Tse^VF-Tsi^VF, a novel bacteriolytic effector-immunity pair of Vibrio fluvialis VflT6SS2, provides a fitness advantage in microbial competition

Highlight

• •T6SS effector TseVF demonstrates apparent bacterial killing activity in the periplasm.
• •TseVF interacts with phosphate lipid molecules and compromises the integrity of the cell membrane.
• •The C-terminal region of TseVF is responsible for its bactericidal activity. 4. TsiVF binds to and neutralizes TseVF toxicity.
• •TseVF secretion and bactericidality requires the C terminus of a cognate VgrG protein (TssI2_b).

Abstract

Vibrio fluvialis (V. fluvialis) is a halophilic Gram-negative bacterium regarded as an emerging pathogen that can cause severe cholera-like diarrhea and various extraintestinal infections. Type VI secretion systems (T6SSs) are nano-weapons used by bacteria to deliver toxic effectors into host cells or antagonize competitors. VflT6SS2 of V. fluvialis is involved in bacterial pathogenicity and competitive environmental survival. Previously, we have identified some regulatory factors of VflT6SS2 and a pesticin domain-containing effector, TssI2. Here we reported a novel effector-immunity pair of VflT6SS2, namely Tse^VF-Tsi^VF, the homologs of which are widely distributed in Vibrio. Tse^VF mediates killing effect toward co-existing bacteria while Tsi^VF confers protection against Tse^VF. Tse^VF biologically functions in the periplasmic space and its C-terminal region (residues 965–1141), which adopts a helix-turn-helix fold composed of eight helices and fully accounts for the bactericidal activity. Periplasmic expression of Tse^VF965–1141 compromises the integrity of the cell membrane, leading to the leakage of cellular contents, and two highly conserved residues W1056 and W1091, are important for the bactericidality. The membrane-disrupting function of Tse^VF was further supported by its interaction with phosphate lipid phosphatidylethanolamine and phosphocholine. The immunity Tsi^VF directly binds to and neutralizes Tse^VF toxicity through its TPR-like domain. The C terminus of TssI2_b specifically mediates the loading or secretion of Tse^VF. In conclusion, our study identified a new class of T6SS pore-forming effector-immunity pair Tse^VF-Tsi^VF and revealed its molecular function and secretion mechanism in V. fluvialis, further enhancing our understanding of the diverse effectors of T6SS.

Graphical abstract

1. Introduction

Type VI secretion system (T6SS) is a highly versatile secretion system in Gram-negative bacteria that delivers toxic effector proteins into neighboring cells or into the extracellular milieu. It plays a crucial role in bacterial virulence and participates in multiple functions such as interbacterial competition, pathogenesis, metal ion uptake, and host immune response (Jana and Salomon, 2019; Yang et al., 2021). T6SS assembles from three major parts: a membrane complex, a phage-like baseplate, and a needle-like sheath-tube polymer with injection functions (Nazarov et al., 2018). The TssJLM membrane complex recruits the TssEFGK-VgrG baseplate on the cell envelope which services as an assembly platform for sheath-tube polymerization (Brunet et al., 2015). The sheath-tube polymer consists of an inner tube (Hcp) wrapped with a TssBC contractile sheath, tipped with a VgrG-PAAR spike, and anchored on the baseplate and membrane complex (Shneider et al., 2013; Cherrak et al., 2019; Lin et al., 2021). Contraction of the TssBC sheath propels the inner tube and membrane-puncturing spike out of the bacterium, delivering VgrG, PAAR and accompanying effectors to the target cells (Cherrak et al., 2019).

A well-characterized function of T6SS-associated effectors is antibacterial toxicity, mediating the role of T6SS in inter-species competition (Yang et al., 2018). To protect themselves from self-intoxication by toxic effectors, bacteria encode cognate immunity proteins that usually antagonize the toxic activities of effectors via direct interaction (Yang et al., 2018). Therefore, toxic effectors and their cognate immunity proteins typically exist in pairs (Yang et al., 2018). These bactericidal effectors elicit toxicity through the disruption of different cellular structures, including the cell wall, membrane, and nucleic acids. Pore-forming toxins (Miyata et al., 2011) and lipases (Russell et al., 2013) are two main classes of T6SS-associated effectors targeting the cell membranes. The first pore-forming T6SS effector to be described was VasX from Vibrio cholerae, which contains predicted C-terminal transmembrane helices and displays toxicity to amoebae (Miyata et al., 2011). Upon delivery to the periplasm, VasX can compromise the integrity of inner-membrane, as evidenced by the loss of membrane potential and increase of cellular permeability to intercalating dye (Miyata et al., 2013). P. aeruginosa Tse4 is another well-studied pore-forming toxin that contains six transmembrane domains and spontaneously integrates into bacterial membranes, creating tiny, ion-selective channels with a preference for cations over anions that cause membrane depolarization followed by K⁺ efflux, ultimately leading to bacterial death (LaCourse et al., 2018; Rojas-Palomino et al., 2025). Other T6SS effectors were suggested to be pore-forming effectors based on homology to pore-forming colicins and the presence of transmembrane helices at their predicted C-terminal toxin domains, but their activities have yet to be demonstrated (Salomon et al., 2014; Dar et al., 2018). Five widespread and divergent families of lipase effectors, named Tle1–5 (Type VI lipase effector), were uncovered. Tle1–4 share a GxSxG conserved motif and a Ser–Asp–His catalytic triad, while Tle5 contains a dual HxKxxxxD motif (Russell et al., 2013). A variety of T6SS nuclease effectors containing different conserved motifs such as HNH, HxxD, and D - (D/E) xK were identified to degrade cellular DNA (Koskiniemi et al., 2013; Ma et al., 2014, 2017; Pissaridou et al., 2018; Jana et al., 2019).

Effectors can be transferred to target cells by directly or indirectly binding to PAAR, VgrG, and Hcp. When associated with Hcp, the effector is embedded in the lumen of the hexameric ring. However, the available space in the lumen of the Hcp ring limits the size of the effector to be transported, which is estimated to be <25 kDa (Silverman et al., 2013). Therefore, most effectors are delivered into neighboring cells through interaction with VgrG or PAAR (Shneider et al., 2013; Cianfanelli et al., 2016; M. Liu et al., 2023). Some Hcp (Ma et al., 2017), VgrG and PAAR (Pissaridou et al., 2018; Quentin et al., 2018) proteins can fuse toxic domains at C-termini to function as “specialized” effectors, such as V. cholerae VgrG1, which cross-links actin in eukaryotic cells through its C-terminal actin cross-linking domain (Pukatzki et al., 2007; Durand et al., 2012), and VgrG3, which carries a C-terminal extension with peptidoglycan glycoside hydrolase activity (Brooks et al., 2013; M. Liu et al., 2023).

Two T6SS gene clusters (VflT6SS1 and VflT6SS2) have been identified in V. fluvialis, and VflT6SS2 is functional in mediating interbacterial competitiveness (Huang et al., 2017). The genetic organization and coding sequence of VflT6SS2 are highly homologous to the T6SS of V. cholerae, except harbors three “auxiliary” or “orphan” hcp-vgrG alleles which were designated as tssD2_a-tssI2_a, tssD2_b-tssI2_b, and tssD2_c-tssI2_c, respectively (Huang et al., 2017). We have shown that environmental factors such as warm temperature, high salinity/osmolarity and transcriptional regulators including integration host factor (IHF) and σ⁵⁴-dependent VasH are involved in activating the expression of VflT6SS2 (Huang et al., 2017; Pan et al., 2018). Also, the crosstalk between the quorum sensing system (QS) and T6SS was revealed on the transcriptional regulatory role of the CqsA/LuxS-HapR (Liu et al., 2021) and VfqI-VfqR QS circuit (Han et al., 2022) on VflT6SS2 function. Two toxic effectors were identified, TssI2 and Tle1^VF. TssI2, encoded by the main cluster, is a “specialized” effector that displays PG hydrolase activity via a C-terminal pesticin domain (Huang et al., 2022). Tle1^VF is encoded by the auxiliary cluster a (tssD2_a-tssI2_a) and has phospholipase activity (unpublished results, manuscript in preparation). Here, our further exploration revealed a novel effector-immunity pair, named Tse^VF-Tsi^VF, in auxiliary cluster b (tssD2_b-tssI2_b). We demonstrated that Tse^VF, as a novel pore-forming effector, can disrupt the integrity of cell membrane through the C-terminus (residues 965–1141) and confer competitive advantages to V. fluvialis. Tsi^VF directly binds and counteracts the virulence of Tse^VF via its TPR-like domain. And Tse^VF interacts with the C terminus of TssI2_b to load and exert functional activity. These results help us better understand the function of VflT6SS2 and its diverse effectors in mediating interspecies and intraspecies antibacterial competitiveness.

