HrpB7 from Xanthomonas campestris pv. vesicatoria is an essential component of the type III secretion system and shares features of HrpO/FliJ/YscO family members

The Gram‐negative bacterium Xanthomonas campestris pv. vesicatoria translocates effector proteins via a type III secretion system (T3SS) into eukaryotic cells. The T3SS spans both bacterial membranes and consists of more than 20 proteins, 9 of which are conserved in plant and animal pathogens and constitute the core subunits of the secretion apparatus. T3S in X. campestris pv. vesicatoria also depends on nonconserved proteins with yet unknown function including HrpB7, which contains predicted N‐ and C‐terminal coiled‐coil regions. In the present study, we provide experimental evidence that HrpB7 forms stable oligomeric complexes. Interaction and localisation studies suggest that HrpB7 interacts with inner membrane and predicted cytoplasmic (C) ring components of the T3SS but is dispensable for the assembly of the C ring. Additional interaction partners of HrpB7 include the cytoplasmic adenosinetriphosphatase HrcN and the T3S chaperone HpaB. The interaction of HrpB7 with T3SS components as well as complex formation by HrpB7 depends on the presence of leucine heptad motifs, which are part of the predicted N‐ and C‐terminal coiled‐coil structures. Our data suggest that HrpB7 forms multimeric complexes that associate with the T3SS and might serve as a docking site for the general T3S chaperone HpaB.


