Bacterial strains and plasmids used in this study are listed in Table S1 . Escherichia coli strains were grown at 37°C in lysogeny broth (LB) medium, and X. euvesicatoria at 30°C in nutrient-yeast extract-glycerol (NYG) medium. Antibiotics were added to the media at the following final concentrations: ampicillin, 100 μg/ml; kanamycin, 25 μg/ml; rifampicin, 100 μg/ml; spectinomycin, 100 μg/ml; streptomycin, 2.5 µg/ml and gentamicin, 15 μg/ml. Plasmids were introduced into E. coli by transformation and into X. euvesicatoria by electroporation or triparental conjugation, using pRK2013 as helper plasmid.

For infection assays, X. euvesicatoria strains were resuspended in 1 mM MgCl₂ at an optical density (OD_600nm) of 0.1 which corresponds to 1×10⁸ colony-forming units (CFU) ml⁻¹. Bacterial suspensions were infiltrated into leaves of the pepper cultivar Early Cal Wonder (ECW) using a needleless syringe. Infected plants were incubated in growth chambers for 16 hours of light at 28°C and 8 hours of darkness at 22°C. Disease symptoms were photographed seven days post inoculation (dpi). In planta growth curves were performed in three ECW plants as described (Bonas et al., 1991). To monitor translocation of dTALE-2, X. euvesicatoria strains were resuspended in 1 mM MgCl₂ at a density of 4×10⁸ CFU ml⁻¹ and infiltrated into leaves of gfp-transgenic Nicotiana benthamiana plants (Werner et al., 2011). Infected N. benthamiana plants were incubated for 16 hours of light at 20°C and 8 hours of darkness at 18°C. GFP fluorescence was documented four dpi using a chemiluminescence/fluorescence imager (Vilber Fusion FX Edge). Infection experiments were performed at least three times with different transconjugants; representative results are shown.

To generate X. euvesicatoria T4S deletion mutants, DNA fragments flanking individual (virB4, virD4, virG, virA and icmE) or multiple (virB11virB1, virB6virB8virB9virB10 and dotCB) T4S genes were amplified by PCR and cloned into the Golden Gate-compatible suicide vector pOGG2 using the type IIs restriction enzyme BsaI and ligase (Engler et al., 2008). The resulting constructs were transferred into strain 85-10 by triparental conjugation. Transconjugants were selected as described previously and double crossovers resulted in T4S deletion mutant strains (Huguet et al., 1998). Multiple deletions of T4S genes or gene regions resulted from the successive introduction of corresponding deletion constructs into X. euvesicatoria strains.

To insert the gentamicin resistance gene into the genome of X. euvesicatoria strain 85-10, the gene was amplified by PCR from plasmid pBRM and cloned into the Golden Gate-compatible suicide vector pLAND downstream of the lac promoter using BsaI and ligase. pLAND allows the insertion of genes into the hpaFG region adjacent to the T3S gene cluster (Lorenz et al., 2012). The corresponding construct pLAND-gent^R was introduced into strain 85-10 by conjugation and homologous recombination resulted in strain 85-10::gent^R as described previously (Huguet et al., 1998; Lorenz et al., 2012).

To insert a kanamycin resistance gene and gfp (green fluorescent protein) into X. euvesicatoria, DNA fragments flanking a non-coding region at nucleotide position 2831 on plasmid pXCV183 were amplified by PCR from strain 85-10. Similarly, the promoter of the gentamicin resistance gene on plasmid pBRM, the kanamycin resistance gene from plasmid pKT25 and the gfp gene including the lac promoter were amplified by PCR using plasmid pBRM-sfgfp as template. Flanking BsaI sites with matching overhangs were introduced by the primer sequences. The corresponding amplicons were first subcloned in vector pICH41021 using the blunt-end cutter SmaI and ligase in a cut-ligation. All five amplicons were subsequently assembled in the Golden Gate-compatible suicide vector pOGG2 using BsaI and ligase. The resulting construct pOGG2-kan^Rgfp(pXCV183) was transferred into strain 85-10 by triparental conjugation. Double crossovers resulted in X. euvesicatoria strain 85-10 containing the kanamycin resistance gene and gfp on plasmid pXCV183.

For the generation of XCV0332, XCV1120 and XCV3751 expression constructs, genes were amplified by PCR from X. euvesicatoria strain 85-10 and individually cloned into the Golden Gate-compatible expression vector pBRNM using BsaI in a single restriction/ligation reaction. Vector pBRNM allows the expression of genes under control of a lac promoter in frame with an N-terminal 3 × c-Myc epitope-encoding sequence. To generate the traI expression construct, the promoter of XCV4361, the 3 × c-Myc epitope-encoding sequence and traI were amplified by PCR. Primer sequences introduced BsaI restriction sites with matching overhangs. When compared with the lac promoter, the promoter upstream of XCV4361 results in lower expression levels of downstream genes as was previously shown for the predicted xylanase gene XCV4360 (Solé et al., 2015). All three DNA fragments were assembled by Golden Gate cloning in vector pBRM-P which contains an integrated stop codon downstream of the insert.