2. Materials and methods

2.1. Bacterial strains, culture conditions, and plasmids

V. fluvialis wild-type (WT) strain 85003 and its derivative mutants were routinely grown in Luria-Bertani (LB) broth (pH 7.4) with 2 % NaCl (340 mM) at 30 °C. Escherichia coli strains were usually cultured at 37 °C. All plasmids and bacterial strains used in this study are listed in Tables S1 and S2. Antibiotics were added to the culture medium as needed: ampicillin (Amp, 100 μg/mL), kanamycin (Km, 50 μg/mL), streptomycin (Sm, 100 μg/mL), rifampicin (Rfp, 50 μg/mL), tetracycline (Tc, 10 μg/mL for E. coli, 2.5 μg/mL for V. fluvialis), chloramphenicol (Cm, 10 μg/mL for E. coli, 2.5 μg/mL for V. fluvialis). Isopropyl-β-D-thiogalactopyranoside (IPTG) was added at a final concentration of 0.5 mM if necessary.

2.2. Mutations construction

V. fluvialis ΔvasH mutant was constructed previously (Huang et al., 2017). ΔvasK, ΔtssI2_b, Δtap^VF-tse^VF, and Δtse^VF-tsi^VF were constructed using the same method. Briefly, flanking sequences of the target gene were amplified by polymerase chain reaction (PCR) using high-fidelity polymerase (TaKaRa, China) and spliced by overlapping PCR (Wu et al., 2015). The primers used are listed in Table S2. The resultant fragments were cloned into pWM91 suicide plasmid. Recombinant plasmids were transferred from E. coli SM10λpir to WT strain by conjugation. Exconjugants were counter-selected by streaking on NaCl-free LB agar containing 8–10 % sucrose. Colonies that are sucrose-resistant and sensitive to Amp were verified by PCR and further confirmed by sequencing.

2.3. Plasmid construction

Complementation plasmids pSRKTc-tsi^VF, pSRKTc-tsi^{VFΔ100–180}, pFlag-CTC-tssI2_b and pFlag-CTC-tssI2_bΔC were constructed by cloning the coding sequences of tsi^VF, tsi^{VFΔ100–180}, tssI2_b and tssI2_bΔC which were amplified with specific primers into pSRKTc or pFlag-CTC, respectively. These target genes were expressed from the lac promoter with the induction of IPTG. All related primers used are listed in Table S2.

2.4. Bacterial toxicity assay

pBAD‑sec-tse^VF and its truncated variants were constructed by multiple PCR to add the Sec signal peptide sequence to the 5′ end of tse^VF or its truncated versions, followed by cloning the target sequences into the pBAD24 vector. Primers used are listed in Table S2. Then the recombinant plasmid was introduced into E. coli MG1655. Cells were normalized to 3.5 McF and inoculated at 1:100 into 5 mL LB broth for 1 h (T 0). Then, 1 mL of the bacterial suspensions with a proper final concentration (w/v) of L-arabinose (for induction) or D-glucose (for repression) was incubated for 3 h (T 3). The resultant bacterial suspensions were 10-fold serially diluted and 10 µL of each dilution was spotted onto Amp resistant LB agar plates for colony counting.

2.5. Bacterial competition assay and self-intoxication assay

Bacterial competition assays were performed as previously described (Huang et al., 2022). Briefly, overnight cultures of the attacker (V. fluvialis, Sm^R) and prey (E. coli and other Vibrio spp., Rfp^R) strains were normalized to 1.5 McF, then mixed at a ratio of 9:1 (attacker:prey) in a final total volume of 1 mL. A 10 µL aliquot of this mixture was then spotted onto 2 % NaCl LB agar plates and incubated at 30 °C for 5 h, or for 12 h with additional 1.0 mM IPTG where induction was required. Immediately take an appropriate volume of the remaining mixture, perform a 10-fold serial dilution, and plate onto Rfp- or Sm- resistant agar plate to determine the initial colony forming unit (CFU) counts of the prey or attacker (T 0). After co-incubation (T 5/T 12), bacteria were harvested from the competition spot, resuspended, serially diluted, and plated onto selective agar plates to determine final prey or attacker CFU counts. Prey survival was assessed by comparing the final (T 5 or T 12) CFU counts to the initial (T 0) counts while ensuring consistent numbers of attacker in each group. Assays were repeated three times independently, and representative results were shown.

The self-intoxication assay was performed using the same method with Δtse^VF-tsi^VF (V. fluvialis WT strain with tse^VF and tsi^VF dual deletion, Rfp^r) or its complementation derivative as preys.

2.6. Membrane permeabilization assay

E. coli MG1655 carrying either a control plasmid (pBAD24) or pBAD‑sec-tse^VF965–1141 construct was grown overnight in LB broth at 37 °C with shaking. The cultures were normalized to 3.5 McF, diluted 1:100 in fresh LB broth, and incubated for 1 h (T 0). Then, 1 mL bacterial suspension with proper final concentration of L-arabinose or D-glucose was continually incubated for 1.5 h and 3 h. The cells were pelleted, washed twice with phosphate-buffered saline (PBS), and then resuspended in 1 mL of PBS supplemented with 2.5 μM propidium iodide (PI). After incubation at 37 °C for 20 min in the dark, the fluorescence (excitation ƛ: 535 nm; emission ƛ: 617 nm) and OD₆₀₀ were measured using Infinite M200 Pro (Tecan, Austria).

2.7. Transmission electron microscope (TEM)

TEM images were obtained by the negative-staining method. First, the overnight bacterial cultures were normalized to 3.5 McF, diluted 1:100 in fresh LB broth, and incubated for 1 h. Subsequently, the bacterial suspension was incubated for 3 h with either 0.2 % L-arabinose or 0.2 % D-glucose. The bacteria were then fixed by adding formaldehyde to a final concentration of 4 %. The fixed samples were subsequently transferred onto the copper grids. After 1 min, excess culture liquid was wicked off with filter paper, and the grid was rinsed with fresh deionized water and stained with 0.5 % phosphotungstic acid (pH 6.8) for 1 min. Finally, the grid was dried at room temperature and observed for bacterial morphology under a transmission electron microscope (FEI, Tecnai12, USA). A CCD digital camera (EMSIS, model MORADA-G3, Germany) was used to take photos and record the results.

2.8. Protein expression and purification

The complete open reading frame of tse^VF and the coding sequence of tsi^VF (residues 19–333, without the N-terminal signal peptide) were amplified and individually cloned into the pET-30a(+) expression vector, generating the recombinant plasmids pET30a-tse^VF-his and pET30a-tsi^VF19–333-his, respectively. E. coli BL21 (DE3) harboring either plasmid was cultured in LB broth to OD₆₀₀ of 0.5 with shaking at 37 °C and then shifted to 16 °C, induced by 0.5 mM IPTG overnight. Cells were collected by centrifugation and resuspended in Lysis buffer (25 mM Tris–HCl pH 8.0, 150 mM NaCl, 10 mM imidazole). After sonication on ice, the cell debris was removed by centrifugation.

According to the manufacturer’s instructions (Novagen, Germany), affinity chromatography was performed using Ni-IDA resin to adsorb the 6 × His-tagged Tse^VF or Tsi^VF19–333 in the supernatant. The resin was washed sequentially with Lysis Buffer supplemented with 20 mM, 40 mM, and 60 mM imidazole. The target proteins were subsequently eluted with elution buffer (25 mM Tris–HCl pH 8.0, 50 mM NaCl, 300 mM imidazole). The buffer of the eluted protein was exchanged into PBS using a centrifugal filter device (Millipore, USA) with a 100 kDa molecular weight cutoff. Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and Coomassie blue staining were used to check protein purity. The purified proteins were quantified using a BCA Protein Assay Kit (TaKaRa, China) and stored at −80 °C.