| INTRODUCTION
The Gram-negative plant pathogenic bacterium Xanthomonas campestris pv. vesicatoria (Xcv; also designated X. euvesicatoria [Jones, Lacy, Bouzar, Stall, & Schaad, 2004]) is the causal agent of bacterial spot disease on pepper and tomato plants and one of the model organisms for the analysis of bacterial infection strategies (Büttner & Bonas, 2010). Pathogenicity of Xcv depends on the translocation of effector proteins into plant cells where they interfere with essential plant cellular processes and thus promote bacterial proliferation (Büttner, 2016;Büttner & Bonas, 2010;Dean, 2011). Effector protein delivery is mediated by a type III secretion (T3S) system, which is present in many Gram-negative plant and animal pathogenic bacteria and is related to the bacterial flagellum. Both systems are, therefore, referred to as translocation-associated and flagellar T3SS (Abby & Rocha, 2012;Büttner, 2012). The membrane-spanning secretion apparatus of both systems contains at least eight conserved core components, which are known as secretion and cellular translocation (Sct) proteins in animal pathogenic bacteria followed by a letter corresponding to the nomenclature of T3SS components from Yersinia spp. (Hueck, 1998).
Structural studies of isolated T3SSs from animal pathogenic bacteria revealed the presence of ring components in both bacterial membranes (Büttner, 2012;Deng et al., 2017). The outer membrane (OM) ring, also termed secretin, consists of members of the SctC protein family and is connected on the extracellular side to a pilus-like structure known as T3S needle in animal and T3S pilus in plant pathogenic bacteria. T3S needles or pili serve as transport channels for effector proteins to the eukaryotic plasma membrane (Büttner, 2012;Deng et al., 2017). On the periplasmic side, the OM ring is in contact with an inner rod structure and the inner membrane (IM) rings, which are assembled by SctD and SctJ proteins (Büttner, 2012;Deng et al., 2017;Diepold & Wagner, 2014). Embedded in the IM rings is the export apparatus, which presumably forms a transport channel for secreted proteins and consists of members of the SctU, SctV, SctR, SctS, and SctT families (Büttner, 2012;Deng et al., 2017;Lara-Tejero & Galan, 2019). The export apparatus associates with the cytoplasmic adenosinetriphosphatase (ATPase) complex (SctN and SctL), which presumably unfolds T3S substrates and is surrounded by a predicted cytoplasmic ring (C ring) or pod-like structures as was described for several T3SSs from animal pathogenic bacteria. C ring or pod-like structures are assembled by members of the SctQ family and are presumably involved in substrate recognition (Diepold et al., 2015;Hu et al., 2015;Lara-Tejero, 2019;Makino et al., 2016;Morita-Ishihara et al., 2006;Spaeth et al., 2009).
To date, the architecture and mode of action of T3SSs have been intensively studied in animal pathogenic bacteria, whereas much less is known about T3SSs from plant pathogenic bacteria. T3S genes from plant pathogens were designated hypersensitive response and pathogenicity (hrp) because they are essential for disease symptoms on susceptible plants and the induction of the effector-triggered immunity in resistant plants (Büttner & He, 2009;Schmidt & Hensel, 2004;Tampakaki et al., 2010). Effector-triggered immunity depends on the recognition of individual effectors in plants with cognate resistance genes and often leads to the induction of the hypersensitive response (HR), a local cell death at the infection site, which restricts bacterial multiplication (Gill et al., 2015). Hrp-T3SSs from plant pathogenic bacteria have been classified into Hrp1-(in Erwinia spp. and Pseudomonas syringae pathovars) and Hrp2-T3SSs (in Xanthomonas spp. and Ralstonia solanacearum) according to similarities in the genetic organisation and regulation of hrp genes (Alfano & Collmer, 1997;Bogdanove et al., 1996;Troisfontaines & Cornelis, 2005). In Xcv, hrp genes are activated when the bacteria enter the plant or are cultivated in special minimal media. hrp gene expression depends on two regulatory proteins, HrpG and HrpX, which are encoded outside the hrp gene cluster Büttner & Bonas, 2010;. HrpG is an OmpRtype response regulator that perceives an environmental signal via a yet unknown mechanism and activates HrpX, an AraC-type transcriptional regulator .
HrpX binds to conserved DNA motifs in the promoter regions of most hrp operons and activates hrp gene expression (Koebnik, et al., 2006;Noël et al, 2001).
The hrp gene cluster from Xcv contains eight operons with 25 genes including 11 conserved hrc (hrp conserved), 7 nonconserved hrp, and 6 hpa (hrp associated) genes as well as the effector gene xopF1 . hrc genes encode the core components of the T3SS such as the ATPase, the membrane rings, the predicted C ring, and the ATPase complex. The single-letter nomenclature of Hrc proteins refers to the corresponding Sct proteins from animal pathogens (Bogdanove et al., 1996;. In contrast to Hrc proteins, Hrp proteins are not widely conserved, and homologues are mainly found within one bacterial species. Hrp proteins from Xcv and other Xanthomonas spp. include the predicted periplasmic inner rod proteins HrpB1 and HrpB2, the pilus protein HrpE, the translocon protein HrpF, and proteins of yet unknown function such as HrpB4, HrpB7, and HrpD6 Hartmann et al., 2012;Hausner et al., 2013;. Hrc and Hrp proteins are essential for T3S as structural components, whereas the accessory Hpa (Hrp associated) proteins are often involved in the control of T3S. One example is the T3S chaperone HpaB, which binds to type III effectors and presumably targets them to the T3S ATPase (Büttner et al., 2004;Büttner et al., 2006;Lonjon et al., 2017;Lorenz & Büttner, 2009;Prochaska et al., 2018;Scheibner et al., 2018). HpaB contributes to the translocation of most effector proteins and is, therefore, essential for pathogenicity.
Focus of the present study is the functional characterisation of HrpB7, which is encoded downstream of the ATPase gene hrcN by the seventh gene in the hrpB operon. HrpB7 is conserved in Xanthomonas spp. and shares amino acid sequence identity with HrpD from R. solanacearum. In Xcv, HrpB7 is the only hrp gene product with predicted N-and C-terminal coiled coils and contains features of HrpO/FliJ/YscO family members, which are small α-helical coiled-coil proteins encoded downstream of the T3S-ATPase genes from plant and animal pathogenic bacteria. In the present study, we show that HrpB7 from Xcv is required for secretion and translocation of early and late T3S substrates. HrpB7 forms stable protein complexes even in the absence of the T3SS and interacts with the ATPase and the predicted C ring, suggesting that it is a cytoplasmic structural component of the T3SS. The analysis of HrpB7 mutant derivatives by in vitro and in vivo studies revealed that protein function and complex formation depend on the N-and C-terminal leucine heptad motifs, which are part of the predicted coiled-coil structures of HrpB7 and are presumably involved in protein-protein interactions.
2 | RESULTS 2.1 | HrpB7 contains leucine heptad motifs and regularly spaced cysteine residues HrpB7 from Xcv strain 85-10 is a small protein (169 amino acids), encoded downstream of the ATPase gene hrcN by the seventh gene of the hrpB operon ( Figure 1a). The N-terminal region of HrpB7 (amino acids 28-57) contains a motif of five regularly spaced cysteine residues (C-X 6 ) 5 , each separated from the neighbouring cysteine residue by six amino acids. This motif overlaps with a leucine-heptad motif (L-X 6 -L-X 6 -L) consisting of leucine residues at every seventh position (amino acids 26 to 40; Figure 1a Heptad motifs (also referred to as [abcdefg] n ) are often part of alphahelical coiled-coil structures and form two helical turns (3.5 amino acids per turn) with amino acids at positions "a" and "d" being generally hydrophobic, for example, leucine, valine, or isoleucine (Lupas & Gruber, 2005;Mason & Arndt, 2004). Given the threedimensional structure of the helix, residues at positions "a" and "d" are in close proximity to each other at one side of the helix, thus forming a hydrophobic region that stabilises the helical structure and can promote protein dimerisation or oligomerisation via hydrophobic interactions (Mason & Arndt, 2004;Figure 1b,c). In contrast to the "a" and "d" residues, positions "e" and "g" are usually occupied by solvent-exposed polar or charged residues (Mason & Arndt, 2004).
Opposite charges at positions "e" and "g" within one helix or in   F I G U R E 1 HrpB7 contains leucine heptad motifs and regularly spaced cysteine residues. (a) Schematic representation of the hrp gene cluster and amino acid sequence of HrpB7 (accession number CAJ22064). hrc, hrp, and hpa genes of the hrp gene cluster are represented by red, purple, and yellow arrows, respectively. The direction of transcription of single hrp operons is indicated. The xopF1 operon between hpaE and hrpF, which encodes the effector protein XopF1 and two putative associated chaperones, is replaced by a dashed line. The amino acid sequence of HrpB7 is given below the hrp gene cluster. HrpB7 contains leucine heptad motifs (indicated by red brackets) and five N-terminal regularly spaced cysteine residues, which are shown in bold letters. Numbers refer to amino acid positions. (b) Helical wheel representation of the N-terminal leucine heptad motifs in HrpB7. The letters A-G in the inner circle refer to amino acid positions 1-7 in the heptad motifs (see text), respectively. Amino acids are shown in coloured circles; numbers refer to the amino acid positions. Red circles represent amino acids with nonpolar residues. Blue, green, and purple circles refer to amino acids with polar, positively, and negatively charged amino acids, respectively, as indicated. Cysteine residues are represented by yellow circles. (c) Helical wheel representation of the C-terminal leucine heptad motifs in HrpB7. Amino acids are presented as described in the helical wheel representation of the N-terminal leucine heptad motifs in HrpB7 (b) neighbouring helices likely stabilise the coiled coil structure (Mason & Arndt, 2004). In the N-terminal two leucine heptad motifs of HrpB7, these criteria are fit when the leucine residues are located at position "a" and the regularly spaced cysteine residues of the (C-X 6 ) 5 motif at position "d" of each repeat (Figure 1b). In the C-terminal leucine heptad motif, position "d" is occupied by the nonpolar amino acids alanine, valine, and leucine ( Figure 1c).
In agreement with the presence of leucine heptad motifs in HrpB7, the analysis of the predicted secondary protein structure by Phyre2 (Kelley, et al., 2015) and SWISS-MODEL (Waterhouse et al., 2018) revealed a predominantly alpha-helical conformation of HrpB7 ( Figure S1). Furthermore, coiled-coil structures were predicted for regions spanning amino acids 15 to 50 and 88 to 156 of HrpB7 (prediction by programme coiled coil prediction; https://npsa-prabi.ibcp.fr/cgi-bin/npsa_automat.pl?page=npsa_ lupas.html). F I G U R E 2 Analysis of an Xcv hrpB7 deletion mutant. (a) HrpB7 is essential for pathogenicity of Xcv. Strains 85-10 (wt), 85* (wt), and corresponding hrpB7 deletion mutants (ΔhrpB7) with or without (−) expression constructs encoding HrpB7 as indicated were infiltrated at a density of 10 8 CFU ml −1 into leaves of susceptible ECW and resistant ECW-10R pepper plants. Disease symptoms were photographed 7 days post inoculation (dpi; for derivatives of strain 85*) and 10 dpi (for derivatives of strain 85-10). For the better visualisation of the HR, leaves were destained in ethanol 2 dpi. Dashed lines indicate the inoculated areas. HrpB7 was stably synthesised as is shown in Figure S2. (b) In planta growth of the hrpB7 deletion mutant. Xcv wild-type strain 85-10, the hrpB7 deletion mutant 85-10ΔhrpB7, and the ATPase-deficient strain 85-10ΔhrcN were infiltrated at a density of 10 4 CFU ml −1 into leaves of susceptible pepper plants.  Fenselau & Bonas, 1995). To further assess the contribution of HrpB7 to pathogenicity, we deleted codons 7 to 164 of hrpB7   Rossier et al., 1999;Wengelnik et al., 1999). The hrpB7 mutant phenotype was complemented by ectopic expression of hrpB7 under control of the lac promoter, suggesting that it was specifically caused by the absence of hrpB7 and not by a polar effect of the deletion on the expression of the downstream hrcT gene ( Figure 2a). This is an important control because hrpB7 from Xanthomonas oryzae pv. oryzicola was reported to contain a promoter for the downstream hrcT gene, which encodes an essential component of the export apparatus (Liu et al., 2014).
We also performed complementation studies with C-terminally c-Myc epitope-tagged derivatives of HrpB7 from Xcv and X. campestris pv. campestris (Xcc; HrpB7 Xcc ), which share 70% amino acid identity (77% sequence similarity) and show sequence variations in some of the N-terminal leucine and cysteine residues ( Figure S1). Both proteins restored pathogenicity in strain 85*ΔhrpB7, suggesting that they were functional ( Figure 2c). In contrast to HrpB7 Xcv -c-Myc, HrpB7 Xccc-Myc also fully complemented the phenotype of strain 85-10ΔhrpB7 with respect to disease symptoms and the HR ( Figure S2). Given that the ectopic expression of hrpB7 Xcv -c-myc in strain 85-10 suppressed the elicitation of plant reactions, we assume that enhanced levels of HrpB7 Xcv -c-Myc but not of HrpB7 Xcc -c-Myc interfere with pathogenicity in strain 85-10 ( Figure S2). Immunoblot analysis showed that all

