Identification of divergent type VI secretion effectors using a conserved chaperone domain

Significance

How different microbial species compete for specific niches is not fully understood. Here, we focus on the type VI secretion system (T6SS), a specialized protein delivery system, which many Gram-negative bacteria use to kill eukaryotic and prokaryotic competitors by translocating toxic protein molecules to target cells. Identification of effectors is required for understanding the pivotal role that the T6SS plays in dictating interbacterial and bacterial–host dynamics. In this study, we describe a new approach to systematically identify T6SS effectors. We also demonstrate that secretion of effectors requires interaction with a set of cognate effector-binding chaperone proteins that we define in this study, providing important insights for understanding the mechanism of T6SS effector delivery.

Abstract

The type VI secretion system (T6SS) is a lethal weapon used by many bacteria to kill eukaryotic predators or prokaryotic competitors. Killing by the T6SS results from repetitive delivery of toxic effectors. Despite their importance in dictating bacterial fitness, systematic prediction of T6SS effectors remains challenging due to high effector diversity and the absence of a conserved signature sequence. Here, we report a class of T6SS effector chaperone (TEC) proteins that are required for effector delivery through binding to VgrG and effector proteins. The TEC proteins share a highly conserved domain (DUF4123) and are genetically encoded upstream of their cognate effector genes. Using the conserved TEC domain sequence, we identified a large family of TEC genes coupled to putative T6SS effectors in Gram-negative bacteria. We validated this approach by verifying a predicted effector TseC in Aeromonas hydrophila. We show that TseC is a T6SS-secreted antibacterial effector and that the downstream gene tsiC encodes the cognate immunity protein. Further, we demonstrate that TseC secretion requires its cognate TEC protein and an associated VgrG protein. Distinct from previous effector-dependent bioinformatic analyses, our approach using the conserved TEC domain will facilitate the discovery and functional characterization of new T6SS effectors in Gram-negative bacteria.

Protein secretion systems play a pivotal role in bacterial interspecies interaction and virulence (1, 2). Of the known secretion systems in Gram-negative bacteria, the type VI secretion system (T6SS) enables bacteria to compete with both eukaryotic and prokaryotic species through delivery of toxic effectors (2–4). The T6SS is a multicomponent nanomachine analogous to the contractile bacteriophage tail (5). First characterized in Vibrio cholerae (6) and Pseudomonas aeruginosa (7), the T6SS has now been identified in ∼25% of Gram-negative bacteria, including many important pathogens (2, 8), and has been implicated as a critical factor in niche competition (9–11).

The T6SS structure is composed of an Hcp inner tube, a VipAB outer sheath that wraps around the Hcp tube, a tip complex consisting of VgrG and PAAR proteins, and a membrane-bound baseplate (2, 4, 12). Sheath contraction drives the inner Hcp tube and the tip proteins, VgrG and PAAR, outward into the environment and neighboring cells (13, 14). The contracted sheath is then dissembled by an ATPase ClpV and recycled for another T6SS assembly and contraction event (12, 15, 16). Two essential T6SS baseplate components, VasF and VasK, are homologous to the DotU and IcmF proteins of the type IV secretion system (T4SS) in Legionella pneumophila (17).

Bacteria often possess multiple copies of VgrG and PAAR genes that form the tip of T6SS, and deletion of VgrG and PAAR genes abolishes T6SS secretion (14). Some VgrG and PAAR proteins carry functional extension domains and thus act as secreted T6SS effectors, as exemplified by the VgrG1 actin cross-linking domain (6), VgrG3 lysozyme domain in V. cholerae (18, 19), and the nuclease domain of the PAAR protein RhsA in Dickeya dadantii (20). Known T6SS effectors can target a number of essential cellular components, including the actin and membrane of eukaryotic cells (18, 21, 22) and the cell wall, membrane, and DNA of bacterial cells (3, 18–20, 23, 24). Each antibacterial effector coexists with an antagonistic immunity protein that confers protection during T6SS-mediated attacks between sister cells (3, 18, 24). Interestingly, T6SS-mediated lethal attacks induce the generation of reactive oxygen species in the prey cells (25), similar to cells treated with antibiotics (26, 27).