2.9. Liposome sedimentation assay

Liposome sedimentation assay was performed with reference to the literature with slight modifications (Jia et al., 2022). Commercially available phospholipids were used: DOPE (1,2-dioleoyl-sn‑glycero-3-phosphoethanolamine), DOPC (1,2-dioleoyl-sn‑glycero-3-phosphocholine), Brain SM (Sphingomyelin [Brain, Porcine]), POPC (1-palmi-toyl-2-oleoyl‑glycero-3-phosphocholine), and POPE(1-palmitoyl-2-oleoyl-sn‑glycero-3-phosphoethanolamine) (Jia et al., 2022), which were purchased from Avanti Polar Lipids, Inc. Lipids were prepared with liposome buffer (25 mM Bis-tris, 109.5 mM NaCl, 5.4 mM KCl, 0.4 mM MgSO₄, 0.45 mM CaCl₂, pH 6.5) in a clean glass vial to a final concentration of 5 mM at room temperature with vigorous vortexing for 1 h, stored at 4 °C, and used within 48 h (Jia et al., 2022). Liposome (40 μL) and purified Tse^VF−His₆ protein (5 μL, 2 mg/mL) were co-incubated at 28 °C for 1 h, with parallel negative controls containing liposome buffer only. After incubation, samples were centrifuged at 15,294 × g for 1 h at 4 °C (Jia et al., 2022). The resulting pellet and the supernatant were processed separately: pellet was resuspended with 10 μL 1 × loading buffer; supernatant was heated at 100 °C with 1 × loading buffer until a volume of about 10 μL was reached (Jia et al., 2022). Protein composition was assessed by SDS-PAGE with Coomassie blue. Gel images were acquired using Bio-Rad Chemidoc and the intensities of protein bands were quantified using ImageJ software. For quantification, gel images were first converted to 8-bit grayscale, followed by uniform background subtraction using the “Subtract Background” function (rolling-ball algorithm, radius = 50 pixels). Subsequently, to ensure consistency, a rectangular region of interest (ROI) was drawn to tightly enclose a representative band in a control lane (Lane Buffer S), thereby minimizing the inclusion of surrounding background. This exact ROI was then applied to the identical position in all lane for measurement. The Integrated Density (IntDen) was recorded for each band. The binding percentage for each condition was calculated as: Binding ( %) = [IntDen (Pellet) / (IntDen (Pellet) + IntDen (Supernatant))] × 100 %. Each assay was performed in triplicate.

2.10. Bacterial adenylate cyclase-based two-hybrid (BACTH) assay

Target proteins were fused to the isolated T18 or T25 catalytic domains of adenylate cyclase (CyaA) in BACTH vectors (Karimova et al., 2017). Recombinant BACTH plasmids were first maintained in E. coli K12 recA strain (XL1-Blue) before being introduced into E. coli BTH101 reporter strain via transformation (Karimova et al., 2017). The transformants were then plated on LB agar plates containing the appropriate antibiotics, IPTG (1 mM) and bromo‑chloro‑indolyl-galactopyranoside (X-gal, 40 mg/mL), followed by incubation at 30 °C for 24 h to 48 h (Huang et al., 2022). To ensure reproducibility, the assay was conducted in triplicate. Data from one representative trial were presented.

2.11. Co-immunoprecipitation (Co-IP) and protein pull-down assay

The recombinant plasmids pBAD33-tsi^VF19–333-myc, pBAD33-tsi^{VFΔ100–180}-myc, pSRKTc-tse^VF, pET30a-tse^VF-his, pFlag-CTC-tssI2_a, pFlag-CTC-tssI2_bΔC and pFlag-CTC-tssI2_b were constructed as described above. The Co-IP and pull-down assays were performed as previously described (Huang et al., 2022). Briefly, E. coli BL21 (DE3) cells harboring the relevant plasmids were grown in 200 mL fresh LB with appropriate antibiotics to an OD₆₀₀ of 0.5. Protein expression was induced with 0.2 % L-arabinose or 0.5 mM IPTG, followed by 16 °C overnight shaking at 150 rpm. Cells were harvested and resuspended in 5 mL lysis buffer (50 mM Tris–HCl pH 7.5, 150 mM NaCl, 2 % (v/v) glycerol, 1 mM PMSF), and lysed by sonication. The lysates were clarified by centrifugation (12,000 × g, 30 min, 4 °C). Input samples were prepared by mixing 100 μL aliquot of supernatant with SDS loading buffer and boiled for 10 min. For immunoprecipitation, lysates were incubated with Pierce™ Anti-c-Myc Agarose (Thermo, USA), or Protein A/G PLUS Agarose (Santa Cruz Biotechnology, USA) conjugated with Flag or Tse^VF antibodies. For pull-down, lysates were incubated with HisPur™ Ni-NTA resin (Thermo, USA). After overnight incubation (4 °C, rotation), beads/resin were pelleted (1000 × g, 4 °C), washed three times with lysis buffer, and resuspended in 50 μL lysis buffer containing SDS loading buffer. Samples were boiled and analyzed by Western blotting (WB). All experiments were repeated in two independent experiments (representative data shown).

2.12. Isothermal titration calorimetry (ITC)

The binding interaction between Tse^VF and Tsi^VF19–33 was characterized using MicroCal iTC₂₀₀ instrument (GE Healthcare, USA) at 25 °C, following established experimental procedures similar to previous studies (Huang et al., 2022). For the measurements, the microcalorimetric cell contained a low concentration of Tse^VF-His₆ protein (25 μM), while a high concentration of Tsi^VF19–333-His₆ protein (279 μM) was filled into the injection syringe. Under constant stirring, 3 μL of Tsi^VF19–333-His₆ was titrated into the Tse^VF-His₆-containing sample pool at every 150 s, with each titration lasting 6 s. Calculate the titration heat to eliminate the effect of heat generated by titrating Tsi^VF19–333-His₆ into PBS buffer, and subtract these values from the experimental data. Titration data were analyzed using MicroCal-enabled Origin™ software (OriginLabs) and fitted to the One Set of Sites binding model, assuming a fixed N value of 1 to calculate the value of the equilibrium dissociation constant (K_d) (Huang et al., 2022).

2.13. Ultra-performance liquid chromatography (UPLC)-tandem mass spectrometry (MS) analysis

Peptidoglycan (PG) was isolated from stationary-phase E. coli MG1655 cultures using an established protocol (Huang et al., 2022). Briefly, cell pellets were resuspended in PBS (pH 7.4) and lysed by dropwise addition to boiling 5 % SDS, followed by continuous heating and stirring for 1.5 h. After overnight stirring at room temperature, PG sacculi were collected via ultracentrifugation (150,000 × g, 40 min, 20 °C) and extensively washed with distilled water to remove residual SDS. Proteins were removed by Pronase E digestion (1 mg/mL, 56 °C, overnight), followed by SDS inactivation (100 °C, 5 min). Purified PG was washed and resuspended at 300 mg/mL (wet weight).

For UPLC-MS analysis, reaction systems (final volume 105 μL) containing 100 mg/mL PG in 10 mM NaAc buffer (pH 4.9) were supplemented with 10 mM NaCl, 3 mM MgCl₂ and 0.1 % Triton X-100. PG samples were incubated overnight at 37 °C with 1 mg/mL Lysozyme, Tse^VF, or TssI2 variant protein (TssI2^R1000A) (Huang et al., 2022). The reactions were terminated by boiling (100 °C, 5 min), and the insoluble debris was removed by centrifugation (12,000 × g, 5 min). Processed samples were analyzed by UPLC-MS as previously described (Huang et al., 2022).

2.14. Analysis of vflt6ss2 expression and secretion

To examine VflT6SS2 expression and secretion, overnight cultures of V. fluvialis WT and its derivative mutants were diluted 1:100 in 20 mL LB (340 mM NaCl) and grown at 30 °C shaking to OD₆₀₀=1.5. Proteins from cell pellets and cell-free supernatants were prepared as previously described and quantified with the BCA™ protein assay kit (Thermo Fisher Scientific, USA) (Huang et al., 2017; Pan et al., 2018; Liu et al., 2021). WB analysis of10 µg protein per sample was performed with polyclonal rabbit Hcp antibody and E. coli cyclic AMP receptor protein (CRP) antibody (BioLegend, USA) (Huang et al., 2017).

2.15. Bioinformatics and protein structure analysis

Protein domains and functions were predicted using InterPro (https://www.ebi.ac.uk/interpro/search/sequence/). Transmembrane structures were forecasted with Phobius (https://phobius.sbc.su.se/). Secondary structure of Tse^VF and Tsi^VF were visualized using IBS software.

Due to the lack of available templates for constructing a homology model, the full-length Tse^VF structure was predicted using the I-TASSER server (https://zhanglab.ccmb.med.umich.edu/) and AlphaFold2 (https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb#scrollTo=_sztQyz29DIC). The resulting models were evaluated using the SAVES server (https://saves.mbi.ucla.edu/). The highest scoring model (ERAT2 model score of 87.6083) was selected for Fig. 2B and subsequently optimized using Gromacs 2018–4.

Fig. 2.

Tse^VF mediates the disruption of cell envelope integrity.

