and P. acnes are not present in the sample; othe rwise, much higher coverage of these genomes would be expected if these bact eria were present in citrus phloem. In comparison, even the non-HLB-associated actinobacterial relative of P. acnes, Micrococcus luteus had 0.038-fold coverage when assembled with the citrus metagenomic sequences, double the fold coverage of P. acnes (Table 3-7). Seeing a non-HLB a ssociated bacterium with a higher fold coverage than a bacterium reportedly associat ed with HLB indicates that the genome of any bacteria will match such a large amount of metagenomic sequences due to alignment with conserved genes. Upon closer inspection, 84 to 99% of the seque ncing reads that matched the genomes of these three bacteria aligned to regions in the 16S and 23S rRNA genes. If P. acnes was truly present in the sample, a more even distribution of reads aligning over the genome would be expected. It is likely that a large number of these sequences matching the 16S and 23S rRNA genes are derived from the rRNA genes of citrus chloroplast a nd mitochondria, which would be present in citrus tissu e in higher numbers and, therefore, inflate the fold coverage of any bacterial genome compared to the metagenome. To test this idea, rRNA regions in the reference assemblies with many metagenomic seque nce matches were found to have greater than 90% and, in some cases, 100%, iden tity with sequences from the C. sinensis chloroplast genome. Furthermore, when the reference assemblies were repeated with reads that had the C. sinensis chloroplast genome filtered out, the number of matching bases and fold coverage of Ca. P. asteri, P. acnes, and M. luteus, as well as the genomes from 5 reference alphaproteobacteria, 53

PAGE 54

were significantly reduced (Table 3-8). In contrast, the fold coverage of Ca. L. asiaticus contigs remained above 26 even with chloroplas t sequences rem oved. These analyses validate the observation that the majority of sequences matching these genomes were chloroplast in nature and these other bacterial species were not present in the phloe m sample because any bacterial genome is likely to ha ve a low level of fold coverage when compared against such a large number of metagenomic sequences due to the conserved regi ons in rRNA genes. In order to test the hypothesis that Ca. P. asteri and P. acnes were not present in citrus phloem, a detection limit was calculated to dete rmine the threshold level of sequencing matches required to be 99% confident that a bacterial genome is truly present in the sample. The detection limit of these metagenomic sequences wa s set at three standard deviations above the mean fold coverage of seven bacterial genomes referenced against th e entire metagenomic dataset. At this level, any genome with a fold coverage of 0.29 or greater was considered to be above the background level of false positive sequen ce matches and deemed to be present in the citrus phloem. This corresponds to 0.0021% of a ll bases and reads in the metagenomic dataset, or approximately 8,144 of the 386,745,534 total read s and 285,030 bases out of the total 13.6 Gb of metagenomic data. Ca. L. asiaticus was the only genome analyzed with a fold coverage greater than this threshold. Neither Ca. P. asteri nor P. acnes had enough sequencing reads match their genomes to achieve a fold covera ge approaching this limit. Therefore, Ca. L. asiaticus was the only one present within these phloem samples. Approximately 0.23% of the 13.65 billion base s sequenced in this work are from Ca. L. asiaticus. Assuming a diploid ge nome size of citrus at 900 Mb, these data represent a 15-fold coverage of the diploid citrus genome or can be considered the full chromosomal content of 15 citrus phloem cells. Assuming a Ca. Liberibacter genome size of 1.21 Mb, the 26-fold 54

PAGE 55

coverage of the Ca. Liberibacter genom e obtained here can be considered to be the full chromosomal content of 26 Ca L. asiaticus cells. This s uggests that the phloem sample sequenced here possessed 1.7 Ca L. asiaticus cells per phloem cell. Based on our detection limit of 0.29-fold coverage, the metagenomic a pproach described here could detect one Ca L. asiaticus cell for every 52 phloem cells. DNA Viruses and Viroids Not Present Because Kochs postu lates have not been completed for HLB, it is possible that other agents, such as a virus or viroid, may be involve d in the development of disease symptoms. To determine whether any such agent is present in HLB-infected phloem, all metagenomic sequence data was compared against a virus and viroid da tabase. When a reference assembly of greening infected phloem sequences against the viroid database was performed, no sequencing reads assembled, indicating that there are no known viroid s present within the phl oem. This result is not unexpected as the phloem sample sequenced was of DNA extracted directly from the plant tissue and no reverse-transcriptase steps to obtain RNA sequences were performed. The reference assembly against the virus databa se did yield matches, but none showed good coverage. The highest fold coverage to any member of the virus database was 2.04-fold but the consensus sequence only spanned less than 1% of the reference that it matched. Considering the short nature of the consensus sequence and the fact that the match was to Hepatitis, a human virus, it is unlikely that this virus is pres ent in the phloem sample. While the amount of sequencing in this study may not be sufficient to detect low abunda nce viral partic les, the amount of sequence obtained for this work should be su fficient to detect viru s particles that are sufficiently abundant to cause disease. Therefor e, DNA viruses are not likely to play a role in HLB in the phloem. 55

PAGE 56

High Coverage of Citrus The DNA sequenced for this metagenomic analysis was extracted from total citrus phloem; therefore, it was expected that the datase t would include a high coverage of the C. sinensis genome in addition to the genomes of any microorganisms present. In order to confirm this, all phloem sequences were run through a reference asse mbly against a citrus expressed sequence tag (EST) database. Of the 34,505 ESTs in the database, 79.85% had hits from the phloem sequence, with an average fold c overage of 13.19 (results not shown). Discussion Most of the research res ults that associate Ca. Liberibacter spp. to HLB diseases in citrus involve microscopic examinations, cloning and sequencing of PC R amplified DNA from infected tissue, and other PCR-based techniques. In this study, a PCR independent metagenomic approach was taken to co nfirm the presence of Ca L. asiaticus contigs, determine if other organisms may be present, and assess the microbia l diversity in HLB-infected phloem. Prior to confirming the role of Ca L. asiaticus in infected citrus phl oem, it was first necessary to verify that the contigs for this strains genome in GenBa nk were only present in infected tissue. To that end, PCR performed with primers designed to each of the Ca L. asiaticus contigs confirmed that 30 of the 32 contigs were present in HLB-infected but not healthy citrus tissue. Once it was confirmed that the Ca L. asiaticus contigs were present only in HLB-infected tissue, the presence of those cont igs in infected citrus phloem was examined by metagenomics. In this analysis, all contigs were present in inf ected phloem with fold coverage ranging from 20.642 to 60.067. Such high fold covera ge of these contigs from HLB-infected phloem further confirms the validity of th ese contigs belonging to the genome of Ca L. asiaticus and a causative agent of HLB. It is unlikely that th e high coverage of Ca L. asiaticus is due to the presence of another alphaproteobacterium in the tissue since a selection of 56

PAGE 57

genom es from 5 different alphapr oteobacterial relatives to Ca. Liberibacter spp. did not have greater than 0.2-fold coverage. Because Kochs postulates have not been completed for Ca. Liberibacter spp. in HLB disease, it is possible that anothe r pathogen or pathogens may play a role in disease development. For this reason, the composition of the microbial co mmunity in infected citrus phloem was also investigated by assembling the citrus phloe m metagenome sequences against the 16S rRNA database from RDP. This analysis was run with data from three different high-throughput sequencing methods separately as well as with all the data combined and, in each case, the Ca L. asiaticus was found to be the only bacterium w ith significant coverage of its 16S rRNA gene. For all other bacteria in the RDP database, metagenomic sequences matched no greater than 13.5% of the 16S gene or a fold coverage of 1.45. If any other bacteria were present within the citrus phloem, a more even coverage over the entire 16S rRNA gene would be expected because the total DNA extract from infected phloem was sequenced without using any PCR amplification that would create biases. Therefore, it can be said with confidence that the matches of the contigs and 16S rRNA gene with the metagenom ic dataset from HLB-infected phloem are indeed due to the presence of Ca L. asiaticus in inf ected phloem and not another bacterium. In addition, no known DNA virus was identified in the metagenomic analysis, although, in the field, most HLB-infected trees are also infected with CTV. However, the samples examined were chosen because they are free of CTV. We have examined the interaction of CTV and HLB and found little or no synergistic intera ctions (Folimonova et al. in pres s). Therefore, viruses were discounted as playing a role in HLB. The results from this study differ slightly from others assessing the microbial diversity of HLB-infected citrus plants. In the current study, the only bacterium found to be present in 57

PAGE 58

infected citrus was Ca L. asiaticus whereas Sagaram and associates (2009) found a large diversity of microorganisms, co mprising 47 orders of bacteria. However, two very different samples were analyzed in these studies. Sagara m and associates (2009) were looking at entire leaf midribs whereas the current study looke d at phloem scraped from the inside of C. sinensis bark that was peeled from growing trees. Th e entire midrib includes the apoplast, which is known to harbor bacterial endophytes. It is likely that a larger a rray of bacteria are present in leaf tissue than are actually inside the phloem cells of the plant vascular system. The causative agent of HLB is thought to be restricted to phloem sieve tubes; therefore, this tissue is ideal for examining HLB-associated DNA sequences. Because Ca. Liberibacter spp. is intracellular within phloem cells, being spread systemically from the site of infection to other tissues, there will be fewer bacteria present in this environment as any present would have to bypass more plant defense mechanisms to enter the cell. Although midribs do contain phloem tissue, other leaf tissue is also present in such preparations and would contain addition al bacteria that would inflate the estimate of microbial diversity directly in contact with Ca L. asiaticus. In addition, Chen and associates (2009) reported a strain of Ca. P. asteri associated with HLB-infected citrus that was al so not identified in the curren t study. Because that study also examined leaf midribs as opposed to phloem cells from the bark, differences in the bacteria identified are not surprising. Although this st udy also reported observing phytoplasma in phloem sieve tube cells from midribs using electron microscopy, the majority of the microscopic examinations were performed in periwinkle, which showed an enrichment of the bacterium compared to citrus. Furthermore, the identi fication of phytoplasmas in sieve tubes was done on the basis of cell wall thickness, without any molecular means of confirming their identity such as a phytoplasma-specific antibody. Thus, it is not certain that phytoplasmas are present in the 58

PAGE 59

sieve tube cells of citrus. However, these results do not preclude the poss ibility that phytoplasmas can cause citrus greening-like symptoms in citrus in some areas. In this work, citrus is shown to have these symptoms with Ca. Liberibacter spp. exclusively. In another study, Davis and associates (2008) looked specifically at phloem tissue in an attempt to culture Ca L. asiaticus and obtained a co-culture of Ca. L. asiaticus with actinobacteria related to P. acnes based on 16S rRNA analysis, with a 1,006 bp 16S rRNA sequence from the co-culture having 100% identity to P. acnes. These observations are in contrast to the results of the present study b ecause there was no significant coverage of the P. acnes genome in the metagenomic sequences. Although Ca L. asiaticus was reported to have been co-cultured with P. acnes, it took 31 culturing attempts to obtain a second co-culture with these bacteria and 12 additional attempts to obtain a third co-culture from HLB-infected plants (Davis et al. 2008). The number of repeated attempts necessary to obtain co-cultures with P. acnes from HLB-infected tissue indicates this ba cterium may not be a common inhabitant of citrus. If P. acnes was commonly found in citrus pl ants and capable of supporting Ca L. asiaticus in co-culture a higher abundance of P. acnes DNA would be expected in these plants. In addition, Davis and associates (2008) used P. acnes specific primers to detect the bacterium in DNA extracts from citrus but onl y 3 healthy plants were re ported to be positive for P. acnes Also, whereas Davis and associates (2008) state they identified P. acnes in their co-cultures, they did not specify whether PCR was performed to identify P. acnes in DNA directly isolated from HLB-infected plants. Furthermore, P. acnes is a commensal of huma n skin (Brggemann et al. 2004), and its identification in plant tissues may be the result of cont amination while handling the samples. In this work, P. acnes was not found in the metagenomic sequences and may not be associated with Ca L. asiaticus in planta. 59

PAGE 60

Given the large am ount of sequencing da ta generated by the three high-throughput sequencing methods, there is a hi gh likelihood that the genome of any bacteria present in the phloem sample would have been detected. To test this, a detection limit of 3 standard deviations above the mean observed fold coverage of a b acterial genome was used to distinguish between the false and genuine sequence matches to the Ca L. asiaticus genome. Only the Ca L. asiaticus genome sequences met this standard. In addition, the level of 16S rRNA coverage by other bacteria was also much lower than expected for a bacterium to be present in these samples. In both cases, when looking at entire genom es or specifically at 16S rRNA genes, Ca. Liberibacter was the only bacterium identified above these detection limits. Furthermore, fold coverage of the Ca. Liberibacter 16S rRNA and genome were 12 and 90 times higher than these limits, respectively. Because Ca L. asiaticus occurs in very low concentrations in infected citrus (Tatinen i et al. 2008) and was identified at su ch high levels above the detection limit, we are confident that any other low-abunda nce bacteria present in the tissue would have been detected by this method. This article presents a new approach to evalua ting the microbial diversity present in citrus tissue by a PCR-independent metagenomic method. This method has the advantage of avoiding biases based on PCR amplification and cloni ng (Polz and Cavanaugh 1998). In addition, by using high-throughput sequencing me thods, a greater sequencing depth of the entire citrus phloem sample was obtained at a lower per-bas e cost than traditional Sanger sequencing. Similar metagenomic studies have also been applied to investigate the caus ative agents of other diseases. Using 454 pyrosequencing, colony collapse disorder (CCD) in honey bees ( Apic mellifera ) was found to be associated with Israeli acu te paralysis virus (IAPV), an unclassified dicistrovirus (Cox-Foster et al. 2007). In that study, results from the metagenomic analysis 60

PAGE 61

contrasted w ith previously proposed causative agents of CCD. Although IAPV sequences were found only in CCD cases, Nosema ceranae and an iflavirus sp. proposed to be associated with the disease were identified in both CDD and non -CCD cases (Cox-Foster et al. 2007). As a result, this demonstrates how the metagenomics a pproach can be used to pinpoint disease agents free from culturing, PCR, and cloning biases. In summary, this study used a cultureand PCR-independent meta genomic approach to verify the validity in citrus of the reported Ca L. asiaticus contigs th at were obtained from infected psyllids, determine that it is the major and perhaps the only bacterium present in the phloem of infected citrus phloem, and thereby substantiate its proposed role as the likely causative agent of HLB. W ith the decreasing costs in sequencing due to high-throughput methods and the level of sequence coverage of samples analyzed, metagenomics will become a most effective method for studying uncultured plan t pathogens in the future. This will allow researchers to examine a broader view of the co mplexity of organisms present in the diseased state. Materials and Methods Sample Preparation and DNA Extraction Eleven citru s plants, including C. volkameriana C. aurantifolia cv. Swingle, C. limon cv. Eureka, Hirado Buntan Pink, and C. sinensis cv. Valencia, symptomatic for citrus greening, were sampled. Branches were cut from plants and surface sterilized by wiping with tissue paper and 70% ethanol. The bark, which was slippi ng in growing plants, was removed and the phloem cells were scraped off by using a sterile, disposable scalpel. DNA was extracted directly from the phloem cells. DNA was isolated from the collected phloem tissue samples (approximately 150 mg) using the FastDNA Kit (Qbiogene, Inc., Calif.). After the DNA extraction, samples were purified with 61

PAGE 62

the DNeasy Tissue kit (Qiagen, Valencia, CA.) follo wing the m anufacturers instructions. This procedure was repeated to obtained enough DNA for 454, SOLiD, and Solexa sequencing. PCR to Confirm Validity of Ca L. asiaticus Contigs in GenBank DNA extracted from healthy and infected C. sinensis cv. Valencia (as described above) was used as template in PCR reactions with primer pairs for each Ca. L. asiaticus contig in GenBank. Primers were designed to amplify a 0.5, 1, or 2 kilobase (kb) fragment from each contig using Primer3 (v. 0.4.0) ( http://frodo.wi.mit.edu/ ) (Table 3-9). PCR reactions contained 4 L of 5X HF Phusion Buffer (New England Biolabs), 200 M of each dNTP, 2.5 M of each prim er, 0.4 U of Phusion polymerase, and approximately 50 ng DNA template in a final volume of 20 l. PCR reactions were held at 98 C 30 seconds (s); followed by 30 cycles of 98 C for 10 s, 60 C for 30 s, and 72 C for 1 minutes (min); with a final extension at 72 C for 10 min. PCR products from healthy and HLB-infected sa mples were resolved on a 1% agarose gel and examined for bands in the expected size range. ARISA Analysis, Cloning and Sequencing for In itia l Assessment of Bacterial Diversity and to Confirm Presence of Ca L. asiaticus Bacterial community composition was assess ed by ARISA (Fisher and Triplett 1999). ARISA is a molecular technique for constructing bacterial comm unity fingerprints based on the length heterogeneity of the intergenic transc ribed spacer region of bacterial rRNA operons (Bosshard et al. 2000). Treating the elements of ARISA profiles as opera tional taxonomic units allows for bacterial community comparisons. In this study, ARISA profiles were assumed to be indicative of bacterial community composition, and differences in ARISA profiles were assumed to reflect variation in the composition of the respective bacterial communities. PCR reaction mixtures contained 5 l of 10X PCR buffer (Promega), 200 M of each dNTP, 100 M of each primer, 2.5 U of Taq polymerase, and approximately 100 ng of DNA 62

PAGE 63

tem plate (estimated by using NanoDrop ND-1000 spectrophotometer) in a final volume of 50 l. The primers used were 787F (5-ATTAGATA CCCNGGTAG-3) (Roesch et al. 2007) and L-D-Bact-132-a-A-18 (5 -CCGGGTTTCCCCATTCGG-3 ) (Ranjard et al. 2001). Reaction mixtures were held at 94C for 3 minutes; followed by 30 cycles of amplification at 94C for 45 s, 55C for 1 min., and 72C for 2 min; and a final extensio n of 72C for 7 min. Sample fragments were then discriminated by using on-chip gel electrophoresis with the Agilent 2100 Bioanalyzer and DNA LabChip Kit 7500. Briefly, a ladder with known fragment sizes is loaded and a standard cu rve of migration time versus frag ments size is plotted. The size of each fragment in the sample is then calcu lated by taking into account the migration times measured. Lower and upper marker standards are also run with each sample in order to align the ladder data with data from the sample wells. Data is translated into gel-like images (bands) and electrophelograms (peaks). After the dominant peaks were detected, the PCR products were loaded in a 1.2% agarose gel and the bands were identified and exci sed. The bands were purified using QIAEX II Gel Extraction Kit (Qiagen, Valencia, CA) following the manufactures instru ctions. The purified fragments were cloned into a TOPO TA Cloning (Invitrogen, Carlsbad, CA). Plasmids were purified with the QIAprep spin miniprep kit (Qiagen, Valencia, CA) and gene fragments were cycle sequenced using T7 primer in a DYEna mic ET terminator cycle sequencing kit (GE Healthcare), on a PTC200 thermocycler (B ioRad) and run on a 96 well MegaBACE 1000 capillary sequencer (GE Healthca re). Once the presence of Ca. Liberibacter spp. was confirmed by sequencing, DNA from 1 positive tree wa s used for further analysis. Sequences of ARISA bands were deposited onto GenBa nk under accession numbers GQ254546 to GQ254633. 63

PAGE 64

ARISA Sequence Analysis and Classification The sequences obtained for each excised band we re initially analyzed by using VecScreen, an on-line tool for identifying segments of a nucle ic acid sequence that ma y be of vector origin ( http://www.ncbi.nlm.nih.gov/VecScreen/VecScreen.htm l ). After manual edition for the elim ination of potentially chimeric sequences phylogenetic analyzes were conducted using MEGA version 4 (Tamura et al. 2007). The e volutionary distance among the sequences was inferred using the Neighbor-Joining method (Saitou and Nei 1987), and a bootstrap test (500 replicates) was conducted in order to ca lculate the confidence limits of the phylogeny (Felsenstein 1985). All positions containing gaps and missing data were eliminated from the dataset (Complete deletion option). High-Throughput Sequencing Three high-throughput sequencing m e thods were used to assess th e bacterial diversity of a symptomatic C. sinensis cv. Valencia plant. DNA from one Ca. Liberibacter sp. positive plant was used for 454 pyrosequencing, Solexa, and SO LiD sequencing. Analyses were performed with DNA extracted from an infected citrus tree positive for citrus gr eening according to the ARISA analysis, cloning, and sequencing of the DNA fragments. For 454, 10 g of DNA were sent to the Interdisciplinary Center for Biotechnology Research (ICBR) at the University of Florida ( http://www.biotech.ufl.edu/ ), prepared according to the m anufacturers instructions and sequenced using the 454 Genome Sequencer FLX System (454 Life Sciences, Branford, CT). An additional 10 g of DNA were sent to ICBR at the University of Florida and used for SOLiD sequencing according to the manufacturer protocol s (Applied Biosystems, Carlsbad, CA). For Solexa sequencing, 10 g of DNA were sent to the Ce nter for Genome Research and Biocomputing at Oregon State University ( http://www.cgrb.oregonstate.edu/ ) where it was 64

PAGE 65

prepared using the m anufacturers standard protocols and run on the Illumina 1G Genome Analyzer (Illumina, Hayward, CA). All metageno mic sequences are available from the National Center for Biotechnology Information (NCBI) Short ReadArchive, accession numbers SRR016809, SRR016816, and SRR017902. Sequence Analysis Reads obtained from each sequencing met hod were analyzed using CLC Genomics Workbench Version 3.2 ( www.clcbio.com ). Parameters used were 99% similarity for 454 reads, a lim it of 1 for Solexa and SOLiD, and a mismatch cost of 3. All other parameters were set to the programs default. Analyses of the data from each sequencing method were run individually and pooled together. To determine microbial community composition within C. sinensis phloem, all sequence data was first run through a reference assembly against the C. sinensis chloroplast genome from NCBI (NC_008334) to e liminate false positives from chloroplast ribosomal RNA. Sequencing reads that did not a ssemble to the chloroplast genome were used in a reference assembly against the 16S database from the Ribosomal Database Project ( http://rdp.cme.msu.edu ). The reference assemblies were done on the entire dataset as well as individually for 454, Solexa, and SOLiD data. T o determine if any viruses or viroids were present in the phloem of the gree ning infected plant, all sequenc ing reads were run in reference assemblies against viroid and virus databases. The viroid database was composed of all 39 completed viroid genomes from NCBI and the virus database was composed of all annotated virus genomes from the Viral Bi oinformatics Resource Center ( http://athena.bioc.uvic.ca/ ). To estim ate the level of Ca. L. asiaticus genome coverage present in the phloem sequences, a reference assembly of the data was run against all Ca. L. asiaticus str. psy62 contigs available in GenBank (NZ_ABQW00000000). Four reference assemblies were performed with the putative Ca L. asiaticus contigs, one each for 454, SOLiD, and Solexa 65

PAGE 66

data, as well as a collective reference assem b ly with all data from each method. Prior to performing the reference assemblies, the 34 Ca L. asiaticus contigs from GenBank were assembled in SeqMan Pro Ver. 7.2.1 (DNAstar, Lasergene), which reduced the number of contigs to 32. In order to confirm the Ca L. asiaticus matches seen in citrus phloem were not the result of other alphaproteobacteria pres ent in citrus phloem, reference assemblies with the sequenced genomes of 5 alphaproteobacterial relatives were performed. The ge nomes used in this analysis were selected by performing a local BLAST search of the Ca. L. asiaticus contigs against a database of 40 fully sequenced alphaproteobacteria, including a representative from each fully sequenced genus on NCBI (Accessi on numbers NC_009484 .1, NC_003062.2, NC_004842.2, NC_009937.1, NC_008783.1, NC_010581.1, NC_004463.1, NC_010742.1, NC_002696.2, NC_009952.1, NC_007354.1, NC_007722.1, NC_011365.1, NC_006677.1, NC_008343.1, NC_008358.1, NC_007802.1, NC_007626.1, NC_008347.1, NC_002678.2, NC_010172.1, NC_007798.1, NC_007964.1, NC_007794.1, NC_009668, NC_009488.1, NC_008686.1, NC_009719.1, NC_007205.1, NC_011144.1, NC_008380.1, NC_007494.1, NC_008435.1, NC_007643.1, NC_009881.1, NC_008209.1, NC _009511.1, NC_002978.6, NC_009720.1, and NC_006526.1). The 5 species that we re most closely related to Ca. Liberibacter were chosen for this study. Those 5 genomes were R. leguminosarum bv. viciae 3841 (NC_008378, NC_008379, NC_008380, NC_008381, NC _008382, NC_008383, & NC_008384), Bartonella bacilliformis KC583 (NC_008783.1), Caulobacter crescentus CB15 (NC_002696.2), Brucella abortus S19 (NC_010740 & NC_010742), and Orientia tsutsugamushi str. Boryong (NC_009488.1). Ca. P. asteri and P. acnes have also been reported to be associated with HLB infection in citrus; therefore, reference assemb lies were performed with fully sequenced strains 66

PAGE 67

of these species, Ca. P. asteris AY WB (N C_007716, NC_007717, NC_007718, NC_007719, NC_007720) and P. acnes KPA171202 (NC_006085). An additional reference assembly was performed against the genome of the non-HLB-associated bacterium, M. luteus NCTC 2665 (NC_012803), as a comparison. To determine the detection limit for identifying a bacterial genome present in the metagenomic data, the mean and standard deviations for the fold coverage of R. leguminosarum bv. viciae 3841, B. bacilliformis KC583, C. crescentus CB15 B. abortus S19 O. tsutsugamushi str. Boryong Ca. P. asteris AYWB, and P. acnes KPA171202 were calculated. As 99% of observed genomes are likely to fall within 3 standard deviations of the mean assuming a normal distribution, any genome with enough matching reads and bases to have a fold coverage greater than 3 standard deviations above the mean was considered significant. The detection limit for identifying a true match from the RDP databa se was determined using the same method by calculating the mean and standard deviation of the fold coverage from each match in the database. With such a large amount of sequence data from citrus phloem, it is expected that the dataset also contains a high fold coverage of th e citrus genome. To confirm this, a reference assembly of all citrus phloem data was performed against a citrus EST database. The citrus EST database was composed of all C. sinensis cv. Valencia ESTs from KEGG ( http://www.genome.jp/kegg-bin/show_organism?org=ecsi ). These ESTs were run in a de novo assem bly on CLC Genomics Workbench to e liminate duplicates. A ssembled contigs and non-assembled ESTs were used as the reference in the assembly with the complete citrus phloem metagenome dataset. 67

