HrcU and HrpP are pathogenicity factors in the fire blight pathogen Erwinia amylovora required for the type III secretion of DspA/E

Background Many Gram-negative bacterial pathogens mediate host-microbe interactions via utilization of the type III secretion (T3S) system. The T3S system is a complex molecular machine consisting of more than 20 proteins. Collectively, these proteins translocate effectors across extracellular space and into the host cytoplasm. Successful translocation requires timely synthesis and allocation of both structural and secreted T3S proteins. Based on amino acid conservation in animal pathogenic bacteria, HrcU and HrpP were examined for their roles in regulation of T3S hierarchy. Results Both HrcU and HrpP were shown to be required for disease development in an immature pear infection model and respective mutants were unable to induce a hypersensitive response in tobacco. Using in vitro western blot analyses, both proteins were also shown to be required for the secretion of DspA/E, a type 3 effector and an important pathogenicity factor. Via yeast-two hybridization (Y2H), HrpP and HrcU were revealed to exhibit protein-protein binding. Finally, all HrcU and HrpP phenotypes identified were shown to be dependent on a conserved amino acid motif in the cytoplasmic tail of HrcU. Conclusions Collectively, these data demonstrate roles for HrcU and HrpP in regulating T3S and represent the first attempt in understanding T3S heirarchy in E. amylovora. Electronic supplementary material The online version of this article (doi:10.1186/s12866-016-0702-y) contains supplementary material, which is available to authorized users.


Background
The type III secretion (T3S) system is a common feature of Gram-negative bacterial pathogens. The T3S system functions to facilitate the translocation of bacterial effector proteins into eukaryotic host cells where they suppress host defense responses, facilitate colonization, and promote disease development [1]. Consequently, T3S has been the focus of intensive research in both animal and plant pathosystems.
The T3S system is a complex proteinaceous machine consisting of more than 20 components. Because the successful translocation of bacterial effectors necessitates a functioning multipartite machine, the production of structural and secreted T3S system components has been assumed to be hierarchical. Recent analyses have confirmed the hierarchical nature of T3S in a few animal and plant pathogens [2][3][4][5][6][7]. Characterization of this hierarchy has revealed multiple substrate classes. Early substrates are involved in pilus formation while late substrates, like effectors, are secreted after the assembly of a complete T3S system.
The Gram-negative plant pathogenic bacterium Erwinia amylovora is the causative agent of fire blight, a disease of rosaceous species including apple and pear. Disease development by E. amylovora requires a functioning T3S system [26]. In E. amylovora, the T3S system is known to secrete at least 12 proteins including the harpins HrpN and HrpW as well as the effector DspA/E (hereafter termed DspE), a pathogenicity factor [27][28][29]. To date, little is known about how secretion hierarchy is regulated in E. amylovora. While HrpJ is required for secretion of translocators HrpN and HrpW, nothing is known about how E. amylovora regulates the substrate specificity of DspE, the most important component of the T3S system for fire blight disease development [29]. In E. amylovora, YscU/FlhB and TS34 proteins are represented by HrcU (EAM_2905) and HrpP (EAM_2900), respectively. Here, HrcU and HrpP are explored for roles in T3S system regulation in E. amylovora.

Results
HrcU exhibits a conserved NPTH motif required for pathogenicity in E. amylovora The NPTH motif in YscU/FlhB proteins is the site of autoproteolytic cleavage and conformational change required for protein function [4,15]. This NPTH motif is conserved in all known YscU/FlhB proteins [3,5,30]. Bioinformatic analysis of HrcU from E. amylovora using a dense alignment surface algorithm predicted that, like YscU/FlhB homologs, HrcU encodes four transmembrane domains as well as a cytoplasmic Cterminal tail (Fig. 1a) [10,31]. Using T-Coffee multiple alignment software, the amino acid sequence of HrcU was compared to multiple homologs in T3S systems of plant and animal bacterial pathogens as well as in the flagellum ( Table 1) [32]. The E. amylovora HrcU NPTH motif (HrcU NPTH ) was found to be conserved in E. amylovora and in all analyzed homologs (Fig. 1b).
