T6SS mutation enhances EHEC cytotoxicity in intestinal epithelial cells

To evaluate T6SS-mediated cytotoxicity, the trypan blue staining was used to assess cell death in the human colon carcinoma cell line (Caco-2) infection model. As shown in Fig 1A, the ΔT6SS mutant exhibited significantly higher cytotoxicity than wild-type (WT) bacteria. The cytotoxicity of EHEC in Caco-2 cells was quantified using flow cytometry. Flow cytometry revealed that after 11 h co-culture, ΔT6SS-infected Caco-2 cells exhibited 92.9% mortality, versus 40.1% in WT-infected controls (Fig 1B and 1C). Based on a previous report on T6SS mutants attenuating EHEC pathogenicity in murine models [27], we extended our analysis to MC-38 murine colon adenocarcinoma cells. As shown in Fig 1D, ΔT6SS elicited significantly enhanced cytotoxicity in MC-38 cells within 5 h post-infection compared to WT. Notably, MC-38 cell death occurred more rapidly than that in Caco-2 cells, suggesting a discordance between ΔT6SS-associated cytotoxicity and murine pathogenicity. To further validate T6SS function, we constructed a clpV deletion mutant. Similar to ΔT6SS phenotypes, ΔclpV infection caused 84.4% host cell death versus 48.1% in WT controls (Fig 1E and 1F). Complementation with clpV on a low-copy plasmid reversed this hypercytotoxic phenotype (50.4% mortality), confirming that T6SS dysfunction potentiates EHEC cytotoxicity (Fig 1E and 1F).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Virulence of EHEC wild type and T6SS-deficient mutants toward intestinal epithelial cells.

(A) Trypan blue exclusion assay assessing cytotoxicity. Human intestinal epithelial cells (Caco-2) were seeded in 12-well plates at 2 × 10⁵ cells/well and cultured in 5% CO₂ at 37°C for 48 h. Cells were challenged with 10⁷ CFU/well bacteria for 8 h prior to viability assessment. (B) Flow cytometric quantification of cytotoxicity. The infection protocol was identical to that in (A), with co-culture extended to 11 h for optimal detection. (C) Comparative mortality rates of Caco-2 cells infected with EHEC wild type versus ΔT6SS strains. (D) Time-course cytotoxicity profiles in MC-38 and Caco-2 cell lines exposed to EHEC. (E) Flow cytometry for cytotoxicity assessment of EHEC genetic derivatives. (F) Statistical comparison of cytotoxic effects across EHEC genetic derivatives. (G) The growth of EHEC wild type and T6SS deficient mutant during co-culture with Caco-2 cells. (H) Total bacterial counts after 4 h and 8 h of co-culture with Caco-2 cells, determined by plate counting. (I) Adherent bacterial counts after 4 h and 8 h of co-culture, determined by plate counting. (J) Bacterial adherent assay for EHEC and its derivatives. WT, wild type; ΔT6SS, T6SS deletion strain; ΔclpV, clpV mutant; ΔclpV/pclpV, clpV mutant strain complemented with clpV; vector (pACYC184) was used as a negative control. Data in (C, D, F, G-J) are presented as mean ± SD from three independent experiments. P values in (H) and (I) were determined by two-tailed unpaired Student’s t-tests; those in (C, F, J) by one-way ANOVA with Dunnett’s or Tukey’s multiple comparisons test. *p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant.