PAGE 68

Figure 3-1. Automated ribosomal intergenic spacer analysis (ARISA) of healthy and Huanglongbing infected Citrus sin ensis phloem. Example of the ARISA profiles for the DNA isolated from the phloem tissue of citrus samples, A) non-symptomatic and B) symptomatic, generated by the Agilent 2100 bioanalyzer. Base pair sizes are indicated adjacent to the peaks. The left -most (50 bp) and the right-most (10,380 bp) peaks represent the markers used to align the ladder data with data from the sample wells. The peaks of approximately 740 bp represent sequences amplified from the C. sinensis chloroplast, the peaks of a pproximately 1,330 bp represents the sequences amplified from the mitochondria l gene for mitochondrial RNA, and the peak of approximately 1,600 bp corresponds of the sequences amplified from Candidatus Liberibacter asiaticus. 68

PAGE 69

Table 3-1. Overview of meta genomic sequence statistics Sequencing technique Number of reads Nu mber of gigabases Average read length 454 246,355 0.058 238.48 Solexa 49,423,731 1.8 36 SOLiD 337,075,448 11.8 35 All data 386,745,534 13.6 35.26 Table 3-2. Fold coverage of each Candidatus Liberib acter asiaticus contig determined by reference assembly with 454, Solexa, and SOLiD metagenomic data from Huanglongbing-infected citrus phloem Reference sequencesa 454 Solexa SOLiD All datasets NZ_ABQW01000001 0.054 14.314 12.221 26.595 NZ_ABQW01000002 0.056 15.028 12.005 27.089 NZ_ABQW01000003 0.08 13.949 9.146 23.175 NZ_ABQW01000004 0.083 15.048 13.107 28.272 NZ_ABQW01000005_33b 0.057 15.414 11.843 27.314 NZ_ABQW01000006 0.081 14.459 10.881 25.421 NZ_ABQW01000007 0.061 16.453 12.288 28.802 NZ_ABQW01000008 0.074 13.405 9.257 22.736 NZ_ABQW01000009 0.101 13.624 11.64 25.348 NZ_ABQW01000010 0.026 13.835 8.996 22.855 NZ_ABQW01000011 0.055 12.495 15.286 27.789 NZ_ABQW01000012 0.033 13.196 9.464 22.694 NZ_ABQW01000013 0.046 14.242 11.206 25.494 NZ_ABQW01000014 0.083 13.883 9.675 23.641 NZ_ABQW01000015 0.049 14.103 8.957 23.109 NZ_ABQW01000016 0.07 19.019 9.921 29.009 NZ_ABQW01000017 0.064 14.895 10.927 25.886 NZ_ABQW01000018 0.112 10.782 16.71 27.599 NZ_ABQW01000019 0.124 14.705 9.243 24.071 NZ_ABQW01000020 0.197 14.564 15.644 30.404 NZ_ABQW01000021 0.055 13.022 11.739 24.816 NZ_ABQW01000022_31c 0.341 22.494 29.517 52.363 NZ_ABQW01000023 0.164 16.9 20.3 37.364 NZ_ABQW01000024 0.088 11.734 13.978 25.8 NZ_ABQW01000025 0.099 26.224 14.892 41.211 NZ_ABQW01000026 0.219 10.862 18.077 28.904 NZ_ABQW01000027 0.106 15.764 17.352 33.209 NZ_ABQW01000028 0 15.026 5.616 20.642 NZ_ABQW01000029 0 14.958 11.047 26.042 NZ_ABQW01000030 0 13.966 8.51 22.477 NZ_ABQW01000032 0.333 28.444 31.289 60.067 NZ_ABQW01000034 0 15.95 12.042 27.913 aEach contig represented by its GenBank Ref Seq number. bContig sequence assembled from NZ_ABQW01000005 and NZ_ABQW01000033 using SeqMan Pro. cContig sequence assembled from NZ_ABQW01000022 and NZ_ABQW01000031 using SeqMan Pro. 69

PAGE 70

Table 3-3. Percent length of each Candidatus Liberibacter asiaticus contig covered by 454, Solexa, and SOLiD metagenomic data from Huanglongbing-infected citrus phloem as determined by reference assembly Reference sequencesa 454 Solexa SOLiD All datasets NZ_ABQW01000001 4.951 99.947 94.585 99.989 NZ_ABQW01000002 4.966 99.973 94.272 99.994 NZ_ABQW01000003 6.826 99.93 91.246 99.966 NZ_ABQW01000004 7.413 99.671 93.607 99.784 NZ_ABQW01000005_33b 5.022 99.936 92.817 99.947 NZ_ABQW01000006 7.205 99.991 92.966 99.999 NZ_ABQW01000007 4.917 99.927 92.189 99.946 NZ_ABQW01000008 6.635 99.97 91.62 99.995 NZ_ABQW01000009 7.100 99.891 93.702 99.997 NZ_ABQW01000010 2.545 99.928 91.094 99.964 NZ_ABQW01000011 3.075 99.393 95.527 99.691 NZ_ABQW01000012 3.295 99.927 93.125 100.002 NZ_ABQW01000013 3.899 99.964 93.291 99.964 NZ_ABQW01000014 3.223 99.952 91.694 99.952 NZ_ABQW01000015 4.942 99.964 91.593 100 NZ_ABQW01000016 6.984 99.991 91.402 99.995 NZ_ABQW01000017 5.456 99.989 93.582 99.989 NZ_ABQW01000018 7.112 99.469 98.340 100 NZ_ABQW01000019 10.086 99.983 93.184 99.983 NZ_ABQW01000020 16.012 99.939 97.128 99.983 NZ_ABQW01000021 4.232 99.893 94.603 100 NZ_ABQW01000022_31c 28.651 99.525 97.375 99.835 NZ_ABQW01000023 14.74 99.88 97.985 100 NZ_ABQW01000024 4.783 100 97.285 100 NZ_ABQW01000025 9.843 99.837 91.628 99.837 NZ_ABQW01000026 21.144 98 95.869 99.951 NZ_ABQW01000027 10.614 100 99.321 100 NZ_ABQW01000028 0 99.854 82.389 99.854 NZ_ABQW01000029 0 99.849 96.939 100 NZ_ABQW01000030 0 99.873 90.361 100 NZ_ABQW01000032 25.874 99.563 100 100 NZ_ABQW01000034 0 100 96.902 100 aEach contig represented by its GenBank Ref Seq number. bContig sequence assembled from NZ_ABQW01000005 and NZ_ABQW01000033 using SeqMan Pro. cContig sequence assembled from NZ_ABQW01000022 and NZ_ABQW01000031 using SeqMan Pro. 70

PAGE 71

Table 3-4. All metagenomic phloem seque nces compared to the contigs of Candidatus Liberibacter asiaticus or th e fully sequenced genomes of other alphaproteobacteria by reference assembly Bacterial species Genome size (bp) Number of matching reads Number of matching bases Fold coverage Candidatus Liberibacter asiaticus str. psy62 1,217,424 900,124 32,071,418 26.34 Rhizobium leguminosarum bv. viciae 3841 5,892,908 8,027 281,989 0.048 Bartonella bacilliformis KC583 1,445,021 7,209 253,180 0.175 Caulobacter crescentus CB15 4,016,947 5,969 209,512 0.052 Brucella abortus S19 2,238,636 7,505 263,576 0.118 Orientia tsutsugamushi str. Boryong 2,127,051 2,339 82,333 0.039 Table 3-5. Fold coverage of the top 16S rR NA RDP database m atches with 454, Solexa, and SOLiD metagenomic data from Huanglongbing-infected citrus phloem Reference sequences 454 Solexa SOLiD All Data Sets Candidatus Liberibacter 0.328 5.23 17.365 23.173 Nitrospina gracilis 0 0 1.427 1.451 Candidatus Burkholderia 0 0.375 0.885 1.301 Syntrophus gentianae 0 0.046 0.603 1.028 Nitratifractor salsuginis 0 0 0.942 0.991 Syntrophus buswellii 0 0.022 0.51 0.895 Pseudomonas rhizosphaerae 0 0.023 0.616 0.886 Hoeflea marina 0 0.073 1.229 0.85 Thermonema rossianum 0 0 0.766 0.837 Thermodesulfovibrio islandicus 0 0.231 0.645 0.804 Leucothrix mucor 0 0.094 0.685 0.802 Desulfobacter halotolerans 0 0.073 0.448 0.732 Pseudomonas argentinensis 0 0 1.049 0.685 Desulfospira joergensenii 0 0.048 0.506 0.627 Martelella mediterranea 0 0 0.993 0.615 Marinomonas aquimarina 0 0 0.509 0.554 Mycoplasma cloacale 0 0.025 0.427 0.527 Phaeospirillum fulvum 0 0.051 0.448 0.525 Desulfotignum phosphitoxidans 0 0.074 0.508 0.508 Desulfomonile tiedjei 0 0.14 0.369 0.508 Desulfobacula toluolica 0.03 0 0.223 0.345 71

PAGE 72

Table 3-6. Comparison of Candidatus Liberibacter 16S rRNA to 454, Solexa, and SOLiD metagenomic sequences from Huanglongbing-in fected phloem by reference assembly Sequencing technique Number of matching reads Fold coverage Length of consensus (bp) % 16S gene matching consensus sequence 454 2 0.328 379 26.17 Solexa 211 5.23 950 65.61 SOLiD 720 17.37 1,118 77.21 All Data 944 23.17 1,212 83.7 Table 3-7. Comparison of Candidatus Liberibacter asiatic us and other reported Huanglongbing-associated bacteria to all m etagenomic phloem sequences Bacterial species Genome size (bp) Number of matching reads Number of matching bases Fold coverage Candidatus Liberibacter asiaticus str. psy62 1,217,424 900,124 32,071,418 26.34 Candidatus Phytoplasma asteris 723,970 3,926 138,117 0.191 Propionibacterium acnes KPA171202 2,560,265 1,744 61,452 0.024 Micrococcus luteus NCTC 2665 2,501,097 2,677 94,125 0.038 Table 3-8. Comparison of metagenomic assemblies with to tal metagenomic data to metagenomic data with chloroplast sequences removed Bacterial species Total fold coverage Fold coverage minus chloroplast Total bases matched Bases matched minus chloroplast Candidatus Liberibacter asiaticus str. psy62 26.34 26.179 32,071,418 31,871,539 Rhizobium leguminosarum bv. viciae 3841 0.048 0.018 281,989 104,598 Bartonella bacilliformis KC583 0.175 0.058 253,180 83,988 Caulobacter crescentus CB15 0.052 0.014 209,512 57,130 Brucella abortus S19 0.118 0.038 263,576 84,194 Orientia tsutsugamushi str. Boryong 0.039 0.017 82,333 35,218 Candidatus Phytoplasma asteris 0.191 0.016 138,117 11,356 Propionibacterium acnes KPA171202 0.024 0.009 61,452 22,352 Micrococcus luteus NCTC 2665 0.038 0.004 94,125 9,822 72

PAGE 73

Table 3-9. Candidatus Liberibacter asiatic us contig primers Primer name Sequence (5 3) NZ_ABQW01000001F CTGGTGTTCGCAAGGGTATT NZ_ABQW01000001R ACACCC GAACGCATTAACTC NZ_ABQW01000002F ACTGCCCTGCTTTTTCTTCA NZ_ABQW01000002R ACA GGTCGTGCTATGGTTCC NZ_ABQW01000003F TGGA AATTTCGGAGAGGATG NZ_ABQW01000003R GAAATGTTGCTGCGACTTCA NZ_ABQW01000004F TGATTCTCCTCATCCCCAAG NZ_ABQW01000004R CTTTG GTGGTATTCGCTGGT NZ_ABQW01000006F GATTTAGCGGAATCGGATCA NZ_ABQW01000006R TCAT GCCTTGCTTGAAACAG NZ_ABQW01000007F CGCATCCACAGTTTATGGTG NZ_ABQW01000007R AGCCGCTACAGCAATCCTAA NZ_ABQW01000008F GGTTCCAGAAGTTGCTGCTC NZ_ABQW01000008R CAACACGCCCCTTATCTTGT NZ_ABQW01000009F ACGAGCGTAACCAACAATCC NZ_ABQW01000009R AGGCT CTTGGACTCACGAAA NZ_ABQW01000010F CTGCTTCAATAAGCGGGAAG NZ_ABQW01000010R CTGAGCAACATGGCTCAAAA NZ_ABQW01000011F CTGTTACGGTTGCTCCCATT NZ_ABQW01000011R CGAATTGTCTAGCAGCCACA NZ_ABQW01000012F GGGCGTACAACCCAACTCTA NZ_ABQW01000012R TTTAACGAGCCGCTTCAAGT NZ_ABQW01000013F TACGCCTGCATATGCTTCAG NZ_ABQW01000013R TGTC ATGAAGCGAGAAAACG NZ_ABQW01000014F ACCCAAAAGTCGGTGTCTTG NZ_ABQW01000014R CCGTAAACACGAAAGGCAAT NZ_ABQW01000015F GGGGC AAGCTCATGATAAAA NZ_ABQW01000015R AGTTG CGAGAAGGCTTGGTA NZ_ABQW01000016F CTTGATCTCGCCACTGTTGA NZ_ABQW01000016R CATTCGCAGTGCTATGGCTA NZ_ABQW01000017F ACGACAGGGGTGCAAGTATC NZ_ABQW01000017R TGCATC AGAAAGCGCAATAC NZ_ABQW01000018F ACGGGGCTTGGTAAAGAAGT NZ_ABQW01000018R CAAACT CCCGTTTCACACCT NZ_ABQW01000019F AGAGT CATGGAAGCGGCTAA NZ_ABQW01000019R CCGAGGGATATCAGCGATTA NZ_ABQW01000020F GAAGC AATATGGGCTGGAAA NZ_ABQW01000020R TTCA GAAGACATGCCACGAG NZ_ABQW01000021F AATTGGAACTTCCCCCAAAC NZ_ABQW01000021R ACGA TGCTTTGCTCGATCTT NZ_ABQW01000023F CCCTCTTCCAATCCCTTCTC NZ_ABQW01000023R CTATATCGGCAGGTCGTGGT NZ_ABQW01000024F CTCTTGCATCCAACGACAGA 73

PAGE 74

Table 3-9. Continued Primer name Sequence (5 3) NZ_ABQW01000024R GGACGGTTGTCTTACCCTCA NZ_ABQW01000025F GGGAGT ACGGAGAGTGGTGA NZ_ABQW01000025R TTGATAAAGACGCGCTGTTG NZ_ABQW01000026F CGGCGAAAGAGCTTTACAAC NZ_ABQW01000026R CTGTGGCAATAGCGTTGAGA NZ_ABQW01000027F CGAGAAGGGGGAGGATTAAG NZ_ABQW01000027R AGCGTTATAGGCTCCAGCAA NZ_ABQW01000028-2kF TGCAATCAAGGCTCACCATA NZ_ABQW01000028-2kR TCCT ACTCCTTCCGCCATAA NZ_ABQW01000029-2kF CAATGAGAGATAAGACAATCCTGTTT NZ_ABQW01000029-2kR GCTT AAAGCTGATAAACCCATCA NZ_ABQW01000030F-1kF TCCAGAATCCCGTCTTGTTC NZ_ABQW01000030R-1kR GT GGCTTTGGGTTTGTCTGT NZ_ABQW01000032F-1kF GGG CATTGAGCCTGAATCTA NZ_ABQW01000032R-1kR TGCA AAACGAGAAACTGACG NZ_ABQW01000034F-1kF CTTGTCGGCAAGTTGGTTTC NZ_ABQW01000034R-1kR ATGG CGAAAAACGACAGTTC NZ_ABQW01000022_31F GGAAC AAAGCAAACGAGAGC NZ_ABQW01000022_31R CAAAAGGAGGAAGTCGCAAG NZ_ABQW01000033_05F GTTGTGTGCCGAGGATAGGT NZ_ABQW01000033_05R GAGGA CAGACCGCTCTCAAC 74

PAGE 75

CHAP TER 4 OPTICAL MAPPING OF GL UCONACETOBACTER DIAZOTROPHICUS PAL 5 REVEALS CHROMOSOMAL REARRANGEMENTS IN COMPLETED GENOME SEQUENCE Background Gluconacetobacter diazotrophicus P Al 5 is a bacterial endophyte of sugarcane, originally isolated in Brazil (Gillis et al 1989). This endophyte is of ag ricultural significance due to its ability to provide fixed nitrogen to its host plant, in addition to in creasing plant growth by mechanisms independent of nitrogen fixation (Lee et al. 2004; Sevilla et al. 2001). The ability of G. diazotrophicus to increase growth and reduce plant de pendence on nitrogen fertilization also makes it important to increasing the efficiency of biofuel production from sugarcane (Boddey 1995). Since it was first isolated, other strains of G. diazotrophicus have been isolated in several other countries and plant hosts (Boddey et al. 199 5; Fuentes-Ramirez et al. 1993; Hoefsloot et al. 2005; Paula et al. 1991; Jimenez-Sa lgado et al. 1997; Tapia-Herna ndez et al. 2000). As a result, there has been great interest in sequencing the genome of G. diazotrophicus to guide further research on this bacterium and to better unde rstand endophytic nitrogen fixation by comparative genomics with other sequenced nitrogen fixing endophytic bacteria. The genomic sequencing of G. diazotrophicus PAl 5 was undertaken by two independent groups, RioGene in Brazil, funded by FAPRJ, an d the United States Department of Energys Joint Genome Institute (JGI) in California. Bo th groups have since closed the genomic sequence of G. diazotrophicus PAl 5 and deposited it on GenBank (Accession numbers CP001189.1 and AM889285.1). Interestingly, though both groups repor ted sequencing the PA l 5 strain, the two genome sequences varied greatly between each othe r in gene arrangement and plasmid content, indicating the presence of sequencing and/or assembly errors in one of the genomic sequences. To ascertain which genome assembly more clos ely matches the physical map of PAl 5, optical mapping was utilized. 75

PAGE 76

Optical m apping serves to create a physical restriction map of a genome assembled from deoxyribonucleic acid (DNA) molecules immobilized on a glass slide prior to digestion with a selected restriction enzyme, maintaining the orig inal order of restriction fragments. After digestion, DNA is stained and visualized by fluor escent microscopy, and the resulting digitized images are analyzed in an assembly program to construct an optical restriction map of the genome of interest (Aston et al. 1999; Zhou et al 2004a). These optical maps can be compared to in silico digests of DNA sequences and have been ut ilized in many sequencing studies, serving as scaffolds for contig alignment, as well as an independent means of identifying errors (inversions, insertions, deletions translocations, etc.) in pr eviously assembled sequences (Latreille et al. 2007; Lim et al 2001; Reslewic et al. 2005; Wu et al. 2009; Zhou et al. 2007; Zhou et al. 2002; Zhou et al. 2004b). Therefore, optical mapping was deemed to be an ideal tool to elucidate which PAl 5 genome sequence most closely matched the physical DNA. Results Optical Map of G. diazotrophicus PAl 5 A BglII optical m ap of G. diazotrophicus PAl 5 (ATCC 49037) was constructed in order to determine which genome assembly was the most accurate representation of the strain. The optical map was 3,845,512 basepairs (bp) in length and composed of 424 restriction fragments, with an average fragment size of 9,070 bp (Table 4-1). In comparison, the in silico map of the JGI sequence was 3,887,492 bp in length, while the RioGene map was 3,944,163 bp (Table 4-1). The average fragment length of both in silico maps is over 1,000 bp shorter than the average fragment length of the optical map (Table 41). These differences between the optical and in silico maps are likely due to the fact that re striction fragments shorter than 500 bp are not detected by optical mapping owing to such short fragments being washed off the optical slide (Meng et al. 1995). 76

PAGE 77

Identification of Sequence Rearrangements Using Optical Mapping Once the BglII optical map of PAl 5 was aligned to in silico BglII restriction maps generated from the two separate genome sequences it was readily apparent that the sequence from RioGene contained numerous chromosomal rearrangements (Figure 4-1a). Comparison of the optical map to the RioGene in silico map revealed the presence of 2 large inverted regions (Figure 4-1b). These inversions were 555.9 and 564.3 kilobases (kb) in length, together spanning close to 28% of the genome sequen ce (Table 4-2). In addition, numerous translocations were identified in the RioGen e sequence. One large translocation spanning 865.8 kb of the genome (Figure 4-1c) and 5 smaller translocations ranging in size from 69.8 to 330.8 kb (Figure 4-1d) were identifie d (Table 4-2). From these de terminations, it appears that 74% of the PAl 5 genome sequence proposed by RioGene is rearranged compared to the physical map of the PAl 5 genome. In contrast, the in silico map of the JGI PAl 5 sequence showed a higher alignment to the optical map (Figure 41a). Only 3 small inversions and 1 small translocation were detected. These regions only accounted for 5.6% of the genomic sequence. Additionally, several regions of the RioGene BglII in silico map did not align to the optical map (Figure 4-1a). Together, these regi ons totaled 1,053,347 bp, or 26.7% of the genome sequence (Table 4-3). In comparison, the regions of the JGI in silico map that did not align to the optical map were composed largely of singl e restriction fragments, most of which were 500 bp in length or less (Table 4-3), which is the below the detection threshold of optical mapping technology. All told, it appears that n early all of the PAl 5 genome sequence from RioGene was either rearranged or did not align to the optical map of PAl 5. From these analyses, it was concluded that the sequence from JGI is a more accurate representation of PAl 5s genome. 77

PAGE 78

Differences in Annotation Between Genome Sequences Given the surprisingly high level of chromo somal rearrangements and non-aligned regions between genomic sequences reported from the same strain, the annotations of both PAl 5 sequences were determined using the Rapid Annotation using Subsystem Technology (RAST) (Aziz et al. 2008) web based annot ation service to ascertain what effect these rearrangements have on gene calling. With a tota l of 8 rearrangements in the Ri oGene sequence, there are up to 16 locations where coding sequen ces (CDSs) could have been di srupted. Interestingly, 168 and 187 of the CDSs identified in the RioGene and JG I genomes, respectively, were unique, sharing 0% identity with CDSs in the other genome (Tab le 4-4). In total, 247 of the CDSs in the RioGene sequence shared less than 50% identity w ith CDSs in the JGI sequence (Table 4-4). This number of differences between the two ge nome sequences was over 10 times greater than expected from the observed inversions and translocations in the RioGene sequence. Since both genomic sequences are from the same strain, a similar complement of genes was expected. However, only 90% of the CDSs predicted in each genome shared greater than 90% identity (Table 4-4). Considering both genome sequences were generated from the same strain, the level of differences between them at the sequence level was surprising. Fewer differences were seen between the two genomes when looking at the functional roles genes were assigned to. In total, 13 and 21 functional roles identified were found to be unique to the RioGene and JGI se quences, respectively (Table 4-5) Interestingly, two of the unique functional roles identified in the RioGene sequence were Cand N-terminal sections of a transketolase enzyme, though complete transketolase genes were identified in both the genome sequences. While this observation initially s uggested a rearrangement had occurred within a transketolase gene in the RioGene sequence, ex amination of the genome region containing these transketolase genes showed that the C-term inal gene was immediately upstream of the 78

PAGE 79

N-term inal section. This confi guration of transketolase genes was also observed in 3 other bacteria in the SEED database ( Ruegeria sp. PR1b, Desulforubris audaxviator and Carboxydothermus hydrogenoformans ), indicating that this sp lit is not the result of a chromosomal rearrangement. In confirmation of this, none of the inversion or translocation break points fell between these two genes. Given the number of inversions and translocations in the RioGene sequence, the annotation was also checked for transposases that could contribute to chromosomal rearrangements. The RioGene PAl 5 sequence was found to possess 110 transposase genes, wh ile the JGI sequence only contained 59 transposase genes, almost half that amount. A large number of these were putative transposases, though severa l insertion sequence (IS)3, IS4, and IS5 family proteins were also identified as well (Table 4-6). Interestingl y, only 2 translocated regions had transposases less than 10 kb from each of their ends. A 69.7 kb translocation had an IS4 transposase family protein 6.9 kb upstream and an IS5 transposase family protein 814 bp downstream of the region. Another 171 kb rearrangement had putative tr ansposases 193 bp and 6.9 kb up and downstream of the translocated region. A third transl ocated region had a putative transposase 348 bp upstream, but the nearest downstream transposase was greater than 30 kb away. In all other cases, the nearest transposase to an inverted or translocated region was over 18 kb away. Discussion The production of two different genom e sequences from the same bacterium, G. diazotrophicus PAl 5, demonstrated the need to confir m the assembly of these genomes through an independent method. Optical restriction mapping has been used by many groups as an independent method of verifying sequences and iden tifying assembly errors, due to the fact that it maintains the order of restric tion fragments in the mapping process. To that end, these two genome assemblies were compared to a BglII optical map of the PAl 5 bacterium, leading to the 79