For non-VgrG/PAAR–related effectors, their translocation requires either binding to the inner tube Hcp proteins as chaperones or binding to the tip VgrG proteins (2, 14, 28). T6SS-dependent effectors can be experimentally identified by comparing the secretomes of WT and T6SS mutants (3, 29–31) and by screening for T6SS-encoded immunity proteins (18). Because known effectors lack a common secretion signal, bioinformatic identification of T6SS effectors is challenging. A heuristic approach based on the physical properties of effectors has been used to identify a superfamily of peptidoglycan-degrading effectors in bacteria (32). A recent study identified a common N-terminal motif in a number of T6SS effectors (31). However, this motif does not exist in the T6SS effector TseL in V. cholerae (18).

In this study, we report that VC1417, the gene upstream of tseL, encodes a protein with a highly conserved domain, DUF4123. We show that VC1417 is required for TseL delivery and interacts with VgrG1 (VC1416) and TseL. Because of the genetic linkage of VC1417 and TseL and its importance for TseL secretion, we postulated that genes encoding the conserved DUF4123 domain proteins are generally located upstream of genes encoding putative T6SS effectors. Using the conserved domain sequence, we bioinformatically predicted a large family of effector proteins with diverse functions in Gram-negative bacteria. We validated our prediction by the identification and characterization of a new secreted effector TseC and its antagonistic immunity protein TsiC in Aeromonas hydrophila SSU. Our results demonstrate a new effective approach to identify T6SS effectors with highly divergent sequences.

Results

Secretion of TseL Requires VC1417 and VgrG1 in V. cholerae.

TseL (VC1418) is located in the V. cholerae hcp1 operon consisting of seven genes (Fig. 1A), three of which (VC1417, VC1420, and VC1421) encode proteins with unknown functions. We first tested whether any of these three genes were required for TseL secretion. By comparing WT and a mutant lacking VC1417 to VC1421 (33), we found that TseL cannot be secreted in the mutant (Fig. 1B), indicating that at least one of the three genes, VC1417, VC1420, and VC1421, is required for TseL secretion. We reasoned that, if complementation of a gene can restore TseL secretion in the VC1417-21 mutant that lacks the immunity gene VC1419, it would be highly toxic when coexpressed with TseL due to sister cell-to-sister cell delivery of TseL. Indeed, we found that coexpression of VC1417 and TseL reduced cell survival by 10³-fold whereas expressing VC1417 or TseL alone had little effect on cell viability (Fig. 1C), indicating that VC1417 is required for TseL secretion. Interestingly, VC1417 itself is not a secreted substrate of T6SS (Fig. 1D). Expression of VC1420 or VC1421 did not restore TseL secretion.

Fig. 1.

VC1417 is important for TseL secretion in V. cholerae. (A) Operon structure of the VC1415-VC1421 region. Three of the seven genes, highlighted in black, have unknown functions. (B) Immunoblotting analysis of TseL::3V5 secretion as detected in the cytosolic (Cell) and supernatant (Sec) fractions. RpoB is an abundant cytosolic protein and is used as a control for cell lysis. (C) VC1417 is required for delivery of TseL. The VC1417-21 mutant was complemented with a plasmid-borne VC1417 or TseL alone or two plasmids carrying both VC1417 and TseL. Survival of the VC1417-21 mutant was tested by plating strains on LB medium containing 0.1% arabinose. Restoration of TseL delivery by VC1417 resulted in toxic delivery of TseL between sister cells, thereby leading to cell death. Mean values and SEs of triplicate samples are indicated. (D) VC1417 is not secreted by T6SS. VC1417::3V5 was overexpressed using a pBAD arabinose-inducible vector, and secretion was tested by immunoblotting. Wild type, V. cholerae T6SS active strain V52; ∆T6SS, the vasK deletion mutant of V52. (E) Effects of VgrG proteins on TseL delivery. Prey cell is the VC1417-21 mutant lacking the cognate immunity gene tsiV1 to TseL. Killer cells are WT V52 and its derivative vgrG and vasK mutants as indicated. (F) Bacterial two-hybrid analysis of VC1417 interaction with VgrG1 and TseL. VasH is a DNA-binding sigma54-dependent regulator and was used as a control. (G) Coimmunoprecipitation of VC1417 with VgrG1 and TseL. Cell lysates containing VC1417-FLAG alone or together with VgrG1-6His or TseL-6His were incubated with anti-His magnetic beads, and bound proteins were analyzed by Western blot using an anti-FLAG monoclonal antibody.