To determine the role of HrcU in disease development, a chromosomal deletion of hrcU was created in E. amylovora Ea1189. Ea1189ΔhrcU was confirmed to be The NPTH motif is labeled in red and the arrow represents the site of cleavage and conformational change reported in homologous proteins. b T-coffee multiple sequence alignment of C-terminal NPTH motif in HrcU homologs. Weblogo was used to visualize an alignment of HrcU homologs from animal pathogens Yersinia enterocolitica, Shigella flexneri and Escherichia coli, plant pathogens Erwinia amylovora, Xanthomonas campestris and Pseudomonas syringae as well as a flagellar homolog from Salmonella enteric. NPTH is conserved in all HrcU homologs nonpathogenic due to a lack of symptom development 6 days post inoculation (dpi) in an immature pear infection model (Fig. 2a). In trans expression of hrcU via the plasmid pRRM1 was able to successfully complement the mutant strain restoring full virulence to Ea1189ΔhrcU (Fig. 2a).
To ascertain the importance of HrcU NPTH in E. amylovora, HrcU was subjected to site-directed mutagenesis. The asparagine residue of the NPTH motif is required for YscU/FlhB protein function in assayed homologs [4,5,17]. Consequently, the conserved asparagine residue located at position 266 in the amino acid sequence of HrcU was mutated to encode a codon corresponding to alanine. This hrcU mutant allele (HrcU N266A ) was cloned into an expression vector creating pRRM2 [33].
To determine the role of HrcU N266A in host-microbe interactions, Ea1189ΔhrcU/pRRM1 and Ea1189ΔhrcU/ pRRM2 were inoculated into immature pear fruits. While plasmid-borne hrcU was able to re-establish wild type (WT) virulence levels to Ea1189ΔhrcU, Ea1189ΔhrcU/pRRM2 was unable to restore pathogenicity 6 dpi in immature pear fruits (Fig. 2a). This indicates that HrcU NPTH is required for HrcU function and that HrcU NPTH is necessary to mediate compatible host interactions.
HrcU NPTH is required for the elicitation of the hypersensitive response The hypersensitive response (HR) is a hallmark of incompatible plant-microbe interactions. The HR is characterized by rapid, localized programmed cell-death in response to pathogen-associated proteins frequently represented by T3S system substrates. HR elicitation in E. amylovora requires a functional T3S system [34]. E. amylovora Ea1189 strains were inoculated into Nicotiana benthamiana mesophyll tissue and, 16 h post inoculation (hpi), results revealed that E. amylovora Ea1189 requires HrcU, and specifically HrcU NPTH , for HR development (Fig. 2b). While WT Ea1189 and complemented Ea1189ΔhrcU/pRRM1 induced robust HR symptoms in N. benthamiana, Ea1189ΔhrcU and Ea1189ΔhrcU expressing HrcU N266A failed to trigger an incompatible defense response (Fig. 2b). As the HR in response to E. amylovora infection requires T3S, these results suggest that the inability of HrcU N266A to complement Ea1189ΔhrcU is due to the disrupted function of HrcU NPTH in mediating T3S.

HrpP is required for pathogenicity and hypersensitive response induction
In YscU/FlhB proteins, the NPTH motif is required for the regulation of T3S hierarchy [3][4][5]12]. T3S system hierarchy regulation is mediated via direct and indirect interactions with T3S4 proteins [3,4,16,25]. In E. amylovora, HrpP (EAM_2900) is a predicted T3S4 protein. Bioinformatic analyses of the HrpP amino acid sequence are in accordance with previous observations that T3S4 proteins are poorly conserved  Fig. 2 Phenotypic characterization of HrcU-related mutant strains in Ea1189. WT Ea1189, Ea1189ΔhrcU and Ea1189ΔhrcU stains expressing native HrcU from pRRM1 or HrcU N266A from pRRM2 were inoculated into a immature pear fruits and b Nicotiana benthamiana. Pear fruit necrosis was recorded 6 days post inoculation while the hypersensistive response in N. benthamiana was observed 16 h post inoculation. Ea1189ΔhrcU was non-pathogenic and unable to elicit a hypersensitive response while pRRM2 was unable to complement the hrcU null mutation between species and that, in bacterial plant pathogens, T3S4 proteins are N-terminally truncated relative to homologs in animal pathogenic bacteria and the flagellum (Fig 3a) [8]. Beginning at amino acid position 98 though, HrpP does exhibit a modified Pro-X-Leu-Gly motif that is characteristic of T3S4 with alanine replacing leucine at position 100 (Pro-Glu-Ala-Gly) (Fig 3b) [35].