PAGE 80

determ ination that the sequence reported by JGI is a more accurate representation of the PAl 5 strain from the American Type Culture Colle ction (ATCC) while th e sequence reported by RioGene contained numerous rearrangements, including 2 large inversions and several translocations, when compared to the physical map of the PAl 5 strain. In the current study, optical mapping was used to distinguish between discordant genomic assemblies of the same bacterial strain. The size and number of chromosomal rearrangements identified in the RioGene sequence of G. diazotrophicus PAl 5 was high, with nearly all of the sequence composed of regions that were inverted, translocated, or not aligned to the optical map of the PAl 5 strain. In contrast, only a few sma ll inversions were detect ed when the JGI PAl 5 sequence was compared to the optical map. In addition, annotation of the two genome sequences found that approximately 5% of the CDSs in each were unique to each genome sequence. This is a surprisingly high amount considering the two ge nomes are reported to be from the same strain and much greater than would be expected from the observed sequence rearrangements. There are a few possibilities for the differences between these two PAl 5 genome sequences. One explanation for the differences between the RioGene and JGI se quences is assembly errors. While many studies have reported using optical maps to aid in genome assembly and identification of assembly errors prior to comple tion, fewer have reported using this technique to identify errors in previously completed genomes After the successful us e of optical mapping to aid in assembling the Xenorhabdus nematophila genome, Latreille and associates (2004) used the same technique on another Xenorhabdus species, X. bovienii that had been previously sequenced, identifying a large inversion in the genome assembly that had been considered finished. In addition, optical ma pping has also been used to veri fy assemblies between strains of the same species. In the case of Mycobacterium avium subspecies paratuberculosis an optical 80

PAGE 81

m ap of the ATCC type strain was used to reveal the presence of an inversion in the genome of the sequenced strain, which was determined to be due to an assembly error rather than genomic variation between strains (Wu et al. 2009). These two instances illustrate how even closed and published genomes may contain sign ificant assembly errors, indi cating that caution should be taken when looking at assemblies where optical mapping was not used. Another reason for the differences between th e RioGene and PAl 5 sequences is natural divergence and evolution that can occur during culturing, though the extremely high level of differences between the two sequences indicates other factors may be involved. The amount of sequence rearrangement seen betw een the PAl 5 optical map and the RioGene PAl 5 sequence is higher than the level of rearrangements seen betw een different strains within the same species. Comparison of an optical map of E. coli H10407 to the sequence of E. coli K-12 showed no major structural differences between the two strains (Chen et al. 2006). In M. avium subspecies paratuberculosis only 1 inversion between the sequenced strain, K-10, and the optically mapped strain, ATCC 19698, was detected and that inversion was subsequently determined to be due to an assembly error rather than a true chromoso mal rearrangement (Wu et al. 2009). In addition, when comparing Shigella flexneri strains 2457T and 301, Zhou and associates (2004a) found 3 inversions that were 876, 72, and 20 kb in le ngth. Given these num bers, if the sequence inversions and translocations seen between th e RioGene PAl 5 sequence and the PAl 5 optical map are due to true chromosomal rearrangements that occurred as a result of evolution during culturing, one must question whether they can still be considered the same strain. If the breakpoints of assembly errors such as inversions or translocations occur within a coding region, such errors could alter the annot ation of the genome. For example, when the previously mentioned inversion in the sequence of M. avium subspecies paratuberculosis K-10 81

PAGE 82

was corrected, 2 new genes were identified (W u et al. 2009). As a result, the annotation of the PAl 5 sequences from both RioGene and JGI we re determined and compared using the RAST on-line annotation pipeline (Aziz et al. 2008). Both sequences were reanno tated instead of using the original annotation provide d on GenBank to avoid any bias es based on differences in annotation methods. Six percent of the CDSs fr om each genome shared less than 50% identity when compared against each other. Interestin gly, approximately 5% of the CDSs showed 0% identity when the two genomes were compare d, equating to 168 and 187 unique genes in each sequence. Again, this level of sequence and gene difference was surprising from what is supposed to be the same strain, even with ch romosomal rearrangements. Possible reasons for such differences could be evolution of the strain, or sequencing errors. Annotation of both genomes also revealed that the RioGene sequenc e possessed almost twice as many transposases as the JGI sequence. The stri kingly high number of transposas es in the RioGene sequence in relation to the JGI seque nce suggests the possibility that some of the sequence rearrangements seen may be the result of transposition. Alternatively, since 16 of the transposases originated from IS sequences, which are flanked by invert ed repeats (Mahillon and Chandler 1998), it is also possible that these repeated regions caused errors in assembly. The G. diazotrophicus PAl 5 optical map was constructe d from a PAl 5 isolate obtained directly from the ATCC and maintained in -80oC freezer stocks until cu ltured to obtain DNA for the optical map. Therefore, it can be said with confidence that the PA l 5 optical map is an accurate representation of the original PAl 5 strain submitted to ATCC. If the observed chromosomal inversions and transl ocations are the result of evolution in culture, this would indicate that the bacterium RioG ene sequenced has evolved to the point that it may no longer be considered the same as the original PAl 5 stra in. If sequencing errors are at fault, only 82

PAGE 83

exam ination of the raw sequencing data could id entify such a cause. The true origin of the differences between these two sequences, whethe r it be genuine chromosomal rearrangements that occurred during culturing of the bacteria or a combination of errors in sequencing and assembly, is difficult to ascertain without exam ining the raw sequencing reads or the specific isolate used by RioGene, which were unavailable Given the degree of rearrangements seen between these two sequences, it is possible a combination of chromosomal rearrangement and assembly errors may be the cause. Conclusions The genom ic sequence of G. diazotrophicus PAl 5 produced by JGI was deemed to be the most accurate representation of the genome as determined by optical mapping, while the RioGene sequence contained numerous rearrangements Therefore, subsequent studies of this bacterium involving examination of the genom e should utilize the JGI sequence. The observations made here further confirm the utility of optical mapping in determining proper assembly of genomic sequences and identifying po tential chromosomal rearrangements. It also highlights the need to provide raw reads and quality scores when submitting genomes to allow for independent confirmation of assembly. As technology advances and new and improved annotation programs are develope d, data from instances where contradictory sequences are observed could be reanalyzed in order to clarify results. The rearrangements in the RioGene sequence of G. diazotrophicus PAl 5 may not have been identified had JGI not released a conflic ting genome sequence of the same strain that prompted further investigation. As the genome sequencing of a single bacterial strain is not usually performed independently by different groups, the possibility remains that other previously released, closed genomes could c ontain similar differences compared to other bacterial isolates under the same strain designation main tained in different laboratories. Such 83

PAGE 84

rearrangem ents in genome sequences of the same strain could confound future work by researchers using comparative genomics to look for variations between closely related organisms. In such cases, again, the best tool to distinguish actual vari ations between organisms will be optical mapping. Therefore, it is propose d that raw sequencing reads with quality scores be made available when publishing complete d genomes and that optical mapping become a regular tool in genome assembly projects to ensu re that differences which arise between genomic sequences from the same or related strains ar e genuine chromosomal rearrangements as opposed to assembly or sequencing errors. Methods Bacterial Strain The bacte rial strain used in this study was Gluconacetobacter diazotrophicus PAl 5 obtained from the American Type Culture Collection (ATCC 49037). G. diazotrophicus PAl 5 was cultured on yeast mannitol agar (YMA) and broth at 30oC. Preparation of Cells for Optical Mapping G. diazotrophicus PAl 5 was grown in a 5 m L YM broth until the cells reached a density of 109 colony forming units per mL. The culture wa s dispensed into five 1.5 mL microcentrifuge tubes in 1 mL aliquots. Tubes were then centr ifuged at 6,000 rpm for 10 minutes to pellet the cells. Tubes with cell pellets were shipped on dry ice to OpGen Technologies, Inc. (Madison, Wisconsin) for optical mapping. Optical Mapping and Analysis A BglII optical m ap of G. diazotrophicus PAl 5 was constructed by OpGen Technologies, Inc. (Madison, Wisconsin). In silico BglII restriction maps of the two complete G. diazotrophicus PAl 5 genomic sequences on GenBank (Accession numbers CP001189 and AM889285) were constructed from each sequences GenBank file and compared to the BglII 84

PAGE 85

optical m ap of PAl 5 using MapViewer versi on 2.1.1 (OpGen Technologie s, Inc.). Plasmid sequences associated with each genome assembly did not align to the optical map and were therefore not included the analysis. Comparison of Annotation The annotations of the two genom ic assemb lies were determined using RAST ver. 2.0 (Aziz et al. 2008). Genome and plasmid sequences for RioGene (Accession numbers AM889285, AM889286, and AM889287) and JGI (Accession numbers CP001189 and CP001190) were concatenated into single FASTA f iles prior to RAST analysis. Annotations determined by RAST were compared using the SEED viewer (ver. 2.0) (Overbeek et al. 2005) based on percent identity between CDSs and the functional roles assigned to annotated genes. 85

PAGE 86

A B C D Figure 4-1. Alignment of G. diazotrophicus PAl 5 optical map with in s ilico maps of genome sequences. A) The BglII optical map of PAl 5 aligned against in silico optical maps calculated from the genome sequence proposed by RioGene (AM889285) and JGI (CP001189). B-D) Rearrangements in the RioGene PAl 5 sequence when aligned against the optical map. B) Two large inve rsions in RioGene sequence. C) Large translocation in RioGene sequence. D) Fi ve translocations in RioGene sequence. Dark blue represents cut sites, light blue represents aligned regions, red represents regions aligning to both sequences, and white represents unaligned regions. Alignment lines for inversions and transloc ations are highlighted in pink. Inverted and translocated regions highlighted in yellow. 86

PAGE 87

Table 4-1. Optical and in silico BglII restriction maps for G. diazotrophicus PAl 5 Optical map JGI in silico map RioGene in silico map Map length (bp) 3,845,512 3,887,492 3,944,163 Number of fragments 424 486 503 Average fragment length (bp) 9,070 7,999 7,841 Maximum fragment length (bp) 52,064 51,728 50,690 Minimum fragment length (bp) 562 24 28 Table 4-2. Rearrangement positions in G. diazotrophicus PAl 5 genom e sequence from RioGene Rearrangement type Start position ( bp) Stop position (bp) Length (bp) Inversion 391,267 955,614 564,347 Inversion 3,078,324 3,634,241 555,917 Translocation 149,268 358,682 209,414 Translocation 1,115,930 1,446,765 330,835 Translocation 1,581,823 1,651,595 69,772 Translocation 1,627,253 1,796,352 169,099 Translocation 1,796,352 2,662,168 865,816 Translocation 3,706,896 3,878,171 171,275 Table 4-3. Regions of in silico m aps not aligned to the G. diazotrophicus PAl 5 optical map JGI RioGene Total length of unaligne d regions (bp) 27,540 1,053,347 Average unaligned fragment length (bp) 574 5,885 Maximum unaligned fragment length (bp) 1,341 32,719 Minimum unaligned fragment length (bp) 24 28 Table 4-4. Comparison of coding sequen ces between two genom ic sequences of G. diazotrophicus PAl 5 based on percent identity Percent identity to comparison genome Number CDS from JGI Percent CDS from JGI Number of CDS from RioGene Percent of CDS from RioGene 100% 2,024 56.7 2,069 56.0 99% 2,812 79.2 2,876 77.8 90% 3,190 89.8 3,313 89.6 >75% 3,267 92.0 3,402 92.0 50% 3,326 93.7 3,449 93.3 <50% 225 6.3 247 6.7 0% 187 5.3 168 4.5 87

PAGE 88

Table 4-5. Unique functional roles between G. diazotrophicus PAl 5 genome sequences Roles unique to JGI Ro les unique to RioGene Ribose ABC transport system, periplasmic ribose-binding protein RbsB (TC 3.A.1.2.1) Sorbitol dehydrogenase (EC 1.1.1.14) D-alanine-D-alanine ligase (EC 6.3.2.4) Tr ansketolase, C-terminal section (EC 2.2.1.1) UDP-N-acetylenolpyruvoylglucosamine reductase (EC 1.1.1.158) Transketolase, N-terminal section (EC 2.2.1.1) Organic hydroperoxide resistan ce protein COG0028: Thiamine pyrophosphate-requiring enzymes Organic hydroperoxide re sistance transcriptional regulator D-galactonate regulator, IclR family Molybdenum cofactor biosynthesis prot ein B Epi-inositol hydrolase (EC 3.7.1.-) Flagellar biosynthesis protein fliL Chromosome partition protein smc Flagellar hook-associated protein flgL dTDP-rhamnosyl transferase RfbF (EC 2.-.-.-) Deoxyuridine 5-triphosphate nucleotidohydrolase (EC 3.6.1.23) Protein of unknown function DUF374 Aminopeptidase S (Leu, Val, Phe, Tyr preference) (EC 3.4.11.-) Nicotinate-nucleotide adenylyltransferase (EC 2.7.7.18) Leucyl/phenylalanyl-tRNAprotein transferase (EC 2.3.2.6) DNA repair exonuclease family protein YhaO Cysteinyl-tRNA synthetase (EC 6.1.1.16) ATP-dependent DNA helicase UvrD/PcrA, proteobacterial paralog tRNA:Cm32/Um32 methyltransferase Ou ter membrane lipoprotein carrier protein LolA DNA-binding response regulator KdpE Osmosensitive K+ channel histidine kinase KdpD (EC 2.7.3.-) Potassium-transporting ATPase A chain (EC 3.6.3.12) (TC 3.A.3.7.1) Potassium-transporting ATPase B chain (EC 3.6.3.12) (TC 3.A.3.7.1) Beta-hexosaminidase (EC 3.2.1.52) Potassium-transporting ATPase C chain (EC 3.6.3.12) (TC 3.A.3.7.1) Protein-export membrane protein secD (TC 3.A.5.1.1) H+/Clexchange transporter ClcA 88

PAGE 89

89 Table 4-6. Transposases in G. diazotrophicus P Al 5 genome sequences JGI RioGene Total transposase genes 59 110 Transposase 6 19 Transposase (class II) 1 2 Transposase (class III) 1 0 Transposase (class IV) 1 0 Putative transposase 27 64 Transposase IS3 family protein 2 4 Transposase IS3/IS911 family protein 1 0 Transposase IS4 family protein 6 4 Transposase IS5 family protein 4 7 Transposase IS256 1 0 Transposase IS630 0 1 Isrso16-transposase OrfA protein 1 0 Transposase and inactivated derivative 2 1 Transposase mutator type 5 6 Probable insertion sequence transposase protein 1 0 TRm2011-2a transposase 0 2

PAGE 90

CHAP TER 5 ENDOPHYTE MEDIATED PLANT GROWTH PROMOTION Introduction It has been long known that som e bacteria have beneficial effects on plant growth. Some of these bacteria are believed to enhance growth by providing fi xed nitrogen to plants, but are also known to increase growth yield by mechan isms independent of nitrogen fixation. These bacteria can be rhizobacteria th at live in the soil ar ound plant roots, or en dophytes, living inside the plant tissue without forming symbiotic structures or causing disease. One such endophytic bacteria, Klebsiella pneumoniae 342 (Kp342), was isolated from a nitrogen efficient line of maize (Chelius and Triplett 2000). Kp342 is a mode l for endophytic bacteria because it has been shown to colonize the interior of seve ral plant species, including soybean ( Medicago sativa and M. truncatula ), wheat ( Triticum aestivum), rice ( Oryza sativa ), and Arabidopsis thaliana (Dong et al. 2003a and b). In addition, Kp342 increases the yield of maize grown in the field and Arabidopsis grown in the greenhouse (R iggs et al. 2001 and 2002). These increases in plant growth were independent of nitrogen fixati on since the plants tested were grown under conditions of nitrogen fertilization when nitrogen fixation would be inhi bited (Riggs et al. 2001 and 2002). Kp342 has also been found to fix nitr ogen within wheat. It has been shown to produce the nitrogenase enzyme within the cortex of maize and wheat, and inoculation with this bacterium relieves nitrogen deficiency symptoms in Trenton wheat (Chelius and Triplett 2000; Iniguez et al. 2004). In the course of studying this bacteriu m, confocal microscopy showed that Kp343 colonization is localized around points of lateral r oot emergence, indicating that the bacteria may be entering plants through cracks formed as lateral roots emerge from the primary root (Dong et al. 2003a; Iniguez et al. 2004). During an experime nt to determine if it enters plants in this 90

PAGE 91

m anner, Kp342 was discovered to increase th e number of lateral roots formed by Arabidopsis This unexpected result yielded an interesti ng development in the study of Kp342s nitrogen fixation independent growth promo tion. First, the direct consequen ce of a larger, more branched root system is that plants can obtain more nut rients and water from the soil, enhancing the growth of the plant. Second, th e increase in lateral root numbe rs provided a quick assay for future experiments to determine Kp342s growth-promoting mechanis ms, since the signal responsible for lateral ro ot promotion may also be involved in an overall increase in growth yield. Lateral Root Development Lateral roots develop from a group of differen tiated, pericycle cells in contact with xylem pole cells called founder cells, and their development into lateral roots is regulated by auxin (Malamy and Benfey 1997; Casimiro et al. 2003). These pericycle cells ar e signaled to become lateral root primordia when they pass through the root elongation and di fferentiation zones, but lateral root primordium formation is also sign aled by a second check point that is dependent upon environmental factors later in development (Dubrovsky et al 2000). Lateral root formation begins when lateral root prim ordia develop from founder cells that undergo antic linal divisions resulting in 2 short daughter cells surrounded by 2 long daughter cells (Stage I primordia) and then undergo periclinal divisions, creating a second layer of cells resulting in stage II primordia (Malamy and Benfey 1997). In addition to cell cy cle activation, the fate of these pericycle cells must be respecified in order to undergo latera l root initiation (Vanne ste et al. 2005). After initiation, cells in the lateral root primordia continue to divide and progress to stage VII and emerge from the primary root by cell expansi on (Malamy and Benfey 1997). Upon emergence, the lateral root apical meristem is formed (Malamy and Benfey 1997). It is at this point, when 91

PAGE 92

cells at the lateral roo t tip are short and more numerous, that the primordium is considered to have developed into a lateral root (Malamy and Benfey 1997). The molecular events between auxin signaling an d lateral root formation are not as well understood as the stages of morphological developm ent, partially because the few cells involved in the process are embedded in the root and difficult to access (Himanen et al. 2004). To overcome this obstacle, Himanen and associates (2004) synchronously induced pericycle cells to form lateral roots, increasing th e number of cells undergoing th e process in order to detect changes in gene expression associated with late ral root initiation. They performed microarray transcript profiling on these root s and found that 4 stages prec ede pericycle cell division: G1 cell cycle block, signal transducti on, and progression through the G1/S and G2/M transitions. One of the genes observed in this study, Krp2 a cyclin-dependent kinase inhibitor that negatively regulates cell cycle activity and is transcip tionally down regulated by auxin, was strongly expressed in most pericycle cells but repressed in cells in the early stages of lateral root initiation, indicating that KRP2 is involved in blocking the G1/S transition in the pericycle cells not forming lateral roots (Himanen et al. 2002 and 2004). Other gene products involved in regulating the auxin signal in lateral root formation are NAC1, a transcriptional activator that is induced by auxin and promotes lateral root development (Xie et al. 2000), and SINAT5, a ubiquitin ligase that targets NAC1 for proteolysis, attenuating the auxin signal in pericycle cells not destined to become lateral roots (Xie et al. 2002). In addition, there ar e myriad other proteins and transcription factors involved in the complex signaling pathways in lateral root development. Modes of Action of Plant Growth-Promoting Bacteria Secretion of plant hormones Many plant growth-promoting bact eria have been shown to secrete plant horm ones, such as auxin, gibberellins, and cytoki nins (Table 5-1). These hormone s have been implicated as a 92

PAGE 93

cause of plant growth prom otion by bacteria sinc e the effects of bacterial inoculation on root architecture (such as increased root hair and lateral root number) are similar to the effects of exogenously applying a solution of these hormones (Tien et al. 1979). Most research on plant growth promotion by bacteria re sulting from the secretion of plant hormones has focused on auxin. It has been estimated that production of the auxin, indole-3-acetic acid (IAA), is common among bacteria in the rhizosphere, with approxim ately 80% being capable of synthesizing this hormone (Patten and Glick 1996). Although there is a lot of coro llary evidence for auxins role in bacterial plant growth promotion, such as the similar effects bacterial inoculation and applicatio n of IAA have on plant roots, it does not exclude the possibility that the bacteria may be producing other growth-promoting factors that act similarly to or activate the same plant developmental pathways as auxin. In order to confirm that auxin is responsible for changes in root morphology, genetic analysis is also required. For example, when investigating how Pseudomonas thivervalensis causes shorter, more branched root systems, mutant analysis involving the screening of Arabidopsis mutants showed that plants insensitive to auxin did not experience the same root system shortening as wild type plants, indica ting that auxin plays a role in this response (Persello-Cartieaux et al. 2001). A lternatively, other genetic analyses of auxins role in bacterial growth promotion have focused on the bacterial side, looking at mutants with reduced auxin production. Most plant growth-promoting bacter ia secrete auxin via the indole-3-pyruvate pathway, and the key enzyme in this pathwa y, indole-3-pyruvate de carboxylase, is the rate-limiting step (Lambrecht et al. 2000; Koga et al. 1994). This enzyme, encoded by the ipdC gene, is present in both K p342 and another plant growth-p romoting endophytic bacteria, Enterobacter cloacae P101 (Table 5-2). This gene has been the target for knocking out auxin 93

PAGE 94

synthesis in several plant growth-p ro moting bacteria. In the case of A. brasilense knocking out the ipdC gene reduces auxin production by 90%, resulti ng in a reduction of la teral root length and number, root hair formation, and primary r oot length inhibition compared to wild type strains (Barbieri and Galli 1993; D obbelaere et al. 1999). Even in ipdC mutant experiments, the mutants still result in some changes in root arch itecture compared to uninoculated plants. This observation is likely due to the fact that the ipdC mutation does not completely abolish auxin production since there are multiple pathways for a uxin synthesis that bacteria could use. For instance, an ipdC mutant of Pseudomonas putida still increased the numbe r of adventitious roots formed by mung bean cuttings compared to uninoc ulated controls (Patte n and Glick 2002). In addition, the effect of the ipdC mutation on auxin production in A. brasilense depends on the carbon source the bacteria are grown on, with muta nts grown on pyruvate, lactate, or fumarate not producing significantly different amounts of auxin than wild type strains, suggesting an alternate tryptophan-dependent pathway of auxin synthesis (Carreno-Lopez et al. 2000). Since there are multiple pathways for auxin synthesis (Bartel 1997), it has proven difficult to obtain bacterial mutants in which auxi n production has been completely abolished. As a result, the involvement of other yet uncharacterized grow th-promoting signals secreted by the bacteria cannot be discounted. As mentioned previously, plan t growth-promoting bacteria have also been shown to produce gibberellins and cytokinins. In the case of gibberellin, Fulchieri and associates (1993) found that putting GA3 on roots in concentrations similar to those produced by bacteria increased root growth in maize seedlings a nd that inoculation with different Azospirillum strains increased GA3 levels in maize roots. Application of GA3 to culture media also results in increased lateral root numbers in pearl millet plants (Tien et al. 1979). Although these results are intriguing, GA 94

PAGE 95

synthesis genes were not identified or knocked out in the Azo spirillum strain in order to directly demonstrate a role for gibberellin synthesis by the bacter ia in the observed growth response. In the case of cytokinin, which is commonly believed to have an inhibitory effect on lateral root formation, kinetin and trans-zea tin inhibit the formation of ne w lateral root primordia, but increase the elongation of existing lateral roots in rice (Debi et al. 2005). If plant growth-promoting bacteria were producing th is hormone, the elongation of existing root primordia may appear to be an increase in la teral root number when viewed by the naked-eye. In addition to secreting plant hormones, some plant growth-promoting bacteria are known to produce enzymes that interfere with the form ation of plant hormones. For example, ACC deaminase, an enzyme that lowers the concen tration of ethylene in plants by cleaving its precursor, 1-aminocyclopropane-1-carboxylate (ACC), is found in several plant growth-promoting bacteria (Glick 2005). Since ethylene is know n to inhibit root elongation (Abeles et al. 1992), bacteria that lower the levels of this hormone in the plant would be expected to result in plants with longer root systems. This role of ACC deam inase in altering root morphology has been supported by analysis of bacter ial mutants. Plants inoculated with ACC deaminase producing strains of P. putida have longer roots than uninoculated plants while mutant bacteria lacking ACC de aminase do not increase root le ngth (Hall et al. 1996). In addition, transferring the A CC deaminase gene into a P. fluorescens strain lacking this enzyme confers the ability to increase root length in canola while the wild type strain has no effect, suggesting that bacterial ACC deaminase lowers ethylene levels inside the plant, resulting in an increase in root length (Wa ng et al. 2000). Neither Kp342 nor E. cloacae P101 possess this enzyme (Table 5-2), so expressing this gene in these strains may further enhance their growth-promoting ability. 95

PAGE 96

Production of volatile compounds Some plant growth-promoting bacter ia elicit increased growth in Arabidopsis by emitting volatile compounds (Ryu et al. 2003). Bacillus subtilis GB03 and Bacillus amyloliquefacians IN937a increase total leaf surface area in Arabidopsis when grown together on divided Petri dishes that allow only volatile signals to pass between the plants and bacteria (Ryu et al. 2003). Gas chromatography-mass spectrometry (GC-MS) analysis of the volat ile compounds produced by these bacteria identified two compounds, 3-h ydroxy-2-butanone (acetoin) and 2,3-butanediol, that appear to elicit the growth response (Ryu et al. 2003). Exogenous application of 2,3-butanediol onto plants and inoculation with b acteria blocked in the synthesis of these two substances confirmed that they play a role in plant growth promotion (Ryu et al. 2003). E. cloacae P101 and E. coli K12 both possess an operon responsi ble for the production of these two compounds, but Kp342 appears to have only one ge ne from this operon (Table 5-2). Since acetoin and 2,3-butanediol are products of the butanediol fermentation pathway present in many enteric bacteria (Madigan and Martinko 2006), these compounds may not be unique to plant growth promotion. Making mineral nutrients more available to plants Phytases from plant growth-prom oting Bacillus strains degrade myo-inositol hexakisphosphate under low phosphate conditions, making phosphate available to plant roots from soil phytate, and Bacilli lack ing a functional phytase do not elicit the same growth response as wild type strains (Idriss et al. 2002). Neither Kp342 nor E. cloacae P101 possess this enzyme (Table 5-2), though it could be a candidate for being engineered into these strains to further enhance plant growth promotion. Phosphate ava ilability alters root architecture, with low phosphate conditions yielding higher numbers and densities of lateral roots (Lopez-Bucio et al. 2002; Williamson et al. 2001). In contrast to these results, Chevalier and associates (2003) found 96