https://doi.org/10.1371/journal.ppat.1014039.g001

Impact of a T6SS mutation on bacterial growth rate and adhesion

In LB medium, the growth rates of the EHEC wild-type and ΔT6SS mutant were nearly identical (S1 Fig). However, under co-culture conditions with Caco-2 cells, the growth rate of the ΔT6SS mutant was slightly higher than that of the WT at the 4-hour time point, with no significant difference observed after 8 hours (Fig 1G and 1H). Nevertheless, the ΔclpV mutant, which also lacks a functional T6SS, displayed a growth rate identical to the WT yet caused stronger cytotoxicity, indicating that the observed growth advantage is not the primary driver of enhanced virulence. Regarding bacterial adherence, the ΔT6SS mutant demonstrated a stronger adhesion capacity compared to the WT at both 4 and 8 hours (Fig 1I). Similar to ΔT6SS phenotypes, ΔclpV also showed stronger adhesion capacity than WT and complementation with clpV on a low-copy plasmid could reduce the bacterial adhesion capacity (Fig 1J). Collectively, these results demonstrate that the T6SS mutation has a negligible effect on growth, but promotes EHEC adherence.

T6SS mutation promotes Φstx2 lysis and Stx2 liberation

Given that Φstx2 is lytically induced during EHEC infection in murine models [35,36], we hypothesized that T6SS inactivation would modulate Φstx2 excision during host colonization. Therefore, we quantified the released Φstx2 particles in EHEC-Caco-2 co-culture supernatants using quantitative real-time PCR (qPCR) targeting the excised phage genomes. Compared to WT strain, ΔT6SS and ΔclpV mutants exhibited 8.78-fold and 5.50-fold increases in excised Φstx2 genomes, respectively (Fig 2A), demonstrating enhanced prophage lytic activation upon T6SS disruption. Genetic complementation of clpV in the ΔclpV mutant restored Φstx2 lysis to WT levels. Furthermore, western blot analysis of Stx2A subunits confirmed an increase in mature toxin release. As shown in Fig 2B, the ΔT6SS and ΔclpV mutants exhibited 1.77-fold and 1.56-fold increases in Stx2A accumulation compared to the WT. Complementation of clpV in the ΔclpV background reversed Stx2A overproduction (Fig 2B). Collectively, these data establish that T6SS dysfunction potentiates Φstx2 excision and Stx2 release during host-pathogen interaction.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Identification of virulence factors associated with T6SS-deficient cytotoxicity in EHEC.

(A) Quantification of Stx2 phage particles released into the co-culture supernatants of EHEC and Caco-2 cells, measured by qPCR. (B) Immunoblot analysis of the secreted Stx2A subunit (toxin) using an anti-Stx2A monoclonal antibody. (C) Flow cytometric analysis of cytotoxicity in Caco-2 cells following co-culture with EHEC genetic variants. (D) Statistical comparison of cytotoxic effects across EHEC mutant strains. (E) Expression levels of recA and key T3SS genes, analyzed by qPCR. WT, wild type; ΔT6SS, T6SS mutant; ΔclpV, clpV mutant; ΔclpV/ΔT3SS, clpV and T3SS double mutant; ΔclpV/ΔΦ, clpV and Φstx2 double mutant; ΔclpV/ΔΦ/ΔT3SS, clpV, Φstx2 and T3SS triple mutant. ΔclpV/pclpV, clpV mutant complemented with clpV; a vector (pACYC184) was used as a negative control. Data in (B, D, E) are presented as the mean ± standard deviation (SD) from three independent experiments. P values were determined by one-way ANOVA with Dunnett’s or Tukey’s multiple comparisons test and two-tailed unpaired t-tests. *p < 0.05; ****p < 0.0001; ns, not significant.

https://doi.org/10.1371/journal.ppat.1014039.g002

Φstx2 as an essential component and the T3SS as the primary virulence factor within the cytotoxic network of T6SS-deficient EHEC

We hypothesized that the two major EHEC virulence factors, the Φstx2 prophage and the T3SS, mediate the enhanced cytotoxicity resulting from T6SS inactivation. To test this, we constructed isogenic mutants of Φstx2 and the T3SS core component escN and assessed their cytotoxic effects. Cytotoxicity quantification revealed a significant increase in host cell mortality from 53.4% for the wild type (WT) to 82.0% for the ΔclpV mutant (Fig 2C and 2D). In contrast, the Φstx2 single mutant (53.3%) did not significantly alter cytotoxicity. Notably, the ΔΦstx2/ΔclpV double mutant (58.3%) failed to show the hyper-cytotoxic phenotype of the ΔclpV single mutant, indicating that Φstx2 is required for the full manifestation of T6SS-deficient cytotoxicity.

