Bacterial strains and plasmids used in this study are listed in Table 1. Escherichia coli strains were cultivated at 37°C in lysogeny broth (LB) medium and X. campestris pv. vesicatoria strains at 30°C in nutrient-yeast extract-glycerol (NYG) medium (Daniels et al., 1984) or minimal medium A (Ausubel et al., 1996) at pH 7.0 supplemented with 10 mM sucrose and 0.3% casamino acids.

X. campestris pv. vesicatoria strains were inoculated into leaves of the near-isogenic pepper cultivars Early Cal Wonder (ECW), ECW-10R and ECW-30R, and Bs3- or gfp-transgenic Nicotiana benthamiana plants at concentrations of 4 × 10⁸ colony-forming units (CFU) ml⁻¹ in 1 mM MgCl₂ if not stated otherwise (Minsavage et al., 1990; Bonas et al., 1991; Kousik and Ritchie, 1998; Scheibner et al., 2016). Agrobacterium tumefaciens strains were infiltrated in 1 mM MgCl₂ at a density of 8 × 10⁸ CFU ml⁻¹. Infected pepper plants were incubated for 16 h of light at 28°C and 65% humidity, and 8 h of darkness at 22°C and 65% humidity. N. benthamiana plants were incubated for 16 h of light at 20°C and 75% humidity, and 8 h of darkness at 18°C and 70% humidity. The appearance of plant reactions was scored over a period of one to 12 dpi. For the better visualization of the HR, leaves were destained in 70% ethanol. Experiments were repeated at least twice; representative plant reactions are shown.

For the generation of expression constructs encoding AvrBs3Δ2 fusion proteins under control of the lac promoter, avrBs3 gene fragments were amplified by PCR from X. campestris pv. vesicatoria and cloned into the BsaI sites of plasmid pBR356, which contains the avrBs3Δ2 reporter gene (Scheibner et al., 2016). For the generation of constructs encoding AvrBs3Δ63 or AvrBs3Δ63 fusion proteins, modules encoding N-terminal, central and C-terminal regions of AvrBs3Δ63 and the respective fusion partners were cloned into the BsaI sites of plasmid pBRM. The stop codon of avrBs3 was included in these constructs. xopJ_1-155-avrBs3Δ2 and xopJ_1-155-avrBs3Δ63 were expressed under control of the native xopJ promoter in plasmid pBRM-P. For this, fragments containing the first 155 codons of xopJ and 720 bp of the upstream region including the xopJ promoter (Noël et al., 2003) were ligated with fragments encoding the N-terminal, central and C-terminal regions of AvrBs3Δ63.

For the generation of binary constructs for A. tumefaciens-mediated transient gene expression, avrBs3_1-10 and avrBs3_1-20 were amplified by PCR and ligated with modules encoding N-terminal, central and C-terminal regions of AvrBs3Δ2 into the BsaI sites of the binary vector pGGA3 downstream of the 35S promoter. Similarly, modules encoding N-terminal, central and C-terminal regions of AvrBs3 were cloned into pGGA3. The module encoding the C-terminal region of AvrBs3 lacked the native stop codon and was therefore expressed in fusion with a FLAG epitope-encoding sequence provided by vector pGGA3.

Expression constructs encoding dTALE-2 and derivatives thereof were generated by Golden-Gate assembly of individual DNA modules encoding N-terminal, central and C-terminal regions of dTALE-2, dTALE-2ΔN as well as N-terminal regions of the respective fusion partners. All primers used in this study are listed in Table 2. Plasmids were introduced into X. campestris pv. vesicatoria by electroporation or by conjugation using pRK2013 as a helper plasmid in triparental matings (Figurski and Helinski, 1979).

For the generation of X. campestris pv. vesicatoria deletion mutants (see Table 1), derivatives of the suicide vector pOK1, which contained the flanking regions of the deleted genes, were transferred to X. campestris pv. vesicatoria recipient strains by triparental conjugation. Double crossovers resulted in deletion mutants which were selected as described previously (Huguet et al., 1998).