To establish the role of HrpP in mediating plantmicrobe interactions, a chromosomal deletion of HrpP was synthesized, and relevant strains were inoculated into host and non-host plant species. Like Ea1189ΔhrcU strains expressing HrcU N266A , Ea1189ΔhrpP was nonpathogenic 6 dpi in immature pear fruit and unable to elicit a HR in N. benthamiana (Fig 3c).

HrcU and HrpP interact in E. amylovora
While not all T3S4 proteins have been observed to interact directly with YscU/FlhB counterparts, direct interactions have been recorded between the T3S4 proteins FliK and HpaC [3,16]. To explore the possibility of HrpP interactions with HrcU in E. amylovora, hrpP and hrcU constructs were cloned into Y2H vectors and assayed in Saccharomyces cerevisiae AH109 via survival on minimal medium and α-galactosidase activity. In the Y2H assay, hrcU alleles featuring the HrcU N266A point mutation were included along with N-terminal hrcU deletions (HrcU-CT). HrcU-CT constructs were included due to reported transmembrane domain interference with protein-protein interactivity in homologous YscU/FlhB proteins [3,16]. Alongside HrpP, the T3S protein HrpJ was also screened for the ability to interact with HrcU in yeast as HrpJ is a demonstrated regulator of T3S system hierarchy in E. amylovora and homologs are required for late substrate secretion due to their roles as T3S inner rod proteins [29]. Another example of this occurs in Pseudomonas syringae where HrpJ functions within the bacterial cell to control secretion of translocator proteins such as the harpins HrpZ1 and HrpW1 [36].
In all cases, full-length HrcU encoding N-terminal transmembrane domains were unable to interact with either HrpP or HrpJ (Fig 4). HrpJ exhibited a very weak interaction with both HrcU-CT and HrcU-CT N266A (Fig 4). Conversely, HrpP interacted strongly with HrcU-CT in Y2H experiments (Fig 4). While the HrcU NPTH motif was not absolutely required for interactions with HrpP, HrcU-CT N266A displayed less α-galactosidase activity in the presence of HrpP than did HrcU-CT (Fig 4). This indicates that HrpP does interact with HrcU and that HrcU NPTH -mediated conformational changes in HrcU affect HrpP binding in Y2H assays. All qualitative Y2H results were assessed quantitatively using image analysis software ImageJ and shown to be statistically significant.
HrcU NPTH and HrpP are required for the secretion of DspE The T3S system effector DspE is a pathogenicity factor of E. amylovora and the translocation of DspE is required for fire blight disease development [27,28,37,38]. Mutations affecting the T3S system that result in a loss-ofpathogenicity phenotype are consequently hypothesized to be attributed to decreased DspE translocation by E. amylovora. To determine if HrcU NPTH and HrpP are involved in regulating DspE secretion, E. amylovora strains were transformed with pLRT201 to express a DspE-CyaA fusion protein and incubated in vitro in hrp-inducing minimal medium (HrpMM) used to mimic conditions of the plant apoplast [39]. Proteins were extracted 48 hpi and subjected to one-dimensional SDS-PAGE separation and western blot analysis using an anti-CyaA antibody. As predicted, an Ea1189 strain harboring native hrcU secreted DspE in vitro while Ea1189ΔhrcU failed to secrete any DspE protein (Fig 5a). Likewise, an Ea1189 strain synthesizing HrcU N266A and Ea1189ΔhrpP were also unable to secrete DspE (Fig 5a). Using SDS-PAGE analysis, we also show that Ea1189ΔhrcU complemented with the fulllength hrcU on pRRM1 secreted the native DspE protein, while Ea1189ΔhrcU complemented with hrcU N266A on pRRM2 was unable to secrete DspE (Fig 5b). Finally, the wild-type Ea1189 strain containing pRRM2 could still secrete DspE, indicating that HrcU N266A does not exhibit a dominant-negative effect on HrcU (Fig 5b). These results show that HrcU NPTH and HrpP are required for DspE secretion in vitro and suggest that Ea1189ΔhrpP and Ea1189ΔhrcU/pRRM2 are nonpathogenic due to loss of DspE secretion and translocation capability.