A. Bacterial toxicity assay of Tse^VF. E. coli MG1655 harboring pBAD-tse^VF (cytosolic) or pBAD‑sec-tse^VF (periplasmic) were induced by 0.02 % and 0.2 % L-arabinose or suppressed by 0.2 % D-glucose for 3 h in LB broth. Bacterial survival was counted by gradient dilutions of bacterial suspensions.

B. Domain organization and structural prediction of Tse^VF. The non-cytoplasmic region at the N-terminus (residues 1–969) and the cytoplasmic region at the C-terminus (residues 970–1197) are marked with arrows at the top. Structurally defined domains are colored as follows: the LysM domain (residues 1–59) in blue and the transmembrane domain (residues 941–969) in green. Within the I-TASSER predicted active region (residues 666–1141), the area (residues 965–1141) experimentally showing the strongest bactericidal activity is colored yellow, the area (residues 666–941) with no killing effect is in red. The remaining regions are shown in rosiness.

C. Bactericidal activities of various Tse^VF truncated constructs against E. coli MG1655. Survival of E. coli MG1655 containing the indicated construct was assessed before (T 0) and after (T 3) 0.2 % and 0.02 % L-arabinose induction or 0.2 % D-glucose inhibition.

D. Membrane permeabilization assay of Tse^VF. E. coli MG1655 harboring pBAD24 (control) or pBAD‑sec-tse^VF965–1141 was induced by 0.2 % L-arabinose or inhibited by 0.2 % D-glucose, and incubated for 1.5 h and 3 h, followed by collecting pellets and staining by 25 μM PI. The fluorescence values were measured and normalized to the values of OD₆₀₀ (Fluorescence/OD₆₀₀).

E. TEM analysis of bacterial morphology mediated by Tse^VF. E. coli MG1655 harboring pBAD24 (control) or pBAD‑sec-tse^VF965–1141 was induced by 0.2 % L-arabinose or inhibited by 0.2 % D-glucose for 3 h, followed by adding 4 % formaldehyde and observing bacterial morphology by negative staining. Fields of view with different scales were shown (from left to right, 1 μm, 500 nm, and 500 nm, respectively).

F. Liposome sedimentation assay of Tse^VF. SDS-PAGE analysis (a) and quantification of the percentage of precipitated protein (b). The liposomes used were 100 % POPE, DOPE, POPC, DOPE or SM. S: supernatant; P: pellet. The percentages of Tse^VF precipitated by the lipids (Tse^VF binding) were calculated from the grayscale value ratios of the Tse^VF bands in the pellet and supernatant. Data in (b) represent the mean ± SD from three independent experiments. Statistical significance was determined by t-test (*, P < 0.05; **, P < 0.01).

To comprehensively analyze the evolutionary conservation and functional diversity of the Tse^VF C-terminal domain, a systematic Position-Specific Iterated BLAST (PSI-BLAST) search was performed against the NCBI non-redundant (NR) protein database (accessed December 7–9, 2023) using an initial e-value cutoff of 10⁻⁶ and a maximum of 5000 hits per iteration. The resulting sequences were filtered based on three stringent criteria: (1) identity ≥30 %, (2) coverage ≥50 %, and (3) e-value threshold ≤ 10⁻⁹ (Huang et al., 2022). Subsequently, taxonomic distribution analysis was performed on the identified homologs and visualized using the SankeyMATIC software (http://sankeymatic.com/). Domain prediction and annotation for the 277 retrieved sequences were performed using the Batch CD-Search Tool (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi) (Wang et al., 2023).

The GC content of the region from auxiliary cluster b (tssD2_b-tssI2_b) was calculated by DNA Features Viewer 3.0.1 package in Python v.3.7 (Zulkower and Rosser, 2020). The homologous sequences of Tse^VF were identified by BLASTp based on NR database.

3. Result

3.1. Tse^VF-Tsi^VF is a novel VflT6SS2 antibacterial effector-immunity pair

Our previous work has shown that VflT6SS2 is functionally expressed and is closely related to the competitive survival of V. fluvialis (Huang et al., 2017). We assumed that besides the identified effectors TssI2 (Huang et al., 2022) and Tle1^VF (manuscript in preparation), there are other potential effectors contributing to the competitive advantage of V. fluvialis. Comparison of V. fluvialis orphan cluster b and the counterpart of V. cholerae (auxiliary cluster 2) revealed that the three downstream genes within the vgrG/tssI polycistrons share no homologies (Fig. 1A) (Zhang et al., 2021). In V. cholerae, vasW, vasX and tsiV-2, the downstream genes of vgrG-2, respectively, encode a T6SS adaptor, a pore-forming effector and its immunity (Joshi et al., 2017). While in V. fluvialis, the follow-up genes of tssI2_b encode three unknown function proteins. Further sequence analysis showed that the GC contents of AL536_07425, AL536_07420 and AL536_07415 are 38 %, 43 % and 41 %, respectively, much lower than those of the flanking sequences (around 50 %) (Fig. 1B), indicating that these genes are probably acquired horizontally and potentially encode additional effector and immunity proteins. The gene product of AL536_07425 is 176 amino acid-long and residues 86 to 132 have a 29.79 % identity with V. cholerae chaperone protein Tap-1 (Unterweger et al., 2015), hence we named it tap^VF in V. fluvialis 85,003. AL536_07420 and AL536_07415 are overlapping genes on the same strand with a three-nucleotide overlap. AL536_07420 protein has a total length of 1197 amino acid residues, with a LysM domain (residues 2–49) at the N-terminus and no other predicted domains. AL536_07415 protein is 333 amino acid-long and contains a signal peptide (residues 1–18) and a tetratricopeptide repeat-like (TRP-like) domain (residues 17–258), which is commonly involved in protein-protein interaction (Suzuki et al., 2003). We reasoned that AL536_07420 and AL536_07415 encode a potential T6SS effector and immunity pair in V. fluvialis and therefore named them as tse^VF and tsi^VF.

Fig. 1.

Tse^VF-Tsi^VF is an antibacterial effector-immunity pair.

A. Genetic organization of the orphan cluster b (Orph. cluster b) of VflT6SS2 in V. fluvialis and that of the auxiliary cluster 2 (Aux. cluster 2) of T6SS in V. cholerae N16961.

B. GC content analysis of the orphan cluster b (Orph. cluster b).

C-D. Tse^VF mediated bacterial killing activity. Viable counts of E. coli MG1655 (C) and Δtse^VF-tsi^VF (D) before (T 0) and after (T 5) co-incubation with the indicated attackers (WT and its mutants) on LB agar plates containing 340 mM NaCl at 30 °C. PBS was used as blank control, and ΔvasK as T6SS⁻ control. Statistical analysis was performed by t-test with the data of survived E. coli MG1655 (C) or Δtse^VF-tsi^VF (D) at T 5 (*, P < 0.05; **, P < 0.01, ***, P < 0.001).

E. Viable counts of Δtse^VF-tsi^VF containing complement plasmid pSRKTc-tsi^VF or pSRKTc empty vector before (T 0) and after (T 12) co-incubation with the indicated attackers (WT and its mutants) on LB agar plates containing 340 mM NaCl and 1 mM IPTG at 30 °C. Statistical analysis was performed by t-test with the data of survived preys between groups at T 12 (***, P < 0.001).

F. Viable counts of different Vibrio strains (I-V) before (T 0) and after (T 5) co-incubation with the V. fluvialis WT and its mutants on LB agar plates containing 340 mM NaCl at 30 °C. Statistical analysis was performed by t-test with the data of survived preys between groups at T 5 (**, P < 0.01).

To verify the above assumption, we constructed Δtap^VF-tse^VF and Δtse^VF-tsi^VF mutants based on V. fluvialis WT and tested their bactericidal activity. Mutants ΔvasK or ΔvasH were used as T6SS negative controls (Huang et al., 2022). The bacterial competition assay showed that the WT could strongly inhibit E. coli MG1655 survival following 5 h of co-incubation (Fig. 1C). Both Δtap^VF-tse^VF and Δtse^VF-tsi^VF mutants exhibited similar reductions in antibacterial activity relative to WT (Fig. 1C). It is rational that the decrease in bactericidal effects of Δtap^VF-tse^VF and Δtse^VF-tsi^VF was not as strong as ΔvasK mutant where VflT6SS2 is fully unfunctional and other bactericidal effectors like TssI2 (Huang et al., 2022) still play a role in above double mutants (Fig. 1C). So, based on the structural and functional predictions of Tap^VF, Tse^VF and Tsi^VF, as well as the consistently reduced bactericidal activity of Δtap^VF-tse^VF and Δtse^VF-tsi^VF, we can reasonably infer that Tse^VF is an effector, with Tsi^VF being the immunity of Tse^VF. However, the role of Tap^VF as a dedicated chaperone warrants further investigation.

