Generation of a modular xps-T2S gene cluster by Golden Gate cloning

The Xps-T2S system from X. euvesicatoria strain 85-10 is encoded by a chromosomal gene cluster which contains eleven genes organized in three predicted operons comprising xpsE, xpsF and xpsGHIJKLMCD (Fig 1). The corresponding gene products constitute the components of the assembly platform, the pseudopilus, the ATPase and the secretin of the T2S system (S1 Fig). In previous studies, we showed that the ATPase XpsE and the secretin XpsD are essential for T2S [29]. For this, the corresponding genes were deleted from the genome of X. euvesicatoria strain 85-10 by homologous recombination which is a time-consuming approach including several selection steps and the PCR-based analysis of recombination events. Furthermore, complementation of the respective mutant phenotypes often involves the reintroduction of the deleted gene on a plasmid which leads to higher gene copy numbers and thus possibly interferes with protein function. To facilitate mutagenesis of xps genes and complementation studies, we generated a modularized xps-T2S gene cluster by hierarchical Golden Gate-based cloning and assembly of gene and promoter fragments [33–35]. The modular design allows genetic manipulations by rapid exchange of single gene modules in a multi-gene construct. For this, individual promoters and genes were subcloned and subsequently assembled by Golden Gate-based modular cloning (MoClo), using the type IIs restriction enzymes BsaI and BpiI. Both enzymes cut DNA outside of their recognition sites. The resulting 4-bp overhangs allow the ordered assembly of DNA fragments by one-pot restriction/ligation reactions [33,34]. MoClo destination vectors designated level −2, −1, 0, 1 and M are available and contain the lacZα fragment flanked by BsaI and BpiI sites in inverse orientations and alternating order (Fig 1). This design allows the stepwise cloning of operons and multi-gene constructs in which the internal BsaI and BpiI sites have been removed by PCR-based mutagenesis prior to assembly (Fig 1). For modular cloning, single gene and predicted promoter fragments of the xpsG to xpsD operon were subcloned into level −2 vectors, preassembled in level −1 vectors and transferred to a level 0 vector (Fig 1). The corresponding construct was subsequently assembled with additional level 0 modules containing xpsE and xpsF genes to generate the complete xps-T2S gene cluster. Two dummy modules were included upstream and downstream of the xps-T2S gene cluster, which can be replaced by genes or reporter fusions in the final construct. A series of level 1 vectors with different BpiI fusion sites allows the ordered assembly of modules in a final level M vector (Fig 1). The 3’ BpiI site of the last level 1 module is connected to the level M vector by an end-linker (Fig 1). A similar method was previously used to generate a modular T3S gene cluster from X. euvesicatoria [36].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Schematic representation of the xps-T2S gene cluster and overview of cloning steps for the generation of the modular xps-T2S gene cluster.

Genes are represented by arrows, promoters by orange boxes. Letters refer to the nomenclature of xps genes. The position of internal BsaI and BpiI restriction sites is indicated. Numbers below the arrows refer to the length of the single genes in base pairs. For the modular assembly, gene and promoter fragments were amplified by PCR and cloned into level −2 vectors (xpsG to xpsD) or level 0 vectors (xpsE and xpsF) as indicated. Restriction/ligation using the type IIs restriction enzymes BsaI and BpiI in alternating order led to the assembly of transcription units in level 0 and level 1 vectors and of the entire xps-T2S gene cluster in the level M vector. The final level M construct contains an end-linker (EL) and two dummy modules (grey boxes) which can be replaced by additional genes or reporter fusions.

https://doi.org/10.1371/journal.ppat.1013008.g001

The modular Xps-T2S system is functional in X. euvesicatoria

For functional studies, the final level M expression construct containing the modularized T2S gene cluster was introduced into strain 85-10∆xps, in which the entire chromosomal xps gene cluster had been deleted from the chromosome. When bacteria were syringe-infiltrated into leaves of susceptible pepper plants, the wild-type strain 85-10 induced disease symptoms in form of water-soaked lesions whereas reduced symptoms were observed after infiltration of strain 85-10∆xps as expected (Fig 2A) [29]. The wild-type phenotype was restored upon introduction of the modular T2S gene cluster on a low copy level M vector (Fig 2A). We also conducted dip infection assays which simulate more natural conditions compared to the syringe infiltration experiments. Dip infection of pepper leaves with the wild-type strain 85-10 led to the formation of bacterial spot symptoms whereas spot formation was significantly reduced after infection with strain 85-10∆xps (S2 Fig). As with the phenotypes observed after syringe infiltration, the modular T2S gene cluster restored wild-type levels of spot formation when analysed in the xps deletion mutant (S2 Fig). In addition to infection experiments, we monitored extracellular protease activity on milk-containing agar plates. As reported previously, strain 85-10 led to the formation of a cleared halo around the inoculation site due to the degradation of milk proteins whereas halo formation was severely reduced for strain 85-10∆xps [29,37] (Fig 2A). In the presence of the modular T2S gene cluster, extracellular protease activity was restored and even more prominent than in strain 85-10, suggesting that the plasmid-encoded modular xps-T2S gene cluster is functional (Fig 2A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The modular xps-T2S system is functional in X. euvesicatoria strain 85-10

∆xps. (A) The wild-type strain 85-10 and the T2S deletion mutant 85-10∆xps (∆xps) with (+) or without (−) the modular xps-T2S expression construct (+ pT2S) were analysed for bacterial virulence and extracellular protease activity. For the analysis of xpsD and xpsE mutants, modular T2S gene cluster constructs lacking xpsD (∆xpsD) or xpsE (∆xpsE) were introduced into strain 85-10∆xps. For complementation studies, xpsD (+XpsD) or xpsE (+XpsE) expression cassettes were inserted at position 1 of the respective constructs as depicted in (B). Bacteria were infiltrated at a density of 2 × 10⁷ CFU (colony-forming units) ml⁻¹ into leaves of susceptible pepper plants and disease symptoms were photographed 8 dpi (days post inoculation). Dashed lines indicate the infiltrated areas. For the analysis of extracellular protease activity, bacteria were grown on 1% (w/v) milk plates for 2 days and halo formation was documented. Bacteria were removed from the holes prior to documentation. The numbers refer to the width of the halos in millimeters, with mean values calculated from three replicates. (B) Overview on modular level M T2S constructs generated for the analysis of xpsE and xpsD mutants. Genes are represented by arrows, promoters by orange boxes. Deletions are indicated by black boxes. For complementation studies, xpsE and xpsD were reinserted at position 1 of the corresponding modular T2S gene cluster constructs as indicated.