PAGE 97

that phosphorus starvation reduced prim ary root length and latera l root number. Although there is still ambiguity and contradiction over the effect of phosphate on root architecture, these observations indicate the possibili ty that plant growth-promoting Bacilli may be altering root architecture by increasing the amoun t of phosphate available to plants. In addition to phosphate, nitrogen has also been shown to alter root architecture. Since many plant growth-promoting bact eria, including Kp342, are known to fix nitrogen, it is possible that nitrogen fixation may be involved in the modification of plant root systems by these bacteria. High sucrose to nitrogen ratios inhibi t lateral root initiation (Malamy and Ryan 2001). In addition, localized low nitrat e concentrations have been s hown to increase lateral root formation while higher concentrations increase th e rate of lateral root elongation (Zhang et al. 1999). Therefore, growth-promoting bacterial strains could also enhance growth by making nutrients more available to their plant hosts. Secretion of other compounds In addition to plant hormones and nutrients, pl ant growth-promoting ba cteria m ay also be secreting other novel compounds that could be eliciting increased late ral root numbers and growth responses. For instance, the antibiotic 2,4-diacetylphlorogluci nol (DAPG) produced by a strain of Pseudomonas fluorescens has been shown to increase root length and weight and transiently increase lateral root formation in pea plants, indicating this molecule may act similarly to plant hormones (De Leij et al. 2002 ). Another example of a compound that acts similarly to a plant hormone is affinin, an alkamide that occurs naturally in plants (Ramirez-Chavez et al. 2004). The exogenous application of this compound results in increased lateral root density and enhanced primary root and root hair growth in low concentrations (Ramirez-Chavez et al. 2004). Alth ough these effects are similar to those of auxin, experiments using auxin-inducible genes fused to a -glucuronidase (GUS) repor ter and auxin-resistant 97

PAGE 98

mutants have dem onstrated that affinins mode of action is independent of auxin (Ramirez-Chavez et al. 2004). Amidenin, anothe r alkamide that is produced by the fungus Amycolatopsis sp., has been shown to promote growth in rice seedlings (Kanbe et al. 1993). Therefore, the presence of ot her plant growth-promoting com pounds in bacteria cannot be discounted, even when the plant growth response is similar to that seen in response to plant hormones like auxin. Results Increased Lateral Root Number in Arabidopsis thaliana Kp342 was first observed to increase lateral root num ber during an experiment to determine its route of entry in to plants, which was hypothesized to be through cracks formed at the base of lateral roots. To test th is hypothesis, Kp342 was inoculated onto an Arabidopsis mutant, xbat 32, which has a reduced number of lateral roots (Nodzon et al. 2004), to see if the bacteria colonized the mutant to a lesser exte nt than wild type pl ants. However, after inoculation, the number of la teral roots on inoculated xbat 32 mutants was increased relative to uninoculated controls, nearly re scuing the mutant phenotype (Figur e 5-1b). In addition, the wild type plants also displayed an increase in lateral root number (Figure 5-1a, b). Once observed, this lateral root increase by K p342 indicated a possible cause of this bacteriums nitrogen fixation independent growth promotion. Lateral Root Increase Not Due to Nitrogen Fixation Before m aking any attempts to identify the la teral root-increasing si gnal produced by the bacteria, it was necessary to firs t confirm that this growth res ponse was not a result of nitrogen fixation, since changes in nitrate concentrations are known to alter lateral root development. To see if nitrogen fixation by Kp342 was responsible for its lateral root increasing phenotype, a nifH mutant of the bacteria which does not fix nitr ogen in plant tissue (In iguez et al. 2004) was 98

PAGE 99

inoculated onto Arabidopsis and found to increase lateral root num bers compared to uninoculated plants (Figure 5-2). Al though the plants inoculated with the nifH mutant showed a slightly reduced number of latera l roots compared to those inocul ated with wild type Kp342, this difference was small and possibly due to enhanced nutrition provided to the plants in the form of fixed nitrogen produced by Kp342. As a result, it was concluded that the primary cause of the lateral root increase wa s not nitrogen fixation. Lateral Root Increasing Phenotype is Strain Specific In addition to determ ining if the lateral root increase was not due to of nitrogen fixation, it was also important to determine if this trait is common among enteric bacteria, or specific to Kp342 and other plant growth-promoting bacteria In order to determine this, Kp342 was compared with other enteric bacteria, including E. coli K12 and K. pneumoniae type strain 13883 (Kp13883) obtained from the American Type Culture Collection (ATCC). E. coli K12 was used to test if other enteric bacteria possess the ge nes required for eliciting a lateral root increase and the type strain of K. pneumoniae was used to test if this trait is common to the Klebsiella genus. The results of these inoculati ons demonstrated Kp342 increased lateral root number more than Kp13883, while E. coli inoculated plants were not statistically different from uninoculated plants (Figure 5-3a). Although, Kp13883 s lightly increased late ral root numbers to a certain extent, it was significantly less than Kp342. In addition, Kp342 inoculation was also comp ared to inoculation with another plant growth-promoting bacteria, E. cloacae P101 (Riggs et al. 2001), in or der to determine if this trait is present in other enteric pl ant growth promoters. The results from this experiment demonstrated that P101 does indeed increase lateral root number to the same extent as Kp342 (Figure 5-3b). These results indicate that the la teral root increasing pheno type is not a ubiquitous trait among other enteric bacteria, but is more commonly found in strains isolated from plant 99

PAGE 100

tissue, possibly developed by the ba cteria as an adaptation to livi ng in association w ith plants. Although the type strain, Kp13883, sli ghtly increased lateral root number, the intensity of this response is specific to plant growth-promoting strains like Kp342 and P101. The difference in the intensity of the response may be due to di fferent regulation of growth-promoting genes in these bacteria or multiple plant growth-promoti ng pathways present in Kp342 and P101 that are absent in Kp13883. Lateral Root Promotion Due to Secreted Diffusible Product Another basic question about the growth-p rom oting phenotype was whether Kp342 elicits the lateral root increase as a response to a secreted and di ffusible compound produced by the plant or whether the bacteria must be present within plant tissue to elicit a response. To test if the bacteria were secreting a diffusible product responsible for increasi ng lateral root number, Kp342 was inoculated onto plant roots covered by a 0.2 m membrane. This membrane served to prevent the bacteria from coming into direct contact with the roots while still allowing chemical signals to pass between the plants and the bacteria (Figure 5-4a). As controls, plants were also treated with only the membrane or with Kp342 inoculated under the membrane. Since Kp342 increased lateral root number to the same extent both under and on top of the membrane, it was concluded that it does secrete a product that can diffuse th rough the membrane to increase lateral root formation on the plants (Figure 5-4b). The results from this experiment also showed that the membrane itself altered lateral root development. Uninoculated plants under the membrane had higher numbers of lateral roots than uninoculated plants without a membrane (Figure 5-4b). In addition, inoculated membrane -covered plants had lower lateral root numbers than plants treated with Kp342 alone (Figure 5-4b). The eff ects the membrane had on root numbers were probably due to the fact that the membrane limited the amount of light and the gas exchange the root systems were exposed to. 100

PAGE 101

Secretion of Lateral Root Promot ing Compound is Plant Inducible In the first attempt to isolate growth-promoting compounds secreted by Kp342, concentrated extracts from culture supernatants of overnight broths were assayed for lateral root increasing activity. Initially, the supernatant from Kp342 grown on Luria-Bertani (LB) broth was filter sterilized and fractionated on C18 resin with increasing concentrations of methanol to isolate lateral root promoting compounds. These fractions were then concentrated on a rotary evaporator so that the concentration of a ny growth-promoting compounds would be high enough to elicit a growth response, since previous experiments showed unconcentrated Kp342 culture supernatants do not elicit a growth response. These LB supernatant fractions were assayed on plants to check for lateral root induction compared to Kp342 inoculation. All the fractions appeared to increase lateral root numbers to some degree but a subsequent experiment yielded conflicting and variable results, demonstrating that more contro lled bacterial culture conditions were required to reliably elicit the production of lateral root inducing compounds. In order to analyze a more controlled growth environmen t, another experiment was performed with murashige and skoog (MS) media, a defined plant growth media that Kp342 grows on after inoculation. Use of this defined media simplif ied the composition of the culture extract for easier analysis. When plants were treated with the concentrated MS culture supernatant, they did not increase lateral root numbers compared to pl ants treated with concentrated media (Figure 5-5). Since MS is a defined media, the possibil ity that it may not contai n the substrate required to produce growth-promoting compounds was consid ered. The bacteria may require certain substrates or signals from the host plant in or der to produce compounds that elicit a pl ant growth response. 101

PAGE 102

Lateral Root Promotion May Involve Secretion of a Plant Hormone When lateral root promotion was first obs erved in response to Kp342 inoculation, the immediate candidate suspected to be invol ved was auxin. The Kp342 genome has been sequenced and the annotation has identified the presence of an ipdC auxin synthesis gene. The observations that Kp342 possesses this auxin sy nthesis gene and that another strain of K. pneumoniae has been shown to secrete auxin (El-Khawas and Adachi 1999), coupled with the fact that exogenous application of auxin is known to increase lateral roots on plants, further substantiates the hypothesis that Kp342 may be producing auxin to elic it this lateral root promotion. In an attempt to confirm if auxi n and other plant hormones are involved in this response, Arabidopsis mutants described as being insensitive to certain plant hormones were tested to see if any of these mutations abolis hed the plants response to Kp342 (Table 5-3). When assayed, most of these mutants res ponded to Kp342 inoculation (Figure 5-6). Unfortunately, the response of these mutants to exogenous application of the hormone they are insensitive to has not been studied in regards to lateral root formation. Most were only examined for reduced inhibition of primary root length. As a result these data must be interpreted with caution, since these genes may not be involved to lateral root formation. For example, the axr2 1 mutant is insensitive to auxin in that it s primary root elongates in the presence of high auxin concentrations, but it is se nsitive to auxins effect on late ral root formation, forming more lateral roots when trea ted with auxin (Knee and Hangarter 1996). Therefore, these hormones could not be discounted as being involved in lateral root promo tion. In order to obtain more conclusive data, mutants insensitive to auxin that do not significantly increase lateral root number in response to exogenous hormone application were select ed for further experiments. One of these mutants, nph41/ arf19 4 (Wilmoth et al. 2005), was tested and did not increase 102

PAGE 103

lateral num ber after Kp342 inoculat ion, indicating that auxin may be involved in the lateral root response (Figure 5-6). Exogenous Application of Auxin Partially Mimics Lateral Root Promotion by Kp342 W ith the difficulty in extracting growth-p romoting compounds from Kp342 supernatants and the uncertainty of the conclusions drawn fr om lateral root promotion on auxin insensitive mutants, Kp342 inoculation was compared directly to exogenous application of auxin on wild type Col-0 Arabidopsis plants. The root systems of Kp342 in oculated plants were compared to their auxin treated counterparts in regards to both latera l root number and primary root length. When looking at the number of lateral roots pr oduced by these plants, both Kp342 inoculation and treatment with 0.1 M IAA increased lateral root number to the same extent (Figure 5-7a). In contrast, exogenously applie d auxin significantly inhibited primary root length while Kp342 inoculated plants did not significantly differ from uninoculated nega tive controls in that regard (Figure 5-7b). Therefore, exogenous auxin application partially, but not completely, mimics the effects of Kp342 inoculation. Role of ipdC in Auxin Production Initially, attempts were made to delete the ipdC gene in Kp342. However, the construction of in-frame deletions in Kp342 proved to be prob lematic due to a variety of problems. These problems include the presence of a wide range of antibiotic resist ance genes commonly used in gene deletion protocols and the in efficient transformation efficien cy of the Kp342 strain. As a result, further genetic work on determining the role of the ipdC gene and IAA production on lateral root promotion was carried out in E. cloacae P101. As P101 also increases lateral root numbers in Arabidopsis similar to Kp342 (Figure 5-3b) and its genome annotation revealed the presence of the ipdC gene, it served as an ideal substitute. 103

PAGE 104

An in-fram e deletion of the ipdC gene in P101 was constructed using a modified Red recombinase-mediated homologous recombinati on method (Jantama et al. 2008). Briefly, the ipdC gene was first replaced with an antibioti c resistance cassette via a double-crossover homologous recombination, which was later removed by an additional double-crossover recombination event with a deoxyribonucleic acid (DNA) fragment containing a clean, in-frame deletion of the gene. When th e culture supernatant of the ipdC mutant was assayed, IAA production was lowered to 1 g/mL, nearly 4-fold lower than th e wild type P101 strain (Figure 5-8). This loss of auxin production in the ipdC mutant supported its pr oposed role in auxin production by P101. To further confirm the role of ipdC in IAA production, the ipdC mutant was complemented with a plasmid-born copy of ipdC on pHLT14. When assayed, IAA production in the complemented ipdC strain was restored, with a concentration eight times higher than the wild type strain (Figure 5-8). The fact that the complemented mutant produced eight times more auxin than wild type P101 is probably due to the higher copy number of the plasmid relative to the single copy number of the ipdC gene found on the chromosome. Regardless, restoration of auxin production in ipdC by pHLT14 further confirmed the ipdC genes role in auxin production in P101. pHLT14 was constructed to include the par locus for plasmid stabil ity so that it could be maintained by bacteria inside plant tissue in the ab sence of selection pressure. To verify stability of the plasmid in the absence of selection pr essure, the numbers of antibiotic resistant and sensitive cells were counted af ter two subculturings in LB me dia without antib iotics. pHLT14 cultures had the same number of an tibiotic resistant and sensitive cells, indicating the bacteria had retained the plasmid (Figure 5-9). In co mparison, cultures carrying a complement plasmid lacking the par locus had significantly fewer antibiotic resistant cells than non-resistant cells, 104

PAGE 105

indicating the plasm id was lost (Figure 5-9). Therefore, the par locus served its intended purpose of maintaining pHLT14 in ipdC in the absence of selection pressure. ipdC Mutants on Plants When the ipdC m utant in P101 was included in experiments on Arabidopsis plants, no differences were seen in late ral root numbers between any tr eatment, even uninoculated and P101 wild type treated plants. With the loss of this phenotype, the role of ipdC catalyzed auxin production in lateral root promotion previously observed in response to bacterial inoculation could not be verified. In th e event that loss of the phenotype was due to a change in P101, Kp342 was retested on plants sin ce the possibility of two bact erial strains losing the same phenotype was unlikely. Neither wild type P101 nor Kp342 increased late ral root numbers when inoculated on Arabidopsis Col-0 as previously observed, indi cating loss of lateral root induction was not due to a change in th e bacterium and, instead, might be due to an artifact in plant culturing. Since the loss of the lateral root inducing phenot ype was roughly preceded by the purchase of new MS media, the possibility that the media had been contaminated was considered; therefore, new MS media was procured. Even when new media was used, plants still did not increase lateral root numbers in response to bacterial inoculation relative to uninoculated plants. Because the lateral root response was first observed in xbat 32 mutants and the intensity of the response was more dramatic in this mutant, experiments were repeated on this Arabidopsis line, but again, inoculated and uninoculated plants did not differ from each other. In all new cases, uninoculated plants were seen to have more lateral roots than previously observed. Due to the inability to restore lateral root promotion by changing the strain, plant culture media, or Arabidopsis line, it is believed the loss of this phenotype is due to gradual changes and improvement in handling of the plants over time. 105

PAGE 106

Discussion The endophytic bacterium, K. pneumoniae 342, is of interest as a model endophyte due to its ability to colonize the interi or of plant tissues and increase plant growth yield. While this endophyte has been found to provide fixed nitrogen to plants and resu lts in a generalized enhancement in plant growth, further studies on Kp342 found that it could enhance lateral root growth in A. thaliana by a mechanism independent of nitrogen fixation. This lateral root-enhancing phenotype was only observed in other plant growth-promoting bacteria and deemed to be an ideal assay for identify ing the mechanism of Kp342s nitrogen fixation independent growth-promoting mechanism. As outlined earlier in this chapter, plant growth-promoting bacteria posse ss a variety of mechanisms that enhance plant growth. Attempts to isolate growth-promoting compounds fr om Kp342 culture supern atants failed, so the genome sequences of Kp342 and another endophytic bacterium, E. cloacae P101, were examined for genes associated with these mechanisms. As a result of this examination, auxin production by indole-pyruvate deca rboxylase was deemed to be an ideal candidate for further study of these two endophytes growth-promoting effect on plants. Initial attempts to determine if auxin or another plant hormone pl ays a role in Kp342s lateral root promotion involved assaying a wide array of hormone insensitive mutants for their response to bacterial inoculation. It was expect ed that if Kp342 were el iciting a growth response by secreting one of these hormones, that response would be diminished or eliminated in a mutant insensitive the hormone. In nearly every muta nt tested, Kp342 increased the number of lateral roots on plants. While this observation initiall y indicated that the growth-promoting response was due to a factor other than one of the hormones examined, the complexity of hormone signaling in Arabidopsis confounds the interpreta tion of these results. Phytohormones play multiple roles in plant growth and development. As a result, there are many different receptors 106

PAGE 107

that respond to phytohormones in different signaling pathways, l eading to different effects on plant b iology. Therefore, a mutant that is insensitive to one effect of a hormone may still respond to that hormone in other developmental pathways. This fact was demonstrated by Knee and Hangarter (1996), who found seve ral supposedly auxin insensitive Arabidopsis mutants that still increased lateral root formation when the hor mone was applied to plant growth media. As most of the hormone insensitive mutants used in this study were originally assayed for insensitivity based on root elonga tion rather than lateral root response to hormone application, the increase in lateral root numbers seen on thes e mutants is inconclusive Therefore, mutants specifically insensitive to horm one application in regards to lateral root production were selected. The search for such mutants proved to be difficult owing to the essential and redundant nature of genes involved in lateral root ini tiation (Malamy and Ryan 2001). One of the few lateral root specific mutants obtained was nph41/arf194 (Wilmoth et al. 2005). This auxin insensitive mutant did not increase lateral root numbers in response to Kp342, indicating that auxin production by Kp342 could be th e cause of lateral ro ot promotion. It should be noted that, given the severity of lateral root initiation mutants, it is possible that no growth-promoting compound could rescue it. Therefore, more eviden ce was required to confirm the role of auxin production in Kp342-mediated lateral root promotion. Interestingly, while Kp342 increased lateral root number simila r to exogenous auxin application, it did not in hibit primary root length as was observed upon auxin treatment. One possible reason for the partial mimicry of auxin a pplication is that Kp342 s ecrets auxin at such a low level that it promotes lateral root branch ing, but does not inhibit root elongation. Because Kp342 is present in and on the plan ts, it could produce a constant s ource of auxin so that even low levels produced by the bacteria could increas e lateral root growth without reaching an 107

PAGE 108

inhibitory concentration. Exogenous applicatio n of IAA involves treating plants with higher concentrations in the m edia, since low concen trations may breakdown or be metabolized by the plant over time. As a result, exogenous IAA c ould not completely simulate auxin production by Kp342, so differences between bacterial inocula tion and exogenous auxin application are not unexpected. Therefore, auxin remains a likely candidate for Kp342s la teral root promoting effect on plants. In order to confirm auxins ro le in lateral root promoti on, experiments involving strains abolished in auxin production were needed. Analysis of the Kp342 genome indicated the presence of the ipdC auxin synthesis gene, which was sele cted as the target for knocking out auxin production in this strai n. Unfortunately, genetic manipulations in Kp342 proved to be problematic. The strain possesses resistances to multiple antibiotics commonly used as selectable markers in molecular biology, includ ing ampicillin, tetracycline, and chloramphenicol (Appendix A), so special care had to be taken in selecting plasmid vectors to use in gene knockout approaches. The approach selected to delete the ipdC gene is a Red recombinase-mediated homologous recombination method (Jantama et al. 2008). Unfortunately, Kp342 was unable to be transformed with a plasmi d carrying the Red recombinase genes. This problem was initially believed to be due to plasmid incompatibly with one of Kp342s native plasmids, but several attempts to transform the bacteria with the Red recombinase genes on other plasmids with different origins of replication al so failed. Further inves tigation determined that the problem lies with the arabi nose induction and Red recombinase genes themselves. The entire set of Red recombinase genes plus the upstrea m arabinose induction genes were cloned into pCR4.1, a vector previously transformed into Kp342, but transformation of the recombinase carrying pCR4 still yielded no transformants. Therefore, it appears that either the Red 108

PAGE 109

recom binase or arabinose induction genes themselv es are incompatible with Kp342. In addition to this difficulty, Kp342 also has a low transfor mation efficiency, which is a common trait in members of the Klebsiella genus (Regue et al. 1992; Fournet-Fa yard et al. 1995). To ensure that the screening of transformants for mutants in any gene deletion method was successful, high transformation efficiencies would be needed. Therefore, further at tempts to make gene deletions in Kp342 were halted in favor of working with a more gene tically tractable bacterium. Because P101 enhances latera l root growth similar to Kp342 and also possessed the ipdC gene, it was selected for genetic studies on the role of auxin production in the lateral root response. P101 proved to be more amenable genetic manipulation than Kp342 and an ipdC mutant in this strain was constructed with rela tive ease. This mutant was deficient in auxin production compared to the wild type st rain, and that production was restored upon complementation with a plasmid-born copy of the gene. Unfortunately, once the ipdC mutant was constructed, the lateral root growth respons e seen on plants had di minished and significant differences in lateral root numbers between treatments were no longer observed, not even between uninoculated and wild type P101 or Kp342 treated plan ts. The sudden loss of this previously reliable and reproducib le lateral root promoting phenot ype was baffling, and repeated troubleshooting measures were unable to identif y the source of the problem. Because the phenotype was no longer observed in response to either P101 or Kp342, loss of lateral root induction due to evolution of the bacteria was discounted, as it is unlikely that two different strains would lose the same phe notype. Therefore, problems stemming from plant culturing were considered. The loss of th e lateral root induc tion roughly followed the purchase of new MS media, so the possibility that the media had been contaminated by a synthetic auxin was considered. If auxin were present in the media, even uninoculated plants would be stimulated to 109

PAGE 110

produce lateral roots, confounding any results compar ing treated and untreated plants. But even on new m edia, lateral root inducti on was still not observed. One fi nal possibility for the loss of lateral root promotion was handling of the plants and human error. Working with Arabidopsis can be difficult owing to the small and delicate na ture of the plants root system, but over time handling of the plants became easie r with practice and experience. It is possible that plants in earlier experiments were stre ssed by handling, and such stre ss could have altered root development in the plant. After handling of the plants became more routine, it is possible that they were less stressed and root development less responsive to b acterial inoculation. Therefore, the changes in the lateral root numbers produced by plants may be an artifact of gradual changes in handling that occurred over time. Regardless of the loss of the late ral root promoting phenotype, th e work in this chapter also resulted in the construction of plasmids carry ing known plant growth-promoting genes that could have potential uses in future studies. The ipdC gene, known to be involved in growth promotion by several plant-associated st rains, was cloned into a low copy number plasmid that was modified with a plasmid stability locus, allowing it to be maintained by bacteria in plant tissues in the absence of selection pre ssure. In addition, though not utili zed in the current experiments, another plasmid carrying the acdS gene encoding ACC deaminase, a known plant growth-promoting factor, was also constructed. This acdS gene was placed immediately downstream of the ipdC gene in the original complement plasmid so that both genes would be under control of the same promoter. This new plasmid construct with two known plant growth-promoting genes could be transformed into other plan t-associated bacteria and potentially increase positive effects on plant gr owth. In addition, since ACC deaminase acts by breaking down the precursor of the plant hormone ethylene, whic h is known to act in plant 110

PAGE 111

defense pathways (Glick 2005; Inig uez et al. 2005; Pieterse et al 1998), this plasm id may also lower plant defense responses to endophytic bact eria, resulting in increased colonization and a concomitant increase in growth promotion. Ther efore, the plasmids constructed in this study have potential for future studies in plant growth-promoting bacteria. Materials and Methods Plant Culturing A. thaliana seeds were obtained from the Arabidopsis Biological Resource Center, w ith the exception of nph41/arf9 4, which was provided by Jason Reed at the University of North Carolina, and xbat 32, which was provided by Wen-Yuan Sun at the University of Florida. Arabidopsis seeds were surface sterilized by s ubmersion in 70% ethanol for 5 minutes (min), rinsing 4 times in sterile water, submersion in 10% bleach for 20 min, and rinsing 4 more times in sterile water. Seeds were germinated on half strength MS media (Sigma) with 10 g L-1 sucrose and 0.9% noble agar and cold treated at 4oC in the dark for 3 days. After cold treatment, seeds were transferred to a 22oC growth chamber with an 11 hour daylight cycle. Plates with seeds were oriented vertically in the chamber so plant roots would grow along the surface of the agar. After 8 days, seedlings we re transferred to MS media w ithout sucrose and treated 1 day later. Lateral root number and primary root lengt h were counted 9 days after inoculation. All experiments assaying late ral root growth used 30 plants per treatment. Bacterial Strains and Inoculum Preparation Bacterial s trains used were K. pneumoniae 342, K. pneumoniae ATCC13883, E. cloacae P101, and E. coli K12. All bacteria were cultured on Luria-Bertani (LB) medium at 37oC, with the exception of strain s carrying pLOI341, which were cultured at 30oC. Inocula were prepared by scraping cells from LB agar plates and suspen ding them in sterile phosphate buffered saline (PBS). Bacteria were then diluted to 104 colony forming units (CFU)/mL and 10 L were 111