Discussion
In this study the roles of HrcU and HrpP in regulating the T3S system in E. amylovora Ea1189 were explored. Using site-directed mutagenesis, phenotypic analyses, Y2H assays and protein visualization, HrcU and HrpP were shown to interact and mediate host-microbe interactions via the regulation of T3S system substrates like the effector DspE. HrcU and HrpP were both confirmed to be pathogenicity factors in E. amylovora Ea1189. Ea1189ΔhrcU and Ea1189ΔhrpP were both unable to cause disease in immature pear fruits ( Fig. 2a and 3c). Likewise, Ea1189ΔhrcU and Ea1189ΔhrpP were also unable to elicit a HR after inoculation into N. benthamiana (Fig 2b and 3c). These results are in agreement with previous observations regarding HrcU in P. syringae and X. campestris [5,40]. Interestingly, while HrpP in E. amylovora and P. syringae are both required for disease and HR induction, the T3S4 homolog HpaC is not a pathogenicity factor in X. campestris [41,42].
The important influence of HrcU and HrpP in facilitating disease development is hypothesized to stem from roles in regulating T3S hierarchy. In YscU/FlhB proteins, the regulation of T3S hierarchy hinges on a conserved NPTH amino acid motif [4,5,9,13]. The cytoplasmic C-terminus of HrcU in E. amylovora encodes an NPTH motif (Fig. 1b). Notably, a site-directed mutation of hrcU resulting in the construct HrcU N266A was unable to complement Ea1189ΔhrcU suggesting that the role of HrcU in mediating plant-microbe interactions requires the presence of an asparagine residue at position 266 (Fig. 2). Ea1189ΔhrcU strains expressing HrcU N266A were nonpathogenic and here we report via western blot analysis that HrcU-mediated secretion of DspE was dependent on the integrity of its conserved NPTH motif (Fig. 5). While full-length hrcU was able to complement Ea1189ΔhrcU in trans and restore DspE secretion, hrcU N266A failed to rescue Ea1189ΔhrcU  Fig. 4 HrcU yeast two-hybrid interaction assays (Y2H). Native HrcU, HrcU N266A and C-terminal (CT) truncations of both proteins were cloned into the bait vector pGBKT7. HrpJ and HrpP were expressed from the prey vector pGADT7. Blue coloration indicates the strength of protein-protein interactions. All Y2H interactions were quantified relative to an empty-vector control using ImageJ software and tested for significance using Kendall rank correlation coefficient τ tests. Both full-length and truncated HrcU N266A exhibited impaired interactions with HrpP and HrpJ relative to native HrcU constructs mutant phenotypes ( Fig. 2 and 5). These phenotypes are likely linked as DspE secretion is required for disease development [38]. Collectively, results illustrating the roles of HrcU and HrcU N266A in E. amylovora reinforce data highlighting the importance of the NPTH motif in YscU/FlhB proteins as synonymous mutations in X. campestris, enteropathogenic E. coli and Y. enterocolitica also abolish disease development [5,43,44]. Notably though, while the NPTH motif is required for T3S-dependent disease development, YscU/FlhB mediated regulation of T3S differs between bacterial species.
In Salmonella, the flagellar protein FlhB functions to establish hook assembly prior to filament secretion. Consequently, FlhB N269A mutants fail to terminate hook protein secretion and initiate export of flagellin [43]. YscU in turn regulates the secretion of late substrates including translocators and effectors [4,45].
One of the lesser-described proteins in the YscU/FlhB family is EscU from Escherichia coli EPEC strain E2348/69. EscU is particularly relevant to discussions of E. amylovora. Thomassin et al. [46] observed that EscU N262A poorly secretes effectors while differentially regulating the secretion of effector chaperones. While chaperone EspC was secreted at wild-type levels, EspA, EspB and Tir were poorly secreted in vitro. The authors also revealed that Tirinduced actin polymeration was comparably reduced in infected HeLa cells. E. amylovora also utilizes a large consortium of chaperones to regulate effector secretion. Special attention should be given to how chaperones interact with HrcU to regulate secretion hierarchy [38,47].
