Exploring the Genetic Basis for Host Specific Virulence and Pathogenicity in Erwinia Tracheiphila

Iowa State University Capstones, Theses and Graduate Theses and Dissertations Dissertations

2020

Exploring the genetic basis for host specific virulence and pathogenicity in Erwinia tracheiphila

Olakunle Olawole Iowa State University

Follow this and additional works at: https://lib.dr.iastate.edu/etd

Recommended Citation Olawole, Olakunle, "Exploring the genetic basis for host specific virulence and pathogenicity in Erwinia tracheiphila" (2020). Graduate Theses and Dissertations. 18577. https://lib.dr.iastate.edu/etd/18577

This Dissertation is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State University Digital Repository. It has been accepted for inclusion in Graduate Theses and Dissertations by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. Exploring the genetic basis for host specific virulence and pathogenicity in Erwinia tracheiphila

Olakunle I. Olawole

A dissertation submitted to the graduate faculty in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Major: Plant Pathology

Program of Study Committee: Gwyn A. Beattie, Major Professor Mark L. Gleason Steven A. Whitham Alison E. Robertson Kyaw J. Aung

The student author, whose presentation of the scholarship herein was approved by the program of study committee, is solely responsible for the content of this thesis. The Graduate College will ensure this thesis is globally accessible and will not permit alterations after a degree is conferred.

DEDICATION To God Almighty – my strength, shield and exceeding great reward; without You by my side, the journey wouldn’t have been this glorious. To my wife, Funmi, who sacrificially stood by me through the thick and thin of graduate school, often offering herself as a ‘cheap labor’ to help with data entry for my numerous plant studies; without you, I wouldn’t have come this far.

Thank you so much for your love, patience, understanding, prayers and encouragement during the darkest moments of this journey. To my parent, thank you for your sacrifice to ensure I have access to invaluable educational opportunities. To folks of the New Life Church and the United

Nations for Christ, thank you for your spiritual support, and for being such a great family to hang out with.

iii

TABLES OF CONTENT Page ACKNOWLEDGEMENTS v

ABSTRACT vi

CHAPTER 1. GENERAL INTRODUCTION 1 Dissertation organization 2 Literature review 3 Research rationale and significance 14 References 15

CHAPTER 2. SUBSPECIES DELINEATION OF ERWINIA TRACHEIPHILA 21 STRAINS ON THE BASIS OF WHOLE-GENOME SEQUENCE AND

PHYSIOLOGICAL DATA

Abstract 21 Introduction 22 Materials and methods 24 Results 34 Discussion 48 References 54

CHAPTER 3. THE EFFECTORS EOP1 AND DSPE ARE DRIVERS OF HOST- 58 SPECIFICITY AND PATHOGENICITY AMONG ERWINIA TRACHEIPHILA STRAINS Abstract 58 Introduction 59 Materials and methods 63 Results 70 Discussion 83 References 88

CHAPTER 4. ERWINIA TRACHEIPHILA SUBSPECIES DIFFERENCES IN HOST 94 SPECIFICITY ARE ASSOCIATED WITH HOST-SPECIFIC DIFFERENCES IN HRPA EXPRESSION Abstract 94 Combined Introduction, Results and Discussion 95 References 102

CHAPTER 5. IN PLANTA EXPRESSION PROFILING AND FUNCTIONAL 104 ANALYSIS OF ERWINIA TRACHEIPHILA EFFECTORS Abstract 104 Introduction 105 Materials and methods 108 Results 117 Discussion 129 134 References

CHAPTER 6. CONCLUSIONS AND FUTURE DIRECTIONS 141 References 146

ACKNOWLEDGEMENTS

A big thank you to my advisor, Dr. Gwyn A. Beattie, for her patience and support as a mentor for research, teaching, and professional development. I have learned a lot from your wealth of knowledge and experience in this field. Thank you so much! To my committee members Dr. Mark L. Gleason, Dr. Steven A. Whitham, Dr. Alison E. Robertson and Dr. Kyaw

J. Aung, thank you for your feedback and guidance.

I am grateful to all of the past and present colleagues in the Beattie and Halverson laboratories. Your suggestions and comments during joint laboratory meetings are greatly appreciated. Specifically, I would like to thank Dr. Chiliang Chen for his guidance on optimizing bacterial mutagenesis protocols which set a very good background for me to work on this project. I also appreciate the assistance of Dr. Benzhong Fu, Sharon Raquel Badilla

Arias and Kephas Mphande during my gazillion plant studies.

Finally, I would like to thank my funding sources – College of Agriculture and Life

Sciences, Iowa State University and the Iowa State University Presidential Fellowship.

ABSTRACT

Erwinia tracheiphila is among the few plant pathogenic species of Erwinia retained in this genus after several other species have been re-assigned to new genera. E. tracheiphila strains fall into clades that exhibit differences in host-specific virulence based on their ability to rapidly wilt muskmelon but not wilt squash (Et-melo clade) or to rapidly wilt squash and less rapidly wilt muskmelon (Et-C1 and Et-C2 clades). We characterized whole genome sequences and physiological traits of 24 E. tracheiphila strains in two of these clades, leading us to propose that the Et-melo and Et-C1 clades be delineated as E. tracheiphila subsp. tracheiphila and E. tracheiphila subsp. lentilata, respectively. Here, we explored how the genetic basis of E. tracheiphila sub-speciation is associated with host-specific virulence. We identified the distribution of the type III secreted effector (T3SE) repertoire of two host-specific strains, SCR3

(Et-melo) and BHKY (Et-C1), and characterized the expression profile of selected effectors in planta using RT-qPCR. We explored the role of one highly expressed effector, Eop1, as a host- specificity candidate, and found that loss of eop1 did not alter the virulence of two Et-melo strains or an Et-C1 strain on their respective hosts. However, over-expression of eop1 from an

Et-melo strain in an Et-C1 strain increased its virulence on muskmelon, but not on squash, indicating that Eop1 functions as a host-specific virulence factor. The expression profiling in planta also highlighted poor expression of a key T3SS pilus gene, hrpA, in Et-melo in squash.

We demonstrated experimentally that over-expression of hrpA in an Et-melo strain enabled this vii strain to infect squash, demonstrating that host-specific hrpA expression is another factor contributing to host-specific virulence in E. tracheiphila. We also examined the role of the effector DspE in pathogenicity, and found that loss of dspE from Et-melo strains significantly reduced, but did not eliminate, virulence on muskmelon, whereas loss of dspE from Et-C1 strains resulted in a complete loss of pathogenicity on squash and muskmelon. Thus, DspE has distinct roles in the two E. tracheiphila clades. This work represents the first characterization of the major molecular drivers of host specificity and pathogenicity among E. tracheiphila strains.

Finally, using pyramided mutant analysis, we demonstrated that, consistent with their expression profiles in planta, effectors DspE, OspG, Eop1 and AvrB4 contributed to the virulence of Et- melo on muskmelon, whereas DspE, OspG, Eop1, HopL1 and HopO1 contributed to the virulence of Et-C1 on squash and muskmelon, with highly overlapping functions among many of the effectors. This work thus illustrates the contribution of distinct sets of effectors to the virulence of E. tracheiphila strains on their hosts. This work sets the foundation for a better understanding of the molecular interaction between the E. tracheiphila pathosystem and its cucurbit hosts which may help guide mechanistic approaches that could lead to the development of durable resistant plant cultivars.

CHAPTER 1

GENERAL INTRODUCTION The genus Erwinia contains 44 validly published species that are included in the list of prokaryotic names with standing in nomenclature (LPSN), including both pathogenic and non- pathogenic Enterobacteria. Among these, only a few species are known to be plant pathogens, including Erwinia amylovora, E. pyrifoliae, E. psidii, E. papaya, E. aphidicola and E. tracheiphila. These are the causative agents of fire blight of rosaceous plants, bacterial shoot blight of pear in Asia, branch, flower and fruit rot in guava, bacterial crown rot of papaya and fruit spot of pepper and bacterial wilt of cucurbits, respectively. Other species range from commensals to epiphytes.

Unlike most plant pathogenic Erwinia species, which cause necrotic diseases on their respective hosts, E. tracheiphila causes a vascular disease. E. tracheiphila is a xylem pathogen that is vectored by cucumber beetles. It replicates in the plant xylem and can induce wilt.

Moreover, there is evidence that it can replicate in the digestive tract of its insect vectors.

Adaptation to the low-nutrient conditions in the xylem may be one factor driving the observed differences between E. tracheiphila and other plant-associated Erwinia species.

Insights from the approximately 5-Mb genome of E. tracheiphila revealed that it is distinct from those of other sequenced Erwinia species in exhibiting pseudogenization of about one-fifth of its coding sequences, influx of mobile genetic elements, and massive acquisition of horizontally transferred genes, including several putative virulence factors. These features suggest that E. tracheiphila recently evolved from a free-living lifestyle outside of the plant into its current lifestyle in which it is restricted to the xylem or vectors that transmit it. 2

Similar to the closely related and well-studied pathogen E. amylovora, E. tracheiphila strains infect their hosts (cucurbits from Cucumis or Cucurbita genera) in a host-specific manner.

Specifically, wilt symptoms often progress faster when seedlings are inoculated with strains that originated from the same crop host genus. However, while the molecular mechanism of host- specific virulence and pathogenicity has been reasonably established in E. amylovora, nothing is currently known of the virulence factors and mechanisms of host specificity in the E. tracheiphila pathosystem. This is the major focus of this project.

DISSERTATION ORGANIZATION

This dissertation features six chapters. Chapter one contains an introduction, literature review, and research rationale and significance. Chapter two presents the delineation of subspecies of Erwinia tracheiphila strains on the basis of whole genome sequence and physiological data analyses. Chapter three demonstrates the roles of the effectors Eop1 and DspE in host specific virulence and pathogenicity, respectively, among Erwinia tracheiphila strains.

Chapter four demonstrates how the lack of expression of a critical type III secretion protein in one Erwinia tracheiphila subspecies in the xylem of squash contributes to its lack of pathogenicity on squash. Chapter five describes the in planta expression profiling and functional analyses of E. tracheiphila effectors. Finally, chapter 6 presents general conclusions from these studies and future directions.

LITERATURE REVIEW

Erwinia tracheiphila, an economically important bacterial pathogen of cucurbits

The annual production of muskmelon (Cucumis melo), squash (Cucurbita spp.), cucumber (Cucumis sativa), pumpkin (Cucurbita pepo) and other cucurbits in the United States is valued at more than $1.8 billion (52). While production covers about a total of 113,000 ha of agricultural land, more than half is cultivated in the Northeastern and Midwestern regions.

Cucurbit cultivation is threatened by a number of bacterial diseases, including angular leaf spot of cucumber caused by Pseudomonas amygdali pv. lachrymans, bacterial fruit blotch of watermelon caused by Acidovorax citrulli, bacterial leaf spot of pumpkin caused by

Xanthomonas cucurbitae and bacterial wilt of cucurbits (BWC) caused by Erwinia tracheiphila.

Among these diseases, BWC is the most important because of its wide host range, as it infects muskmelon, cucumber, squash and pumpkin, although watermelon is highly resistant (11; 77).

Squash and pumpkin are native to North and South America (20; 34), whereas Cucumis species (muskmelon, cucumber and watermelon) were introduced into the United States from

India and Africa (65). However, the introduced Cucumis host plants species, especially cucumber and muskmelon, are the most susceptible to BWC (67). The production of susceptible cucurbit cultivars in the United States is severely threatened by BWC, which is mostly endemic in the regions (Northeast and Midwest) where more than half of the nation’s total acreage is cultivated, causing yield losses of up to 80% (59; 61).

Management of bacterial wilt primarily depends on frequent insecticide (mainly neonicotinoid) applications to reduce cucumber beetle population. However, because of the danger posed by these insecticides to insect pollinator health and insectivorous birds, the probability of their continued use remains very uncertain (19; 29; 32). A few alternative

4 management strategies can help manage bacterial wilt, including perimeter trap cropping, which entices beetles to an alternate crop (15), and use of row covers to prevent beetle feeding (42), among others. The most effective strategy for managing bacterial plant diseases is the use of resistant cultivars (77) and application of copper compounds (16). However, bacterial wilt- resistant cucurbit cultivars are currently not available (48), and although copper supplements get into plant xylem, they do not inhibit bacterial replication (28). Therefore, a better understanding of the molecular basis for virulence among E. tracheiphila strains could provide insights into identifying resistance genes in cucurbits that could be used to breed against bacterial wilt. Also, most farmers cultivate cucurbits on a small scale, usually in an organic system (14; 42), thus necessitating a need for more diverse management options that could be tailored towards different agricultural systems.

Symptoms and signs of bacterial wilt of cucurbits

Erwinia tracheiphila is a xylem pathogen, and the name “bacterial wilt” is derived from the characteristic wilting of infected cucurbits, which often moves from the leaves to the stems, followed by foliar necrosis and eventual collapse and death of entire plant (70). Wilt symptoms have historically been attributed to the occlusion of xylem vessels (38); however, our current understanding of the complexity of virulence among bacterial phytopathogens, including E. tracheiphila, has shown that pathogenesis is driven by more than a single factor among the xylem dwellers (74).

A diagnostic sign of infection includes visible strands of bacterial slime when wilting stems are cut from the crown and slowly pulled apart (38). This sign is more useful in diagnosing

E. tracheiphila-infected cucumber and muskmelon than in squash and pumpkin (62), where the

5 appearance of visible strands is not consistent. This sign also distinguishes BWC from the wilt symptoms induced by the soil-borne fungus Fusarium oxysporum formae speciales, which instead induces a light brown discoloration of the xylem vessels (2; 23).

Transmission of Erwinia tracheiphila

E. tracheiphila is transmitted by the striped or spotted cucumber beetles (Acalymma vittatum and Diabrotica undecimpunctata howardi, respectively) (57; 70). The striped cucumber beetles are cucurbit specialists, and as such they are the most important vectors of E. tracheiphila. They are also the primary overwintering reservoir of E. tracheiphila, and thus link infection from one season to another (57; 62). There is a positive correlation between the population density and behavior of the striped cucumber beetles and bacterial wilt epidemics, further reinforcing their importance as the primary vector (25). The spotted cucumber beetles which are polyphagous, are also found in cucurbit fields (24; 49).

Cucumber beetles generally survive the winter as adults and become active in the spring early in the growing season when established plants are more susceptible to bacterial wilt (43;

51). Transmission occurs when E. tracheiphila-infested frass from beetles make contact with feeding wounds or flower nectaries (40; 57; 64). Extracts of mouthparts and intestinal tracts of infested cucumber beetles have been demonstrated to cause wilt symptoms when inoculated to cucumber seedlings (57).

Impact of cucurbit growth stage on speed and severity of bacterial wilt

The speed and severity of symptom development and bacterial disease progression in plants are generally influenced by the disease triangle, which is an interplay between the host, pathogen and environment (33; 74). Among the host factors which include susceptibility, growth

6 stage, population density and general fitness (33), the impact of host growth stage has been found to be particularly critical to the progression of bacterial wilt. For instance, wilt symptoms occurred more rapidly in melon plants inoculated at 2 to 4 weeks than at 6 to 8 weeks after emergence (45). Also, the cotyledon stage of pumpkin was susceptible to bacterial wilt, whereas the 3-4 leaf stage was resistant (10; 73).

Concept of host specific-virulence among Erwinia tracheiphila strains

E. tracheiphila was one of the first bacterial phytopathogens ever described. As early as

1911, E. tracheiphila was reported to infect cucurbits in a host-specific manner; that is, strains isolated from cucumber showed a rapid wilt on cucumber, but slight or no symptoms when inoculated into squash or pumpkin (70). Over a hundred years after this first report, individual E. tracheiphila strains have been repeatedly shown to differ in their ability to induce a rapid wilt, or even a wilt, in distinct cucurbita genera.

Like the 1911 findings, Rojas and associates (63) confirmed that wilt symptoms progressed faster when seedlings were inoculated with strains that originated from the same crop host genus. The authors investigated genetic variability using rep-PCR among 69 E. tracheiphila strains isolated from 5 different host genera, and reported that strains fell into two distinct groups based on their rep-PCR fingerprints. Moreover, these fingerprints correlated with virulence patterns on squash and melon. Nazareno and Dumenyo (53) confirmed a similar incidence of host-specific virulence among E. tracheiphila strains.

Recently, Shapiro et al (67) showed that E. tracheiphila strains clustered into three phylogenetic clades, which they named Et-melo, Et-C1 and Et-C2 based on similarity across the genomes in draft genome sequences. All Et-melo strains were isolated from Cucumis host

7 species, whereas Et-C1 and Et-C2 strains were isolated from both Cucumis and Cucurbita host species. The authors found that the Et-melo strains only infected cucumber and melon but not squash, whereas the Et-C1 and Et-C2 strains infected squash but were reduced in virulence on melon and cucumber.

Despite the lines of evidence supporting host-specific virulence among E. tracheiphila strains, the mechanism behind such remains unknown. With no cucurbit cultivars known to be resistant to bacterial wilt (48), a better understanding of the mechanisms of host specific- virulence and pathogenesis among E. tracheiphila strains could be a step towards identifying and boosting resistance in cucurbits (7). The role of certain virulence factors that are secreted via the type III secretion system in host-specific virulence and pathogenicity among E. tracheiphila strains was explored in this project (Chapter 3).

The Type III secretion system in xylem-inhabiting pathogenic bacteria Type III secretion systems (T3SSs) are complex membrane-spanning bacterial structures inject secreted effector proteins directly into the host cell cytoplasm (1; 12; 17; 35). This system is essential for the virulence of a wide array of animal and plant pathogenic bacteria (12).

Phytopathogenic bacteria have T3SSs that can be grouped into at least three different

T3SS families based on phylogeny and synteny. These include the Hrp1 (hypersensitive response and pathogenicity) T3SS, which are found in Pseudomonas syringae and Erwinia spp; Hrp2

T3SS, which are found in Ralstonia solanacearum, Xanthomonas spp., Acidovorax and

Burkholderia; and a rhizobial-like Hrp3 T3SS which is found only within some P. syringae strains (1; 27). The majority of studies on the T3SS and their cognate effector proteins in bacterial virulence have been on necrotic pathogens including P. syringae and many

Xanthomonas spp., but the T3SS is also important to the virulence of some xylem pathogens including Xanthomonas oryzae and Ralstonia solanacearum (74).

Xylem pathogens are endogenous, living inside xylem cells or tracheary elements, as compared to exogenous bacteria which live in the apoplast or on leaf surfaces (9). Only three bacterial xylem pathogens are known to have a functional Hrp T3SS: Pantoea stewartii subsp. stewartii (18), Ralstonia solanacearum (4; 47) and Xanthomonas campestris pv. campestris (56).

These bacteria respectively cause Stewart’s bacterial wilt and leaf blight of maize, bacterial wilt of over 200 plant species of important economic crops, and crucifer black rot disease. Several other xylem pathogens do not encode a Hrp T3SS in their genomes. These include Xylella fastidiosa, the causal agent of pierce’s disease of grapes; Xanthomonas albineans, the causal agent of leaf scald of sugarcane; and Clavibacter michiganensis subsp. michiganensis (9; 22).

Pantoea stewartii subsp. stewartii has a second T3SS, a Salmonella pathogenicity island-

1 (SPI-1), called Pantoea secretion Island 2 (PSI-2), which is required for persistence in the flea beetle vector but not for pathogenesis of maize (18). Xanthomonas albineans similarly has a SPI-

1 type T3SS, but not a Hrp T3SS (54). The E. tracheiphila genome encode both a SPI-1 T3SS and a Hrp T3SS (67). Work in this thesis examined the functional role of the Hrp T3SS in E. tracheiphila, but whether the SPI-1 T3SS in E. tracheiphila has a role in enhancing its persistence inside the digestive tract of the cucumber beetle vectors remains to be addressed.

Candidate effectors relevant to the E. tracheiphila pathosystem

Effectors are defined as proteins secreted and delivered into the host cells through the

T3SS, and are thus called type III secreted effectors (T3SEs) (12). They are known to have target molecules within plants, and can be recognized by host resistance (R) genes to trigger a defense

9 response against the pathogen (17). Members of the Xanthomonas spp. and Ralstonia solanacearum have a special type of effector called transcription activator-like effectors

(TALEs) which localize to the host nucleus to activate genes that contribute to disease or turn off genes contributing to defense (49).

The genomes of these xylem pathogens encode varying numbers of T3SEs, ranging from

70 to 75 in Ralstonia solanacearum (50; 55) to 47 in Xanthomonas campestris pv. campestris

(30), 23 in E. tracheiphila (67) and one in Pantoea stewartii subsp. stewartii (31). E. tracheiphila contains effectors that have been implicated as host-specific virulence and pathogenicity factors in other pathosystems. However, nothing is currently known about the functions of these effectors in the E. tracheiphila pathosystem. The two most important effectors in the closely-related pathogen E. amylovora are Eop1 (Erwinia outer protein-1) and DspE

(disease specific protein-E), which function as host-range limiting (5) and pathogenicity (8) factors, respectively. However, while the E. amylovora genome encodes about four T3SS effectors (13), the E. tracheiphila genome suggests the presence of at least 23 effectors, including Eop1, DspE, OspG, AvrB4, HopL1, and HopO1 among others (67).

Similar to the E. tracheiphila pathosystem, strains of E. amylovora fall into two groups based on host-range, including the Spiraeoideae and Rubus strains. Spiraeoideae strains have a broad host range and can infect plants in many rosaceous genera, including apple and pear; while

Rubus strains exclusively infect plants from the genus Rubus, especially raspberry and blackberry. Based on unusually divergent Eop1 sequences between both groups, the authors investigated the role of Eop1 in host specificity and found that an eop1-deficient mutant from the

Rubus strains became virulent to immature pear fruit (5), demonstrating that Eop1 has a role in restricting the ability of E. amylovora to infect a specific host.

A comparison of the Eop1 sequences of Cucumis- and Cucurbita-derived strains of E. tracheiphila showed sequence differences associated with strains that differ in their virulence on distinct Cucurbita genera (44) . This association coupled with the role of Eop1 in host-specific restriction of virulence in E. amylovora suggests that Eop1 could have a role in determining host specific virulence in E. tracheiphila. We looked at this further in this thesis.

Eop1 is in the YopJ family of type III secreted effectors

Eop1 belongs to the YopJ/AvrRxv family of type III secreted T3SEs and homologues are found across several species of the Proteobacteria, including Pseudomonas (HopZ),

Xanthomonas (AvrBsT, AvrRxv, AvrXv4 and XopJ), Erwinia (Eop1), Ralstonia (PopP1), and the plant symbiont Rhizobium (Y4LO) (41; 58).

Although Xanthomonas campestris pathovars contain several YopJ homologues, two

(AvrBsT and AvrXv4) have been demonstrated to have roles that are consistent with host- specificity. AvrBsT has an avirulence function in pepper isogenic lines that carry the Bs1 resistance gene (47), whereas AvrXv4 from X. campestris pv. vesicatoria has an avirulence function in wild tomato and tobacco (58). In contrast, Xanthomonas campestris pv. vesicatoria

XopJ has a virulence function in tobacco (71).

YopJ homologues in Pseudomonas syringae include three major forms: HopZ1, HopZ2 and HopZ3, with at least one of these, HopZ1 diversified into three distinct forms (HopZ1a,

HopZ1b and HopZ1c) (46). HopZ2 is closely related to the Xanthomonas YopJ homologues, while HopZ3 is more closely related to those of Erwinia (46). Interestingly, similar to the role of

AvrBsT and AvrXv4 in Xanthomonas, HopZ2 functions as an avirulence gene in common bean cultivars (3). Whether HopZ3 functions as a determinant of host-specificity or avirulence among

Pseudomonas syringae pathovars is currently unknown, although it triggers defense responses in common bean seedlings, which in turn can be suppressed by other effectors when delivered from heterologous strains (60).

Ralstonia solanacearum is the only member of the β-proteobacteria whose YopJ homologue (PopP1) has been functionally characterized. PopP1 functions in R-gene mediated resistance, as a PopP1-deficient strain gained the ability to cause disease on a Petunia line for which the wild-type strain wouldn’t normally cause disease (39). The role of the Rhizobium

YopJ homolog Y4LO is yet unknown. A group initially reported it as a symbiotic determinant involved in the differentiation of symbioses in Rhizobium sp. strain NGR234 (α-proteobacteria)

(75), but it was later retracted due to the deletion of the wrong gene (76).

The genetic basis for host-specific virulence among E. tracheiphila strains was explored in this project by investigating the role of Eop1 either as a host-limiting factor as in E. amylovora, or as a putative avirulence gene as in Ralstonia solanacearum, Pseudomonas syringae and Xanthomonas campestris.

The AvrE superfamily of type III secreted effectors includes critical pathogenicity factors among phytopathogenic bacteria

Almost all type III-dependent phytopathogenic bacteria encode members of the AvrE superfamily of T3SEs (21). These AvrE homologues are core effectors and are known to be required for, or significantly contribute to, the virulence of these bacteria (21).

WtsE is the homologue of AvrE in Pantoea stewartii subsp. stewartii. It is the only Hrp-

T3SE known to be secreted by this bacterial pathogen, and it is absolutely required for its pathogenicity based on that a WtsE-deficient mutant became non-pathogenic on sweetcorn (31).

Deleting AvrE homologues in E. amylovora (8) and Pectobacterium carotovorum (36) also resulted in a non-pathogenic phenotype in these bacterial plant pathogens. However, deleting

AvrE homologues only reduced the virulence of Pseudomonas syringae (6), Ralstonia solanacearum (34) and Xanthomonas campestris pv. campestris (30) on their respective hosts.

Effectors of this family are very large proteins of about 192Da, which may also contribute to their essential roles among these plant bacteria (21). Like many effectors, most members of this effector family are associated with a type III secreted chaperone gene, which is required for efficient folding, stability and secretion of the effector (26). However, the gene for a chaperone is lacking for AvrE homologues in Ralstonia solanacearum (PopS) and Xanthomonas campestris pv. campetris (XopAM), suggesting that they may be utilizing other chaperones elsewhere within their genomes for stabilization and secretion (21).

The role of DspE and its chaperone DspF in the pathogenesis of the E. tracheiphila- cucurbit system is addressed in this thesis. Many of the effectors of individual bacterial strains often have incremental effects, such that the deletion of one does not result in a different phenotype from the wide-type strain (37). The incremental effect of E. tracheiphila effectors on their melon and squash is also explored in this thesis.

Genomic resources for Erwinia tracheiphila

Complete and draft genome sequences of phytopathogenic bacteria, including E. tracheiphila, are available due to the advent of next-generation sequencing. The publicly available genome sequences of E. tracheiphila strains include the PacBio-generated sequences of both a Cucurbita-derived strain, BuffGH (68), and a Cucumis-derived strain, MDCuke (66), and five illumina-generated draft sequences of Cucurbita strains (69). BuffGH was assembled into

13 seven contigs, whereas MDCuke assembled into five contigs. The Illumina-generated draft sequences of the five Cucurbita strains consisted of between 1,140 and 3,334 contigs, which were de novo assembled (69).

A comprehensive description of the genome and draft genome of the Cucurbita-derived strains revealed a high proportion of horizontally-acquired genes and an increased mobile element invasion and a greater expansion of gene content than other closely related species of

Erwinia (69).. The authors also showed that about one-fifth of the coding sequences of these genomes may have lost their functions as a result of truncation by pseudogenization.

Chromosomal rearrangements associated with high number of phage and transposable elements within E. tracheiphila strains were reported as major drivers of differences in genetic architecture between E. tracheiphila versus E. amylovora and E. billingiae strains (69).

Recently (67), a more comprehensive sequencing approach was used to confirm the previous work (63) on genetic diversity among E. tracheiphila strains. The authors performed a phylogenomic analysis on the draft genome sequences of 88 E. tracheiphila isolates collected from the Northeastern and Midwestern regions of the United States. The analysis clustered all of the strains into three distinct phylogenetic lineages, designated the Et-melo, Et-C1 and Et-C2 clades (67).

The low diversity of the genome sequences of strains within each E. tracheiphila clade suggests that the clades are monomorphic (67; 69). The E. tracheiphila genome is completely missing genes for the clustered regularly interspaced short palindromic repeats (CRISPR/Cas system), which most bacteria use as a defense mechanism against foreign DNA. This may explain the reason for the influx of phage genes into its genome as compared to other closely related Erwinia species (69).

The post-genomics era offers a greater possibility of a better understanding of the genetic basis for host-specific virulence and pathogenicity among E. tracheiphila strains, even as more genomes are being sequenced and made publicly available. Unlike most members of the

Enterobacteria such as Escherichia coli, the genetic tractability of E. tracheiphila has been difficult as most mutagenesis protocols needed optimization for this species. This thesis work involved optimizing the molecular tools, resulting in our ability to generate deletion mutants and pyramided deletion mutants in E. tracheiphila. The current knowledge of genetic variability among E. tracheiphila strains supports the potential sub-speciation of these strains, an activity that requires a more comprehensive comparison of their physiology and genome sequences. The second chapter of this thesis focuses on delineation of subspecies within the E. tracheiphila species, and the fourth chapter explores differences between two subspecies in their repertoire of functional T3SS effectors.

RESEARCH RATIONALE AND SIGNIFICANCE

Cucurbits are an economically important crop in the United States. Annually, over 110,000 ha of melon, cucumber, pumpkin, and squash are cultivated for fresh and processing markets (72).

Although bacterial wilt disease is limited primarily to the Northeastern and Midwestern United

States, it causes losses up to 80% at an annual monetary cost of over 25 million US dollars, and occurs over about 68% of the total acreage where cucurbits are grown (62; 63). The management of E. tracheiphila has often relied on the use of insecticides to control the cucumber beetle vector.

However, this poses a great risk of toxicity to natural pollinators of cucurbits, aquatic organisms, human and other non-targets (62). It also usually requires multiple applications within a single season and can be erratic in its effectiveness (29). A better understanding of the genetic basis for

15 host-specific virulence and pathogenicity among E. tracheiphila strains may provide insights that lead to strategies for more affordable and more eco-friendly disease management.