PAGE 112

inocula ted per plant. To test for a secreted product, a sterile membrane filter with a 0.2 m pore size was placed over the root system of plants an d bacteria were inoculated on top. As a positive control, bacterial inoculum was applied underneath the membrane. Uninoculated plants with just the membrane were used as a negative control. Extraction of Compounds from Culture Supernatant Kp342 was grown in 4 L of LB brot h or MS broth overnight at 28oC with shaking. Culture supernatants were collected after centrifuging at 5000 rpm for 30 min. Additional bacteria were removed from the supernatan t by filtering through a 0.2 m membrane. The LB supernatant was fractionated on C18 resin with increasing con centrations of methanol. Briefly, the LB supernatant was poured over C18 and solutes we re eluted with 500 mL each of 25, 50, 75, and 100% methanol. The MS supernatant and LB meth anol fractions were concentrated on a Buchi Rotavapor at 90 rpm, with the water bath set at 50oC and the condenser cool ed with water chilled to 4oC. The volume of each sample was concen trated down to 5 mL and stored at -80oC until use in plant assays. Comparison of Bacterial Inoculation to Kp342 Supernatant Extracts and Exogenous IAA Col-0 Arabidopsis plants were cultivated and inoculated as described previously. For com parison to concentrated culture supernatants, plants were treated with 10 L of supernatant extracts instead of bacterial suspensions. Fo r comparison to exogenous application of auxin, varying amounts of IAA were added to MS me dia from a 1 mM IAA stock solution in 100% ethanol immediately prior to pouring plates. The resulting IAA con centrations in the media were 0.1, 1, and 10 M IAA. MS agar with plain ethanol added in the same volume as the IAA stocks was used as a negative media control. Plants were transferred to fresh media 9 days after germination, and were either inoculated with Kp342 as described previously or transferred to 112

PAGE 113

plates with IAA containing m edia. Lateral root number and prim ary root length were measured 9 days after inoculation. Construction of ipdC In-Frame Deletion in E. cloacae P101 P101 DNA was extracted using the FastDNA Ki t. All plasmids were extracted using a QIAprep Spin miniprep kit. The ipdC gene in E. cloacae P101 was knocked out using a method involving two double-crossover hom ologous recombinations with linear, polymerase chain reaction (PCR) -generated construc ts (Jantama et al. 2008). The ipdC gene plus 300 basepairs (bp) up and downstream was PCR amplified with Ex Taq (Takara) using primers A_P101_ipdC and D_P101_ipdC (Table 5-4). Reactions contained 5 L of 10x Ex Taq buffer, 200 M of each dNTP, 500 ng of P101 DNA, 2 M of each primer, and 1.25 units of Ex Taq in 50 L. Cycle conditions were as follows: 94oC for 30 seconds (s), 30 cycles of 94oC for 10 s, 55oC for 30 s, and 72oC for 2 min 40 s, followed by a 5 min final extension at 72oC. The resulting PCR product was cloned into the pCR4.1 TOPO cloning vector yielding pHLT4. An inverted PCR of the pHLT4 plasmid was performed with Phusion polym erase (New England Biolabs) using primers F2-JMP101ipdC and R2-JMP101ipdC (Table 5-4). The reaction contained 10 L of 5x HF buffer, 200 M of each dNTP, 500 ng of pHLT4, 2 M of each primer, and 2 units of Phusion polymerase in a final volume of 50 L. The reaction was cycled under the following conditions: 98oC for 30 s, 30 cycles of 98oC for 10 s, 58oC for 30 s, and 72oC for 1 minute 15 s, followed by a 5 min final extension at 72oC. The primers in this reaction were designed to include the start and stop codons of the ipdC gene and yield a product contai ning the plasmid flanked by the up and downstream regions of ipdC This product was blunt ligated with a cat/sacB cassette excised from pLOI4162 (Jantama et al. 2008). The result ing plasmid, pHLT6, was used as a template for a PCR with primers A_P101_ipdC and D_P101_i pdC as described above, yielding a linear 113

PAGE 114

fragm ent containing th e upand downstream ipdC regions with the cat/sacB cassette inserted in place of the gene. This linear construct was electroporated into P101 cells expressing Red recombinase from plasmid pLOI3421 (Wood et al. 2005). Red recombinase expression was under control of the arabinose pr omoter and induced in P101 by growth in media containing 5% arabinose. Insertions into ipdC were selected by screening tr ansformants for chloramphenicol resistance and sucrose sensitivity. To remove the selectable marker, the cat/sacB cassette was digested out of pHLT6 using PacI. The plasmid was then religated, leaving a short scar sequence with stop codons in all 6 reading frames. This new plasmid, pHLT7, was used as template in another PCR with primers A_P101_ipdC and D_P 101_ipdC to generate a new linear fragment with the up and downstream regions of ipdC flanking the stop codon scar sequence. This fragment was electroporated into P101 ipdC::cat/sacB expressing Red recombinase. Resulting transformants were selected for sucrose resistan ce and screened for chloramphenicol sensitivity, insuring loss of the cat/sacB cassette in the genome and a non-polar deletion of the ipdC gene. The red recombinase carrying plasmid, pLOI3421, wh ich has a temperature sensitive origin of replication, was cured from th e strain by growth at 37oC overnight, yielding E. cloacae P101 ipdC Complementation of P101 ipdC The ipdC mutant of P101 was complem ented with a plasmid-born copy of the ipdC gene. The vector selected for complementation was pAYC184 (Chang and Cohen 1978). To construct the complement plasmid, the ipdC gene was amplified using primers A_P101_ipdC and D_P101_ipdC. These primers amplified 300 bp upstream of the gene to ensure the native promoter was obtained. Each pr imer contains a BamHI restrict ion site for easy cloning. The ipdC PCR product was digested with BamHI and then ligated into the BamHI site in the 114

PAGE 115

tetracycline gene of pACYC184, yielding pHLT12 (Figure 5-10a). Transfor mants were screened for chloramphenicol resistance and tetracycline sensitivity to insure insertion of the ipdC gene into the plasmid. In order for the plasmid to be maintained in plant tissue in the absence of selection pressure, the partitioning genes for plasmid stability from pTR102 were insert ed (Weinstein et al. 1992). To insert the partitioning genes into pHLT12, the par locus was first amplified from pTR102 with Phusion polymerase using primers par1 sa and par2sa (Table 5-4). The resulting PCR product was blunt ligated into the ScaI site in the ampic illin gene of pHLT4, yielding pHLT13. Primers tpkanpar1 and tpkanpar2 (Table 5-4) were then designed to amplify the region of pHLT13 that includes the inserted par locus and the adjacent kanamycin resistance gene. This step was performed to obtain a par kan cassette that could be used as a selectable marker for the insertion of the par locus. The par kan cassette from pHLT13 was amplified with Phusion polymerase and used as the inse rt in a blunt ligation into th e ScaI site of pHLT12, yielding pHLT14 (Figure 5-10b). Transformants were screened for kanamycin resistance. Determination of Plasmid Stability Stability of pHLT14 wa s determined by growing ipdC with pHLT14 in the absence selection pressure. ipdC with pHLT12, which lacks the par locus for plasmid stability, was used as a negative control. ipdC pHLT14 was streaked out on LB plates with 50 g/mL kanamycin and ipdC pHLT12 was streaked out on LB with 20 g/mL chloramphenicol. One colony from each plate was used to inoculate 3 mL of plain LB Cultures were grown at 37oC with shaking for 24 hours and 30 L were taken to subculture into new 3 mL broths. After 2 subculturings, dilution spread -plating was performed on plain LB and LB with the appropriate 115

PAGE 116

antib iotic. Colonies on each media type were co unted to determine the number of CFU/mL in each culture that were still resistant, a nd therefore, still ca rried the plasmid. Auxin Assay P101, ipdC and ipdC pHLT14 cultures were grown at 30oC in M9 minimal media supplemented with 100 g/mL tryptophan. At 12 hours intervals during the growth curve, culture OD620 was measured. Two milliliters of culture supernatant were collected by centrifugation at 13,000 rpm for 1 minute and filter sterilized with a 0.2 m membrane to remove remaining cells. Supernatants were assayed for IAA using Salkowskis colorimetric assay (Gordon and Weber 1951). The Salkowski r eagent was composed of 1 mL of 0.5 M FeCl3 and 50 mL of 35% HClO4. For the assay, 0.5 mL of Salkowsk i reagent was added to 1 mL of culture supernatant, vortexed, a nd incubated at room temperature for 20 minutes. Absorbance was measured at 530 nm on a Shimadzu UV-160 sp ectrophotometer. Concentration of IAA in culture supernatants was determined by comp arison to a standard curve of 1, 2.5, 5, 7.5, 10, 15, 20, 30, 50, and 75 g/mL IAA diluted in M9 media with 100 g/mL of tryptophan. Insertion of ACC Deam inase into pHLT14 ACC dea minase was added to pHLT14 by inserting the acdS gene from P. putida UW4 immediately upstream of ipdC so both genes would be under control of the same promoter. To do this, primers ACCd-FXhoI and ACCd-RSbfI (Table 5-4) were designed to amplify the coding region of acdS, including the ribosome binding site. Primers pACYCipdcRXhoI-1 and ACCd-RSbfI (Table 5-4) were designed to amplify pHLT14, excluding the upstream region of ipdC to eliminate any transcripti onal terminators downstream of ipdC In order for the acdS gene to insert in the same orientation as ipdC the forward and reverse pr imers in each pair were designed to include an XhoI or Sb fI restriction site, ensuring that the gene could only insert in 116

PAGE 117

one orientation. The pHLT14 and acdS PCR products were digested with XhoI and SbfI and liga ted together, resulting in plasmid pHLT16 (Figure 5-10c). 117

PAGE 118

A B Figure 5-1. Lateral root pr om otion by Kp342. A) Uninoculated and Kp342 inoculated Arabidopsis Col-0. B) Lateral root numbers on Kp342 inoculated Arabidopsis compared to uninoculated plants in Col-0 and xbat 32. Plants were inoculated with 10 cells 9 days after germination and r oots counted 9 days post inoculation. Statistical difference denot ed by different letters. Figure 5-2. Effect of nitrogen fi xation on lateral root prom otion. Lateral roots per cm primary root on Arabidopsis inoculated with wild type Kp342 or a nifH mutant compared to uninoculated plants. Plants were inoculat ed with 10 cells 9 days after germination and roots counted 9 days post inoculation. Statistical difference denoted by different letters. 118

PAGE 119

A B Figure 5-3. Bacterial strain specificity of lateral root prom otion. A) Lateral root per cm primary root on Kp342 inoculated plants compared to plants inoculated with Kp13883 and E. coli K12. B) Lateral root per cm primary root on Kp342 inoculated plants compared to E. cloacae P101 inoculation. Arabidopsis plants were inoculated with 10 cells 9 days after germination and roots counted 9 days post inoculation. Statistical difference denoted by different letters. 119

PAGE 120

A B Figure 5-4. Lateral root promo tion due to a secreted product. A) Diagram of m embrane experiment. B) Lateral roots per cm primary root on Arabidopsis with Kp342 inoculated directly on roots or with a 0.2 m membrane between the roots and the bacteria. Controls include uninoculated pl ants with and without the membrane and Kp342 inoculated under the membrane. Plants were inoculated with 10 cells 9 days after germination and roots counted 9 days post inoculation. Statistical difference denoted by different letters. 120

PAGE 121

Figure 5-5. Lateral root promoti on in response to culture supern atants. Lateral roots per cm primary root on Arabidopsis treated with concentrated Kp342 culture supernatant or uninoculated MS media. Kp342 inoculated and uninoculated plants used as positive and negative controls. Plants were tr eated 9 days after germination and roots counted 9 days post inoculation. Statisti cal difference denoted by different letters. Figure 5-6. Response of hormone insensitive Arabidopsis m utants to Kp342. Lateral roots per cm primary root on uninoculated (blue) a nd Kp342 inoculated (red) plants. Plants were inoculated with 10 cells 9 days afte r germination and roots counted 9 days post inoculation. Asterisks indicat e inoculated plants were st atistically different from uninoculated controls. 121

PAGE 122

A B Figure 5-7. Comparison of Kp342 in oculation to exogenous IAA treat m ent. A) Lateral roots per cm primary root and B) primary r oot length on Kp342 inoculated Col-0 Arabidopsis compared to plants grown on media supplemented with 0.1, 1, and 10 M IAA. Plants were treated 9 days after germ ination and roots counted 9 days post inoculation. Statistical differe nce denoted by different letters. 122

PAGE 123

Figure 5-8. Auxin production in P 101 strains. Concentration of indole-3-acetic acid in culture supernatants of P101, a ipdC mutant, and the ipdC mutant complemented with a copy of the ipdC gene carried on plasmid pHLT14. Bacteria were cultured in M9 media supplemented with 100 g/mL tryptophan and supernatants were collected 46 hours after inoculation. IAA concentrati on adjusted for differences in culture OD620 in order to correct for differences in growth between cultures. Statistical difference denoted by different letters. Figure 5-9. Plasmid stability in P101 ipdC Cultures were grown in LB without selection pressure. CFU/ mL was counted after 2 subculturings by dilution spread-plating on media with and without antibiotics to de termine how many cells carried the plasmid and were resistant. pHLT14 carries the ipdC complement with the par locus for plasmid stability. pHLT12 lacks the par locus and was used as a negative control. 123

PAGE 124

124 A B C Figure 5-10. Diagrams of selected plasmid cons tructs. A) pHLT12 is a derivative of pACYC184 with the ipd C gene from E. cloacae P101 cloned into the BamH I resistriction site. B) pHLT14 is a derivati ve of pHLT12 with the par locus for plasmid stability from pTR102 and the kanamycin resistance gene from pCR4 cloned into the ScaI site. C) pHLT16 is a derivative of pHLT14 with the acdS gene for ACC deaminase from P. putida UW4 cloned immediately downstream of the ipdC gene.

PAGE 125

Table 5-1. Examples of plant hormones secreted by bacteria Bacteria Hormone(s) Reference(s) Gluconacetobacter diazotrophicus Auxin, Gibberellin Bastian et al. 1998 Azospirillum sp. Auxin, Cytokinin, Gibberellin Crozier et al. 1988; Ca cciari et al. 1989; Bottini et al. 1998 Azotobacter chroococcum Cytokinin Nieto and Frankenberger 1989 Bacillus sp. Gibberellin Gutierrez-Manero et al. 2001 Herbaspirillum seropedicae Auxin, Gibberellin Bastian et al. 1998 Klebsiella pneumoniae Auxin El-Khawas and Adachi 1999 Paenibacillus polymyxa Auxin, Cytokinin Lebuhn et al. 1997; Timmusk et al. 1999 Pseudomonas flourescens Cytokinin Garcia de Salamone et al. 2001 Table 5-2. Comparison of growth-promoting mechanisms in K. pneumoniae 342, E. cloacae P101, and E. coli K12 Gene product K. pneumoniae 342 E. cloacae P 101 E. coli K12 Indole-pyruvate decarboxylase ( ipdC ) Present Present Absent ACC deaminase Absent Absent Putative 2,3-butanediol production ( budABC ) Absent (only budA ) Present Absent Nitrogenase ( nif ) Present (all 16 nif genes) Absent (only nifU ) Absent (only nifA and nifS ) Phytase Absent Absent Absent Table 5-3. Hormone insensitive Arabidopsis m utants used to examine lateral root promotion Mutant Hormone(s) Reference axr1-12 Auxin Lincoln et al. 1990 axr2-1 Auxin; Ethylene; Abscisic Acid Wilson et al. 1990 axr4-2 Auxin Hobbie and Estelle 1995 axr4-2,1-3 Auxin Hobbie and Estelle 1995 bri1,gl Brassinosteroid Clouse et al. 1996 ein2 Ethylene Alonso et al. 1999 etr1-1 Ethylene Bleecker et al. 1988; Guzman and Ecker 1990 gai-1 Gibberellin Koorneef et al. 1985 cyr1-1 Cytokinin Deikman and Ulrich 1995 abi1-1 Abscisic Acid Koornneef et al. 1984 abi4-1 Abscisic Acid Finkelstein 1994 jar1-1 Jasmonic Acid Staswick et al. 1992 125

PAGE 126

126 Table 5-4. Gene deletion a nd plasm id construction primers Primer Sequence (5 3) A_P101_ipdC CGCGGATCCT ATAACCCGCTTTGGTCGAG D_P101_ipdC CGCGGATCCG GATGATTGCGGTATGGTG F2-JMP101ipdC CATAACAGGTGTCCTTCTGAC R2-JMP101ipdC GCGCGTAATAACGCCTGATTA par1sa GGGGTCGACGACTACACCGAGGGGGAAAG par2sa CTGGTCGACGGGT AAGTACGCCATCAGGA tpkanpar1 TCAGCTACTGGGCTATCTGGA tpkanpar2 CTGGATCTCAACAGCGGTAAG pACYCipdcRXhoI-1 CCTCGAGTCAGGCGTTATTACGCGCTT pACYC184-FSbfI ACTGGCCCTGCAGGTTTGAGAAGCACACGGTCAC ACCd-FXhoI GCTCGAGGATATCACAACAATCAAGGA ACCd-RSbfI CCTGTGCCTGCAGGACTGAATTGAACCCGAAACG

PAGE 127

CHAP TER 6 CONCLUSION Bacteria can interact with plan ts in a variety of different ways. Som e act as pathogens while others live in association with plants wi thout causing disease. Many plant-associated bacteria have beneficial effects on plants. Such bacteria ca n live as epiphytes on the surface of plant tissues, or they can live within plant tissue, as symbi onts or endophytes. The common theme driving the research described in this work is the examination bacteria that inhabit the interior of plant tissues, whether they are a pathogen or growth promoter. This work also aimed to draw upon a wide range of research techniques and approaches, from utilizing bioinformatics with new high-throughput sequencing and optical mapping, to using classic molecular biology and plant culturing methods to answer questi ons about these plantinhabiting bacteria. The presence of bacteria in plant tissues is a common occurrence. In the case of the disease, Citrus Greening, also known as Huanglongbing (HLB), the natu re of the microbial community within infected tissue is still open to debate. Since the bacterium believed to cause this disease, Candidatus Liberibacter spp., has yet to be isolated in pure culture, Kochs postulates have not been fulfilled and its role in disease development has not been confirmed. As a result, the presence and iden tity of any other bacteria with in infected tissue have been investigated, since other pathogens may contribute to development of the disease. Different bacteria have been identified or cultured from HLB-infected plants in various citrus growing regions throughout the world. In China and Braz il, phytoplasmas have b een identified in the midribs of leaves from infected citrus plants (Chen et al. 2009; Teixeira et al. 2008). In addition, Sagaram and associates (2009) re ported a wide array of bacter ial diversity in midribs from infected citrus plants in Florid a, identifying 47 orders of bacteria in 15 different phyla using 16S ribosomal ribonucleic acid (rRNA) microarrays. 127

PAGE 128

In the research described here, a m etagenom ic approach using hi gh-throughput sequence data from HLB-infected tissue was employed to ascertain the microbial community within infected tissue by polymerase ch ain reaction (PCR) independent means, thus removing primer based biases from the analysis. In contrast to the other studies, Ca L. asiaticus was the only bacterium identified in infected tissue. This differe nce is believed to be due to the nature of the tissues sampled in the different studies. The other studies examined leaf midribs whereas the research performed here looked solely at phloem tissue scraped from the inside of bark. Past studies on HLB-infected plants indicate the causative agent is limited to phloem sieve tubes, therefore, specifically focusing on phloem tissu e was deemed the best choice for identifying bacteria associated with the disease. The mi dribs examined in the other studies would have contained leaf tissue, such as the apoplast, which is known to ha rbor bacterial endophytes. As a result, the presence of other bacteria in those tissues could vary based on other factors and may not play a role in the developm ent of HLB. Given the large amount of sequence data obtained from the high-throughput methods used, extensiv e genome coverage of any bacterium present within infected phloem was expected, and Ca L. asiaticus was the only bacterium with such coverage. As the genome sequence of Ca L. asiaticus used in this study was isolated from the insect vector that transmits the disease, finding the same sequence in tissue from an infected citrus plant further subs tantiates the etiology of Ca L. asiaticus in HLB disease. As previously mentioned, many bacteria that can live within plant tissue do not cause disease, and instead have bene ficial effects on their plant hos t. Therefore, research on endophytic bacteria that live in pl ant tissue and increase plant grow th was another main focus in this work. There are several bacterial endophytes th at have been studied for their roles in plant growth promotion, including Gluconacetobacter diazotrophicus PAl 5, Azoarcus sp. BH72, 128

PAGE 129

Herbaspirillum seropedicae Z67, and Klebsiella pneumoniae 342. The genom es of these strains have been or are currently being sequenced. In addition, genomes of close, non-endophytic relatives of these bacteria are also available. With the increasing availability of genomic information, comparative genomics can be used to identify genes unique to the endophytic and growth-promoting lifestyle. In regards to the genome of G. diazotrophicus PAl 5, one of the fi rst nitrogen fixing endophytic bacteria isolated, two different group s have sequenced this strain. Strangely, considerable differences were observed between the closed sequences reported by these two groups. Given this information, it appeared that one of these se quences might contain assembly or sequencing errors. Consideri ng the genomic sequence of this st rain will eventually be used for comparative genomics and the planning new experiments with the bacterium, it was important to determine which of these sequen ces is the best repres entation of the actual G. diazotrophicus PAl 5 genome. To that end, an optical map of the G. diazotrophicus PAl 5 chromosome was made in order to obtain a phy sical map to which the two sequences could be compared. From that comparison, one genomic sequence was found to have significant chromosomal rearrangements. In addition, ex amination of the annot ation identified many differences in the predicted genes in the two se quences. Therefore, the underlying cause of the disparity between the two genomic sequences of the PAl 5 strain has resulted in significant differences at the gene level th at could confound future studies. Possible explanations for the discordant PAl 5 sequence are that the bacterium sequenced has evolved in culture to the point that it is no longer the same as the original PA l 5 strain or that errors occurred during the sequencing and assembly. This situation surrounding the two PAl 5 genome sequences demonstrates the importance of using optical maps when assembling genomes and providing raw 129

PAGE 130

sequencing reads with qu ality scores when subm itting genomes to public databases. With this information, independent verification can be made when contradictory sequences arise and it can be determined if the differences are due to strain evolution or sequencing errors. In either case, the analysis performed in this work demonstrat es which of the two genomic sequences should be used to guide future research on G. diazotrophicus PAl 5. New high-throughput techniques allowing examinati on of plant-associated bacteria at the whole genome level has opened up avenues for comparative genomics, but the ultimate goal of these techniques is still to identify gene target s in the bacteria to examine how they colonize plants and increase their growth. There is only so much information that sequence analysis can reveal about an organism without performing la boratory experiments to confirm the roles of these genes in plant growth and colonization. For this reason, another focus of the research described here involved working directly with endophytes, looking at their effect on root growth in Arabidopsis as an assay for studying their plant gr owth-promoting effects. A variety of experiments were performed on the root growth phenotype. Because genomes of the two endophytic bacteria examined were found to possess a gene for auxin synthesis, this gene was selected as a target for studying growth pr omotion by these strains. While unexpected difficulties arose in the reproducib ility of the lateral root promoting phenotype, a knockout and complementing plasmids of the auxin synthe sis gene were constructed in the endophyte, E. cloacae P101. Therefore, genetic techniques and v ectors for studying the roles of other genes of interest in P101 are available for future studies. The other endophytic bacterium studied in this research, K. pneumoniae 342 (Kp342), was not amenable to genetic manipulation. In addi tion, examination of its genome revealed the presence of numerous pathogenicity and antib iotic resistance genes (Fouts et al. 2008). 130

PAGE 131

131 Subsequent benchwork on this bacterium confir med the antibiotic resistance and pathogenicity of this strain. Therefore, Kp342 is not a suitab le candidate for agricult ural application, though it may be useful as an experimental endophytic model should a convenient system for making in-frame deletions ever be devel oped to work reliably in this st rain. In the current study, given its pathogenic nature, it was d eemed inefficient to expend more time and effort on developing such a system for Kp342. As a result, subseque nt studies focused on another, more genetically tractable endophyte. In closing, the research described here enco mpasses work done on a variety of different plant-associated bacteria. In addition, di fferent techniques were employed to examine plant-microbe interactions over a range of depths. The ultimate goal of this research was to utilize different techniques and technologies for an integrated approach to examining plant-microbe interactions, from genomics a nd metagenomics to the examination of specific genes identified in plant-associated bacteria.