For in cis expression of virB4 in X. euvesicatoria, virB4 was amplified by PCR and cloned into the Golden Gate-compatible suicide vector pLAND downstream of the lac promoter and in frame with a C-terminal 3 x c-Myc epitope-encoding sequence using BsaI and ligase. To insert virB4-c-myc into the genome of X. euvesicatoria, pLAND-virB4 was transferred into strain 85-10ΔvirB4. Double homologous recombination events led to the insertion of virB4-c-myc into the hpaFG region. For the in cis expression of icmE, icmE and the promoter of XCV0160 were amplified by PCR from strain 85-10 and assembled in the Golden Gate-compatible suicide vector pLAND-P in frame with a C-terminal 3 × c-Myc epitope-encoding sequence using BsaI and ligase. The promoter of XCV0160 was previously shown to mediate gene expression in X. euvesicatoria and presumably results in lower expression levels than the constitutive lac promoter (Scheibner and Büttner, unpublished data). The resulting construct pLAND-P-icmE was transferred into strain 85-10ΔicmE and double homologous recombination events led to the insertion of icmE-c-myc into the hpaFG region.

For the generation of promoter-reporter constructs, upstream regions of virB2, virB5, virD4, virG, dotA, dotD, icmL, hrpB1 as well as 129 bp downstream of the annotated start codon of icmL were amplified by PCR. Each amplicon was ligated with the sfgfp reporter gene in the Golden Gate-compatible vector pBRM-P+T, which contains a terminator upstream of the cloning site. The resulting constructs were transferred into X. euvesicatoria strains 85-10 and 85-10ΔvirG. vir and icm/dot promoter regions were additionally cloned upstream of the dTALE-2 (designer transcription activator-like effector 2) gene in a Golden Gate-compatible expression vector using the modular cloning (MoClo) system (Weber et al., 2011). The MoClo system is based on different cloning vectors (designated level -2, -1, 0, 1 and 2 vectors) which allow the stepwise assembly of gene and promoter modules into multigene constructs by the alternate use of BsaI and BpiI restriction enzymes (Weber et al., 2011). vir and icm/dot promoter regions were first cloned into vector pAGM9121 using BpiI and ligase, thus resulting in level 0 constructs, and were subsequently transferred to vector pICH47732 using BsaI and ligase to create level 1 promoter constructs. For the assembly of dTALE-2, modules encoding the N-terminal region (pICH70781), the central repeats (pICH73079, pICH73081, pICH73093) and the C-terminal region of dTALE-2 (pICH72151) were assembled in the level 0 construct pICH73103. The resulting insert was subsequently cloned into vector pICH50251 using BsaI and ligase to create the level 1 dTALE-2 construct pICH79631. For the generation of dTALE-2 expression constructs containing vir or icm/dot promoters, modules with promoter fragments, the dTALE-2-encoding sequence (pICH79631), and a transcription terminator (pICH50122) were assembled in the level 2 vector pICH77739 using BpiI and ligase.

For bacterial adenylate cyclase-based two-hybrid (BACTH) assays, T4S genes were cloned into the Golden Gate-compatible vectors pUT18GG, pUT18CGG, pKT25GG and pKNT25GG using BsaI and ligase. Additionally, the putative T4S substrate genes XCV0332, traI, XCV3751 and XCV1120 were amplified by PCR from X. euvesicatoria strain 85-10, subcloned into pICH41021 as blunt-end fragments using SmaI and ligase, and the resulting inserts were cloned into the BsaI sites of vectors pUT18GG, pUT18CGG, pKT25GG and pKNT25GG by Golden Gate cloning. Primers and plasmids used in this study are listed in Tables S1, S2 .

For in vitro T4S assays, X. euvesicatoria strains were grown over night in NYG medium with antibiotics, resuspended in fresh NYG medium containing 50 μg ml⁻¹ bovine serum albumin (BSA) at a cell density of 2 × 10⁸ CFU/ml, and incubated on a rotary shaker at 30°C for 4 h. Bacterial cells and secreted proteins were separated by filtration as described previously (Rossier et al., 1999). 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 analyzed by SDS-PAGE and immunoblotting, using antibodies directed against the c-Myc epitope (Sigma Aldrich) and the β-subunit of the RNA polymerase (Invitrogen), respectively. Horseradish peroxidase-labelled anti-mouse and anti-rabbit antibodies were used as secondary antibodies. Binding of antibodies was visualized using a chemiluminescence imager (Vilber Fusion FX Edge). Results were reproduced at least two times.

BACTH assays were performed using the Euromedex BACTH system kit and Golden Gate-compatible expression vectors ( Table S1 ). To monitor protein synthesis, expression constructs encoding T25 and T18 fusion proteins were transformed into JM109 E. coli cells, and gene expression was induced with IPTG (isopropyl-β-D-thiogalactopyranoside; 2 mM final concentration) at an OD₆₀₀ of 0.6 - 0.8. After induction, cultures were incubated on a rotary shaker for 2 h at 37°C. Bacterial cells were collected by centrifugation, resuspended in Laemmli buffer, and analyzed by immunoblotting, using a FLAG epitope-specific antibody.