Next, we examined the role of the T3SS. Inactivation of the T3SS in the ΔclpV background (ΔclpV/ΔescN double mutant) abolished the enhanced cytotoxicity, reducing mortality to 25.3%. This level was comparable to that of the ΔescN single mutant (25.6%) and the ΔescN/ΔclpV/ΔΦstx2 triple mutant (17.1%) (Fig 2D). These results demonstrate that while Φstx2 is an essential component of the cytotoxic network, the T3SS serves as the primary virulence factor executing cell death in T6SS-deficient EHEC in our infection model.

To investigate the link between T6SS deficiency and T3SS expression, we assessed the expression of three key T3SS genes using qPCR. As shown in Fig 2E, two of the three tested T3SS genes—ler (a positive regulator) and espA (a translocator)—were significantly upregulated in the ΔclpV mutant compared to the WT. This upregulation was reversed upon genetic complementation with clpV, indicating that T6SS inactivation specifically promoted the expression of these core T3SS components during cellular infection.

Furthermore, we examined the expression of recA, a central SOS response gene that can trigger Φstx2 prophage induction [37]. Consistently, recA expression was significantly elevated in the ΔclpV mutant and reduced to near WT levels following clpV complementation (Fig 2E). These results indicate that T6SS deficiency not only stimulates T3SS but also activates the SOS response, positioning recA as a potential regulatory node linking T6SS loss to downstream virulence pathways.

We also considered bacterial endotoxins as potential contributors to the enhanced cytotoxicity of T6SS-deficient EHEC. TAK-242 is a selective inhibitor of Toll-like receptor 4 (TLR4) signal transduction and suppresses lipopolysaccharide (LPS)-induced inflammation [38]. To this end, we added TAK-242 (final concentration: 10 μM) to the EHEC and Caco-2 cell co-culture model to determine whether cell death was mediated by endotoxin release. The results showed that TAK-242 neither reduced overall EHEC cytotoxicity nor diminished the cytotoxic difference between the ΔT6SS mutant and the WT strain (Fig 3A and 3B), thus ruling out endotoxin as a major driver of the enhanced T6SS-deficient phenotype.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Contribution of host ROS and bacterial endotoxin to the cytotoxicity of T6SS-deficient EHEC.

The roles of host-derived reactive oxygen species (ROS) and bacterial endotoxin were assessed using the antioxidant vitamin C (0.5 mM) and the TLR4 inhibitor TAK-242, respectively. (A) Flow cytometric analysis of cytotoxicity in Caco-2 cells co-cultured with EHEC and treated with 0.5 mM vitamin C or TAK-242. (B–D) Effects of vitamin C or TAK-242 on (B) cytotoxicity, (C) Stx2 phage liberation, and (D) expression of recA and key T3SS genes in wild type versus T6SS mutant EHEC. (E) Bacterial susceptibility to exogenous oxidative stress. (F) Φstx2 prophage liberation induced by oxidative stress. WT, wild type; ΔT6SS, T6SS mutant; ΔclpV, clpV mutant; ΔclpV/pclpV, complemented mutant; a vector (pACYC184) was used as a negative control. Data in (B-F) are presented as mean ± SD from three independent experiments. P values in (B-D) were determined by two-tailed unpaired Student’s t-tests; those in (E) and (F) by one-way ANOVA with Dunnett’s or Tukey’s multiple comparisons test. **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant.

https://doi.org/10.1371/journal.ppat.1014039.g003

Reactive Oxygen Species (ROS) as a key host signal driving T6SS-deficient cytotoxicity