PAGE 132

APPENDIX A GENOME OF KLEBSIELLA PNEUMONIAE 342 In order to direct and facilitate future studies on Klebsiella pneumoniae 342 (Kp342), the genom e of this bacterium was sequenced. With the genome sequence in hand, Kp342 was compared to other plant-associated bacteria in order to identify potentia l plant colonization and growth promotion genes. Given its plant origin s and its relation to ot her human pathogens, the genome was also examined to determine its pathogenic potential. The Kp342 genome was examined for the presence of antibiotic resi stance genes and pathogenicity genes in other K. pneumoniae strains. The ultimate goal of the genome sequencing project wa s to identify targets for gene knockout once a genetic system for K342 is developed. The following work includes text excerpts, data, and analyses that were contributed to th e Kp342 genome paper (Fouts et al. 2008). Bioinformatic analyses were performed by Derric Fouts. Pathogenicity assays were performed by Carston Struve. Heather Tyle r wrote about the biol ogical significance of plant-induced genes identified, performed antibio tic resistance assays on Kp342, and performed statistical analysis on the pathogenicity data. Plant Induced Genes Found in Kp342 Bioinformatic analysis of the Kp342 genome performed by Derrick Fouts demonstrated that this endophyte possesses many genes that sh are homology with known plant-induced genes. Several of these genes were selected as targets for future mutagenesis to confirm their role in endophytic colonization by Kp342. Of these gene s, many appear to be involved in the adaptation of bacteria to conditions within plan t tissue, such as limited amino acid concentration and carbon sources. Several amino acid and nucle otide biosynthesis genes present in Kp342 were found to be induced in Ralstonia solanacearum and Pseudomonas syringae pv. tomato upon plant colonization. These ge nes include CTP synthase ( pyrG ), acetyl-CoA 132

PAGE 133

acety ltransferase, amidophosphoribosyltransferase ( purF ), argininosuccinate synthase ( argG ), a diaminopimelate decarboxylase ( lysA ), and acetolactate synthase large subunit ( ivlI ) (Boch et al. 2002; Brown and Allen 2004). The importance of amino acid biosynthesis in plant-microbe interactions is supporte d by the observation that P. syringae mutants impaired in the biosynthesis of some amino acids were unable to cause diseas e symptoms on tomato (Cuppels 1986). Further gene deletion studies in Kp342 woul d be able to confirm if thes e genes also play a role in endophytic adaptation to plant tissue in addi tion to their role in plant pathogens. Other plant-inducible genes th at Kp342 shares homology with are thought to be involved in adapting to plant defenses. Puta tive stress response ge nes turned on in R. solanacearum upon plant colonization also found in Kp342 include an Ada regulatory protein, excinuclease ( uvrA ), deoxyribonucleic acid (DNA)-dam age-inducible protein F ( dinF ), fumerate hydratase ( fumC), and an acriflavin resistance protein A ( acrA ) (Brown and Allen 2004). Several of these genes have been implicated in bacterial response to oxidative stress and DNA damage due to plant defense responses, some of which are involved in DNA repair. For example, the Ada protein is required to activate the transcri ption of genes involved in adap tive response to DNA methylation damage caused by alkylating agents, and has also been shown to be activated by nitric oxide (Landini and Volkert 1995; Nakabeppu and Sekiguchi 1986; Vasilieva and Maschkovskaya 2005). In addition, exinuclease ( uvrA ) functions in ultraviolet ( UV) induced DNA repair, but has also been shown to participate in hydrogen peroxide and toxic chemical induced DNA damage repair, indicating that this gene may act to protect the bacteria against DNA-damaging compounds produced by plants (Asad et al. 1994 ; Mikulasova et al. 2005; Rupp et al. 1982). These oxidative response genes are not limited to DNA repair pathways. Fumarate hydratase ( fumC), which is synthesized highest under conditions when superoxide radicals accumulate, is 133

PAGE 134

part of the tricarboxylic acid cycl e (Park and Gunsalus 1995). This gene appears to be a back up for the m ore abundant FumA, which is inactivated under oxidative conditions (Park and Gunsalus 1995; Ueda et al. 1991). Since an earl y plant defense response involves the increase of reactive oxygen species, induction of oxidative st ress related genes indicate the bacteria are actively evading this defense mechanism while col onizing plants. Acriflavine resistance protein A ( acrA ) is another stress response gene induced upon plant colonization, but does not appear to be triggered by oxidative stre ss. The product of this gene encodes a component of the AcrAB-TolC efflux pump that is important in toxi c waste removal in bacteria and shows increased expression under stress conditions (Helling et al. 2002; Ma et al. 1995). In the plant pathogen, Erwinia amylovora, AcrAB is induced by phytoalexin plant defense compounds from apple and is required for resist ance to these toxins (Burse et al. 2004). These genes were identified as being induced in plant pathogens so their role in an endophytic bacteriums response to plant defense should be confirmed. A gene believed to be involved in plant att achment has also been identified for further study. This plant-inducible haemagglutinin gene in R. solacacearum is homologous to a Kp342 filamentous haemagglutinin (HecA) protein (Brown and Allen 2004). The hecA gene is part of a HecA/B hemolysin/hemagglutinin secretion operon. The HecA/B proteins make up a two-partner secretion (TPS) system in which a Tp sA family exoprotein with specific conserved secretion signals is transported across the membrane by a TpsB family channel-forming transporter that recognizes the secretion signal (Jacob-Dubuisson et al. 2001). In Erwinia chrysanthemi a mutant in the hecA gene that encodes an adhesin had reduced attachment, cell aggregate formation, and virulence on Nicotinia clevelandii (Rojas et al. 2002). Homologs of this gene appear in both pl ant and animal pathogens (Rojas et al. 2002). A TPS operon ( hlpAB ) 134

PAGE 135

has also been identified in the nonpathogenic plant colonizing bacteria, P seudomonas putida KT2440, and is necessary for competitive root col onization (Molina et al. 2006). The presence of another TPS operon important to colonizatio n in a nonpathogenic plant-associated bacteria gives support the likelihood that the HecA/B homologue in Kp342 is a promising candidate for future study. Antibiotic Resistance in Kp342 Annotation of the Kp342 genome also revealed th e presence of several antibiotic resistance genes. The majority of these genes are effl ux pumps and beta-lactamase genes that confer resistance to a wide array of antibiotics. Resistance of Kp342 to antibiotics conferred by these genes was further confirmed in the lab by the Ki rby-Baur disc method, testing members of all major antibiotic families (Table A-1). Since Kp342 is a plant isolate as opposed to a clinical isolate, the presence of so many broad-spectru m antibiotic resistances in this strain was intriguing. One possible explanation for the maintenance of such resistance genes in Kp342 is for the removal of toxic metabolites produced by plants. Another possible explanation for the antibiotic resistances in Kp342 is exposure to the plant signaling molecule, salicylic acid. Salicylate is known to enhance antibiotic resist ance when co-applied to bacterial cultures of Serratia marcescens and Escherichia coli (Berlanga and Vias 2000; C ohen et al. 1993). It has also been found to increase many an tibiotic resistances in clinical K. pneumoniae strains, including resistance to cefazolin, cefoper azone, norfloxacin, doxycycline, mezlocillin and trimethoprim-sulphamethoxasole (Domenico et al. 1990). Because salicylic acid is pivotal in signaling many plant processes, such as responses to abiotic stress a nd defense responses to pathogens (Raskin 1992), Kp342 would have be en exposed to this hormone during its endophytic lifestyle inside plant tissue. As a result, long-term exposure to this hormone within 135

PAGE 136

the plant could have contributed to the m aintenance of so many antibiotic resistance genes in the genome. Kp342 Pathogenicity Given the number of antibiotic resistance ge nes found in Kp342 and the fact that other strains of K. pneumoniae are known to be pathogenic to human s, the virulence and pathogenicity of Kp342 needed to be investigated before consid ering its use in an agricultural setting. Several virulence factors in pathogenic K. pneumoniae strains have been iden tified by attenuation of disease in signature-tagged mutagenesis studies (L awlor et al. 2005; Struve et al. 2003; Lai et al. 2001). Therefore, these genes were searched for in the Kp342 genome. Several of these pathogenicity genes were identified in the K p342 genome, and with the identification of known virulence genes in Kp342, it was deemed prudent to verify the level of pathogenicity of this endophyte in an animal model. In order to determine pathogenicity in anim als, Kp342 was assayed in urinary tract and lung infection models in mice by a collaborator, Carsten Struve, at the Statens Serum Institute in Copenhagen, Denmark. For comparis on, another clinical isolate of K. pneumoniae C3091, was included in the study. In the urinar y tract model, five out of six mice had infected bladders three days after inoculation with K p342 in numbers similar to those seen when inoculated by the clinical isolate (Table A-2). Kp342 was also able to ascend from the bladder to the kidneys, but at a level 28 times lower than the clinical strain. In the lung infection model, all Kp342 inoculated mice were infected two days after in halation, but infection was 49 times lower than in mice inoculated with C3091. When looking at systemic spreading of Kp342 from the lungs, only one in five mice had infections that spread to the liver and none of the mice had infected spleens (Table A-2). In contra st, three and two of the five C 3091 inoculated mice had infected 136

PAGE 137

livers and spleens, respectively (Table A-2). Therefore, while Kp342 has the potential to be pathogenic and is capable of causi ng infection in an anim al host, its effects are attenuated and less virulent compared to clinical isolates. 137

PAGE 138

138 Table A-1. Kp342 Antibiotic Resistance Profile Drug ( g) Phenotype Observed zone of inhibition diameter (mm)a Interpretive standards (mm)a Novobiocin (30) Resistant 0 17 resistant Gentamicin (10) Intermediate 14 13-14 intermediate Kanamycin (30) Sensitive 30 18 sensitive Neomycin (30) Resistant 9 12 resistant Cefotaxime (30) Sensitive 25 23 sensitive Cefoperazone (75) Sensitive 24 21 sensitive Cefazolin (30) Sensitive 20 18 sensitive Ceftriaxone (30) Intermediate 15 14-20 intermediate Cefuroxime (30) Intermediate 15 15-17 intermediate Cephalothin (30) Resistant 14 14 resistant Moxalactam (30) Intermediate 22 15-22 intermediate Ampicillin (10) Resistant 0 15 resistant Mezlocillin (75) Intermediate 18 18-20 intermediate Penicillin (10)b Resistant 0 Piperacillin (100) Intermediate 19 18-20 intermediate Ticarcillin (75) Resistant 7 14 resistant Azithromycin (15)b Undetermined 10 Erythromycin (15)b Resistant 0 Ciproflaxacin (5) Intermediate 18 16-20 intermediate Nalidixic acid (30) Resistant 8 13 resistant Norfloxacin (10) Resistant 0 12 resistant Oxolinic acid (2) Resistant 7 10 resistant Sulfisoxazole (0.25) Resistant 8 12 resistant Trimethoprim (5) Resistant 0 10 resistant Minocycline (30) Resistant 8 14 resistant Oxytetracycline (30)b Resistant 0 Tetracycline (30) Resistant 10 14 resistant Rifampin (5)b Resistant 0 Spectinomycin (100)b Resistant 0 aObserved zones of inhibition were compared to interpretive standards reported for members of the Enterobacteriaceae. bNo interpretive standards for Enterobacteriaceae were reported. Table A-2. Infection of K. pneumoniae 342 and C3091 in a Mouse Infection Model Model Tissue Log CFU Kp342a Log CFU C3091a Urinary track infection Bladder 3.40 +/0.72 3.94 +/0.40 Urinary track infection Kidneyb 2.43 +/0.40 3.87 +/0.45 Lung infection Liver 0.44 +/0.44 1.99 +/1.02 Lung infection Lungb 4.63 +/0.41 6.32 +/0.38 Lung infection Spleen 0 0.54 +/0.35 aMean of the log of colony formi ng units (CFU) recovered per organ plus or minus the standard error. bTissues had statistically si gnificant difference between K p342 and C3091 infection at the 5% level as determined by Fishers Least Significant Difference Test.

PAGE 139

LITERATURE CITED Abeles, F. B., Morgan, P. W., and Salveit, M. E. Jr. 1992. The biosynthesis of ethylene. Pages 26-55 in: Ethylene in Plant Biology, 2nd ed. Academic Press, San Diego, CA. Alonso, J. M., Hirayama, T., Roman, G., Nouri zadeh, S., and Ecker, J. R. 1999. EIN2, a bifunctional transducer of ethyl ene and stress responses in Arabidopsis Science 284:2148-2152. Anderson, D. M., and Mills, D. 1985. The use of transposon mutagenesis in the isolation of nutritional and virulence mu tants in 2 pathovars of Pseudomonas syringae Phytopathology 75:104-108. Armon, R., Dosoretz, C. G., Azov, Y., and Shelef G. 1994. Residual contamination of crops irrigated with effluent of different qualities-a fieldstudy. Water Sci. Technol. 30:239-248. Asad, L. M. B. O., Dealmeida, C. E. B., Dasilv a, A. B., Asad, N. R., and Leitao, A. C. 1994. Hydrogen-peroxide induces the repair of UV-damages DNA in Escherichia coli A LexAindependent but UvrA-dependent and RecA-dep endent mechanism. Curr. Microbiol. 29:291-294. Aston, C., Mishra, B., and Schwartz, D. C. 1999. Optical mapping and its potential for largescale sequencing projects. Trends Biotechnol. 17:297-302. Auty, M., Duffy, G., OBeirne, D., McGovern, A ., Gleeson, E., and Jordan, K. 2005. In situ localization of Escherichia coli O157: H7 in food by confocal scanning laser microscopy. J. Food Prot. 68:482-486. Aziz, R. K., Bartels, D., Best, A. A., DeJongh, M., Disz, T., Edwards, R. A., Formsma, K., Gerdes, S., Glass, E. M., Kubal, M., Meyer, F., Olsen, G. J., Olson, R., Osterman, A. L., Overbeek, R. A., McNeil, L. K., Paarmann, D., Paczian, T., Parrello, B., Pusch, G. D., Reich, C., Stevens, R., Vassieva, O., Vonste in, V., Wilke, A., and Zagnitko, O. 2008. The RAST Server: Rapid annotations using subsystems technology. BMC Genomics 9:75. Babic, I., Roy, S., Watada, A. E., and Wergin, W. P. 1996. Changes in microbial populations on fresh cut spinach. Int. J. Food Microbiol. 31:107-119. Barak, J. D., Gorski, L., Naraghi-Arani P., and Charkowski, A. O. 2005. Salmonella enterica virulence genes are required fo r bacterial attachment to plant tissue. Appl. Environ. Microbiol. 71:5685-5691. Barak, J. D., Whitehand, L. C., and Charkowski, A. O. 2002. Differences in attachment of Salmonella enterica serovars and Escherichia coli O157: H7 to alfalfa sprouts. Appl. Environ. Microbiol. 68:4758-4763. 139

PAGE 140

Barbieri, P., and Galli, E. 1993. Effect on wheat root development of inoculation with an Azospirillum brasilense mutant with altered indole-3 -acetic acid production. Res. Microbiol. 144:69-75. Bardwell, J. C. A., McGovern, K., and Beckwith J. 1991. Identification of a protein required for disulfide bond formation in vivo. Cell 67:581-589. Bartel, B. 1997. Auxin Biosynthesis. Annu. Re v. Plant Physiol. Plant Mol. Biol. 48:51-66. Bastian, F., Cohen, A., Piccoli, P., Luna, V., Baraldi, R., and Bottini, R. 1998. Production of indole-3-acetic acid and gibberellins A1 and A3 by Acetobacter diazotrophicus and Herbaspirillum seropedicae in chemically-defined culture media. Plant Growth Regul. 24:7-11. Berlanga, M., and Vias, M. 2000. Salicylate induction of phenotypic re sistance to quinolones in Serratia marcescens. J. Antimicrob. Chemother. 46:279-282. Bernier, S. P., Silo-Suh, L., Woods, D. E., Ohma n, D. E., and Sokol, P. A. 2003. Comparative analysis of plant and animal models for characterization of Burkholderia cepacia virulence. Infect. Immun. 71:5306-5313. Bernstein, N., Sela, S., Pinto, R., and Ioffe, M. 2007. Evidence for internalization of Escherichia coli into the aerial parts of maize via the root system. J. Food Prot. 70:471-475. Besser, T. E., LeJeune, J. T., Rice, D. H., Ber g, J., Stilborn, R. P., Kaya, K., Bae, W., and Hancock, D. D. 2005. Increasing prevalence of Campylobacter jejuni in feedlot cattle through the feeding period. Appl. Environ. Microbiol. 71:5752-5758. Bleecker, A. B., Estelle, M. A., Somerville, C., and Kende, H. 1988. Insensitivity to ethylene conferred by a dominant mutation in Arabidopsis thaliana Science 241:1086-1089. Boch, J., Joardar, V., Gao, L., Robertson, T. L., Lim, M., and Kunkel, B. N. 2002. Identification of Pseudomonas syringae pv. tomato genes induced during infection of Arabidopsis thaliana. Mol. Microbiol. 44:73-88. Boddey, R. M. 1995. Biological ni trogen-fixation in sugar-cane a key to energetically viable biofuel production. Crit. Rev. Plant Sci. 14:263-279. Boddey, R. M., de Oliveira, O. C., Urquiaga, S., Re is, V. M., de Olivares, F. L., Baldani, V. L. D., and Dbereiner, J. 1995. Biological nitr ogen-fixation associated with sugar cane and rice -contributions and prospects for improvement. Plant Soil 174:195-209. Bohm, M., Hurek, T., and Reinhold-Hurek, B. 2007. Twitching motility is essential for endophytic rice colonization by the N2-fixing endophyte Azoarcus sp. strain BH72. Mol. Plant-Microbe Inte ract. 20:526-533. 140

PAGE 141

Bosshard, P. P., Stettler, R., and Bachofen, R. 2000. Seasonal and spatial comm unity dynamics in the meromictic Lake Cada gno. Arch. Microbiol. 174:168-174. Bottini, R., Fulchieri, M., Pearce, D., and Pharis R. P. 1989. Identification of gibberellin-A1, gibberellin-A3, and iso-A3 in cultures of Azospirillum lipoferum Plant Physiol. 90:45-47. Bov, J. M. 2006. Huanglongbing: A destruc tive, newly-emerging, century-old disease of citrus. J. Plant. Pathol. 88:7-37. Brandl, M. T. 2006. Fitness of human enteric pathogens on plants and implications for food safety. Annu. Rev. Phytopathol. 44:367-392. Brandl, M.T., and Mandrell, R. E. 2002. Fitness of Salmonella enterica serovar Thompson in the cilantro phyllosphere. Appl Environ. Microbiol. 68:3614-3621. Brown, D. G. and Allen, C. 2004. Ralstonia solanacearum genes induced during growth in tomato: an inside view of bacterial wilt. Mol. Microbiol. 53:1641-1660. Brggemann, H., Henne, A., Hoster, F., Liesegang, H., Wiezer, A., Strittmatter, A., Hujer, S., Drre, P., and Gottschalk, G. 2004. The complete genome sequence of Propionibacterium acnes, a commensal of human skin. Science 305:671-673. Burnett, S. L., Chen, J. R., and Beuchat, L. R. 2000. Attachment of Escherichia coli O157: H7 to the surfaces and internal structures of apples as detected by confocal scanning laser microscopy. Appl. Environ. Microbiol. 66:4679-4687. Burse, A., Weingart, H., and Ullrish, M. S. 2004. The phytoalexin-inducible multidrug efflux pump AcrAB contributes to virulenc e in the fire blight pathogen, Erwinia amylovora Mol. Plant-Microbe Interact. 17:43-54. Cacciari, I., Lippi, D., Pietrosanti, T., and Piet rosanti, W. 1989. Phytohormone-like substances produced by single and mixed diazotrophic cultures of Azospirillum and Arthrobacter. Plant Soil 115:151-153. Calvin, L. 2007. Outbreak linked to spinach forces reassessment of food safety practices. Amber Waves 5:24-31. Capoor, S. P., Rao, D. G., and Viswanat, S. M. 1967. Diaphorina citri Kuway., a vector of greening disease of citrus in India. Indian J. Agr. Sci. 37:572-576. Carreno-Lopez, R., Campos-Reales, N., Elmeric h, C., and Baca, B. E. 2000. Physiological evidence for differently regulated tryptophandependent pathways fo r indole-3-acetic acid synthesis in Azospirillum brasilense Mol. Gen. Genet. 264:521-530. Casimiro, I., Beeckman, T., Graham, N., Bhalerao R., Zhang, H., Casero, P., Sandberg, G., and Bennett, M. J. 2003. Dissecting Arabidopsis lateral root development. Trends Plant Sci. 8:165-171. 141

PAGE 142

Chang, A. C. Y., and Cohen, S. N. 1978. Cons truction and characteri zation of a mplifiable multicopy DNA cloning vehicles derived from the P 15A cryptic miniplasmid. J. Bacteriol. 134:1141-1156. Charkowski, A. O., Barak, J. D., Sarreal, C. Z., and Mandrell, R. E. 2002. Differences in growth of Salmonella enterica and Escherichia coli O157: H7 on alfalfa sprouts. Appl. Environ. Microbiol. 68:3114-3120. Chelius, M. K., and Triplett, E. W. 2000. Immunolocalization of dinitrogenase reductase produced by Klebsiella pneumoniae in association with Zea mays L. Appl. Environ. Microbiol. 66:783-787. Chelius, M. K., and Triplett, E. W. 2001. The diversity of archaea and b acteria in association with the roots of Zea mays L. Microb. Ecol. 41:252-263. Chen, J., Pu, X., Deng, X., Liu, S., Li, H., and Civerolo, E. 2009. A phytoplasma related to Candidatus Phytoplasma asteri detected in citr us showing Huanglongbing (Yellow Shoot Disease) symptoms in Guangdong, P. R. China. Phytopathology 99:236-242. Chen, Q., Savarino, S. J., and Venkatesan, M. M. 2006. Subtractive hybridization and optical mapping of the enterotoxigenic Escherichia coli H10407 chromosome: isolation of unique sequences and demonstration of signific ant similarity to the chromosome of E. coli K-12. Microbiology 152:1041-1054. Chevalier, F., Pata, M., Nacry, P., Doumas, P ., and Rossignol, M. 2003. Effects of phosphate availability on root system architecture: la rge-scale analysis of the natural variation between Arabidopsis accessions. Plant Cell Environ. 26:1839-1850. Clouse, S. D., Langford, M., and McMorris, T. C. 1996. A brassinosteroidinsensitive mutant in Arabidopsis thaliana exhibits multiple defects in growth and development. Plant Physiol. 11:671-678. Coenye, T., Vandamme, P., Govan, J. R. W ., and Lipuma, J. J. 2001. Taxonomy and identification of the Burkholderia cepacia complex. J. Clin. Microbiol. 39:3427-3436. Cohen, S. P., Levy, S. B., Foulds, J., and Rosner J. L. 1993. Salicylat e induction of antibiotic resistance in Escherichia coli : activation of the mar operon and a mar -independent pathway. J. Bacteriol. 175:7856-7862. Cooley, M. B Chao, D., and Mandrell, R. E. 2006. Escherichia coli O157: H7 survival and growth on lettuce is altered by the presence of epiphytic bacteria. J. Food Prot. 69:2329-2335. Cooley, M. B., Miller, W. G., and Ma ndrell, R. E. 2003. Colonization of Arabidopsis thaliana with Salmonella enterica and enterohemorrhagic Escherichia coli O157: H7 and competition by Enterobacter asburiae. Appl. Environ. Microbiol. 69:4915-4926. 142

PAGE 143

Cox-Foster, D. L., Conlan, S., Hol mes, E. C., Pal acios, G., Evans, J. D., Moran, N. A., Quan, P. L., Briese, T., Hornig, M., Geiser, D. M., Martinson, V., vanEngelsdorp, D., Kalkstein, A. L., Drysdale, A., Hui, J., Zhai, J. H., Cui, L. W., Hutchison, S. K., Simons, J. F., Egholm, M., Pettis, J. S., and Lipkin, W. I. 2007. A metagenomic survey of microbes in honey bee colony collapse disorder. Science 318:283-287. Crozier, A., Arruda, P., Jasmim, J. M., Montei ro A. M., and Sandberg, G. 1988. Analysis of indole-3-acetic acid and related indoles in culture medium from Azospirillum lipoferum and Azospirillum brasilense Appl. Environ. Microbiol. 54:2833-2837. Cuppels, D. A. 1986. Generation and charac terization of Tn5 inse rtion mutations in Pseudomonas syringae pv. tomato Appl. Environ. Microbiol. 51:323-327. Davis, M. J., Mondal, S. N., Chen, H. Q ., Rogers, M. E., and Brlansky, R. H. 2008. Co-cultivation of Candidatus Liberibacter asiaticus' with Actinobacteria from Citrus with Huanglongbing. Plant Dis. 92:1547-1550. De Leij, F. A. A. M., Dixon-Hardy, J. E ., and Lynch, J. M. 2002. Effect of 2,4-diacetylphloroglucinol-producin g and non-producing strains of Pseudomonas fluorescens on root development of pea seedlings in three different soil types a nd its effect on nodulation by Rhizobium Biol. Fertil. Soils. 35:114-121. Debi, B. R., Taketa, S., and Ichii, M. 2005. Cytokinin inhibits lateral root initiation but stimulates lateral root elongation in rice ( Oryza sativa ). J. Plant Physiol. 162:507-515. Deikman, J., and Ulrich, M. 1995. A novel cytokinin-resistant mutant of Arabidopsis with abbreviated shoot developm ent. Planta 195:440-449. Dobbelaere, S., Croonenborghs, A., Thys, A., Va nde Broek, A., and Vanderleyden, J. 1999. Phytostimulatory effect of Azospirillum brasilense wild type and mutant strains altered in IAA production on wheat. Plant Soil 212:155-164. Domenico, P., Hopkins, T., and Cunha, B. A. 1990. The effect of sodium salicylate on antibiotic susceptibility and synergy in Klebsiella pneumoniae J. Antimicrob. Chemother. 26:343-351. Dong, Y. M., Iniguez, A. L., Ahmer, B. M. M., a nd Triplett, E. W. 2003a. Kinetics and strain specificity of rhizosphere and endophytic col onization by enteric bacteria on seedlings of Medicago sativa and Medicago truncatula Appl. Environ. Mi crobiol. 69:1783-1790. Dong, Y. M., Iniguez, A. L., and Triplett, E. W. 2003b. Quantitative assessments of the host range and strain specificity of endophytic colonization by Klebsiella pneumoniae 342. Plant Soil 257:49-59. Doddapaneni, H., Liao, H., Lin, H., Bai, X., Zha o, X., Civerolo, E. L., Irey, M., Coletta-Filho, H., and Pietersen, G. 2008. Comparative phy logenomics and multi-gene cluster analyses of the Citrus Huanglongbing (H LB)-associated bacterium Candidatus Liberibacter. BMC Res. Notes 1:72. 143