Protein-protein interaction studies were performed as described previously (Karimova et al., 2005; Otten and Büttner, 2021). Briefly, expression constructs encoding T18- and T25-fusion proteins were cotransformed into chemically competent BTH101 E. coli cells and transformants were plated on LB plates containing kanamycin and gentamicin. At least three colonies per transformant were cultivated over night at 30°C on a rotary shaker in liquid LB medium with appropriate antibiotics. Two microliters of each culture were spotted on LB plates containing kanamycin, gentamicin, X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranosid; 40 μg/ml) and 2 mM IPTG. Colonies were photographed over a period of three to 5 days. Every co-transformation was performed at least three times; representative results are shown.

Growth of E. coli during cocultivation with X. euvesicatoria was analyzed as described previously (Bayer-Santos et al., 2019). X. euvesicatoria and E. coli strains were grown overnight in NYG or LB medium, respectively, with appropriate antibiotics. Cells were harvested by centrifugation, resuspended at OD_600nm of 0.3 in fresh medium and grown to exponential phase for 4 h on a rotary shaker. The OD_600nm of all cultures was measured and adjusted to 1. Serial dilutions (1:2) of E. coli cultures were prepared in 96 well plates and equal volumes of E. coli and X. euvesicatoria cultures were mixed in each well. Five microliters of every suspension were spotted on LB agar plates containing 100 μM IPTG and 40 μg/mL X-gal. Plates were incubated for 24 h at 30˚C and photographed. The experiment was performed three times with similar results.

For quantitative growth competition assays, X. euvesicatoria and E. coli were cocultivated according to a previously described protocol (Souza et al., 2015). For this, X. euvesicatoria strains were grown over night on NYG agar plates containing rifampicin and E. coli strain Top 10 was grown over night in 5 ml LB liquid medium in the presence of streptinomycin. Cells were resuspended to an OD_600nm of 0.2 in liquid NYG or LB medium and cultures were incubated for five hours on a rotary shaker at 30°C for X. euvesicatoria or at 37°C for E. coli. After incubation, the OD_600nm of all cultures was measured and X. euvesicatoria was mixed with E. coli in triplicates in a final volume of 1 ml at a ratio of 100: 1 (X. euvesicatoria: E. coli) corresponding to a final OD_600nm for X. euvesicatoria of 0.05 and for E. coli of 0.0005. To measure bacterial growth at 0 hours, we prepared serial dilutions of 100 µl of the mixtures which were plated on LB plates containing streptinomycin for selection of Top 10 and on NYG plates containing rifampicin for selection of X. euvesicatoria. For the analysis of bacterial growth after cocultivation of X. euvesciatoria and E. coli, 100 µl of the mixtures were spotted on LB agar plates and cultivated at 30°C. After 38 hours, bacteria were resuspended in 1 ml of NYG medium and all mixtures were adjusted to the same OD_600nm. Serial dilutions were subsequently plated on LB plates containing streptinomycin for selection of Top 10 and on NYG plates containing rifampicin for selection of X. euvesicatoria and the final X. euvesicatoria – E. coli ratios were calculated after counting the colonies.

Plasmid transfer between X. euvesicatoria strains was analyzed according to a protocol by Stall et al. (1986). For this, selected X. euvesicatoria strains were grown over night on NYG agar plates containing appropriate antibiotics. Cells were resuspended at an OD_600nm of 0.3 in liquid NYG medium without antibiotics. After six hours of incubation at 30°C on a rotary shaker, donor and recipient strains were mixed in 1 ml of NYG medium at a final OD_600nm of 0.4 for each strain. Bacterial suspensions were subsequently dropped on a cellulose nitrate filter (NC20, membrane filters, 0.2 µm, diameter 50 mm, Whatman) which was transferred onto an NYG agar plate containing rifampicin. Bacteria were covered with 1 ml 1% water agar and incubated over night at 30°C. Bacterial conjugation mixtures were then transferred from the cellulose nitrate filters to 1 ml of NYG medium using a spatula and adjusted to an OD_600nm of 0.05. Serial dilutions were plated on NYG plates (1.5% agar) with appropriate antibiotics. The plates were incubated for two days at 30°C and the transconjugants were counted. The number of transconjugants obtained with the wild-type strain as donor was set to 100%.

For the isolation of native plasmids, X. euvesicatoria strains were resuspended at an OD_600nm of 0.05 in 50 ml NYG medium containing rifampicin and grown over night at 30°C. Plasmids were isolated using the GeneJET plasmid midiprep kit (Thermo Scientific) according to the manufacturer’s instructions. Twenty microliters of each plasmid preparation were loaded on an 0.7% agarose gel containing 5% ethidium bromide. The electrophoresis was performed in 1 x TBE (Tris-borate-EDTA [ethylene diamine tetraacetic acid]) buffer at 120 mA for 25 min and DNA was visualized using an UV imager. The agarose gel electrophoresis was performed three times with similar results.