https://doi.org/10.1371/journal.ppat.1013008.g002

We next analysed the suitability of the modular T2S gene cluster for mutational approaches and complementation studies in X. euvesicatoria strain 85-10∆xps. We previously showed that T2S depends on the secretin XpsD and the ATPase XpsE [29,30]. As proof of concept, we individually replaced the corresponding gene modules by short linker sequences and introduced the reassembled modified T2S gene clusters into strain 85-10∆xps (Fig 2B). The absence of xpsE or xpsD led to reduced virulence and extracellular protease activity as expected (Fig 2A) [29]. For complementation studies, xpsE and xpsD were inserted under control of their native promoters at position 1 upstream of the modular xps gene clusters in the final level M constructs (Fig 2B). The corresponding expression constructs fully restored virulence and extracellular protease activity in strain 85-10∆xps, suggesting that xpsE and xpsD were functional when expressed next to the modular xps gene cluster under control of their native promoters in the final level M construct (Fig 2A). Taken together, these experiments demonstrate the suitability of the modular T2S gene cluster for mutant and complementation studies in X. euvesicatoria.

Generation of single gene deletions in the modular T2S gene cluster reveals an essential role of predicted assembly platform components

According to similarities between Xps and Gsp proteins, the T2S system from X. euvesicatoria consists of the OM secretin XpsD, a central periplasmic pseudopilus and an assembly platform containing the IM protein XpsF which is associated with the cytoplasmic ATPase XpsE and surrounded by XpsC, XpsL and XpsM [3]. Complexes of XpsC, XpsL and XpsM homologs have been detected in E. amylovora and X. campestris pv. campestris, however, their structure and contribution to the formation of the assembly platform is largely unknown [38,39]. In the present study, we analysed the functions and interactions of XpsC, XpsL and XpsM using the modular T2S gene cluster from X. euvesicatoria. For this, we generated level M expression constructs in which the individual genes were replaced by short linker sequences (S3 Fig). For complementation studies, individual assembly platform genes were reinserted under control of the native xpsG promoter at position 1 of the corresponding modular gene clusters (S3 Fig). When analysed in strain 85-10∆xps, modular T2S gene cluster constructs deleted in xpsC, xpsL or xpsM led to reduced virulence and extracellular protease activity, indicative of a loss of T2S (Fig 3A and 3B). In cis expression of xpsC and xpsM in the corresponding mutant T2S gene clusters restored the wild-type phenotype, suggesting that reduced virulence and loss of detectable extracellular protease activity were specifically caused by the absence of XpsC or XpsM and not by a polar effect of the deletions on neighbouring genes (Fig 3A). In case of the xpsL mutant, however, reinsertion of xpsL under control of the native promoter did not restore the wild-type phenotype in the xpsL mutant (Fig 3B). To exclude that the lack of complementation was caused by a polar effect of the deletion in xpsL on the expression of downstream genes in the xps-T2S gene cluster, we generated an additional xpsL mutant containing a nonsense mutation at codon position 147 of xpsL and repeated the infection and complementation studies. Similar to the xpsL deletion, the stop codon in xpsL resulted in reduced virulence and extracellular protease activity when the corresponding modular xps-T2S gene cluster construct was analysed in strain 85-10∆xps (Fig 3B). However, in cis expression of xpsL under control of the xpsG operon promoter in the corresponding modular expression construct did not complement the mutant phenotype (Fig 3B). To investigate whether the lack of complementation was caused by alterations in the expression level of xpsL, we replaced the predicted Shine Dalgarno sequence in the xpsG promoter by the corresponding sequence which is present upstream of xpsL (Fig 3B). The resultling xpsL expression cassette restored the wild-type phenotype, suggesting that the function of XpsL is tightly linked to its level of translation which is controlled by the Shine Dalgarno sequence (Fig 3B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The predicted assembly platform components XpsC, XpsL and XpsM are essential for T2S in X. euvesicatoria.

(A) Complementation studies with xpsC and xpsM mutants. Infection and protease activity assays were performed with derivatives of X. euvesicatoria strain 85-10∆xps with modular expression constructs containing the wild-type (WT) T2S gene cluster (pT2S) or derivatives thereof lacking xpsC (∆xpsC) or xpsM (∆xpsM). For complementation studies, expression cassettes encoding XpsC (+XpsC) or XpsM (+ XpsM) under control of the xpsG promoter were inserted at position 1 of T2S gene cluster constructs lacking xpsC and xpsM, respectively. For infection studies, bacteria were infiltrated at a density of 2 × 10⁷ CFU (colony-forming units) ml⁻¹ into leaves of susceptible pepper plants and disease symptoms were photographed 8 dpi (days post inoculation). Dashed lines indicate the infiltrated areas. For the analysis of extracellular protease activity, bacteria were grown on 1% (w/v) milk plates for 2 days and halo formation was documented. The numbers refer to the width of the halos in millimeters, with mean values calculated from three replicates. (B) xpsL is essential for T2S and depends on its native Shine Dalgarno sequence for efficient expression. The contribution of xpsL to virulence and extracellular protease activity was analysed in derivatives of strain 85-10∆xps containing the wild-type T2S gene cluster or derivatives thereof lacking xpsL (∆xpsL) with or without expression cassettes encoding XpsL under control of the native xpsG promoter (+XpsL). In a second construct, the predicted Shine Dalgarno (SD) sequence of xpsG (indicated in bold letters) was exchanged by the corresponding sequence upstream of xpsL (XpsL_+SD) as illustrated. The translation initiation codon (ATG) is underlined. For complementation studies, bacteria were infiltrated into leaves of susceptible pepper leaves and were grown on milk-containing agar plates. Protease activity was monitored as described in (A).