We then tested whether the vgrG1 gene, upstream of VC1417, is involved in TseL secretion in V. cholerae. Using the VC1417-21 mutant as a prey, we found that WT V. cholerae killed the prey efficiently whereas the vgrG2 mutant could not kill (Fig. 1E). This observation is consistent with previous reports that VgrG2 is essential for T6SS secretion (18, 33). Interestingly, whereas the vgrG3 mutant exhibited impaired killing, killing was completely abolished in the vgrG1 mutant (Fig. 1E). As a control, the vgrG1 mutant efficiently killed Escherichia coli as a prey (Fig. S1), consistent with a previous report (33). These results indicate that VgrG1 is not essential for T6SS activity but is required for delivery of TseL. We previously found that VgrG3 interacts with TseL in V. cholerae but not in E. coli (18). Because VgrG proteins likely form a heterotrimer in V. cholerae (21, 33), our results suggest that the previously reported interaction between VgrG3 and TseL is likely through VgrG1.

Interaction Between TseL, VC1417, and VgrG1.

The requirement of VC1417 and VgrG1 for TseL delivery suggests that these proteins may interact with one another. To test this hypothesis, we used a bacterial two-hybrid assay based on the functional complementation of the two T18 and T25 fragments of Bordetella pertussis adenylate cyclase (34). Protein interaction functionally reconstitutes the activity of adenylate cyclase that subsequently results in a LacZ⁺ phenotype on LB supplemented with X-gal. Because VgrG1 carries a large C-terminal actin–cross-linking domain that can be swapped by beta-lactamase without affecting secretion (35), we reasoned that the C-terminal extension domain is not required for delivery and thus expressed only the highly conserved N-terminal sequence (1-638aa) of VgrG1 for testing protein–protein interaction. We found LacZ⁺ phenotypes when VC1417 was coexpressed with TseL or VgrG1, suggesting direct interaction (Fig. 1F). In contrast, the negative control VasH, a DNA-binding sigma54-dependent regulator (36–38), exhibited LacZ⁻ when coexpressed with the other proteins tested. Using coimmunoprecipitation assays, we confirmed the interaction of VC1417 with VgrG1 and TseL (Fig. 1G).

VC1417 Belongs to a Large Family of Proteins with a Conserved Domain.

By searching the protein sequence of VC1417 in the Pfam protein database (39), we found that VC1417 carries a conserved domain DUF4123. DUF4123 was found in 818 protein sequences in 344 bacterial species, 342 of which belong to Proteobacteria, including Gammaproteobacteria (69%) and Betaproteobacteria (26%) (Fig. 2A and Dataset S1). Although DUF4123 is the only domain in the majority of these proteins, a few proteins carry an additional forkhead-associated (FHA) domain that is often involved in regulatory functions through phosphorylation. Interestingly, the Fha1 protein in P. aeruginosa is required for activating the T6SS cluster 1 (40). In addition, over 90% of DUF4123-encoding bacterial genomes also carry hallmark T6SS proteins, VipA (the outer sheath), and Hcp (the inner tube), indicating a strong association between the presence of DUF4123 and T6SS genes (Fig. 2B).

Fig. 2.

Distribution of the DUF4123 domain in bacterial species. (A) Distribution of DUF4123 in Proteobacteria. DUF4123 species were retrieved from the Pfam database. (B) Comparison of bacterial genomes possessing the DUF4123 domain proteins and the T6SS hallmark proteins VipA and Hcp.

The genome of V. cholerae encodes another DUF4123 domain protein, VasW, which is known to be required for secretion of its downstream effector, VasX (23). Because the DUF4123 domain proteins are widely distributed in Gram-negative bacteria, we reasoned that we could use this conserved domain as a signal to find highly divergent T6SS effectors. Two previously characterized effectors in V. cholerae, TseL and VasX, share little sequence similarity, but both have the DUF4123 domain-containing genes upstream, validating this method as a potential strategy for T6SS effector identification.

As a proof of principle, we first examined DUF4123 and downstream genes in a number of known T6SS-active bacteria, including P. aeruginosa (41), Agrobacterium tumefaciens (10), D. dadantii (20), and Aeromonas hydrophila (42). Indeed, genes encoding the DUF4123 domain are found upstream of genes encoding known T6SS effectors. Notably, DUF4123 genes are also often located together, with at least of one of the genes encoding T6SS secreted proteins Hcp, VgrG, or PAAR (Fig. 3).

Fig. 3.