We hypothesized that host-derived reactive oxygen species (ROS) play a key role in enhancing the cytotoxicity of T6SS-deficient EHEC. To test this hypothesis, we used the antioxidant vitamin C to neutralize ROS during infection. Supplementing co-cultures with 0.5 mM vitamin C significantly attenuated the cytotoxicity of both the WT and ΔT6SS strains, with a more pronounced suppressive effect on the ΔT6SS mutant (Fig 3A and 3B). Concomitantly, it markedly reduced Φstx2 prophage liberation in both strains (Fig 3C). A similar pattern was observed at the transcriptional level; vitamin C downregulated the expression of recA and the key T3SS genes ler and espA in both genetic backgrounds (Fig 3D). Collectively, these data establish host-derived ROS as the primary driver of enhanced cytotoxicity, prophage induction, and virulence gene upregulation in T6SS-deficient EHEC.

To assess the susceptibility of T6SS-mutated EHEC and its associated Φstx2 prophage to oxidative stress, we employed hydrogen peroxide (H₂O₂), a major reactive oxygen species (ROS). We first assessed bacterial growth under H₂O₂ stress. As shown in Fig 3E, in the absence of H₂O₂ treatment, all strains reached a final OD₆₀₀ of approximately 5. Upon treatment with 1.5 mM H₂O₂, the growth of all the strains was inhibited to varying degrees. The ΔclpV mutant showed a significant reduction in mean OD₆₀₀ (1.374 ± 0.044) compared to the WT (1.778 ± 0.033). This growth defect was partially rescued by clpV complementation, with the mean OD₆₀₀ recovering to 1.687 ± 0.045. These data demonstrate that T6SS deficiency renders EHEC more susceptible to oxidative stress. We next quantified Φstx2 prophage lysis and genome release. The ΔclpV mutant showed an approximately 15.96-fold increase in excised Φstx2 genomes compared to the WT, indicating enhanced prophage lytic activation upon T6SS disruption. Complementation with clpV significantly reduced this effect (Fig 3F). This demonstrates that the T6SS mutant is more sensitive to Φstx2 induction and liberation under oxidative stress.

In silico and functional analysis of EHEC T6SS

The EHEC strain EDL933 harbored a T6SS main gene cluster and two orphan loci encoding vgrG and vgrE (Fig 4A). Within the T6SS main gene cluster (Z0244-Z0272), the structural gene region (Z0248-Z0267) exhibited 99.72% nucleotide identity with its homologs in the aEPEC strain CB9615. A total of 26 mutated loci including 23 loci in protein-coding regions and 3 loci in non-coding regions were identified in this gene cluster during evolution (S1 Dataset). Among these loci, 23 loci were single nucleotide mutations, whereas three involved sequence insertions or deletions (S1 Dataset). Among the structural protein-coding mutations (19 in total), 4 were synonymous, whereas 15 resulted in non-synonymous amino acid substitutions (S1 Dataset). In vipB (Z0262), a CAG (Gln)→TAG (stop codon) nonsense mutation at positions 1390–1392 truncated the protein by 84 residues compared to ancestral vipB’ (G2583_0238) (S2A Fig, S1 Dataset). The gene tssM, which encodes an inner membrane protein with ATPase activity to energize T6SS secretion, was mutated by insertion of 28 bp repeated sequences (TGCTGATACTGGCGTGGATTTTTCTGCT). This mutation led to a change in the open reading frame (ORF) of EHEC (Fig 4B, S1 Dataset). clpV, a cytoplasmic AAA⁺ ATPase involved in protein degradation and disaggregation, was mutated by a 12 bp deletion. This mutation led to four amino acids (Ser₅₃₁-Glu₅₃₂-Ser₅₃₃-Glu₅₃₄) being deleted in ClpV of EHEC (S2B Fig, S1 Dataset).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Analysis of the T6SS degeneration gene cluster and its predicted effectors in EHEC EDL933.