For protein analysis, bacteria were cultivated over-night in liquid NYG medium and cells were harvested by centrifugation. Equal amounts of proteins adjusted according to the optical densities of the cultures were analyzed by immunoblotting, using AvrBs3- or HrpF-specific antibodies. For the analysis of in vitro T3S, bacteria were incubated in MA medium at pH 5.3 for 1.5 h in the presence of thiamine and bovine serum albumin as described previously (Rossier et al., 1999). Secreted proteins were separated from bacterial cells by filtration and precipitated by trichloroacetic acid. Equal amounts of bacterial total cell extracts and culture supernatants (adjusted according to the optical densities of the cultures) were analyzed by SDS-PAGE and immunoblotting using antibodies specific for AvrBs3, the translocon protein HrpF, the inner membrane ring protein HrcJ and the predicted periplasmic inner rod protein HrpB1, respectively (Knoop et al., 1991; Rossier et al., 2000; Büttner et al., 2002). Horseradish peroxidase-labeled anti-rabbit antibodies (GE Healthcare) were used as secondary antibodies. Experiments were repeated twice.

For transcript analysis via semiquantitative reverse-transcription PCR (RT-PCR), strain 85^* and derivatives thereof ectopically expressing avrBs3 were inoculated at concentrations of 4 × 10⁸ CFU ml⁻¹ and derivatives of strain 82-8 at concentrations of 8 × 10⁸ CFU ml⁻¹ into leaves of ECW-30R pepper plants or Bs3-transgenic N. benthamiana plants. 16 leaf discs (2 mm diameter) were harvested 8 h (for derivatives of strain 85^*) or 24 h (for derivatives of strain 82-8) post infiltration, pooled and immediately frozen in liquid nitrogen. RNA was isolated with the QIAGEN RNeasy Plant Mini Kit, and samples were treated with DNase I (Roche) for 30 min. cDNA was synthesized from 2 μg of total RNA with the RevertAid H Minus First Strand cDNA synthesis kit (Thermo Scientific) using random hexamer primers. Four microlitres of a 1:50 diluted cDNA solution were used as template in an RT-PCR with up to 40 cycles of denaturation, annealing and elongation. The constitutively expressed gene Elongation factor 1α (EF1α) was amplified as control (Römer et al., 2007). To exclude contaminations by genomic DNA, all reactions were also performed in the absence of reverse transcriptase. The experiments were repeated at least twice with similar results.

To localize the T3S signal in AvrBs3, we generated fusion proteins consisting of the first 10, 20, or 30 amino acids, respectively, of AvrBs3 and AvrBs3Δ2, which is an N-terminal deletion derivative of AvrBs3. AvrBs3Δ2 lacks amino acids 2–152 and thus the T3S and translocation signal. However, AvrBs3Δ2 contains the effector domain and induces the HR in AvrBs3-responsive plants when delivered as fusion partner of a functional translocation signal (Szurek et al., 2002; Noël et al., 2003; Drehkopf et al., 2017). AvrBs3Δ2 fusion proteins were analyzed in X. campestris pv. vesicatoria strain 85^*, which is a derivative of the wild-type strain 85-10 and contains HrpG^*, a constitutively active version of the hrp gene regulator HrpG (Rossier et al., 1999; Wengelnik et al., 1999). When bacteria were cultivated in secretion medium, AvrBs3₁₋₁₀-, AvrBs3₁₋₂₀- and AvrBs3₁₋₃₀-AvrBs3Δ2 fusion proteins were detected in the culture supernatants (Figure 1A). Similar results were obtained for strain 85^*ΔhpaB, which lacks the T3S chaperone HpaB (Figure 1A; Büttner et al., 2004). As controls, we analyzed the inner membrane-associated protein HrcJ and the predicted periplasmic inner rod protein HrpB1. Both proteins were only detected in cell extracts of strains 85^* and 85^*ΔhpaB, suggesting that no cell lysis had occurred (Figure 1A). We conclude from these data that the N-terminal 10–30 amino acids of AvrBs3 target AvrBs3Δ2 for secretion, even in the absence of the T3S chaperone HpaB. This is in contrast to the full-length AvrBs3 protein, which depends on HpaB for efficient secretion (Büttner et al., 2004).