Previous genome sequence analyses identified predicted virB/virD4 and icm/dot T4S genes on the chromosome and on plasmids pXCV38 and pXCV183 of X. euvesicatoria strain 85-10 (Thieme et al., 2005). The chromosomal T4S gene cluster contains virB7, virB8, virB9, virD4 and four copies of virB6 ( Figure 1A ). virB7, virB8, virB9 and virD4 genes encode predicted core components of the T4S system which are homologous (76 – 100% amino acid identity) to corresponding proteins from X. axonopodis pv. citri ( Figure 1A ; Table S3 ). VirB6 proteins are less conserved (40 – 42% amino acid identity) ( Figure 1A ). In contrast to the chromosomally encoded VirB/VirD4 T4S system from X. axonopodis pv. citri, which was previously shown to deliver toxins into bacterial cells (Souza et al., 2015), the chromosomal vir gene cluster from X. euvesicatoria is incomplete because it lacks virB1 to virB5 as well as virB10 and virB11 ( Figure 1A ). Chromosomal vir genes of X. euvesicatoria, however, include virA and virG ( Figure 1A ), which encode putative regulators with homology to the corresponding sensor and response regulator of a VirA/VirG two-component system from A. tumefaciens (Gelvin, 2003; Wise and Binns, 2015). VirA from A. tumefaciens is autophosphorylated and subsequently phosphorylates VirG which binds to vir box elements (consensus [5’-TG(A/T)AA(C/T)-3’]) in promoters of target genes (Winans et al., 1987; Jin et al., 1990; Tamamoto et al., 1990; Winans, 1990). Notably, predicted vir boxes are located upstream of several vir and icm/dot genes from X. euvesicatoria ( Table 1 ).

In addition to chromosomal vir genes, X. euvesicatoria contains a second virB T4S gene cluster with virB1 – virB11 genes on plasmid pXCV38 ( Figure 1B ). The region adjacent to virB1 encodes predicted proteins with homology to TraI and TraG, which are involved in DNA transfer by T4S systems (Audette et al., 2007; Waksman, 2019). TraI belongs to the group of relaxases, which cut DNA prior to transfer at the origin of transfer (oriT) and remain attached to the DNA during T4S-dependent transport (Waksman, 2019; Costa et al., 2021). The plasmid-encoded Vir proteins from X. euvesicatoria share an overall amino acid identity between 32 – 98% with corresponding proteins encoded by the virB gene cluster located on plasmid pXAC64 of X. axonopodis pv. citri strain 306 ( Figure 1 ). In contrast, the amino acid identity between chromosomally and plasmid-encoded Vir proteins is significantly lower in both X. euvesicatoria and X. axonopodis pv. citri and ranges between 20 – 50% ( Figure 1 ). In both pathogens, VirD4 is encoded as a single copy gene on the chromosome.

Notably, X. euvesicatoria also contains a predicted icm/dot T4S gene cluster on plasmid pXCV183 (Thieme et al., 2005) ( Figure 2A ). The corresponding gene products are homologous to major structural components of Icm/Dot T4S systems and share between 22 – 46% amino acid identity with respective Icm/Dot proteins from L. pneumophila ( Figure 2A , Table S5 ). The organization of the 16 putative icm/dot genes from X. euvesicatoria including icmL, K, E, G, C, D, J as well as dotD, C, B, O and L is similar to the order of the corresponding genes in L. pneumophila ( Figure 2B ). However, genes encoding predicted IcmF, H, N, M, Q, R, S, V and X proteins are missing in X. euvesicatoria.

To investigate whether vir and icm/dot T4S genes from X. euvesicatoria are expressed, we generated reporter constructs containing sfgfp (superfolder green fluorescent protein) downstream of predicted promoters of virB2, virB5, virD4, virG, dotA, dotD and icmL (see above, Table 1 ). Constructs were transferred into X. euvesicatoria strains 85-10 and 85-10ΔvirG, which lacks the predicted regulatory gene virG. The lac promoter, which activates gene expression in X. euvesicatoria, served as positive control (Szczesny et al., 2010). As negative control, we cloned sfgfp downstream of the hrpB1 promoter, which is inactive in NYG medium (Wengelnik and Bonas, 1996). hrpB1 is a gene of the T3S gene cluster and is only activated in planta or when bacteria are cultivated in special minimal media (Wengelnik and Bonas, 1996). For the analysis of promoter activities, bacteria were cultivated in NYG medium and bacterial cell extracts were analyzed by immunoblotting using a GFP-specific antibody. As expected, sfGFP was detectable in cell extracts of cultures containing sfgfp under control of the lac but not of the hrpB1 promoter ( Figure 3A ). All tested vir promoters as well as the dotA and dotD promoters led to sfGFP synthesis. Similar sfGFP levels were obtained when constructs were analyzed in strains 85-10 and 85-10ΔvirG, suggesting that VirG is dispensable for the activation of gene expression ( Figure 3A ). Notably, the predicted promoter region spanning 229 bp upstream of the annotated start site of icmL did not lead to detectable sfgfp expression ( Figures 3A, B ).