PAGE 144

Dorr, J., Hurek, T., and Reinhold-H urek, B. 1998. Type IV pili are involved in plant-microbe and fungus-microbe interactions. Mol. Microbiol. 30:7-17. Duan, Y., Zhou, L., Hall, D. G., Li, W., Doddapa neni, H., Lin, H., Liu, L., Vahling, M., Gabriel, D. W., Williams, K. P., Dickerman, A., Sun, Y., and Gottwald, T. 2009. Complete genome sequence of citrus Huanglongbing bacterium, Candidatus Liberibacter asiaticus obtained through metagenomics. Mol. Pl ant Microbe Intera ct. 22(8):1011-1020. Duarte, X., Anderson, C. T., Grimson, M., Barabote R. D., Strauss, R. E., Gollahon, L. S. and San Francisco, M. J. D. 2000. Erwinia chrysanthemi strains cause death of human gastrointestinal cells in culture and express an intimin-like protein. FEMS Microbiol. Lett. 190:81-86. Dubrovsky, J. G., Doerner, P. W, Colon-Carmona A., and Rost, T. L. 2000. Pericycle cell proliferation and lateral root initiation in Arabidopsis. Plant Physiol. 124:1648-1657. Eblen, B. S., Walderhaug, M. O., Edelson-Mammel, S ., Chirtel, S. J., De Jesus, A., Merker, R. I., and Buchanan, R. L., and Miller, A. J. 2004. Potential for interna lization, growth, and survival of Salmonella and Escherichia coli O157: H7 in oranges. J. Food Prot. 67:15781584. El-Khawas, H., and Adachi, K. 1999. Identification and quantification of auxins in culture media of Azospirillum and Klebsiella and their effect on rice r oots. Biol. Fertil. Soils 28:377-381. Emsley, P., Charles, I. G., Fairweather, N. F., and Isaacs, N. W. 1996. Structure of Bordetella pertussis virulence factor P.69 pe rtactin. Nature 381:90-92. Engelhard, M., Hurek, T., and Reinhold-Hurek B. 2000. Preferential occurrence of diazotrophic endophytes, Azoarcus spp., in wild rice spec ies and land races of Oryza sativa in comparison with modern races. Environ. Microbiol. 2:131-141. Espinosa-Urgel, M., and Ramos, J. L. 2004. Cell density-dependent gene contributes to efficient seed colonization by Pseudomonas putida KT2440. Appl. Environ. Microbiol. 70:5190-5198. Felsenstein, J. 1985. Confiden ce limits on phylogenies: an approach using the bootstrap. Evolution 39:783-791. Finkelstein, R. R. 1994. Mutations at two new Arabidopsis ABA response loci are similar to the abi3 mutations. Plant J. 5:765-771. Fisher, M. M. and Triplett, E. W. 1999. Auto mated approach for ribosomal intergenic spacer analysis of microbial diversity and its applic ation to freshwater bacterial communities. Appl. Environ. Microbiol. 65:4630-4636. 144

PAGE 145

Folim onova, S. Y., Robertson, C. J., Garnsey, S. M., Gowda, S., and Dawson. W. O. 2009. Examination of the responses of different genotypes of citrus to Huanglongbing (citrus greening) under different conditions. Phytopathology. In press. Fournet-Fayard, S., Joly, B., and Forestier, C. 1995. Transformation of wild type Klebsiella pneumoniae with plasmid DNA by electroporation. J. Microbiol. Methods 24:49-54. Fouts, D. E., Tyler, H. L., DeBoy, R. T., Daugherty, S., Ren, Q., Badger, J. H., Durkin, A. S., Huot, H., Shrivastava, S., Kothari, S., D odson, R. J., Mohamoud, Y., Khouri, H., Roesch, L. F. W., Krogfelt, K. A., Struve, C., Triplett, E. W., and Meth, B. A. 2008. Complete genome sequence of the N2-fixing broad host range endophyte Klebsiella pneumoniae 342 and virulence predictions verified in mice. PLoS Genet. 4:e1000141. Franz, E., van Diepeningen, A. D., de Vos, O. J., van Bruggen, A. H. C. 2005. Effects of cattle feeding regimen and soil management type on the fate of Escherichia coli O157: H7 and Salmonella enterica serovar typhimurium in manure, manure-amended soil, and lettuce. Appl. Environ. Microbiol. 71:6165-6174. Franz, E., Visser, A. A., van Diepeningen, A. D., Klerks, M. M., Termorshuizen, A. J., and van Bruggen, A. H. C. 2007. Quantification of contamination of lettuce by GFP-expressing Escherichia coli O157: H7 and Salmonella enterica serovar Typhimurium. Food Microbiol. 24:106-112. Fuentes-Ramirez, L. E., Jimenez-Salgado, T., Abarcao-Campo, I. R., and Caballero-Mellado, J. 1993. Acetobacter diazotrophicus an indoleacetic acid produc ing bacterium isolated from sugarcane cultivars of Mexico. Plant Soil 154:145-150. Fulchieri, M., Lucangeli, C., and Bottini, R. 1993. Inoculation with Azospirillum lipoferum affects growth and gibberellin status of co rn seedling roots. Plant Cell Physiol. 34:1305-1309. Gagliardi, J. V., and Karns, J. S. 2002. Persistence of Escherichia coli O157: H7 in soil and on plant roots. Envir on. Microbiol. 4:89-96. Garcia de Salamone, I. E., Hynes, R. K., and Nelson, L. M. 2001. Cytokinin production by plant growth promoting rhizobacteria and selected mutants. Can. J. Microbiol. 47:404-411. Garnier, M. and Bov, J. M. 1983. Transmissi on of the organism associated with citrus greening disease from sweet orange to periwinkle by dodder. Phytopathology 73:1358-1363. Garnier, M., Danel, N., and Bov, J. M. 1984. Aetiology of citrus greening disease. Annales De Microbiologie 135A:169-179. Ghosh, S. K., Giannotti, J., and Louis, C. 1978. Multiplication intensive des prokaryotes associs aux maladies de type greening des ag rumes dans les cellules cribles de cuscute. Annales de Phytopathologie 9:525-530. 145

PAGE 146

Gillis, M., Kersters, K., Hoste, B., Janssens, D., Kroppensted t, R. M., Stephan, M. P., Teixeira, K. R. S., Dbereiner, J., and De Ley, J. 1989. Acetobacter diazotrophicus sp. nov., a nitrogen-fixing acetic-acid bacteriu m associated with sugarcane. Int. J. Syst. Bacteriol. 39:361-364. Girardin, H., Morris, C. E., Albagnac, C., Dreux, N., Glaux, C., and Nguyen-The, C. 2005. Behaviour of the pathogen surrogates Listeria innocua and Clostridium sporogenes during production of parsley in fields fertilized with contaminated amendments. FEMS Microbiol. Ecol. 54:287-295. Glick, B. R. 2005. Modulation of plant ethy lene levels by the bacterial enzyme ACC deaminase. FEMS Microbiol. Lett. 251:1-7. Gordan, S. A., and Weber, R. P. 1951. Colori metric estimation of i ndoleacetic acid. Plant Physiol. 26:192-195. Govan, J. R. W., Hughes, J. E., and Vandamme, P. 1996. Burkholderia cepacia : medical, taxonomic and ecological issues. J. Med. Microbiol. 45:395-407. Guo, X., Chen, J. R., Brackett, R. E., and Beuchat, L. R. 2001. Survival of Salmonellae on and in tomato plants from the time of inoculat ion at flowering and early stages of fruit development through fruit ripening. A ppl. Environ. Microbiol. 67:4760-4764. Gutierrez-Manero, F. J., Ramos-Solano, B., Proba nza, A., Mehouachi, J., Tadeo, F. R., and Talon, M. 2001. The plant-gr owth-promoting rhizobacteria Bacillus pumilus and Bacillus licheniformis produce high amounts of physiologically activ e gibberellins. P hysiol. Plant. 111:206-211. Guzman, P. and Ecker, J. R. 1990. Exploiting the triple response of Arabidopsis to identify ethylene-related mutants. Plant Cell 2:513-523. Gyaneshwar, P., James, E. K., Mathan, N., Reddy, P. M., Reinhold-Hurek, B., and Ladha, J. K. 2001. Endophytic colonization of ri ce by a diazotrophic strain of Serratia marcescens J. Bacteriol. 183:2634-2645. Hall, J. A., Peirson, D., Ghosh, S., and Glick, B. R. 1996. Root elongation in various agronomic crops by the plant growth promoting rhizobacterium Pseudomonas putida GR12-2. Isr. J. Plant Sci. 44:37-42. Hardt, W. D., and Galan, J. E. 1997. A secreted Salmonella protein with homology to an avirulence determinant of plant pathogenic bacteria. Proc. Natl. Acad. Sci. USA 94:9887-9892. Hassan, A. N., and Frank, J. F. 2004. Attachment of Escherichia coli O157: H7 grown in tryptic soy broth and nutrient broth to apple and le ttuce surfaces as rela ted to cell hydrophobicity, surface charge, and capsule production. Int. J. Food Microbiol. 96:103-109. 146

PAGE 147

He, J. X., Baldini, R. L., Deziel, E., Saucier, M. Zhang, Q. H., Liberati, N. T., Lee, D., Urbach, J., Goodm an, H. M., and Rahme, L. G. 2004. The broad host range pathogen Pseudomonas aeruginosa strain PA14 carries two pathoge nicity islands harboring plant and animal virulence genes. Proc Natl. Acad. Sci. USA 101:2530-2535. Helling, R. B., Janes, B. K., Kimball, H., Tra n, T., Bundesmann, M., Check, P., Phelan, D., and Miller, C. 2002. Toxic waste disposal in Escherichia coli J. Bacteriol. 184:3699-3703. Himanen, K., Boucheron, E., Vanneste, S., de Almeida Engler, J., Inze, D., and Beeckman, T. 2002. Auxin-mediated cell cycle activation during early lateral root initiation. Plant Cell 14:2339-2351. Himanen, K., Vuylsteke, M., Vanneste, S., Vercruysse, S., Boucheron, E., Alard, P., Chriqui, D., van Montagu, M., Inze, D., and Beeckman, T. 2004. Transcript profiling of early lateral root initiation. Proc. Natl Acad. Sci. USA 101:5146-5151. Hobbie, L., and Estelle, M. 1995. The axr4 auxin-resistant mutants of Arabidopsis thaliana define a gene important for root gravitropis m and lateral root initiation. Plant J. 7:211-220. Hoefsloot, G., Termorshuizen, A. J., Watt, D. A ., and Cramer, M. D. 2005. Biological nitrogen fixation is not a major contributor to the ni trogen demand of a commercially grown South African sugarcane cultivar Plant Soil 277:85-96. Hontzeas, N., Saleh, S. S., and Glick, B. R. 2004. Changes in gene expression in canola roots induced by ACC-deaminase-containing plantgrowth-promoting bacteria. Mol. PlantMicrobe. Interact. 17:865-8671. Howard, M. B., and Hutcheson, S. W. 2003. Growth dynamics of Salmonella enterica strains on alfalfa sprouts and in wast e seed irrigation water. A ppl. Environ. Microbiol. 69:548553. Hrabak, E. M., and Willis, D. K. 1992. The lemA gene required for pathogenicity of Pseudomonas syringae pv. syringae on bean is a member of a family of two-component regulators. J. Bacteriol. 174:3011-3020. Hrabak, E. M., and Willis D. K. 1993. Involvement of the lemA gene in production of syringomycin and protease by Pseudomonas syringae pv. syringae Mol. Plant-Microbe Interact. 6:368-3675. Ibekwe, A. M., Watt, P. M., Shouse, P. J., and Grieve, C. M. 2004. Fate of Escherichia coli O157: H7 in irrigation water on soils and plan ts as validated by culture method and realtime PCR. Can. J. Microbiol. 50:1007-1014. Ibenyassine, K., AitMhand, R., Karamoko, Y., Cohen, N., and Ennaji, M. M. 2006. Use of repetitive DNA sequences to determine the persistence of enteropathogenic Escherichia coli in vegetables and in soil gr own in fields treated with contaminated irrigation water. Lett. Appl. Microbiol. 43:528-533. 147

PAGE 148

Ibenyassine, K., Mhand, R. A., Karamoko, Y., Anajjar, B., Chouibani, M., and Ennaji, M. M. 2007. Bacterial pathogens recovered from vegeta bles irrigated by wastewater in Morocco. J. Environ. Health 69:47-51. Idriss, E. E., Makarewicz, O., Farouk, A., Rosner, K., Greiner, R., Bochow, H., Richter, T., and Borriss, R. 2002. Extrace llular phytase activity of Bacillus amyloliquefaciens FZB45 contributes to its plant-growth-promo ting effect. Microbiology 148:2097-2109. Iglewski, B. H., and Kabat, D. 1975. NAD-de pendent inhibition of protein synthesis by Pseudomonas aeruginosa toxin. Proc. Natl. Acad. Sci. USA 72:2284-2288. Iniguez, A. L., Dong, Y. M., Carter H. D., Ahmer, B. M. M., Stone, J. M., and Triplett, E. W. 2005. Regulation of enteric endophytic bacterial colonization by plant defenses. Mol. Plant-Microbe. Interact. 18:169-178. Iniguez, A. L., Dong, Y., and Triplett, E. W. 2004. Nitrogen fixation in wheat provided by Klebsiella pneumoniae 342. Mol. Plant-Microbe Interact. 17:1078-1085. Islam, M., Doyle, M. P., Phatak, S. C., Milln er, P., and Jiang, X. P. 2005. Survival of Escherichia coli O157: H7 in soil and on carrots and onions grown in fields treated with contaminated manure composts or irriga tion water. Food Microbiol. 22:63-70. Islam, M., Morgan, J., Doyle, M. P., Phatak, S. C ., Millner, P., and Jiang, X. P. 2004. Fate of Salmonella enterica serovar Typhimurium on carrots and radishes grown in fields treated with contaminated manure composts or irri gation water. Appl. Environ. Microbiol. 70:2497-2502. Itoh, Y., Sugita-Konishi, Y., Kasuga, F., Iwaki, M., Hara-Kudo, Y., Saito, N., and Noguchi, Y. 1998. Enterohemorrhagic Escherichia coli O157: H7 present in radish sprouts. Appl. Environ. Microbiol. 64:1532-1535. Jablasone, J., Warriner, K., and Griff iths, M. 2005. Interactions of Escherichia coli O157: H7, Salmonella typhimurium and Listeria monocytogenes plants cultivated in a gnotobiotic system. Int. J. Food Microbiol. 99:7-18. Jacob-Dubuisson, F., Locht, C., and Antoine, R. 2001. Two-partner secretion in Gram-negative bacteria: a thrifty, specific pathway for la rge virulence proteins. Mol. Microbiol. 40:306-313. Jagoueix, S., Bov, J. M., and Garnier, M. 1994. The phloem-limited bacterium of greening disease of citrus is a member of the subdivision of the Proteobacteria Int. J. Syst. Bacteriol. 44:379-386. Jagoueix, S., Bov, J. M., and Garnier, M. 1997. Comparison of the 16S/23S ribosomal intergenic regions of '' Candidatus Liberobacter asiaticum'' and '' Candidatus Liberobacter africanum,'' the two species asso ciated with citrus Huanglongbing (greening) disease. Int. J. Syst. Bacteriol. 47:224-227. 148

PAGE 149

Janes, M. E., Ki m, K. S., and Johnson, M. G. 2005. Transmission electron microscopy study of enterohemorrhagic Escherichia coli O157: H7 in apple tissue. J. Food Prot. 68:216-224. Jantama, K., Zhang, X., Moore, J. C., Shanmugam, K. T., Svoronos, S. A., and Ingram, L. O. 2008. Eliminating side products and increasing su ccinate yields in e ngineered strains of Escherichia coli C. Biotechnol. Bioeng. 101:881-893. Jeter, C., and Matthysse, A. G. 2005. Characterization of the binding of diarrheagenic strains of E. coli to plant surfaces and the role of curli in the interaction of the bacteria with alfalfa sprouts. Mol. Plant-Microbe Interact. 18:1235-1242. Jha, A. K., Bais, H. P., and Vivanco, J. M. 2005. Enterococcus faecalis mammalian virulence-related factors exhibi t potent pathogenicity in the Arabidopsis thaliana plant model. Infect. Immun. 73:464-475. Jimenez, B., Austin, A., Cloete, E., and Ph asha, C. 2006. Using Ecosan sludge for crop production. Water Sci. Technol. 54:169-177. Jimenez-Salgado, T., Fuentes-Ramirez, L. E., Ta pia-Hernandez, A., Masc arua-Esparza, M. A., Martinez-Romero, E., and Caballero-Mellado, J. 1997. Coffea arabica L., a new host plant for Acetobacter diazotrophicus and isolation of other n itrogen-fixing acetobacteria. Appl. Environ. Microbiol. 63:3676-3683. Johannessen, G. S., Bengtsson, G. B., Heier, B. T., Bredholt, S., Wasteson, Y., and Rorvik, L. M. 2005. Potential uptake of Escherichia coli O157: H7 from organic manure into crisphead lettuce. Appl. Envir on. Microbiol. 71:2221-2225. Kanbe, K., Naganawa, H., Okamura, M., Sasaki, T ., Hamada, M., Okami, Y., and Takeuchi, I. 1993. Amidenin, a new plant growth-re gulating substance isolated from Amycolatopsis sp. Biosci. Biotechnol. Bi ochem. 57:1261-1263. Karadeniz, A., Topcuoglu, S. F., and Inan, S. 2006. Auxin, gibberellin, cy tokinin and abscisic acid production in some bacteria. Wo rld J. Microbiol. Biotechnol. 22:1061-1064. Kenney, S. J., Anderson, G. L., Williams, P. L., Millner, P. D., and Beuchat, L. R. 2006. Migration of Caenorhabditis elegans to manure and manure compost and potential for transport of Salmonella newport to fruits and vegetables. Int. J. Food Microbiol. 106:61-68. Kenney, S. J., Burnett, S. L., and Beuchat, L. R. 2001. Location of Escherichia coli O157: H7 on and in apples as affected by bruising, washing, and rubbing. J. Food Prot. 64:1328-1333. Klerks, M. M., Franz, E., van Gent-Pelzer, M., Zijlstra, C., van Bruggen, A. H. C. 2007. Differential interaction of Salmonella enterica serovars with lettuce cultivars and plant-microbe factors influencing the col onization efficiency. ISME J. 1:620-631. 149

PAGE 150

Knee, E. M., and Hangarter, R. P. 1996. Di fferential IAA dose res ponse relations of the axr1 and axr2 m utants of Arabidopsis Physiol. Plant. 98:320-324. Koga, J., Syono, K., Ichikawa, T., and Adachi, T. 1994. Involvement of L-tryptophan aminotransferase in indole-3acetic acid biosynthesis in Enterobacter cloacae Biochim. Biophys. Acta 1209:241-247. Koornneef, M., Elgersma, A., Hanhart, C. J., van Leonen-Martinet, E. P., van Rign, L., and Zeevaart, J. A. D. 1985. A gi bberellin insensitive mutant of Arabidopsis thaliana Physiol. Plant. 65:33-39. Koornneef, M., Reuling, G., and Karssen, C. M. 1984. The isolation and characterization of abscisic acid-insensitive mutants of Arabidopsis thaliana Physiol. Plant. 61:377-383. Krause, A., Ramakumar, A., Bartels, D., Battistoni F., Bekel, T., Boch, J., Bhm, M., Friedrich, F., Hurek, T., Krause, L., Linke, B., McHardy, A. C., Sarkar, A., Schneiker, S., Syed, A. A., Thauer, R., Vorhlter, F. J., Weidner, S ., Phler, A., Reinhold-Hurek, B., Kaiser, O., and Goesmann, A. 2006. Complete genome of the mutualistic, N2-fixing grass endophyte Azoarcus sp strain BH72. Nat. Biotechnol. 24:1385-1391. Kuklinsky-Sobral, J., Araujo, W. L., Mendes, R., Geraldi, I. O., Pizzirani-Kleiner, A. A., and Azevedo, J. L. 2004. Isolation and characte rization of soybean-associated bacteria and their potential for plant growth promotion. Environ. Microbiol. 6:1244-1251. Kutter, S., Hartmann, A., and Schmid, M. 2006. Colonization of barley ( Hordeum vulgare ) with Salmonella enterica and Listeria spp. FEMS Microbiol. Ecol. 56:262-271. Lacava, P. T., Araujo, W. L., and Azevedo, J. L. 2007. Evaluation of endophytic colonization of Citrus sinensis and Catharanthus roseus seedlings by endophytic bacteria. J. Microbiol. 45:11-14. Lai, Y. C., Peng, H. L., and Chang, H. Y. 2001. Identification of genes induced in vivo during Klebsiella pneumoniae CG43 infection. Infect. Immun. 69:7140-7145. Lambrecht, M., Okon, Y., Vande Broek, A., and Va nderleyden, J. 2000. Indole-3-acetic acid: a reciprocal signaling molecule in bacteria-plant interactions. Trends Microbiol. 8:298-300. Landini, P. and Volkert, M. R. 1995. Transcriptional activation of the Escherichia coli adaptive response gene aidB is mediated by binding of methylat ed Ada protein. J. Biol. Chem. 270:8285-8289. Latreille, P., Norton, S., Goldman, B. S., Henkha us, J., Miller, N., Barbazuk, B., Bode, H. B., Darby, C., Du, Z. J., Forst, S., Gaudriault, S ., Goodner, B., Goodrich-Blair, H., and Slater, S. 2007. Optical mapping as a routine tool for bacterial genome sequence finishing. BMC Genomics 8:321. 150

PAGE 151

Laville, J., Voisard, C., Keel, C., Maurhofer, M. Defago, G., and Haas, D. 1992. Global control in Pseudomonas fluorescens m ediating antibiotic synthesis and suppression of black root rot of tobacco. Proc. Natl. Acad. Sci. USA 89:1562-1566. Lawlor, M. S., Hsu, J., Rick, P. D., and Miller, V. L. 2005. Identification of Klebsiella pneumoniae virulence determinants using an intranas al infection model. Mol. Microbiol. 58:1054-1073. Lebuhn, M., Heulin, T., and Hartmann, A. 1997. Production of auxin and other indolic and phenolic compounds by Paenibacillus polymyxa strains isolated from different proximity to plant roots. FEMS Microbiol. Ecol. 22:325-334. Lee, S., Flores-Encarnacin, M., Contreras-Zentella, M., Garcia-Flores, L., Escamilla, J. E., and Kennedy, C. 2004. Indole-3-acetic acid biosynthesis is deficient in Gluconacetobacter diazotrophicus strains with mutations in cytochrome c biogenesis genes. J. Bacteriol. 186:5384-5391. Li, J. P., Ovakim, D. H., Charles, T. C., a nd Glick, B. R. 2000. An ACC deaminase minus mutant of Enterobacter cloacae UW4 no longer promotes root elongation. Curr. Microbiol. 41:101-105. Lim, A., Dimalanta, E. T., Potamousis, K. D., Yen, G., Apodoca, J., Tao, C. H., Lin, J. Y., Qi, R., Skiadas, J., Ramanathan, A., Perna, N. T., Plunkett, G., Burland, V., Mau, B., Hackett, J., Blattner, F. R., Anantharaman, T. S., Mi shra, B., and Schwartz, D. C. 2001. Shotgun optical maps of the whole Escherichia coli O157 : H7 genome. Genome Res. 11:1584-1593. Lin, H., Doddapaneni, H., Bai, X. J., Yao, J. Q., Zhao, X. L., and Civerolo, E. L. 2008. Acquisition of uncharacterized sequences from Candidatus Liberibacter, an unculturable bacterium, using an improved genomic walk ing method. Mol. Cell. Probes 22:30-37. Lincoln, C., Britton, J. H., and Estelle, M. 1990. Growth and development of the axr1 mutants of Arabidopsis Plant Cell 2:1071-1080. Loncarevic, S., Johannessen, G. S., and Rorvi k, L. M. 2005. Bacteriological quality of organically grown leaf lettuce in Norw ay. Lett. Appl. Microbiol. 41:186-189. Lopez-Bucio, J., Hernandez-Abreu, E., SanchezCalderon, L., Nieto-Jacobo, M. F., Simpson, J., and Herrera-Estrella, L. 2002. Phosphate ava ilability alters architecture and cause changes in hormone sensitivity in the Arabidopsis root system. Plant Physiol. 129:244-256. Loubens, I., Debarbieux, L., Bohin, A., Lacroix, J. M., and Bohin, J. P. 1993. Homology between a genetic locus ( mdoA ) involved in the osmoregulated biosynthesis of periplasmic glucans in Escherichia coli and a genetic locus ( hrpM ) controlling pathogenicity of Pseudomonas syringae Mol. Microbiol. 10:329-340. 151