https://doi.org/10.1371/journal.ppat.1013008.g003

Structural modeling reveals a trimeric XpsCLM complex which is conserved in different bacterial species

XpsC, XpsL and XpsM are predicted IM proteins with periplasmic domains and likely surround the central XpsF component of the assembly platform. To predict the architecture of the XpsCLM complex in the context of the T2S system, we modeled the corresponding protein structures in silico using the AlphaFold2 structure prediction algorithm and the molecular visualization program UCSF ChimeraX [40,41]. When included in one structural model, XpsC, XpsL and XpsM form a trimeric complex with a pTM (predicted template modelling) value of 0.56 and an ipTM (inter-chain predicted template modelling) value of 0.62 (see Material and Methods, Figs 4 and S6). The pTM score assesses the confidence in the relative arrangement of domains within a single protein chain whereas the ipTM score evaluates the reliability of predicted conformations and interactions between protein chains in a multimer structure. pTM and ipTM values range from 0 to 1, and values closer to 1 indicate higher confidence. According to the AlphaFold2 predictions, XpsC, XpsL and XpsM insert into the IM with an N-terminal transmembrane helix followed by a large C-terminal periplasmic region. The cytoplasmic domain of XpsL is predicted to associate with the putative ATPase XpsE, albeit with lower pTM and ipTM values of 0.48 and 0.51, respectively (Figs 4 and S6). The periplasmic regions of XpsL and XpsM adopt potential ferredoxin-like folds containing two α helices and an anti-parallel β sheet (S4 Fig). According to the structural model, XpsL and XpsM interact with each other via their transmembrane and periplasmic regions including the predicted ferredoxin-like fold as was previously described for GspL and GspM from Klebsiella oxytoca and Dickeya dadantii [14,39]. Additionally, both XpsL and XpsM likely interact with the periplasmic region of XpsC (S5 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. XpsCLM form a predicted trimeric complex which is structurally conserved in different bacterial species.

The structure of a complex consisting of the periplasmic region of XpsD (N1 and N0 domains), XpsC, XpsL, XpsM and XpsE was predicted using the AlphaFold2 algorithm and the molecular visualization program UCSF ChimeraX. Proteins are represented in different colours as indicated. The HR and 2P domains of XpsC, the periplasmic (peri) and cytoplasmic (cyto) domains of XpsL and XpsM and the position of the inner membrane (IM) are indicated. The N termini of XpsC, XpsL and XpsM are located in the cytoplasm. XpsE consists of two N-terminal (N1E and N2E) and one C-terminal (CTE) domain. Ribbon diagrams for the XpsCLM complex and for single proteins are shown in S4 Fig. The per-residue model confidence score (pLDDT) of single protein complexes is shown in S6 Fig. AlphaFold2 predictions are also presented for the GspCLM proteins XcpPYZ from P. aeruginosa strain PAO1, OutCLM from D. dadantii strain 3937 and EpsCLM from V. cholerae strain N16961. HR and 2P domains of GspC proteins are indicated. The 2P domains of OutC and EpsC contain a PDZ fold. Numbers refer to the amino acid identities with the corresponding XpsCLM proteins from X. euvesicatoria. The following proteins were used for the topology models: XcpP (accession number CAA48581), XcpY (accession number AAG06484) and XcpZ (accession number AAG06483) from P. aeruginosa strain PAO1, OutC (accession number CAA46369), OutL (accession number ADM99368) and OutM (accession number ADM99367) from D. dadantii strain 3937 and EpsC (accession number P45777), EpsL (accession number P45782) and EpsM (accession number P41851) from V. cholerae strain N16961.

https://doi.org/10.1371/journal.ppat.1013008.g004

XpsC was previously designated XpsN but is a homolog of GspC [32]. The periplasmic region of XpsC contains an HR domain which is formed by β sheets and is also present in other GspC proteins including XcpP from P. aeruginosa, OutC from D. dadantii and EpsC from V. cholerae (Figs 4 and S5). The HR domain of XpsC is predicted to interact with the periplasmic domain of the secretin XpsD (Fig 4). A second periplasmic region, referred to as 2P domain, contains a single predicted α helix as described for XcpP from Pseudomonas aeruginosa [42]. In contrast, a PDZ fold containing α helices and β sheets is present in the 2P domains of the GspC proteins OutC and EpsC from D. dadantii and V. cholerae [12,43] (Figs 4 and S5). PDZ domains also occur in T2S system-unrelated proteins and are often involved in protein-protein interactions [44,45]. Interestingly, AlphaFold modeling revealed a similar structure of trimeric GspCLM complexes from the bacterial pathogens P. aeruginosa (XcpPYZ), D. dadantii (OutCLM) and V. cholerae (EpsCLM), despite limited amino acid sequence similarities of the corresponding proteins. The pTM values of 0.42 for all three complexes and the ipTM values of 0.41 (for XcpPYZ), 0.42 (for OutCLM) and 0.48 (for EpsCLM) were slightly lower than the corresponding values for XpsCLM from X. euvesicatoria (see above). Notably, however, significant portions of the GspC homologs in all models were highly disordered, which might result in lower pTM and ipTM values (see S6 Fig). Taken together, we conclude from these structure predictions that the architecture of XpsCLM complexes is likely conserved across highly divergent T2S sytems (Figs 4 and S6).