HrcU from X. campestris pv. vesicatoria represents the most characterized YscU/FlhB protein in plant pathogenic bacteria. Like YscU, HrcU Xcv inhibits the secretion of late substrates. HrcU Xcv NPTH mutants in turn over-secrete early T3S substrates analogous to increased hook secretion exhibited by FlhB flagellar mutants [5,9]. Research concerning HrcU Xcv has developed to reveal that while the NPTH motif in YscU/FlhB proteins has been the focus of much attention, additional HrcU domains and amino acid residues play a role in regulating T3S. While this research is the first demonstration of HrcU Ea controlling substrate specificity in hrp1-T3S systems via the NPTH motif, more analyses are required to understand the full scope of HrcU-mediated T3S regulation.
Examinations of the T3S4 protein HrpP in E. amylovora revealed that, while exhibiting a c-terminal P-X-L-G motif, HrpP is diminutive like T3S4 proteins in other plant pathogenic bacteria. Conversely, mammalian bacterial pathogens exhibit T3S4 proteins up 3X in length. While structurally distinct, all T3S4 proteins share some commonalities. Here we present for the first time that HrpP is a pathogenicity factor in E. amylovora Ea1189. Ea1189ΔhrpP was unable to generate disease development on immature pear fruit or induce a HR in N. benthamiana (Fig. 3c). Like HrcU N266A , Ea1189ΔhrpP was also unable to secrete the T3S effector DspE as displayed using an in vitro western blot (Fig. 5).
Other T3S4 proteins, similarily to HrpP in E. amylovora, function to promote the secretion of late T3S substrates. Null mutations in YscP from Y. enterocolitica, HpaC from X. campestris and HrpP from P. syringae all fail to secrete late T3S substrates [16,23,41,42]. More . Native HrcU, HrpP and the HrcU NPTH domain were required for DspE secretion as Ea1189ΔhrcU/pRRM2 expressing HrcU N266A was unable to complement DspE secretion into the supernatant. HrcU and HrpP did not affect the production of DspE in the cell pellet. b Separate images composing (a). c T3S secretome in Ea1189 strains. All strains were cultured in minimal medium and after processing, were separated via one-dimensional SDS-PAGE and stained with silver nitrate. DspE is represented by a band corresponding to~200 kDa and marked with an arrow. Both WT Ea1189 and Ea1189ΔhrcU/pRRM1 secrete DspE in vitro. Ea1189ΔhrcU and Ea1189ΔhrcU/pRRM2 cannot secret DspE indicating that HrcU N266A is required for DspE secretion. WT Ea1189 producing HrcU N266A does not exhibit a dominant negative effect on DspE secretion notably though, many T3S4 proteins also function to actively suppress the secretion of early T3S substrates and null mutations result in increased secretion of pilus subunits and inner rode proteins. FliK from the flagellum suppresses the secretion of the inner rod-like hook protein FlhB [13,48]. Mutations in YscP trigger hypersecretion of the pilin protein YscF and the inner rod protein YscI while HpaC mutants secrete more inner rod protein HrpB2 than wild-type [16,23,45].
Conversely, HrpP from P. syringae pv. tomato appears to function atypically relative to other know T3S4 proteins. While previously-described T3S4 proteins actively suppress early T3S events, PstΔhrpP poorly secretes the early substrate HrpA, a T3S pilus subunit protein [42]. Consequently, HrpP Pst may be more accurately described as a post-translational activator of T3S as opposed to a T3S4 protein. More experimentation will be required to determine if HrpP Ea also functions as a post-translation activator though it is important to note that HrpP Pst has not been observed to interact with HrcU Pst while evidence suggests that HrpP Ea binds HrcU Ea as has been reported for other canonical T3S4 proteins [25,[49][50][51][52].
Using Y2H analysis to explore protein-protein interactions, we demonstrate that HrpP and the cytoplasmic tail of HrcU bind when co-expressed in Saccharomyces cerevisiae (Fig. 4). These results conform to previous observations in other plant and animal pathogenic bacteria. For example, in Salmonella, FlhB binds directly to FliK [50] and in Xanthomonas, HrcU Xcv directly binds HpaC [5,9]. Noteably, YscU from Yersinia has never been shown to interact with YscP indicating potential variability in T3S hierarchy regulation [17]. Despite some variability, to date, all known YscU/FlhB NPTH domains are required for T3S function and disease development.