In addition to the in vitro studies, we analyzed a potential in planta activity of vir and icm/dot promoters. For this, we generated expression constructs containing vir and icm/dot promoters upstream of an artificial TAL (transcription activator-like) effector gene (dTALE-2, designer TAL effector-2). dTALE-2 is a derivative of the type III effector protein AvrBs3 from X. euvesicatoria which is imported into the plant cell nucleus where it acts as a transcription factor and activates plant gene expression (Boch et al., 2014). TAL effectors bind to defined nucleotide sequences in the promoters of target genes via their central region, which consists of a variable number of amino acid repeats and allows the base-specific binding to DNA (Boch et al., 2009; Moscou and Bogdanove, 2009). dTALE-2 is used as a reporter for T3S-dependent protein translocation because it binds to a dTALE-2-responsive promoter which controls the expression of gfp on a stably integrated viral vector construct in transgenic Nicotiana benthamiana plants (Werner et al., 2011; Scheibner et al., 2016). vir and icm/dot promoter-driven expression in X. euvesicatoria, which leads to the translocation of dTALE-2 by the T3S system, therefore activates gfp expression in transgenic N. benthamiana plants ( Figures 3C, D ). For these assays, expression constructs encoding dTALE-2 under control of vir or icm/dot promoters were transferred into X. euvesicatoria strain 85-10ΔxopQ which lacks the T3E XopQ and induces disease symptoms in N. benthamiana plants (Adlung et al., 2016). Inoculation of gfp-transgenic N. benthamiana plants with xopQ deletion mutants containing dTALE-2 expression constructs led to GFP fluorescence 4 dpi when dTALE-2 was expressed under control of the virB5, virG or hrpB1 promoter ( Figure 3D ). Expression of dTALE-2 under control of the virB2, dotA and dotD promoters led to weaker fluorescence which was detectable 7 dpi ( Figure 3D ). No GFP fluorescence was observed when dTALE-2 was expressed under control of the virD4 or the icmL promoter ( Figure 3D ). The lack of detectable GFP fluorescence after virD4 promoter-driven expression of dTALE-2 is in contrast to the in vitro activity of the virD4 promoter and suggests that virD4 is not or only weakly expressed in planta ( Figures 3A, D ). In case of the predicted icmL promoter, we assume that the region 229 nucleotides upstream of the annotated GTG start codon of icmL does not activate in vitro or in planta reporter gene expression. Interestingly, DNA sequence analyses revealed the presence of an additional putative ATG translation initiation codon 129 nucleotides downstream of the predicted GTG start codon of icmL ( Figure 3B ). To investigate whether the icmL promoter is located between the annotated GTG and the alternative ATG start codon, we generated an additional expression construct containing dTALE-2 under control of a 129-bp DNA fragment upstream of the ATG codon. This led to detectable in vitro and in planta reporter gene expression, indicative of a promoter downstream of the annotated GTG translation start site ( Figures 3A, D ). In agreement with this finding, a promoter and a vir box was predicted in the region spanning 129 bp downstream of the annotated GTG start codon ( Table 1 and Table S4 ).

To analyze a potential virulence function of T4S genes in X. euvesicatoria strain 85-10, we deleted single and multiple vir and icm/dot genes encoding predicted essential components of the T4S systems such as ATPases (VirD4, VirB4, VirB11 and DotB), the predicted structural components VirB6, VirB8, VirB9, VirB10, DotC and IcmE, and the putative regulators VirA and VirG ( Figures 1 , 2 ). For infection experiments, the wild-type X. euvesicatoria strain 85-10 and corresponding deletion mutants were infiltrated into leaves of susceptible ECW pepper plants and disease symptoms were inspected over a period of five to eight days. As expected, the wild-type strain 85-10 induced disease symptoms in form of water-soaked lesions that became necrotic five to seven dpi ( Figure 4A ). No reduction in symptom formation was detectable after infiltration of mutant strains deleted in single or multiple vir or icm/dot genes ( Figure 4A ). To investigate a potential redundant virulence function of both T4S systems, we also combined deletions in vir and icm/dot genes, thus generating strains 85-10ΔvirB4ΔicmE, 85-10ΔdotCBΔicmEΔvirB4 and 85-10ΔvirB6-10ΔicmEΔdotCB. However, no significant reduction in disease symptoms was observed ( Figure 4A ).

In addition to phenotypic analyses, we performed in planta growth curves with strains 85-10, 85-10ΔvirB6-10, 85-10ΔdotCBΔicmE and 85-10ΔvirB6-10ΔdotCBΔicmE. For this, bacteria were infiltrated into leaves of susceptible ECW pepper plants, and bacterial growth was analyzed over a period of twelve days. As control, we included the non-pathogenic strain 85-10ΔhrcN which lacks the ATPase HrcN of the T3S system (Lorenz and Büttner, 2009). The wild-type strain 85-10 multiplied up to 10⁸ cfu ml^-1 whereas a significant reduction in bacterial growth was observed for strain 85-10ΔhrcN as expected ( Figure 4B ). T4S mutant strains displayed a wild-type growth rate, suggesting that vir and icm/dot genes do not significantly contribute to bacterial multiplication in planta ( Figure 4B ).