Bacterial two-hybrid studies revealed interactions between XpsC, XpsL, XpsM and XpsE

To experimentally validate the predicted interactions between XpsC, XpsL, XpsM and XpsE, we performed protein-protein interaction studies using the bacterial adenylate cyclase two-hybrid (BACTH) system. This method is based on the reconstitution of the catalytic domain of the adenylate cyclase (Cya) from two subdomains (T18 and T25) in the bacterial cytoplasm and is suitable for the analysis of transmembrane proteins [46,47]. The T18 and T25 subdomains of Cya were analysed as N-terminal fusion partners of predicted assembly platform components to allow their localization in the cytoplasm (Fig 4). According to topology predictions, the C-terminal regions of XpsC, XpsL and XpsM are located in the periplasm (Fig 4). A C-terminal T18 or T25 fusion partner would, therefore, prevent the formation of a functional catalytic Cya domain in the cytoplasm. For the putative ATPase XpsE, however, which is located in the cytoplasm, T18 and T25 subdomains were analysed as both N- or C-terminal fusion partners. Immunoblot analysis of bacterial cell extracts showed that all proteins were stably synthesized (S7 Fig). For BACTH assays, T18 and T25 fusions were co-expressed in the E. coli reporter strain DHM1, which lacks the native cya gene. Protein-protein interactions led to cAMP production and thus to lacZ gene expression which was monitored when bacteria were grown on reporter plates containing X-Gal. We detected self-interactions of XpsE and XpsC, interactions of XpsE with XpsL as well as of XpsC with both XpsM and XpsL (Fig 5A). The analysed fusion proteins did not bind to the T18 or T25 subdomains alone, suggesting that the observed interactions were specific (Fig 5A). We did not detect interactions between XpsM and XpsL in strain DHM1 and, therefore, repeated the BACTH assays using the more sensitive reporter strain BTH101, which grows faster than strain DHM1 and leads to enhanced LacZ activity [46]. When analysed in strain BTH101, XpsM and XpsL interacted with themselves and with each other, suggesting that they can form homo- and heterooligomeric complexes (Fig 5A). To confirm the interaction between XpsE and XpsL, we performed in vitro GST (glutathione S-transferase) pull-down assays. For this, GST and GST-XpsE were immobilized on glutathione sepharose and incubated with bacterial lysates containing an N-terminally c-Myc epitope-tagged derivative of XpsL. Immunoblot analyses showed that c-Myc-XpsL coeluted with GST-XpsE but not with GST alone, suggesting a direct interaction between XpsE and XpsL (Fig 5B). Our results are in agreement with the previous finding that the ATPase XpsE from X. campestris pv. campestris oligomerizes and interacts with XpsL [48].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. In vitro protein-protein interaction studies with predicted assembly platform components from X. euvesicatoria.

(A) BACTH assays with XpsC, XpsL, XpsM and XpsE. T25 and T18 fusions of XpsC, XpsL, XpsM and XpsE as indicated were analysed in the E. coli reporter strains DHM1 and BTH101 as indicated. Transformants were incubated on indicator plates containing X-Gal and IPTG and representative colonies were photographed after five days. All fusions proteins were stably synthesized as was shown by immunoblot analysis (S7 Fig). Experiments were performed at least three times with similar results. One representative colony is shown. n.t., not tested. (B) In vitro interaction of XpsL and XpsE. GST and GST-XpsE were immobilized on a glutathione sepharose matrix and incubated with a bacterial lysate containing c-Myc-XpsL. The total cell extract (TE) and eluted proteins (eluate) were analysed by immunoblotting using c-Myc- or GST-specific antibodies. Signals corresponding to GST and GST-XpsE are indicated by asterisks. The experiment was performed three times with similar results. One representative result is shown.

https://doi.org/10.1371/journal.ppat.1013008.g005

Stability of XpsC and XpsL depends on XpsM

Next, we analysed predicted components of the assembly platform in vivo in the context of the native T2S system. To detect XpsL, we introduced a c-Myc epitope at the N terminus because a C-terminal epitope might interfere with the predicted interaction of XpsL with XpsC and XpsM. The corresponding expression cassette containing c-myc-xpsL under control of the xpsG promoter was inserted into the flanking region of modular xps gene cluster constructs, which were mutated in xpsL or additionally lacked xpsC, xpsM, xpsE, xpsF or xpsD (S3 Fig). The resulting constructs were transferred into strain 85-10∆xps and assessed for T2S system activity and the synthesis of c-Myc-XpsL.

Protease activity assays showed that c-Myc-XpsL restored extracellular protease activity in the xpsL mutant, suggesting that it was functional (Fig 6A). When analysed by immunoblotting, c-Myc-XpsL was detected at the expected molecular size of 48 kDa (Fig 6A). An additional protein at a size of approximately 37 kDa, hereafter referred to XpsL’, presumably represents a degradation or cleavage product of c-Myc-XpsL lacking the C-terminal protein region (Fig 6A). Reduced amounts of c-Myc-XpsL were detected in strains lacking the complete xps-T2S gene cluster (∆xpsE-D) or xpsM, suggesting that XpsM contributes to XpsL stability (Fig 6A). Interestingly, however, in contrast to the full-length c-Myc-XpsL protein, the levels of c-Myc-XpsL’ appeared to be unaffected, suggesting that the N-terminal region of XpsL is stable in the absence of XpsM (Fig 6A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. In vivo crosslinking experiments reveal the presence of a trimeric XpsCLM complex which assembles independently of the secretin XpsD.

(A) Detection of XpsL-containing complexes after in vivo crosslinking depends on XpsC and XpsM. Derivatives of strain 85-10∆xps (∆xps) containing the wild-type (WT) modular T2S gene cluster (+pT2S) or derivatives thereof deleted in xpsL, xpsE-D, xpsC, xpsM, xpsE, xpsF or xpsD and encoding c-Myc-XpsL as indicated were grown in NYG medium. Equal amounts of cell cultures were centrifuged and cells were either resuspended in Laemmli buffer at 99°C (total extract) or incubated with formaldehyde and resuspended in Laemmli buffer at 37°C (crosslinked). Proteins were analysed by immunoblotting using antibodies specific for the c-Myc epitope or the RNA polymerase β to ensure equal loading. Signals corresponding to the expected sizes of c-Myc-XpsL (48 kDa; black arrow), a putative degradation product of c-Myc-XpsL (XpsL’, 37 kDa; white arrow) and a protein complex (100 kDa; asterisk) are indicated. The molecular weight of predicted components of the assembly platform is indicated below the blots. The dotted line indicates that the blot has been cut to exclude an additional lane. For the analysis of T2S system activity, bacteria were grown on milk protein-containing agar plates to demonstrate extracellular protease activity. Halo formation was documented two days after incubation. Experiments were performed three times with similar results. One representative example is shown. The numbers refer to the width of the halos in millimeters, with mean values calculated from three replicates. (B) The formation of XpsC-containing complexes depends on XpsM. Derivatives of strain 85-10∆xps (∆xps) containing the wild-type modular T2S gene cluster or derivatives thereof deleted in xpsC, xpsE-D, xpsM, xpsL or xpsD and encoding XpsC-c-Myc as indicated were grown in NYG medium and analysed as described in (A). The signals corresponding to c-Myc-XpsC (black arrow), a potential N-terminally extended derivative or c-Myc-XpsC (XpsC”, white arrow) and c-Myc-XpsC-specific complexes (asterisks) are indicated.