To further investigate the function of Tse^VF and Tsi^VF, we performed self-intoxication assay with Δtse^VF-tsi^VF mutant as the prey. As presented in Fig. 1D, WT attacker efficiently killed Δtse^VF-tsi^VF, but both Δtse^VF-tsi^VF and Δtap^VF-tse^VF attackers failed to do so, similar to ΔvasK. These results established that Tse^VF had the bactericidal ability and the lack of Tsi^VF resulted in the loss of protection against Tse^VF-mediated self-intoxication, indicating that Tsi^VF is the immunity of Tse^VF effector. Consistently, relative to pSRKTc vector control, complementation with pSRKTc-tsi^VF significantly enhanced the resistance of Δtse^VF-tsi^VF to WT attacker, providing definitive evidence that Tsi^VF is the cognate immunity for Tse^VF (Fig. 1E). But we noticed that the Tsi^VF conferred full protection requires a longer incubation period (12 h) and the standard 5 h offers only partial protection, which is likely attributable to the different expression levels of Tsi^VF. Together, these findings established that Tse^VF-Tsi^VF is a novel effector-immunity pair that contributes to the antibacterial activity of VflT6SS2 in V. fluvialis.

To assess the ecological relevance of Tse^VF-mediated antagonism, we examined its activity against co-occurring Vibrio pathogens, including V. cholerae (toxigenic and non-toxigenic), V. vulnificus, V. parahemolyticus, and Vibrio alginolyticus. In bacterial competition assays, the WT exhibited stronger antibacterial activity against all tested strains than the VflT6SS2-disabled ΔvasH. For Δtse^VF-tsi^VF, the loss of Tse^VF partially alleviated the killing ability of V. fluvialis against non-toxigenic V. cholerae 7743, V. vulnificus ABH2018-w-021, V. alginolyticus ATCC17749, and V. parahemolyticus 1200, but showed no inhibitory effect on the killing ability against V. cholerae toxigenic strain A1552 (Fig. 1F). The differential killing outcomes likely reflect the diverse effector and immunity protein repertoires among Vibrio species. V. cholerae A1552 may harbor a Tse^VF-Tsi^VF homolog, thereby exhibiting resistance to Tse^VF killing, while other tested Vibrio strains lack such matching homologs and are susceptible to Tse^VF toxicity. These findings demonstrated that VflT6SS2 provides V. fluvialis with a competitive advantage in shared ecological niches, mediated through the species- and strain-specific bactericidal activity of Tse^VF effector.

3.2. The C-terminal of Tse^VF exerts bactericidal effect in periplasmic space and disrupts cell envelope integrity

To further investigate the bactericidal mechanism and functional cellular localization of Tse^VF, we heterologously expressed it in E. coli MG1655 (Fig. 2A). Results showed that only when fused with an N-terminal Sec signal peptide, Tse^VF exhibited bactericidal activity, indicating that Tse^VF functions in the periplasmic interstitium. Then we scanned the InterPro database for putative domain in Tse^VF and transmembrane structures were determined with the Phobius tool. The results revealed that Tse^VF consists of an N-terminal non-cytoplasmic region (residues 1–969) and a C-terminal cytoplasmic region (residues 970–1197). In terms of the N-terminal, the region of residues 1–59 contains a LysM domain and the region of residues 941–969 possesses a transmembrane domain (Fig. 2B). Meanwhile, three-dimensional (3D) structural prediction using I-TASSER showed that residues 666–1141 are in the active region of Tse^VF (Fig. 2B). Based on these information, a series of recombinant plasmids expressing different fragments or domains were constructed and introduced into E. coli MG1655 to measure their bactericidal activities. As shown in Fig. 2C, periplasmic expression of LysM domain (pBAD‑sec-lysM) had no toxic effect on E. coli MG1655, and the absence of LysM in Tse^VF (pBAD‑sec-tse^VFΔlysM) had no significant influence on bactericidal activity compared to its full-length counterpart (pBAD‑sec-tse^VF in Fig. 2A). Therefore, the LysM domain does not seem to be necessary for Tse^VF to function in the periplasmic. The heterologous introduction of pBAD‑sec-tse^VF1–947 (the N-terminal non-cytoplasmic region excluding transmembrane domain), pBAD‑sec-tse^VF1–969 (the N-terminal non-cytoplasmic region including transmembrane region) and pBAD24‑sec-tse^VF965–1197 (the C-terminal cytoplasmic region) showed varying degrees of inhibitory effects on the growth of E. coli MG1655 (Fig. 2C). Compared to the full-length construct pBAD‑sec-tse^VF (Fig. 2A), the bactericidal activity of pBAD‑sec-tse^VF1–947 and pBAD‑sec-tse^VF1–969 was generally weaker, but that of pBAD‑sec-tse^VF965–1197 was the strongest. Therefore, we inferred that the primary functional domain of Tse^VF may is likely located in its C-terminal cytoplasmic region. Then, pBAD‑sec-tse^VF666–964 and pBAD‑sec-tse^VF965–1141 were constructed for further evaluation. As shown, the bactericidal activity of pBAD‑sec-tse^VF965–1141 was similar, or even stronger than that of pBAD‑sec-tse^VF965–1197, strongly indicating that the C-terminal fragment of residues 965–1141 is the dominant contributor to Tse^VF's bactericidal effect, while the N-terminal region may harbor additional, independent toxic function.

Periplasmic-acting effector generally targets the membrane or PG layer of cell wall. To further determine the specific function of Tse^VF, we first cloned and purified Tse^VF with a C-terminal 6× His tag and tested its ability to catalyze PG hydrolysis. The results of UPLC-MS analysis showed that Tse^VF-His₆ cannot hydrolyze cell wall PG (Fig. S1). Then, the membrane permeabilization assay using PI was performed to evaluate membrane damage upon Tse^VF965–1141 exposure, where PI fluorescence indicates cytoplasmic entry due to membrane damage. As shown in Fig. 2D, compared to control vector pBAD24 and glucose-repressed condition, induction of pBAD‑sec-tse^VF965–1141 with 0.2 % L-arabinose caused significantly high level of fluorescence intensity, thus indicating the clear membrane injury of MG1655. In accordance with this, under arabinose-induced condition, we observed that the cell contents of E. coli MG1655 containing pBAD‑sec-tse^VF965–1141 seemed to be leaked out and the cells underwent significant morphological changes from rod-shaped state to oval cell shape under TEM (Fig. 2E). This morphological change is different from that caused by the effector proteins targeting the cell wall (M. Liu et al., 2023), and is more similar to the morphological phenotype when using 8 % SDS to destroy other cell components (such as lipids), retaining only cell wall sacculus (Wang and Dong, 2021). Therefore, we hypothesized that Tse^VF might interact directly with cell membrane. To confirm this hypothesis, we performed liposome sedimentation assay (Jia et al., 2022) to test the interaction between Tse^VF and membrane lipids. As shown in Fig. 2F, Tse^VF-His₆ was precipitated after incubation with liposomes containing phosphatidylcholine (i.e., DOPC and POPC) and phosphatidylethanolamine (i.e., DOPE, POPE). In contrast, no strong binding was observed between Tse^VF-His₆ and Sphingomyelin (SM). This means that Tse^VF can bind to bacterial cell membrane components to achieve bactericidal effects. Together, these results demonstrated that Tse^VF likely functions as a membrane pore-forming toxin and the C-terminal residues 965–1141 are essential for its bactericidal activity.

3.3. Two TRP residues are important for the bactericidal function of the C-terminal of Tse^VF

Through the above research, we have preliminarily determined that the main functional domain of Tse^VF is located at the C-terminal region comprising amino acids 965–1141, which has only 177 residues, accounting for only 14.78 % of the total length. Therefore, to further dissect the sequence feature, PSI-BLAST was performed to search for more homologs of the Tse^VF C-terminal region in the NR protein database of NCBI (accessed on Dec 27, 2023) and 277 homologs were identified. Sequence alignment of these 277 homologs showed many conserved residues at certain sites (Fig. 3A). Based on polarity and hydrophobicity, five amino acid residues were selected for site-directed mutagenesis to evaluate their roles in the bactericidal activity of Tse^VF965–1141, these are E1009, E1071, R1076, W1056 and W1091. Their corresponding recombinant plasmids were introduced into E. coli MG1655 for bacterial toxicity assay. Our results showed that only the W1056A and W1091A mutants displayed significantly reduced bactericidal activity compared to their wild type, indicating that the two Trp residues are critical for the bactericidal activity of Tse^VF (Fig. 3B).