The chromosomally encoded VirB/VirD4 T4S system from X. axonopodis pv. citri and the corresponding T4S system from the related pathogen S. maltophilia were previously shown to deliver toxins into Gram-negative bacteria such as E. coli, thus resulting in cell lysis (Souza et al., 2015; Bayer-Santos et al., 2019). To investigate a similar toxic effect of X. euvesicatoria on E. coli, we incubated X. euvesicatoria strain 85-10 and deletion mutants thereof lacking single or multiple vir or icm/dot genes in the presence of decreasing concentrations of E. coli strain BL21. Co-cultures were spotted on LB agar plates containing X-Gal which is a chromogenic artificial substrate of the β-galactosidase from E. coli and is converted into a blue pigment. The formation of blue colonies by E. coli indicates bacterial multiplication and was unaffected upon cocultivation with X. euvesicatoria ( Figure 5A ).

In additional experiments, we incubated X. euvesicatoria wild-type and T4S mutant strains with E. coli strain Top10 in a ratio of 100:1 on solid LB medium and plated the bacteria 0 and 38 hours later for quantitative analysis. The X. euvesicatoria/E. coli ratio decreased 38 h after incubation, indicating a faster growth rate of E. coli compared to X. euvesicatoria ( Figure 5B ). However, no significant difference with X. euvesicatoria wild-type and T4S mutant strains was observed, suggesting that the T4S systems from X. euvesicatoria do not exert a negative effect on the growth of E. coli upon cocultivation under the tested conditions ( Figure 5B ). This is in contrast to the VirB/VirD4 T4S system from X. axonopodis pv. citri which exerts a toxic effect on E. coli (Souza et al., 2015).

The association of traI and traG genes with the plasmid-localized vir gene cluster from X. euvesicatoria prompted us to investigate the role of the predicted VirB/VirD4 T4S system in plasmid transfer. It was previously reported that X. euvesicatoria and X. citri pv. citri transfer plasmids and chromosomal DNA into other Xanthomonas strains (Basim et al., 1999; El Yacoubi et al., 2007). In the present study, we analyzed the transfer of a derivative of plasmid pXCV38 (hereafter referred to as pXCV38_82-8::spec^R ) which originates from strain 82-8 and contains a spectinomycin resistance gene inserted into the avrBs3 T3E gene (Bonas et al., 1989). Donor and recipient strains with plasmid pXCV38_82-8::spec^R were, therefore, selected on medium containing spectinomycin. For conjugation experiments, strain 82-8 with plasmid pXCV38_82-8::spec^R was incubated with a derivative of strain 85-10 containing a gentamicin resistance gene inserted into the genome in the hpaFG region adjacent to the hrp gene cluster. The hpaFG region is dispensable for virulence and serves as a landing platform for gene insertion (Lorenz et al., 2012). Cocultivation of strains 82-8 pXCV38_82-8::spec^R and 85-10::gent^R resulted in transconjugants which were gentamicin- and spectinomycin-resistant, suggesting that pXCV38_82-8::spec^R had been transferred from strain 82-8 into strain 85-10::gent^R ( Figure 6A and Table S6 ). The presence of pXCV38_82-8::spec^R in the recipient strains was confirmed by PCR amplification of the 5’ region of avrBs3 and by agarose gel electrophoresis of isolated plasmids ( Figures 6B and C ). pXCV38_82-8::spec^R from strain 82-8, which has not yet been sequenced, is larger than pXCV38 from strain 85-10. Both plasmids could, therefore, be separated by agarose gel electrophoresis and were present in the resulting transconjugants ( Figure 6C ). Strain 82-8 thus likely contains a functional T4S system which acts as a conjugation machine and mediates transfer of plasmid pXCV38_82-8::spec^R . To investigate whether strain 82-8 contains vir genes, we performed PCR analysis using primers specific for vir genes from strain 85-10. Notably, we did not amplify virB4, virB10 and virB11 from strain 82-8, suggesting that the primers did not efficiently bind to the template DNA ( Figure 6D ). In contrast, the chromosomal virD4 gene was amplified from strains 85-10 and 82-8 ( Figure 6D ). We assume that strain 82-8 contains a functional T4S system because it mediates transfer of plasmid pXCV38_82-8::spec^R (see above).

To investigate a possible contribution of the T4S systems to plasmid transfer in strain 85-10, pXCV38_82-8::spec^R was transferred to derivatives of strain 85-10::gent^R deleted in virD4 or virB4 ( Figure 6A ). The resulting T4S wild-type and mutant derivatives of strain 85-10::gent^R containing pXCV38_82-8::spec^R served as donor strains in additional conjugation experiments, using a kanamycin-resistant derivative of strain 85-10 (85-10::kan^R ) as recipient ( Figure 6A ). Transconjugants grew on medium containing kanamycin and spectinomycin but were sensitive to gentamicin, indicating a transfer of pXCV38_82-8::spec^R from strain 85-10::gent^R to strain 85-10::kan^R (data not shown). PCR analyses confirmed the presence of the kanamycin resistance gene and the 5’ region of avrBs3 in the transconjugants ( Figure 6B ). pXCV38_82-8::spec^R was most efficiently transferred by strain 85-10::gent^R , whereas significantly less transformants were obtained when strain 85-10ΔvirD4::gent^R was used as donor strain ( Figure 6E ). Residual plasmid transfer was observed with strain 85-10ΔvirB4::gent^R , resulting in approximately 20% of the transconjugants obtained with the wild-type strain ( Figure 6E ). It is possible that the loss of VirB4 was partially compensated by functionally redundant proteins encoded in the genome of the donor strain or on plasmid pXCV38_82-8::spec^R , which was used to analyze the plasmid transfer. However, given the severe reduction in plasmid transfer by the virD4 mutant, we assume that the VirB/VirD4 T4S system acts as a conjugation machine and mediates horizontal gene transfer.