| HrpB7 is essential for T3S and effector translocation
In addition to infection studies, we investigated the contribution of HrpB7 to the T3S-dependent delivery of the transcription activatorlike (TAL) effector protein AvrBs3, which activates expression of the resistance gene Bs3 in AvrBs3-responsive ECW-30R pepper plants (Boch et al., 2009;Römer et al., 2009). Furthermore, we used the Nterminal deletion derivative AvrBs3Δ2, which lacks a native export signal, as a reporter to monitor the translocation of the effector proteins XopC and XopJ. When AvrBs3 and AvrBs3Δ2 fusion proteins were analysed in strain 85*, which naturally lacks the avrBs3 gene, they induced the AvrBs3-specific HR in ECW-30R pepper plants as expected (Minsavage et al., 1990;Noël et al., 2003;Römer et al., 2009;Szurek et al., 2002). No HR induction was observed with strain 85*ΔhrpB7, suggesting that translocation depends on HrpB7 ( Figure 3a). We also analysed the influence of HrpB7 on the export of the early T3S substrate HrpB2, which is translocated in the absence of the control proteins HpaA, HpaB, and HpaC (Lorenz & Büttner, 2011;Rossier et al., 2000;Scheibner et al., 2016;.  XopJ 1-155 -AvrBs3∆2