Fig. 3.

Tse^VF homologous proteins are widespread distributed T6SS effectors with two Trp residues critical for the bactericidal function.

A. Conservative site analysis of the C-terminal of Tse^VF. Multiple sequence alignment of Tse^VF homologs retrieved from PSI-BLAT was used to identify conserved sites. Conservative sites were displayed using WebLogo (https://weblogo.berkeley.edu/). The red arrows represent the conserved sites selected for subsequent experiments.

B. Bactericidal activities mediated by the C-terminal of Tse^VF and its mutants. Survival of E. coli MG1655 containing indicated Tse^VF constructs was counted by gradient dilutions of bacterial suspensions before (T 0) and after (T 3) 0.2 % L-arabinose induction.

C. Taxonomic distribution of Tse^VF C-terminal homologs. The distribution of homologs is displayed at different taxonomic levels (from left to right, order, family, genus and species, respectively).

D. The co-occurrence types of tse^VF and T6SS related genes. (a-b) The genetic organization of tse^VF homologs is consistent with the organization in V. fluvialis 85,003. Polymorphism can be seen in the flanking genes of tse^VF (b); (c) tse^VF homologous genes are distributed near the main cluster of T6SS; (d) tse^VF homologous genes are distributed downstream of vgrG or paar. Homologous genes are color-matched; hypothetical genes are shown in white.

3.4. Tse^VF homologous proteins are all potential effectors of T6SS

We annotated the above 277 proteins retrieved from PSI-BLAST at the genus level, with 229 belonging to Vibrio (82.67 %), 30 belonging to Salinivibrio (10.83 %), 10 belonging to Enterovibrio (3.61 %), 6 belonging to Marinomonas (2.17 %), 1 belonging to Grimentia (0.36 %), and 1 belonging to Shewanella (0.36 %) (Fig. 3C). Within the Vibrio genus, V. cholerae (44.10 %) and V. fluvialis (20.52 %) were the most prevalent species. Interestingly, we found that except for two homologous sequences from Enterovibrio, the N-terminus of the remaining 275 sequences containing the Tse^VF965–1141 homologous region also contained LysM domains.

By aligning 277 homologous sequences and removing those shorter than 200 amino acids or lacking complete flanking regions, 52 sequences were finally retained and their gene organization structures were drawn (Fig. 3D). All 52 tse^VF homologous genes co-existed with tsi^VF and tap^VF homologs located around T6SS related genes. The genomic arrangement of tse^VF can be classified into three forms (Fig. 3D). The first type is that the organization structure of tse^VF is consistent with the V. fluvialis 85,003, where tse^VF located in an orphan cluster, between the tap^VF and tsi^VF and after the hcp-vgrG (Fig. 3D (a-b)). In V. cholerae DRC187 and Enterovibrio norvegicus 1F-211, a new triad of chaperone-effector-immunity is linked behind and forward the tse^VF gene cluster, respectively. And in V. cholerae 3566–06 and RFB05, the new triad of chaperone-effector-immunity replaces the triad with high homology to Tap^VF-Tse^VF-Tsi^VF. In V. cholerae DRC187 and RFB05, there are autonomous transposable elements (ISPsv4, ISPcc6, ISVa18) nearby the new triad of chaperone-effector-immunity, suggesting their horizontally transferred feature. The second type of the distribution of tse^VF is characterized by its near location with the main cluster (Fig. 3D (c)). For Marinomonas foliarum strain JZW, the distribution pattern of tse^VF homologous gene is similar to the first type, which is located downstream of hcp-vgrG in the form of tap^VF-tse^VF-tsi^VF, and the orphan cluster (hcp-vgrG-tap^VF-tse^VF-tsi^VF) is located upstream of the T6SS main gene cluster. While in Vibrio viridastus LJC006, the tse^VF homologous gene is directly located downstream of the major cluster vgrG in the form of tap^VF-tse^VF-tsi^VF. The third type of tse^VF arrangement is marked by associating with single vgrG or paar (Fig. 3D (d)). Overall, these analyses suggest that Tse^VF homologous proteins are effectors of T6SS and exist as the tap^VF-tse^VF-tsi^VF form near T6SS related genes (Fig. 3D).

3.5. Tsi^VF directly interacts with Tse^VF through the middle region of TRP-like domain

The immunity Tsi^VF is located downstream of and adjacent to Tse^VF and can efficiently restrain the Tse^VF-mediated self-toxication in V. fluvialis 85003 (Fig. 1A and 1E). In addition, the N-terminal region (residues 1–18) of Tsi^VF contains a signal peptide that could help transport Tsi^VF to the periplasmic space, consistent with the results of cellular localization and functional activity of Tse^VF heterologous expression in E. coli (Fig. 2). The organizational and functional association between Tse^VF and Tsi^VF implies the interaction between them. The generated molecular docking model predicted that the middle region (residues 100–180) of the TRP-like domain (residues 17–258) of Tsi^VF can directly bind to the region of residues 113–853 in Tse^VF (Fig. 4A, 4D and Table S3). Therefore, bacterial two-hybrid assay was performed to confirm that Tsi^VF can directly interact with Tse^VF (Fig. 4B). Based on the functional complementation of T18 and T25 fragments of Bordetella pertussis adenylate cyclase, the interacting proteins can functionally reconstitute the activity of adenylate cyclase, which results in a Cya+ phenotype, i.e., blue colonies on LB agar plates supplemented with X-gal. Thus, Tsi^VF without the N-terminal signal sequence (residues 1–18) was cloned into the C-terminal of the T25 polypeptide in pKT25 or the N-terminal of it in pKTN25, while Tse^VF was cloned into the C-terminal of the T18 polypeptide in pUT18C or the N-terminal of it in pUT18. Cya+ phenotype was observed when T25-Tsi^VF19–333 or Tsi^VF19–333-T25 was co-expressed with T18-Tse^VF, suggesting a specific interaction between Tse^VF and Tsi^VF19–333 in the cytoplasm (Fig. 4B). We interpreted that the lack of interaction between Tse^VF-T18 and T25-Tsi^VF19–333 or Tsi^VF19–333-T25 is probably due to the steric hindrance somehow caused by the fusion of the adenylate cyclase T18 domain to the C-terminus of Tse^VF (Park et al., 2018). We further examined the interaction intensity between Tse^VF and Tsi^VF19–333 by ITC analysis, and the result revealed a high binding ability between them with a dissociation constant (K_d) of 0.15 ± 0.18 μM (Fig. 4C). Subsequently, we performed Co-IP assay to further confirm the specific interaction between Tse^VF and Tsi^VF19–333. Our results demonstrated that Tse^VF can directly interact with Tsi^VF19–333, but not with Tsi^{VFΔ100–180} which is lack of the region of residues 100–180 (Fig. 4E). Consistently, self-intoxication assay also confirmed that the deletion of this region of Tsi^VF abolished its protection against the bacterial killing effect of Tse^VF (Fig. 4F). Therefore, these results demonstrated that the intermediate region of the TRP-like domain of the cognate immunity Tsi^VF specifically binds to and antagonizes the bactericidal activity of Tse^VF effector.

Fig. 4.

Tse^VF can directly interact with the middle region of Tsi^VF.

A. Predicted molecular docking model of the Tse^VF-Tsi^VF complex. Tse^VF is shown in pink and Tsi^VF is shown in indigo.

B. BACTH analysis of the interaction between Tse^VF and Tsi^VF. Protein-protein interaction was detected in E. coli BTH101 cells by the functional complementation of T18 and T25 fragments.

C. ITC analysis of the specific binding between Tse^VF and Tsi^VF. The top panel shows the raw calorimetric data of the interaction, and the bottom panel shows the integrated heat variation. The measured K_d is indicated.

D. Domain architecture and structural features of Tsi^VF. The secondary structure schematic and a predicted structure of Tsi^VF are shown. The region encompassing residues 100–180 is highlighted.

E. Co-IP analysis of the interaction between Tse^VF and Tsi^VF. Cells co-expressing the pSRKTc vector and pBAD33-tsi^VF19–333-myc were used as negative control.

F. Viable counts of Δtse^VF-tsi^VF complemented with pSRKTc-tsi^VF or pSRKTc-tsi^{VFΔ100–180} were assessed before (T 0) and after (T 12) co-incubation with the indicated attacker strains (WT or ΔvasK). Statistical analysis was performed by t-test with the data of survived preys between groups at T 12 (***, P < 0.001).