References

1. Alfano JR, Collmer A. 1997. The type III (Hrp) secretion pathway of plant pathogenic bacteria: Trafficking harpins, Avr proteins, and death. J. Bacteriol. 179:5655-62 2. Altinok HH, Erdogan O. 2015. Determination of the In vitro effect of Trichoderma harzianum on phytopathogenic strains of Fusarium oxysporum. Not. Bot. Horti Agrobot. Cluj-Na. 43:494-500 3. Arnold DL, Jackson RW, Fillingham AJ, Goss SC, Taylor JD, et al. 2001. Highly conserved sequences flank avirulence genes: Isolation of novel avirulence genes from Pseudomonas syringae pv. pisi. Microbiology-(UK) 147:1171-82 4. Asolkar T, Ramesh R. 2018. Development of T3SS mutants (hrpB (-) and hrcV (-) ) of Ralstonia solanacearum, evaluation of virulence attenuation in brinjal and tomato-a pre- requisite to validate T3Es of R. solanacearum. Indian J. Microbiol. 58:372-80 5. Asselin JE, Bonasera JM, Kim JF, Oh CS, Beer SV. 2011. Eop1 from a Rubus strain of Erwinia amylovora functions as a host-range limiting factor. Phytopathology 101:935-44 6. Badel JL, Shimizu R, Oh HS, Collmer A. 2006. A Pseudomonas syringae pv. tomato AvrE1/HopM1 mutant is severely reduced in growth and lesion formation in tomato. Mol. Plant-Microbe Interact. 19:99-111 7. Baumler A, Fang FC. 2013. Host specificity of bacterial pathogens. Cold Spring Harb. Perspect. Med. 3:19 8. Bogdanove AJ, Bauer DW, Beer SV. 1998. Erwinia amylovora secretes DspE, a pathogenicity factor and functional AvrE homolog, through the hrp (type III secretion) pathway. J. Bacteriol. 180:2244-7 9. Bove JM, Garnier M. 2003. Phloem-and xylem-restricted plant pathogenic bacteria (vol 163, pg 1083, 2002). Plant Sci. 164:421 10. Brust GE. 1997. Differential susceptibility of pumpkins to bacterial wilt related to plant growth stage and cultivar. Crop Prot. 16:411-4 11. Brust GE, Rane KK. 1995. Differential occurrence of bacterial wilt in muskmelon due to preferential striped cucumber beetle feeding. Hortscience 30:1043-5 12. Buttner D. 2016. Behind the lines-actions of bacterial type III effector proteins in plant cells. Fems Microbiol. Rev. 40:894-937 13. Castiblanco LF, Triplett LR, Sundin GW. 2018. Regulation of effector delivery by type III secretion chaperone proteins in Erwinia amylovora. Front. Microbiol. 9:10

14. Caudle JR, Coolong T, Williams MA, Vincelli P, Bessin R. 2013. Development of an organic muskmelon production system against bacterial wilt disease. In Ii International Organic Fruit Symposium, ed. D Granatstein, PK Andrews, SD Bishop, W Janisiewicz, 1001:249-54. Leuven 1: Int Soc Horticultural Science. Number of 249-54 pp. 15. Cavanagh A, Hazzard R, Adler LS, Boucher J. 2009. Using trap crops for control of Acalymma vittatum (Coleoptera: Chrysomelidae) reduces insecticide use in butternut squash. Journal of Economic Entomology 102:1101-7 16. Cobine PA, Cruz LF, Navarrete F, Duncan D, Tygart M, De La Fuente L. 2013. Xylella fastidiosa differentially accumulates mineral elements in biofilm and planktonic cells. PLoS One 8:10 17. Coburn B, Sekirov I, Finlay BB. 2007. Type III secretion systems and disease. Clin. Microbiol. Rev. 20:535–49 18. Correa VR, Majerczak DR, Ammar ED, Merighi M, Pratt RC, et al. 2012. The bacterium Pantoea stewartii uses two different Type III secretion systems to colonize its plant host and insect vector. Appl. Environ. Microbiol. 78:6327-36 19. Cresswell JE, Thompson HM. 2012. Comment on "a common pesticide decreases foraging success and survival in honey bees". Science 337:2 20. Decker DS. 1988. Origin(s), evolution, and systematics of Cucurbita pepo (Cucurbitaceae). Econ. Bot. 42:4-15 21. Degrave A, Siamer S, Boureau T, Barny MA. 2015. The AvrE superfamily: ancestral type III effectors involved in suppression of pathogen-associated molecular pattern- triggered immunity. Mol. Plant Pathol. 16:899-905 22. Eichenlaub R, Gartemann KH. 2011. The Clavibacter michiganensis subspecies: Molecular investigation of gram-positive bacterial plant pathogens. In Annual Review of Phytopathology, Vol 49, ed. NK VanAlfen, G Bruening, JE Leach, 49:445-64. Palo Alto: Annual Reviews. Number of 445-64 pp. 23. Elkabetz M, Paris HS, Burger Y, Hanan A, Cohen R. 2016. Two genes for resistance to Fusarium oxysporum f. sp radicis cucumerinum in melon (Cucumis melo, Cucurbitaceae). Sci. Hortic. 201:57-60 24. Fisher JR, Branson TF, Sutter GR. 1984. Use of common squash cultivars, Cucurbita spp for mass collection of corn rootworm beetles, Diabrotica spp (Coleoptera, Chrysomelidae). J. Kans. Entomol. Soc. 57:409-12 25. Fleischer SJ, de Mackiewicz D, Gildow FE, Lukezic FL. 1999. Serological estimates of the seasonal dynamics of Erwinia tracheiphila in Acalymma vittata (Coleoptera : Chrysomelidae). Environ. Entomol. 28:470-6 26. Gaudriault S, Paulin JP, Barny MA. 2002. The DspB/F protein of Erwinia amylovora is a type III secretion chaperone ensuring efficient intrabacterial production of the Hrp- secreted DspA/E pathogenicity factor. Mol. Plant Pathol. 3:313-20

27. Gazi AD, Sarris PF, Fadouloglou VE, Charova SN, Mathioudakis N, et al. 2012. Phylogenetic analysis of a gene cluster encoding an additional, Rhizobial-like type III secretion system that is narrowly distributed among Pseudomonas syringae strains. BMC Microbiol. 12:15 28. Ge Q, Cobine P, De La Fuente L. 2018. Xylella fastidiosa virulence in planta with copper supplementation. Phytopathology 108:148- 29. Goulson D. 2013. Review: An overview of the environmental risks posed by neonicotinoid insecticides. J. Appl. Ecol. 50:977-87 30. Guy E, Genissel A, Hajri A, Chabannes M, David P, et al. 2013. Natural genetic variation of Xanthomonas campestris pv. campestris pathogenicity on Arabidopsis revealed by association and reverse genetics. mBio 4:13 31. Ham JH, Majerczak D, Ewert S, Sreerekha MV, Mackey D, Coplin D. 2008. WtsE, an AvrE-family type III effector protein of Pantoea stewartii subsp stewartii, causes cell death in non-host plants. Mol. Plant Pathol. 9:633-43 32. Henry M, Beguin M, Requier F, Rollin O, Odoux JF, et al. 2012. A common pesticide decreases foraging success and survival in honey bees. Science 336:348-50 33. Horsfall JG, Dimond AE. 1959. Plant pathology: an advanced treatise. Vol. I: the diseased plant. Academic Press, New York and London. xiv + 674 pp. pp. 34. Jacobs JM, Milling A, Mitra RM, Hogan CS, Ailloud F, et al. 2013. Ralstonia solanacearum requires PopS, an ancient AvrE-family effector, for virulence and to overcome salicylic acid-mediated defenses during tomato pathogenesis. mBio 4:12 35. Jin QL, He SY. 2001. Role of the Hrp pilus in type III protein secretion in Pseudomonas syringae. Science 294:2556-8 36. Kim HS, Thammarat P, Lommel SA, Hogan CS, Charkowski AO. 2011. Pectobacterium carotovorum elicits plant cell death with DspE/F but the P. carotovorum DspE does not suppress callose or induce expression of plant genes early in plant-microbe interactions. Mol. Plant-Microbe Interact. 24:773-86 37. Kvitko BH, Park DH, Velasquez AC, Wei CF, Russell AB, et al. 2009. Deletions in the repertoire of Pseudomonas syringae pv. tomato DC3000 Type III secretion effector genes reveal functional overlap among effectors. PLoS Pathog. 5:16 38. Latin RX. 2000. Bacteria Wilt. APSnet Feature, online: http://www.apsnet.org/publications/apsnetfeatures/Pages/BacterialWilt.aspx 39. Lavie M, Shillington E, Eguiluz C, Grimsley N, Boucher C. 2002. PopP1, a new member of the YopJ/AvrRxv family of type III effector proteins, acts as a host-specificity factor and modulates aggressiveness of Ralstonia solanacearum. Mol. Plant-Microbe Interact. 15:1058-68 40. Leach JG. 1964. Observations on cucumber beetles as vectors of cucurbit wilt. Phytopathology 54:606-&

41. Lewis JD, Lee A, Ma WB, Zhou HB, Guttman DS, Desveaux D. 2011. The YopJ superfamily in plant-associated bacteria. Mol. Plant Pathol. 12:928-37 42. Lilley JM, Sanchez ES. 2016. The potential of strip tillage and rowcovers for organic cucurbit production. HortTechnology 26:628-36 43. Liu Q, Beattie GA, Rojas ES, Gleason ML. 2018. Bacterial wilt symptoms are impacted by host age and involve net downward movement of Erwinia tracheiphila in muskmelon. Eur. J. Plant Pathol. 151:803-10 44. Liu Q, Beattie GA, Saalau Rojas E, Gleason ML. 2015. Bacterial wilt symptoms are impacted by host age and involve net downward movement of Erwinia tracheiphila in muskmelon. Eur. J. Plant Pathol. 151:803-10 45. Liu Q, Beattie GA, Salaau Rojas E, Gleason ML. 2018. Bacterial wilt symptoms are impacted by host age and involve net downward movement of Erwinia tracheiphila in muskmelon. Eur. J. Plant Pathol. 151:803-10 46. Ma WB, Dong FFT, Stavrinides J, Guttman DS. 2006. Type III effector diversification via both pathoadaptation and horizontal transfer in response to a coevolutionary arms race. PLoS Genet. 2:2131-42 47. Minsavage GV, Dahlbeck D, Whalen MC, Kearney B, Bonas U, et al. 1990. Gene-for- gene relationships specifying disease resistance in Xanthomonas campestris pv vesicatoria pepper interactions. Mol. Plant-Microbe Interact. 3:41-7 48. Mohammadi AS, Maynard ET, Foster RE, Egel DS, McNamara KT. 2018. Muskmelon cultivar attractiveness to striped cucumber beetle and susceptibility to bacterial wilt. Hortscience 53:782-7 49. Moscou MJ, Bogdanove AJ. 2009. A simple cipher governs DNA recognition by TAL effectors. Science 326:1501- 50. Mukaihara T, Tamura N, Iwabuchi M. 2010. Genome-wide identification of a large repertoire of Ralstonia solanacearum type III effector proteins by a new functional screen. Mol. Plant-Microbe Interact. 23:251-62 51. Munroe DD, Smith RF. 1980. A revision of the systematics of Acalymma sensu stricto barber (Coleoptera, Chrysomelidae) from North America including Mexico. Mem. Entomol. Soc. Can.:1-92 52. NASS. 2019. Vegetables 2018 Summary. USDA, National Agricultural Statistics Service 53. Nazareno ES, Dumenyo CK. 2015. Modified inoculation and disease assessment methods reveal host specificity in Erwinia tracheiphila-Cucurbitaceae interactions. Microb. Pathog. 89:184-7 54. Pieretti I, Pesic A, Petras D, Royer M, Sussmuth RD, Cociancich S. 2015. What makes Xanthomonas albilineans unique amongst Xanthomonads? Front. Plant Sci. 6:7 55. Poueymiro M, Genin S. 2009. Secreted proteins from Ralstonia solanacearum: a hundred tricks to kill a plant. Curr. Opin. Microbiol. 12:44-52

56. Qian W, Jia YT, Ren SX, He YQ, Feng JX, et al. 2005. Comparative and functional genomic analyses of the pathogenicity of phytopathogen Xanthomonas campestris pv. campestris. Genome Res. 15:757-67 57. Rand FV, Enlows EMA. 1916. Transmission and control of bacterial wilt of Cucurbits. J. Agric. Res. 6:417-34 58. Roden J, Eardley L, Hotson A, Cao YY, Mudgett MB. 2004. Characterization of the Xanthomonas AvrXv4 effector, a SUMO protease translocated into plant cells. Mol. Plant-Microbe Interact. 17:633-43 59. Rojas ES, Batzer JC, Beattie GA, Fleischer SJ, Shapiro LR, et al. 2015. Bacterial Wilt of Cucurbits: Resurrecting a Classic Pathosystem. Plant Disease 99:564-74 60. Rufian JS, Lucia A, Rueda-Blanco J, Zumaquero A, Guevara CM, et al. 2018. Suppression of HopZ effector-triggered plant immunity in a natural pathosystem. Front. Plant Sci. 9:16 61. Saalau-Rojas E, Gleason ML. 2013. Erwinia tracheiphila: Getting around with a little help from my friends. Phytopathology 103:171-2 62. Saalau Rojas E, Batzer JC, Beattie GA, Fleischer SJ, Shapiro LR, et al. 2015. Bacterial wilt of Cucurbits: Resurrecting a classic pathosystem. Plant Disease 99:564-74 63. Saalau Rojas E, Dixon PM, Batzer JC, Gleason ML. 2013. Genetic and virulence variability among Erwinia tracheiphila strains recovered from different cucurbit hosts. Phytopathology 103:900-5 64. Sasu MA, Seidl-Adams I, Wall K, Winsor JA, Stephenson AG. 2010. Floral transmission of Erwinia tracheiphila by Cucumber beetles in a wild Cucurbita pepo. Environ. Entomol. 39:140-8 65. Sebastian P, Schaefer H, Telford IRH, Renner SS. 2010. Cucumber (Cucumis sativus) and melon (C. melo) have numerous wild relatives in Asia and Australia, and the sister species of melon is from Australia. Proc. Natl. Acad. Sci. U. S. A. 107:14269-73 66. Shapiro LR, Andrade A, Scully ED, Rocha J, Paulson JN, Kolter R. 2018. Draft genome sequence of an Erwinia tracheiphila isolate from an infected muskmelon (Cucumis melo). Microbiol. Resour. Ann. 7:2 67. Shapiro LR, Paulson JN, Arnold BJ, Scully ED, Zhaxybayeva O, et al. 2018. An introduced crop plant is driving diversification of the virulent bacterial pathogen Erwinia tracheiphila. mBio 9:20 68. Shapiro LR, Scully ED, Roberts D, Straub TJ, Geib SM, et al. 2015. Draft genome sequence of Erwinia tracheiphila, an economically important bacterial pathogen of cucurbits. Microbiol. Resour. Ann. 3:2 69. Shapiro LR, Scully ED, Straub TJ, Park J, Stephenson AG, et al. 2016. Horizontal gene acquisitions, mobile element proliferation, and genome decay in the host-restricted plant pathogen Erwinia tracheiphila. Genome Biol. Evol. 8:649-64

70. Smith EF. 1911. Bacteria in relation to plant diseases. Nature 74:294 71. Ustun S, Bartetzko V, Bornke F. 2015. The Xanthomonas effector XopJ triggers a conditional hypersensitive response upon treatment of N. benthamiana leaves with salicylic acid. Front. Plant Sci. 6:11 72. Vrisman CM, Deblais L, Rajashekara G, Miller SA. 2016. Differential colonization dynamics of cucurbit hosts by Erwinia tracheiphila. Phytopathology 106:684-92 73. Watterson JC, Williams PH, Durbin RD. 1971. Response of cucurbits to Erwinia tracheiphila. Plant Disease Reporter 55:816-9 74. Yadeta K, Thomma B. 2013. The xylem as battleground for plant hosts and vascular wilt pathogens. Front. Plant Sci. 4:12 75. Yang FJ, Cheng LL, Zhang L, Dai WJ, Liu Z, et al. 2009. RETRACTED: Y4lO of Rhizobium sp strain NGR234 Is a symbiotic determinant required for symbiosome differentiation (Retracted article. See vol. 197, pg. 3026, 2015). J. Bacteriol. 191:735-46 76. Yang FJ, Cheng LL, Zhang L, Dai WJ, Liu Z, et al. 2015. Y4lO of Rhizobium sp. strain NGR234 Is a symbiotic determinant required for symbiosome differentiation (Retraction of vol 191, pg 735, 2009). J. Bacteriol. 197:3026- 77. Yao CB, Zehnder G, Bauske E, Klopper J. 1996. Relationship between cucumber beetle (Coleoptera: Chrysomelidae) density and incidence of bacterial wilt of Cucurbits. Journal of Economic Entomology 89:510-4

CHAPTER 2 SUBSPECIES DELINEATION OF ERWINIA TRACHEIPHILA STRAINS ON THE

BASIS OF WHOLE-GENOME SEQUENCE AND PHYSIOLOGICAL DATA

Olakunle I. Olawole1, Qian Liu1, Yun Xu1, Benzhong Fu1, Mark L. Gleason1 and Gwyn A.

Beattie1.

1Department of Plant Pathology and Microbiology, Iowa State University, Ames, IA, U.S.A.

All of the data presented in this chapter were generated and analyzed by O.O. with the exception of the bacterial growth rate data (generated and analyzed by Q.L.) and the virulence data

(generated and analyzed collaboratively with Y.X.); the DNA for two of the strains was extracted by B.F. prior to assembly and analysis by O.O.

Abstract

Erwinia tracheiphila strains fall into phylogenetic clades that differ in host-specific virulence. We used a polyphasic approach in which we examined assembled genome sequences, selected functional groups, plasmid content, physiological traits and virulence to evaluate support for subspeciation within E. tracheiphila using 24 strains belonging to the Et-melo and

Et-C1 clades. The Et-melo strains were clearly differentiated from the Et-C1 strains based on whole genome sequence comparisons. The two clades differed greatly in their gene content, with almost 30% of the ~8,000 genes in the E. tracheiphila pangenome specific to the Et-melo clade and 20% specific to the Et-C1 clade. The two clades shared only a small fraction of the 44 toxin- antitoxin (TA) modules within the strains, and only two potential prophages among the 81 total or putative prophages. Whereas the Et-melo clade genomes had significantly more TA modules

22 and prophage than the Et-C1 clade strains, the Et-C1 clade had significantly more insertion sequence elements than the Et-melo strains. The profile of carbohydrate-active enzymes at the family level was similar in the two clades, but the two clades notably differed in the number of enzymes in each of several glycoside hydrolase subfamilies. The Et-melo clade was deficient in the ability to metabolize seven substrates that could be utilized by the Et-C1 clade. Strains of the

Et-C1 clade grew slower in culture, wilted muskmelon plants slower and had fewer large plasmids, on average, than Et-melo strains. In addition, cbsA, a 1,4-β-cellobiosidase that was present in Et-melo strains but absent in Et-C1 strains, modulated the virulence of both an Et-melo strain (upon deletion) and an Et-C1 strain (upon over-expression). We propose that the xylem- inhabiting, vector transmitted bacteria of cucurbits presently described as Erwinia tracheiphila

Et-melo and Et-C1 strains be delineated as Erwinia tracheiphila subsp. tracheiphila and Erwinia tracheiphila subsp. lentilata, respectively.

Introduction

E. tracheiphila is a Gram-negative, xylem-inhabiting and obligately vector-transmitted bacterial pathogen. This pathogen causes bacterial wilt of cucurbits in both the Cucurbita genus

(squash and pumpkin) and the Cucumis genus (muskmelon and cucumber). E. tracheiphila was one of the first bacterial phytopathogens to be described (40). The genus Erwinia includes many other plant pathogens, such as Erwinia amylovora, E. pyrifoliae, E. mallotivora, E. papayae, and

E. rhapontici (15; 18; 43). It also includes insect-associated species such as E. typographi and E. iniecta (25), epiphytes such as E. billingiae and E. tasmaniensis (25), and species such as E. toletana and E. oleae that can modulate the virulence of pathogens in mixed communities (8).

The genus Erwinia has a rich taxonomic history. It currently contains 44 validly published species based on the list of prokaryotic names with standing in nomenclature (28).

Members of this genus are characteristically heterogeneous, and, as such, have undergone numerous taxonomic changes over the past century (15) as illustrated by the reassignment of several species from the genus Erwinia into new genera, including Pectobacterium, Pantoea, and

Dickeya (7; 15). Among the Enterobacteriaceae, the species E. tracheiphila, E. amylovora, E. rhapontici, E. persicinus, E. psidii and E. mallotivora form a distinct clade based on 16S rDNA sequence similarities (15). Interestingly, whereas all these species cause necrotic diseases on their respective hosts, E. tracheiphila is the primary member known to cause a vascular disease.

As early as 1911, E. tracheiphila was reported to infect cucurbits in a host-specific manner; for example, strains isolated from cucumber showed a rapid wilt on cucumber, but slight or no symptoms when inoculated into squash or pumpkin (40). Over a hundred years after this first report, E. tracheiphila strains are still recognized to fall into distinct functional groups as evidenced by their ability to induce a rapid wilt, or even a wilt, in one cucurbit genus as compared to another. Since 1911, several research teams (26; 32) have reported differences among groups of E. tracheiphila strains in their virulence on distinct host plants, with one strain group wilting only members of the Cucumis genus and another wilting members of both the

Cucurbita and Cucumis genera. Saalau Rojas and associates (32) investigated the potential for genome sequence differences among 69 E. tracheiphila strains representing a variety of geographic and host origins. They discovered that these strains fell into two groups based on their rep-PCR fingerprints, which indicated genetic variation throughout their genomes; moreover, these fingerprints correlated with their virulence patterns on squash and muskmelon.

After expanding this E. tracheiphila strain collection further, Shapiro and colleagues (36) performed phylogenetic analyses based on draft genome sequences of 88 strains and found three clades, which they designated Et-melo, Et-C1 and Et-C2. The Et-melo strains had relatively broad geographic origins, having been isolated from infected plants from the middle and eastern regions of the U.S. Moreover, they were all isolated from Cucumis host species (C. melo and C. sativus), and were capable of wilting these Cucumis species but not squash (Cucurbita pepo)

(36). The Et-C1 and Et-C2 strains had narrower geographic origins, having been isolated from infected plants only in the mid-Atlantic and northeastern U.S., with the Et-C2 strains originating only from the northeastern U.S. The Et-C1 and Et-C2 strains were isolated from C. sativus and

Cucurbita spp., and were capable of wilting all of these hosts, although they induced a noticeably slower wilt than the Et-melo strains on Cucumis spp.

Here, we evaluated evidence to support a subspecies delineation of the Et-melo and Et-C1 clades using 12 representative strains per clade. We examined assembled genome sequences, predicted functions, plasmid content, physiological traits and virulence to evaluate support for designating these E. tracheiphila strains as distinct subspecies. Based on our results, we propose that the two clades, Et-melo and Et-C1, of the xylem-inhabiting, vector-transmitted bacterial pathogen of cucurbits presently described as Erwinia tracheiphila be designated as E. tracheiphila subspecies tracheiphila and E. tracheiphila subspecies lentilata, respectively.

Materials and methods

Bacterial strains, growth conditions and inoculum preparation

The E. tracheiphila strains for this study are described in Table 1. E. tracheiphila strains were grown in King’s B medium (17) at 28°◦C unless otherwise described. Escherichia coli

25 strains were grown in Luria medium at 37°C. The following antibiotics were added, when needed, at 50µg/ml: rifampin (Rif), kanamycin (Km), spectinomycin (Spc), and ampicillin

(Amp).

DNA extraction, library preparation and genome sequencing and assembly using the

PacBio platform

Single colonies of the E. tracheiphila strains SCR3 and BHKY were grown in King’s B broth to late-log phase, after which the cultures were adjusted to an OD600 of 1.0 using King’s B broth. Genomic DNA was purified using the DNeasy Blood and Tissue kit (Qiagen, Venlo,

Limburg) according to the manufacturer’s instructions, and included steps for Proteinase K and

RNase treatments, as recommended for PacBio genomic DNA extractions (23). Library preparation and sequencing were done at the Iowa State University DNA facility using the

SMRTbell Template Prep Kit (Pacific Biosciences, Menlo Park, CA) according to the PacBio standard protocol. Libraries of 20-kb size-selected fragments were generated using the

BluePippin Size-selection System (Sae Science Inc, Beverly, MA), which included steps for

DNA damage and end repair and ligation to hairpin adapters. Libraries were subsequently sequenced on the PacBio RSII instrument and P6-C4 chemistry with one SMRTcell per strain.

The HGAPv3 protocol of the Pacific Biosciences SMRTportal software was used to assemble each genome. Raw SMRT sequencing reads were trimmed, corrected, and assembled de novo using default parameters with a genome size estimate of 5 Mb (38).

Table 1: Bacterial strains and plasmids used in this study

Rep-PCR Location and year of 2 Strain Host 1 Clade Ref isolation Pattern BHKY Cucurbita moschata Kentucky (2010) B Et-C1 32,36 BoCa4-1b Cucumis melo Iowa (2008) A Et-melo 32,36 BoCu1-2 Cucumis sativus Iowa (2008) A Et-melo 32,36 BoCu2-1 Cucumis sativus Iowa (2008) A Et-melo 32,36 BoCu3-1b Cucumis sativus Iowa (2008) A Et-melo 32,36 BontCu Cucumis sativus Iowa (2010) A Et-melo 32,36 Cucurbita pepo ssp. BuffGH Pennsylvania (2009) B Et-C1 texana 32,36 Cuke1IN Cucumis sativus Indiana (2010) A Et-melo 32,36 FCa2-3 Cucumis melo Iowa (2008) A Et-melo 32,36 FCu1-3 Cucumis sativus Iowa (2008) A Et-melo 32,36 FCu3-3 Cucumis sativus Iowa (2008) A Et-melo 32,36 Fish3-2 Cucumis sativus Iowa (2009) A Et-melo 32,36 FishCu3-1 Cucumis sativus Iowa (2009) A Et-melo 32,36 FishCu1-5 Cucumis sativus Iowa (2009) A Et-melo 32,36 GHM3-1 Cucumis melo Iowa (2008) A ND3 32,36 GrinCu Cucumis sativus Iowa (2010) A Et-melo 32,36 GuthCu Cucumis sativus Iowa (2010) A ND3 32,36 GZ4 Cucurbita pepo New York (2009) B Et-C1 32,38 HCa1-5 Cucumis melo Iowa (2009) ND Et-melo 22 HCu Cucumis sativus Iowa (2008) A Et-melo 32,36 HCu1-4 Cucumis sativus Iowa (2008) A Et-melo 32,36 HFCu Cucumis sativus Iowa (2010) A Et-melo 32,36 HFMusk Cucumis melo Iowa (2010) A Et-melo 32,36 HM2-2 Cucumis melo Iowa (2008) A Et-melo 32,36 KYMusk Cucumis melo Kentucky (2009) A ND3 32,36 LamMusk1 Cucumis melo Iowa (2010) A Et-melo 32,36 LamMusk2 Cucumis melo Iowa (2010) A Et-melo 32,36 LICuke14 Cucumis sativus New York (2010) A Et-C1 32,36 LICuke24 Cucumis sativus New York (2010) A Et-C1 32,36 LIMusk1 Cucumis melo New York (2010) A Et-melo 32,36 LIMusk2 Cucumis melo New York (2010) A ND3 32,36 LIMusk3 Cucumis melo New York (2010) A Et-melo 32,36 LISumSq1 Cucurbita pepo New York (2010) B Et-C1 32,36 LISumSq3 Cucurbita pepo New York (2010) B Et-C1 32,36 M2Ca2 Cucumis melo Iowa (2008) A ND3 32,36 MaMax Cucumis melo Maryland (2010) A ND3 32,36 MBrut1 Cucumis melo Oklahoma (2009) A Et-melo 32,36 MBrut3 Cucumis melo Oklahoma (2009) A Et-melo 32,36 MBrut4 Cucumis melo Oklahoma (2009) A Et-melo 32,36 MBrut6 Cucumis melo Oklahoma (2009) A Et-melo 32,36 MBrut7 Cucumis melo Oklahoma (2009) A Et-melo 32,36 MCa1-1 Cucumis melo Iowa (2008) A Et-melo 32,36 MCa4-2 Cucumis melo Iowa (2008) A Et-melo 32,36 McM1-1 Cucumis melo Iowa (2008) A Et-melo 32,36 McM2-4 Cucumis melo Iowa (2009) A Et-melo 32,36

Table 1. Continued

Rep-PCR Location and year of 2 Strain Host 1 Clade Ref isolation Pattern MDCuke Cucumis sativus Maryland (2010) A ND3 32,36 MISpSq Cucurbita pepo Michigan (2009) B Et-C1 32,36 Musk1IN Cucumis melo Indiana (2010) A Et-melo 32,36 NYAcSq1 Cucurbita pepo New York (2009) B Et-C1 32,36 NYAcSq2 Cucurbita pepo New York (2009) B Et-C1 32,36 NYZuch1 Cucurbita pepo New York (2009) B Et-C1 32,36 NYZuch2 Cucurbita pepo New York (2009) B Et-C1 32,36 OKDH1 Cucumis melo Oklahoma (2010) A ND3 32,36 OKMusk1 Cucumis melo Oklahoma (2010) A Et-melo 32,36 OKMusk2 Cucumis melo Oklahoma (2010) A Et-melo 32,36 OKMusk3 Cucumis melo Oklahoma (2010) A Et-melo 32,36 PPHow1 Cucurbita pepo Pennsylvania (2009) B ND5 32,36 PPHow2 Cucurbita pepo Pennsylvania (2009) B Et-C1 32,36 SCR3 Cucumis melo Iowa (2009) ND Et-melo 32 TedCu Cucumis sativus Iowa (2009) A Et-melo 32,36 TedCu(10) Cucumis sativus Iowa (2010) A Et-melo 32,36 TPINCu1 Cucumis sativus Indiana (2009) A Et-melo 32,36 UnisCa1-5 Cucumis melo Iowa (2009) A Et-melo 32,36 UnisCu1-1 Cucumis sativus Iowa (2009) A Et-melo 32,36 ZimmMusk Cucumis melo Iowa (2010) A Et-melo 32,36 ZittCuke14 Cucumis sativus New York (2010) A Et-C1 32,36 ZittCuke24 Cucumis sativus New York (2010) A Et-C1 32,36

1 These Rep-PCR patterns were identified based on BOXA1R and ERIC1-2 primers by Saalau Rojas et al. (32); ND, not determined. 2 These clades were assigned based on draft genome sequences by Shapiro et al. (36); ND, not determined. 3 These strains were tentatively assigned to the Et-melo clade in this work based on their Rep-PCR pattern A and the strong association between Rep-PCR group A and the Et-melo clade. 4 Although the strains LICuke1, LICuke2, ZittCuke1 and ZittCuke2 were identified as having the Rep-PCR pattern A in (32), sequence data at two loci and virulence on squash indicated that LICuke2 had been mis-assigned (22), and the four strains were subsequently assigned to the Et-C1 clade based on their draft genomes (36). This clade assignment is supported by the growth rates and pathogenicity data reported in the current study. 5 This strain was tentatively assigned to the Et-C1 clade in this work based on its Rep-PCR pattern B and the strong association between Rep-PCR group B and the Et-C1 clade.

DNA extraction, library preparation and genome sequencing and assembly using the Illumina platform Single colonies of 22 E. tracheiphila strains (Table 1) were grown and genomic DNA was isolated, as described above. Library preparation was done in collaboration with staff at the

Iowa State University DNA facility using the Nextera DNA Flex Library Prep kit (Illumina, San

Diego, CA), and sequencing was performed using the Illumina HiSeq3000 platform to generate

100-bp paired-end reads. DNA isolation and sequencing of the strains MBrut4 and LIMust3 were

28 performed separately from the other strains. The adaptors were trimmed and quality filtered using Trim galore (33) and FastQC 0.11.7 (39). The Illumina paired-end reads were assembled by mapping them to the SCR3 and BHKY reference genomes using the Burrows–Wheeler

Alignment (BWA) 0.7.17 and Bowtie2 2.3.4.1 toolkits (6). A pileup was created with SAMTools

1.9 (20), which also removed unmapped reads, and variants were called using VCFtools 0.1.14

(11). Finally, a consensus sequence file was generated for each strain using the bcftools (19). The strain-specific regions were assembled by extracting the unmapped reads, which were subsequently assembled de novo using Spades 3.11.1 (5). The mapped and de novo assemblies were concatenated for each strain and used for all downstream analyses. This resulted in 8 to 93 contigs for all but two strains, which were sequenced separately and had 311 and 403 contigs.

Genome annotation and analysis of the pan-genome, prophage, insertion sequences, carbohydrate enzymes and toxin-antitoxin systems

Annotation of assembled genome sequences was carried out using the rapid prokaryotic genome annotation (Prokka) (34) and rapid annotation using subsystem technology (RAST) (4) pipelines. The annotated files from the Prokka were used for a pan-genome analysis using

ROARY, which compiles orthogroups based on the presence or absence of orthogroup members

(27). The minimum sequence length was set to 120 bp and only sequences containing both start and stop codons were used. Iterative clustering steps starting at 100% identity and 100% length, with reductions down in 0.5% steps to 98%, were used to remove core genes present in all isolates. The genes in the pangenome, core and accessory genomes were aligned and maximum likelihood trees were produced using RAxML (42); trees were viewed in FigTree 1.4.4 (30).

Genome sequences were further compared using the BLAST Ring Image Generator (BRIG)

29 software (1), which is a cross-platform application that enables the interactive generation of comparative genomic images via a graphical user interface. Prophage analysis was performed using PHASTER (3). Insertion sequence elements were identified using ISEScan (48).

Carbohydrate-active enzymes were identified using the dbCAN2 database (49). Toxin-antitoxin systems were identified using TADB 2.0 (47), a database of toxin-antitoxin loci.

Plasmid profiling

Plasmid DNA was extracted from two-day old cultures of E. tracheiphila strains by using the Qiagen midi-kits (Qiagen, Venlo, Limburg) according to manufacturer’s instructions. Pulsed- field gel electrophoresis was performed using the CHEF Mapper XA System (Bio-Rad) by running 500ng of plasmid DNA on a 1% agarose gel dissolved in 0.5X TBE buffer.

Electrophoretic parameters include a voltage of 5V/cm, pulse angle of 120°C and pulse time of 1 s to 100 s, all in a 14°C chilled 0.5X TBE buffer (29; 31). The plasmid DNA samples used as controls were extracted from Pseudomonas syringae pv. tomato (DC3000) and P. syringae pv. phaseolicola (1448A) strains (13). At least 3 independent experiments were used to evaluate plasmid DNA from each E. tracheiphila strain.

Linear PCR products were generated and introduced via electroporation into E. tracheiphila cells that had been transformed with pKD46. These linear PCR products contained a kanamycin cassette (kan) flanked by flippase recognition target (FRT) sites, as amplified from plasmid pKD13. The linear PCR products included 90-nt regions flanking cbsA in the MDCuke genome. Cells of MDCuke(pKD46) were grown at 28°C in King’s B broth amended with Amp.

When cells reached an OD600 of 0.3, freshly-made L-arabinose was added to a final concentration of 10mM, and cell growth was continued to an OD600 of 0.5. Cells were harvested

30 and washed three times with room temperature sterile distilled water. The room temperature wash was adopted because a cold-water wash appeared to reduce E. tracheiphila competency.

Cells were suspended in sterile distilled water, placed on ice for 15 to 20 min, and the chimeric linear fragments generated above were introduced into the chilled cells at a concentration of 2 to

3 μg DNA per 100 μl of cells in a 0.2-cm cuvette using a gene pulser electroporator (Bio-Rad,

Hercules, CA). Electroporated cells were recovered with 1ml of ice-cold KB broth and incubated for 6 to 10 h, after which cells were centrifuged and pellets were plated on Km-supplemented solid KB agar to select for successful mutants. After confirmation of a MDCuke cbsA::kan mutant by PCR, the kan cassette was removed by introduction of pFlp2Ω (41), which encodes the Flp recombinase to promote recombination at the FRT sites.

To express cbsA from MDCuke in an E. tracheiphila strain lacking this gene, the cbsA gene from MDCuke with its native promoter was amplified by PCR and cloned into the EcoRV site of pN (10), which is a derivative of the broad-host-range vector pME6041 (16) that has a

733-bp nptII promoter inserted upstream of the multiple cloning site. The resulting construct, pN-cbsA contained cbsA under the control of a fusion of its native promoter and the nptII promoter. The plasmid pN-cbsA was introduced by electroporation into E. tracheiphila strain

BHKY cells, prepared for transformation as described above for the MDCuke cells.

Physiological characteristics of E. tracheiphila strains

A total of 65 strains of E. tracheiphila (Table 1) strains were evaluated for their growth rates. Strains were grown on Nutrient Agar Peptone (NAP) medium (46) for 3-4 days, and four independent suspensions were prepared for each strain. The cell suspensions were adjusted to an

OD600 of 0.2 and, for each suspension, 30 μl was introduced into 150 μl of Nutrient Broth (NB)

31 in a well of a microtiter plate. The plates were incubated at 25°C with constant shaking and cell growth was monitored on an EL 340 Microplate Biokinetics Reader (Bio-tek Instruments,

Winooski, VT) until cells reach at least early stationary phase. The growth rates of each culture were calculated based on the slope of the growth curve during exponential growth.

Phenotype MicroArrays (Biolog, Hayward, CA) were used to determine the ability of four E. tracheiphila strains, MDCuke, SCR3, BuffGH and BHKY, to utilize 190 carbon sources.