https://doi.org/10.1371/journal.ppat.1013008.g006

For the analysis of XpsC, a gene module encoding XpsC-c-Myc under control of the xpsG promoter was inserted into level M constructs containing modular xps-T2S gene clusters with deletions in single (xpsC) or multiple xps genes as described above (S3 Fig). XpsC-c-Myc restored protease activity in a strain lacking the native xpsC gene in the modular T2S gene cluster, suggesting that the C-terminal c-Myc epitope did not significantly interfere with XpsC function (Fig 6B). When analysed by immunoblotting in X. euvesicatoria protein extracts, XpsC-c-Myc was detected at a size of approximately 40 kDa instead of the expected size of 32 kDa (28 kDa for XpsC and 4 kDa for the c-Myc epitope) (Fig 6B). Furthermore, an additional XpsC-c-Myc-specific signal (referred to as XpsC”-c-Myc) was present at a size of 55 kDa (Fig 6B). Inspection of the xpsG promoter, which is the native promoter of the predicted xpsG-D operon and was used to express xpsC (see above, Fig 1), revealed the presence of additional potential start codons upstream of the translation initiation site of xpsC. Alternative translation initiation might lead to the synthesis of an N-terminally extended derivative of XpsC-c-Myc and could thus explain the detection of XpsC”-c-Myc. In agreement with this hypothesis, the introduction of a nonsense mutation upstream of the start codon of xpsC abolished the detection of XpsC” (S8 Fig). These experiments were performed with an N-terminally c-Myc epitope-tagged derivative of XpsC, which was functional and restored the extracellular protease activity in the xpsC mutant (S8 Fig).

Identification of the XpsCLM complex by in vivo crosslinking experiments

To investigate a possible complex formation by XpsC and XpsL in X. euvesicatoria, we performed crosslinking experiments with formaldehyde (FA) which penetrates cells and often interacts with the amino group of lysine. FA introduces inter- or intramolecular covalent crosslinks between amino acid residues of proteins which are in close proximity (2.3–2.7 Å) to each other [49,50]. Immunoblot analysis of cell extracts containing c-Myc-XpsL revealed a specific signal at approximately 100 kDa and additional less prominent signals at 60 and 70 kDa after FA treatment, suggesting that c-Myc-XpsL integrates into oligomeric protein complexes (Fig 6A). XpsL-containing complexes were also observed upon deletion of xpsE, xpsF or xpsD but were significantly reduced in the absence of the entire xps-T2S gene cluster as well as in xpsC or xpsM mutants (Fig 6A). Given the contribution of XpsC and XpsM to complex formation, we speculate that the signal at 100 kDa corresponds to the predicted trimeric complex of c-Myc-XpsL (48 kDa), XpsC (28 kDa) and XpsM (24 kDa). This hypothesis is supported by the results of our BACTH assays which showed interactions of XpsC, XpsM and XpsL (see above, Fig 5A).

In addition to c-Myc-XpsL, we analysed complex formation by XpsC-c-Myc after FA crosslinking. Immunoblot analyses revealed the presence of XpsC-c-Myc-specific protein complexes at molecular sizes of approximately 100 and 120 kDa after FA treatment (Fig 6B). The signal at 120 kDa likely resulted from incorporation of the N-terminally extended version of XpsC-c-Myc into the complex because it was absent upon introduction of a nonsense mutation upstream of the xpsC start codon (see above, S8 Fig). The 100 kDa complex containing XpsC-c-Myc was also detected in a secretin mutant but not in a strain lacking xpsE-D (Fig 6B). Furthermore, deletion of xpsM abolished complex formation, whereas the absence of xpsL led to a smaller complex at a size of approximately 60 kDa (Fig 6B). Given that the 100 kDa complex corresponds to the size of XpsC-c-Myc (40 kDa), XpsL (37 kDa) and XpsM (24 kDa) and was detected after crosslinking for both c-Myc-XpsL and XpsC-c-Myc, we speculate that XpsL and XpsC associate with XpsM in X. euvesicatoria and form the predicted XpsCLM trimer. Future crosslinking and mass spectrometry analysis will provide insights into complex formation in vivo and will help to identify the interaction sites of the predicted assembly platform components.

The XpsCLM complex and the secretin assemble independently of each other

It was previously reported that the HR domain of GspC proteins interacts with the periplasmic N0 domain of the secretin and thus connects the IM assembly platform to the OM secretin channel [11]. Our crosslinking experiments with XpsC-c-Myc and c-Myc-XpsL suggested that XpsD is dispensable for the formation of the XpsCLM complex (see above, Fig 6A and 6B). To investigate whether the assembly platform is required for secretin assembly, we performed fluorescence microscopy-based in vivo localization studies with XpsD and the red fluorescent reporter mCherry which is active in the periplasm [51]. Fluorescent reporter fusions allow to study the assembly of protein secretion systems and were previously used to analyse the formation of the cytoplasmic sorting platform of the T3S system from X. euvesicatoria [36]. Reporter fusion proteins, which are located in the cytoplasm, produce fluorescent signals throughout the entire bacterial cells, whereas reporter fusions integrated into the protein secretion system allow the detection of fluorescent foci which are indicative of complex formation. mCherry was used as C-terminal fusion partner of XpsD because the N terminal region of XpsD is part of a cleavable Sec signal. According to topology predictions, both the N- and C-termini of XpsD are located in the periplasm. Complex formation of XpsD-mCherry should result in the formation of fluorescent foci at the cell periphery as shown previously for PulD of K. oxytoca [52].