N. benth. gfp
HrpB7 is essential for T3S-dependent protein delivery. (A) HrpB7 is essential for translocation of early and late T3S substrates. Strains 85* and 85*ΔhrpB7 (ΔhrpB7) with expression constructs encoding AvrBs3 or the AvrBs3Δ2 fusion proteins XopC 1-200 -AvrBs3Δ2 and XopJ 1-155 -AvrBs3Δ2 as indicated were infiltrated at a density of 4 × 10 8 CFU ml −1 into leaves of AvrBs3-responsive Early Cal Wonder-30R pepper plants. For the better visualisation of the hypersentitive response, leaves were bleached in ethanol 2 days post inoculation. For the analysis of the early T3S substrate HrpB2, strains 85*ΔhpaABC (ΔABC) and 85*ΔhpaABCΔhrpB7 (ΔABCΔhrpB7) containing HrpB2 1-40 -dTALE-2ΔN were inoculated at a density of 4 × 10 8 CFU ml −1 into leaves of gfp-transgenic Nicotiana benthamiana plants. Leaves were photographed 7 days post inoculation. Dashed lines indicate the inoculated areas. AvrBs3 and fusion proteins were stably synthesised in all strains (data not shown). (b) HrpB7 is essential for T3S and is itself not secreted by the T3SS. Strains 85* (wt) and 85*ΔhrpB7 (ΔhrpB7) with (+) or without (−) expression constructs encoding HrpB7 or AvrBs3 as indicated were incubated in secretion medium. TE and culture SN were analysed by immunoblotting using antibodies specific for HrpB7, the translocon protein HrpF, the effector AvrBs3, and the inner membrane ring protein HrcJ, respectively. The upper protein detected by the AvrBs3-specific antibody corresponds to full-length AvrBs3, lower signals likely result from the detection of degradation products. TE, total cell extracts; SN, supernatants To study the influence of HrpB7 on in vitro T3S, bacteria were incubated in secretion medium, and total cell extracts and culture supernatants were analysed by immunoblotting. The translocon protein HrpF and the effector protein AvrBs3 (ectopically expressed from a plasmid) were detected in the culture supernatant of strain 85* but not of strain 85*ΔhrpB7 (Figure 3b). In contrast, the predicted IM ring protein HrcJ, which was analysed as lysis control, was only detected in total cell extracts as expexted (Figure 3b). We also investigated a possible secretion of HrpB7 in strains 85* and 85*ΔhrpB7 containing hrpB7 or hrpB7-c-myc expression constructs. HrpB7 and HrpB7-c-Myc were detected in total cell extracts but not in the culture supernatants, suggesting that HrpB7 is not secreted under the conditions tested (Figures 3b and S3).

| HrpB7 forms stable protein complexes and localises to bacterial membranes
As mentioned above, HrpB7 presumably forms complexes corresponding to mono-or heterooligomers in the presence of SDS and Similar results were obtained when bacteria were incubated under T3S nonpermissive conditions ( Figure S4).
To further investigate the subcellular localisation of HrpB7, we generated N-and C-terminal fusions of HrpB7 to an N-terminal deletion derivative of the alkaline phosphatase PhoA (PhoA Δ2-120 ) from Escherichia coli, which is active when located in the periplasm and used as a reporter to analyse protein topology (Berger et al., 2010;Manoil & Beckwith, 1986). HrpB7-PhoA Δ2-120 and PhoA Δ2-120 -HrpB7 both complemented the hrpB7 mutant phenotype, suggesting that the PhoA Δ2-120 fusion partner did not interfere with HrpB7 function ( Figure 5a). When analysed in Xcv strain 85*ΔphoA, which lacks the native phoA gene (Hausner et al., 2017), no PhoA activity of both fusions was detectable, suggesting that the N-and C-termini of HrpB7 are not located in the periplasm (Figure 5b). Given the absence of predicted transmembrane domains in HrpB7, we conclude that HrpB7 localises to the cytoplasmic side of the IM.

| Cysteine and leucine residues in HrpB7 contribute to protein function and complex formation
To analyse the contribution of single amino acids in HrpB7 Xcv to protein function, we exchanged one or several cysteine residues in the N-terminal (C-X 6 ) 5 motif against alanine and also introduced mutations leading to the substitution of leucine residues at positions 33 and 40 and/or 111 and 118 by serine or glycine (Table 1). The resulting HrpB7 derivatives were analysed as C-terminally c-Myc epitope-tagged derivatives for their ability to complement the hrpB7 mutant phenotype. Exchange of the first three cysteine residues of the (C-X 6 ) 5 motif or the central cysteine residue at position 90 did not significantly affect HrpB7 function, whereas HrpB7 derivatives with mutations in all five cysteine residues of the (C-X 6 ) 5 motif did not complement the hrpB7 mutant phenotype (Figure 6a; Table 1).

Immunoblot analyses of bacterial protein extracts showed that all
HrpB7 derivatives were stably synthesised and that formation of HrpB7-specific complexes was significantly reduced or not detectable for HrpB7 C1-5A -c-Myc and HrpB7 C1-6A -c-Myc, which was additionally mutated in C90 (Figure 6c; Table 1). When HrpB7 derivatives were analysed as untagged proteins, HrpB7 C1-5A and HrpB7 C1-6A partially complemented the phenotype of strain fusion proteins as indicated were incubated in secretion medium, and phosphatase activity was analysed as described in Section 4. Proteins were detected by immunoblotting as described in panel (a). PhoA Δ2-120 fusions are indicated by asterisks. The fusion between the periplasmic predicted lytic transglycosylase HpaH and PhoA Δ2-120 was analysed as positive control (Hausner et al., 2017). ECW, Early Cal Wonder 85*ΔhrpB7, suggesting that protein function was not completely abolished ( Figure S5; Table 1).
In contrast, HrpB7 derivatives in which the leucine residues in both heptad motifs were exchanged against glycine (HrpB7 L12G ) only partially complemented the hrpB7 mutant phenotype when analysed as untagged proteins ( Figure S6). No complementation was observed when HrpB7 L12G was analysed as C-terminally c-Myc epitope-tagged derivative ( Figure 6b; Table 1). This is presumably caused by a negative effect of the c-Myc epitope on protein function (see also Figure S2). Immunoblot analysis showed that complex formation was anti-c-Myc HrpB7-c-Myc