The phenotype microarray plates 1 PM1 and PM2A were used according to the manufacturer’s instructions. Briefly, each bacterial strain was suspended in the Biolog inoculation fluid (IF-0) and the OD600 was normalized to 0.36 (42% transmittance). Cells were diluted with IF-0 amended with Biolog Dye-Mix A in a 1:5 ratio (diluted cells:dye solution) and inoculated into the microarray wells (100 ul/well). Plates were incubated for 48 h and a positive reaction was based on the detection of purple coloration, which indicated respiration in the presence of the tetrazolium dye and thus carbon utilization. Each strain was examined in two independent plates of each type.

Virulence assay

A total of 59 strains of E. tracheiphila were evaluated for their relative virulence on muskmelon, a host that supports the virulence of all of the strains. Strains that were altered in cbsA, as described above, were examined for their virulence on muskmelon and squash. Seeds of muskmelon (Cucumis melo cv. Athena) and squash (Cucurbita pepo cv. Parthenon) were sown in 10-cm pots containing a 1:1:1 mix of peat moss, coarse perlite and Metro-Mix 300 (Sun Gro

Horticulture, Canada Ltd.; Vancouver, BC, Canada). Seedlings were maintained in a growth chamber at 28°C under a photoperiod of 12 h of light and 12 h of darkness and 70% relative

32 humidity. Plants were watered every other day and fertilizer solution (NPK: 15-5-15, Peters

Excel) was added only once, at two days before inoculation.

To prepare cells for plant inoculation, E. tracheiphila strains were recovered from cryogenically preserved glycerol stocks (-80°C) on King’s B agar, which was amended with Rif.

The plates were incubated for 3-4 days, after which a single colony was transferred to a fresh plate to make a lawn. Cells from a two-day old lawn were suspended in 10 mM PB and normalized to an OD600 of 0.5. A nonionic organosilicone surfactant (Silwet L-77) was added to each bacterial suspension at 0.02% (vol/vol).

E. tracheiphila cells were inoculated into two-week old seedlings by puncturing the site closest to the petiole of the adaxial surface of the youngest fully expanded leaf with a 28.6-mm- diameter florist’s pin frog (Kenzan Pin Frog, sold by www.save-on-crafts.com), and 200 μl of a cell suspension, prepared as described above, was applied to the punctured site. Plants were inoculated with PB buffer as controls. Inoculated plants were incubated as described for the seedlings, above, and rated daily for wilt development based on the total number of leaves and the number of leaves wilted per inoculated plant. Once plants reached 100% wilt, newly emerging leaves were not counted and the plants were removed from the experiment. Data were collected for two weeks. The virulence of each strain was examined in at least three independent assays.

Construction of a cbsA deletion mutant and cbsA expression construct

The primers used in this study are shown in Table 2. To delete cbsA in E. tracheiphila strain MDCuke, we used the lambda Red recombinase system, which is comprised of

Table 2. Plasmids and primers used in this study.

Primer Sequences1

Primers for constructing deletion mutants cbhA-Km-F 5’GCCAGTAAAACCACGAACTTCAGTGACTGTCACCGGGAGGAG GCAAGCCGGATGAACCGGTTTCGCCGACCCCTGACCCGACGGGA AAAGACGTCTTGAGCGATTGTGTAGGCT-3’ cbhA-Km-R 5’CCGCCTTATAGATGGCAGCTTTTTACTTTACCCAGCCATTGCAG GGGATTCCTCAGTAATGGGCTGCTTCGTTCCCGAACGTGCTCAA GAGTGATTGCGCCTACCCGGATATT-3’ cbhA-KmF ACCTTGCCGGATGGTCATCTG cbhA-KmR AACTTCAGTGACTGTCACCGGGAG

Primers for constructing chbA-overexpression fusions cbhAMDCUKE (KpnI-F) AGTGCAATGATGGGCGATACGG

cbhAMDCUKE (NcoI-R) CGTTTCTGGCTTACGCAACCAAC pN-ME60-F AAGGTCATCCACCGGATCAATTCC pN-ME60-R TCTATCGATGCATGCCATGGTACC

1Regions underlined in the primer sequence are targeted to the kan cassette on pKD13 (12). recombinase components expressed under the control of an arabinose-inducible promoter on the plasmid pKD46 (12).

Statistical analyses For the virulence assay, the total number and wilted number of leaves were used to calculate the proportion of wilted leaves. This proportion was further subjected to an arcsine- square root transformation before analysis, as is common for data expressed as proportions, and these values were plotted over time to generate disease progress curves. Virulence was quantified based on the area under disease progress curve (AUDPC) values. Graphs were generated using

SigmaPlot 14 and statistical differences among treatments were evaluated using JMP Pro 12

(SAS Institute Inc., Cary, NC). For both the growth rate and AUDPC values, the strains were compared based on an analysis of variance (ANOVA), with mean values compared using a Least

Significant Difference (LSD) test at a 5% level of significance for studies with more than two strains, and with a Student’s t-test for studies with only two strains. Values were plotted as the mean and standard error of the mean (SEM) of data from individual experiments.

Results

Comparative genomic analysis of 24 strains of Erwinia tracheiphila supports division into two distinct groups of low genetic diversity

We selected 12 strains that were assigned to, or predicted to be in, the Et-melo cluster

(32; 36), and 12 of the 13 strains that were assigned to the Et-C1 cluster (32; 36), for comparative genomics (Tables 1 and 3). We assembled genome sequences from long-read data for the Et-melo strain SCR3 and Et-C1 strain BHKY and used these as reference genomes for the assembly of genome sequences from short-read data for the other 22 strains. For these 22 strains, we also performed de novo assembly of the unmapped reads and concatenated these with the mapped assemblies. Although short-read sequences were publicly available for five of the Et-C1 strains (38), and genome sequences derived from long-read data were publicly available for two of the 22 strains (35; 37), short-read data were obtained on these sequences here to foster side- by-side assemblies and comparisons. The genome sequences were assembled into a chromosome and three additional contigs for SCR3, of which all three are likely plasmids (128,183 bp, 67,812 bp and 42,229 bp), and two additional contigs for BHKY, of which one is likely a plasmid

(42,095 bp) and the other a potential phage (22,063 bp) based on the presence of phage genes, with duplicate genes on the chromosome. The 42-kb plasmids in SCR3 and BHKY are highly distinct in their gene content. The genome sequences for the other strains assembled from short- read data ranged from 8 to 93 contigs for all but two strains, which were sequenced separately and had 311 and 403 contigs. Although the strains of the two clades had similar genome sizes, approximately 5.0 Mb, the Et-melo strains had significantly more coding sequences (CDS)

(5,547 ± 78) than the Et-C1 strains (5,486 ± 28) (P = 0.017) and a significantly lower % GC content (50.45 ± 0.02) than the Et-C1 strains (50.53 ± 0.01, respectively) (P < 0.001) (Table 3).

Table 3. An overview of the assembly metrics of the sequenced genomes of E. tracheiphila Et-melo and Et-C1 clade strains Total contig N50 contig Clade Strains # contigs length (bp) size, (bp) GC (%) # CDS Et-melo SCR31 4 4,910,513 4,672,289 50.46 5,587 Et-melo UnisCa1-5 26 4,979,064 4,672,598 50.42 5,489 Et-melo TedCu(10) 31 4,990,562 4,672,630 50.46 5,494 Et-melo TPINCu1 32 4,967,781 4,672,609 50.47 5,447 Et-melo OKMusk1 35 5,035,441 4,672,541 50.47 5,556 Et-melo FCu1-3 38 4,963,203 4,672,601 50.48 5,443 Et-melo LIMusk1 39 4,980,296 4,672,600 50.46 5,470 Et-melo HFCu 82 5,069,002 4,672,559 50.44 5,589 Et-melo HCa1-5 88 5,074,084 4,672,565 50.45 5,581 Et-melo MDCuke 93 5,071,587 4,672,594 50.44 5,588 Et-melo LIMusk3 311 5,141,950 4,672,763 50.46 5,645 Et-melo MBrut3 403 5,179,373 4,672,584 50.41 5,680 Et-C1 BHKY1 3 4,988,831 4,924,673 50.53 5,518 Et-C1 MISpSq 8 4,989,931 4,924,792 50.53 5,458 Et-C1 LISumSq1 13 5,024,823 4,924,794 50.53 5,493 Et-C1 GZ4 15 5,034,725 4,924,786 50.51 5,504 Et-C1 NYZuch1 16 5,035,273 4,924,819 50.51 5,507 Et-C1 BuffGH 18 4,991,925 4,924,811 50.53 5,443 Et-C1 NYAcSq2 18 4,991,773 4,924,834 50.53 5,442 Et-C1 PPHow1 21 5,021,915 4,924,825 50.53 5,479 Et-C1 ZittCuke1 23 5,022,915 4,924,838 50.53 5,480

Et-C1 NYZuch2 26 5,047,178 4,924,820 50.52 5,527 Et-C1 LICuke2 29 5,028,911 4,924,798 50.53 5,499 Et-C1 PPHow2 31 5,025,084 4,924,799 50.53 5,480

Et-melo MDCuke2 5 5,015,952 4,891,633 50.48 5,116 Et-C1 BuffGH2 7 5,130,443 4,281,223 50.6 5,414 1Sequence data were generated used the PacBio platform with RSII chemistry. Sequence data on all other strains were generated using the HiSeq 3000 Illumina platform. 2Data from two publicly available genome sequences generated using the PacBio platform with RSII chemistry. MDCuke (35); BuffGH (37).

We aligned whole genome sequences of all of the 24 strains to either strain BHKY (Fig 1A) or strain SCR3 (Fig 1B) as the reference genome and visualized the genomes using the BLAST

Ring Image Generator (BRIG) software (1). This visualization supported the division of the strains into two clades with a low genetic diversity within each group, as previously reported

(36). We observed four variable regions (VR) for which the variation was detected as the absence or presence of relatively large regions among strains within a clade (Fig 1). Three variable regions (VR1, VR2 and VR3) were identified among the Et-melo strains, including VR1 in 4 of the 12 strains, VR2 in 3 strains, and VR3 in two strains (Fig 1A). Variation was detected in only three of the Et-C1 strains (VR4) (Fig 1B). A maximum likelihood tree of whole genome sequences similarly highlighted the separation of the strains into two clusters (Fig 2), consistent with the Et-melo and Et-C1 clades previously reported (36).

Pangenomic analysis reveals strong clade-specificity in the accessory genome

A pangenome analysis of the 24 strains using ROARY (27) identified a total of 8,064 genes, of which 47.3% (3,812 genes) were shared among at least 23 of the strains and thus comprised the core genome (Fig 3). The remaining 52.7% (4,252 genes) comprised the accessory genome. Within the accessory genome, the majority of genes (89%) was present in strains of only one clade, indicating that the clades differ greatly in gene content. We identified 2,368 genes that were present only in the Et-melo clade. Of these, the majority (53%) were present in all 12 strains, and 25% were present in only 1 strain. Similarly, we identified 1,586 genes that were present only within the Et-C1 clade. Of these, the majority (83%) were present in all 12 strains, and 4% were present in only 1 strain. These results illustrate a divergence in gene content between the two clades, a larger number of clade-specific genes in the Et-melo than Et-C1 clade, and a low genetic diversity among strains within each clade.

Figure 1. A BRIG visualization illustrates the clade-specificity of 24 E. tracheiphila strains based on the alignment of their genome sequences to (A) strain BHKY (clade Et-C1) and (B) strain SCR3 (clade Et-melo) as reference genomes. The innermost circles represent the GC content (black) and GC skew (purple/green) of the reference strains. Data reflect unfiltered BLASTn searches. The red rings represent the strains assigned to the Et-C1 clade, whereas the blue rings represent the strains assigned to the Et-melo clade (Table 1). In the outermost blue and red rings, the color intensity is proportional to the BLASTn identity to either reference strain, with red or blue regions indicating the highest nucleotide identity to the reference, and white most dissimilar. The indicated regions of variability in (A) and (B) are distinct regions. VR, variable region.

Figure 2. A phylogenetic tree of E. tracheiphila whole genome sequences constructed based on the Maximum Likelihood method supports clustering these 24 strains into two clades. The genome sequences were aligned and maximum likelihood trees generated using RAxML (42) and trees were viewed using FigTree 1.4.4 (30).

Figure 3: Pangenome and functional group distribution among 24 Erwinia tracheiphila strains. Summary of pangenome analysis, showing the pangenome size, relative contribution of the core and accessory genes, and the clade-specific genes.

The two E. tracheiphila clades differ in their gene content in selected functional categories The genome sequences of strains of both clades indicated a number of shared functional traits. These include genes involved in flagellar motility and chemotaxis, type IV pili, two type

III secretion systems, a type VI secretion system, copper efflux and resistance, and six complete ribosomal RNA operons. Strains of both clades also have genes for the synthesis of the quorum signal autoinducer-2, as well as two to three acyl-homoserine lactone (AHL) synthase genes, although only the Et-C1 clade has a gene for the cognate AHL-binding transcriptional regulator.

Here we evaluated potential clade-specificity in the gene content within several functional categories, namely toxin-antitoxin modules, prophages, carbohydrate-active enzymes and insertion sequence elements.

Toxin-antitoxin (TA) modules appeared to be particularly abundant in the E. tracheiphila genomes, consistent with the abundance of mobile genetic elements in this species (38). The Et- melo strains encode 30 TA modules, which is almost twice as many as the Et-C1 strains, which encode 17 (Fig 4). Interestingly, only four of the 44 TA modules are shared between the two clades. Although the majority of the toxins in both clades are predicted to be in the RelE superfamily, the Et-melo clade was distinct in having several toxins in the HicA family. Among the antitoxins, we observed that the Xre and PHD antitoxins were the dominant type in the Et- melo strains, whereas the RHH antitoxin was the most abundant among the Et-C1 strains (Fig 4).

The clades thus have highly distinct sets of TA modules.

Figure 4. The E. tracheiphila clades are distinct in their profile of toxin-antitoxin (TA) systems. TA systems were identified based on the TADB 2.0 database of bacterial type II toxin-antitoxin loci (47). The identity of the TA systems based on TADB 2.0 is shown at the bottom. Green, TA system is present. Yellow, TA system is absent. E. tracheiphila genomes encode many prophages (2; 35; 37; 38); here, we profiled the

genomewide prophage repertoire of all 24 strains using PHASTER (3). The majority of the

identified prophages are predicted to be Mu-like (Fig 5), and thus tailed phage with features

similar to phage in the Myoviridae and Siphoviridae families. A few members of the Podoviridae

and the filamentous prophage family Inoviridae were also identified. A total of 51 prophages

Figure 5. The E. tracheiphila clades are distinct in the repertoire of prophages identified in their genomes. Prophage were predicted in the genome sequences using PHASTER (3). The prophages were designated as intact, incomplete, or questionable prophages, and were assigned unique identifiers only within each category. The predicted family and length of each prophage is shown at the bottom. Green, prophage is present. Yellow, prophage is absent.

(27 intact, 15 incomplete, and 9 questionable) was predicted among the Et-melo strains; whereas

32 prophages (18 intact, 6 incomplete, and 8 questionable) were predicted among the Et-C1

strains (Fig 5). Of these 83 potential prophages, only two phages that were considered

questionable (9.4-kb and 35-kb), were shared by all 24 strains. These findings show that, like

with the TA modules, the two clades have distinct sets of prophages, with the Et-melo clade

42 having significantly more phage (15.7 ± 0.9 intact prophage and 24.2 ± 1.1 intact and incomplete prophage per strain) than the Et-C1 clade (14.0 ± 0.0 intact prophage and 18.3 ± 0.5 intact and incomplete prophage per strain) (P < 0.0001), based on phage predictions by PHASTER.

The carbohydrate-active enzymes (CAZymes) include enzymes that could be involved in degrading plant carbohydrates associated with the xylem. We performed a genomewide characterization of these enzymes in the 24 strains using the dbCAN2 database of CAZymes

(49). We found a total of 55 CAZymes, including 28 glycoside hydrolases (GHs), 14 glycosyl transferases (GTs), 5 carbohydrate binding modules, 3 polysaccharide lyases (PLs), 3 auxiliary activity families and 2 carbohydrate esterases. The two clades differed primarily in their GH profiles (Table 5), with the subfamilies GH19, GH23, and GH73 enriched in the Et-C1 strains, and GH13_5 present only in the Et-C1 strains. Similarly, the subfamilies GH6 and GH24 are enriched in the Et-melo strains, and GH33 and GH77 are present only in the Et-melo strains.

Table 5. Carbohydrate active enzymes that differ in number between the two E. tracheiphila clades.

No. of enzymes2

Enzyme family1 Enzyme subfamily Et-melo clade Et-C1 clade P-value3

Glycoside hydrolase GH23 10.0 ± 0.6 11.2 ± 0.4 0.0005 Glycoside hydrolase GH73 3.0 ± 0.0 4.0 ± 0.0 <0.0001 Glycoside hydrolase GH19 1.8 ± 0.4 3.0 ± 0.0 <0.0001 Glycoside hydrolase GH13_5 0.0 ± 0.0 1.0 ± 0.0 <0.0001 Glycoside hydrolase GH6 5.0 ± 0.0 3.0 ± 0.0 <0.0001 Glycoside hydrolase GH33 1.0 ± 0.0 0.0 ± 0.0 <0.0001 Glycoside hydrolase GH77 1.0 ± 0.0 0.0 ± 0.0 <0.0001 Glycosyltransferase GT2 11.0 ± 0.0 8.9 ± 0.3 <0.0001 Polysaccharide lyase PL4 1.0 ± 0.0 2.0 ± 0.0 <0.0001 1 Carbohydrate-active enzymes (CAZymes) were predicted using dbCAN (49). Although E. tracheiphila has 5 CAZymes in the carbohydrate-binding module family, 2 in the carbohydrate esterase family, and 3 in the auxiliary activities family, strains in the two E. tracheiphila clades had similar numbers of enzymes in each of these three families. 2 The mean ± SD is shown for the number of enzymes predicted by the genome sequences of 12 strains within each clade. 3 The P-value was derived from a students’ t-test comparing the two clades for the number of enzymes in the indicated subfamily.

In addition, the two clades differed in the abundance of one subfamily in each the GT and PL

families (Table 5). These differences in the repertoire of the polysaccharide-degrading enzymes in the two clades

could suggest differences in their ability to degrade the xylem intervessel pit-membranes in

various cucurbit hosts.

Finally, we characterized the insertion sequence elements (ISEs) of the 24 strains using

ISEScan (48). The clades both had exceptionally large number of ISEs, as expected based on a

previous report (38), but the Et-C1 clade, which had 659 ± 1 total ISEs, had significantly more

than the Et-melo clade, with 588 ± 11 (P < 0.0001). This difference was due to slightly larger

numbers in many ISE families (Table 6), but a particularly large difference in the IS4 family,

showed more than 4-fold more ISEs in the Et-C1 than the Et-melo strains. In contrast, there

were more ISEs that could not be placed in families in Et-melo strains than in Et-C1 strains

Table 6. Insertion sequence element (ISE) composition of the two E. tracheiphila clades. Number of ISEs2 ISE families1 Et-melo Et-C1 P-value3 IS481 96.2 ± 4.8 114.2 ± 1.0 <0.0001 IS1 83.4 ± 0.7 96.3 ± 0.5 <0.0001 IS91 86.4 ± 2.3 83.3 ± 0.5 <0.0005 IS256 57.8 ± 1.3 75.1 ± 0.3 <0.0001 IS200/IS605 66.8 ± 1.6 61.0 ± 0.0 <0.0001 IS3 50.8 ± 2.0 54.0 ± 0.0 <0.0002 IS4 9.0 ± 0.0 42.3 ± 0.5 <0.0001 IS5 38.2 ± 0.7 31.0 ± 0.0 <0.0001 IS6 27.4 ± 1.2 29.0 ± 0.0 0.001 ISL3 19.0 ± 0.0 28.0 ± 0.0 <0.0001 IS110 17 ± 0.3 17.1 ± 0.3 <0.0001 IS21 6.1 ± 0.3 11.0 ± 0.0 <0.0001 ISNCY 13.2 ± 0.4 9.0 ± 0.0 <0.0001 IS630 5.1 ± 0.3 6.0 ± 0.0 <0.0001 IS66 2.2 ± 0.6 1.0 ± 0.0 <0.0001 Unknown 8.0 ± 0.0 1.0 ± 0.0 <0.0001 TOTAL 588 ± 11.1 659 ± 1.44 <0.0001 1 ISEs for individual genome sequences were predicted using the ISEScan software (48). 2 The mean ± SD is shown for the number of ISEs predicted by the genome sequences of 12 strains within each clade. 3 The P-value was derived from a students’ t-test comparing the two clades for the number of ISEs in the indicated family.

(Table 6), although these unknown ISEs were rare in both clades. Overall, these differences in the gene content in these selected functional categories between the Et-melo and Et-C1 strains illustrate some of the genetic distinctions between these two groups.

The two E. tracheiphila clades differ in their plasmid content

A pulsed-field gel electrophoretic analysis of 21 E. tracheiphila strains showed the presence of plasmids in all of the strains (dark bands in Fig 6), with many light bands that we predict are prophage that were induced during growth in culture. The Et-melo strains generally contained 7-8 plasmids, with one strain, MBrut3 (lane 13), having as many as 10 plasmids.

The Et-melo plasmids included the 67.8-kb and 42.2-kb plasmids that were predicted by the genome sequence of SCR3 (lane 8), with the predicted 128.2-kb plasmid likely too large to be visualized in this gel. The Et-melo strains, including SCR3, generally harbored three small plasmids that would have been excluded during size fractionation of DNA for genome

Figure 6. The E. tracheiphila clades differ in their plasmid content. Plasmid DNA was subjected to pulsed-field gel electrophoresis for visualization. The samples were as follows: lane 1, 12 kb-plus DNA ladder; lane 2, EcoR1- digested lambda DNA; lane 3, Pseudomonas syringae pv. tomato (DC3000) plasmids (67kb, 73kb); lane 4, P. syringae pv. phaseolicola (1448A) plasmid (52kb); and lane 5, a mixture of DC3000 and 1448A plasmids. The Et- melo clade strains included: lane 6, MDCuke; lane 7, KYMusk; lane 8, SCR3; lane 9, TPINCu1; lane 10, TedCu(10); lane 11, OKMusk3; lane 12, Fca2-3; lane 13, MBrut3; lane 14, Musk1IN; and lane 15, LIMusk1. The Et-C1 clade strains included lane 16, BuffGH; lane 17, BHKY; lane 18, LISumSq1; lane 19, NYAcSq1; lane 20, NYAcSq2; lane 21, MISpSq; lane 22, NYZuch1; lane 23, NYZuch2; lane 24, GZ4; lane 25, PPHow1; and lane 26, PPHow2.

45 sequencing; these were approximately 11 and 7 kb (Fig 6) and 2 kb (data not shown). In contrast, the Et-C1 strains generally contained only 1-2 plasmids, and these included the 42.1-kb plasmid predicted by the genome sequence of BHKY (lane 17). The Et-C1 strains generally lacked small plasmids like those in the Et-melo strains. These distinct plasmid profile patterns again highlight the distinct genomic features of these two E. tracheiphila clades.

The E. tracheiphila strains exhibit differences in growth rate and carbon utilization profile

We investigated the growth rates in nutrient broth of 65 E. tracheiphila strains in our collection (Table 1), including 49 Et-melo strains and 16 Et-C1 strains. We found that 90% of the

Et-melo strains grew faster than the fastest-growing Et-C1 strain (Fig 7). Overall, the Et-melo strains showed significantly faster growth rates (0.0038 ± 0.0015 generations/hour) than the Et-

C1 strains (0.0016 ± 0.0002 generations/hour) (P<0.0001). Although more Et-melo strains than

Et-C1 strains were examined for this assay, the coefficient of variation (CV) indicated greater variation in growth rate among the Et-melo than among the Et-C1 strains (CV of 0.41 versus

0.14, respectively).

We also examined the carbon utilization profiles of two strains representing each of the two E. tracheiphila clades. Both strains utilized only 17 of the 192 carbon substrates provided in the Biolog Phenotype MicroArrays PM1 and PM2A (Table 7), reflecting a limited metabolic potential; this is consistent with a previous report that compared E. tracheiphila to other Erwinia spp. (14). There was complete congruence within a clade in the use of these 17 carbon sources, but the two clades differed in their use of 41% of these sugars. Specifically, the two Et-C1 strains utilized all of these carbon sources, whereas the two Et-melo clade strains were deficient in the utilization of seven of them (Table 7). The collective profile of growth rates and carbon

46 utilization abilities of the tested strains provide physiological support for the division of E. tracheiphila strains into at least two distinct groups.

Figure 7. The E. tracheiphila Et-C1 clade strains grow slower, on average, than the Et-melo clade strains. The growth rates (generations per hour) of four independent cultures of each strain were measured in a nutrient broth medium at 25°C with shaking.

Table 7. Carbon substrate utilization ability of selected strains of E. tracheiphila. Et-melo strains Et-C1 strains Substrates1 MDCuke SCR3 BuffGH BHKY

N-Acetyl-D-Glucosamine, D-Galactose, D-Mannitol, Glucose-6-Phosphate, D-Fructose, α-D-Glucose, Sucrose, + + + + Glucose-1-Phosphate, β-Methyl-D-Glucoside, Pectin

D-Saccharic acid, D-Trehalose, Glycerol, D, L-α- Glycerol-phosphate, L-Glutamic Acid, Fructose-6- -- -- + + Phosphate, M-Inositol

1 Data were obtained using the Biolog PM1 and PM2A plates. Substrates are listed only if they were utilized by at least one of the four strains tested.

The E. tracheiphila strains exhibit clade-specific differences in virulence on muskmelon

E. tracheiphila strains have been reported to differ in the rate at which they wilt muskmelon (26; 32; 36; 45), but few strains have been examined at the same time in any one

47 study. To evaluate the correlation between virulence and clade assignment, we profiled the rate at which the strains wilted muskmelon using 46 Et-melo strains and 13 Et-C1 strains. Due to high strain-to-strain variation in the rate of wilt, the clades did not form distinct groups when wilt was expressed as the area under the disease progress curve (AUPDC) (Fig 8). However, the Et- melo strains generally caused a more rapid wilt than the Et-C1 strains, and this was reflected in a significantly higher average AUPDC value for the Et-melo clade (10.1 ± 1.3) than for the Et-C1 clade (7.3 ± 2.6) (P = 0.006).

Figure 8. The E. tracheiphila Et-C1 clade strains induced wilt on muskmelon slower, on average, than Et-melo clade strains. The strains were inoculated into muskmelon (Cucumis melo cv. Athena), which is a host for strains in both clades. The experiment was repeated two times with 9 replicates per treatment. Two replicates of the strains SCR3 and BHKY were included; these replicates are designated as SCR3-1, SCR3-2, BHKY-1 and BHKY-2.

A glycoside hydrolase differentially modulates the virulence of strains of the two E. tracheiphila clades In addition to clade differences among the subfamilies of glycoside hydrolases, we found that one of the enzymes within the GH6 subfamily (Table 5) was present in all of the Et-melo

48 strains but was absent in all of the Et-C1 strains. This enzyme is predicted to be a 1,4-β- cellobiosidase (EC 3.2.1.4). As an enzyme that could be involved in xylem colonization and thus could contribute to the differential virulence of the clades on squash and muskmelon, we investigated the impact of modifying this gene, which we designated cbsA (24), on the virulence of E. tracheiphila on both hosts. We constructed a deletion mutant of cbsA in the Et-melo strain

MDCuke and introduced an over-expression construct of cbsA into the Et-C1 strain BHKY, which naturally lacks the gene. Virulence assays with these constructs showed that loss of CbsA reduced the virulence of MDCuke on muskmelon (Fig 9A), consistent with a potential contribution to Et-melo colonization of the xylem. Surprisingly, overexpression of cbsA in

BHKY reduced its virulence on squash (Fig 9B) and muskmelon (Fig 9C), indicating that cbsA interfered with the virulence of this Et-C1 clade strain. Thus, the function of this Et-melo- specific gene depends on its genomic context, which highlights functional differences among virulence genes in the two clades.

Discussion Among the many plant- and insect-associated Erwinia species, E. tracheiphila is distinct in being the only phytopathogen that is xylem-limited. We currently have little understanding of its evolutionary transition to the xylem habitat. The high number of mobile elements in the genomes of E. tracheiphila strains, when coupled with the pseudogenization of the genomes, may reflect a recent transition to a host-restricted lifestyle (38). Moreover, the radiation of the species into multiple clades, each of which exhibits low genetic diversity, was proposed to result from a population bottleneck followed by a population expansion, with the bottleneck potentially driven by a host jump from native Cucurbita spp. into an introduced, susceptible host species, C.

49 sativus, and subsequent expansion of the clades as they adapted to agriculturally propagated

Figure 9. The carbohydrate-active enzyme, cellulase (cbsA), uniquely modulates the virulence of strains representing each E. tracheiphila clade. (A) The cbsA gene is present in the Et-melo clade strain MDCuke, and deletion reduced virulence. (B, C) The cbsA gene is absent in the Et-C1 clade strain BHKY, and expression of cbsA from MDCuke in BHKY reduced virulence. Strains were inoculated into (A, C) muskmelon, which is a host for strains in both clades, and (B) squash, which is a host only for strains in the Et-C1 clade. Data shown include the proportion of wilted leaves (transformed for normality) on the left, and the area under disease progress curve on the right. Different letters indicate significant differences based on a Least Significant Difference test (P = 0.05). Each experiment was repeated at least twice with 9

replicates per treatment.

Cucumis and Cucurbita spp. (36). Here, we explored the nature and extent of the differences between the two largest and well-characterized E. tracheiphila clades, with a focus on whether these differences warrant delineation of the species into subspecies. We used a polyphasic approach in which we examined assembled genome sequences, selected functional groups, plasmid content, physiological traits and virulence to evaluate support for subspeciation within the E. tracheiphila species.

E. tracheiphila strains infect cucurbits in a host-specific manner, as illustrated by one group of strains showing a rapid wilt on muskmelon (Cucumis melo) but slight or no symptoms when inoculated into squash or pumpkin (Cucurbita spp.) (32), and a second group wilting both hosts, but with only a slow wilt on muskmelon. Strains have been grouped based on genetic variability as reflected via rep-PCR fingerprints (32) and phylogeny derived from draft genome sequences (36). The latter study identified three phylogenetic clades and designated them Et- melo, Et-C1 and Et-C2 (36). In this study, we attempted a subspecies delineation of the Et-melo and Et-C1 clades, using 12 representative strains per clade.

The Et-melo strains were clearly differentiated from the Et-C1 strains based on whole genome sequence comparisons, as visualized by BRIG analysis and a maximum likelihood tree.

Few variable regions were identified within the clades, corroborating the previous finding that individual clades exhibit low genetic diversity (32; 36). Moreover, the two clades differed greatly in their gene content. Almost 30% of the ~8,000 genes in the E. tracheiphila pangenome was specific to the Et-melo clade and 20% was specific to the Et-C1 clade, whereas the rest was shared, with only 47% of the pangenome comprising the core genome. Most of these clade- specific genes were present in all of the strains of each clade, again confirming the low within- clade genetic diversity. The two clades were similar in their average genome sizes, but the Et-

51 melo strains had slightly more coding sequences, comprised a greater portion of the clade- specific pangenome, and had a slightly lower GC content than the Et-C1 strains.

We also showed that the two clades were distinct in their gene content in selected functional categories. Whereas both clades had an abundance of TA modules, prophages, ISEs and CAZymes, they differed in their repertoire of genes within each category. These differences were most pronounced with the TA modules and the prophage, as the gene content in these categories was very nearly non-overlapping. The two clades shared only a small fraction of the

44 TA modules within the strains, and only two potential prophages among the 81 total or putative prophages. Whereas the Et-melo clade genomes had significantly more TA modules and prophage than the Et-C1 clade strains, the Et-C1 clade had significantly more ISEs than the Et- melo strains.

Like the ISEs, the profile of CAZymes at the family level was similar in the two clades.

However, the clades notably differed in the number of enzymes in each of several glycoside hydrolase subfamilies. These differences could have functional implications because of the role of these enzymes in polysaccharide degradation. The movement of E. tracheiphila in the xylem requires bacterial passage through the pit membrane pairs, which are specialized plant cell wall structures that permit xylem sap movement from one xylem vessel to its neighbors (44).

Polysaccharides present in these pit membranes therefore are potential substrates for the carbohydrate-active enzymes of E. tracheiphila and other xylem pathogens (9). We do not know if the clade-specific differences in the glycosyl hydrolase subfamilies confer differences in function. However, when we tested for functional differences conferred by cbsA, a 1,4-β- cellobiosidase in the GH6 subfamily that was present in Et-melo strains but absent in Et-C1 strains, we found evidence that it modulated the virulence of both an Et-melo strain (upon

52 deletion) and an Et-C1 strain (upon over-expression). The positive contribution to Et-melo virulence is consistent with the role of a cbsA homologue in the wilt pathogen Ralstonia solanacearum for which a cbsA deletion mutant reduced virulence on tomato plants (21).

The relative abundance and strain-specificity of the E. tracheiphila CAZymes further suggests roles for these enzymes in host-specific virulence, as they provide the opportunity to degrade polysaccharides that may differ among hosts. This possibility is supported by the association of distinct pit membrane polysaccharides in grapevine genotypes with resistance to the xylem pathogen Xylella fastidiosa(44). We hypothesize that the rate at which E. tracheiphila strains are able to wilt a host, and thus the host-specificity of the clades, is correlated with the presence of CAZymes that enable them to depolymerize and metabolize the polysaccharides in the pit membranes of a cucurbit host. The identification of clade-specific CAZymes in this report may help in testing this hypothesis in the future.

The physiological traits examined showed congruence with the clade designations. In particular, the carbon substrate utilization profiles highlighted a clear distinction between the two clades. Based on two representative strains per clade, the Et-melo clade was deficient in the ability to metabolize seven substrates that could be utilized by the Et-C1 clade. The identification of 10 substrates that are metabolized by strains from both groups supports the possibility of using these substrates to improve the in vitro growth of E. tracheiphila, which grows slower than most other enteric in vitro. Measurements of growth rates in culture and wilt rates on muskmelon both indicated substantial variation among strains, but also that strains of the Et- C1 clade grew slower and wilted muskmelon plants slower, on average, than Et-melo strains.