For the analysis of translocation signals in AvrBs3, we performed in vivo translocation assays with AvrBs3Δ2 fusion proteins. For this, strain 85^* ectopically expressing individual avrBs3Δ2 fusions was infiltrated into leaves of AvrBs3-responsive ECW-30R pepper plants. AvrBs3₁₋₅₀-AvrBs3Δ2 induced the HR in AvrBs3-responsive ECW-30R pepper plants when delivered by strain 85^*, suggesting that the N-terminal 50 amino acids contain a functional translocation signal (Figure 1B). HR induction by AvrBs3₁₋₅₀-AvrBs3Δ2 was dependent on the T3S system because it was macroscopically not detectable when AvrBs3₁₋₅₀-AvrBs3Δ2 was analyzed in the T3S-deficient strain 85^*ΔhrcV (Figure 1C). In contrast to AvrBs3₁₋₅₀-AvrBs3Δ2, AvrBs3Δ2 fusions containing the N-terminal 10, 20, 30, or 40 amino acids of AvrBs3 did not induce a visible HR when analyzed in strain 85^*, suggesting that they were not or not efficiently translocated (Figure 1B). Immunoblot analysis of bacterial cell extracts revealed that all fusion proteins were stably synthesized (Figure 1B).

As control, bacteria were infiltrated into leaves of ECW-10R pepper plants, which contain the resistance gene Bs1 and initiate the HR upon recognition of the effector protein AvrBs1 (Minsavage et al., 1990). All strains induced the AvrBs1-specific HR in leaves of ECW-10R plants, indicating that AvrBs1 was efficiently translocated. This suggests that the AvrBs3Δ2 fusion proteins did not interfere with the activity of the T3S system (Figure 1B).

Next, we investigated the influence of the T3S chaperone HpaB on the translocation of AvrBs3Δ2 fusion proteins. Notably, AvrBs3₁₋₃₀-AvrBs3Δ2 was translocated by strain 85^*ΔhpaB but did not induce a macroscopically visible HR in AvrBs3-responsive pepper plants when analyzed in strain 85^* or the T3S-deficient strain 85^*ΔhpaBΔhrcV (Figures 2A,B; see above). Similar data were obtained for the hrpG wild-type strain 85-10ΔhpaB, suggesting that the translocation of AvrBs3₁₋₃₀-AvrBs3Δ2 by hpaB deletion mutants was not caused by the overexpression of the T3S genes in the presence of HrpG^* (Figure S1). Hence, the N-terminal 30 amino acids of AvrBs3 contain a translocation signal which is recognized in hpaB deletion mutants but is not sufficient for efficient translocation in the wild-type strain. This signal will hereafter be referred to as minimal translocation signal. Translocation in the absence of HpaB was also observed for AvrBs3₁₋₄₀-AvrBs3Δ2. In contrast, translocation of AvrBs3₁₋₅₀-AvrBs3Δ2 was significantly reduced in strain 85^*ΔhpaB (Figure 2A) and not detectable in the hrpG wild-type strain 85-10ΔhpaB (Figure S1), suggesting that the N-terminal 50 amino acids of AvrBs3 depend on HpaB for efficient translocation.