Prerequisite for the formation of fluorescent foci is the stability and functionality of the XpsD-mCherry fusion protein. For functional studies, the native xpsD gene was deleted from the modular T2S gene cluster and an expression cassette encoding XpsD-mCherry under control of the native promoter was inserted at position 1 of the corresponding level M construct. When bacteria were analysed by infection studies, deletion of xpsD led to reduced bacterial virulence in susceptible pepper plants as expected [29] (Fig 7A). XpsD-mCherry restored the wild-type phenotype when encoded in the flanking region of the modular T2S gene cluster lacking xpsD, suggesting that the mCherry fusion partner did not interfere with XpsD function (Fig 7A). Immunoblot analyses of bacterial cell extracts showed that XpsD-mCherry was stably synthesized and forms high molecular weight complexes which were detected in the stacking gel and presumably correspond to XpsD-mCherry multimers (Fig 7B). Complex formation was also detected in the absence of xpsE – xpsD genes, suggesting that the secretin assembles independently of other components of the T2S system (Fig 7B). Similar findings were observed for a C-terminally c-Myc epitope-tagged derivative of XpsD which was expressed in trans under control of a native Xe-specific promoter in xpsD and xpsE-D deletion mutants (S9 Fig). XpsD-c-Myc complemented the xpsD mutant phenotype in planta, restored extracellular protease activity and formed high molecular weight complexes, even in the absence of other T2S system components, as was shown by immunoblot analysis of bacterial cell extracts (S9 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Assembly of the OM secretin XpsD is independent of T2S system components.

(A) Complementation studies with XpsD-mCherry in X. euvesicatoria. Strain 85-10∆xps carrying the wild-type (WT) modular T2S gene cluster (pT2S) or derivatives thereof deleted in xpsD (∆xpsD) or xpsE – xpsD (∆xpsE-D) and encoding XpsD-mCherry as indicated were infiltrated at a density of 2 × 10⁷ CFU (colony-forming units) ml⁻¹ into leaves of susceptible pepper plants and disease symptom formation was documented 8 dpi (days post inoculation). Dashed lines indicate the infiltrated areas. For the analysis of extracellular protease activity, bacteria were grown on 1% (w/v) milk plates. Halo formation resulting from the degradation of milk proteins was photographed after two days. Experiments were performed three times with similar results. Results from one representative experiment are shown. The numbers refer to the width of the halos in millimeters, with mean values calculated from three replicates. (B) XpsD-mCherry is stably synthesized and forms complexes. Bacterial strains as described in (A) were cultivated in NYG medium and equal amounts of protein extracts were analysed by SDS-PAGE and immunoblotting using mCherry-specific antibodies. The signals labeled with asterisks correspond to XpsD-mCherry-specific complexes which were detected in the stacking gel. The dashed line indicates different exposure times for the lower and upper part of the blot. Blots were reprobed with antibodies directed against GroEL, which was analysed as a loading control. (C) XpsD-mCherry forms fluorescent foci. For fluorescence microscopy studies, strain 85-10∆xps (∆xps) containing modular T2S gene cluster constructs (pT2S) deleted in xpsD (∆xpsD) or in the entire cluster (∆xpsE-D) and encoding XpsD-mCherry as indicated were incubated in NYG medium for 2 hours. Foci formation was analysed by fluorescence microscopy using a Zeiss LSM 780 AxioObserver Z1 microscope at 60 x magnification. The scale bar corresponds to 3 µm. Bright field images are shown in the upper panel, the signals from the fluorescent channel for mCherry in the lower panel. Fluorescent foci and their positions in the bright field images are indicated with white arrows. The experiment was performed three times with similar results. One representative image for each strain from one experiment is shown. (D) Quantitative analysis of foci formation. Fluorescent foci were counted in at least 100 cells per strain in three different transconjugants. Mean values and standard deviations are shown as percentage of bacterial cells. Asterisks indicate a significant difference between the number of foci with a p-value < 0.05 based on the results of a chi-squared test.

https://doi.org/10.1371/journal.ppat.1013008.g007

To confirm XpsD complex formation in X. euvesicatoria T2S mutants, we performed fluorescence microscopy studies with bacteria containing XpsD-mCherry in the presence or absence of other Xps proteins. One to three fluorescent foci per cell were detected in both strains, indicative of the formation of XpsD-mCherry oligomers and thus the assembly of the secretin (Fig 7C and 7D). Foci formation was specific for the presence of XpsD and was not detected in bacteria containing mCherry fused to the effector protein XopB (S10 Fig). No significant differences in foci formation were detectable in the presence or absence of the modular xps-T2S gene cluster (Fig 7C and 7D). This confirms the above hypothesis that the secretin assembles independently of other components of the T2S system.

Bacterial strains and growth conditions

Bacterial strains and plasmids used in this study are listed in S1 Table. Escherichia coli strains were cultivated at 37°C in lysogeny broth (LB) or terrific broth (TB) medium and X. euvesicatoria strains at 30°C in nutrient-yeast extract-glycerol (NYG) medium. Plasmids were introduced into chemically competent E. coli strains by heat shock and into X. euvesicatoria strains by electroporation. Antibiotics were added to the media at the following final concentrations: ampicillin, 100 μg ml⁻¹; gentamycin 15 μg ml⁻¹; kanamycin, 25 μg ml⁻¹; rifampicin, 100 μg ml⁻¹ and spectinomycin, 100 μg ml⁻¹.

Deletion of the chromosomal xps-T2S gene cluster in X. euvesicatoria strain 85-10

To generate strain 85-10Δxps, in which the complete T2S gene cluster is deleted, DNA fragments flanking the xps gene cluster were amplified from X. euvesicatoria by PCR using the primer pairs Δxps-5’-fw/ Δxps-5’-rv and Δxps-3’-fw/ Δxps-3’-rv (S2 Table). PCR amplicons were cloned into the suicide vector pOGG2 by Golden Gate cloning using BsaI and ligase [33,66]. The resulting deletion construct pOGG2Δxps was introduced into X. euvesicatoria strain 85-10 by triparental conjugation using the helper plasmid pRK2013 and transconjugants were selected as described previously [67]. Double crossovers resulted in strain 85-10Δxps which contains a 11,750-bp deletion in the xps-T2S gene cluster.