Collectively, the differences between the Et-melo and Et-C1 clade strains in genomic architecture, pangenomic features, gene content, plasmid content, carbon substrate utilization

53 profile, average growth rate, and virulence support differentiation of E. tracheiphila strains into two subspecies. We propose that this xylem-inhabiting, vector-transmitted bacterial pathogen of cucurbits be recognized to include at least two subspecies. In particular, we propose that the previously designated Et-melo and Et-C1 clades be designated as Erwinia tracheiphila subsp. tracheiphila and Erwinia tracheiphila subsp. lentilata, respectively. The subspecies designation tracheiphila is based on the convention in which the first subspecies of a species be given the species name. The subspecies designation lentilata is due to two traits exhibited by these strains, namely their slow growth in culture, as represented by the Latin word lenti for “slow”, and their broader geographical range than the closely related Et-C2 clade strains (36), as represented by the Latin word lata for “broad”. An appropriate designation for the Et-C2 clade of E. tracheiphila strains, although not characterized here, would be Erwinia tracheiphila subsp. lentangustus based on the Latin word angustus for “narrow”.

Description of Erwinia tracheiphila subsp. tracheiphila

Cells are Gram-negative, motile, and rod-shaped with peritrichous flagella. Mature colonies appear creamy white, circular, smooth, and shiny on King’s B agar after 3 to 4 days. The optimal in vitro growth temperature range is 25°C to 30°C, with 8°C and 37°C as the minimum and maximum, respectively. The cells induce a rapid wilt on the Cucumis species muskmelon (C. melo) and cucumber (C. sativus), but no wilt on Cucurbita species. Cells are able to metabolize

N-acetyl-D- lucosamine, D-galactose, D-mannitol, glucose-6-phosphate, D-fructose, α-D- glucose, sucrose, glucose-1-phosphate, β-methyl-D-glucoside and pectin but not D-saccharic acid, D-trehalose, glycerol, D,L-α-glycerol-phosphate, L-glutamic acid, fructose-6-phosphate, and M-inositol.

Description of Erwinia tracheiphila subsp. lentilata

Cells are Gram-negative, motile, and rod-shaped with peritrichous flagella. Mature colonies appear creamy white, circular, smooth, and shiny on King’s B agar after 4 to 6 days.

The optimal in vitro growth temperature range is 25°C to 30°C, with 8°C and 37°C as the minimum and maximum, respectively. The cells induce a slow wilt on muskmelon (Cucumis melo) and a rapid wilt on Cucurbita species. Cells are able to metabolize N-acetyl-D- glucosamine, D-saccharic acid, D-galactose, D-trehalose, glycerol, D,L-α-glycerol-phosphate, D- mannitol, L-glutamic acid, glucose-6-phosphate, D-fructose, α-D-glucose, sucrose, glucose-1- phosphate, fructose-6-phosphate, β-methyl-D-glucoside, M-inositol and pectin.

References

1. Alikhan NF, Petty NK, Ben Zakour NL, Beatson SA. 2011. BLAST Ring Image Generator (BRIG): Simple prokaryote genome comparisons. BMC Genomics 12 2. Andrade-Dominguez A, Kolte R, Shapiro LR. 2018. Complete genome sequence of EtG, the first phage sequenced from Erwinia tracheiphila. Microbiology Resource Announcements 6 3. Arndt D, Grant JR, Marcu A, Sajed T, Pon A, et al. 2016. PHASTER: A better, faster version of the PHAST phage search tool. Nucleic Acids Research 44:W16-W21 4. Aziz RK, Bartels D, Best AA, DeJongh M, Disz T, et al. 2008. The RAST server: Rapid annotations using subsystems technology. BMC Genomics 9 5. Bankevich A, Nurk S, Antipov D, Gurevich AA, Dvorkin M, et al. 2012. SPAdes: A New Genome Assembly Algorithm and Its Applications to Single-Cell Sequencing. J. Comput. Biol. 19:455-77 6. Bock C. 2012. Analysing and interpreting DNA methylation data. Nat. Rev. Genet. 13:705-19 7. Brady CL, Cleenwerck I, Denman S, Venter SN, Rodriguez-Palenzuela P, et al. 2012. Proposal to reclassify Brenneria quercina (Hildebrand and Schroth 1967) Hauben et al. 1999 into a new genus, Lonsdalea gen. nov., as Lonsdalea quercina comb. nov., descriptions of Lonsdalea quercina subsp quercina comb. nov., Lonsdalea quercina

subsp iberica subsp nov and Lonsdalea quercina subsp britannica subsp nov., emendation of the description of the genus Brenneria, reclassification of Dickeya dieffenbachiae as Dickeya dadantii subsp dieffenbachiae comb. nov., and emendation of the description of Dickeya dadantii. Int. J. Syst. Evol. Microbiol. 62:1592-602 8. Buonaurio R, Moretti C, da Silva DP, Cortese C, Ramos C, Venturi V. 2015. The olive knot disease as a model to study the role of interspecies bacterial communities in plant disease. Front. Plant Sci. 6 9. Busk PK, Lange L. 2015. Classification of fungal and bacterial lytic polysaccharide monooxygenases. BMC Genomics 16:13 10. Chen C, Malek AA, Wargo MJ, Hogan DA, Beattie GA. 2010. The ATP-binding cassette transporter Cbc (choline/betaine/carnitine) recruits multiple substrate-binding proteins with strong specificity for distinct quaternary ammonium compounds. Mol. Microbiol. 75:29-45 11. Danecek P, Auton A, Abecasis G, Albers CA, Banks E, et al. 2011. The variant call format and VCFtools. Bioinformatics 27:2156-8 12. Datsenko KA, Wanner BL. 2000. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proceedings of the National Academy of Sciences of the United States of America 97:6640-5 13. Freeman BC, Chen C, Yu XL, Nielsen L, Peterson K, Beattie GA. 2013. Physiological and transcriptional responses to osmotic stress of two Pseudomonas syringae strains that differ in epiphytic fitness and osmotolerance. Journal of Bacteriology 195:4742-52 14. Gardan L, Christen R, Achouak W, Prior P. 2004. Erwinia papayae sp nov., a pathogen of papaya (Carica papaya). Int. J. Syst. Evol. Microbiol. 54:107-13 15. Hauben L, Moore ERB, Vauterin L, Steenackers M, Mergaert J, et al. 1998. Phylogenetic position of phytopathogens within the Enterobacteriaceae. Syst. Appl. Microbiol. 21:384- 97 16. Heeb S, Itoh Y, Nishijyo T, Schnider U, Keel C, et al. 2000. Small, stable shuttle vectors based on the minimal pVS1 replicon for use in Gram-negative, plant-associated bacteria. Mol. Plant-Microbe Interact. 13:232-7 17. King EO, Ward MK, Raney DE. 1954. Two simple media for the demonstration of pyocyanin and fluorescein. J. Lab. Clin. Med. 44:301-7 18. Kwon SW, Go SJ, Kang HW, Ryu JC, Jo JK. 1997. Phylogenetic analysis of Erwinia species based on 16S rRNA gene sequences. International Journal of Systematic Bacteriology 47:1061-7 19. Li H. 2011. A statistical framework for SNP calling, mutation discovery, association mapping and population genetical parameter estimation from sequencing data. Bioinformatics 27:2987-93 20. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, et al. 2009. The sequence alignment/map format and SAMtools. Bioinformatics 25:2078-9

21. Liu HL, Zhang SP, Schell MA, Denny TP. 2005. Pyramiding, unmarked deletions in Ralstonia solanacearum shows that secreted proteins in addition to plant cell-wall- degrading enzymes contribute to virulence. Molecular Plant-Microbe Interactions 18:1296-305 22. Liu Q. 2015. Bacterial wilt of cucurbits: Impact of plant age on symptom progression and pathogen movement and locating genes associated with host preference and pathogenesis in E. tracheiphila. Iowa State University. 87 pp pp. 23. Mayjonade B, Gouzy J, Donnadieu C, Pouilly N, Marande W, et al. 2016. Extraction of high-molecular-weight genomic DNA for long-read sequencing of single molecules. Biotechniques 61:203-5 24. Meinke A, Gilkes NR, Kwan E, Kilburn DG, Warren RAJ, Miller RC. 1994. Cellobiohydrolase-a (Cbha) from the cellulolytic bacterium Cellulomonas fimi is a beta- 1,4-exocellobiohydrolase analogous to Trichoderma reesei CBH-II. Molecular Microbiology 12:413-22 25. Mergaert J, Hauben L, Cnockaert MC, Swings J. 1999. Reclassification of non- pigmented Erwinia herbicola strains from trees as Erwinia billingiae sp. nov. International Journal of Systematic Bacteriology 49:377-83 26. Nazareno ES, Dumenyo CK. 2015. Modified inoculation and disease assessment methods reveal host specificity in Erwinia tracheiphila-Cucurbitaceae interactions. Microb. Pathog. 89:184-7 27. Page AJ, Cummins CA, Hunt M, Wong VK, Reuter S, et al. 2015. Roary: Rapid large- scale prokaryote pan genome analysis. Bioinformatics 31:3691-3 28. Parte AC. 2018. LPSN - List of prokaryotic names with standing in nomenclature (bacterio.net), 20 years on. International Journal of Systematic and Evolutionary Microbiology 68:1825-9 29. Pisabarro A, Correia A, Martin JF. 1998. Pulsed-field gel electrophoresis analysis of the genome of Rhodococcus fascians: Genome size and linear and circular replicon composition in virulent and avirulent strains. Current Microbiology 36:302-8 30. Rambaut A. 2008. FigTree. http://tree.bio.ed.ac.uk/software/figtree/ 31. Rolle RL, Ejiofor AO, Johnson TL. 2005. Determination of the plasmid size and location of delta-endotoxin genes of Bacillus thuringiensis by pulse field gel electrophoresis. African Journal of Biotechnology 4:580-5 32. Saalau Rojas E, Dixon PM, Batzer JC, Gleason ML. 2013. Genetic and virulence variability among Erwinia tracheiphila strains recovered from different cucurbit hosts. Phytopathology 103:900-5 33. Schrader KA, Cheng DT, Joseph V, Prasad M, Walsh M, et al. 2016. Germline variants in targeted tumor sequencing using matched normal DNA. JAMA Oncol. 2:104-11 34. Seemann T. 2014. Prokka: Rapid prokaryotic genome annotation. Bioinformatics 30:2068-9

35. Shapiro LR, Andrade A, Scully ED, Rocha J, Paulson JN, Kolter R. 2018. Draft genome sequence of an Erwinia tracheiphila isolate from an infected muskmelon (Cucumis melo). Microbiol. Resour. Ann. 7:2 36. Shapiro LR, Paulson JN, Arnold BJ, Scully ED, Zhaxybayeva O, et al. 2018. An introduced crop plant is driving diversification of the virulent bacterial pathogen Erwinia tracheiphila. mBio 9:20 37. Shapiro LR, Scully ED, Roberts D, Straub TJ, Geib SM, et al. 2015. Draft genome sequence of Erwinia tracheiphila, an economically important bacterial pathogen of cucurbits. Microbiology Resource Announcements 3:2 38. Shapiro LR, Scully ED, Straub TJ, Park J, Stephenson AG, et al. 2016. Horizontal gene acquisitions, mobile element proliferation, and genome decay in the host-restricted plant pathogen Erwinia tracheiphila. Genome Biol. Evol. 8:649-64 39. Simon A. 2010. FastQC: a quality control tool for high throughput sequence data. https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ 40. Smith EF. 1911. Bacteria in relation to plant diseases. Nature 74:294 41. St-Pierre F, Cui L, Priest DG, Endy D, Dodd IB, Shearwin KE. 2013. One-step cloning and chromosomal integration of DNA. ACS Synthetic Biology 2:537-41 42. Stamatakis A. 2014. RAxML version 8: A tool for phylogenetic analysis and post- analysis of large phylogenies. Bioinformatics 30:1312-3 43. Starr MP, Chatterjee AK. 1972. Genus Erwinia: Enterobacteria pathogenic to plants and animals. Annu. Rev. Microbiol. 26:389-426 44. Sun QA, Greve LC, Labavitch JM. 2011. Polysaccharide compositions of intervessel pit membranes contribute to Pierce's disease resistance of grapevines. Plant Physiology 155:1976-87 45. Vrisman CM, Deblais L, Rajashekara G, Miller SA. 2016. Differential colonization dynamics of cucurbit hosts by Erwinia tracheiphila. Phytopathology 106:684-92 46. Wright HD. 1934. The preparation of nutrient agar with special reference to pneumococci, streptococci and other Gram-positive organisms. Journal of Pathology and Bacteriology 39:359-73 47. Xie YZ, Wei YQ, Shen Y, Li XB, Zhou H, et al. 2018. TADB 2.0: An updated database of bacterial type II toxin-antitoxin loci. Nucleic Acids Research 46:D749-D53 48. Xie ZQ, Tang HX. 2017. ISEScan: Automated identification of insertion sequence elements in prokaryotic genomes. Bioinformatics 33:3340-7 49. Zhang H, Yohe T, Huang L, Entwistle S, Wu PZ, et al. 2018. dbCAN2: A meta server for automated carbohydrate-active enzyme annotation. Nucleic Acids Research 46:W95- W101

CHAPTER 3 THE EFFECTORS EOP1 AND DSPE ARE DRIVERS OF HOST-SPECIFICITY AND

PATHOGENICITY AMONG ERWINIA TRACHEIPHILA STRAINS

Olakunle I. Olawole1, Qian Liu1, Chilang Chen1, Mark L. Gleason1 and Gwyn A. Beattie1.

1Department of Plant Pathology and Microbiology, Iowa State University, Ames, IA, U.S.A.

All of the data presented in this chapter were generated and analyzed by O.O. with the exception of the initial Eop1 and DspE sequence comparisons and lambda-red-recombinase-transformed constructs (generated by Q.L.) and eop1 and dspE deletion mutants in SCR3 (collaboratively generated with C.C.).

Abstract

Erwinia tracheiphila is a xylem-inhabiting and economically important bacterium that causes wilt of cucurbits in the Cucumis (muskmelon and cucumber) and Cucurbita (squash and pumpkin) genera. Two clades of E. tracheiphila strains, Et-melo and Et-C1, exhibit differences in host-specific virulence based on the ability of Et-melo to wilt Cucumis but not Cucurbita species, and Et-C1 to wilt Cucurbita species, designated primary hosts, more rapidly than

Cucumis species, designated secondary hosts. To identify the molecular mechanisms underlying this host specificity, we investigated the effector proteins Eop1 and DspE as host-specific virulence and pathogenicity candidates, respectively. Deletion of eop1, which encodes an effector involved in host-specific virulence in Erwinia amylovora, did not alter the virulence of two Et-melo strains, SCR3 and MDCuke, or an Et-C1 strain, BHKY, on muskmelon or squash.

However, over-expression of eop1 from Et-melo MDCuke in Et-C1 BHKY increased the

59 virulence of BHKY on its secondary host muskmelon, but not its primary host squash, indicating that Eop1 functions as a host-specific virulence factor. Loss of dspE from the Et-melo strains significantly reduced, but did not eliminate, virulence on muskmelon and cucumber, whereas loss of dspE from the Et-C1 strains resulted in a complete loss of pathogenicity on squash and muskmelon. Thus, DspE has distinct roles in the two E. tracheiphila clades. Moreover, the similarity in virulence of Et-melo mutants lacking either dspE or the adjacent chaperone- encoding gene dspF supports DspF as the cognate chaperone for DspE. Collectively, these results highlight Eop1 as a molecular driver of host specificity in E. tracheiphila and DspE as an effector with distinct roles in distinct clades of E. tracheiphila strains.

Introduction

Erwinia tracheiphila is a xylem pathogen that causes bacterial wilt of cucurbits in the

Cucurbita (squash and pumpkin) and Cucumis (muskmelon and cucumber) genera (53). This pathogen is transmitted by striped and spotted cucumber beetles (Acalymma vittatum and

Diabrotica undecimpunctata, respectively). When these beetles feed on leaves or flowers of cucurbits, they can deposit E. tracheiphila-infested frass onto fresh feeding wounds, allowing bacterial entry into the xylem (45). E. tracheiphila cells multiply inside the xylem, producing exopolysaccharides that likely help obstruct the vascular system, and promote wilt, collapse and eventual death of plants (49; 63). Although cucurbits are grown throughout the United States, bacterial wilt is restricted to the Northeastern and Midwestern regions where approximately 68% of the nation’s total cucurbit acreage is cultivated (Rojas, Dixon et al. 2013), and crop losses can reach 80% (12).

Management of bacterial wilt primarily depends on frequent insecticide applications to reduce the cucumber beetle populations. However, this can pose a danger to the health of insect pollinators and insectivorous birds, thus creating regulatory uncertainty as to their continued use

(59; 61). A few alternative management strategies include the establishment of row tunnels to prevent beetle feeding (20) and perimeter trap crops to lure beetle vectors to an alternate crop (7;

60)

E. tracheiphila strains infect cucurbits in a host-specific manner as wilt symptoms often progress faster when seedlings are inoculated with strains that originated from the same crop host genus (42; 50; 53). The genetic variability of 69 strains isolated from 5 distinct host genera showed that all of the strains fell into two groups based on their rep-PCR fingerprints (50). This genetic variability correlated with host-specific virulence patterns of the strains on muskmelon and squash.

Recently, based on similarity across the genome sequences, 88 E. tracheiphila strains were found to cluster into three phylogenetic clades (Et-melo, Et-C1 and Et-C2) (51). The Et-melo and Et-C1 clades reflected the original groups identified by rep-PCR, and the Et-C2 clade contained a newly isolated set of strains with a similar host specificity to Et-C1 but from a more localized geographic range (51). Although these studies have consistently linked genetic variability to host-specific patterns in virulence, a deeper understanding of the mechanisms underlying these patterns has been hindered by the lack of adaptation of genetic tools for manipulating E. tracheiphila.

In this study, we utilized a modified lambda red recombineering system (11) and a splice overlap extension PCR mutagenesis protocol (22) to explore the genetic basis of host-specific virulence and pathogenicity among E. tracheiphila strains. The genome of the closely-related pathogen E. amylovora encodes two major effectors, Eop1 (Erwinia outer protein-1) and DspE

(disease specific protein-E), that function as host-range limiting (2) and pathogenicity (15)

61 factors, respectively. The Eop1 homologue PopP1 in Ralstonia solanacearum similarly functions as a host-specificity factor (31). In addition, DspE homologues of many phytopathogenic bacteria play essential roles in virulence and pathogenicity (4; 10; 24).

Effectors are proteins secreted and delivered into the host cells through the type III secretion system (T3SS), which is a complex membrane-spanning bacterial structure (1; 6; 9;

26). T3SSs can be grouped into at least three different families based on phylogeny and synteny.

These include the Hrp1 (hypersensitive response and pathogenicity) T3SS, which is found in

Pseudomonas syringae and Erwinia spp; the Hrp2 T3SS, which is found in Ralstonia solanacearum, Xanthomonas spp., Acidovorax and Burkholderia; and a rhizobial-like Hrp3

T3SS which is found in rhizobia and some P. syringae strains (1; 17). Eop1 belongs to the

YopJ/AvrRxv family of type III secreted effectors (T3SEs) and homologues are found across several species of the Proteobacteria, including Pseudomonas (HopZ), Xanthomonas (AvrBsT,

AvrRxv, AvrXv4 and XopJ), Erwinia (Eop1), Ralstonia (PopP1), and the plant symbiont

Rhizobium (Y4LO) (32; 46). Effectors of this family have been demonstrated to have either a cysteine protease or an acetyltransferase activity (38; 41; 56; 64). YopJ homologues in

Pseudomonas syringae include three major forms, HopZ1, HopZ2 and HopZ3, with at least one of these, HopZ1, diversified into three distinct forms, HopZ1a, HopZ1b and HopZ1c (36). The

YopJ homologues AvrBsT and AvrXv4 in Xanthomonas species, PopP1 in Ralstonia solanacearum and HopZ2 in Pseudomonas syringae function in R-gene mediated resistance (33;

55; 58), and HopZ3 triggers defense responses that are suppressed by other effectors (48).

DspE belongs to the AvrE superfamily of effectors, which is a superfamily that is present in almost all T3SS-dependent phytopathogenic bacteria. This superfamily includes DspA/E in E. amylovora, DspE in Pectobacterium species, DspE/A in Pantoea agglomerans pv. gypsophilae,

WtsE in Pantoea stewartii subsp. stewartii, AvrE in Pseudomonas syringae, PopS in Ralstonia solanacearum and XopAM in Xanthomonas vesicatoria (13). Members of the AvrE superfamily are core effectors generally critical to the virulence of the bacteria that encode them. The AvrE homologue in Pantoea stewartii is the only T3SE known to be secreted by this bacterial pathogen, and it is absolutely required for its pathogenicity on sweetcorn (19). Although E. amylovora (4; 15; 52) and Pectobacterium carotovorum (27) have multiple effectors, loss of

DspE or DspA/E resulted in complete loss of pathogenicity. In contrast, the loss of AvrE homologues in P. syringae (3), R. solanacearum (24), and X. campestris pv. campestris (18) reduced, but did not eliminate, virulence on their respective hosts. AvrE effectors are very large proteins of about 192Da, and this large size could be associated with these contributions to pathogenicity and virulence (13). Most AvrE/DspE/WtsE homologues are co-transcribed with cognate chaperones (16); however, R. solanacearum PopS and X. campestris pv. campetris

XopAM both lack genes for clear candidate chaperones, suggesting that they do not need these chaperones or are utilizing chaperones encoded elsewhere within their genomes (13).

To evaluate the roles of Eop1 and DspE in host-specific virulence and pathogenicity of E. tracheiphila, respectively, we constructed deletion mutants lacking eop1 or dspE (with or without the chaperone, dspF) in selected Et-melo and Et-C1 strains, and analyzed virulence on squash, muskmelon and cucumber plants. The role of Eop1 in avirulence was also evaluated through homologous and heterologous over-expression of eop1 in these strains. Our result shows that Eop1 functions in host-specific virulence, whereas DspE exhibits distinct roles in virulence among E. tracheiphila strains. This provides the first characterization of molecular drivers of host specificity and pathogenicity within Erwinia tracheiphila strains.

Materials and methods

Plant growth conditions

The cucurbit cultivars used in this study include muskmelon (Cucumis melo cv. Athena), summer squash (Cucurbita pepo cv. Early Summer Crookneck), zucchini squash (Cucurbita pepo cv. Partenon F1 organically grown) and cucumber (Cucumis sativus cv. Marketmore, organically grown). Seeds were sown in 10-cm pots containing a matrix of peat moss, coarse perlite and Metro-Mix 300 (Sun Gro Horticulture, Canada Ltd.; Vancouver, BC, Canada).

Seedlings were maintained in a growth chamber (Percival Scientific Inc.) at 28°C under a photoperiod of 12 h of light and 12 h of darkness and 70% relative humidity (RH). Plants were watered every other day and fertilizer solution (NPK: 15-5-15: Peters Excel, ICL UK/Ire) was added, at two days before inoculation.

Bacterial strains, growth conditions and inoculum preparation

The bacterial strains and plasmids for this study are described in Table 1. E. tracheiphila strains were grown in King’s B (KB) medium (28) at 28◦C, unless otherwise described.

Escherichia coli strains were grown in Luria medium at 37◦C. The following antibiotics were added at 50µg/ml: rifampin (Rif), kanamycin (Km), spectinomycin (Spc), and ampicillin (Amp) when needed. To prepare cells for plant inoculation assays, E. tracheiphila strains were recovered from cryogenically preserved glycerol stocks (-80◦C) on KB agar amended with Rif.

The cells were incubated for 3-4 days, after which a single colony was transferred to a fresh plate to make a lawn. Cells from a two-day old lawn were suspended in 10 mM phosphate-buffer (PB)

8 and normalized to an optical density at 600nm (OD600) of 0.5 (approximately 2.5 x 10 CFU/ml).

A non-ionic organosilicone surfactant (Silwet L77) was added to each bacterial suspension at

0.02% (vol/vol).

Table 1: Bacterial strains and plasmids used in this study

Antibiotic Source or Strains and plasmids Description resistance reference

Erwinia tracheiphila SCR3 Et-melo strain isolated from Cucumis melo in Rif (50) Iowa MDCUKE Et-melo strain isolated from Cucumis sativus in None (50) Maryland UnisCu1-1 Et-melo strain isolated from Cucumis sativus in None (50) Iowa HCa1-5 Et-melo strain isolated from Cucumis sativus in None (50) Iowa BHKY Et-C1 strain isolated from Cucurbita moschata Rif (50) in Kentucky MISpSq Et-C1 strain isolated from Cucurbita pepo in Rif (50) Michigan BuffGH Et-C1 strain isolated from Cucurbita pepo in Rif (50) Pennsylvania SCR3∆eop1 SCR3 eop1 deletion mutant Rif This study BHKY∆eop1 BHKY eop1 deletion mutant Rif This study MDCUKE∆eop1 MDCuke eop1 deletion mutant Spc This study BHKY∆eop1(pN-eop1BHKY) BHKY∆eop1 overexpressing eop1 from BHKY Rif, Km This study BHKY∆eop1(pN-eop1MDCUKE) BHKY∆eop1 overexpressing eop1 from Rif, Km This study MDCuke MDCUKE∆eop1(pN-eop1MDCUKE) MDCuke∆eop1 overexpressing eop1 from Km This study MDCuke MDCUKE∆eop1(pN-eop1BHKY) MDCuke∆eop1 overexpressing eop1 from Km This study BHKY SCR3∆dspF SCR3 dspF deletion mutant Rif This study SCR3∆dspE SCR3 dspE deletion mutant Rif This study SCR3∆dspEF SCR3 dspEF deletion mutant Rif This study MDCUKE∆dspF MDCuke dspF deletion mutant Rif This study MDCUKE∆dspE MDCuke dspE deletion mutant Rif This study MDCUKE∆dspEF MDCuke dspEF deletion mutant Rif This study BHKY∆dspEF BHKY dspEF deletion mutant Rif This study BuffGH∆dspEF BuffGH dspEF deletion mutant Rif This study

Plasmids pN pME6041 with nptII promoter next to a Km (8) multiple cloning site pKD13 Template for Kan cassette bordered by FLP Amp, Cm (11) recombination target sites pKD46 λ Red recombinase used to promote Amp (11) recombination of linear PCR products, maintained in E. tracheiphila. pFlp2Ω Encodes Flp recombinase that displaces Kan Amp (62) cassette. pTOK2T pTOK2 with restored lacZ activity Tet (8)

Inoculation and disease assessment

Two-week old seedlings of each cucurbit cultivar were inoculated with E. tracheiphila.

The site closest to the petiole of the adaxial surface of the youngest fully expanded leaf was punctured with a 28.6-mm-diameter florist’s pin frog (Kenzan Pin Frog, sold by www.save-on- crafts.com), and 200 μl of cell suspension, prepared as described above, was applied to the punctured site. Plants inoculated with PB in the same manner as described above were used as controls. Inoculated plants were incubated as described for the seedlings, above, and rated daily for wilt development based on the total number of leaves and the number of leaves wilted per inoculated plant. Once plants reached 100% wilt, newly emerging leaves were not counted and the plants were not monitored further. Data were collected daily for two weeks.

Construction of deletion mutants and complementation constructs

The primers used in this study are shown in Table 2. For each gene targeted for deletion, primers were targeted to regions that were conserved across the Et-melo and Et-C1 clades.

Deletion mutants were generated by using the lambda Red recombinase system, which is comprised of recombinase components expressed under the control of an arabinose-inducible promoter on the plasmid pKD46 (11). Linear PCR products were generated and introduced via electroporation into E. tracheiphila cells that had been transformed with pKD46. These linear

PCR products contained a kanamycin cassette (kan) flanked by flippase recognition target (FRT) sites, as amplified from plasmid pKD13. The linear PCR products used to generate the dspEF mutants included 1-kb regions flanking dspEF in the genomes, whereas the linear PCR products used to generate the eop1, dspE and dspF mutants included 90-nt regions flanking these genes in the genomes.

Table 2. Primers used for this study

Primer Sequences1

Primers for constructing deletion mutants 5’- CCCACGTTGATGGCGCAGGGGTTAACACTCAAAGCAGCGGGGAGAA eop1-Km-F TTCGACGTCTTGAGCGATTGTGTAGGCT-3’ 5’- CAGGAAAAAAAGCACGGGTATTAACGGTGCGTTTTATTAAAGGGGT eop1-Km-R AAAAAGTGATTGCGCCTACCCGGATATT-3’ 5’- AATATCCGGGTAGGCGCAATCACTTTTTACCCCTTTAATAAAACGCA F2-eop1 -3’ R2-eop1 5’-TTTACCGAATAAACGGCTGAAACT-3’ F1-dspE 5’-AAAATG TGCGGATCCCGTCAGAG-3’ 5’- AGCCTACACAATCGCTCAAGACGTGACCGCCCCCGTTCCCACATTA R1-dspE A-3’ 5’- AATATCCGGGTAGGCGCAATCACTTTAAGCCTCATCCCTGATATCGC F2-dspE -3’ R2-dspE 5’-CTTCCCTTAAAATCGCAGGCTAT-3’ F3-Km 5’-ACGTCTTGAGCGATTGTGTAGGCT-3’ R3-Km 5’-AGTGATTGCGCCTACCCGGATATT-3’ GACTGAAGAAAGAAGGGTTTGAAATGAAGGGCTGATTGACGTAGCC GGTCAGACTCAGATTCGATAATCACCACATTAAGGAGTGACATGAC dspF-KmF GTCTTGAGCGATTGTGTAGGCT GCCAGCAACAACCTTATCCCCTCGCCGTTGAGAACAGCTGAAGAGC GGTATACGCCCGTGCGTTAACGGCACTGGCGTATACCGTTATCCAGT dspF-KmR GATTGCGCCTACCCGGATATT dspF-Ex-F ATACGCCTGCAATGTTACTGCG dspF-Ex-R TGATTCCGTTTCGTCAAGCGC TGCTGTTTTCCCAACTCTATTTTCTGCGACGGGGAACCGTCGAATTA AGAGCACTACATAAACAAAATTAATGTGGGAACGGGGGCGGTCAC dspE-KmF GTCTTGAGCGATTGTGTAGGCT AGCAGCTTTTCAACCTGCTGTTGTGTCGGTGTCATCATGTCACTCCTT AATGTGGTGATTATCGAATCTGAGTCTGACCGGCTACGTCAAAGTG dspE-KmR ATTGCGCCTACCCGGATATT dspE-ExF ATGTTAGTGTGGTGAGACAATCGTCTC dspE-ExR ACGCCATCCTTTAACTGCAAAGG

Primers for constructing eop1-overexpression fusions esc1-eop1-MDCUKE-F AGTGCAATGATGGGCGATACGG esc1-eop1-MDCUKE-R CGTTTCTGGCTTACGCAACCAAC esc1-eop1-BHKY-F TGATGGGTGATACGGGTAACAGCAAC esc1-eop1-BHKY-R GCTGGAGGAGAAATTCGACGTTCTC pN-ME60-F AAGGTCATCCACCGGATCAATTCC pN-ME60-R TCTATCGATGCATGCCATGGTACC pN(eop1)-F ATGGACAGCAAGCGAACCGG pN(eop1)-R CTAATACTGCCCTTGATGAACTGGCCG 1 Regions underlined in the primer sequence are targeted to the kan cassette on pKD13. Ex = primers ‘external’ to targeted genes to be deleted. esc = eop1 chaperone

To generate mutants, E. tracheiphila cells containing pKD46 were grown at 28°C for 2-3

days in KB broth amended with Amp. When cells reached an OD600 of 0.3, freshly-made L- arabinose was added to a final concentration of 10mM to induce expression of the lambda phage Red recombinase genes. Cells were harvested at an OD600 of 0.5 and washed three times with room temperature sterile distilled water. The room temperature wash was adopted because

E. tracheiphila competency appeared to be reduced by washing in cold sterile water. After the washes, the cells were suspended in sterile distilled water and placed on ice for 15 to 20 minutes.

The chimeric linear fragments that contained the flanked kan cassette, as described above, were introduced into the ice-cold electrocompetent cells at a concentration of 2 to 3 μg per 100 μl of cells in a 0.2-cm cuvette using a gene pulser electroporator (Bio-Rad, Hercules, CA).

Electroporated cells were transferred to 1ml of ice-cold KB broth and incubated for 6 to 10 hours, after which cells were centrifuged and pellets were plated on Km-supplemented solid KB agar to select for successful mutants. After confirmation of the mutants by PCR, the kan cassette was removed by introduction of pFlp2Ω (62), which encodes a recombinase, the Flp recombinase, that recognizes the FRT sites for recombination.

A linear fragment for constructing the ΔdspEF mutants was generated by using splice- overlap-extension PCR (22). In short, 1-kb regions were amplified from E. tracheiphila strain

BuffGH with primers targeting the regions upstream and downstream of dspEF; these regions were conserved across E. tracheiphila strains. These fragments were ligated to the FRT-kan-FRT cassette, generated using pKD13 as a template. The resulting fragment was cloned into pTOK2T

(8) such that the SmaI site was preserved, and this plasmid was used as a template in a PCR reaction with the F1-dspE and R2-dspE primer pairs (Table 3). The resulting linear fragment was electroporated into cells of E. tracheiphila strains MDCuke(pKD46) and BHKY(pKD46). The linear fragments used to generate the Δeop1, ΔdspE and ΔdspF mutants contained the FRT-kan-

FRT cassette that was amplified from plasmid pKD13 using primers that incorporated into the final product 90-bp sequences homologous to the regions immediately flanking the target genes in the genomes. These fragments were electroporated into E. tracheiphila cells expressing pKD46, after which the kan cassette was removed. To evaluate complementation of the Δeop1 mutants, the eop1 genes from MDCuke and BHKY were PCR-amplified with their native promoters and cloned into the EcoRV site of pN (8), which is a derivative of the broad-host- range vector pME6041 (21) that has a 733-bp nptII promoter inserted upstream of the multiple cloning site. The resulting constructs, pN-eop1MDCuke and pN-eop1BHKY, contained eop1 under the control of the native promoter fused in tandem with the nptII promoter. These plasmids were introduced into the SCR3Δeop1 and BHKYΔeop1 mutants by electroporation.