AvrBs3Δ2 fusion proteins containing the N-terminal 10 or 20 amino acids of AvrBs3 did not induce a visible HR when analyzed in strain 85^*ΔhpaB (Figure 2A). To exclude that the N-terminal 10 or 20 amino acids of AvrBs3 interfere with the activity of the AvrBs3Δ2 reporter, both gene fusions were expressed under control of the 35S promoter in AvrBs3-responsive pepper plants after A. tumefaciens-mediated gene delivery. Transient expression of avrBs3_1-10- and avrBs3_1-20-avrBs3Δ2 led to the induction of the Bs3-specific HR, indicating that the AvrBs3Δ2 reporter was functional when present as a fusion partner of the N-terminal 10 or 20 amino acids of AvrBs3 (Figure 2C). Thus, the lack of HR induction by AvrBs3₁₋₁₀- and AvrBs3₁₋₂₀-AvrBs3Δ2 fusion proteins was presumably not caused by a misfolding of the reporter. Taken together, we conclude from these data that the N-terminal 50 amino acids of AvrBs3 contain a translocation signal whereas the N-terminal 30 amino acids promote translocation of AvrBs3 in the absence of HpaB. The N-terminal 10 or 20 amino acids did not target the AvrBs3Δ2 reporter for detectable translocation, however, it cannot be excluded that this region contains a translocation signal which is inactive in the context of the AvrBs3Δ2 fusion.

In contrast to AvrBs3₁₋₅₀-AvrBs3Δ2, the full-length AvrBs3 protein induced the HR when analyzed in strains 85^*ΔhpaB and 85-10ΔhpaB, suggesting that it was efficiently translocated in the absence of HpaB (Figure 2A; Figure S1). Similar findings were observed for strain 82^*ΔhpaB, which is a derivative of strain 82-8 and naturally expresses avrBs3 (Figure 2D). This indicates that the HpaB-independent translocation of AvrBs3 was not caused by the overexpression of T3S genes or of avrBs3. We, therefore, wondered whether the differences in HR induction by AvrBs3 and AvrBs3₁₋₅₀-AvrBs3Δ2 were caused by differences in protein activities. Previous reporter assays revealed that the N-terminal deletion derivative AvrBs3Δ2 led to reduced activation of an AvrBs3-responsive promoter when compared with the full-length AvrBs3 protein (Schreiber et al., 2015). In contrast, transcription activation by an AvrBs3 derivative, which was deleted in the N-terminal 63 amino acids, was like wild type, suggesting that the region between amino acids 63 and 152 of AvrBs3 contributes to protein activity (Schreiber et al., 2015). To investigate whether different activities of AvrBs3 derivatives result in different HR intensities, we compared the HR induction by AvrBs3Δ2 and AvrBs3Δ63 fusion proteins. As fusion partner, which provides the T3S and translocation signal, we chose the N-terminal 155 amino acids of the effector protein XopJ (Noël et al., 2003). When strain 85^* or 85-10 delivering XopJ₁₋₁₅₅-AvrBs3Δ2 or XopJ₁₋₁₅₅-AvrBs3Δ63 was infiltrated into leaves of AvrBs3-responsive pepper plants, no differences in the HR intensities were detected, even when bacteria were infiltrated at lower optical densities (Figure 3A). As both fusion proteins were stably synthesized at comparable levels (Figure 3A), we conclude that the previously observed reduced transcription activation activity of AvrBs3Δ2 did not lead to a macroscopically detectable reduction of the AvrBs3-induced HR. The efficient translocation of AvrBs3 by hpaB deletion mutants might, therefore, be caused by the presence of additional export signals which promote translocation in the absence of HpaB and are absent in AvrBs3Δ2 fusion proteins.