In E. amylovora, virulence and DspE secretion assays are consistent with previous observations concerning the functional importance of the HrcU NPTH domain ( Fig. 2 and 5). Our Y2H results reveal however that while HrcU NPTH affects HrpP Ea interactions, the domain is not required for binding (Fig. 4). Confirmation of observed HrcU Ea -HrpP Ea interaction data in yeast will require future use of more sensitive and specific techniques such as co-immunoprecipitation assays. In addition, work by Haunser and Buttner [9] also indicates that HrcU Xcv exhibits multiple amino acid residues, in addition to the NPTH domain, with functional significance for plant disease outcomes and, in response, a more through mutational analysis will be required to understand the role of HrcU Ea in T3S regulation and interactions with HrpP Ea . Considering that both HrcU and HrpP are pathogenicity factors in E. amylovora and as both Ea1189ΔhrcU and Ea1189ΔhrpP exhibit impaired DspE secretion, it is tempting to speculate that interactions between HrcU and HrpP may be important for their relative roles in pathogenicity.

Conclusions
Here we report the first information regarding the roles of HrcU and HrpP in regulating T3S of DspE in E. amylovora. Both proteins were shown to be required for pathogenesis in E. amylovora Ea1189, were required for DspE secretion, and show evidence of inter-protein interactions. Future work should focus on how HrcU and HrpP regulate the secretion of the inner rod protein HrpJ and the needle protein HrpA as well as how regulator copy number influences HrpA secretion. In addition, as effector chaperones are known to be regulated by YscU/FlhB proteins in animal pathogenic bacteria, identifying a role for HrcU in regulating plant pathogenic effector chaperones would be a novel contribution to the plant pathology community. Table 2 lists bacterial strains and plasmids used in this study. Unless otherwise referenced, bacterial strains were grown in Luria Bertani (LB) broth supplemented with 50 μg ml −1 ampicillin, 20 μg ml −1 chloramphenicol, 12 μg ml −1 oxytetracycline or 30 μg ml −1 kanamyacin where appropriate. All strains were cultured at 28°C in a shaking incubator.

DNA manipulation and cloning
Restriction enzyme digestion, T4 DNA ligation, and PCR amplification of genes were carried out using standard molecular techniques [53]. DNA extraction, PCR purification, plasmid extraction, and isolation of DNA fragments from agarose were performed with related kits (Qiagen, Valencia, CA). The sequences of oligonucleotide primers used in this study are listed in Additional file 1: Table S1. All DNA was sequenced at the Research Technology Support Facility at Michigan State University. Double digestion and directional ligation into pBBR1MCS3 [54] with PCR-generated gene sequences was utilized for mutant strain complementation. Final constructs were transformed into competent Ea1189 by electroporation and screened on LB agar plates amended with oxytetracycline.

Virulence and hypersensitive response assays
The virulence of E. amylovora Ea1189 was assayed using a standard immature pear fruit assay as described previously [58]. In brief, bacterial strains were cultured overnight, washed, and resuspended in 0.5x phosphate buffered saline (PBS) to 1 × 10 3 to 1 × 10 4 CFU/ml. Immature pear fruits (Pyrus communis L. cv. Bartlett) were surface sterilized with 10 % bleach, dried in laminar flow hood, and pricked with a needle prior to application of 2 μl bacterial suspension. Inoculated pears were incubated at 28°C in humidified chambers. Symptoms were recorded 6 days post inoculation. The experiments were repeated three times with six replications per experiment. To study elicitation of the HR during incompatible interactions, E. amylovora strains were cultured over night in LB broth. Bacterial cells were collected via centrifugation and washed twice with 0.5X PBS. Cells were resuspended and adjusted to a final concentration of 1 × 10 7 CFU ml −1 in 0.5X PBS. 100 μl of cell suspension were in turn infiltrated into 9-week-old N. benthamiana leaves using a syringe and HR was observed 16 hpi.