Generation of expression constructs

To generate expression constructs for BACTH assays, xps genes were amplified by PCR from X. euvesicatoria using gene-specific primers (S2 Table). PCR amplicons were cloned into vector pAGM9121 using BpiI and T4 ligase or into vectors pICH41021 and pUC57 by blunt-end cloning using SmaI and EcoRV, respectively. Genes were subsequently transferred to the Golden Gate-compatible BACTH vectors pKT25_GG, pKNT25_GG, pUT18_GG or pUT18C_GG using BsaI and ligase as described previously [33] (S1 Table). For GST pull-down assays, xpsE and xpsL were amplified by PCR and subcloned into vectors pICH41021 (xpsL) and pUC57 (xpsE) using SmaI or EcoRV and ligase. xpsL was subsequently cloned into vector pBRNM downstream of an N-terminal c-Myc epitope-encoding sequence using BsaI and ligase. xpsE was assembled by Golden Gate cloning with an expression cassette encoding GST under control of the ptac promoter in the final destination vector pBRM-Pstop. All constructs are listed in S1 Table.

To generate the expression construct encoding XpsD-c-Myc, the 5’ and 3’ region of xpsD was amplified by PCR from X. euvesicatoria strain 85-10 and the corresponding amplicons were individually cloned into vector pICH41021 using SmaI and ligase in a cut/ligation reaction. The resulting modules were assembled with the promoter of gene XCV4361 in vector pBRM-P by Golden Gate cloning, thus generating construct pB-PxpsD (S1 Table).

Generation of the modular xps-T2S gene cluster by Golden Gate cloning

The xps-T2S gene cluster was assembled from single promoter and gene fragments by Golden Gate-based modular cloning [35,53]. For this, the native promoters and coding sequences of xpsE and xpsF were amplified by PCR using gene-specific primers listed in S2 Table. The resulting PCR amplicons were cloned into the level −1 vector pAGM1311 using BsaI and T4 ligase, resulting in constructs pT2S001 (xpsE and promoter) and pT2S005 (xpsF and promoter). Similarly, the native promoter of the xpsG-xpsD operon as well as xpsG, xpsH, xpsI, xpsJ, xpsK, xpsL, xpsM, xpsC and xpsD were amplified by PCR and cloned into the level −2 vector pAGM9121 using BpiI and T4 ligase, resulting in constructs pT2S011 to pT2S026 (Fig 1). Constructs pT2S011 – pT2S014, pT2S016 – pT2S020 and pT2S022 – pT2S026 were assembled by Golden Gate cloning in the level −1 vector pAGM1311 using BsaI and T4 ligase, thus generating constructs pT2S015 (xpsG, xpsH and xpsI including the native promoter), pT2S021 (xpsJ, xpsK and xpsL) and pT2S028 (xpsM, xpsC and xpsD). For the generation of level 0 constructs, gene modules from constructs pT2S001 (xpsE and promoter), pT2S005 (xpsF and promoter) as well as pT2S015, pT2S021 and pT2S028 were inserted into vector pICH41331 using BpiI and ligase, thus leading to the level 0 constructs pT2S002 (xpsE and promoter), pT2S006 (xpsF and promoter) and pT2S030 (xpsG - xpsD operon and promoter). Inserts of level 0 constructs were ligated into different level 1 vectors which determine the positions of the individual modules in the final level M construct [53]. This resulted in constructs pT2S32 (xpsE and promoter in vector pICH47742), pT2S34 (xpsF and promoter in vector pICH47751) and pT2S36 (xpsG-xpsD operon and promoter in vector pICH47761) (Fig 1). Additional level 1 constructs pICH54011, pICH54055 and pICH50914 contained dummy modules for insertion at positions 1 and 5, and an end-linker (Fig 1). The final level M construct pT2S038 was constructed by assembly of level 1 modules in vector pAGM8031 using BpiI and T4 ligase. The end-linker connects the 3’ end of the last module to the destination vector. Constructs used for modular cloning are listed in S1 Table. All constructs were verified by test restriction with appropriate enzymes and all plasmids containing PCR amplicons were sequenced by Sanger sequencing. Deletions, expression cassettes and epitope- or mCherry-encoding sequences were introduced into the modular xps-T2S gene cluster as described in S1 Appendix.

Plant material and infection studies

Capsicum annuum cultivar Early Cal Wonder (ECW) plants were grown at 25°C for 16 h of light and 8 h of darkness at 60 – 70% relative humidity. For syringe infections, X. euvesicatoria strains were infiltrated with a needleless syringe into the lower side of pepper leaves at concentrations of 2 × 10⁷ CFU (colony-forming units) ml⁻¹ in 1 mM MgCl₂. Symptoms were photographed 6 – 8 days post inoculation (dpi). For dip infections, leaves were incubated in 1 mM MgCl₂ solutions containing X. euvesicatoria at concentrations of 2 × 10⁷ CFU ml⁻¹ and 0.02% (v/v) Silwet for at least 30 seconds. The formation of bacterial spot symptoms was monitored 4 – 7 weeks after dipping. Spots were counted in seven infected leaf areas per strain. Statistical significance was determined using an ANOVA with post-hoc HSD (P < 0.001). All experiments were performed at least three times with similar results. Representative results from one experiment are shown.

Protease activity assays

For the analysis of extracellular protease activity, bacteria were grown over night in NYG medium with respective antibiotics. Fourty microlitres of the over-night cultures adjusted to a density of 10⁹ CFU ml⁻¹ were pipetted into holes, which were punched out from NYG agar plates containing 1% (w/v) skimmed milk as described previously [29]. Plates were incubated at 30°C for 2 days and bacteria were removed prior to documentation of halo formation. The width of the halos was measured in millimeters and the mean values were calculated from three replicates. Experiments were performed three times with similar results. One representative result is shown.