HrpB7-c-Myc HrpB7-c-Myc
HrpB7-c-Myc severely reduced for HrpB7 derivatives carrying substitutions in both leucine heptad motifs, whereas mutations in one motif did not or only slightly interfere with complex formation (Figure 6c). We conclude that both leucine heptad motifs as well as the N-terminal (C-X 6 ) 5 motif of HrpB7 are required for complex formation and contribute to protein function.
We also investigated whether the leucine heptad motifs contribute to the localisation of HrpB7. Fractionation studies with HrpB7 L12Gc-Myc in strain 85*ΔhrpB7 grown under T3S-permissive conditions revealed that HrpB7 L12G -c-Myc was predominantly detected in the soluble fraction (Figure 6d). This is in contrast to HrpB7-c-Myc, which was equally distributed in the soluble and membrane fraction (see above; Figure 4d) and suggests that the leucine heptad motifs promote the association of HrpB7 with the bacterial membranes.

| HrpB7 interacts with components of the T3SS
Next, we analysed whether HrpB7 interacts with components of the T3SS and performed in vitro glutathione S-transferase (GST) pulldown assays. GST and GST-HrpB7 were immobilised on glutathione sepharose and incubated with bacterial lysates containing Cterminally c-Myc epitope-tagged derivatives of potential interaction partners. When eluted proteins were analysed by immunoblotting, the ATPase HrcN, the IM ring protein HrcD, and the predicted C ring protein HrcQ were detected in the eluate of GST-HrpB7 but not of GST, suggesting that they interact with HrpB7 (Figure 7a,b). Similar results were obtained for a C-terminally c-Myc epitope-tagged derivative of the T3S chaperone HpaB (Figure 7a,b). We also investigated a possible contribution of the leucine heptad motifs of HrpB7 to proteinprotein interactions. The results of GST pull-down assays suggest that the leucine heptad motifs contribute to the interaction of HrpB7 with HrcQ, HrcD, and HpaB ( Figure 7b). This is in agreement with the predicted role of coiled-coil motifs in protein-protein interactions and supports the hypothesis that coiled-coil motifs contribute to HrpB7 function.

| HrpB7 is dispensable for the formation of the predicted C ring
Given the potential interaction of HrpB7 with HrcQ, we investigated whether HrpB7 is required for the assembly of the predicted C ring. For this, a fusion protein between HrcQ and the superfolder green fluorescent protein (sfGFP) was analysed in Xcv using a modular T3S gene cluster as described previously (Hausner et al., 2019).