To investigate the presence of additional export signals outside the N-terminal 50 amino acids of AvrBs3, we analyzed the translocation of fusion proteins between the N-terminal 10 or 30 amino acids of AvrBs3 and AvrBs3Δ63, which lacks the T3S and translocation signal (Schreiber et al., 2015; Scheibner et al., 2016; Figure 3B). Notably, AvrBs3₁₋₁₀- and AvrBs3₁₋₃₀-AvrBs3Δ63 induced the AvrBs3-specific HR when delivered by strains 85^* and 85^*ΔhpaB (Figure 3C). This is in contrast to the corresponding AvrBs3Δ2 fusions and suggests that the region between amino acids 64 and 152 of AvrBs3 contains a signal which promotes translocation even in the absence of HpaB.

It was previously reported that the translocation of effector proteins from Xanthomonas spp. depends on the translocon protein HrpF (Büttner et al., 2002; Szurek et al., 2002; Hotson et al., 2003; Thieme et al., 2007; Jiang et al., 2009; Teper et al., 2016). To confirm the contribution of HrpF to the translocation of AvrBs3, we performed translocation studies with hrpF deletion mutants. Unexpectedly, strain 85^*ΔhrpF ectopically expressing avrBs3 induced a browning of the infected leaf tissue, which was visible after destaining of the leaves in ethanol 3 dpi (Figure 4A). A similar phenotype was observed with strain 85^*ΔhrpFΔxopA, which additionally lacks the secreted XopA protein (Figure 4A; Noël et al., 2002). XopA contributes to pathogenicity and is homologous to harpins, which are proposed to be involved in effector protein translocation (Noël et al., 2002; Kim et al., 2004). No browning of the leaf tissue was visible when avrBs3 was ectopically expressed in strains 85^*ΔhrpE and 85^*ΔhrcN, which are deleted in the pilus gene hrpE and the ATPase gene hrcN, respectively (Figure 4A).

To investigate whether the browning of the infected leaf tissue could have resulted from the induction of the R gene Bs3, we performed RT-PCR studies. The Bs3 transcript was amplified from leaf tissue infected with strain 85^* ectopically expressing avrBs3 but not from tissue infected with the corresponding hrcN deletion mutant (Figure 4B). Reduced amounts of the Bs3 transcript were detected in leaf material infected with the hrpF deletion mutant, suggesting that AvrBs3 entered the plant cells in the absence of the translocon protein HrpF and induced the expression of Bs3, albeit in reduced amounts (Figure 4B). We assume that the HrpF-independent entry of AvrBs3 into plant cells was not sufficient for the induction of a macroscopically visible HR reaction in Bs3 pepper plants but resulted in a browning of the infected leaf tissue. The phenotypes were not caused by the overexpression of avrBs3 or the T3S genes, because browning of the leaf tissue was also observed after inoculation of Bs3 pepper plants with the hrpF deletion mutant strain 82-8ΔhrpF, which contains the hrpG wild-type gene and naturally expresses avrBs3 (Figure S2). In agreement with these observations, the Bs3 transcript was detectable in leaf material infected with strain 82-8ΔhrpF but not with the T3S-deficient strain 82-8ΔhrcV (Figure S2).

To investigate whether the translocon-independent entry of AvrBs3 into plant cells also occurs in non-host plants, we performed infection assays with Bs3-transgenic N. benthamiana plants. As reported previously, strain 85^* induced a non-host HR on N. benthamiana plants, which is visible as necrotic area (Figure 5A; Metz et al., 2005; Scheibner et al., 2016). No macroscopic HR was observed after infection with the hrpF deletion mutant strain 85^*ΔhrpF, suggesting that the induction of the non-host HR in N. benthamiana depends on the T3S translocon (Figure 5A). Strain 85^*ΔhrpF ectopically expressing avrBs3, however, induced the HR in Bs3-transgenic N. benthamiana plants, suggesting that AvrBs3 entered the plant cells in the absence of HrpF (Figure 5A). The reaction was dependent on the T3S system because AvrBs3 did not induce a visible HR when analyzed in strain 85^*ΔhrcN, which lacks the T3S ATPase (Figure 5A).