Immunoblot analysis of bacterial protein extracts

For immunoblot analysis of protein extracts, bacterial cell pellets were resuspended in Laemmli buffer, denatured by boiling and analysed by SDS-PAGE and immunoblotting, using antibodies specific for the c-Myc epitope (polyclonal antibody from rabbit, Sigma-Aldrich), the FLAG epitope (monoclonal antibody from mouse, Sigma-Aldrich), mCherry (polyclonal antibody from rabbit, Abcam Limited), GroEL (polyclonal antibody from rabbit, Enzo Life Sciences), RNA polymerase β (monoclonal antibody from mouse, Invitrogen), and GST (glutathione S-transferase; polyclonal antibody from goat, Cytiva), respectively. Horseradish peroxidase-labelled anti-mouse, anti-rabbit and anti-goat antibodies were used as secondary antibodies. Binding of antibodies was visualized by enhanced chemiluminescence using a Vilber FUSION-FX6 chemiluminescence imager. Results were reproduced at least twice. One representative blot is shown.

Analysis of protein-protein interactions using the BACTH system

For protein-protein interaction studies, we used a modified version of the Euromedex BACTH system which allowed Golden Gate cloning of genes of interest in fusion with T18- and T25-encoding fragments [68] (see above). For the analysis of protein synthesis, BACTH constructs were introduced into E. coli strain JM109 and gene expression was induced by IPTG (isopropyl-β-D-thiogalactopyranoside) for 2 h at 37°C after the optical densities of the bacterial cultures at 600 nm (OD₆₀₀) had reached values between 0.6 and 0.8. Cells were resuspended in Laemmli buffer and analysed by immunoblotting using a FLAG epitope-specific antibody. For interaction studies, the expression constructs encoding T18 and T25 fusion proteins were introduced into the E. coli reporter strains DHM1 or BTH101 as indicated and transformants were cultivated on LB agar plates containing kanamycin and gentamycin [69]. At least three transformants for each interaction were grown over night in liquid LB medium with appropriate antibiotics and 2 µl of each over-night culture were spotted on LB agar plates containing kanamycin, gentamycin, X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranosid; 40 μg/ml) and 2 mM IPTG (isopropyl-β-D-thiogalactopyranoside). Plates were incubated for up to seven days at room temperature and the colour of bacterial colonies was documented by photographs. Experiments were performed at least three times with bacteria from independent transformations. One representative colony is shown.

GST pull-down assays

For GST pull-down assays, E. coli TOP10 cells with expression constructs encoding GST, GST-XpsE and c-Myc-XpsL were grown in LB medium at 37°C. When the cultures reached an OD₆₀₀ of 0.6 – 0.8, gene expression was induced by addition of IPTG at a final concentration of 2 mM for two hours. Cells were harvested by centrifugation, resuspended in 1 x PBS buffer and lysed with a French press. Insoluble cell debris were removed by centrifugation and soluble GST and GST-XpsE fusion proteins were immobilized on a glutathione sepharose matrix according to the manufacturer’s instructions (GE Healthcare). The glutathione sepharose with immobilized GST and GST fusion proteins was washed with 1 x PBS buffer and incubated with bacterial lysates containing c-Myc-XpsL for 1 h at 4°C on an overhead shaker. Unbound proteins were removed by washing in 1 x PBS. Bound proteins were eluted with Laemmli buffer and analysed by SDS-PAGE and immunoblotting, using GST- and c-Myc epitope-specific antibodies. Experiments were repeated at least two times with similar results.

In vivo crosslinking experiments

In vivo protein complex formation was analysed after crosslinking with FA as described [70]. Briefly, X. euvesicatoria strains were grown over night in NYG medium, resuspended at an OD₆₀₀ of 0.3 in fresh NYG medium and incubated on a tube rotator at 30°C. After 3 – 4 hours, cells from 1 ml cultures were collected by centrifugation, resuspended in Laemmli buffer and boiled for 10 minutes (total cell extract). In parallel, cells from 1 ml of cultures were resuspended in 1 ml of 1 mM sodium phosphate buffer (pH 6.8) containing 1% (v/v) FA. After incubation for 20 minutes at room temperature, cells were collected by centrifugation, washed in 1 ml of 1 mM sodium phosphate buffer (pH 6.8) without FA and resuspended in Laemmli buffer. Protein samples were incubated at 37°C for 20 minutes to maintain FA cross-links. Total cell extracts and crosslinked samples were subsequently analysed by SDS-PAGE and immunoblotting, using antibodies specific for the c-Myc epitope and the RNA polymerase β, which was analysed as loading control. Experiments were performed three times with similar results. One representative blot for each experiment is shown.

Fluorescence microscopy studies

To analyse the localization of an XpsD-mCherry fusion protein, bacteria were grown over night in NYG medium, resuspended in fresh NYG medium at an OD₆₀₀ of 0.15 and incubated on a tube rotator at 30°C for 2 hours. Bacterial cultures were placed on microscopy slides on top of a pad of 1% (w/v) agarose dissolved in NYG medium as described previously [36]. mCherry fluorescence was visualized with a confocal laser scanning microscope (Zeiss LSM 780 AxioObserver. Z1) using filter sets for mCherry (excitation at 587 nm; emission at 610 nm). Experiments were performed with three different transconjugants per construct and repeated two times with comparable results. Fluorescent foci of at least 100 cells per transconjugant were counted. Results from one representative experiment are shown.

In-silico modeling of protein structures using the AlphaFold2 algorithm

The AlphaFold2 algorithm was used for protein structure prediction using the Colabfold server and the molecular visualization program UCSF ChimeraX [40,41,71]. Single structure predictions were performed for XpsE and XpsD, and full-length predictions for complexes of XpsL-XpsE, XpsC-XpsL-XpsM and XpsC_36-169-XpsD_22-252. For the generation of an in silico model of T2S system core components, the structural model of XpsD was superimposed on a pentadecameric crystal structure of the homologous Pseudomonas aeruginosa PAO1 XcpQ secretin channel (PDB: 5WLN) to generate a secretin channel. Furthermore, the XpsE structure was superimposed on the hexameric crystal structure of the homologous ATPase GspE from V. cholerae (PDB: 4KSR). The XpsE-XpsL structure prediction was used to align six predicted XpsCLM complexes to the hexameric XpsE structure and the XpsC_36-169-XpsD_22-252 prediction was used to align the HR domains of the resulting hexameric XpsCLME complex to the N0 domains of the pentadecameric XpsD complex.