Known and predicted DUF4123-associated effectors. DUF4123-containing genes were retrieved from the Pfam database. Protein functions were predicted using the HHpred program. Colors indicate genes encoding functionally similar proteins from a diverse set of Gram-negative bacteria.

Next, we tested whether the DUF4123 domain can predict unknown T6SS effectors. P. aeruginosa PA14 carries four DUF4123 proteins, one of which is located upstream of a known T6SS effector RhsP2 (41). The other three DUF4123 genes are located immediately downstream of genes encoding VgrG or PAAR proteins (Fig. 3). Using a structural prediction program, HHpred (43), we found that the three genes downstream of each of the DUF4123 genes encode a hydrolase, a endonuclease, and a colicin, respectively, suggesting these gene products are T6SS effectors. There are two DUF4123 domain proteins in A. hydrophila SSU (Fig. 3). Sequence analyses using HHpred (43) and Phyre2 (44, 45) show that the ORF2402 carries a putative colicin domain (HHpred probability 21% and Phyre2 confidence 78%) and that the ORF0928 carries a TC toxin domain (HHpred probability 100% and Phyre2 confidence 100%). The TC toxin complexes are important virulence factors in many bacterial pathogens, including Photorhabdus luminescens and Yersinia pestis, which target insects and mammalian cells (46, 47). We thus named ORF2402 TseC for its colicin domain and ORF0928 TseI for its potential insecticidal activity.

We expanded our analysis to 43 representative bacterial species with completely annotated genomes that encode the DUF4123 proteins (Dataset S2). For genes immediately upstream of the 133 DUF4123 genes analyzed, 70% possess an upstream vgrG and 7.5% an upstream PAAR (Dataset S2). For DUF4123 downstream genes, whereas the majority encode unknown functions, genes with known/predicted functions encode putative TC-toxin, lipase, nuclease, and hydrolase.

TseC Is a T6SS-Dependent Antibacterial Effector in A. hydrophila SSU.

To functionally validate our predictive method, we used A. hydrophila SSU as a model. The T6SS of SSU is known to target eukaryotic cells (42, 48), but T6SS-mediated antibacterial activities have not been demonstrated. To assess the function of the T6SS in A. hydrophila, we constructed a T6SS-null mutant lacking the vasK gene essential for T6SS functions (42). Using a bacterial killing assay (18), we found that WT SSU killed E. coli by 10,000-fold in comparison with the vasK mutant (Fig. 4A), indicating that the T6SS of SSU is highly effective in interbacterial competition.

Fig. 4.

Identification of an effector-immunity pair TseC-TsiC in SSU. (A) T6SS-dependent killing of E. coli by SSU. WT SSU and its T6SS null mutant vasK were mixed with prey E. coli at a 10:1 ratio and spotted on LB medium. Survival of E. coli was enumerated by serial dilution. (B) T6SS-dependent secretion of TseC. TseC carrying a 3V5 tag was detected in whole cell lysate or the supernatant (secreted) by immunoblotting analysis using an anti-V5 antiserum. (C) Operon structure of SSU tseC-tsiC. TseC carries a predicted colicin domain. (D) Expression of the SSU TseC colicin domain (768–891) is toxic in E. coli. The colicin domain was cloned into a pBAD vector under an arabinose-inducible promoter. A twin-arginine translocation (Tat) signal sequence was added to the N-terminal to deliver the colicin domain to the periplasm. (E) TsiC is the cognate immunity protein to TseC. The SSU deletion mutant of tseC tsiC was sensitive to T6SS killing whereas complementation with TsiC restores protection. Killer cells are WT SSU and its vasK mutant. Prey cells are the tseC tsiC double mutant and the tseC tsiC mutant complemented with a plasmid-borne tsiC. (F) ORF2403 and VgrG1 (ORF2404) are required for killing the tseC tsiC mutant but not E. coli. Killer cells are WT SSU and its mutant derivatives as indicated. Prey cells are the tseC tsiC double mutant of SSU and E. coli K-12 MG1655. (G) Bacterial two-hybrid assay of the interaction of ORF2403 with VgrG1(ORF2404) and TseC. VasH, a DNA-binding regulator in V. cholerae, is included as a negative control. (H) Coimmunoprecipitation analysis of the interaction of ORF2403 with VgrG1(ORF2404) and TseC. Cell lysates containing ORF2403-6His alone or with VgrG1-3V5 or TseC-3V5 were incubated with magnetic beads conjugated with a monoclonal anti-V5 antibody. Bound proteins were detected using a monoclonal antibody to 6His.