3.6. Tse^VF delivery requires VgrG_B (TssI2_b)

Since we have determined that Tse^VF is a novel effector protein, we next explored the secretion characteristics of Tse^VF to gain a better understanding of its function. In general, the delivery of T6SS effectors requires interaction with adjacent structural proteins such as Hcp, PAAR, and VgrG. VflT6SS2 harbors at least three VgrGs: the main cluster VgrG (TssI2) encoded by tssI2 (Huang et al., 2017, 2022), and two orphan VgrGs (TssI2_a and TssI2_b), which are encoded by tssI2_a and tssI2_b and located within separate orphan hcp-vgrG clusters (Huang et al., 2017). Notably, tssI2_b is positioned directly upstream of the tse^VF (Fig. 1A). We assessed the effect of the upstream protein VgrG_b (TssI2_b) on Tse^VF delivery via bacterial competition assays. As expected, the absence of VgrG_b (TssI2_b) effectively reduced the killing ability towards Δtse^VF-tsi^VF to the same level as T6SS-negative mutant ΔvasH (Fig. 5A), while the deletion of the main gene cluster VgrG (TssI2) had no effect, and the deletion of VgrG_a (TssI2_a) partially compromised the ability, indicating that Tse^VF delivery mainly depends on VgrG_b (TssI2_b) to fulfil its bactericidal activity. To further rule out the possibility that the defect of killing activity in the ΔtssI2_a and ΔtssI2_b mutants was due to impaired T6SS assembly or function, we examined the expression and secretion of Hcp in corresponding mutants. As shown in Fig. 5B, no significant changes were observed in the absence of three VgrGs compared to the WT. This confirms that the overall T6SS machinery remains intact in these mutants, and thus the reduced antibacterial activity is likely due to specific defects in effector delivery rather than a global inactivation of the system.

Fig. 5.

The secretion of Tse^VF depends on VgrG_b (TssI2_b).

A. Survival of the Δtse^VF-tsi^VF prey strain was assessed before (T 0) and after (T 5) co-incubation with the indicated attackers (WT or its mutant). Statistical analysis was performed by t-test with the data of survived preys between groups at T 5 (****, P < 0.0001).

B. Analysis of Hcp expression and secretion. WB analysis of Hcp in whole-cell lysates (pellet) and culture supernatants (supernatants) from the indicated strains. Equal amounts of total protein (10 µg) were loaded per lane. The positions of molecular mass markers (kDa) and the Hcp and Crp proteins are indicated.

C. Sequence alignment analysis of TssI2_a and TssI2_b. The dotted underline represents the conserved domains of VgrG predicted by Interpro. The underlined double line represents the region predicted to be the gp5 domain. The region framed by black line indicates the variable region where TssI2_a and TssI2_b are different.

D. Survival of E. coli MG1655 was assessed before (T 0) and after (T 12) co-incubation with the indicated attackers (WT or its mutant harboring corresponding plasmids). Statistical analysis was performed by t-test with the data of survived preys between groups at T12 (****, P < 0.0001).

E. Pull-down analysis of the interaction between Tse^VF and VgrG_b. His-tagged Tse^VF specifically binds to Flag-tagged TssI2_b instead of Flag-tagged TssI2_a and Flag-tagged TssI2_bΔC.

Through sequence alignment analysis, we found that VgrG_a (TssI2_a) and VgrG_b (TssI2_b) share high homology (90.92 %) except for their polymorphic C-terminal 60 amino acids (Fig. 5C). To further confirm that Tse^VF depends on the C-terminus of VgrG_b (TssI2_b) for delivery and bactericidal activity, we constructed recombinant plasmids expressing VgrG_a (TssI2_a), VgrG_b (TssI2_b), and VgrG_bΔC (TssI2_bΔC) lacking the C-terminus and tested for their ability to complement the killing defect of ΔtssI2_b with bacterial competition assay (Fig. 5D). Complementation of the ΔtssI2_b mutant with full-length tssI2_b construct pFlag-tssI2_b restored bactericidal activity to the WT level, whereas complementation with pFlag-tssI2_a or pFlag-tssI2_bΔC did not. Furthermore, we noticed that the complementation of pFlag-tssI2_bΔC leads to an even lower level of bactericidal activity compared to control vector and pFlag-tssI2_a. The reason behind this phenomenon remains unclear, we preliminarily infer that the high quantity of truncated TssI2_bΔC due to overexpression could probably compete with TssI2_a and/or TssI2 for forming VgrG spike, thus interfering with the loading and therefore the killing activity of corresponding effectors they carried. At the same time, we performed pull-down analysis to verify the physical interaction between Tse^VF and TssI2_b (VgrG_b). The results showed that Tse^VF can interact with the full length of TssI2_b (VgrG_b), but not with VgrG_a (TssI2_a) and VgrG_bΔC (TssI2_bΔC), as evidenced by the presence and absence of the corresponding co-precipitated protein bands (Fig. 5E). Therefore, we demonstrated that the C-terminal of TssI2_b (VgrG_b) is involved in the secretion and bactericidal function of Tse^VF by mediating its loading to the VflT6SS2.

4. Discussion

This work identified a novel effector-immunity pair Tse^VF-Tsi^VF encoded by the orphan cluster b of VflT6SS2 and further analyzed the underlying mechanism of this pair in bacterial competition. The effector Tse^VF exerts bacterial killing activity in the periplasm, mainly through the C-terminal cytoplasmic domain (residues 965–1141), which disrupts the integrity of cell membrane, leading to the leakage of cellular contents to mediate the killing effect toward co-existing bacteria. Two Trp residues (Trp-1056 and Trp-1091) are important for the bactericidal function of Tse^VF965–1141 region. Immunity protein Tsi^VF directly interacts with Tse^VF through the middle region of its TRP-like domain, antagonizing the bactericidal activity of Tse^VF. Tse^VF, as a “cargo” effector, interacts with the C-terminal of TssI2_b (VgrG_b) to load itself and to fulfill the bactericidal activity.

T6SS-wielding bacteria typically employ various effectors for the interbacterial competition and host pathogenicity. VflT6SS2 endows V. fluvialis with a survival advantage by employing effectors TssI2 and Tse^VF to inhibit other species competing for a common niche. Interestingly, we noticed that the interbacterial competitions mediated by the two effectors display species- and strain-specific effects. TssI2 shows interbacterial toxicity to the pandemic V. cholerae strain A1552 and V. alginolyticus strain ATCC17749, but not to V. vulnificus strain ABH2018-w-021, pandemic V. cholerae strain C7258 and non-toxigenic strain 93,097 (Huang et al., 2022). By contrast, Tse^VF cannot mediate the killing to the V. cholerae pandemic strain A1552, but can competitively inhibits the V. cholerae non-toxigenic strain 7743, V. alginolyticus ATCC17749, V. parahemolyticus 1200 and V. vulnificus ABH2018-w-021 (Fig. 1F). This observed specificity aligns with the paradigm that T6SS competition outcomes are governed by effector-immunity pair compatibility (Unterweger et al., 2014). The differential susceptibility among Vibrios suggests that natural populations harbor diverse effector-immunity modules. Our bioinformatic analysis provides a potential explanation for the observed broad activity of Tse^VF. A search for homologs of its functional domain (Tse^VF965–1141) retrieved a limited set of only 177 sequences in nature (Fig. 3C). Drawing an analogy to intraspecific competition in V. cholerae, we hypothesize that this scarcity may underlie Tse^VF's effectiveness: many Vibrio strains, having never or rarely encountered this specific module, likely lack the cognate immunity protein Tsi^VF, rendering them susceptible (Unterweger et al., 2014). Conversely, the more constrained activity profile of TssI2 may reflect a scenario where its corresponding immunity is more widespread due to a broader distribution of TssI2-like modules (Huang et al., 2022). Thus, the complementary targeting strategies of TssI2 and Tse^VF—shaped by the distinct prevalence and distribution of their corresponding immunity genes—together furnish V. fluvialis with a versatile competitive arsenal in its ecological niche.