To confirm these results, we performed translocation assays with X. campestris pv. vesicatoria strains ectopically expressing dTALE-2 (designer TAL effector-2), which encodes an artificial derivative of AvrBs3 with modified RVDs (Weber et al., 2011b). To analyse the translocation of dTALE-2, bacteria were infiltrated into leaves of transgenic N. benthamiana plants, which encode the green fluorescent protein (GFP) under control of a dTALE-2-responsive promoter on a stably integrated viral vector construct (Werner et al., 2011; Scheibner et al., 2016). dTALE-2 induced GFP fluorescence when delivered by strain 85^* (Figure 5B; Scheibner et al., 2016). Reduced fluorescence was observed when dTALE-2 was analyzed in strains 85^*ΔhrpF and 85^*ΔhrpFΔxopA whereas no fluorescence was detectable after infiltration of the T3S-deficient strain 85^*ΔhrpE (Figure 5B). Similar data were obtained for derivatives of the wild-type strain 85-10, suggesting that the HrpF-independent entry of dTALE-2 into plant cells was not caused by the overexpression of the T3S genes (Figure 5C). Taken together, these data suggest that dTALE-2 can enter the plant cell in a translocon-independent manner.

We also performed infection experiments with strain 85-10ΔhrpFΔxpsDΔxcsD, which lacks hrpF and functional T2S systems. Type II-secreted cell wall-degrading enzymes were proposed to facilitate T3S pilus assembly and thus presumably contribute to the HrpF-independent passage of AvrBs3 across the plant plasma membrane (Szczesny et al., 2010b). In agreement with this hypothesis, dTALE-2 induced reduced GFP fluorescence when analyzed in strain 85-10ΔhrpFΔxpsDΔxcsD (Figure 5D).

Next, we investigated the contribution of the translocation signal to the translocon-independent entry of AvrBs3 into plant cells. Infection experiments showed that the N-terminal deletion derivative dTALE-2ΔN did not induce detectable GFP fluorescence in gfp-transgenic N. benthamiana plants when analyzed in strains 85^* and 85^*ΔhrpF (Figure 6A). As dTALE-2ΔN is deleted in the T3S and translocation signal (Schreiber et al., 2015; Scheibner et al., 2016), the lack of GFP fluorescence suggests that the translocon-independent delivery of dTALE-2 depends on its passage through the T3S system. In contrast to dTALE-2ΔN, AvrBs3₁₋₅₀-dTALE-2ΔN induced GFP fluorescence when analyzed in strains 85^*, 85^*ΔhpaB and 85^*ΔhrpF (Figure 6B). Notably, GFP fluorescence induced by strain 85^* was reduced because of the induction of the non-host HR by strain 85^* which leads to tissue necrosis (see above; Figure 5A). Our observations confirm the results obtained for the AvrBs3₁₋₅₀-AvrBs3Δ2 fusion (see above) and suggest that the N-terminal 50 amino acids of AvrBs3 contain a translocation signal, which can target dTALE-2 for translocon-independent translocation.

Translocation assays with AvrBs3₁₋₃₀-dTALE-2ΔN revealed that the N-terminal 30 amino acids of AvrBs3 targeted dTALE-2ΔN for translocation in strain 85^*ΔhpaB whereas no fluorescence was observed with the wild-type strain or the hrpF deletion mutant (Figure 6B). This confirms the finding that the N-terminal 30 amino acids of AvrBs3 contain a minimal translocation signal, which is recognized in the absence of HpaB. Reduced GFP fluorescence, however, was detected when AvrBs3₁₋₃₀-dTALE-2ΔN was analyzed in strain 85^*ΔhrpFΔhpaB (Figure 6B). Thus, AvrBs3₁₋₃₀-dTALE-2ΔN can enter the plant cell in the absence of the translocon when delivered by the hpaB deletion mutant (Figure 6B). Taken together, these data suggest that the translocon-dependent and -independent delivery of AvrBs3 into plant cells depends on similar targeting signals.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.