To test whether the predicted effector TseC is secreted by A. hydrophila T6SS (Fig. 3), we expressed an epitope-tagged TseC in the WT and the vasK mutant. Western blot analysis showed that TseC was secreted only by the WT but not the vasK mutant (Fig. 4B). TseC is predicted to carry a colicin domain (Fig. 4C). Because colicins attack E. coli through binding to cell membranes (49), we determined TseC antibacterial toxicity by expressing the colicin domain in the periplasm of E. coli using a twin-arginine secretion signal (50, 51). Periplasmic expression of the colicin domain reduced the survival of E. coli to 1% after induction (Fig. 4D), indicating that A. hydrophila TseC is potently antibacterial.

Because previously characterized T6SS-dependent toxic effectors coexist with antagonistic immunity proteins that are encoded by downstream genes (18, 24), we predicted that the gene downstream of tseC is the cognate immunity gene, hereafter referred to as tsiC. If TsiC is the immunity protein to TseC, the tsiC mutant would be susceptible to WT T6SS-mediated killing by delivery of TseC. We tested this hypothesis by constructing a double knockout mutant lacking both tseC and tsiC. Indeed, the tseC tsiC mutant was efficiently killed by 10⁴-fold when exposed to WT SSU, and complementation with a plasmid-borne tsiC fully protected the tseC tsiC mutant from killing (Fig. 4E), indicating that TsiC is the cognate immunity protein to TseC.

Delivery of TseC Requires VgrG1 and DUF4123 Protein ORF2403.

Upstream of SSU tseC are vgrG1 (ORF2404) (48) and the DUF4123 gene ORF2403. We postulated that the secretion of TseC requires VgrG1 and ORF2403 in SSU. To test this hypothesis, we constructed deletion mutants of vgrG1 and ORF2403 and tested their effects on killing the tseC tsiC double mutant. Neither mutant could kill the tseC tsiC double mutant (Fig. 4F). In contrast, E. coli was efficiently killed by these mutants, indicating that ORF2403 and VgrG1 are not required for T6SS functions. Given the dependency of TseC delivery on ORF2403 and VgrG1, we sought to determine whether these proteins interact. Using the bacterial two-hybrid assay (Fig. 4G) and coimmunoprecipitation (Fig. 4H), we identified the interaction of ORF2403 with VgrG1 and TseC. In addition, the tseC mutant could still kill E. coli, suggesting the existence of other T6SS-dependent antibacterial effectors in SSU (Fig. S2).

Discussion

Since the discovery of the T6SS in V. cholerae and P. aeruginosa, considerable effort has been made toward understanding the delivery mechanism and the physiological functions (2, 4, 14, 52). Previous research highlights that numerous human pathogens use the T6SS to deliver toxic effectors to their bacterial competitors or eukaryotic hosts (2, 4). Recent reports on T6SS function in the Bacteroidetes (11) and Agrobacterium (10) further underline the importance of T6SS in dictating bacterial dynamics in complex communities, such as the microbiota in humans and plants. Despite their importance, the identification and assignment of enzymatic function to T6SS effectors still remains challenging. Comparative analysis of effector sequences from different species could be used to identify potential homologs. However, systematic identification of effectors using bioinformatics is difficult because known T6SS effectors are highly diverse in sequence and function. Although previous studies have successfully identified a number of effectors based on the physical characteristics of known effectors (32) and an N-terminal sequence marker (31), respectively, neither method could identify TseL in V. cholerae.

In this study, instead of relying on the diverse effector sequences, we demonstrate an effective approach of using a conserved domain (DUF4123) to identify the associated downstream effectors. Our results show that DUF4123 proteins directly interact with the cognate VgrG and effector proteins and play an essential role in effector delivery but that DUF4123 proteins are not secreted or required for effector activities. DUF4123 may thus function similarly to the chaperone proteins of T4 phage, gp38 (53, 54) and gp63 (55, 56), which are important for tail fiber assembly and attachment but are not components of the mature phage particle (57). In addition, the secretion of many effectors of the type 3 secretion (TTS) system is dependent on specific interaction with cognate chaperone proteins that are present in the cytosol but not secreted (58, 59). Interestingly, TTS chaperone proteins generally have low molecular weight and acidic isoelectric point (pI <5), and the chaperone genes are often found next to the genes encoding cognate effectors (58, 59). We found that DUF4123 proteins also exhibit low pI values (∼5) and that the domain has several highly conserved residues (Fig. S3). Because of the abovementioned characteristics of DUF4123 proteins, we postulate that DUF4123 proteins function as T6SS effector chaperones (TECs) and thus name the DUF4123 domain TEC, the VC1417 protein TecL, and the SSU2403 protein TecC.