Evaluation of in vitro and in planta gene expression

To evaluate the in vitro expression of eop1 in various strain constructs, three independent cultures of BHKY, BHKYΔeop1, and BHKYΔeop1(pN-eop1) were grown in KB medium for 2-

3 days. Bacterial cultures were normalized to an OD600 of 1.0, and using an RNeasy Mini Kit

(Qiagen), RNA was extracted from 1ml of each culture with DNA removed using on-column

DNase I digestion (Qiagen, Venlo, Limburg). To evaluate eop1 expression in planta, two-week old seedlings of muskmelon were inoculated as described above with BHKY, BHKYΔeop1,

BHKYΔeop1(pN-eop1MDCuke) and BHKYΔeop1(pN-eop1BHKY). Inoculated plants were incubated until the first appearance of wilt symptoms, at which time xylem sap was collected from infected plants, following surface sterilization of excised segment with 70% ethanol. The xylem sap was collected by manually squeezing the aerial half of the cut stem and collecting the sap with a 100 µl pipette. Immediately following xylem sap collection, the sap was combined

69 with RNAProtect for 10 minutes at room temperature to preserve RNA integrity. The cells in this mixture were collected by centrifugation and pellets were either used immediately for RNA extraction using an RNeasy Mini Kit or were frozen at -80OC for RNA extraction at a later time.

The eop1 transcripts were measured using quantitative reverse transcription-PCR (RT- qPCR) with the qScript One-Step SYBR Green RT-qPCR kit (Quantabio, Gaithersburg, MD).

Relative expression was calculated from the cycle threshold (Ct) values, using expression of the recA gene as the internal control. The primers 5’- GGGGTATGATGAATGAGATTCGGC -3’ and 5’- ATTTCAGCGCATTACTCAGCGC -3’ were used to evaluate expression of eop1, and

5’- CCATATCTACGGGTTCTCTCTCGC -3’ and 5’- AGCGATAACCTGCAAAGTCAGC -3’ were used to evaluate expression of recA. One hundred nanograms of total RNA was used for each of three technical replicates, each in a total reaction of 25 µl. Real-time PCR was performed using the Mastercycler® ep realplex (Eppendorf), under the following conditions: one cycle at

O ◦ ◦ 95 C for 2 min and 30 s, 40 cycles at 95 C for 15 s and 60 C for 30 s. The resulting Ct values were calculated using the relative standard curve method (37) in which the Ct values of each tested gene was normalized to the Ct values of recA.

Sequence alignment and generation of phylogenetic trees

The amino acid sequences of Eop1 or AvrE of E. tracheiphila strains were generated from their whole genome sequences from this study and publicly available genome sequences.

Proteins sequences were aligned using MUSCLE (14), phylogenetic trees were produced using

RAxML (54) and trees were viewed in FigTree 1.4.4 (44). A maximum likelihood tree of the

YopJ sequences was generated and supported with a bootstrap of 1,000 replicates.

Statistical analyses

Total and wilted number of leaves were used to calculate the proportion of wilted leaves, which was further subjected to an arcin-squareroot transformation for normalization, as is common for data expressed as proportions. Virulence was visually represented as disease progress curves (Arcsin (square root (proportion wilted)) over time), which were quantified based on the area under disease progress curve (AUDPC). Graphs were generated using

SigmaPlot 14 and statistical comparisons were made using JMP Pro 12 (SAS Institute Inc., Cary,

NC). AUPDC values were subjected to analysis of variance (ANOVA) and values were separated by using the Fisher’s Least Significant Difference (L.S.D.) test at 5% level of significance. Results are the mean and standard error of mean (SEM) of data from an experiment. For the RT-qPCR data, the average of Ct values of three replicates were calculated and values were used to generate bar charts. All Ct values were subjected to ANOVA and values were separated by Fisher’s L.S.D. test at 5% level of significance.

Results

Eop1 sequence differences correlate with Erwinia tracheiphila clades

Eop1 was identified as a putative factor influencing host-specific virulence among E. tracheiphila strains based on the role of an Eop1 homologue as a host-range limiting factor in E. amylovora (2) and host-specificity factor in R. solanacearum (31). The genome of E. tracheiphila encodes an Eop1 homologue that shares 59% sequence similarity with E. amylovora

Eop1 (Fig 1 and Table 2). The Eop1 sequences from selected strains of the Et-melo and Et-C1 clades exhibited amino acid sequences within each clade but differences between clades (Fig 2).

Moreover, eop1 gene exhibits conserved synteny within the type III secretion system loci of

Figure 1: Alignment of Eop1 sequences of Erwinia amylovora with an Et-melo strain SCR3 and an Et-C1 strain BHKY strains of Erwinia tracheiphila. Amino acid sequences were aligned using Clustal Omega in MEGA-X. (Note: This figure will be in the supplemental data in the submitted manuscript)

Erwinia species (Fig 3). We developed a maximum-likelihood phylogenetic tree in MEGA-X, of

Eop1 protein sequences from selected Et-melo and Et-C1 strains, and strains of Erwinia,

Pseudomonas, Burkholderia and Xanthomonas species (Fig 4). The E. tracheiphila strains formed a unique clade, with the Et-melo strains clearly separated from the Et-C1 strains. Other

Eop1 homologues formed distinct clades within distinct orders, although families within the

Enterobacteriales order fell into two distinct clades, indicating that members of this effector family are widely distributed and evolutionarily diverse. Based on the possibility of conservation

Table 3: Percent Identity Matrix of Alignment of Eop1 sequences of Erwinia amylovora with an Et- melo strain SCR3 and an Et-C1 strain BHKY strains of Erwinia tracheiphila (Note: This table will be in the supplemental data in the submitted manuscript)

E. amylovora Et-melo (SCR3) Et-C1 (BHKY) E. amylovora 100.00 58.78 58.52 Et-melo (SCR3) 58.78 100.00 99.27 Et-C1 (BHKY) 58.52 99.27 100.00

Figure 2: Et-melo and Et-C1 strains of Erwinia tracheiphila exhibit clade-specific sequence differences. Eop1 sequences of three Et-C1 strains (BuffGH, BHKY and MISpSq) and four Et-melo strains (UnisCu1-5, Hcal-5, SCR3 and MDCuke) exhibited 99.97% sequence identity within each group. However, there were 3 amino acid differences and one 3-amino acid-insertion/deletion event (highlighted in yellow) between the two groups. Amino acid sequences were aligned using Clustal Omega in MEGA-X. of function among diverse taxa, we hypothesized that Eop1 has a role in determining host- specific virulence in E. tracheiphila as it does in E. amylovora and R. solanacearum.

Figure 3. Illustration of syntenic regions within the type III secretion system loci of Erwinia species with complete genome sequences. (Note: This figure will be in the supplemental data in the submitted manuscript). Reproduced from the thesis of Qian Liu (34).

Figure 4: Members of the YopJ effector family are evolutionarily conserved. Phylogenetic analysis clusters known YopJ effector family members into distinct clades, with the E. tracheiphila eop1 being clearly separated among the two subspecies and from the eop1 orthologs of other species in the Erwiniaceae family. The Pseudomonadales, Burkholderia species complex and the Xanthomonadales each formed a unique group. Proteins sequences were aligned using MUSCLE (14), phylogenetic tree was produced using RAxML (54) and trees were viewed in FigTree 1.4.4 (44). Generated maximum likelihood tree of the YopJ sequences was supported with a bootstrap of 1000 replicates. Eop1 sequences of BHKY and SCR3 were generated from their complete genomes which we sequenced using the PacBio platform; while Eop1 sequences of UnisCu1-1, Hca1-5 and MISpSq were generated via Sanger sequencing (34). Salmonella enterica was selected as the outgroup lineage based on its status as the species outside the Erwinia genus, that had the top-scoring match in the BLAST searches of the selected proteins. Sequence divergence is represented by the bar.

Deletion of eop1 did not alter the virulence of E. tracheiphila strains on muskmelon or

squash

Deletion mutants were generated by expressing the lambda red recombinase system (11)

on a stable plasmid, thus enabling this recombinase rather than a native recombinase to promote

homologous recombination. The eop1 gene of several E. tracheiphila strains was first replaced

74 with a kan cassette, and this was then removed to generate an unmarked ∆eop1 deletion mutants

(Fig 5).

Figure 5. Illustration of eop1 and dspE deletions and expression constructs generated in this study. The eop1-dspE locus contains genes for the HrpN, Eop1, and DspE effectors, as well as a putative Eop1-specific chaperone esc1, a chaperone for the HrpW effector, although the hrpW gene is absent in E. tracheiphila genome, and the DspE chaperone, DspF. Deletions of the eop1, dspF, dspE and dspEF genes are shown as dotted lines. The region cloned to test for the effects of eop1 expression is shown as a fusion to the constitutive promoter PnptII.

Deleting eop1 from the Et-melo strains MDCuke (MDCuke∆eop1) and SCR3 (SCR3∆eop1) did not alter their virulence on a muskmelon host (Fig 6), nor did it enable them to infect squash

(data not shown). Similarly, deletion of eop1 from the Et-C1 strain BHKY (BHKY∆eop1) did not alter its virulence on either its primary host squash or secondary host muskmelon (Fig 6).

These findings indicate that in contrast to its role in E. amylovora, Eop1 is not a host-limiting factor in E. tracheiphila. However, Eop1 could contribute to virulence, and even host-specific virulence, but this contribution could have escaped detection due to compensation by other effectors, as is common for type III secreted effectors.

Figure 6: Deletion of the effector eop1 gene did not alter the virulence of Erwinia tracheiphila strains on primary or secondary host species. Transformed proportion wilted (left panels) and area under disease progress curve (right panels) of wild-type strains and eop1 deletion mutants of the Et-melo strains SCR3 (A) and MDCuke (B) on their primary host muskmelon; and an Et-C1 strain BHKY on its secondary host muskmelon (C) and primary host squash (D). The proportion wilted was transformed with an arcsine (square root) transformation to achieve normality. Lines represent the standard error of mean within a figure of the transformed proportion wilted while bars represent the mean area under disease progress curve. Different letters indicate significant difference based on a Least Significant Difference test (P = 0.05). Each experiment was repeated at least twice with 9 replicates per treatment. (Note: This figure will be in the supplemental data in the submitted manuscript).

Over-expression of eop1 from an Et-melo strain in an Et-C1 strain increased virulence on the secondary host muskmelon but not the primary host squash

The Et-C1 strain BHKY causes rapid wilt on squash and slow wilt on muskmelon; in contrast, the Et-melo strain MDCuke causes rapid wilt on muskmelon and does not wilt squash. To

Figure 7: Homologous and heterologous over-expression of the effector eop1 gene did not alter virulence of Erwinia tracheiphila strains on their respective primary host species. Transformed proportion wilted (left panels) and area under disease progress curve (right panels) of wild-type strains, ∆eop1 derivatives, and constructs supporting homologous and heterologous expression of eop1 in an Et-melo strain MDCUKE∆eop1 on primary host muskmelon (A); as well as in Et-C1 strain BHKY∆eop1 on its primary host squash (B). Data symbols and statistics are as described in Figure 6. (Note: This figure will be in the supplemental data in the submitted manuscript).

77 determine whether these virulence traits are influenced by clade-based differences in the Eop1 sequence, we constructed MDCuke and BHKY strains over-expressing homologous and heterologous eop1 variants and evaluated virulence. Homologous and heterologous over-

Figure 8: Expression of eop1 from Et-melo in Et-C1 increased virulence on the secondary host muskmelon. (A) Proportion wilted (left panel) and area under disease progress curve (right panel) of the wild-type Et-C1 strain BHKY, BHKY∆eop1 and strains with homologous and heterologous over-expression of eop1 in BHKY∆eop1 on their secondary host muskmelon. The proportion wilted was transformed with an arcsine (square root) transformation to achieve normality. Data symbols and statistics are as described in Figure 6. (B) In `planta expression of eop1 in BHKY, BHKY∆eop1 and BHKYΔeop1(pN-eop1). Expression of eop1 was measured using quantitative reverse transcription-PCR (RT-qPCR) and relative expression was calculated from the cycle threshold (CT) values, using expression of the recA gene as the internal control. Experiment was repeated twice with 3 technical replicates. Bars represent mean of 2-∆∆Ct values. Different letters indicate significant difference based on a Least Significant Difference test (P = 0.05).

78 expression of eop1 in MDCuke did not alter virulence on muskmelon (Fig 7A) or enable infection of squash (data not shown).

Similarly, homologous and heterologous over-expression of eop1 in BHKY did not alter virulence on its primary host, squash (Fig 7B). However, heterologous over-expression of eop1

MDCuke in BHKY increased virulence on the secondary host muskmelon (Fig 8A).

Overexpression of homologous and heterologous eop1 variants in BHKY in planta was confirmed using RNA isolated from the xylem sap of infected muskmelon plants (Fig 8B). These findings demonstrate that Eop1 functions in host-specific virulence, namely virulence specifically on a muskmelon host. Although BHKY(pN-eop1MDCuke) was more virulent than the wild-type BHKY (Fig 8A), it remained less virulent than the Et-melo strains on muskmelon, as reflected by their AUDPC (Fig 6A and 6B versus Fig 8A), confirming eop1MDCuke as a host- specific virulence factor.

DspE contributes to, but is not required for, virulence in Et-melo strains SCR3 and

MDCuke

To evaluate the role of DspE in E. tracheiphila virulence, we constructed deletion mutants that were missing dspEF in Et-melo strains MDCuke and SCR3 and Et-C1 strains

BuffGH and BHKY. Loss of dspE or dspEF reduced the virulence of both Et-melo strains on muskmelon and cucumber (P<0.05) (Fig 9A - 9D). We did not detect difference in virulence when dspE was deleted alone, or in addition with its chaperone, consistent with the non- virulence function of chaperones. The magnitude of reduction suggests a central role for DspE in the virulence of Et-melo strains, while the remaining virulence demonstrates that DspE is not absolutely required for virulence of this E. tracheiphila clade. Evaluation of the role in

avirulence showed that the loss of dspE also did not enable MDCuke or SCR3 to wilt squash

(data not shown), thus consistent with a role of DspE in virulence but not avirulence. Put

together, DspE is a major factor contributing to, but is not required for, Et-melo virulence.

Figure 9: The effector DspE contributes to the virulence of the Et-melo strains SCR3 (A) and MDCuke (B) on the primary host muskmelon. Proportion wilted (left panel) and area under disease progress curve (right panel) of muskmelon infected with wild-type strains and ∆dspEF mutants. SCR3∆dspEF and MDCukeΔdspEF. Data symbols and statistics are as described in Figure 6.

DspE is required for pathogenicity of Et-C1 strains BHKY and BuffGH on both primary and secondary hosts Loss of dspE or dspEF in BuffGH and BHKY completely eliminated the ability to cause

disease on both primary host squash (Fig 10A and 10B) and secondary hosts muskmelon and

cucumber (Fig 11A and 11B).

Figure 10: The effector dspE is required for the pathogenicity of the Et-C1 strains BuffGH (A) and BHKY (B) on the primary host squash. Transformed proportion wilted (left panel) and area under disease progress curve (right panel) of squash seedlings infected with BuffGH∆dspEF and BHKY∆dspEF. Data symbols and statistics are as described in Figure 6.

These results demonstrate that DspE is required for the pathogenicity of this E. tracheiphila

clade. The collective results across the two clades highlight distinct roles of DspE in these

clades. The presence within a species of strains that exhibit both reduced virulence and non-

pathogenic phenotypes upon loss of an AvrE effector family member has not been previously

Figure 11: The effector DspE is required for the pathogenicity of the Et-C1 strains BuffGH (A) and BHKY (B) on the secondary host muskmelon. Transformed proportion wilted (left panel) and area under disease progress curve (right panel) of squash seedlings infected with BuffGH∆dspEF and BHKY∆dspEF. Data symbols and statistics are as described in Figure 6. reported (Fig 12), although a similar difference between two species within the Pectobacterium

genus, P. carotovorum and P. atrosepticum, was observed. These differences in the impact of

DspE loss may reflect the presence of one or more additional effectors with compensatory functions in the strains exhibiting residual virulence.

The chaperone DspF is required for virulence of Et-melo strains

Whereas DspE was required for the pathogenicity of E. amylovora (57) and Pantoea stewartii ssp. stewartii (19), loss of DspF In these pathogens reduced, but did not eliminate their

virulence, indicating that DspF contributes to, but is not absolutely required, for stability and

secretion of DspE in these

Figure 12: Phylogeny of the AvrE effector family and their contributions to virulence. Phylogenetic analysis of selected members of the AvrE family of effectors. DspE sequences are distinct for the Et-melo and Et-C1 clades. Proteins sequences were aligned using MUSCLE (14), phylogenetic tree was produced using RAxML (54) and trees were viewed in FigTree 1.4.4 (44). DspE sequences of BHKY and SCR3 were obtained from their complete genome sequences which we generated using the PacBio platform. Pasg = Pantoea agglomerans pv. Gypsophilae (40), Pnss = Pantoea stewartii subsp. Stewartii (19), Ea (Ea266) = Erwinia amylovora strain Ea266 (5), Pec = Pectobacterium carotovorum ssp carotovorum (27), Pea = Pectobacterium atrosepticum, Pto (DC3000) = Pseudomonas syringae strain DC3000 (3), RS (UW551) = Ralstonia solanacearum strain UW551 (24), Xv = Xanthomonas vesicatoria (18).

pathogens. The virulence of MDCuke∆dspF and SCR3∆dspF was reduced (P<0.05) (Fig 13A

and 13B) in a similar magnitude as their cognate dspE mutants on muskmelon, suggesting that

DspF is absolutely required for the secretion, and hence virulence function of DspE in the Et-

melo strains.

Discussion

E. tracheiphila strains fall into two clades that differ in their virulence patterns across

Cucurbita species (42; 50; 51). This is the first study to explore the genetic basis of this host- specific virulence in E. tracheiphila, as well as the contribution of individual E. tracheiphila effectors to virulence and pathogenicity. Using deletion mutants of E. tracheiphila, we evaluated

Figure 13: The chaperone DspF is required for the virulence function of DspE in the Et-melo strains MDCuke and SCR3 on the primary host muskmelon seedlings. Transformed proportion wilted (left panel) and area under disease progress curve (right panel) of muskmelon seedlings infected with ∆dspF or ∆dspE in MDCuke (A) and SCR3 (B) backgrounds. Data symbols and statistics are as described in Figure 6.

the contributions of the putative host-specific virulence factor Eop1, and the probable

path ogenicity factor DspE, in the virulence and pathogenicity of two clades of E. tracheiphila strains on multiple Cucurbita host species. Our results demonstrate a small, but detectable, role

84 for Eop1 in enhancing virulence specifically on the Cucumis species muskmelon, and show that this enhanced virulence was associated only with the Eop1 variant produced by strains that rapidly wilt Cucumis species. Thus, the Eop1 effector functions as a host-specific virulence factor in E. tracheiphila. We also demonstrated a major role for the effector DspE in E. tracheiphila virulence. Although this role was predicted based on the contribution of DspE family members to the pathogenicity of many phytopathogenic bacteria, this role surprisingly differed in the two E. tracheiphila clades, with one clade requiring this effector for wilt and the other wilting even in its absence. Collectively, these findings illustrate differences between the two clades in their portfolio of effectors influencing virulence.

Eop1 belongs to the YopJ superfamily of effectors (2; 43). This is a widespread superfamily with many members that have been shown to suppress host immune defenses, a function that depends on a triad of amino acids (histidine, glutamate or aspartate, and cysteine) characteristic of the superfamily and often involves inhibiting host proteasome function (32; 35).

The YopJ superfamily has been subdivided into families, with the E. tracheiphila Eop1 effector classed as a member of the YopJ/AvrRxv family. This family has homologues across several species of Proteobacteria, including Pseudomonas syringae (HopZ2, HopZ4), Xanthomonas campestris (AvrBsT, AvrRxv, AvrXv4 and XopJ), Erwinia amylovora (Eop1), Ralstonia solanacearum (PopP1) and the plant symbiont Rhizobium (Y4LO) (32; 46). Although effectors in this family are widely associated with virulence, few have been examined for a role in host- specific virulence. Of these, E. amylovora Eop1 (2) and R. solanacearum PopP1 (31) were shown to help mediate host-specific restriction of virulence. We found, however, that loss of

Eop1 did not enable Et-melo clade strains to infect squash, a non-host of Cucurbita species, nor did it increase the virulence of Et-C1 clade strains on muskmelon, a Cucumis species that we

designated a secondary host because of the slowness of the wilt induced by these strains. These

results demonstrate that Eop1 is not a host-limiting factor in E. tracheiphila, as it is E. amylovora

(2) and R. solanacearum (31).

Our results, however, demonstrated that E. tracheiphila Eop1 functions in host-specific

virulence. Over-expression of eop1 from the Et-melo strain MDCuke in the Et-C1 strain BHKY

increased the rate at which BHKY wilted muskmelon, a secondary host that normally wilts

slowly, but did not increase the rate at which it wilted squash, a host that normally wilts rapidly.

This is the first evidence demonstrating a genetic basis for host-specific virulence among the E.

tracheiphila clades. Moreover, the eop1MDCuke-mediated increase in Et-C1 BHKY virulence on

muskmelon demonstrates that the MDCuke Eop1 allele has a virulence role on muskmelon.

Surprisingly, overexpressing eop1MDCuke in Et-melo MDCuke did not affect its virulence on

muskmelon, although this could be due to factors other than eop1 expression limiting the rate of

wilt by an Et-melo strain. Although we do not know the mechanism by which eop1MDCuke

increased the virulence of BHKY, many members of the YopJ family of effectors exhibit an

acetyltransferase activity associated with disrupting the stability of their targets, with a

consequent reduction in innate immunity (35). The MDCuke and BHKY Eop1 proteins include

Figure 14: Eop1 of the E. tracheiphila Et-melo MDCuke and Et-C1 BHKY strains have a conserved acetyltransferase domain. Prediction of domains of Eop1 sequences of MDCuke and BHKY as Blasted against PFAM database through the motif search online program (https://www.genome.jp/tools/motif/). (Note: This figure will be in the supplemental data in the submitted manuscript).

86 this acetyltransferase motif (Fig 14), thus providing a motif that could be further examined for its contribution to eop1MDCuke-mediated virulence enhancement.

DspE is a member of the AvrE family of T3SS effectors. This effector family is widely distributed among phytopathogenic bacteria, and its members are often either essential for pathogenesis, as in P. stewartii subsp. stewartii and E. amylovora, or major contributors to virulence, as in pathogens outside of the Enterobacteriaceae family like P. syringae and R. solanacearum (13). Effectors of this family are very large proteins of about 192Da, with surprisingly little known of their biochemical function or host target(s) (13). Like many effectors, most members of this effector family are associated with a co-expressed chaperone, which is required for efficient folding, stability and secretion of the effector (16). Loss of dspEF, that is, dspE and the adjacent chaperone-encoding gene dspF, from two Et-melo strains reduced, but did not eliminate, their ability to wilt muskmelon. The magnitude of the reduction in virulence of these mutants is consistent with a central role for DspE in virulence, as expected; however, the remaining virulence activity demonstrates that DspE is not absolutely required for virulence by this E. tracheiphila clade, unlike in most Enterobacteriaceae family phytopathogens. Similar to a Pseudomonas syringae pv. tomato strain in which AvrE is functionally redundant with HopM1 (3), the Et-melo clade of strains may have one or more other effectors that compensate for the loss of DspE. AvrE homologues were also demonstrated to have reduced virulence, but not non-virulent, phenotypes in Pectobacterium atrosepticum (23),

Ralstonia solanacearum strain UW551 (24) and Xanthomonas vesicatoria (25). The loss of

DspE did not enable two Et-melo strains to wilt squash (data not shown), which is consistent with the lack of evidence for a role for loss of DspE in avirulence; however, the P. syringae pv. tomato AvrE conferred an avirulence phenotype when expressed in P. syringae pv. glycine (29).

In contrast to the Et-melo strains, loss of dspEF in the Et-C1 strains BuffGH and BHKY completely eliminated wilt induction on both the primary host squash and secondary hosts muskmelon. This finding is consistent with the absolute requirement for DspE and its homologues in most enteric phytopathogens. Collectively, these results highlight a surprising difference between the two E. tracheiphila clades, thus identifying them as the only phytopathogenic species that includes both strains that do not require an AvrE family member

(Et-melo clade) and strains that do require an AvrE family member (Et-C1 clade) for disease induction. To date, a similar diversity of DspE functions has been detected only at the species level, with Pectobacterium carotovorum and P. atrosepticum strains showing a similar divergence in their requirement for DspE for pathogenicity.

Many T3SS effectors are co-expressed with a chaperone, and this is true for most members of the AvrE effector family as well (19). While DspE was required for their pathogenicity, loss of DspF exhibited a reduced virulence phenotype in E. amylovora (57) and

Pantoea stewartii spp. stewartii (19), suggesting that DspF contributes to, but is not required for stability and secretion of DspE. The virulence of MDCuke∆dspF and SCR3∆dspF was reduced to a level similar in magnitude as their cognate dspE mutants on muskmelon, suggesting that

DspF is absolutely required for the secretion of DspE.

These findings are consistent with the presence of distinct effector repertoires in the two clades, as highlighted in E. tracheiphila genomic sequence data (51), since distinct repertoires may explain the different roles of DspE in the two clades based on functional compensation among effectors in the Et-melo clade. Bacterial wilt-resistant cucurbit cultivars are not known

(39), thus management of bacterial wilt involves a heavy reliance on the use of insecticides to control the vectors of E. tracheiphila (47). The recent identification of the first plant host target

88 of an AvrE homologue (30), when coupled with our evidence a critical role for an AvrE family member in E. tracheiphila virulence, highlights the possibility of a directed search for resistance in cucurbits using genome sequence information. In this manner, this study may help lead to strategies for more affordable and eco-friendly options for managing bacterial wilt of cucurbits.

References 1. Alfano JR, Collmer A. 1997. The type III (Hrp) secretion pathway of plant pathogenic bacteria: Trafficking harpins, Avr proteins, and death. Journal of Bacteriology 179:5655- 62 2. Asselin JE, Bonasera JM, Kim JF, Oh CS, Beer SV. 2011. Eop1 from a Rubus strain of Erwinia amylovora functions as a host-range limiting factor. Phytopathology 101:935-44 3. Badel JL, Shimizu R, Oh HS, Collmer A. 2006. A Pseudomonas syringae pv. tomato avrE1/hopM1 mutant is severely reduced in growth and lesion formation in tomato. Molecular Plant-Microbe Interactions 19:99-111 4. Bogdanove AJ, Bauer DW, Beer SV. 1998. Erwinia amylovora secretes DspE, a pathogenicity factor and functional AvrE homolog, through the Hrp (type III secretion) pathway. Journal of Bacteriology 180:2244-7 5. Boureau T, ElMaarouf-Bouteau H, Garnier A, Brisset MN, Perino C, et al. 2006. DspA/E, a type III effector essential for Erwinia amylovora pathogenicity and growth in planta, induces cell death in host apple and nonhost tobacco plants. Mol. Plant-Microbe Interact. 19:16-24 6. Buttner D. 2016. Behind the lines: Actions of bacterial type III effector proteins in plant cells. Fems Microbiol. Rev. 40:894-937 7. Cavanagh A, Hazzard R, Adler LS, Boucher J. 2009. Using trap crops for control of Acalymma vittatum (Coleoptera: Chrysomelidae) reduces insecticide use in butternut squash. J. Econ. Entomol. 102:1101-7 8. Chen C, Malek AA, Wargo MJ, Hogan DA, Beattie GA. 2010. The ATP-binding cassette transporter Cbc (choline/betaine/carnitine) recruits multiple substrate-binding proteins with strong specificity for distinct quaternary ammonium compounds. Molecular Microbiology 75:29-45 9. Coburn B, Sekirov I, Finlay BB. 2007. Type III secretion systems and disease. Clin. Microbiol. Rev. 20:535-+

10. Coplin DL, Majerczak DR, Ham JH, Mackey D. 2008. The WtsE virulence effector from Pantoea stewartii, a plant signal mimic? In Proceedings of the Eleventh International Workshop on Fire Blight, ed. KB Johnson, VO Stockwell, 793:203-+. Leuven 1: Int Soc Horticultural Science. Number of 203-+ pp. 11. Datsenko KA, Wanner BL. 2000. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. U. S. A. 97:6640-5 12. de Mackiewicz D, Gildow FE, Blua M, Fleischer SJ, Lukezic FL. 1998. Herbaceous weeds are not ecologically important reservoirs of Erwinia tracheiphila. Plant Dis. 82:521-9 13. Degrave A, Siamer S, Boureau T, Barny MA. 2015. The AvrE superfamily: Ancestral type III effectors involved in suppression of pathogen-associated molecular pattern- triggered immunity. Mol. Plant Pathol. 16:899-905 14. Edgar RC. 2004. MUSCLE: Multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Research 32:1792-7 15. Gaudriault S, Malandrin L, Paulin JP, Barny MA. 1997. DspA, an essential pathogenicity factor of Erwinia amylovora showing homology with AvrE of Pseudomonas syringae, is secreted via the Hrp secretion pathway in a DspB-dependent way. Molecular Microbiology 26:1057-69 16. Gaudriault S, Paulin JP, Barny MA. 2002. The DspB/F protein of Erwinia amylovora is a type III secretion chaperone ensuring efficient intrabacterial production of the Hrp- secreted DspA/E pathogenicity factor. Mol. Plant Pathol. 3:313-20 17. Gazi AD, Sarris PF, Fadouloglou VE, Charova SN, Mathioudakis N, et al. 2012. Phylogenetic analysis of a gene cluster encoding an additional, rhizobial-like type III secretion system that is narrowly distributed among Pseudomonas syringae strains. BMC Microbiol. 12:15 18. Guy E, Genissel A, Hajri A, Chabannes M, David P, et al. 2013. Natural genetic variation of Xanthomonas campestris pv. campestris pathogenicity on Arabidopsis Revealed By Association And Reverse Genetics. mBio 4:13 19. Ham JH, Majerczak DR, Arroyo-Rodriguez AS, Mackey DM, Coplin DL. 2006. WtsE, an AvrE-family effector protein from Pantoea stewartii subsp stewartii, causes disease- associated cell death in corn and requires a chaperone protein for stability. Molecular Plant-Microbe Interactions 19:1092-102 20. Hanna HM, Steward BL, Rosentrater KA. 2018. Evaluating row cover establishment systems for cantaloupe and summer squash. Appl. Eng. Agric. 34:355-64 21. Heeb S, Itoh Y, Nishijyo T, Schnider U, Keel C, et al. 2000. Small, stable shuttle vectors based on the minimal pVS1 replicon for use in Gram-negative, plant-associated bacteria. Mol. Plant-Microbe Interact. 13:232-7 22. Ho SN, Hunt HD, Horton RM, Pullen JK, Pease LR. 1989. Site-directed mutagenesis by overlap extension using the polymerase chain-reaction. Gene 77:51-9

23. Holeva MC, Bell KS, Hyman LJ, Avrova AO, Whisson SC, et al. 2004. Use of a pooled transposon mutation grid to demonstrate roles in disease development for Erwinia carotovora subsp atroseptica putative type III secreted effector (DspE/A) and helper (HrpN) proteins. Molecular Plant-Microbe Interactions 17:943-50 24. Jacobs JM, Milling A, Mitra RM, Hogan CS, Ailloud F, et al. 2013. Ralstonia solanacearum requires PopS, an ancient AvrE-family effector, for virulence and to overcome salicylic acid-mediated defenses during tomato pathogenesis. mBio 4:12 25. Jiang W, Jiang BL, Xu RQ, Huang JD, Wei HY, et al. 2009. Identification of six type iii effector genes with the PIP box in Xanthomonas campestris pv. campestris and five of them contribute individually to full pathogenicity. Molecular Plant-Microbe Interactions 22:1401-11 26. Jin QL, He SY. 2001. Role of the Hrp pilus in type III protein secretion in Pseudomonas syringae. Science 294:2556-8 27. Kim HS, Thammarat P, Lommel SA, Hogan CS, Charkowski AO. 2011. Pectobacterium carotovorum elicits plant cell death with Dspe/f but the P. carotovorum dspe does not suppress callose or induce expression of plant genes early in plant-microbe interactions. Molecular Plant-Microbe Interactions 24:773-86 28. King EO, Ward MK, Raney DE. 1954. 2 simple media for the demonstration of pyocyanin and fluorescin J. Lab. Clin. Med. 44:301-7 29. Kobayashi DY, Tamaki SJ, Keen NT. 1989. Cloned avirulence genes from the tomato pathogen pseudomonas-syringae pv tomato confer cultivar specificity on soybean. Proc. Natl. Acad. Sci. U. S. A. 86:157-61 30. Laflamme B, Dillon MM, Martel A, Almeida RND, Desveaux D, Guttman DS. 2020. The pan-genome effector-triggered immunity landscape of a host-pathogen interaction. Science 367:763-8 31. Lavie M, Shillington E, Eguiluz C, Grimsley N, Boucher C. 2002. PopP1, a new member of the YopJ/AvrRxv family of type III effector proteins, acts as a host-specificity factor and modulates aggressiveness of Ralstonia solanacearum. Mol. Plant-Microbe Interact. 15:1058-68 32. Lewis JD, Lee A, Ma WB, Zhou HB, Guttman DS, Desveaux D. 2011. The YopJ superfamily in plant-associated bacteria. Mol. Plant Pathol. 12:928-37 33. Lewis JD, Wilton M, Mott GA, Lu WW, Hassan JA, et al. 2014. Immunomodulation by the Pseudomonas syringae HopZ Type III effector family in Arabidopsis. PLoS One 9:20 34. Liu Q, Beattie GA, Saalau Rojas E, Gleason ML. 2015. Bacterial wilt symptoms are impacted by host age and involve net downward movement of Erwinia tracheiphila in muskmelon. Eur. J. Plant Pathol. 151:803-10 35. Ma KW, Ma WB. 2016. YopJ family effectors promote bacterial infection through a unique acetyltransferase activity. Microbiol. Mol. Biol. Rev. 80:17