Our data establish that Tse^VF is a novel membrane-targeting effector sharing no homology with previously characterized T6SS effectors. With constructs expressing different cellular localized full-length Tse^VF proteins, we firstly confirmed that periplasmically-located Tse^VF exerted bactericidal effect (Fig. 2A). Through screening various Tse^VF truncates, we narrowed the primary functional domain of Tse^VF to its C-terminal 965–1141 amino acids fragment (Fig. 2B and 2C). Then, by employing membrane permeabilization assay, we demonstrated that Tse^VF965–1141 caused cell membrane damage, allowing PI to enter the cells (Fig. 2D). Subsequent TEM analysis of cell morphological change revealed that Tse^VF965–1141 intoxicated cells lost normal rod shape and became rounded and enlarged, while the bacterial cell wall still existed or maintained (Fig. 2E). Consistently, UPLC analysis showed that Tse^VF did not disrupt the peptidoglycan network of the cell wall (Fig. S1). Finally, liposome sedimentation assay verified that Tse^VF could directly interact with phosphatidylcholine and phosphatidylethanolamine, the lipid components of the bacterial cell membrane (Fig. 2F). Collectively, these results demonstrate that Tse^VF is a bacteriolytic effector that targets and disrupts the integrity of the bacterial inner membrane. In addition, due to the lack of the lipase specific motifs, such as GxSxG and Ser-Asp-His catalytic sites or HxKxxxxD motif (Russell et al., 2013), Tse^VF is not a lipase effector. Instead, its functional profile—rapid membrane permeabilization coupled with lipid binding—is reminiscent of pore-forming toxins (PFTs) such as VasX, and its genetic linkage to orphan VgrG_b (Fig. 1A) further supports this functional analogy (Miyata et al., 2011).

Based on its mode of action, we strongly hypothesize Tse^VF might function as a PFT. PFTs are categorized as α-PFTs and β-PFTs based on whether the secondary structure of their transmembrane elements is composed of α-helices or β-barrels, respectively (Dal Peraro and van der Goot, 2016). The α-PFTs undergo dramatic conformational changes and complex oligomerization pathways upon exposure to the cell membrane. The mechanism of pore formation by β-PFTs is relatively straightforward, involving helical organization for membrane insertion. AlphaFold2-predicted structure of Tse^VF showed that α-helix bundle is the main secondary structure of the functional C-terminal region (Fig. 2B and S2). Therefore, we speculate that Tse^VF may form pores on the bacteria cell membrane like an α-PFT. However, Tse^VF showed no correlation with the reported pore-forming T6SS effectors, such as VasX (Miyata et al., 2013), Tse4 (LaCourse et al., 2018), Tme (Fridman et al., 2020), in terms of sequence, predicted functional domain and protein structure. Indeed, the colicin pore-forming structures have been identified in the C-terminal of several predicted transmembrane helices-containing proteins, such as the Colicin B structure of the Vibrio coralliilyticus strain VCR_19,700 and the Colicin A structure of the Pseudoalteromonas haloplanktis strain PHAL_03780 (Salomon et al., 2014). However, Tse^VF was not predicted to contain any relevant structural domains, indicating that it is a potentially novel pore-forming toxin.

Multiple sequence alignment of the Tse^VF965–1141 homologs identified by PSI-BLAST revealed multiple conserved residues at specific sites (Fig. 3A), which are distinct from the almost invariant DxxK motif found in the Tme effector (Fridman et al., 2020). Through mutation analysis of five selected target sites, we further revealed that the conserved residues Trp-1056 and Trp-1091 are required for the bactericidal activity of Tse^VF965–1141, which are respectively located in two α-helices (Fig. 3B and S2). However, the underlying mechanism of W1056A and W1091A mutations on Tse^VF mediated bactericidal activity remains to be investigated. Tryptophan plays a critical role in membrane protein structure and function. Despite being the least abundant amino acid in soluble proteins (∼1.1 % in cytoplasmic proteins), its frequency increases significantly (∼2.9 %) in transmembrane α-helices of membrane proteins (Granseth et al., 2005; Barik, 2020). This enrichment reflects its unique biochemical properties: The indole side chain of Trp features a bulky, hydrophobic dual-ring structure, with the NH group capable of forming hydrogen bonds (Granseth et al., 2005; Barik, 2020). Additionally, its large π-electron surface enables strong hydrophobic and cation-π interactions, providing stabilization energy exceeding that of ionic bridges (Granseth et al., 2005; Barik, 2020). Notably, Trp residues frequently serve as interfacial anchors in lipid bilayers, positioning and stabilizing transmembrane helices in membrane proteins (Granseth et al., 2005; Sanchez et al., 2008). Therefore, we reasoned that the W1056A or W1091A mutation could probably modify the overall or the local conformational structure of Tse^VF965–1141, affecting protein stability or membrane lipids binding and the subsequent pore-formation, thus leading to a decrease in bactericidal phenotype against E. coli. Analogously, Morante et al. corroborated that the residue Phe-16 is critical for the pore formation of fragacea toxin C, an α-helical pore-forming toxin from an actinoporin protein family, in cholesterol-rich membranes (Morante et al., 2015). Quite recently, a pore-forming Vibrio-specific T6SS effector TseVs was reported, which shares no homology with Tse^VF and where several residues were also found to play essential roles in its dimerization and virulence (Liu et al., 2025).

An N-terminal lysin motif (LysM) domain is predicted for Tse^VF, which has been classified as the peptidoglycan-binding module attached to the N- or C-terminus of the catalytic module of a protein. Here, we did not observe obvious influence on the bactericidal activity of Tse^VF (Fig. 2A and 2C) by the deletion of LysM domain. In nature, proteins homologous to the C-terminal of Tse^VF also possess the LysM domain (99.26 %) (Fig. 3C), and these homologous proteins are all connected to T6SS related components (Fig. 3D). They are mainly distributed in three types in the form of tap-tse-tsi (Fig. 3D). The first type is behind the hcp-vgrG in auxiliary cluster (Fig. 3D (a-b)). The second type is that the tse^VF homologous sequence is adjacent to the main cluster of T6SS (Fig. 3D (c)). The third type is directly located downstream of an isolated vgrG or paar (Fig. 3D (d)). The above distributions mean that the tap-tse-tsi triad may be horizontally transferred as a functional unit and that the secretion of these Tse^VF-like effectors is likely dependent on an adjacent VgrG or PAAR. Indeed, in our study, we experimentally demonstrated that the secretion of Tse^VF relies on the direct interaction with the C-terminal of its upstream TssI2_b (VgrG_b) (Fig. 5C-E). The finding that Hcp secretion is unaffected in the ΔtssI2_a mutant (Fig. 5B) indicates that its partial defect in Tse^VF mediated killing is not due to impaired apparatus function. Instead, we speculate that the high sequence homology between TssI2_a and TssI2_b might allow TssI2_a to incorporate into the VgrG spike, and its absence could subtly compromise spike efficiency, leading to reduction in Tse^VF delivery (Liang et al., 2021). This model provides a plausible explanation for the intermediate phenotype (Fig. 5A) and suggests that the composition of the VgrG spike, influenced by homologous orphan VgrGs, could represent a nuanced regulatory layer for fine-tuning effector delivery. The exact molecular details of this interaction warrant further structural and biochemical investigation.

5. Conclusion

Our study identified a new effector-immunity module Tse^VF-Tsi^VF in V. fluvialis and systematically investigated the function and secretion mechanisms of Tse^VF, a novel pore-forming effector of VflT6SS2 sharing no homology with other known T6SS effectors. TRP-like domain-containing immunity Tsi^VF provides cognate protection through physically binding to Tse^VF. These findings further enhanced our understanding of the diversity and functions of T6SS effectors in V. fluvialis.

CRediT authorship contribution statement

Yu Han: Writing – original draft, Visualization, Investigation, Data curation. Weili Liang: Conceptualization, Writing – review & editing, Funding acquisition, Supervision, Resources. Biao Kan: Conceptualization, Funding acquisition, Resources. Yue Xiao: Writing – review & editing, Project administration. Yuanming Huang, Zhe Li: Visualization, Methodology. Ran Duan, Xiaorui Li, Kunkun Wang, Saisen Ji: Validation, Data curation.

Funding

This study was supported by grants from the National Key R&D Program of China (2023YFC2604400 and 2021YFC2300302) and the National Natural Science Foundation of China (82472292).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendices

Supplementary Table S1 Bacterial strains and plasmids used in this study.

Supplementary Table S2 Primers used in this study.

Supplementary Table S3 The molecular docking results of Tse^VF and Tsi^VF.

Supplementary Figure S1 UPLC chromatograms of PG hydrolysis products by lysozyme (red), recombinant Tse^VF (green) and TssI2^R1000A (black) (Huang et al., 2022). The purified PG was digested by Tse^VF, lysozyme or TssI2^R1000A, and reduced by sodium borohydride, then filtered by 0.22 μm filter membrane. The flow-through samples were collected for UPLC analysis.

Supplementary Figure S2 The overall structure of the C-terminal domain of Tse^VF is shown as a ribbon diagram. Five conserved residues selected for site mutation based on multiple sequence alignment of Tse^VF homologs are highlighted in green.