TEC genes are widely distributed in Proteobacteria and are largely located together with an upstream VgrG/PAAR gene. We predicted that downstream of TEC genes are genes encoding candidate T6SS effectors. Using the TEC sequence, we validated our analyses by identifying known effectors, including TseL and VasX in V. cholerae, that share few common features in sequence, function, and structure. We also confirmed a new T6SS-dependent effector-immunity pair, TseC-TsiC, in the A. hydrophila SSU strain.

Although the exact mechanism for T6SS protein export is not fully understood, there are two predicted models. The first model requires that effectors bind to the inner surface of the ring-like Hcp hexamers (52) whereas the second, termed multiple effector translocation VgrG (MERV), involves binding of effectors to the tip VgrG and PAAR proteins (2, 14, 18). The limited inner space of the Hcp hexameric ring likely poses a physical restraint on the size of effectors relying on binding to Hcp as chaperones for delivery (4). In the MERV model, binding to the tip proteins renders more flexibility to accommodate effectors that differ greatly in size and sequence (2, 14). Indeed, a number of effectors with diverse functions have been reported to require VgrG and PAAR proteins for delivery (14, 18, 41). Our results on the VgrG-dependent secretion of TseL in V. cholerae and TseC in A. hydrophila further support the MERV model.

Notably, many bacterial species possess multiple TEC proteins. For example, A. hydrophila SSU has two TEC proteins (Fig. 3). These two TEC proteins cannot functionally complement each other, as evidenced by the loss of killing resulting from deletion in tecC (ORF2403) (Fig. 4F). Because TEC proteins interact with both conserved VgrG proteins and divergent effectors, we propose that the conserved TEC domain is responsible for binding to VgrG/PAAR whereas each TEC protein has acquired specific sequences to accommodate binding to its partner effector. For T6SS-mediated delivery of a given VgrG-binding effector, multiple binding events likely occur in a temporal order that includes effector binding to the cognate VgrG, to the TEC protein, and to the immunity protein (if the immunity protein is present in the cytosol). Because TEC proteins and immunity proteins are not secreted, the separation of effectors from the cognate TEC and immunity proteins probably takes place before binding to VgrG for delivery. It is possible that TEC proteins coordinate the process of effector loading to the VgrG/PAAR spike to prevent premature binding of effectors with VgrG. The formation of T6SS spike might expose the effector-binding site of VgrG that attracts effectors and displaces TEC proteins. Because TEC proteins can bind to both effectors and VgrGs, it is also possible that TEC proteins facilitate the binding of VgrG and effectors by presenting the binding partners in right conformation or maintaining protein stability. Structural analyses of TEC, VgrG, and effector proteins are required to fully understand not only the actions of TEC but also the mechanisms of T6SS effector delivery.

Here, we report a class of diverse and TEC-dependent T6SS effectors. Future studies characterizing the functions of these effectors in different model systems will greatly increase our understanding of the physiological role that T6SS plays in these species. Given the diverse ecological niches that these species occupy in the environment and the host, more novel functions of T6SS effectors are likely to be discovered. Because T6SS effectors are known to target essential cellular functions, including the cell wall, membrane, and DNA/RNA of bacteria and the membrane and cytoskeleton of eukaryotic cells (2, 4), the toxicity of effectors may provide an alternative therapeutic approach for treating bacterial infections or killing specific types of eukaryotic cells.

Materials and Methods

Bacterial Strains and Growth Conditions.

Strains and plasmids used in this study are listed in Table S1. Cultures were routinely grown aerobically at 37 °C in LB (wt/vol 1% tryptone, 0.5% yeast extract, and 0.5% NaCl). Antibiotics and chemicals were used at the following concentrations: ampicillin (100 µg/mL), streptomycin (100 µg/mL), kanamycin (50 µg/mL), tetracycline (10 µg/mL), chloramphenicol (25 µg/mL for E. coli, 2.5 µg/mL for SSU and V52), isopropyl β-d-1-thiogalactopyranoside (IPTG) (1 mM), and arabinose (wt/vol 0.1%). Mutants of SSU and V52 were constructed using crossover PCR and homologous recombination (60, 61). Gene expression vectors were constructed as previously described (38). All constructs were verified by sequencing. Primers for cloning and mutant construction are listed in Dataset S3.