| DISCUSSION
In this study, we identified HrpB7 as a novel complex-forming component of the T3SS from Xcv. Previous transposon mutagenesis approaches already indicated an essential pathogenicity function of HrpB7 Fenselau & Bonas, 1995). This was con- kDa F I G U R E 7 HrpB7 interacts with T3S-associated proteins and is dispensable for cytoplasmic ring formation. (a) In vitro interaction studies with HrpB7. GST and GST-HrpB7 were immobilised on glutathione sepharose and incubated with bacterial lysates containing C-terminally c-Myc epitope-tagged derivatives of the ATPase HrcN, the inner membrane ring protein HrcD, the predicted cytoplasmic ring protein HrcQ, and the T3S chaperone HpaB. TE and eluates were analysed by immunoblotting using GST-and c-Myc epitope-specific antibodies as indicated. One representative blot for the detection of GST and GST-HrpB7 is shown. Experiments were performed at least three times with similar results. (b) The leucine heptad motifs of HrpB7 contribute to the interaction with HrcD, HrcQ, and HpaB. GST, GST-HrpB7, and GST-HrpB7 L12G were immobilised on glutathione sepharose and incubated with bacterial lysates containing C-terminally c-Myc epitope-tagged derivatives of HrcD, HrcQ, and HpaB. TE and eluates were analysed as described in panel (a). L12G, HrpB7 L33,40,111,118G . (c) HrpB7 is dispensable for foci formation by HrcQ-sfGFP. Fluorescent foci were analysed in strain 85*Δhrp_fsHAGX carrying modular level P hrp-HAGX expression constructs containing hrcQ-sfgfp and deletions in hrcQ (ΔhrcQ), in the hrpA to hpaB operons (ΔhrpA-hpaB), as well as in hrcQ and hrpB7 (ΔhrcQΔhrpB7) as indicated. Bacteria were grown in secretion medium, and foci formation was analysed by fluorescence microscopy. The experiment was performed three times with similar results. One representative image for each strain from one experiment is shown. The size bar corresponds to 2 μm. (d) Statistic analysis of HrcQ-sfGFP foci formation. Fluorescent foci were counted in approximately 300 cells per strain in three different transconjugants, and the mean values and standard deviations are shown as percentage of bacterial cells. Asterisks indicate a significant difference between the ΔhrpA-hpaB and the hrcQ single-deletion mutant strain or the ΔhrcQΔhrpB7 and the hrcQ single-deletion mutant strain as indicated by brackets with a p value < .05 based on the results of a χ 2 test. ECW, Early Cal Wonder; GST, glutathione S-transferase; TE, total cell extracts motifs, which are part of the predicted coiled-coil structures. HrpB7 is the only hrp gene product from Xcv with predicted N-and Cterminal coiled coils. This protein motif usually consists of two or more α helices winding around each other and is often involved in protein-protein interactions (Delahay & Frankel, 2002;Woolfson, et al., 2012). In agreement with the presence of predicted coiled coils in HrpB7, our in vitro pull-down assays suggest that HrpB7 interacts with T3SS components including the ATPase HrcN, the predicted C ring protein HrcQ, and the IM ring component HrcD ( Figure 7). Furthermore, HrpB7 forms stable complexes, probably corresponding to dimers and homo-or heterooligomers, suggesting that it is part of an oligomeric substructure of the T3SS (Figure 4).
Mutations in both N-and C-terminal leucine heptad motifs severely reduced complex formation, thus confirming the predicted contribution of the putative coiled-coil structures to the protein-protein interactions ( Figure 6). Given that HrpB7 lacks predicted transmembrane helices and is not secreted and presumably does not localise to the bacterial periplasm, we assume that it associates on the cytoplasmic side with the ATPase complex and the ring structures of the T3SS (Figures 4 and 5).
Notably, complex formation and interaction with the ATPase complex, the C ring and components of the export apparatus were previously reported for T3SS-associated proteins from plant and animal pathogens, which are encoded downstream of the T3S-ATPase genes similarly to HrpB7 (Cherradi et al., 2014;Evans et al., 2006;Fraser et al., 2003;Gazi et al., 2008;Romo-Castillo et al., 2014).
Despite a lack of general sequence conservation, these proteins contain predicted coiled coils and are referred to as HrpO/FliJ/YscO protein family (Gazi et al., 2008;Gazi et al., 2009). In addition to HrpB7, the only other characterised member of this protein family from plant pathogenic bacteria is HrpO from P. syringae. HrpO was described as α helical with characteristics of intrinsically disordered proteins and self-associates as was also observed for HrpB7. Furthermore, HrpO interacts with the predicted regulator of the T3S-ATPase, HrpE (Gazi et al., 2008;Uversky, 2013). An interaction with the ATPase complex was also reported for the HrpO/FliJ/YscO family members FliJ from the flagellar T3SS, Spa13 from Shigella flexneri, and EscO (formerly known as Orf15 or EscA) from enteropathogenic E. coli (Cherradi et al., 2014;Evans et al., 2006;Romo-Castillo et al., 2014). It remains to be investigated whether also HrpB7 from Xcv contributes to the assembly of the ATPase complex and to the regulation of its activity as was reported for FliJ and EscO (Ibuki et al., 2011;Majewski et al., 2019). In preliminary in vitro experiments, however, we did not detect a positive influence of HrpB7 on the activity of the purified ATPase HrcN (J. Hausner and D. Büttner, unpublished data).
Furthermore, our fluorescence microscopy studies showed that the predicted C ring component HrcQ assembles independently of HrpB7 ( Figure 7). Given that a failure in the assembly of the ATPase complex likely results in a loss of C ring formation as shown in Yersinia spp. (Diepold et al., 2010), it is possible that HrpB7 is dispensable for the formation of the ATPase complex. Notably, a similar finding was reported for the HrpO/FliJ/YscO family member YscO from Yersinia spp., suggesting that a role in the assembly of the ATPase does not appear to be a general characteristic function of this protein family (Diepold et al., 2012).
HrpB7 does not only interact with components of the T3SS but also with the T3S chaperone HpaB as revealed by our interaction studies ( Figure 7). Notably, an interaction with T3S chaperones was also described for FliJ, Spa13, and InvI, suggesting that HrpO/FliJ/ YscO family members are involved in a network of interactions with T3SS components and chaperones (Evans et al., 2006;Evans & Hughes, 2009). In Xcv, the T3S chaperone HpaB is essential for the efficient secretion and translocation of multiple effector proteins and was previously identified as interaction partner of T3SS components including the ATPase, the C ring, and the cytoplasmic domains of components of the export apparatus (Büttner et al., 2004;Büttner et al., 2006;Lonjon et al., 2017;Lorenz & Büttner, 2009;Lorenz et al., 2012;Prochaska et al., 2018). The presence of multiple binding sites for HpaB in the T3SS suggests that HrpB7 does not play an exclusive role in the recruitment of chaperone complexes but might rather be of general importance for T3S. This hypothesis is supported by the finding that HrpB7 is not only essential for the delivery of HpaB-dependent effector proteins but also for the translocation of the HpaB-independent early T3S substrate HrpB2 ( Figure 3).

Given the variety of different interaction partners identified for
HrpO/FliJ/YscO family members, it seems likely that the observed protein-protein interactions are stabilised by the coiled-coil motifs.
We, therefore, analysed the predicted contribution of the putative coiled-coil motifs in HrpB7 to protein-protein interactions and protein function. Characteristic features of coiled coils are repetitive heptad motifs (abcdefg) n with hydrophobic amino acids at positions "a" and "d" (Gazi et al., 2009). The presence of unbranched amino acids such as leucine at position "a" as is the case in HrpB7 (Figure 1) might favour the formation of four-stranded parallel coiled coils (Lupas & Gruber, 2005). When we exchanged the leucine residues in the heptad motifs of HrpB7 against glycine, complex formation was severely impaired ( Figure 6). Glycine is a known helix breaker and, therefore, likely affects the helical structures within the predicted coiled coils (Serrano et al., 1992). HrpB7 derivatives with leucine-to-glycine exchanges in both predicted coiled-coil regions did not complement the hrpB7 mutant phenotype and only weakly interacted with HrcQ, HrcD, and HpaB, suggesting an important role of these amino acids in HrpB7 function and protein-protein interactions (Figures 6 and 7).
The exchange of leucine residues in N-and C-terminal heptad motifs against the polar amino acid serine also abolished detectable complex formation ( Figure 6). However, HrpB7 derivatives with leucine-toserine exchanges were partially functional, indicating that complex formation contributes to but is presumably not crucial for HrpB7 function. The observed negative impact of the mutations was strongly reduced when amino acid exchanges were only introduced in either the N-or the C-terminal heptad motif ( Figure 6).
We also investigated the role of the five regularly spaced cysteine residues at position "d" of each N-terminal heptad in HrpB7.
The exchange of all five cysteine residues in the N-terminal (C-X 6 ) 5 motif against alanine significantly reduced complex formation by HrpB7 ( Figure 6). This is in agreement with the hypothesis that cysteine residues stabilise the coiled-coil structure by the formation of disulfide bonds and thus presumably contribute to the assembly of intermolecular or intramolecular complexes (Shen et al., 2005;Zhou et al., 1993). Given that alanine still fits the criteria for position "d" of the leucine heptad motif, the C-to-A exchange likely did not completely prevent coiled-coil formation.
In line with this hypothesis is our finding that the C-to-A mutations impaired but did not abolish HrpB7 function ( Figure 6). We conclude that HrpB7 function and complex formation depend on the leucine and cysteine residues in the N-and C-terminal heptad motifs but that single amino acids can be mutated without causing significant detrimental effects. This is in line with the finding that some of the cysteine and leucine residues in the N-and C-terminal heptad motifs are not conserved in HrpB7 Xcc , which is functional in Xcv.
Taken together, our study provides the first detailed characterisation of a member of the FliJ/HrpO/YscO family from a Hrp2-T3SS.
The experimental results suggest that HrpB7 from Xcv forms an essential oligomeric structure of the T3SS and interacts via coiled coils with the ATPase complex as well as with IM and cytoplasmic rings of the T3SS. In future experiments, we will investigate whether HrpB7 contributes to the assembly of the ATPase complex or additional substructures of the T3SS.