PAGE 152

Lynch, M., Painter, J., Woodruff, R., and Brade n, C. 2006. Surveillance for foodborne-disease outbreaks-United States, 1998-2002. MMWR Surveill. Summ. 55:1-34. Ma, D., Cook, D. N., Al berti, M., Pon, N. G., Nikaido, H., and Hearst, J. E. 1995. Genes acrA and acrB encode a stress-indu ced efflux system in Escherichia coli Mol. Microbiol. 16:45-55. Madigan, M. T., and Martino, J. M. 2006. Brock Biology of Microorganisms. 11th ed. Pearson Prentice Hall, Upper Saddle River, NJ. Mahillon, J., and Chandler, M. 1998. Inserti on sequences. Microbiol. Mol. Biol. Rev. 62:725-774. Malamy, J. E., and Benfey, P. N. 1997. Organiza tion and cell differentiation in lateral roots of Arabidopsis thaliana Development 124:33-44. Malamy, J. E., and Ryan, K. S. 2001. Environm ental regulation of lateral root initiation in Arabidopsis Plant Physiol. 127:899-909. Markova, Y. A., Romanenko, A. S., and Dukhanina, A. V. 2005. Isolation of bacteria of the family Enterobacteriaceae from pl ant tissues. Microbiology 74:575-578. Matos, A., and Garland, J. L. 2005. Effects of co mmunity versus single strain inoculants on the biocontrol of Salmonella and microbial community dynamics in alfalfa sprouts. J. Food Prot. 68:40-48. McClean, A. P. D. and Oberholzer, P. C. J. 1965. Citrus psylla, a vector of the greening disease of sweet orange. S. Afr. J. Agric. Sci. 8:298-298. Melloul, A. A., Hassani, L., and Rafouk, L. 2001. Salmonella contamination of vegetables irrigated with untreated wastewater. World J. Mi crobiol. Biotechnol. 17:207-209. Meng, X., Benson, K., Chada, K., Huff, E. J., a nd Schwartz, D. C. 1995. Optical mapping of bacteriophage-lambda clones using restricti on endonucleases. Nat. Genet. 9:432-438. Mikulasova, M., Vaverkova, S., Birosova, L., and Suchanova, M. 2005. Genotoxic effects of the hydroxycinnamic acid derivatives caffeic, ch lorogenic and cichoric acids. Biologia 60:275-279. Molina, M. A., Ramos, J. L., and Espinosa-Urgel, M. 2006. A two-partner secretion system is involved in seed ad root co lonization and iron uptake by Pseudomonas putida KT2400. Environ. Microbiol. 8:639-647. Monier, J. M., and Lindow, S. E. 2005. Aggregates of resident bacteria facilitate survival of immigrant bacteria on leaf surfaces. Microb. Ecol. 49:343-352. Muirhead, R. W., Collins, R. P., and Bremer, P. J. 2006. Interaction of Escherichia coli and soil particles in runoff. Appl. E nviron. Microbiol. 72:3406-3411. 152

PAGE 153

Nakabeppu, Y., and Sekiguchi, M. 1986. Regulator y m echanisms for induction of synthesis of repair enzymes in response to alkylating agen ts: Ada protein acts as a transcriptional regulator. Proc. Nat. Acad. Sci. USA 83:6297-6301. Nieto, K. F., and Frankenberger, W. T. Jr. 1989. Biosynthesis of cytokinins by Azotobacter chroococcum Soil Biol. Biochem. 21:967-972. Nodzon, L. A., Xu, W. H., Wang, Y ., Pi, L. Y., Chakrabarty, P. K., and Song, W. Y. 2004. The ubiquitin ligase XBAT32 regulates lateral root development in Arabidopsis Plant J. 40:996-1006. Ohman, D. E., Sadoff, J. C., and Iglewski, B. H. 1980. Toxin A-deficient mutants of Pseudomonas aeruginosa PA103: isolation and charac terization. Infect. Immun. 28:899-908. Opelt, K., Berg, C., and Berg, G. 2007. The bryophyte genus Sphagnum is a reservoir for powerful and extraordinary antagonists and potentially facultative human pathogens. FEMS Microbiol. Ecol. 61:38-53. Ostroff, R. M., and Vasil, M. L. 1987. Id entification of a new phospholipase C activity by analysis of an insertional mutation in the hemolytic phospholipase C structural gene of Pseudomonas aeruginosa. J. Bacteriol. 169:4597-4601. Overbeek, R., Begley, T., Butler, R. M., Choudhuri, J. V., Chuang, H. Y., Cohoon, M., de Crcy-Lagard, V., Diaz, N., Disz, T., Edwards, R., Fonstein, M., Frank, E. D., Gerdes, S., Glass, E. M., Goesmann, A., Hanson, A., Iw ata-Reuyl, D., Jensen, R., Jamshidi, N., Krause, L., Kubal, M., Larsen, N., Linke, B ., McHardy, A. C., Meyer, F., Neuweger, H., Olsen, G., Olson, R., Osterman, A., Portnoy, V., Pusch, G. D., Rodionov, D. A., Rckert, C., Steiner, J., Stevens, R., Th iele, I., Vassieva, O., Ye, Y., Zagnitko, O., and Vonstein, V. 2005. The subsystem approach to genome annota tion and its use in the project to annotate 1000 genomes. Nucleic Acids Res. 33:5691-5702. Park, S. J., and Gunsalus, R. P. 1995. O xygen, iron, carbon, and superoxide control of the fumerase fumA and fumC genes of Escherichia coli: Role of the acrA, fnr, and soxR gene products. J. Bacteriol. 177:6255-6262. Patten, C. L., and Glick, B. R. 1996. Bacterial biosynthesis of indole-3acetic acid. Can. J. Microbiol. 42:207-220. Patten, C. L., and Glick, B. R. 2002. Role of Pseudomonas putida indoleacetic acid in development of the host plant root system Appl. Environ. Microbiol. 68:3795-3801. Paula, M. A., Reis, V. M., and Dbereiner, J. 1991. Interactions of Glomus clarum with Acetobacter diazotrophicus in infection of sweet potato ( Ipomoea batatas ), sugarcane ( Saccharum spp.), and sweet sorghum ( Sorghum vulgare ). Biol. Fert. Soils 11:111-115. 153

PAGE 154

Pedrosa, F., and Consortium G. 2005. A report on the genome of Herbaspirillum seropedicae strain Z78. Pages 111-114 in: Biological Ni trogen Nixation, Sustainable Agriculture and the Environment; Proceedings of the 14th International Nitrogen Fixation Congress. Beijing, People's Republic of China. October 27 November 01, 2004. Springer, Dordrecht, Netherlands. Peek, J.A., and Taylor, R. K. 1992. Characterization of a periplasmic thiol:disulfide interchange protein required for the functional matura tion of secreted virulence factors of Vibrio cholerae. Proc. Natl. Acad. Sci. USA 89:6210-6214. Persello-Cartieaux, F., David, P., Sarrobert, C ., Thibaud, M. C., Achouak, W., Robglia, C., and Nussaume, L. 2001. Utilization of mutant s to analyze the interaction between Arabidopsis thaliana and its naturally root-associated Pseudomonas. Planta 212:190-198. Pieterse, C. M. J., van Wees, S. C. M., van Pelt, J. A., Knoe ster, M., Laan, R., Gerrits, N., Weisbeek, P. J., and van Loon, L. C. 1998. A novel signaling pathway controlling induced systemic resistance in Ar abidopsis. Plant Cell 10:1571-1580. Plotnikova, J. M., Rahme, L. G., and Ausube l, F. M. 2000. Pathogenesis of the human opportunistic pathogen Pseudomonas aeruginosa PA14 in Arabidopsis Plant Physiol. 124:1766-1774. Polz, M. F. and Cavanaugh, C. M. 1998. Bias in template-to-product ratios in multitemplate PCR. Appl. Environ. Mi crobiol. 64:3724-3730. Rahme, L. G., Ausubel, F. M., Cao, H., Dr enkard, E., Goumnerov, B. C., Lau, G. W., Mahajan-Miklos, S., Plotnikova, J., Tan, M. W., Tsongalis, J., Walendziewicz, C. L., and Tompkins, R. G. 2000. Plants and animals share functionally common bacterial virulence factors. Proc. Natl. Ac ad. Sci. USA 97:8815-8821. Rahme, L. G., Stevens, E. J., Wolfort, S. F., Shao, J., Tompkins, R. G., and Ausubel, F. M. 1995. Common virulence factors for bacterial pathogenicity in plants and animals. Science 268:1899-1902. Ramirez-Chavez, E., Lopez-Bucio, J., Herrera-E strella, L., and Molina-Torres, J. 2004. Alkamides isolated from plants promote growth and alter root development in Arabidopsis Plant Physiol. 134:1058-1068. Ranjard, L., Poly, F., Lata, J. C., Mougel, C., Thioulouse, J., and Nazaret, S. 2001. Characterization of bacterial and fungal soil communities by automated ribosomal intergenic spacer analysis fingerprints: Bi ological and methodological variability. Appl. Environ. Microbiol. 67:4479-4487. Raskin, I. 1992. Salicylate, A New Pl ant Hormone. Plant Physiol. 99:799-803. 154

PAGE 155

Ray, S. K., Rajeshwari, R., Shar ma, Y., and Sonti, R. V. 2002. A high-molecular-weight outer membrane protein of Xanthomonas oryzae pv. oryzae exhibits similarity to nonfimbrial adhesins of animal pathogeni c bacteria and is required for optimum virulence. Mol. Microbiol. 46:637-647. Regu, M., Enfedaque, J., Camprubi, S., and Toms, J. M. 1992. The O-antigen lipopolysaccharide is the major barrier to plasmid DNA uptake by Klebsiella pneumoniae during transformation by electroporation and os motic shock. J. Microbiol. Methods 15:129-134. Reinhold-Hurek, B., Maes, T., Gemmer, S., Van Montagu, M., and Hurek, T. 2006. An endoglucanase is involved in infection of rice roots by the not-cellulose-metabolizing endophyte Azoarcus sp strain BH72. Mol. Plan t-Microbe Interact. 19:181-188. Reiter, B., Brgmann, H., Burg, K., and Sessitsch, A. 2003. Endophytic nifH gene diversity in African sweet potato. Can. J. Microbiol. 49:549-555. Reslewic, S., Zhou, S., Place, M., Zhang, Y. P., Briska, A., Goldstein, S., Churas, C., Runnheim, R., Forrest, D., Lim, A., Lapidus, A., Han, C. S., Roberts, G. P., and Schwartz, D. C. 2005. Whole-genome shotgun optical mapping of Rhodospirillum rubrum Appl. Environ. Microbiol. 71:5511-5522. Rich, J. J., Kinscherf, T. G., Kitten, T., and W illis, D. K. 1994. Genetic evidence that the gacA gene encodes the cognate response regulator for the lemA sensor in Pseudomonas syringae J. Bacteriol. 176:7468-7475. Riggs, P. J., Chelius, M. K., Iniguez, A. L., Kaeppler, S. M., and Tr iplett, E. W. 2001. Enhanced maize productivity by inoculation w ith Diazotrophic bacter ia. Aust. J. Plant Physiol. 28:829-836. Riggs, P. J., Moritz, R. L., Chelius, M. K., Dong, Y., Iniguez, A. L., Kaeppler, S. M., Casler, M. D., and Triplett, E. W. 2002. Isolation and characterization of diazotrophic endophytes from grasses and their effects on plant growt h. Pages 263-267 in: Nitrogen fixation: Global perspectives: Proceedings of the 13th International Congress on Nitrogen Fixation. Hamilton, Ontario, Canada. July 2-7, 2001. CABI Publishing, New York, NY. Robertson, L. J., Johannessen, G. S., Gjerde, B. K., and Loncarevic, S. 2002. Microbiological analysis of seed sprouts in Norway. Int. J. Food Microbiol. 75:119-126. Roesch, L. F., Fulthorpe, R. R., Riva, A., Casella G., Hadwin, A. K. M., Kent, A. D., Daroub, S. H., Camargo, F. A. O., Farmerie, W. G., and Triplett, E. W. 2007. Pyrosequencing enumerates and contrasts soil microbial diversity. ISME J. 1:283-290. Rojas, C. M., Ham, J. H., Deng, W. L., Doyle, J. J., and Collmer, A. 2002. HecA, a member of a class of adhesins produced by diverse pathoge nic bacteria, contribute s to the attachment, aggregation, epidermal cell kill ing, and virulence phenotypes of Erwinia chrysanthemi EC16 on Nicotiana clevelandii seedlings. Proc. Natl. A cad. Sci. USA 99:13142-13147. 155

PAGE 156

Rosenblueth, M., Martinez, L., Silva, J., and Martinez-Rom ero, E. 2004. Klebsiella variicola a novel species with clinical and plant-associated isolates. Sy st. Appl. Microbiol. 27:27-35. Rosenblueth, M., and Martinez-Romero, E. 2006. Bacterial endophytes a nd their interactions with hosts. Mol. Plant-Mi crobe Interact. 19:827-837. Rupp, W. D., Sancar, A., and Sancar, G. B. 1982. Propertie s and regulation of the UvrABC endonuclease. Biochimie 64:595-598. Ryu, C. M., Fang, M. A., Hu, C. H., Reddy, M. S., We i, H. X., Pare, P. W., and Kloepper, J. W. 2003. Bacterial volatiles promote growth in Arabidopsis Proc. Natl. Acad. Sci. USA. 100:4927-4932. Sagaram, U. S., DeAngelis, K. M., Trivedi, P., Andersen, G. L., Lu, S. E., and Wang, N. 2009. Bacterial diversity analysis of Huanglongbing pathogen-infected citr us, using PhyloChip arrays and 16S rRNA gene clone library sequencing. Appl. Environ. Microbiol. 75:1566-1574. Saitou, N. and Nei, M. 1987. The neighbor-joi ning method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406-425. Sevilla, M., Burris, R. H., Gunapala, N., a nd Kennedy, C. 2001. Comparison of benefit to sugarcane plant growth and 15N2 incorporation following inocul ation of sterile plants with Acetobacter diazotrophicus wild-type and Nifmutant st rains. Mol. Plant-Microbe. Interact. 14:358-366. Shevchik, V. E., Bortoli-German, I., Robe rt-Baudouy, J., Robinet, S., Barras, F., and Condemine, G. 1995. Differential effect of dsbA and dsbC mutations on extracellular enzyme secretion in Erwinia chrysanthemi Mol. Microbiol. 16:745-753. Solomon, E. B., and Matthews, K. R. 2005. Us e of fluorescent microspheres as a tool to investigate bacterial inter actions with growing plants J. Food Prot. 68:870-873. Solomon, E. B., and Matthews, K. R. 2006. Interaction of live and dead Escherichia coli O157: H7 and fluorescent microspheres with lettu ce tissue suggests bacterial processes do not mediate adherence. Lett. Appl. Microbiol. 42:88-93. Solomon, E. B., Yaron, S., and Matthews, K. R. 2002. Transmission of Escherichia coli O157: H7 from contaminated manure and irrigati on water to lettuce plant tissue and its subsequent internalization. Appl Environ. Microbiol. 68:397-400. Song, I., Stine, S. W., Choi, C. Y., and Gerba, C. P. 2006. Comparison of crop contamination by microorganisms during subsurface drip and furrow irrigation. J. Environ. Eng. ASCE 132:1243-1248. Staswick, P. E., Su, W., and Howell, S. H. 1992. Methyl jasmonate inhibition of root growth and induction of a leaf pr otein are decreased in an Arabidopsis thaliana mutant. Proc. Natl. Acad. Sci. U.S.A. 89:6837-6840. 156

PAGE 157

Struve, C., Forestier, C., and Krogfelt, K. A. 2003. Application of a novel m ulti-screening signature-tagged mutagenesis assay for identification of Klebsiella pneumoniae genes essential in colonization and in fection. Microb iology 149:167-176. Surette, M. A., Sturz, A. V., Lada, R. R ., and Nowak, J. 2003. Bacterial endophytes in processing carrots ( Daucus carota L. var. sativus ): their localization, population density, biodiversity and their effects on pl ant growth. Plant Soil 253:381-390. Tamura, K., Dudley, J., Nei, M., and Kumar, S. 2007. MEGA4: Molecular evolutionary genetics analysis (MEGA) software ve rsion 4.0. Mol. Biol. Evol. 24:1596-1599. Tapia-Hernandez, A., BustillosCristales, M. R., Jimenez-Salgado, T., Caballero-Mellado, J., and Fuentes-Ramirez, L. E. 2000. Natural endophytic occurrence of Acetobacter diazotrophicus in pineapple plants. Microb. Ecol. 39:49-55. Tatineni, S., Sagaram, U. S., Gowda, S., Robertson, C. J., Dawson, W. O., Iwanami, T., and Wang, N. 2008. In planta distribution of Candidatus Liberibacter asiaticus as revealed by polymerase chain reaction (PCR) and re al-time PCR. Phytopathology 98:592-599. Teixeira, D. C., Saillard, C., Eveillard, S., Danet, J. L., da Costa, P. I., Ayres, A. J., and Bov, J. 2005. Candidatus Liberibacter americanus', asso ciated with citrus huanglongbing (greening disease) in Sao Paulo State, Brazi l. Int. J. Syst. Evol. Microbiol. 55:1857-1862. Teixeira, D. C., Wulff, E. C., Martins, E. W., Kitajima R., Bassanezi, R., Ayres, A. J., Eveillard, S., Saillard, C., and Bov, J. M. 2008. A phytoplasma closely related to pigeon pea witches'-broom phytoplasma (16Sr IX) is asso ciated with citrus huanglongbing symptoms in the state of So Paulo, Br azil. Phytopathology 98:977-984. Tien, T. M., Gaskins, M. H., and Hubbell, D. H. 1979. Plant growth substances produced by Azospirillum brasilense and their effect on the growth of pearl millet ( Pennisetum americanum ). Appl. Environ. Microbiol. 37:1016-1024. Tyler, H. L., Roesch, L. F. W., Gowda, S., Dawson, W. O., and Triplett, E. W. 2009. Confirmation of the sequence of Candidatus Liberibacter asiaticu s and assessment of microbial diversity in Huanglongbing-infected citrus phloem using a metagenomic approach. Mol. Plant-Micr obe Interact. In press. Tyler, H. L., and Triplett, E. W. 2008. Plan ts as a habitat for be neficial and/or human pathogenic bacteria. Annu. Rev. Phytopathol. 46:53-73. Timmusk, S., Nicander, B., Granhall, U., a nd Tillberg, E. 1999. Cytokinin production by Paenibacillus polymyxa Soil Biol. Biochem. 31:1847-1852. Torres, A. G., Jeter, C., Langley, W., and Ma tthysse, A. G. 2005. Differential binding of Escherichia coli O157: H7 to alfalfa, human epithelia l cells, and plastic is mediated by a variety of surface structures. Appl. Environ. Microbiol. 71:8008-8015. 157

PAGE 158

Ueda, Y., Yum oto, N., Tokushige, M., Fukui, K., and Ohya-Nishiguchi, H. 1991. Purification and characterization of two types of fumarase from Escherichia coli J. Biochem. 109:728-733. Vanneste, S., De Rybel, B., Beemster, G. T. S., Ljung, K., De Smet, I., Van Insterdael, G., Naudts, M., Iida, R., Gruissem, W., Tasaka, M., Inze, D., Fukaki, H., and Beeckman, T. 2005. Cell cycle progression in the peri cycle is not sufficient for SOLITARY ROOT/IAA14-mediated lateral root initiation in Arabidopsis thaliana Plant Cell 17:3035-3050. Vasilieva, S. V., and Moschkovskaya, E. J. 2005. Quasi-adaptive response to alkylating agents in Escherichia coli : A new phenomenon. Russ. J. Genet. 41:484-489. Wachtel, M. R., Whitehand, L. C., and Mandrell, R. E. 2002. Association of Escherichia coli O157: H7 with preharvest leaf lettuce upon exposure to contaminated irrigation water. J. Food Prot. 65:18-25. Wang, C. X., Knill, E., Glick, B. R., and Defago, G. 2000. Effect of transferring 1-aminocyclopropane-1-carboxylic acid (ACC) deaminase genes into Pseudomonas fluorescens strain CHA0 and its gacA derivative CHA96 on thei r growth-promoting and disease-suppressive capacities. Can. J. Microbiol. 46:898-907. Wang, E. T., Tan, Z. Y., Guo, X. W., RodriguezDuran, R., Boll, G., and Martinez-Romero, E. 2006. Diverse endophytic bacteria is olated from a leguminous tree Conzattia multiflora grown in Mexico. Arch Microbiol. 186:251-259. Warriner, K., Spaniolas, S., Dickinson, M., Wright C., and Waites, W. M. 2003. Internalization of bioluminescent Escherichia coli and Salmonella Montevideo in growing bean sprouts. J. Appl. Microbiol. 95:719-727. Watarai, M., Tobe, T., Yoshikawa, M., and Sasakawa, C. 1995. Disulfide oxidoreductase activity of Shigella flexneri is required for release of Ipa proteins and invasion of epithelial cells. Proc. Natl. Aca d. Sci. USA 92:4927-4931. Weinstein, M., Roberts, R. C., and Helinski, D. R. 1992. A region of the broad-host range plasmid RK2 causes stable in plan ta inheritance of plasmids in Rhizobium meliloti cells isolated from alfalfa root nodul es. J. Bacteriol. 174:7486-7489. Williams, A. P., Avery, L. M., Killham, K., and Jones, D. L. 2007. Survival of Escherichia coli O157: H7 in the rhizosphere of maize grown in waste-amended soil. J. Appl. Microbiol. 102:319-326. Williamson, L. C., Ribrioux, S. P. C. P., Fitte r, A. H., and Ottoline-Leyser, H. M. 2001. Phosphate availability regulates root system architecture in Arabidopsis Plant Physiol. 126:875-882. 158

PAGE 159

W ilmoth, J. C., Wang, S., Tiwari, S. B., Joshi, A. D., Hagen, G., Guilfoyle, T. J., Alonso, J. M., Ecker, J. R., and Reed, J. W. 2005. NPH4 /ARF7 and ARF19 promote leaf expansion and auxin-induced lateral root formation. Plant J. 43:118-130. Wilson, A. K., Pickett, F. B., Turner, J. C., and Estelle, M. 1990. A dominant mutation in Arabidopsis confers resistance to auxin, ethylene, and abscisic acid. Molecular and General Genetics. 222:377-382. Wood, B. E., Yomano, L. P., York, S. W., a nd Ingram, L. O. 2005. Development of industrial-medium-required elim ination of the 2,3-butanediol fermentation pathway to maintain ethanol yield in an ethanologenic strain of Klebsiella oxytoca Biotechnol. Prog. 21:1366-1372. Wu, C. W., Schramm, T. M., Zhou, S., Schwartz, D. C., and Talaat, A. M. 2009. Optical mapping of the Mycobacterium avium subspecies paratuberculosis genome. BMC Genomics 10:25. Xie, Q., Frugis, G., Cogan, D., and Chua, N. H. 2000. Arabidopsis NAC1 transduces auxin signal downstream of TIR1 to promote lateral root development. Genes Dev. 14:3024-3036. Xie, Q., Guo, H. S., Dallman, G., Fang, S., Wei ssman, A. M., and Chua, N. H. 2002. SINAT5 promotes ubiquitin-related degr adation of NAC1 to attenuate auxin signals. Nature 419:167-170. Yorgey, P., Rahme, L. G., Tan, M. W., and Ausubel, F. M. 2001. The roles of mucD and alginate in the virulence of Pseudomonas aeruginosa in plants, nematodes and mice. Mol. Microbiol. 41:1063-1076. Zelenev, V. V., van Bruggen, A. H. C., and Semenov, A. M. 2005. Short-term wavelike dynamics of bacterial populations in response to nutrient input from fresh plant residues. Microb. Ecol. 49:83-93. Zhang, H., Jennings, A., Barlow, P. W., and Forde, B. G. 1999. Dual pathways for regulation of root branching by nitrate. Proc. Natl. Acad. Sci. USA. 96:6529-6534. Zhou, S., Bechner, M. C., Place, M., Churas, C. P., Pape, L., Leong, S. A., Runnheim, R., Forrest, D. K., Goldstein, S., Livny, M., and Sc hwartz, D. C. 2007. Validation of rice genome sequence by optical mapping. BMC Genomics 8:278. Zhou, S., Deng, W., Anantharaman, T. S., Lim, A., Dimalanta, E. T., Wang, J., Wu, T., Chunhong, T., Creighton, R., Kile, A., Kvikstad, E., Bechner, M., Yen, G., Garic-Stankovic, A., Severin, J., Forrest, D ., Runnheim, R., Churas, C., Lamers, C., Perna, N. T., Burland, V., Blattner, F. R., Mishra B., and Schwartz, D. C. 2002. A wholegenome shotgun optical map of Yersinia pestis strain KIM. Appl. Environ. Microbiol. 68:6321-6331. 159

PAGE 160

160 Zhou, S., Kile, A., Bechner, M., Place, M., Kvikstad, E., Deng, W., Wei, J., Severin, J., Runnheim, R., Churas, C., Forrest, D., Dimalanta, E. T., Lamers, C., Burland, V., Blattner, F. R., and Schwartz, D. C. 2004a. Singlemolecule approach to bacterial genomic comparisons via optical mapping. J. Bacteriol. 186:7773-7782. Zhou, S., Kile, A., Kvikstad, E., Bechner, M., Se verin, J., Forrest, D., Runnheim, R., Churas, C., Anantharaman, T. S., Myler, P., Vogt, C., Iv ens, A., Stuart, K., and Schwartz, D. C. 2004b. Shotgun optical mapping of the entire Leishmania major Friedlin genome. Mol. Biochem. Parasitol. 138:97-106.

PAGE 161

BIOGRAPHICAL SKETCH Heather Tyler was born and raised in Lake Wa les, Florida. She graduated from Lake Wales High School as co-salutatorian in the spring of 1999 and went on to attend Florida Southern College that fall, where she majore d in biology and minored in chemistry. While attending Florida Southern, Heather became a memb er of the Phi Eta Sigma and Beta Beta Beta honor societies. She also participated in the scho ols chapter of Habitat for Humanity, serving as secretary her senior year. Heather graduated summa cum laude from Florida Southern College in the spring of 2003. That fall, she went on to pursue graduate studies in microbiology at the University of Florida, where her research focused on plant-associated bact eria and their roles in altering and enhancing plant growth. The goal of her work in gra duate school has been to gain knowledge that will ultimately aid in increa sing production of food crops and reducing the dependence of these plants on ex ternally applied fertilizers, th ereby reducing the environmental impact of agricultural land. 161