36. Ma WB, Dong FFT, Stavrinides J, Guttman DS. 2006. Type III effector diversification via both pathoadaptation and horizontal transfer in response to a coevolutionary arms race. PLoS Genet. 2:2131-42 37. Markel E, Stodghill P, Bao Z, Myers CR, Swingle B. 2016. AlgU controls expression of virulence genes in Pseudomonas syringae pv. tomato DC3000. Journal of Bacteriology 198:2330-44 38. Mittal R, Peak-Chew SY, Sade RS, Vallis Y, McMahon HT. 2010. The acetyltransferase activity of the bacterial toxin yopj of Yersinia Is activated by eukaryotic host cell inositol hexakisphosphate. J. Biol. Chem. 285:19927-34 39. Mohammadi AS, Maynard ET, Foster RE, Egel DS, McNamara KT. 2018. Muskmelon cultivar attractiveness to striped cucumber beetle and susceptibility to bacterial wilt. Hortscience 53:782-7 40. Mor H, Manulis S, Zuck M, Nizan R, Coplin DL, Barash I. 2001. Genetic organization of the hrp gene cluster and dspAE/BF operon in Erwinia herbicola pv. gypsophilae. Molecular Plant-Microbe Interactions 14:431-6 41. Mukherjee S, Keitany G, Li Y, Wang Y, Ball HL, et al. 2006. Yersinia YopJ acetylates and inhibits kinase activation by blocking phosphorylation. Science 312:1211-4 42. Nazareno ES, Dumenyo CK. 2015. Modified inoculation and disease assessment methods reveal host specificity in Erwinia tracheiphila-Cucurbitaceae interactions. Microb. Pathog. 89:184-7 43. Palmer LE, Pancetti AR, Greenberg S, Bliska JB. 1999. YopJ of Yersinia spp. is sufficient to cause downregulation of multiple mitogen-activated protein kinases in eukaryotic cells. Infect. Immun. 67:708-16 44. Rambaut A. 2008. FigTree. http://tree.bio.ed.ac.uk/software/figtree/ 45. Rand FV, Enlows EMA. 1916. Transmission and control of bacterial wilt of cucurbits. J. Agric. Res. 6:417-34 46. Roden J, Eardley L, Hotson A, Cao YY, Mudgett MB. 2004. Characterization of the Xanthomonas AvrXv4 effector, a SUMO protease translocated into plant cells. Molecular Plant-Microbe Interactions 17:633-43 47. Rufian JS, Lucia A, Rueda-Blanco J, Zumaquero A, Guevara CM, et al. 2018. Suppression of HopZ effector-triggered plant immunity in a natural pathosystem. Front. Plant Sci. 9:16 48. Saalau Rojas E, Batzer JC, Beattie GA, Fleischer SJ, Shapiro LR, et al. 2015. Bacterial wilt of cucurbits: Resurrecting a classic pathosystem. Plant Dis. 99:564-74 49. Saalau Rojas E, Dixon PM, Batzer JC, Gleason ML. 2013. Genetic and virulence variability among Erwinia tracheiphila strains recovered from different Cucurbit hosts. Phytopathology 103:900-5

50. Shapiro LR, Paulson JN, Arnold BJ, Scully ED, Zhaxybayeva O, et al. 2018. An introduced crop plant is driving diversification of the virulent bacterial pathogen Erwinia tracheiphila. mBio 9:20 51. Siamer S, Gaubert S, Boureau T, Brisset MN, Barny MA. 2013. Mutational analysis of a predicted double -propeller domain of the DspA/E effector of Erwinia amylovora. FEMS Microbiol. Lett. 342:54-61 52. Smith EF. 1911. Bacteria in relation to plant diseases. Nature 74:294 53. Smith EF. 1911. Bacteria in relation to plant diseases. Nature 74:294 54. Stamatakis A. 2014. RAxML version 8: A tool for phylogenetic analysis and post- analysis of large phylogenies. Bioinformatics 30:1312-3 55. Szczesny R, Buttner D, Escolar L, Schulze S, Seiferth A, Bonas U. 2010. Suppression of the AvrBs1-specific hypersensitive response by the YopJ effector homolog AvrBsT from Xanthomonas depends on a SNF1-related kinase. New Phytol. 187:1058-74 56. Tasset C, Bernoux M, Jauneau A, Pouzet C, Briere C, et al. 2010. Autoacetylation of the Ralstonia solanacearum effector PopP2 targets a lysine residue essential for RRS1-R- mediated immunity in Arabidopsis. PLoS Pathog. 6:14 57. Triplett L, Melotto M, He SY, Sundin G. 2008. Translocation and chaperone interaction of the Erwinia amylovora secreted effector DspE. In Proceedings of the Eleventh International Workshop on Fire Blight, ed. KB Johnson, VO Stockwell, 793:231-6. Leuven 1: International Society Horticultural Science. Number of 231-6 pp. 58. Ustun S, Konig P, Guttman DS, Bornke F. 2014. HopZ4 from Pseudomonas syringae, a member of the HopZ type iii effector family from the YopJ superfamily, inhibits the proteasome in plants. Molecular Plant-Microbe Interactions 27:611-23 59. Venier M, Hites RA. 2014. DDT and HCH, two discontinued organochlorine insecticides in the Great Lakes region: Isomer trends and sources. Environment International 69:159- 65 60. Weber DC. 2018. Field attraction of striped cucumber beetles to a synthetic vittatalactone mixture. J. Econ. Entomol. 111:2988-91 61. Wolfram J, Stehle S, Bub S, Petschick LL, Schulz R. 2018. Meta-analysis of insecticides in United States surface waters: Status and future implications. Environmental Science & Technology 52:14452-60 62. Wu L, McGrane RS, Beattie GA. 2013. Light regulation of swarming motility in Pseudomonas syringae integrates signaling pathways mediated by a bacteriophytochrome and a LOV protein. mBio 4 63. Yao CB, Zehnder G, Bauske E, Klopper J. 1996. Relationship between cucumber beetle (Coleoptera: Chrysomelidae) density and incidence of bacterial wilt of cucurbits. Journal of Economic Entomology 89:510-4

64. Zhang ZM, Ma KW, Gao LF, Hu ZQ, Schwizer S, et al. 2017. Mechanism of host substrate acetylation by a YopJ family effector. Nat. Plants 3:10

CHAPTER 4 ERWINIA TRACHEIPHILA CLADE DIFFERENCES IN HOST SPECIFICITY ARE

ASSOCIATED WITH HOST-SPECIFIC DIFFERENCES IN HRPA EXPRESSION

Olakunle I. Olawole1, Mark L. Gleason1 and Gwyn A. Beattie1.

1Department of Plant Pathology and Microbiology, Iowa State University, Ames, IA, U.S.A.

Abstract

Erwinia tracheiphila causes bacterial wilt of cucurbits, especially those in the Cucumis and Cucurbita genera. Strains fall into clades that exhibit differences in host-specific virulence based on the ability of strains of the Et-melo clade to rapidly wilt muskmelon but not wilt squash, and strains of the Et-C1 clade to rapidly wilt squash and less rapidly wilt muskmelon. The use of the hrpA gene encoding a pilin protein as a control gene when characterizing the in planta expression profile of E. tracheiphila genes led to the discovery of possible clade- and host- specific differences in hrpA expression. To confirm these differences and investigate if expression of this critical type III secretion system protein could be a factor influencing host- specific virulence, we measured hrpA expression in cells of Et-melo strain SCR3 and Et-C1 strain BHKY recovered from infected muskmelon and squash plants. hrpA expression in SCR3 was 316-fold higher in muskmelon than in squash, with expression at the limit of detection in squash, whereas hrpA expression in BHKY was similarly high in both hosts. Over-expression of hrpA in SCR3 enabled this strain to wilt squash, providing a clear gain-of-function upon expression of this gene. Overexpression of hrpA in SCR3, however, enabled this strain to wilt

95 squash at a much slower rate than BHKY, indicating that limited hrpA expression is only one factor restricting wilt induction on squash by Et-melo strains.

Erwinia tracheiphila is an economically important xylem-inhabiting bacterium that causes bacterial wilt of cucurbits, especially those in the Cucumis (muskmelon and cucumber) and Cucurbita (squash and pumpkin) genera (16). E. tracheiphila is transmitted by cucumber beetles when bacterial-infested frass contact feeding wounds on leaves or flower nectaries, allowing bacterial entry into the xylem (12). Bacterial wilt is restricted to the Northeastern and

Midwestern regions of the United States (13) and crop losses can reach 80% (4).

Strains of E. tracheiphila fall into clades that exhibit differences in host-specific virulence. This is based on the ability of strains of the Et-melo clade to rapidly wilt muskmelon but not wilt squash, and strains of the Et-C1 clade to rapidly wilt squash and less rapidly wilt muskmelon (10; 13; 15). In chapter 3, we demonstrated that a type III secretion system effector gene, eop1, functions as a host-specificity factor; specifically, over-expression of eop1 from an

Et-melo strain in an Et-C1 strain increased the virulence of the Et-C1 strain on its secondary host, muskmelon, but not on its primary host, squash. Thus, the variant of the Eop1 protein in the

Et-melo clade appears to contribute to aggressive virulence, in Cucumis species such as muskmelon, but not in Cucurbita species such as squash. The question remains, however, as to what factors restrain this Et-melo clade from wilting Cucurbita hosts. We discovered a possible clue to the answer when we used the hrpA gene, which encodes a pilin protein for the type III secretion system (T3SS), as a control gene during the characterization of T3SS effector gene expression in planta; specifically, we observed unexpected clade- and host-specific differences

96 in hrpA expression. Here, we investigated the ensuing hypothesis that poor hrpA expression in the Et-melo clade specifically in squash is responsible for its inability to wilt squash.

Pili are surface appendages that have a range of specialized functions. These include attachment to surfaces (14), transfer of bacteria DNA to a host plant by Agrobacterium tumefaciens (8; 9), and transfer of bacterial proteins to a host plant by many phytopathogens.

The latter involves the hypersensitive response and pathogenicity (Hrp) pilus as a conduit to translocate proteins called effectors directly into host cytoplasm (1; 2; 6). Due to their critical role in the secretion and translocation of effector genes, orthologues of the Hrp pilus are known to be required for phytopathogenicity in pathogens expressing the T3SS (7; 11; 17; 18). The E.

Figure 1: Expression of hrpA in Et-melo strain SCR3 and Et-C1 strain BHKY in the xylem sap of muskmelon and squash. Two-week old seedlings of muskmelon (cv. Athena) or summer squash (cv. Yellow crookneck) were inoculated with SCR3 or BHKY. Inoculated plants were incubated until the first appearance of wilt symptoms, after which xylem sap was collected from infected plants and immediately treated with RNAprotect for 10 min at room temperature. RNAprotect-treated cells from xylem sap were centrifuged and RNA was immediately extracted from pellets using a RNeasy Mini Kit, or frozen at -80OC until when needed. Expression of hrpA was measured using quantitative reverse transcription-PCR (RT- qPCR) with the Qscript one-step RT-qPCR kit (Quanta Biosciences, Gaithersburg, MD). Relative expression was calculated from the cycle threshold (Ct) values, using expression of the recA gene as the internal control.

97 tracheiphila genome encodes a Hrp T3SS locus, which includes genes for effectors and structural components such as the pilin, hrpA (15), that is the major subunit of the T3SS pilus.

After infecting muskmelon and squash plants with the Et-melo strain SCR3 and the Et-C1 strain BHKY, we recovered xylem sap from an infected plant representing each strain-host combination, isolated RNA, and used reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) with recA as an internal control gene to characterize the expression of hrpA

(Fig 1), using the primer pairs recA-Fq/Rq and hrpA-Fq/Rq respectively (Table 1).

Table 1. Primers used for this study

Primer Sequences1 recA-Fq CCATATCTACGGGTTCTCTCTCGC recA-Rq AGCGATAACCTGCAAAGTCAGC hrpA-Fq TTAATGACAGGCAGCACTGGC hrpA-Rq TGATCTCTTTGCTCGCAGAGTCC hrpA-SCR3-pSCR3 - EcoRV - F GGAGGTGATATCCTATACGAACATTGTGTGATGTTCTAGGTC hrpA-SCR3-pSCR3 - BamH1 - R ATCATCGGATCCTCAGAACTGAATGGATTTAGAGGCATTG GGAGGTGATATCAGCACCACTTACTATACGAACATTGTGTGA TGTTCCAGGTCACAAGCTAATCAATTAAATTGGAGAGAGTA hrpA-SCR3-pBHKY - EcoRV - F ATATGAGCGGTTTAATGACAGGCAG hrpA-SCR3-pBHKY - BamH1 - R ATCATCGGATCCTCAGAACTGAATGGATTTAGAGGCATTG hrpA-BHKY-pBHKY - EcoRV - F GGAGGTGATATCAGCACCACTTACTATACGAACATTGTG hrpA-BHKY-pBHKY - BamH1 - R ATCATCGGATCCTCAGAACTGAATGGATTTAGAGGCATTG

The xylem sap was collected when the cell densities were estimated to be greater than 105

cells/ml, based on preliminary studies (data not shown). Despite the inability of the Et-melo strain to induce wilt in squash, the cell densities in squash were sufficiently high to extract RNA, as they were similar in their population dynamics to those in muskmelon. The expression of hrpA in SCR3 was 316-fold higher in muskmelon than in squash, where it was expressed just at the limit of detection (Ct of 316.40 ± 215.13 in muskmelon, versus 0.0102 ± 0.006 in squash) (Fig

1), consistent with our hypothesis that a lack of hrpA expression in SCR3 may explain its lack of pathogenicity on squash. In contrast, hrpA was expressed at relatively high levels compared to recA in all of the strain-host combinations that promoted wilt. Whether the higher expression of

98 hrpA in BHKY on its secondary host muskmelon than on squash is biologically significant is not clear.

To further investigate our hypothesis that poor hrpA expression in the Et-melo clade specifically in squash is responsible for its inability to wilt squash, we first constructed hrpA deletion mutants of SCR3 and BHKY using the lambda Red recombinase system (3). As expected, loss of hrpA eliminated pathogenicity of SCR3 on muskmelon (Fig 2) and of BHKY on host squash and muskmelon (Fig 3). Our findings confirm that E. tracheiphila strains require

Figure 3: The Et-C1 strain BHKY requires a functional type III secretion system for pathogenicity on primary host squash and secondary host muskmelon. Values are as described in the legend to Figure 2.

a functional T3SS for pathogenicity, as predicted based on the presence of a many predicted

Figure 2: The Et-melo strain SCR3 requires a functional type III secretion system for pathogenicity on primary host muskmelon. Transformed proportion wilted (left panel) and area under disease progress curve (right panel) of muskmelon seedlings inoculated with SCR3 lacking the type III secretion system structural gene hrpA. Lines represent the standard error of mean of the arcsine-transformed proportion wilted while bars represent the mean area under disease progress curve. Different letters indicate significant difference based on Least Significant Difference (P = 0.05). Each experiment was repeated at least twice with 10 replicates per treatment. effector genes in the species (10; 13; 15), the critical role of the T3SS effector DspE in pathogenicity and virulence (Chapter 3), and the demonstrated role of the T3SS effector Eop1 in host-specific virulence (Chapter 3). To evaluate if over-expression of hrpA in SCR3 enables it to wilt squash, we developed SCR3 constructs that over-expressed hrpA from SCR3 and BHKY, namely hrpASCR3 and hrpABHKY, respectively, under the control of either their native promoters or the constitutive nptII promoter in the broad-host-range vector pME6041 (5). Interestingly, all

SCR3 strains over-expressing hrpA wilted squash (Fig 4), highlighting a clear gain-of-function and illustrating that the inability of Et-melo to wilt squash is attributable, in part, to insufficient expression of hrpA and potentially other T3SS genes.

The hrpA over-expressing strain did not exhibit levels of wilt equivalent to that of

BHKY, consistent with a role for additional factors in modulating virulence. We did not observe a promoter effect in our hrpA-overexpression constructs. To confirm that the constructs actually over-expressed hrpA, we used RT-qPCR analysis to measure hrpA expression on the constructs both in vitro and in planta. The in vitro expression assay, which involved bacterial cells

100

Figure 4: Overexpression of hrpA in the Et-melo strain SCR3 causes it to infect squash. Transformed proportion wilted (upper panel) and area under disease progress curve (lower panel) of squash seedlings inoculated with SCR3 overexpressing hrpA under the control of different promoters, including the BHKY or SCR3 hrpA promoters or the nptII promoter, all within the broad-host-range vector pME6041 (5). Values are as described in the legend to Figure 2. cultivated in King’s B medium, showed relatively low hrpA expression levels for all of the constructs (Fig 5), which is consistent with the lack of expression of the T3SS genes in many plant pathogens in vitro. Results of the in planta expression assay showed that all constructs expressed hrpA (Fig 5), which is consistent with all of the constructs inducing measurable wilt

(Fig 4). Moreover, higher levels of hrpA expression in planta (Fig 5) did not correlate with measurable increases in wilt (Fig 4), which supports a requirement for a threshold level of expression of hrpA for detectable wilt induction.

A functional T3SS is required for the pathogenesis of most bacterial plant pathogens that have a T3SS, and E. tracheiphila is not an exception. While both clades of E. tracheiphila have a functional T3SS, we demonstrated that the inability of the Et-melo clade to induce wilt on

Cucurbita spp. is due, in part, to poor expression of at least one key T3SS gene, hrpA. The

101

Figure 5: In planta over-expression of hrpA in SCR3 backgrounds is associated with its ability to wilt squash. Expression of hrpA in vitro (brown bars) and in planta (orange bars) in an Et-melo strain SCR3 overexpressing hrpA under the control of different promoters, including the BHKY or SCR3 hrpA promoters or the nptII promoter, all within the broad-host-range vector pME6041 (5). The experimental approach was the same as in Fig 1. ability of the Et-melo clade to induce wilt and express hrpA in a Cucumis spp. but not in

Cucurbita pepo or likely other Cucurbita spp. highlights a major role for host factors in regulating virulence gene expression in E. tracheiphila. Knowledge of these host factors, particularly if host factors are involved in repressing the expression of a key pathogenesis gene in E. tracheiphila, could provide opportunities for cucurbit breeding and chemical control to help expand management options for this plant pathogen.

102

References

1. Abby SS, Rocha EPC. 2012. The non-flagellar type III secretion system evolved from the bacterial flagellum and diversified into host-cell adapted systems. PLoS Genet. 8:15 2. Blocker AJ, Deane JE, Veenendaal AKJ, Roversi P, Hodgkinson JL, et al. 2008. What's the point of the type III secretion system needle? Proc. Natl. Acad. Sci. U. S. A. 105:6507-13 3. Datsenko KA, Wanner BL. 2000. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. U. S. A. 97:6640-5 4. de Mackiewicz D, Gildow FE, Blua M, Fleischer SJ, Lukezic FL. 1998. Herbaceous weeds are not ecologically important reservoirs of Erwinia tracheiphila. Plant Dis. 82:521-9 5. Heeb S, Itoh Y, Nishijyo T, Schnider U, Keel C, et al. 2000. Small, stable shuttle vectors based on the minimal pVS1 replicon for use in Gram-negative, plant-associated bacteria. Molecular Plant-Microbe Interactions 13:232-7 6. Jin QL, He SY. 2001. Role of the Hrp pilus in type III protein secretion in Pseudomonas syringae. Science 294:2556-8 7. Jin QL, Hu WQ, Brown I, McGhee G, Hart P, et al. 2001. Visualization of secreted Hrp and Avr proteins along the Hrp pilus during type III secretion in Erwinia amylovora and Pseudomonas syringae. Molecular Microbiology 40:1129-39 8. Kado CI. 2000. The role of the T-pilus in horizontal gene transfer and tumorigenesis. Current Opinion in Microbiology 3:643-8 9. Koebnik R. 2001. The role of bacterial pili in protein and DNA translocation. Trends in Microbiology 9:586-90 10. Nazareno ES, Dumenyo CK. 2015. Modified inoculation and disease assessment methods reveal host specificity in Erwinia tracheiphila-Cucurbitaceae interactions. Microb. Pathog. 89:184-7 11. Perino C, Gaudriault S, Vian B, Barny MA. 1999. Visualization of harpin secretion in planta during infection of apple seedlings by Erwinia amylovora. Cell Microbiol. 1:131- 41 12. Rand FV, Enlows EMA. 1916. Transmission and control of bacterial wilt of cucurbits. J. Agric. Res. 6:417-34 13. Saalau Rojas E, Dixon PM, Batzer JC, Gleason ML. 2013. Genetic and virulence variability among Erwinia tracheiphila strains recovered from different cucurbit Hosts. Phytopathology 103:900-5 14. Salaau Rojas E, Dixon PM, Batzer JC, Gleason ML. 2013. Genetic and virulence variability among Erwinia tracheiphila strains recovered from different cucurbit hosts. Phytopathology 103:900-5

103

15. Saldana Z, Sanchez E, Xicohtencatl-Cortes J, Puente JL, Giron JA. 2011. Surface structures involved in plant stomata and leaf colonization by Shiga-toxigenic Escherichia coli O157:H7. Front. Microbiol. 2 16. Shapiro LR, Paulson JN, Arnold BJ, Scully ED, Zhaxybayeva O, et al. 2018. An introduced crop plant is driving diversification of the virulent bacterial pathogen Erwinia tracheiphila. mBio 9:20 17. Smith EF. 1911. Bacteria in relation to plant diseases. Nature 74:294 18. Van Gijsegem F, Vasse J, Camus JC, Marenda M, Boucher C. 2000. Ralstonia solanacearum produces Hrp-dependent pili that are required for PopA secretion but not for attachment of bacteria to plant cells. Molecular Microbiology 36:249-60 19. Weber E, Ojanen-Reuhs T, Huguet E, Hause G, Romantschuk M, et al. 2005. The type III-dependent Hrp pilus is required for productive interaction of Xanthomonas campestris pv. vesicatoria with pepper host plants. Journal of Bacteriology 187:2458-68

104

CHAPTER 5 IN PLANTA EXPRESSION PROFILING AND FUNCTIONAL ANALYSIS OF

ERWINIA TRACHEIPHILA EFFECTORS

Olakunle I. Olawole1, Mark L. Gleason1 and Gwyn A. Beattie1.

1Department of Plant Pathology and Microbiology, Iowa State University, Ames, IA, U.S.A.

Abstract

Erwinia tracheiphila is a phytopathogenic bacterium that is restricted to the xylem of its cucurbit host, where it causes bacterial wilt, and it is transmitted by the cucumber beetles. We characterized the predicted type III secreted effector (T3SE) repertoire of two strains, Et-melo strain SCR3 and Et-C1 strain BHKY, using whole genome sequence and showed that the strains have distinct, but overlapping, profiles of predicted T3SEs. We characterized the expression profile of selected effectors in cells of SCR3 and BHKY that were recovered from the xylem sap of infected muskmelon and squash plants. We identified host-specific and strain-specific differences in their T3SE expression, although they both expressed one effector gene, eop1, in multiple plant species. Using combinatorial mutant analyses directed by the expressed gene repertoires, we demonstrated that the effectors DspE, AvrB4, OspG and Eop1 all contribute to the virulence of SCR3 on muskmelon. Similarly, we demonstrated that, in addition to DspE, which we previously showed was required for pathogenicity of BHKY in both muskmelon and squash, the effectors OspG, Eop1, HopL1 and HopO1 contribute to BHKY on both hosts, but with a high level of functional redundancy among them. The loss of DspE or AvrB4 alone reduced virulence, with the additional loss of OspG further reducing virulence, suggesting distinct functions for these three effectors. Virulence reductions associated with the loss of Eop1

105 indicated functional redundancy with other effectors. Additional factors contribute to virulence based on the remaining virulence of a quadruple mutant of SCR3 lacking these four effectors.

Put together, these findings identified the contribution of distinct, but overlapping, effectors contributing to the virulence of the two E. tracheiphila clades, with DspE, Eop1, OspG and

AvrB4 detectably contributing to Et-melo virulence on muskmelon, and DspE, Eop1, OspG,

HopL1 and HopO1 detectably contributing to Et-C1 virulence on squash and muskmelon.

Introduction

The genus Erwinia contains over 40 species, which are mostly plant-associated. While some have either epiphytic or saprophytic relationships with their host plants, a few have adopted a pathogenic lifestyle on economic plants (47). These include Erwinia amylovora, E. pyrifoliae, E. psidii, E. papaya, E. aphidicola and E. tracheiphila, which are the causative agents of fire blight of rosaceous plants, bacterial shoot blight of pear, branch, flower and fruit rot of guava, bacterial crown rot of papaya, fruit spot of pepper, and bacterial wilt of cucurbits, respectively (41). Among the phytopathogenic Erwinia species, the molecular mechanism of pathogenesis of E. amylovora on its Rosaceous host plants is the most well-studied (31; 48; 74).

At present, little is known about the molecular mechanisms of pathogenesis of E. tracheiphila.

E. tracheiphila is a Gram-negative, vascular bacterial plant pathogen that is restricted to the host xylem, and is also vector-transmitted (65). This bacterium causes bacterial wilt of cucurbits in the Cucurbita (squash and pumpkin) and Cucumis (muskmelon and cucumber) genera. E. tracheiphila is transmitted by the striped and spotted cucumber beetles (Acalymma vittatum and Diabrotica undecimpunctata, respectively), particularly when bacterial-infested frass comes in contact with fresh feeding wounds (52). E. tracheiphila cells multiply inside the

106 xylem, produce exopolysaccharides and obstruct the vascular system, which leads to wilting, collapse and eventual death of infected plants (55; 73). Bacterial wilt is restricted to the

Northeastern and Midwestern regions of the United States, where it can cause yield losses of up to 80% (17). Management strategies have relied on frequent applications of insecticides to reduce cucumber beetle populations. The danger posed to the health of insect pollinator and insectivorous birds have, however, threatened their continued use [10-13]. Alternative management options include the establishment of mechanical row covers aimed at preventing beetle feeding [15] and perimeter trap crops to lure beetle vectors to an alternate crop [14], among others.

As one of the first bacterial phytopathogens ever described, E. tracheiphila is known to infect cucurbits in a host-specific manner, as strains isolated from Cucumis spp. showed a rapid wilt on Cucumis spp., but slight or no symptoms when inoculated into Cucurbita spp. (squash or pumpkin) (65). Such host-specific virulence characteristics have been shown to correlate with genetic variability among E. tracheiphila strains on the basis of rep-PCR fingerprint profiles (57) and whole-genome phylogeny (60). The strains have been clustered into three phylogenetic clades, Et-melo, Et-C1 and Et-C2, based on similarity across the genomes in draft genome sequences. Despite the publicly available genome sequences, including PacBio-generated sequences of an Et-C1 strain BuffGH (61) and an Et-melo strain MDCuke (59), the molecular mechanisms of pathogenesis among E. tracheiphila strains remain unknown. With no cucurbit cultivars known to be resistant to bacterial wilt (42), a better understanding of the molecular mechanisms of pathogenesis among E. tracheiphila strains could be a step towards identifying strategies to reduce disease in cucurbits (4).

107

Most Gram-negative phytopathogenic bacteria secrete effector proteins into their host cells through the type III secretion system (T3SS), which is a complex membrane-spanning structure that is essential for the virulence of a wide array of animal and plant pathogenic bacteria (1; 8; 12; 30). T3SSs can be grouped into at least three different families based on phylogeny and synteny. These include the Hrp1 (hypersensitive response and pathogenicity)

T3SS, which is found in Pseudomonas syringae and Erwinia spp; Hrp2 T3SS, which is found in

Ralstonia solanacearum, Xanthomonas spp., Acidovorax and Burkholderia; and a rhizobial-like

Hrp3 T3SS, which is found in rhizobia and within some P. syringae strains (1; 22). The majority of studies on the Hrp1 T3SS and its cognate effector proteins have been on P. syringae and E. amylovora.

Xylem pathogens are endogenous, living inside xylem vessels or tracheary elements, as compared to exogenous bacteria which live in the apoplast or on leaf surfaces (7). Several bacterial xylem pathogens have been shown to have a functional Hrp T3SS, including Pantoea stewartii subsp. stewartii (14), Ralstonia solanacearum (2; 37) and Xanthomonas campestris pv. campestris (51). Pantoea stewartii subsp. stewartii has a second T3SS, a Salmonella pathogenicity island-1 (SPI-1), called Pantoea secretion Island 2 (PSI-2), which is required for persistence in the flea beetle vector but not for pathogenesis of maize (14). Interestingly, the E. tracheiphila genome encodes both a SPI-1 T3SS and a Hrp T3SS (60); however, the roles of these T3SSs in mediating the outcomes of either bacterial-beetle or bacterial-cucurbit interactions are unknown.

The analysis of differentially-expressed effectors during bacterial-plant interactions is important to understanding the infection processes (68). To better understand the interactions between the xylem-restricted Erwinia tracheiphila and its cucurbit hosts, we sequenced the

108 whole genomes of Et-melo strain SCR3 and Et-C1 strain BHKY and identified the effector repertoire of each strain; these genomes were also presented and evaluated for various genomic features in Chapter 2. We characterized the expression of selected effectors in planta and evaluated their roles in virulence through single and pyramided mutational analyses. Our findings showed that SCR3 and BHKY have distinct profiles of predicted T3SEs and exhibit host-specific differences in their in planta expression profiles and the contribution of these effectors to virulence.

Materials and methods

DNA extraction, library preparation and genome sequencing and assembly

Single colonies of the E. tracheiphila strains SCR3 and BHKY were grown in King’s B

(KB) broth to late-log phase, after which the cultures were adjusted to an OD600 of 1.0 using KB broth. Genomic DNA was purified using the DNeasy Blood and Tissue kit (Qiagen, Venlo,

Limburg) according to the manufacturer’s instructions, and included steps for Proteinase K and

RNase treatments, as recommended for PacBio genomic DNA extractions (40). Library preparation and sequencing were done at the Iowa State University DNA facility using the

SMRTbell Template Prep Kit (Pacific Biosciences, Menlo Park, CA) according to the PacBio standard protocol. Libraries of 20-kb size-selected fragments were generated using the

BluePippin Size-selection System (Sae Science Inc, Beverly, MA), which included steps for

DNA damage and end repair and ligation to hairpin adapters. Libraries were subsequently sequenced on the PacBio RSII instrument and P6-C4 chemistry with one SMRTcell per strain.

The HGAPv3 protocol of the Pacific Biosciences SMRTportal software was used to assemble

109 each genome. Raw SMRT sequencing reads were trimmed, corrected, and assembled de novo using default parameters with a genome size estimate of 5 Mb (62).

Genome annotation and identification of type III secretion system effectors (T3SEs)

Assembled genomes were annotated using the RAST pipeline (3) through the

Pathosystems Resource Integration Center (PATRIC vs. 3.5.26) server (69), using default parameters. Annotations were manually curated using Artemis 17.0.1 (54). Predicted coding sequences from each strain of E. tracheiphila were blasted (E-value cutoff of 10-5) against a local database generated from the Pseudomonas Hop protein effector database

(http://www.pseudomonas-syringae.org/T3SS-Hops.xls), which was accessed in April 2018, and expanded to include T3SSEs from many other bacterial plant pathogens.

Plant growth conditions

The cucurbit cultivars used in this study include muskmelon (Cucumis melo cv. Athena), summer squash (Cucurbita pepo cv. Early Summer Crookneck), zucchini squash (Cucurbita pepo cv. Partenon F1, organically grown) and cucumber (Cucumis sativus cv. Marketmore, organically grown). Seeds were sown in 10-cm pots containing a matrix of peat moss, coarse perlite and Metro-Mix 300 (Sun Gro Horticulture, Canada Ltd.; Vancouver, BC, Canada).

Seedlings were maintained in a growth chamber (Percival Scientific Inc.) at 28°C under a photoperiod of 12 h of light and 12 h of darkness and 70% relative humidity. Plants were watered every other day and fertilizer solution (NPK: 15-5-15: Peters Excel, ICL UK/Ire) was added two days before inoculation.

110

Bacterial strains, growth conditions and inoculum preparation

The bacterial strains and plasmids used for this study are described in Table 1. E. tracheiphila strains were grown in KB medium (33) at 28◦C, unless otherwise described.

Escherichia coli strains were grown in Luria medium at 37◦C. The following antibiotics were added at 50µg/ml: rifampin (Rif), kanamycin (Km), spectinomycin (), and ampicillin (Amp), when needed. To prepare cells for plant inoculation assays, E. tracheiphila strains were recovered from cryogenically preserved glycerol stocks (-80◦C) on KB agar amended with Rif.

The cells were incubated for 3-4 days, after which a single colony was transferred to a fresh plate to make a lawn. Cells from a two-day old lawn were suspended in 10 mM phosphate buffer (PB)

8 and normalized to an optical density at 600nm (OD600) of 0.5 (approximately 2.5 x 10 cells/ml).

A non-ionic organosilicone surfactant (Silwet L77) was added to each bacterial suspension at

0.02% (vol/vol).

Inoculation and disease assessment

Two-week old seedlings of each cucurbit cultivar were inoculated with E. tracheiphila.