| Bacterial strains and growth conditions
Bacterial strains and plasmids are listed in Table S1. E. coli strains were cultivated at 37 C in lysogeny broth medium. Xcv strains were grown at 30 C in nutrient yeast extract glycerol medium (Daniels et al., 1984). For T3S assays, we used minimal medium A (MA) supplemented with sucrose (10 mM) and casamino acids (0.3%; Ausubel et al., 1996).

| Plant material and plant infections
For infection studies, bacterial solutions were infiltrated into leaves of the near-isogenic pepper cultivars ECW and ECW-10R with a needle-less syringe (Kousik & Ritchie, 1998;Minsavage et al., 1990).
Plants were inoculated with 1 × 10 8 colony-forming units ( In planta growth curves were performed three times as described . To monitor translocation of dTALE-2, we used gfp-transgenic N. benthamiana plants . Infected N. benthamiana plants were incubated for 16 hr of light at 20 C and 8 hr of darkness at 18 C. GFP fluorescence was documented 4 dpi.

| Generation of Xcv hrpB7 deletion mutant strains
To generate Xcv hrpB7 deletion mutants, DNA fragments flanking hrpB7 and including the first and last 18 nucleotides of the coding sequence were amplified by polymerase chain reaction (PCR) and cloned into the Golden Gate-compatible suicide vector pOGG2 using BsaI and ligase. The resulting construct was transformed into E. coli DH5λpir and transferred into strains 85-10 and 85* by triparental conjugation. Transconjugants were selected as described previously (Huguet, et al., 1998). Double crossovers resulted in hrpB7 deletion mutant strains.

| Generation of expression constructs
To generate a hrpB7 expression construct, hrpB7 was amplified by PCR from Xcv strain 85-10 and cloned into the BsaI sites of vector pBRM, downstream of a lac promoter and in frame with a Cterminal 3 × c-Myc epitope-encoding sequence. Similarly, hrpB7 was cloned into vector pBRM-Stop, which contains a stop codon upstream of the 3 × c-Myc epitope-encoding sequence, resulting in the synthesis of untagged HrpB7. We also generated a pBRM expression construct encoding HrpB7 from Xcv using primers specific for hrpB7 Xcc from strain LMG 568. For the generation of a GST-HrpB7 expression construct, two modules corresponding to hrpB7 and ptac-gst were cloned into vector pBRM-P, resulting in construct pB-P-ptacGST-hrpB7. ptac-gst was amplified by PCR from vector pGEX-2TKM. All PCR amplicons were first subcloned into vector pUC57ΔBsaI or alternatively into vector pICH41021 using SmaI and ligase.
For the exchange of cysteine residues against alanine, 5 0 and 3 0 regions of hrpB7 were amplified by PCR from plasmid pBhrpB7*, in which the internal BsaI site in hrpB7 had been mutated by PCR. (HrpB7 L33,40,111,118G ). hrpB7 L12G was amplified by PCR from pBStophrpB7 L12G and cloned together with ptac-gst into vector pBRM-P, thus generating an expression construct that encodes GST-HrpB7 L12G . In addition to the generation of hrpB7 expression constructs, we amplified hrcD and cloned the corresponding PCR product in vector pBRM. Generated constructs and primers used in this study are listed in Tables S1 and S2, respectively.
For the construction of PhoA Δ2-120 expression constructs, the phoA Δ2-120 gene fragment from E. coli (GenBank accession number M29668) including an additional linker-encoding sequence (amino acids LSLIHISWPMGPG) was amplified using nested forward primers containing the linker-encoding sequence and a phoA-specific reverse primer (see Table S1; Berger et al., 2010). The linker-phoA Δ2-120 gene fragment was subcloned into vector pICH41021 using SmaI and ligase and subsequently ligated with hrpB7 into vector pBRM, resulting in pBhrpB7-linker-phoA Δ2-120 . To generate constructs encoding PhoA Δ2-120 as N-terminal fusion partner of HrpB7, the linker-phoA Δ2-120 gene fragment was amplified from construct pICH41021-linker-phoA Δ2-120 using primers that allow cloning of the linker-phoA Δ2-120 fragment into the 5 0 cloning site 4.5 | Generation of a modular T3S gene cluster construct containing hrcQ-sfgfp and deletions in hrcQ and hrpB7 To introduce a deletion into hrpB7 in the modular T3S gene cluster, hrpB7 was amplified by PCR with primers FShrpB7_F and FShrpB7_R using the level −2 construct pAGB194, which contains hrpB7, as template (Hausner et al., 2019). The resulting PCR amplicon corresponding to hrpB7 with a deletion of bp 15 to 429 was cloned into the pUC19 derivative pAGM9121 (Addgene #51833) using BpiI and ligase, thus generating the level −2 construct pAGB772. In a subsequent Golden Gate reaction, the insert of pAGB772 was assembled with modules from additional level −2 constructs pAGB192 (contains hrcL), pAGB193 (contains hrcN), pAGB195 (contains hrcT), and pAGB196 (contains hrcC) in Level −1 vector pAGM1311 (Addgene #47983) using BsaI and ligase (Hausner et al., 2019). The assembly of the resulting pAGB778 construct with the additional level −1 construct pAGB197, which contains hrpB1 to hrpB4, using BpiI and ligase led to Level 0 construct pAGB784 (corresponding to the hrpA and hrpB operons), which was subsequently cloned into level 1 vector pICH47811 using BsaI and ligase (Addgene #48008), thus generating level 1 construct pAGB790. The assembly by BpiI and ligase with additional level 1 constructs pAGB275 (contains hrpC-hpaB operons with a deletion in hrcQ) and pAGB156 (contains hrpF) led to the generation of level M construct pAGB796, which includes the hrp gene cluster with deletions in hrpB7 and hrcQ (Hausner et al., 2019). Construct pAGB796 was assembled with level M construct pAGB322 (contains xopA, hpaH, hrpX, hrpG*, and hrcQ-sfgfp; Hausner et al., 2019) and end-linker pICH79264 in level P vector pICH75322 using BsaI and ligase, thus resulting in the final level P construct pAGB802.