Mutagenesis
E. amylovora site-directed nonpolar chromosomal mutants were generated using the phage λ Red recombinase system previously described [59]. Briefly, E. amylovora strain Ea1189 harboring pKD46, encoding recombinases red β, γ, and exo, was cultured overnight at 28°C in a shaking incubator. Strains were reinoculated with 0.1 % L-arabinose in LB broth and cultured for four hours to exponential phase. Cells were made electrocompetent and stored at -80°C. Homologous recombination fragments encoding acetyltransferase cassettes were generated via polymerase chain reaction (PCR) using the plasmid pKD3 as a template. A PCR purification kit (Qiagen; Valencia, CA) was using to purify recombination fragments before electroporation into competent Ea1189. LB agar amended with chloramphenicol was used to screen putative mutants and single-gene recombinatorial deletion was confirmed using PCR and functional complementation.
Site-directed point-mutations were introducted into HrcU using and as described by the Altered Sites® II in vitro Mutagenesis System (Promega; Madison, WI). Briefly, full-length HrcU (NC_013971) was cloned into pAlter-Ex1 via NcoI and NsiI restriction sites creating pRRM9. Mutagenic oligonucleotide HrcU_N266A (5′-GACCTGCTGCTGGTCGCTCCCA CGCACTATGCG-3′) was designed to convert the asparagine amino acid at position 266 to an alanine residue. pRRM9 and HrcU_N266A primer were denatured and phosphorylated respectively and via PCR pAlter-Ex1(hrcU N266A ) (pRRM10) was synthesized and transformed into competent E. coli cells.

Yeast two-hybridization
The bait vector pGBKT7 and the prey vector pGADT7 were used for yeast expression and Y2H screening (Clontech, Mountain View, CA, USA). A Frozen-EZ Yeast Transformation II Kit was used to create competent Saccharomyces cerevisiae AH109 and for cotransformation of bait and prey (Zymo Research Corporation, Orange, CA, USA). Transformants were selected on minimal SD agar amended with -Ade/-His/-Leu/-Trp dropout supplement and Mel1 α-galactosidase activity was detected using topically applied X-α-Gal at 4 ug ul −1 (Clontech, Mountain View, CA, USA). The intensity of blue color was quantified using ImageJ (http://imagej.nih.gov/ij/download.html). Kendall rank correlation coefficient τ tests were performed to determine the statistical significance. Equations are listed below:

Secretion assays
Strains were cultured overnight in 50 ml LB broth at 28°C. Cells were washed twice with 0.5X PBS, and resuspended in 50 ml minimal medium, pH 5.7 [60]. Strains were induced for 48 h with shaking, collected by centrifugation, and the supernatant was filtered using 0.22 μm vacuum filtration (Millipore, Billerica, MA, USA). Filtrate was supplemented with 0.5 mM phenylmethylsulfonyl fluoride and concentrated to approximately 500 μl using 10-kDa Amicon centrifugal filter units (Millipore, Billerica, MA, USA). For ease of detection of secreted DspE protein, we used plasmid pLRT201 which is an expression construct that encodes the first 737 amino acids of DspE fused to an adenylate cyclase (CyaA) reporter [38]. We have previously demonstrated secretion of this DspE 1-737 -CyaA fusion protein via the T3S system [38]. DspE secretion was examined in the WT E. amylovora Ea1189/pLRT201, Ea1189ΔhrcU/pLRT201, Ea1189ΔhrpP/pLRT201 and Ea1189ΔhrcU/pRRM2/ pLRT201.
For western blot analysis, proteins were analyzed using anti-CyaA antibody (Santa Cruz Biotechnology, Santa Cruz, CA). For protein visualization, proteins were additionally purified to remove biofilm polysaccharides as previously described [29]. Briefly, protein samples were extracted twice with 0.5 volume of water-saturated phenol and precipitated with by the addition of 5 volumes 100 mM ammonium acetate in methanol. After overnight incubated at −20°C, protein were extracted via centrifugation, resuspended in 50 ul water and reprecipitated in 500 μl of cold acetone. Samples were again incubated overnight at −20°C and protein pellets were collected by centrifugation at 13,000 g at 4°C for 30 min and subsequent resuspension in 50 μl 5 % acetic acid supplement with 0.5 mM PMSF. A bicinchoninic acid (BCA) protein assay kit was used to measure protein concentrations and concentrations were τ ¼ n c −n d n represents sample size 1/2n(n − 1) n c is the number of concordant pairs z ¼ 3 Ã τ Ã √n n−1 ð Þ n d is the number of discordant pairs √ 2(2n + 5)