Next, we investigated whether VirB/VirD4 and Icm/Dot proteins from X. euvesicatoria assemble into oligomeric complexes as would be expected from their predicted functions as components of T4S systems. For protein-protein interaction studies, we used the bacterial adenylate cyclase-based two-hybrid (BACTH) system which detects the interaction of proteins fused to the T18 or the T25 subdomain of the catalytic region of the adenylate cyclase (Cya). Interaction of T18 and T25 fusions leads to the reconstitution of Cya and thus to the production of cAMP. This activates the expression of the lac operon in E. coli reporter strains lacking the native cya gene (Karimova et al., 1998; Battesti and Bouveret, 2012). LacZ activity results in blue colonies, indicative of an interaction between both fusion proteins.

We performed BACTH assays with T18 and T25 fusions of the predicted ATPases VirD4, VirB4 and VirB11 as well as of the predicted core components of the VirB/VirD4 T4S system VirB7 and VirB10 ( Figure 7A ). Corresponding expression constructs were cotransformed into the E. coli BTH101 reporter strain and transformants were grown on plates containing X-Gal. We detected homo- and heterooligomerizations of the putative ATPases VirD4 and VirB11 as well as interactions between all tested putative ATPases with VirB10 ( Figure 7B and Table 2 ). Furthermore, VirB7 interacted with itself and with VirB10 ( Figure 7B and Table 2 ). The interactions were specific because the analyzed T4S system components did not interact with the T25 and T18 domains alone ( Figure 7B ). Immunoblot analysis of bacterial cell extracts showed that most tested T25 and T18 fusion proteins were stably synthesized ( Figure S1 ). T25-VirB4 was not detectable and was, therefore, not included in the analyses.

In addition to components of the predicted VirB/VirD4 T4S system, we performed BACTH assays with Icm/Dot proteins including the predicted ATPases DotO and DotB, which correspond to VirB4 and VirB11, respectively (Sheedlo et al., 2022) ( Figure 7C ). We also performed BACTH assays with the predicted coupling protein DotL as well as the putative major structural component IcmE which is also referred to as DotG and corresponds to VirB10. When analyzed as T25 and T18 fusions, DotO and DotB interacted with themselves as is expected for hexameric ATPases ( Figure 7D and Table 2 ). Furthermore, a weak interaction between T18-IcmE and DotO-T25 was detected and IcmE interacted with itself ( Figure 7D and Table 2 ). No self-interaction was observed for the predicted coupling protein DotL ( Figure 7D ). Yet, it cannot be excluded that N- or C-terminal T25 or T18 fusion partners interfere with the self-interaction of DotL. All tested fusion proteins were synthesized as was revealed by immunoblot analysis ( Figure S1 ). Taken together, we conclude from the results of the BACTH assays that the predicted ATPases of both T4S systems self-interact as expected. Furthermore, VirD4, VirB4 and VirB11 interact with VirB10, which is the putative major structural component of the VirB/VirD4 T4S system and also associates with VirB7 ( Table 2 ). Our results thus suggest that Vir and Icm/Dot proteins are involved in the assembly of heterooligomeric protein complexes, likely corresponding to T4S systems.

To identify possible substrates of the VirB/VirD4 T4S system from X. euvesicatoria, we searched for homologs of known type IV effectors from X. axonopodis pv. citri which are referred to as X-Tfes and were previously identified due to their interaction with the T4CP VirD4 (Alegria et al., 2005). In the present study, we focused on the X-Tfe homologs XCV0332, XCV1120 and XCV3571, which contain C-terminal motifs of XVIPCDs such as blocks of conserved amino acid sequences (GLxRIDHV and FAVQ GxxDPAHxRAHV; x refers to any residue) and a glutamine-rich C-terminal tail (Alegria et al., 2005) ( Tables 3 and S7 ). Given that the XVIPCD provides the binding site for the AAD of VirD4 from X. axonopodis pv. citri (Alegria et al., 2005; Oka et al., 2022), we investigated possible interactions of the identified T4S candidate substrates with VirD4 from X. euvesicatoria. BACTH assays revealed interactions of T25 fusions of XCV3751, XCV1120 and XCV0332 with T18 fusions of VirD4 ( Figure 8A and Table 2 ). The interactions were specific for VirD4, because the analyzed VirD4 interaction partners did not interact with the T18 domain alone ( Figure 8A ). Immunoblot analyses showed that most fusion proteins were stably synthesized ( Figure S1 ). Exceptions were XCV0332-T25 and XCV3751-T25, which were, therefore, not included in the analyses ( Figure S1 ). It was previously reported that the N-terminal region of the XVIPCD from X-Tfes adopts an ααβββ fold which interacts with the AAD of VirD4 (Oka et al., 2022). Interestingly, in silico modelling using the AlphaFold 2 algorithm (Jumper et al., 2021) revealed a similar fold of XVIPCDs in XCV0332, XCV1120 and XCV3751 which might associate with the AAD of VirD4 from X. euvesicatoria ( Figures 8B-D ). As additional candidate T4S substrate, we analyzed TraI which is a predicted relaxase and likely secreted by the VirB/VirD4 T4S system. TraI lacks an XVIPCD but also interacted with VirD4 when analyzed as T25 fusion ( Figure 8A and Table 2 ).