Western Blotting Analysis.

Protein samples were loaded on a precast 4–12% SDS/PAGE gel (Life Technologies), run at 180 V for 40 min and transferred to a PVDF membrane (Millipore) by electrophoresis. The membrane was blocked with 5% (wt/vol) nonfat milk in TBST buffer (50 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH 7.6) for 1 h at room temperature, incubated with primary antibodies at 4 °C overnight, washed three times in TBST buffer, and incubated with an HRP-conjugated secondary antibody (Cell Signaling Technology) for 1 h followed by detection using the ECL solution (Bio-Rad) and a ChemiDoc MP system (Bio-Rad). The monoclonal antibodies to epitope tags, anti-V5, anti-FLAG, and anti-6xHIS were purchased from Sigma Aldrich. The monoclonal antibody to RpoB, the beta subunit of RNA polymerase, was purchased from NeoClone and used as a loading control for Western blot analysis as previously described (62).

Protein Secretion Assay.

Exponential phase cultures (OD₆₀₀ = 0.5) grown in LB were induced by adding 0.1% l-arabinose for 1 h. One-milliliter culture was collected by centrifugation twice at 20,000 × g for 2 min and then filtered through a 0.2-µm filter. The filtered supernatant was combined with 200 µL of 100% ice-cold TCA solution, placed on ice for 2 h, and centrifuged at 15,000 × g for 30 min at 4 °C. The pellet was washed with 1 mL of 100% acetone by centrifugation at 20,000 × g for 5 min, air-dried and mixed with 30 µL of SDS-loading dye, followed by SDS/PAGE and Western blot analyses as described above.

Bacterial Cell-Killing Assay.

The killing assay was performed as previously described (33). Briefly, cultures were mixed together at a ratio of 10:1 (predator to prey), spotted on LB medium for 3 h at 37 °C, and then resuspended in 1 mL of LB. Survival of prey cells was quantified by serial dilution in LB and plating on selective medium.

Bacterial Two-Hybrid Assay.

The two-hybrid assay was performed as described (34, 63). Plasmid vectors carrying the indicated T18 and T25 constructs were transformed to BTH101 (cya-99). Individual colonies were grown in LB for 3 h and then patched on LB medium supplemented with Amp, Kan, X-Gal, and 0.5 mM IPTG. Plates were incubated at room temperature for at least 48 h.

Coimmunoprecipitation Assay.

Genes were cloned into pETDuet-1 and pACYCDuet-1 vectors for expression. E. coli BL21DE3 carrying different gene expression vectors was grown in 50 mL of LB culture to exponential phase OD₆₀₀ = 0.5, and induced by 1mM IPTG for 3 h at 37 °C. Cells were collected by centrifugation at 4,000 × g for 10 min and resuspended in 5 mL of PBS buffer. Cell lysates were prepared by sonication, and cell debris was removed by centrifugation at 20,000 × g for 20 min. Dynabeads Protein G (Life Technologies) was incubated with 5 µg of monoclonal antibodies to 6His or V5 for 2 h at 4 °C, and then mixed with 1 mL of cell lysates for 3 h at 4 °C. The magnetic beads were washed three times with ice-cold PBST (phosphate saline buffer with vol/vol 0.2% Tween 20) and then incubated with 30 µL of SDS-loading buffer, followed by incubation at 70 °C for 10 min to elute bound proteins. Eluted samples were subject to Western blotting analysis as described above.

Bioinformatic Analysis.

Protein sequences were retrieved from the National Center for Biotechnology Information (NCBI) database and analyzed using HHpred (43) and Phyre2 (44, 45) for functional prediction. Representative DUF4123 protein accession numbers and species were downloaded from the Pfam protein database (39). Species carrying the DUF4123 domain, VipA (DFU770), and Hcp (DUF796) were downloaded from the Interpro database and compared using the Gene List Venn Diagram program (genevenn.sourceforge.net/). Using the Pfam-generated species tree of DUF4123, we selected representative species from each genus with fully annotated genomes to characterize the DUF4123 immediate upstream and downstream proteins using the protein annotation in the NCBI database.