The site closest to the petiole of the adaxial surface of the youngest fully expanded leaf was punctured with a 28.6-mm-diameter florist’s pin frog (Kenzan Pin Frog, sold by www.save-on- crafts.com), and 200 μl of cell suspension, prepared as described above, was applied to the puncture site. Plants inoculated with PB in the same manner as described above were used as controls. Inoculated plants were incubated as described for the seedlings, above, and rated daily for wilt development based on the % of wilted leaves, which was calculated as the ratio of wilted leaves divided by the total number of leaves. Once plants reached 100% wilt, newly emerged leaves were not counted and the plants were not monitored further. Data were collected daily for two weeks.

111

Table 1: Bacterial strains and plasmids used in this study Antibiotic Source or Strains and plasmids Description or host/location of resistance reference isolation

Erwinia tracheiphila SCR3 Et-melo strain isolated from Cucumis Rif (56) melo in Iowa SCR3∆eop1 eop1 deletion mutant in SCR3 Rif This study SCR3∆avrB4 avrB4 deletion mutant in SCR3 Rif This study SCR3∆ospG ospG deletion mutant in SCR3 Rif This study SCR3∆dspEF dspEF deletion mutant in SCR3 Rif This study SCR3∆eop1∆ospG eop1 and ospG deletion mutants in Rif This study SCR3 SCR3∆avrB4∆dspEF∆eop1 avrB4, dspEF and eop1 deletion Rif This study mutants in SCR3 SCR3∆avrB4∆dspEF∆ospG avrB4, dspEF and ospG deletion Rif This study mutants in SCR3 SCR3∆avrB4∆dspEF∆eop1∆ospG avrB4, dspEF, eop1 and ospG Rif This study deletion mutants in SCR3 BHKY Et-c1 strain isolated from Cucurbita Rif (56) moschata in Kentucky BHKY∆eop1 eop1 deletion mutant in BHKY Rif This study BHKY∆avrB4 avrB4 deletion mutant in BHKY Rif This study BHKYΔospG ospG deletion mutant in BHKY Rif This study BHKY∆hopL1 hopL1 deletion mutant in BHKY Rif This study BHKY∆hopO1 hopO1 deletion mutant in BHKY Rif This study BHKY∆eop1∆hopL1 eop1 and hopL1 deletion mutants in Rif This study BHKY BHKY∆eop1∆hopO1 eop1 and hopO1 deletion mutants in Rif This study BHKY BHKY∆avrB4∆eop1 avrB4 and eop1 deletion mutants in Rif This study BHKY BHKY∆avrB4∆eop1ΔospG avrB4, eop1 and ospG deletion Rif This study mutants in BHKY BHKY∆avrB4∆eop1∆hopL1ΔospG avrB4, eop1, hopL1 and ospG Rif This study deletion mutants in BHKY BHKY∆eop1∆hopL1ΔospG eop1, hopL1 and ospG deletion Rif This study mutants in BHKY BHKY∆avrB4∆eop1∆hopL1∆hopO1ΔospG avrB4, eop1, hopL1, hopO1 and ospG Rif This study Construction of deletion mutants and complementationdeletion mutants constructsin BHKY

PlasmidsThe primers used in this study are shown in Table 2. For each gene targeted for deletion, pN pME6041 with nptII promoter Km (11) primerspKD13 were targeted to regions that were conservedTemplate foracross the amplificationthe Et-melo ofand the Et -C1Amp, clades. Cm (15) Km cassette bordered by FRT sites pKD46 λ Red recombinase plasmid promotes Amp (15) recombination of linear PCR products pFlp2Ω Flippase helper plasmid Amp, Spc (71) pTOK2T pTOK2 with restored lacZ activity Tet (11)

112

Deletion mutants were generated by using the lambda Red recombinase system, which is comprised of recombinase components expressed under the control of an arabinose-inducible promoter on the plasmid pKD46 [11]. Linear PCR products were generated and introduced via electroporation into E. tracheiphila cells that had been transformed with pKD46. These linear

To generate mutants, E. tracheiphila cells containing pKD46 were grown at 28°C for 2-3 days in KB broth amended with Amp. When cells reached an OD600 of 0.3, freshly-made L- arabinose was added to a final concentration of 10mM to induce expression of the lambda phage Red recombinase genes. Cells were harvested at an OD600 of 0.5 and washed three times with room temperature sterile distilled water. The room temperature wash was adopted because

E. tracheiphila competency appeared to be reduced by washing in cold sterile water. After the washes, the cells were suspended in sterile distilled water and placed on ice for 15 to 20 minutes.

Electroporated cells were transferred to 1ml of ice-cold KB broth and incubated for 6 to

10 hours, after which cells were centrifuged and pellets were plated on Km-supplemented solid

KB agar to select for successful mutants. After confirmation of the mutants by PCR, the kan

113

Table 2. Primers used for this study

Primer Sequencesa Primers for constructing deletion mutants CCCACGTTGATGGCGCAGGGGTTAACACTCAAAGCAGCGGGGAGAATTCGACGTC Eop1-Km-F TTGAGCGATTGTGTAGGCT CAGGAAAAAAAGCACGGGTATTAACGGTGCGTTTTATTAAAGGGGTAAAAAGTGA Eop1-Km-R TTGCGCCTACCCGGATATT F1-dspE AAAATG TGCGGATCCCGTCAGAG R1-dspE AGCCTACACAATCGCTCAAGACGTGACCGCCCCCGTTCCCACATTAA F2-dspE AATATCCGGGTAGGCGCAATCACTTTAAGCCTCATCCCTGATATCGC R2-dspE CTTCCCTTAAAATCGCAGGCTAT F3-Km ACGTCTTGAGCGATTGTGTAGGCT R3-Km AGTGATTGCGCCTACCCGGATATT AAAGCGTGAGTCAGAGTATTACAGCAATAGCTGTAATCTACCGTCAACACTATGGC AGCCTTGCGAACGGGATGTGTTTATATATGAATTACGTCTTGAGCGATTGTGTAGG AvrB4-Km-F CT AAGCTGAATATGGCGTTCTGAATGGTCTTCCACCATGATCGGGGAAGACCCATAAA TAACCGGTAACTGTCAGATCAGGTCTGAGCTGGTAGTGATTGCGCCTACCCGGATA AvrB4-Km-R TT GGAACATCTGCGACTTAAACTCAATACTTTAGCCAGCAGCGGATCGGGACTTACCC ATTACTGGATCGACAGTGGGAGCGTATACAAGAAACGTCTTGAGCGATTGTGTAGG HopL1-Km-F CT CTAAATCAAAACACGCAGATATCATCATCGGCAAAACCGGAGTGTTCAGCAATCA ACGCTGTAACCCGTAAGCGGATTACGGCGCAAATTAGTGATTGCGCCTACCCGGAT HopL1-Km-R ATT CACATTTAGTGATTAAGTGAAATTTATCTTTTGACACGTAAATCATGTGAAAATCTC CTGGATGTTTAAATTGCGCAACCAAGGTCTGTAACGTCTTGAGCGATTGTGTAGGC OspG-Km-F T ACCTCCAGATTTTTGCAACAGTGCCGGTTTTTTTACCATGCGAATGATTCTTCTCTG OspG-Km-R ATCATGGTGACCACTTCATCCTCTACCCACTACAGTGATTGCGCCTACCCGGATATT CAGTGTCGAAATGCGAAAAATCTTTGAGAATTTGCTAAGTGTGGCAGTCGAAATTA AAAAGTCATTCAATTTCAAGTGATAGGTGTATTTACGTCTTGAGCGATTGTGTAGG HopO1-Km-F CT ACCTACAATACTGGAGTTCGCAGGCGTACGGCCTCCAACGCAGAAACTGGACCGTC GCTATATTTTGCGCACGCGCTGCAGTTAAGGCCCAGTGATTGCGCCTACCCGGATA HopO1-Km-R TT

Primers for confirming mutants AvrB4-ExF AGTGTTTCAGGGTGGCCAGC AvrB4-ExR GAGGATCTGCCGTGTGAGCTT HopL1-ExF AGCAGCGGATCGGGACTTAC HopL1-ExR CGCTGTGACGAGCAATCTGTG OspG-ExF AACATATAATTCACATTCTCGCTTGGTGG OspG-ExR CCATGCGAATGATTCTTCTCTGATCATG Eop1/dspE-F TGCGTGCAGATGATCGAAGTGA Eop1/dspE-R CCGTTTCGTCAAGCGCGATTC

114

Table 2. Continued

Primer Sequencesa Primers for in planta gene expression Eop1-Fq GGGGTATGATGAATGAGATTCGGC Eop1-Rq ATTTCAGCGCATTACTCAGCGC RecA-Fq CCATATCTACGGGTTCTCTCTCGC RecA-Rq AGCGATAACCTGCAAAGTCAGC AvrB4-Fq CTCATACAGGCCAGACCAGTCC AvrB4-Rq CGCTACCCCGATAAGAGATTGC OspG-Fq GCTGGTAACCATTACCATGAAGTACTGG OspG-Rq TCACTAAGAGGAGTTCCAGGCAC HopL1-Fq GAGAAGAGCTGTTCGAGCAGGT HopL1-Rq ACTGATATCGTAGCTGTCAGCAGC HopA1-Fq AACTCCTCGATCACGGCTCG HopA1-Rq CGGCAGAGCCCATCATATTTTCG HopE1-Fq GATTTCACGATCTACAGCAATAAGGCG HopE1-Rq CTTTAGCGATAATTCAAACACACCGTGTC Eop3-Fq TATCCATGCCACGTCCGGAG Eop3-Rq CCGCATGAACCGAACAGTTACC HopAF1-Fq CAAACTTAAGGGCAGCGTCCAG HopAF1-Rq CATCTAAGGAAACCATTCGAAGGGTATAG HopG1-Fq GTTTGACTCCACGGCAATGCG HopG1-Rq CCAATACCGATCGTCGGCATATAC HopO1-Fq TTATGTGTATAGTCCGCCAGTTAGCC HopO1-Rq TATCGTGATATACTTCCGCTGCAGC HopV1-Fq GCATTTGCAGGCAGGTAACCAATC HopV1-Rq TCAGTTCCACCAGTGACTTGTTGG DspE-Fq AAGACATCCATGTTGACCAGCAGC DspE-Rq TGTCCATTTCCAGACGTTGCAGATC cassette was removed by introduction of pFlp2Ω (71), which encodes a recombinase, the Flp recombinase, that recognizes the FRT sites for recombination.

A linear fragment for constructing the ΔdspEF mutants was generated by using splice- overlap-extension PCR (29). In short, 1-kb regions were amplified from E. tracheiphila strain

(11) such that the SmaI site was preserved, and this plasmid was used as a template in a PCR reaction with the F1-dspE and R2-dspE primer pairs (Table 2). The resulting linear fragment was

115 electroporated into cells of E. tracheiphila strains SCR3(pKD46) and BHKY(pKD46). The linear fragments used to generate the Δeop1, ΔdspE and ΔdspF mutants contained the FRT-kan-

Evaluation of in vitro and in planta gene expression

To evaluate in planta expression of E. tracheiphila effector genes, two-week old seedlings of muskmelon or squash were inoculated with BHKY or SCR3 as described above.

Inoculated plants were incubated until the first appearance of wilt symptoms (approximately 4 days for muskmelon, and 6 days for squash), at which time xylem sap was collected from infected plants. The xylem sap was collected by manually squeezing the aerial half of the cut stem and collecting the sap with a 100 µl pipette. Immediately following xylem sap collection, the sap was combined with RNAProtect at room temperature to preserve RNA integrity. After a

10-minutes incubation, the cells in this mixture were collected by centrifugation and pellets were either used immediately for RNA extraction using the RNeasy Mini Kit (Qiagen), or were frozen at -80◦C for RNA extraction at a later time.

The transcripts of eop1, avrB4, ospG, hopL1, hopA1, hopE1, eop3, hopAF1, hopG1, hopO1, hopV1 and dspE were measured using quantitative reverse transcription-PCR (RT- qPCR) with the qScript One-Step SYBR Green RT-qPCR kit (Quantabio, Gaithersburg, MD).

116

Relative expression was calculated from the cycle threshold (Ct) values, using expression of the recA gene as the internal control. The primers used to evaluate expression of selected effectors are presented in Table 2. One hundred nanograms of total RNA was used for each of three technical replicates, each in a total reaction of 25 µl. Real-time PCR was performed using the

Mastercycler® realplex (Eppendorf), under the following conditions: one cycle at 95◦C for 2.5

◦ ◦ min, 40 cycles of 95 C at 15 s and one cycle as 60 C for 30 s. The resulting Ct values were calculated using the relative standard curve method (39) in which the Ct values of each tested gene was normalized to the Ct values of recA.

Statistical analyses

For the virulence assay, the total number and wilted number of leaves were used to calculate the proportion of wilted leaves. This proportion was further subjected to an arcsine- square root transformation before analysis, as is common for data expressed as proportions, and these values were plotted over time to generate disease progress curves. Virulence was quantified based on the area under disease progress curve (AUDPC) values. Graphs were generated using

SigmaPlot 14 and statistical differences among treatments were evaluated using JMP Pro 12

(SAS Institute Inc., Cary, NC). For both the gene expression and AUDPC values, the strains were compared based on an analysis of variance (ANOVA), with mean values compared using a

Least Significant Difference (LSD) test at a 5% level of significance for studies with more than two strains, and with a Student’s t-test for studies with only two strains. Values were plotted as the mean and standard error of the mean (SEM) of data from individual experiments.

117

Results

SCR3 and BHKY have distinct profiles of predicted T3SS effectors based on genome sequence

To identify effectors in the two clades, we generated long-read, complete genome sequences for

Et-melo strain SCR3 and Et-C1 strain BHKY. These data double the available long-read, complete genome sequences publicly available for E. tracheiphila (Table 3).

Table 3: An overview of the assembly metrics of Et-melo strain SCR3 and Et-C1 strain BHKY of Erwinia tracheiphila, in comparison with publicly available genome sequences

subsp. melon strains1 subsp. squash strains1 Features MDCuke SCR3 BuffGH BHKY Cucurbita Plant species of origin Cucumis sativus Cucumis melo Cucurbita pepo moschata State of origin Maryland Iowa Pennsylvania Kentucky N50 contig size (bp) 4,891,633 4,672,768 4,281,223 4,924,880

De novo assembly contigs 5 4 7 3

Total size of all contigs (bp) 5,130,743 4,911,102 5,015,962 4,989,113

Coverage (×) 100 100 94 96 tRNAs 64 63 68 65 rRNA operons 6 6 6 6 GenBank Accession numbers CP013970 This work NZ_JXNU01000003 This work 1 Sequencing for all strains was done using the PacBio platform with RSII chemistry

The genome sizes of SCR3 and BHKY are very similar to the published sequences of MDCuke and BuffGH, which are Et-melo and Et-C1 strains respectively (59; 61). SCR3 and BHKY assembled into 4 and 3 contigs respectively; due to the presence of plasmids are likely closed genomes (66). To identify the distribution of the predicted Hrp type III secretion system effector repertoire of these four E. tracheiphila strains, we mined their genomes for putative effector genes using BLASTP and a manually-assembled local effector database. Our results showed that

118 the two E. tracheiphila, as represented by two strains each, clades share a majority of their effector proteins, but each clade also has a distinct subset (Fig 1).

Figure1: Erwinia tracheiphila Et-melo and Et-C1 strains share a majority of their putative hrp type III secretion system effectors (T3SEs). Using a local BLAST database of known effectors, and based on top- scoring BLASTP matches, we identified the distribution of the predicted Hrp T3SE repertoire in SCR3 and BHKY, and used publicly available genomes of the Et-melo strain MDCuke (59) and the Et-C1 strain BuffGH (61) as references.

Our BLASTP results identified other genes that encode effector homologues, as found in a previous work (60), but we omitted them due to little or no evidence for roles as effectors.

These effector homologues include SrfA and SrfB, which are regulators that are encoded as part of the Rcs regulon in Salmonella enterica (21), and Eop2 (also known as HopAK1) and HopAJ1, which are helper proteins in E. amylovora (10; 45) and Pseudomonas syringae (46), respectively.

We also found that HopR1, which belongs to the same redundant-effector group (REG) as AvrE

(35), is truncated with premature stop codons in strains of both clades. Similarly, HopL1 is truncated in the Et-melo strains, but intact in the Et-C1 strains. Put together, genomes of E. tracheiphila Et-melo and Et-C1 strains share 10 T3Es, including AvrB4, DspE, Eop1, HopA1,

HopE1, HopF2, HopG1, HopR1, HopV1 and OspG. In addition, each clade possesses three to

119 four clade-specific effectors, namely Eop3, HopAF1 and Skwp2 for Et-melo; and HopO1,

HopAM1, NleD and HopL1 for Et-C1 (Fig 1).

SCR3 and BHKY each exhibit host-specific differences in the T3SS effector genes expressed in planta Out of the 10 shared effectors among the E. tracheiphila strains, we selected AvrB4,

DspE, Eop1, HopA1, HopE1, HopG1, HopV1 and OspG to examine for expression in planta, as they each have single copies within their genomes. HopF2 and HopR1 were omitted because their sequences had premature stop codons. Among the Et-melo-specific effectors, we only selected Eop3 and HopAF1, as Skwp2 has two copies. Similarly, among the Et-C1-specific effectors, we selected HopL1 and HopO1, as HopAM1 is missing the first 133 N-terminal sequences while NleD has 7 copies.

Overall, we evaluated the in planta expression profiles of 8 shared, 2 Et-melo-specific and 2 Et-C1-specific effectors following recovery of SCR3 and BHKY from infected squash and muskmelon. The xylem sap was collected when the cell densities were estimated to be greater than 105 cells/ml, based on preliminary studies (data not shown). Despite the inability of the Et- melo strain to induce wilt in squash, the cell densities in squash were sufficiently high to extract

RNA, as they were similar in their population dynamics to those in muskmelon. The gene eop1 was the most highly expressed gene in SCR3 in muskmelon, but also in its non-host squash (Fig

2). Interestingly, eop1 was the second most highly expressed gene in BHKY in both hosts (Fig

3); this gene had been examined previously as a host-specific virulence candidate (Chapter 3).

The hopO1 gene, which is present in BHKY but not SCR3, was expressed at an exceptionally high level in both its primary and secondary hosts (Fig 3). Although DspE is a major factor contributing to virulence (Chapter 3), it was expressed at an extremely low level in both strains

120 in both hosts (Fig 2 and 3). This is consistent with previous findings with DspE homologues in other phytopathogens, including E. amylovora (6; 63) and Pantoea stewartii ssp. stewartii (28).

Figure 2: Expression of selected effector genes of Et-melo strain SCR3 on primary host muskmelon and non-host squash. Expression was measured using quantitative reverse transcription-PCR (RT-qPCR). Relative expression was calculated using the ∆∆Ct approach with recA gene as an internal control. Different letters indicate significant difference based on Least Significant Difference (P = 0.05). Different letters within a panel indicate significant differences.

Figure 3: Expression profile of selected effector genes of Et-C1 strain BHKY on primary host squash and secondary host muskmelon. Experimental approach and analysis were the same as in Figure 2. Different letters indicate significant difference based on Least Significant Difference (P = 0.05). Different letters within a panel indicate significant differences.

121

The effectors DspE, Eop1, OspG and AvrB4 and are the primary effectors contributing to the virulence of Et-melo strain SCR3 on its host, muskmelon

The role of Eop1 in virulence was examined in Chapter 3, during which its loss did not alter the virulence of the Et-melo (MDCuke and SCR3) and Et-C1 (BHKY) strains on either muskmelon, cucumber or squash. In this study, we found that SCR3 showed measurable expression of only three effector genes in its host, muskmelon; it showed measurable expression of two of these three effector genes in its non-host, squash (Fig 2). Other than eop1, the SCR3 effector gene that was expressed the most in the primary host was avrB4. AvrB4 contributed to virulence, consistent with it in planta expression pattern, and this was detectable when it was the

Figure 4: The effector AvrB4 contributes to the virulence of the Et-melo strain SCR3 on its primary host. Transformed proportion wilted (left panel) and area under disease progress curve (right panel) of muskmelon seedlings inoculated with an SCR3 deletion mutant of avrB4. The proportion wilted was transformed with an arcsine (square root) transformation to achieve normality. Lines represent the standard error of mean of the transformed proportion wilted while bars represent the mean area under disease progress curve. Different letters indicate significant difference based on Least Significant Difference (P = 0.05). Each experiment was repeated at least twice with 9 replicates. Different letters within a panel indicate significant differences. only effector that was deleted (Fig 4 and Fig 6). The third most expressed gene was ospG, and we looked at the effect of its loss on virulence. Loss of ospG, alone, did not affect virulence (Fig

5 and 6), but loss of this effector in combination with loss of either eop1 (Fig 5) or dspEF (Fig 6)

122 resulted in reduced virulence. The ∆dspEF mutant lacked both the DspE effector and its cognate chaperone, DspF.

Other than eop1, the SCR3 effector gene that was expressed the most in the non-host was

Figure 5: The effector OspG is functionally redundant with Eop1 in the Et-melo strain SCR3 on its primary host. Transformed proportion wilted (left panel) and area under disease progress curve (right panel) of muskmelon seedlings inoculated with deletion mutants of eop1 or ospG in SCR3 background. Data symbols and statistics are as described in Figure 4. ospG . Although it is possible that this effector contributed to the inability of SCR3 to infect squash, that is, it could function as an avirulence factor, this was not evaluated here due to the finding that the expression of the T3SS, itself, was low in squash (Chapter 4).

We evaluated the quantitative contributions of the four SCR3 effectors Eop1, AvrB4,

OspG and DspE to virulence on muskmelon using pyramided mutations. Again, the only major effectors that had measurable individual contributions to virulence were AvrB4 and DspEF, with loss of the effector gene dspE and its chaperone, dspF, having the largest impact on virulence among the single-effector deletion mutants effectors examined. Based on the significant levels of the differences among the pyramided mutants, loss of ospG further reduced the virulence of a dspEF mutant (Fig 6), indicating additive effects of these two effectors. Although not significant

123 at P = 0.05, the loss of avrB4 may have further reduced the virulence of strains lacking dspE and ospG (Fig 6), suggesting additive effects of DspE, OspG and AvrB4.

Figure 6: The effectors of the Et-melo strain SCR3 have additive effects on its virulence on muskmelon. Transformed proportion wilted (left panel) and area under disease progress curve (right panel) of muskmelon seedlings inoculated with combinatorial deletion mutants of eop1 or ospG or avrB4 or dspE in SCR3 background. Data symbols and statistics are as described in Figure 4.

Eop1 showed functional redundancy with OspG when these were the only two effectors deleted (Fig 5) or these were deleted along with dspEF (Fig 6), but ospG had a bigger effect than eop1 when these were deleted along with dspEF and avrB4 (Fig 6). Eop1 also showed functional redundancy with AvrB4 based on the similar effect on virulence of the loss of eop1, avrB4 and both genes in an SCR3∆dspEF background (Fig 6). These results indicate that (i) DspE and

AvrB4, by themselves, measurably contribute to SCR3 virulence, (ii) the relative contribution to virulence of DspE, AvrB4 and OspG is DspE > AvrB4 > OspG based on the virulence of the single and double mutants, and (iii) Eop1 is functionally redundant with OspG and AvrB4, with

OspG contributing more to virulence than Eop1 in the absence of DspEF and AvrB4. The low, but detectable, level of virulence exhibited by a SCR3ΔdspEΔavrB4ΔospGΔeop1 mutant

124

indicates that there are yet additional factors, potentially effectors, that contribute to the

virulence of this strain.

Loss of the individual effectors HopO1, Eop1, HopL1 and OspG did not alter the virulence

of Et-C1 strain BHKY on squash, consistent with likely functional redundancy among

effectors, as observed in other phytopathogens

BHKY expressed only three effector genes in each its primary and secondary host, with

two of the three genes (hopO1 and eop1) the same in both hosts (Fig 3). The effector gene that

was expressed the most in BHKY in both plants was hopO1. Despite this high expression level,

Figure 7: The effector HopO1 did not alter the virulence of Et-C1 strain BHKY on both its primary host squash and secondary host muskmelon. Transformed proportion wilted (left panel) and area under disease progress curve (right panel) of squash (A) or muskmelon (B) seedlings inoculated with deletion mutants of hopO1 in BHKY background. Data symbols and statistics are as described in Figure 4. (Figure will be presented as a Supplementary data in manuscript).

125 loss of hopO1 by itself did not noticeably change the virulence of BHKY on squash and muskmelon (Fig 7).

Similarly, loss of the effector gene that was expressed at the second highest level in

BHKY, eop1, by itself did not noticeably change the virulence of BHKY on either host (Fig 8).

Figure 8: The effectors HopO1 and HopL1 are functionally redundant with Eop1 in the Et-C1 strain BHKY on its secondary host muskmelon. Transformed proportion wilted (left panel) and area under disease progress curve (right panel) of squash or muskmelon seedlings inoculated with combinatorial deletion mutants of eop1 and hopO1 (A) or eop1 and hopL1 in BHKY background. Data symbols and statistics are as described in Figure 4. Each experiment was repeated at least twice with 9 replicates per treatment.

The loss of both hopO1 and eop1 in BHKY detectably reduced virulence on both muskmelon

(Fig 8A) and squash (Fig 9A), indicating that these two effectors overlap in their functions, although the magnitude of the decreased virulence was surprisingly small given the high level of expression of these genes.

126

The effector gene hopL1 was expressed in BHKY only in muskmelon, whereas expression of ospG was specific to the primary host, squash (Fig 3). We examined the impact of

Figure 9: The effectors HopO1, HopL1 and OspG are functionally redundant with Eop1 in the Et-C1 strain BHKY on squash. Transformed proportion wilted (left panels) and area under disease progress curve (right panels) of squash seedlings inoculated with combinatorial deletion mutants of eop1 or hopO1 (A), or hopL1 (B), or ospG (C) in BHKY background. Data symbols and statistics are as described in Figure 4. the loss of these two genes alone and each in combination with eop1 on virulence. Similar to hopO1, loss of hopL1 alone did not alter virulence of BHKY on either host, but the loss of both eop1 and hopL1 reduced virulence on muskmelon (Fig 8A); interestingly, their loss also reduced

127 virulence on squash (Fig 9B), despite the lack of measurable hopL1 expression in squash (Fig 4).

Loss of ospG alone did not alter virulence of BHKY on either its primary or secondary host, but

Figure 10: The effector ospG did not alter the virulence of Et-C1 strain BHKY on its secondary host muskmelon. Transformed proportion wilted (left panel) and area under disease progress curve (right panel) of squash or muskmelon seedlings inoculated with deletion mutants of ospG in BHKY background. Data symbols and statistics are as described in Figure 4. (Figure will be presented as a Supplementary data in manuscript). its loss in combination with eop1 in reduced virulence on squash (Fig 9C) but not on muskmelon

(Fig 10), consistent with its expression profiles in these hosts (Fig 4).

Finally, we investigated the impact of loss of all the highly expressed effectors (HopO1,

Eop1, OspG, and HopL1) on virulence of BHKY in both squash and muskmelon. Apart from the critical role of DspE to its pathogenicity on squash and muskmelon (Chapter 3), we did not detect any other effector that had a measurable individual contribution to BHKY virulence on either squash or muskmelon; however, we did observe significant differences in virulence among the pyramided mutants (Fig 11). Double mutants, which all involved loss of eop1 and one other effector, showed reduced virulence associated with the loss of any of the three effectors tested in squash (Fig 11A), and with the loss of hopO1 or hopL1, but not ospG, in muskmelon (Fig 11B).

The triple and quadruple deletion mutants did not exhibit an additional reduction in virulence beyond the reduction in virulence exhibited by some of the double-deletion mutants.

128

The lack of mutants with all permutations of the lost effectors precluded assessing all pairwise functional redundancies among the effectors. That said, the virulence of the pyramided effector mutants of BHKY demonstrated a high level of redundancy among the effectors Eop1,

Figure 11: The additive effects of effectors of the Et-C1 strain BHKY on its virulence on primary host squash (A) and secondary host muskmelon (B). Transformed proportion wilted (left panels) and area under disease progress curve (right panels) of muskmelon seedlings inoculated with combinatorial deletion mutants of eop1 or ospG or hopL1 or hopO1 in BHKY background. Data symbols and statistics are as described in Figure 4. Each experiment was repeated at least twice with 9 replicates per treatment.

OspG, HopO1 and HopL1 in their contributions to virulence on both hosts, with a redundancy between Eop1 and OspG on squash but not muskmelon being the primary difference I the effector contributions on the two hosts. Moreover, a relatively high level of residual virulence of

129 the mutants lacking all four of these effectors indicates that factors beyond these effectors contribute to virulence on both hosts. One of these factors is DspE, which was not induced here because it is absolutely required for wilt induction.

Discussion

Among the phytopathogenic species of Erwinia, our current understanding of the in planta expression profile and functions of effectors during plant-pathogen interactions has been limited to E. amylovora (6; 36; 48; 50). There has been a paucity of knowledge of such in other species, including E. tracheiphila. In this study, we sequenced the whole genomes of two E. tracheiphila strains, namely an Et-melo strain SCR3 and an Et-C1 strain BHKY, using the

PacBio platform. Using a local BLAST database of known effectors, and based on top-scoring

BLASTP matches, we identified the distribution of the predicted Hrp T3SE repertoire between

SCR3 and BHKY and used publicly available genomes of the Et-melo strain MDCuke (59) and the Et-C1 strain BuffGH (61) as references. We characterized the expression profile of selected effectors, in planta, following recovery of SCR3 from a host muskmelon and a non-host squash, and of BHKY from primary host squash and secondary host muskmelon. Based on expression profile and presence, we demonstrated the roles of selected effectors in virulence, through single and pyramided mutational analyses.

Our findings showed that the Et-melo and Et-C1 strains have distinct profiles of predicted

T3SEs based on genome sequence. Although a recent study reported that E. tracheiphila strains fall into clades that differ in their T3SEs and virulence patterns across Cucurbita species (60), this is the first study to evaluate the in planta expression profile of E. tracheiphila effectors. We selected effectors whose nucleotide sequences are intact and which are present as single copies in

130 the genomes of BHKY and SCR3 for expression measurements. Our results demonstrated that

BHKY and SCR3 exhibited host-specific differences in the in planta expression of their predicted T3SE genes. Although they both expressed one effector gene, eop1, in multiple plant species, the two strains also exhibited strain-specific differences in the effector genes they expressed in planta. Furthermore, our combinatorial mutant analysis demonstrated that, consistent with their expression profiles, effectors DspE, AvrB4, OspG and Eop1 contribute to the virulence of SCR3 on its host, muskmelon. In addition to DspE, which is required for wilt induction by BHKY on both its hosts (Chapter 3), the effectors HopO1, HopL1 and Eop1 contribute to BHKY virulence on both hosts, despite the absence of detectable expression of hopL1 in squash, and the effector OspG contributes to virulence on muskmelon but we found no evidence for activity on squash. We observed functional overlap among many of these effectors.

These findings generally support a correlation between the in planta expression profile and a role for these effectors in virulence.

The pathogenicity of most Gram-negative phytopathogenic bacteria depends on the translocation of effector proteins into plant cells through the T3SS (8). These effectors are known to target and modulate a variety of host cellular components and molecular functions, that favor pathogen establishment and disease (20). The genomes of bacterial xylem pathogens encode varying numbers of T3SEs, ranging from 60-75 in Ralstonia solanacearum (43; 49) to 47 in Xanthomonas campestris pv. campestris (26), 6 in E. amylovora (5; 9; 58), and 1 in Pantoea stewartii subsp. stewartii (27). We found that, in contrast to E. amylovora, genomes of the Et- melo and Et-C1 strains of E. tracheiphila have a total of 17 effectors, of which 10 are shared

(AvrB4, DspE, Eop1, HopA1, HopE1, HopF2, HopG1, HopR1, HopV1 and OspG), 4 are Et- melo-specific (Eop3, HopAF1 and Skwp2) and 3 are Et-C1-specific (HopO1, HopL1, HopAM1

131 and NleD). These findings place E. tracheiphila as the first Erwinia species with a relatively high number of effectors. These also support a previous finding that E. tracheiphila effectors are conserved within their phylogenomic clades based on evaluation of effector repertoire of 88 distinct strains (60).

Our in planta expression results demonstrated that BHKY and SCR3 exhibited some host-specific differences in the in planta expression of their predicted T3SE genes. We found that, among the shared effectors, both strains expressed eop1 on both muskmelon and squash.

However, in contrast to BHKY, whose eop1 expression was similar on both hosts were similar,

SCR3 expressed eop1 about 60-fold more in the non-host squash than it did in the host muskmelon, contradicting the host-specific induction observed for most effectors (34). We found that the T3SS pilin protein of SCR3 was not expressed in squash (Chapter 4), which suggests that SCR3 is unable to secrete and translocate effectors into squash. SCR3 expressed three effectors on its host muskmelon but only two of these on the non-host squash. BHKY expressed hopO1, which is present in BHKY but not SCR3, at an exceptionally high level in both its primary and secondary hosts, along with one shared and one unshared in each squash and muskmelon. Similar to its expression pattern in E. amylovora (6; 63) and Pantoea stewartii ssp. stewartii (28), dspE was expressed at an extremely low level in both strains in both hosts.

Our combinatorial mutant analysis demonstrated that the effectors DspE, AvrB4, OspG and Eop1 all contribute to the fitness of SCR3 on its host, muskmelon. We previously demonstrated that the loss of DspE greatly reduced the virulence of SCR3 on muskmelon, whereas the loss of Eop1 did not (Chapter 3). Here, we evaluated the roles of AvrB4 and OspG as additional virulence factors in SCR3, and used combinatorial mutants to evaluate potential functional redundancies among the effectors. Loss of AvrB4, the second most expressed effector

132 of SCR3 in muskmelon, reduced its virulence on muskmelon. Loss of OspG alone did not affect virulence, but its loss in combination with loss of either Eop1 or DspE resulted in reduced virulence of SCR3 on muskmelon. Although not significant, the loss of AvrB4 may have further reduced the virulence of SCR3 lacking DspE and OspG suggesting additive effects of DspE,

OspG and AvrB4, with Eop1 showing functional redundancy with OspG and AvrB4. These findings are consistent with the additive effects of phytopathogenic bacterial effectors on their hosts (18; 53).