| Analysis of in vitro T3S
In vitro T3S assays were performed as described previously . Briefly, bacteria were grown overnight in MA medium (pH 7.0) supplemented with sucrose (10 mM) and casamino acids (0.3%) and shifted to MA medium (pH 5.3) containing 50-μg ml −1 bovine serum albumin (BSA) and 10-μM thiamine at an optical density (OD 600 nm ) of 0.15. Cultures were incubated on a rotary shaker for 1.5 hr at 30 C, and bacterial cells and secreted proteins were separated by filtration. Proteins in 2 ml of the culture supernatants were precipitated with trichloroacetic acid and resuspended in 20 μl of Laemmli buffer. Total cell extracts and culture supernatants were analysed by SDS-PAGE and immunoblotting, using antibodies directed against the c-Myc epitope as well as against HrpB7, HrpF, HrcJ, and AvrBs3, respectively (Bonas et al., unpublished;Knoop et al., 1991;Rossier et al., 2000). Horseradish peroxidaselabelled anti-mouse and anti-goat antibodies were used as secondary antibodies, and binding of antibodies was visualised by enhanced chemiluminescence.

| GST pull-down assays
For GST pull-down assays, E. coli BL21(DE3) cells containing the expression constructs for the synthesis of GST, GST-HrpB7, GST-HrpB7 L12G , and C-terminally c-Myc epitope-tagged potential interaction partners were grown in lysogeny broth medium until an OD 600 nm of 0.6-0.8. Gene expression, which was driven in all cases by the lac promoter, was induced in the presence of IPTG for 2 hr at 37 C, and after centrifugation bacterial cells were broken with a French press. Cell debris was removed by centrifugation, and soluble GST and GST fusion proteins were immobilised on a glutathione sepharose matrix according to the manufacturer's instructions (GE Healthcare). After washing of the matrix, immobilised GST and GST-HrpB7 were incubated with bacterial lysates containing the predicted interaction partner for 2 hr at 4 C on an overhead shaker. Unbound proteins were removed by washing, and bound proteins were eluted with Laemmli buffer. Cell lysates and eluted proteins were analysed by SDS-PAGE and immunoblotting, using c-Myc epitope-and GST-specific antibodies.

| Fractionation experiments with Xcv strains
For the subcellular localisation of HrpB7, bacteria were grown over- Laemmli buffer and analysed by immunoblotting as described above.

| Phosphatase assays
For the analysis of the PhoA activity, bacteria were resuspended in secretion medium at an OD 600nm of 0.8 and incubated in the presence of 90-μg/ml X-Phos (5-Bromo-4-chloro-3-indolyl phosphate) on a horizontal shaker for 3 hr at 30 C. The colour change was photographed, and bacterial cell extracts were analysed by immunoblotting using a PhoA-specific antibody (Sigma-Aldrich). The phosphatase assay was performed twice with similar results.

| Fluorescence microscopy
For the analysis of HrcQ-sfGFP, bacteria were grown overnight in MA medium (pH 7.0) supplemented with sucrose (10 mM) and casamino acids (0.3%). Cells were then resuspended at an OD 600 nm of 0.15 in MA medium (pH 5.3) supplemented with BSA and thiamine as described above and incubated on a tube rotator at 30 C for 1 hr. Cell suspensions were pipetted on a microscopy slide on top of a pad of 1% agarose dissolved in MA medium (pH 5.3) containing BSA and thiamine as described previously (Hausner et al., 2019). GFP fluorescence was inspected with a confocal laser scanning microscope (Zeiss LSM 780 AxioObserver. Z1) using filter sets for sfGFP (excitation at 485 nm; emission at 510 nm). Experiments were performed with different transconjugants for each strain and repeated twice with similar results.