Given the results of our interaction and mutant studies, we assume that VirD4, which is encoded on the chromosome, acts as coupling protein for the VirB-like T4S system encoded on plasmid pXCV38. It can, however, not be excluded that substrate recognition is mediated by TraG which is encoded next to the virB T4S gene cluster on plasmid pXCV38 and shares 23% amino acid identity with VirD4 ( Figure 1 and Figure S2 ). Since TraG-like proteins were shown to contribute to T4S (Schröder et al., 2002; Khara et al., 2021; Wang et al., 2022), we analyzed a possible interaction of TraG with components and putative T4S substrates of the VirB-like T4S system. When analyzed by BACTH assays, TraG interacted with itself, with the predicted ATPase VirB11 and the structural component VirB10 but not with the potential T4S substrate TraI ( Figure S2 ). This is in contrast to VirD4 and suggests that VirD4 rather than TraG is involved in substrate recruitment.

To analyze a possible in vitro secretion of candidate T4S substrates from X. euvesicatoria, we generated expression constructs encoding XCV3751, XCV1120, XCV0332 and TraI fused to an N-terminal c-Myc epitope under control of the lac promoter. We assume that an N-terminal epitope does not significantly interfere with secretion because T4S signals are often located in the C-terminal protein region (Vergunst et al., 2000; Nagai et al., 2005). When analyzed in strain 85-10, epitope-tagged derivatives of XCV3751, XCV1120 and XCV0332 were stably synthesized. In contrast, c-Myc-TraI was not detected in cell extracts by immunoblotting, suggesting that it was unstable when expressed under control of the lac promoter (data not shown). traI was, therefore, expressed under control of a native promoter from X. euvesicatoria, which resulted in detectable protein synthesis (see below). For in vitro secretion assays, bacteria were cultivated in NYG medium containing BSA which likely stabilizes secreted proteins as was shown for substrates of the T3S system (Rossier et al., 1999). Total cell extracts and culture supernatants were separated by filtration and analyzed by immunoblotting. Notably, all tested proteins were detected in the culture supernatant of strain 85-10, suggesting that they were secreted ( Figures 9A, B ).

To investigate whether secretion of the analyzed proteins was dependent on the T4S systems, we performed in vitro secretion assays with strain 85-10ΔicmEΔvirB4, which lacks predicted essential components of the VirB/VirD4 and Icm/Dot T4S system. Secretion of c-Myc-XCV1120 and c-Myc-XCV3751 was unaffected in strain 85-10ΔicmEΔvirB4, suggesting that the presence of the XVIPCD and the interaction with VirD4 is not sufficient to exclusively target both proteins to the T4S systems ( Figure 9A ). In contrast, significantly reduced amounts of c-Myc-XCV0332 and c-Myc-TraI were detected in the culture supernatant of strain 85-10ΔicmEΔvirB4, indicating a contribution of the T4S systems to secretion ( Figure 9B ). As a control, all supernatants were incubated in the presence of proteinase K to investigate a possible contribution of OM vesicles to secretion. OM vesicles were previously shown to mediate secretion of degradative enzymes from X. euvesicatoria (Solé et al., 2015). Detectable secretion of all tested candidate T4S substrates, however, was abolished upon treatment of the culture supernatants with proteinase K, suggesting that the secreted proteins were not present in OM vesicles ( Figures 9A, B ).

To further elucidate the contribution of the VirB/VirD4 or the Icm/Dot T4S system to secretion of c-Myc-XCV0332 and c-Myc-TraI, we introduced the corresponding expression constructs into virB4 and icmE single deletion mutants. Secretion of c-Myc-TraI was unaffected in both strains, suggesting that it is secreted by either the VirB/VirD4 or the Icm/Dot T4S system ( Figure 9C ). In contrast, secretion of c-Myc-XCV0332 was reduced but not completely abolished in strains 85-10ΔicmE and 85-10ΔvirB4 ( Figure 9D ). We, therefore, assume that the VirB/VirD4 and Icm/Dot T4S systems both contribute to the efficient secretion of XCV0332, which was abolished in the absence of both functional T4S systems ( Figure 9B ). Wild-type secretion levels of c-Myc-XCV0332 were restored when virB4 and icmE were reinserted into the genomes of strains 85-10ΔvirB4 and 85-10ΔicmE, respectively ( Figure 9D ). This suggests that reduced secretion of XCV0332 in the single mutants was specifically caused by the absence of VirB4 and IcmE, respectively, and not by a polar effect of the deletions on other genes. Our findings thus indicate that VirB4/VirD and Icm/Dot T4S systems are involved in in vitro protein secretion and that XCV0332 and TraI are targeted to both T4S systems. In case of TraI, the Icm/Dot T4S system appears to fully compensate for the loss of a functional VirB/VirD4 T4S system.