These findings showed that E. tracheiphila is the first species of Erwinia demonstrated not only to contain the effectors AvrB4 and OspG, but also to have biological functions for these effectors. Pseudomonas syringae pv. phaseolicola strain 1448a has two variants of AvrB4, namely AvrB4-1 and AvrB4-2, which are paralogues of each other with 99% protein sequence similarity (70). Loss of both AvrB4-1 and AvrB4-2 reduced the virulence of the pathogen on bean plants, although loss of either alone did not (38). To the best of our knowledge, little is known of the biochemical or molecular functions of AvrB4, an effector that lacks any known motifs in its protein sequence. Although we did not test for a role in host-specificity, an AvrB4 homologue in Xanthomonas campestris pv. malvacearum strain XcmH1005 is structurally similar to the host-specific pathogenicity gene pthA found in X. citri (67; 72). As an effector kinase from the human pathogen Shigella, OspG exploits host ubiquitin for full virulence (16;

32). Certain phytopathogenic bacterial T3SEs have an E3 ubiquitin ligase domain that has been implicated in the ubiquitination of host substrates; these T3SEs include XopL in Xanthomonas campestris pv. vesicatoria (64) and AvrPtoB in P. syringae pv. tomato strain DC3000 (23). We identified 4 predicted ubiquitination sites and their corresponding substrate motifs in OspG of E.

133

tracheiphila (Fig 12); the role of these sites in the function of OspG and possible ubiquitination

of host proteins to suppress innate immunity would be worth pursuing in the future.

Figure 12: The Erwinia tracheiphila effector OspG has four predicted ubiquitination sites. Ubiquitination sites and corresponding substrate motifs were predicted using the UbiSite online program (http://csb.cse.yzu.edu.tw/UbiSite/). Identified ubiquitination sites are present in the 53rd, 54th, 71st and 148th amino acid residues of the OspG sequence. (Figure will be presented as a Supplementary data in manuscript).

Surprisingly, despite its high expression level, loss of HopO1 by itself did not noticeably

change the virulence of BHKY on squash and muskmelon. Similarly, the loss of HopO1 in P.

syringae did not reduce its virulence on Arabidopsis thaliana, but reduced its ability to multiply

in planta (25). Based on the high expression of both hopO1 and eop1 in BHKY on both hosts,

we examined the impact of the combined loss of these two effectors on virulence. The loss of

both HopO1 and Eop1 BHKY virulence on squash and muskmelon, indicating that these two

effectors overlap in their functions.

Similarly, the loss of HopL1 which was expressed in BHKY only in muskmelon, did not

alter its virulence, but the combined loss of both Eop1 and HopL1 in BHKY reduced virulence

134 on muskmelon, and surprisingly, also on squash. HopL1 is a virulence factor in P. cichorii, based on that a deletion mutant was reduced I virulence on tomato seedlings (44). Consistent with a previous study, DspE was expressed at a level just at the threshold of detection in BHKY in both hosts, perhaps as a result of its toxic nature to plant cells (19), but it is required for the pathogenicity of BHKY on squash and muskmelon. Apart from the critical role of DspE to its pathogenicity, we did not detect any other effector that had a measurable individual contribution to fitness of BHKY on either squash or muskmelon. Although the combined loss of Eop1 and

OspG did not alter the virulence of BHKY on muskmelon, the additional loss of HopL1 reduced virulence, demonstrating complexity in the functional overlap among the effectors of BHKY, similar to what was found among the Pseudomonads (13; 24; 35).

The effector repertoire of E. tracheiphila strains was recently reported (60). In this study, we have provided the first characterization of the E. tracheiphila expression of T3SEs and have demonstrated the virulence functions of these effectors through single and pyramided mutant analyses. Our findings showed that different sets of effectors individually and collectively contributed to SCR3 virulence on muskmelon, versus to BHKY virulence on squash and muskmelon, highlighting the distinct repertoires of effectors contributing to the virulence of each clad of E. tracheiphila.

References 1. Alfano JR, Collmer A. 1997. The type III (Hrp) secretion pathway of plant pathogenic bacteria: trafficking harpins, Avr proteins, and death. J. Bacteriol. 179:5655-62 2. Asolkar T, Ramesh R. 2018. Development of T3SS Mutants (hrpB (-) and hrcV (-) ) of Ralstonia solanacearum, Evaluation of Virulence Attenuation in Brinjal and Tomato-A Pre-requisite to Validate T3Es of R-solanacearum. Indian J. Microbiol. 58:372-80 3. Aziz RK, Bartels D, Best AA, DeJongh M, Disz T, et al. 2008. The RAST server: Rapid annotations using subsystems technology. BMC Genomics 9

135

4. Baumler A, Fang FC. 2013. Host specificity of bacterial pathogens. Cold Spring Harb. Perspect. Med. 3:19 5. Bocsanczy AM, Huguet-Tapia JC, Norman DJ. 2017. Comparative genomics of Ralstonia solanacearum identifies candidate genes associated with cool virulence. Front. Plant Sci. 8:16 6. Boureau T, ElMaarouf-Bouteau H, Garnier A, Brisset MN, Perino C, et al. 2006. DspA/E, a type III effector essential for Erwinia amylovora pathogenicity and growth in planta, induces cell death in host apple and nonhost tobacco plants. Mol. Plant-Microbe Interact. 19:16-24 7. Bove JM, Garnier M. 2003. Phloem-and xylem-restricted plant pathogenic bacteria (vol 163, pg 1083, 2002). Plant Sci. 164:421-+ 8. Buttner D. 2016. Behind the lines-actions of bacterial type III effector proteins in plant cells. Fems Microbiol. Rev. 40:894-937 9. Castiblanco LF, Triplett LR, Sundin GW. 2018. Regulation of effector delivery by type III secretion chaperone proteins in Erwinia amylovora. Front. Microbiol. 9:10 10. Chang JH, Urbach JM, Law TF, Arnold LW, Hu A, et al. 2005. A high-throughput, near- saturating screen for type III effector genes from Pseudomonas syringae. Proc. Natl. Acad. Sci. U. S. A. 102:2549-54 11. Chen C, Malek AA, Wargo MJ, Hogan DA, Beattie GA. 2010. The ATP-binding cassette transporter Cbc (choline/betaine/carnitine) recruits multiple substrate-binding proteins with strong specificity for distinct quaternary ammonium compounds. Mol. Microbiol. 75:29-45 12. Coburn B, Sekirov I, Finlay BB. 2007. Type III secretion systems and disease. Clin. Microbiol. Rev. 20:535-+ 13. Collmer A, Lindeberg M, Petnicki-Ocwieja T, Schneider DJ, Alfano JR. 2002. Genomic mining type III secretion system effectors in Pseudomonas syringae yields new picks for all TTSS prospectors. Trends Microbiol. 10:462-9 14. Correa VR, Majerczak DR, Ammar ED, Merighi M, Pratt RC, et al. 2012. The bacterium Pantoea stewartii uses two different type III secretion systems to colonize its plant host and insect vector. Appl. Environ. Microbiol. 78:6327-36 15. Datsenko KA, Wanner BL. 2000. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. U. S. A. 97:6640-5 16. de Jong MF, Alto NM. 2014. Thinking outside the Osp(G)- kinase activation by E2- ubiquitin. Embo J. 33:403-4 17. de Mackiewicz D, Gildow FE, Blua M, Fleischer SJ, Lukezic FL. 1998. Herbaceous weeds are not ecologically important reservoirs of Erwinia tracheiphila. Plant Dis. 82:521-9

136

18. DebRoy S, Thilmony R, Kwack YB, Nomura K, He SY. 2004. A family of conserved bacterial effectors inhibits salicylic acid-mediated basal immunity and promotes disease necrosis in plants. Proc. Natl. Acad. Sci. U. S. A. 101:9927-32 19. Degrave A, Moreau M, Launay A, Barny MA, Brisset MN, et al. 2013. The bacterial effector DspA/E is toxic in Arabidopsis thaliana and is required for multiplication and survival of fire blight pathogen. Mol. Plant Pathol. 14:506-17 20. Deslandes L, Rivas S. 2012. Catch me if you can: bacterial effectors and plant targets. Trends Plant Sci. 17:644-55 21. Garcia-Calderon CB, Casadesus J, Ramos-Morales F. 2007. Rcs and PhoPQ regulatory overlap in the control of Salmonella enterica virulence. J. Bacteriol. 189:6635-44 22. Gazi AD, Sarris PF, Fadouloglou VE, Charova SN, Mathioudakis N, et al. 2012. Phylogenetic analysis of a gene cluster encoding an additional, rhizobial-like type III secretion system that is narrowly distributed among Pseudomonas syringae strains. BMC Microbiol. 12:15 23. Goehre V, Spallek T, Haeweker H, Mersmann S, Mentzel T, et al. 2008. Plant pattern- recognition receptor FLS2 is directed for degradation by the bacterial ubiquitin ligase AvrPtoB. Current Biology 18:1824-32 24. Greenberg JT, Vinatzer BA. 2003. Identifying type III effectors of plant pathogens and analyzing their interaction with plant cells. Curr. Opin. Microbiol. 6:20-8 25. Guo M, Chancey ST, Tian F, Ge ZX, Jamir Y, Alfano JR. 2005. Pseudomonas syringae type III chaperones ShcO1, ShcS1, and ShcS2 facilitate translocation of their cognate effectors and can substitute for each other in the secretion of HopO1-1. J. Bacteriol. 187:4257-69 26. Guy E, Genissel A, Hajri A, Chabannes M, David P, et al. 2013. Natural Genetic Variation of Xanthomonas campestris pv. campestris Pathogenicity on Arabidopsis Revealed by Association and Reverse Genetics. mBio 4:13 27. Ham JH, Majerczak D, Ewert S, Sreerekha MV, Mackey D, Coplin D. 2008. WtsE, an AvrE-family type III effector protein of Pantoea stewartii subsp stewartii, causes cell death in non-host plants. Mol. Plant Pathol. 9:633-43 28. Ham JH, Majerczak DR, Nomura K, Mecey C, Uribe F, et al. 2009. Multiple Activities of the Plant Pathogen Type III Effector Proteins WtsE and AvrE Require WxxxE Motifs. Mol. Plant-Microbe Interact. 22:703-12 29. Ho SN, Hunt HD, Horton RM, Pullen JK, Pease LR. 1989. Site-directed mutagenesis by overlap extension using the polymerase chain-reaction. Gene 77:51-9 30. Jin QL, He SY. 2001. Role of the Hrp pilus in type III protein secretion in Pseudomonas syringae. Science 294:2556-8 31. Khan MA, Zhao YF, Korban SS. 2012. Molecular mechanisms of pathogenesis and resistance to the bacterial pathogen Erwinia amylovora, causal agent of fire blight disease in rosaceae. Plant Mol. Biol. Rep. 30:247-60

137

32. Kim DW, Lenzen G, Page AL, Legrain P, Sansonetti PJ, Parsot C. 2005. The Shigella flexneri effector OspG interferes with innate immune responses by targeting ubiquitin- conjugating enzymes. Proc. Natl. Acad. Sci. U. S. A. 102:14046-51 33. King EO, Ward MK, Raney DE. 1954. Two simple media for the demonstration of pyocyanin and fluorescin J. Lab. Clin. Med. 44:301-7 34. Kleemann J, Rincon-Rivera LJ, Takahara H, Neumann U, van Themaat EVL, et al. 2012. Sequential Delivery of Host-Induced Virulence Effectors by Appressoria and Intracellular Hyphae of the Phytopathogen Colletotrichum higginsianum. PLoS Pathog. 8 35. Kvitko BH, Park DH, Velasquez AC, Wei CF, Russell AB, et al. 2009. Deletions in the Repertoire of Pseudomonas syringae pv. tomato DC3000 Type III Secretion Effector Genes Reveal Functional Overlap among Effectors. PLoS Pathog. 5:16 36. Lalli DA, Artlip TS, Wisniewski ME, Norelli JL, Farrell RE. 2008. Transgenic expression of Erwinia amylovora effectors Eop1 and HopC(Ea) in apple. In Proceedings of the Eleventh International Workshop on Fire Blight, ed. KB Johnson, VO Stockwell, 793:241-+. Leuven 1: Int Soc Horticultural Science. Number of 241-+ pp. 37. Lu HB, Lema AS, Planas-Marques M, Alonso-Diaz A, Valls M, Coll NS. 2018. Type III secretion-dependent and -independent phenotypes caused by Ralstonia solanacearum in Arabidopsis Roots. Mol. Plant-Microbe Interact. 31:175-84 38. Macho AP, Zumaquero A, Gonzalez-Plaza JJ, Ortiz-Martin I, Rufian JS, Beuzon CR. 2012. Genetic analysis of the individual contribution to virulence of the Type III effector inventory of Pseudomonas syringae pv. phaseolicola. PLoS One 7 39. Markel E, Stodghill P, Bao ZM, Myers CR, Swingle B. 2016. AlgU Controls Expression of Virulence Genes in Pseudomonas syringae pv. tomato DC3000. J. Bacteriol. 198:2330-44 40. Mayjonade B, Gouzy J, Donnadieu C, Pouilly N, Marande W, et al. 2016. Extraction of high-molecular-weight genomic DNA for long-read sequencing of single molecules. Biotechniques 61:203-5 41. Mergaert J, Hauben L, Cnockaert MC, Swings J. 1999. Reclassification of non- pigmented Erwinia herbicola strains from trees as Erwinia billingiae sp. nov. Int. J. Syst. Bacteriol. 49:377-83 42. Mohammadi AS, Maynard ET, Foster RE, Egel DS, McNamara KT. 2018. Muskmelon Cultivar Attractiveness to Striped Cucumber Beetle and Susceptibility to Bacterial Wilt. Hortscience 53:782-7 43. Mukaihara T, Tamura N, Iwabuchi M. 2010. Genome-wide identification of a large repertoire of Ralstonia solanacearum type III effector proteins by a new functional screen. Mol. Plant-Microbe Interact. 23:251-62 44. Nagendran R, Lee YH. 2019. SrfC of Pseudomonas cichorii JBC1 affects its attachment to the host surface and host tissue infection. Plant Pathology 68:1099-108

138

45. Nissinen RM, Ytterberg AJ, Bogdanove AJ, Van Wijk KJ, Beer SV. 2007. Analyses of the secretomes of Erwinia amylovora and selected hrp mutants reveal novel type III secreted proteins and an effect of HrpJ on extracellular harpin levels. Mol. Plant Pathol. 8:55-67 46. Oh HS, Kvitko BH, Morello JE, Collmer A. 2007. Pseudomonas syringae lytic transglycosylases coregulated with the type III secretion system contribute to the translocation of effector proteins into plant cells. J. Bacteriol. 189:8277-89 47. Parte AC. 2018. LPSN - List of prokaryotic names with standing in nomenclature (bacterio.net), 20 years on. International Journal of Systematic and Evolutionary Microbiology 68:1825-9 48. Pester D, Milcevicova R, Schaffer J, Wilhelm E, Blumel S. 2012. Erwinia amylovora Expresses Fast and Simultaneously hrp/dsp Virulence Genes during Flower Infection on Apple Trees. PLoS One 7:9 49. Poueymiro M, Genin S. 2009. Secreted proteins from Ralstonia solanacearum: a hundred tricks to kill a plant. Curr. Opin. Microbiol. 12:44-52 50. Pulawska J, Kaluzna M, Warabieda W, Mikicinski A. 2017. Comparative transcriptome analysis of a lowly virulent strain of Erwinia amylovora in shoots of two apple cultivars - susceptible and resistant to fire blight. BMC Genomics 18:15 51. Qian W, Jia YT, Ren SX, He YQ, Feng JX, et al. 2005. Comparative and functional genomic analyses of the pathogenicity of phytopathogen Xanthomonas campestris pv. campestris. Genome Res. 15:757-67 52. Rand FV, Enlows EMA. 1916. Transmission and control of bacterial wilt of cucurbits. J. Agric. Res. 6:417-34 53. Rufian JS, Lucia A, Rueda-Blanco J, Zumaquero A, Guevara CM, et al. 2018. Suppression of HopZ Effector-Triggered Plant Immunity in a Natural Pathosystem. Front. Plant Sci. 9:16 54. Rutherford K, Parkhill J, Crook J, Horsnell T, Rice P, et al. 2000. Artemis: sequence visualization and annotation. Bioinformatics 16:944-5 55. Saalau Rojas E, Batzer JC, Beattie GA, Fleischer SJ, Shapiro LR, et al. 2015. Bacterial Wilt of Cucurbits: Resurrecting a Classic Pathosystem. Plant Dis. 99:564-74 56. Saalau Rojas E, Dixon PM, Batzer JC, Gleason ML. 2013. Genetic and virulence variability among Erwinia tracheiphila strains recovered from different cucurbit hosts. Phytopathology 103:900-5 57. Salaau Rojas E, Dixon PM, Batzer JC, Gleason ML. 2013. Genetic and virulence variability among Erwinia tracheiphila strains recovered from different cucurbit hosts. Phytopathology 103:900-5 58. Schropfer S, Bottcher C, Wohner T, Richter K, Norelli J, et al. 2018. A single effector protein, AvrRpt2(EA), from Erwinia amylovora can cause fire blight disease symptoms

139

and induces a salicylic acid-dependent defense response. Mol. Plant-Microbe Interact. 31:1179-91 59. Shapiro LR, Andrade A, Scully ED, Rocha J, Paulson JN, Kolter R. 2018. Draft genome sequence of an Erwinia tracheiphila isolate from an infected muskmelon (Cucumis melo). Microbiol. Resour. Ann. 7:2 60. Shapiro LR, Paulson JN, Arnold BJ, Scully ED, Zhaxybayeva O, et al. 2018. An Introduced Crop Plant Is Driving Diversification of the Virulent Bacterial Pathogen Erwinia tracheiphila. mBio 9:20 61. Shapiro LR, Scully ED, Roberts D, Straub TJ, Geib SM, et al. 2015. Draft Genome Sequence of Erwinia tracheiphila, an Economically Important Bacterial Pathogen of Cucurbits. Microbiol. Resour. Ann. 3:2 62. Shapiro LR, Scully ED, Straub TJ, Park J, Stephenson AG, et al. 2016. Horizontal gene acquisitions, mobile element proliferation, and genome decay in the host-restricted plant pathogen Erwinia tracheiphila. Genome Biol. Evol. 8:649-64 63. Siamer S, Patrit O, Fagard M, Belgareh-Touze N, Barny MA. 2011. Expressing the Erwinia amylovora type III effector DspA/E in the yeast Saccharomyces cerevisiae strongly alters cellular trafficking. FEBS Open Bio 1:23-8 64. Singer AU, Schulze S, Skarina T, Xu XH, Cui H, et al. 2013. A Pathogen Type III Effector with a Novel E3 Ubiquitin Ligase Architecture. PLoS Pathog. 9 65. Smith EF. 1911. Bacteria in relation to plant diseases. Nature 74:294 66. Smits THM. 2019. The importance of genome sequence quality to microbial comparative genomics. BMC Genomics 20 67. Swarup S, Yang YN, Kingsley MT, Gabriel DW. 1992. Altered host specificity in Race 1 of Xanthomonas malvaceous by passage through a resistantvariety of Gossypium hirsutum. Mol. Plant-Microbe Interact. 5:204-13 68. Wang H, Yang ZL, Du S, Ma L, Liao Y, et al. 2018. Characterization of Pectobacterium carotovorum proteins differentially expressed during infection of Zantedeschia elliotiana in vivo and in vitro which are essential for virulence. Mol. Plant Pathol. 19:35-48 69. Wattam AR, Davis JJ, Assaf R, Boisvert S, Brettin T, et al. 2017. Improvements to PATRIC, the all-bacterial bioinformatics database and analysis resource center. Nucleic Acids Research 45:D535-D42 70. Wroblewski T, Caldwell KS, Piskurewicz U, Cavanaugh KA, Xu HQ, et al. 2009. Comparative large-scale analysis of interactions between several crop species and the effector repertoires from multiple pathovars of Pseudomonas and Ralstonia. Plant Physiol. 150:1733-49 71. Wu L, McGrane RS, Beattie GA. 2013. Light Regulation of Swarming Motility in Pseudomonas syringae Integrates Signaling Pathways Mediated by a Bacteriophytochrome and a LOV Protein. mBio 4

140

72. Yang YN, Defeyter R, Gabriel DW. 1994. Host-specific symptoms and increased release of Xanthomonas citri and Xanthomonas campestris pv. malvacearum from leaves are determined by the 102 bp tandem repeats of Ptha and AvrB6, respectively. Mol. Plant- Microbe Interact. 7:345-55 73. Yao CB, Zehnder G, Bauske E, Klopper J. 1996. Relationship between cucumber beetle (Coleoptera: Chrysomelidae) density and incidence of bacterial wilt of cucurbits. Journal of Economic Entomology 89:510-4 74. Zhao YF, Blumer SE, Sundin GW. 2005. Identification of Erwinia amylovora genes induced during infection of immature pear tissue. J. Bacteriol. 187:8088-103

141

CHAPTER 6

CONCLUSIONS AND FUTURE REMARKS

E. tracheiphila was one of the first bacterial phytopathogens ever described, and a well- established finding about this pathosystem is its ability to cause bacterial wilt on cucurbits in a host-specific manner; for example, strains isolated from cucumber showed a rapid wilt on cucumber, but slight or no symptoms when inoculated into squash or pumpkin (9-12; 14). In addition, genetic variability among strains has been reported, with evidence from rep-PCR profiles (11) and whole genome phylogeny (12), showing a correlation between genetic variability and virulence patterns. Shapiro et al. [7] showed that E. tracheiphila strains clustered into three phylogenetic clades (Et-melo, Et-C1 and Et-C2) based on similarity across the genomes in draft genome sequences. The authors also found that the genome of E. tracheiphila also contains type III system secreted effectors (T3SEs). However, a fundamental question of the mechanism behind host-specificity and pathogenicity remained unknown.

In this study, we characterized whole genome sequences and physiological traits of 24 E. tracheiphila strains in two of these clades, leading us to propose that the Et-melo and Et-C1 clades be delineated as E. tracheiphila subsp. tracheiphila and E. tracheiphila subsp. lentilata, respectively. Here, we identified the distribution of the T3SE repertoire of two strains, SCR3 (Et- melo) and BHKY (Et-C1), and characterized the expression profile of selected effectors in planta using RT-qPCR. We explored the role of one highly expressed effector, Eop1, as a host- specificity candidate, and found that loss of eop1 did not alter the virulence of two Et-melo strains or an Et-C1 strain on their respective hosts. However, over-expression of eop1 from an

Et-melo strain in an Et-C1 strain increased its virulence on muskmelon, but not on squash, indicating that Eop1 functions as a host-specific virulence factor. The expression profiling in

142 planta also highlighted poor expression of a key type III secretion system (T3SS) pilus gene, hrpA, in Et-melo in squash. We demonstrated experimentally that over-expression of hrpA in an

Et-melo strain enabled this strain to infect squash, demonstrating that host-specific hrpA expression is another factor contributing to host-specific virulence in E. tracheiphila.

Our expression profile data are limited, as they were only targeted towards a group of proteins, effectors. Hence, we are currently limited in our understanding of a genomewide expression pattern for strains from both clades. Eop1 appears to be a global host-specific virulence determinant. Although we found that Eop1 quantitatively contributed to host-specific virulence through heterologous expression of an Et-melo Eop1 variant in an Et-C1 strain, the inability to detect virulence phenotypes in our deletion mutants, unlike what was reported in

Eop1 homologues in E. amylovora (1) and Ralstonia solanacearum (8) suggests that other factors may have a direct and more pronounced impact on host-specific virulence in E. tracheiphila. Findings from our pangenome analysis showed the presence of shared and clade- specific genes. Therefore, deciphering the role of these genes in host-specific virulence, through their global expression patterns, will benefit from a genomewide analysis of transcriptional activity of E. tracheiphila (via RNA-Seq). This is particularly important for E. tracheiphila, which has about 20% of its protein coding sequences pseudogenized (13); a global genomewide expression profile will enhance our understanding of the impact of pseudogenization and mobile element invasion on E. tracheiphila whole-genome expression.

We also examined the role of the effector DspE in pathogenicity and found that loss of dspE from Et-melo strains significantly reduced, but did not eliminate, virulence on muskmelon, whereas loss of dspE from Et-C1 strains resulted in a complete loss of pathogenicity on squash and muskmelon. Thus, DspE has distinct roles in the two E. tracheiphila clades. This work

143 represents the first characterization of the major molecular drivers of host specificity and pathogenicity among E. tracheiphila strains.

Despite the inability of the Et-melo strain SCR3 to induce wilt in squash, the cell densities in squash were sufficiently high to extract RNA, as they were similar in their dynamics to those in muskmelon. Also, SCR3 did not express hrpA within the non-host squash. A fundamental question still remains as to why are the Et-melo strains unable to significantly wilt squash; based on our findings, we propose a combination of host, pathogen and host-pathogen factors as putative factors.

For host factors, we hypothesize that the inability of Et-melo strains to significantly wilt squash is as a result of difference in host xylem sap composition. This is supported by a recent finding where it was shown that maize lines exhibited nutritional immunity in response to

Pantoea stewartii (5), a response that involved altering the xylem sap composition. So, it is conceivable that a similar response in squash alters its xylem sap composition in a way that disfavors the production of extracellular polysaccharide (EPS) by Et-melo strains. Since EPS production is presumed to be the major factor that leads to wilt through xylem occlusion, replication without EPS production may explain their inability to cause significant wilt. To address this question, metabolic assays could be performed on sap collected from infected plants that represent each strain-host combination to see if there is a differential strain-specific and/or host-specific sap composition. Alternatively, the expression of EPS genes in Et-melo cells in planta could be examined.

One of the pathogen factors that could restrict wilt of squash by the Et-melo strains is a difference in quorum-sensing regulation, which could influence the production of EPS, as was reported in strains of Pantoea stewartii (2; 3). This is particularly important since our

144 pangenome analysis showed that strains of both clades have genes for the synthesis of the quorum signal autoinducer-2, as well as two or three acyl-homoserine lactone (AHL) synthase genes; however, only the Et-C1 clade has a gene for the cognate AHL-binding transcriptional regulator (ABTR), which may be required for the activation of expression of a set of genes specifically needed to cause wilt on squash. Although Et-melo strains are able to reach a high density in the non-host squash, a lack of ABTR may prevent them from expressing relevant pathogenicity and/or virulence genes. To test this hypothesis, a genomewide transcriptomic analysis of wild-type strains of both clades from xylem sap collected from melon and squash seedlings, as well as an Et-C1 ABTR deletion mutant and an Et-melo expressing ABTR, would reveal the impact of the presence and/or absence of ABTR in quorum-sensing regulation of pathogenicity genes in squash and also muskmelon.

Table 1: Percent Identity Matrix of aligned CAR1 sequences

Arabidopsis thaliana Cucurbita moschata Cucumis melo Arabidopsis thaliana 100.0 26.7 27.6 Cucurbita moschata 26.7 100.0 53.0 Cucumis melo 27.6 53.0 100.0

Moreover, a host-pathogen factor could be involved based on a recent study that showed that the coiled-coil nucleotide-binding leucine rich repeat host immune receptor CAR1 (CEL-

Activated Resistance 1) recognized the P. syringae effector AvrE (7), a homologue of DspE, and the recognition conferred resistance to Arabidopsis when spray-inoculated with P. syringae. The authors verified this by demonstrating that, when P. syringae was inoculated into an Arabidopsis car1 deletion mutant, it incited yellow chlorotic symptoms as a result of the lost interaction between the effector and its receptor, consistent with the gene-for-gene model of resistance (6).

Interestingly, we found that CAR1 homologues in squash (Cucurbita moschata) and muskmelon

145

(Cucumis melo var. makuwa) genomes shared 26.7% and 27.6% sequence similarity with

Arabidopsis and only 53.0% sequence similarity with each other (Table 1). It is conceivable that such a low level of similarity between CAR1 sequences of squash and muskmelon is a key driver of host-specific virulence, especially as it relates to their abilities to recognize E. tracheiphila effector genes. I hypothesize that the difference in host-specific virulence is determined by a difference in CAR1-DSPE recognition events in squash versus muskmelon. To test this hypothesis, wild type or car1 deletion mutants of both hosts could be inoculated with the wild type and dspE deletion mutants of strains from both clades. The outcome of putative gene-for- gene interactions could be evaluated by disease ratings. It is conceivable that the Et-C1 strains have evolved to avoid DspE recognition by both muskmelon and squash CAR1 receptors, particularly because DspE is the sole pathogenicity determinant among these strains, resulting in the ability to wilt both squash and muskmelon. In contrast, since other unknown effectors compensate for the loss of dspE in the Et-melo strains, the presence of these compensating effectors may influence the difference in ability to be recognized by either muskmelon or squash

CAR1. This study is opined to give insights into how the dynamics of the gene-for-gene model of disease resistance determines the outcome of such interactions.

Finally, using pyramided mutant analysis, we demonstrated that, consistent with their expression profiles in planta, the effectors DspE, OspG, Eop1 and AvrB4 contributed to the virulence of Et-melo on muskmelon, whereas DspE, OspG, Eop1, HopL1 and HopO1 contributed to the virulence of Et-C1 on squash and muskmelon, with highly overlapping functions among many of the effectors. This work thus illustrates the contribution of distinct sets of effectors to the virulence of E. tracheiphila strains on their hosts. This work sets the foundation for a better understanding of the molecular interaction between E. tracheiphila and its

146 cucurbit hosts, although thus far, our work has not provided evidence for the existence of resistance genes in cucurbits that could be used to breed against bacterial wilt development.

Having found evidence for the role of the Hrp T3SS in cucurbit pathogenesis, presumably through its role in the secretion of effectors, the role of the Salmonella type T3SS (ST-T3SS) remains unknown in E. tracheiphila. I hypothesize that this second T3SS plays a role in persistence in cucumber beetle, like it does in Pantoea stewartii in the corn flea beetle (4). To evaluate this hypothesis, a deletion mutant of this locus or a relevant control gene in E. tracheiphila could be used to monitor colonization in cucumber beetles using fluorescently- labelled wild type and mutant strains, with colonization evaluated using confocal laser microscopy. Findings from such a study could set the foundation for developing cucurbit lines with the ability to obliterate the expression of the ST-T3SS in E. tracheiphila upon plant infection, so that even if beetles acquire bacteria from infected plants, they are unable to transmit them to neighboring plants, thus mitigating rates of bacterial transmission. The advent and rapid development of CRISPR-based gene editing tools will likely make this technology available for crop development.

References

1. Asselin JE, Bonasera JM, Kim JF, Oh CS, Beer SV. 2011. Eop1 from a Rubus strain of Erwinia amylovora functions as a host-range limiting factor. Phytopathology 101:935-44

2. Bodman SB, Carlier AL, Stevens AM. 2008. Role of quorum-sensing regulation in pathogenesis of Pantoea stewartii subsp stewartii. Washington: Amer Soc Microbiology. 201-12 pp.

3. Bodman SB, Majerczak DR, Coplin DL. 1998. A negative regulator mediates quorum- sensing control of exopolysaccharide production in Pantoea stewartii subsp. stewartii. Proc. Natl. Acad. Sci. U. S. A. 95:7687-92

147

4. Correa VR, Majerczak DR, Ammar ED, Merighi M, Pratt RC, et al. 2012. The bacterium pantoea stewartii uses two different type III secretion systems to colonize its plant host and insect vector. Appl. Environ. Microbiol. 78:6327-36

5. Doblas-Ibanez P, Deng KY, Vasquez MF, Giese L, Cobine PA, et al. 2019. Dominant, heritable resistance to stewart's wilt in maize is associated with an enhanced vascular defense response to Infection with Pantoea stewartii. Mol. Plant-Microbe Interact. 32:1581-97

6. Flor HH. 1971. Current status of gene-for-gene concept. Annual Review of Phytopathology 9:275-+

7. Laflamme B, Dillon MM, Martel A, Almeida RND, Desveaux D, Guttman DS. 2020. The pan-genome effector-triggered immunity landscape of a host-pathogen interaction. Science 367:763-+

8. Lavie M, Shillington E, Eguiluz C, Grimsley N, Boucher C. 2002. PopP1, a new member of the YopJ/AvrRxv family of type III effector proteins, acts as a host-specificity factor and modulates aggressiveness of Ralstonia solanacearum. Mol. Plant-Microbe Interact. 15:1058-68

9. Liu Q, Beattie GA, Rojas ES, Gleason ML. 2018. Bacterial wilt symptoms are impacted by host age and involve net downward movement of Erwinia tracheiphila in muskmelon. Eur. J. Plant Pathol. 151:803-10

10. Nazareno ES, Dumenyo CK. 2015. Modified inoculation and disease assessment methods reveal host specificity in Erwinia tracheiphila-Cucurbitaceae interactions. Microb. Pathog. 89:184-7

11. Saalau Rojas E, Dixon PM, Batzer JC, Gleason ML. 2013. Genetic and virulence variability among Erwinia tracheiphila strains recovered from different cucurbit hosts. Phytopathology 103:900-5

12. Shapiro LR, Paulson JN, Arnold BJ, Scully ED, Zhaxybayeva O, et al. 2018. An introduced crop plant is driving diversification of the virulent bacterial pathogen Erwinia tracheiphila. mBio 9:20

13. Shapiro LR, Scully ED, Straub TJ, Park J, Stephenson AG, et al. 2016. Horizontal gene acquisitions, mobile element proliferation, and genome decay in the host-restricted plant pathogen Erwinia tracheiphila. Genome Biol. Evol. 8:649-64

14. Vrisman CM, Deblais L, Rajashekara G, Miller SA. 2016. Differential colonization dynamics of cucurbit hosts by Erwinia tracheiphila. Phytopathology 106:684-92