(A) Schematic diagram of the genetic organization of EHEC T6SS. The T6SS gene cluster is divided into the main cluster (Z0244-Z0272), vgrG-2 orphan locus (Z0700-Z0707), and vgrE-2 orphan locus (Z2253-Z2262). Z0248-Z0267, Z0707, and Z2262 encode predicted structural proteins. Z0244-Z0247, Z0268-Z0272, Z0700-Z0706, and Z2253-Z2261 are predicted to encode T6SS effector proteins. etse denotes a predicted EHEC T6SS effector gene, and etsi denotes its cognate immunity gene. (B) Comparative analysis of tssM locus between ancestral aEPEC CB9615 and EHEC EDL933. (C) Comparative analysis of etse4 locus between ancestral aEPEC CB9615 and EHEC EDL933. (D) Assessment of E. coli growth inhibition upon expression of Etse4. (E) Assessment of E. coli growth inhibition upon co-expression of Etse4 and Etsi4 (pET22b-etse-i4). (F) Bacterial competition assay between EHEC (predator) and E. coli K-12 MG1655 (prey). The competition ratio (prey CFU/ predator CFU) was calculated. Comparisons: WT vs. K-12-N (without bile salts); ΔT6SS vs. K-12-N (without bile salts); WT vs. K-12-B (with 50 µg/mL bile salts); ΔT6SS vs. K-12-B (with 50 µg/mL bile salts). Data are presented as mean ± SD from three independent experiments. P values were determined by two-tailed unpaired Student’s t-tests. ns, not significant. (G) Assay for T6SS secretion activity. Hcp1 and Hcp2 are T6SS effectors in EHEC. Hcp2, a known T6SS-dependent secreted effector, served as a control. Under the same conditions, Hcp1 was not detected by Western blot.

https://doi.org/10.1371/journal.ppat.1014039.g004

Flanking the T6SS core structural genes, the effector-encoding regions (Z0244-Z0247 and Z0268-Z0272) exhibited elevated mutation rates (Fig 4A, 4C and S2C). Etse4 (Z0246), a predicted T6SS effector, shares 98.48% amino acid identity with the peptidoglycan-binding protein LysM. They target prokaryotic cells and inhibit the growth of competitors by preventing their cell wall formation. To validate this hypothesis, we heterologously expressed etse4 in E. coli BL21 cells using a pET expression system. While pET28a mediates cytoplasmic expression, pET22b incorporates an N-terminal pelB signal peptide that directs recombinant proteins to the periplasmic space. Our results showed that the expression of etse4 in the periplasmic space (pET22b-etse4) significantly inhibited bacterial growth, whereas negligible growth differences were observed when etse4 was expressed in the cytoplasm (pET28a-etse4) (Fig 4D). This suggests that the effector Etse4 may exert toxicity by targeting the periplasmic space of rival cells and assist EHEC in acquiring more fitness in bacterial competition. In addition, Etse4 toxicity was rescued by the co-expression of etsi4 (Z0245), a predicted cognate immunity protein of Etse4 (Fig 4E). Genomic analysis revealed a TTA (Leu)→TAA (stop codon) nonsense mutation at positions 220–222, resulting in division of the ancestral etse4_ori gene into two truncated ORFs: etse4a (Z0247) and etse4 in EHEC (Fig 4C, S1 Dataset). Despite retaining its functionality in heterologous systems, native Etse4 failed to exhibit antagonistic activity in the bacterial competition assay, even after stimulation with 50 mg/mL bile salt (Fig 4F). This functional discrepancy likely stems from the fragmentation of the ancestral etse4_ori locus, which causes the loss of a Shine-Dalgarno (SD) sequence in etse4, thereby severely impairing its translational efficiency in EHEC. Z0268 was predicted to be a rearrangement hotspot (Rhs)-type T6SS effector-coding gene located on the right side of the T6SS structural protein-coding gene region (Fig 4A). The cognate immunity gene, Z0269, likely confers self-protection against Z0268 toxicity. Comparative genomics showed Z0268 originated through recombination of the 5’-terminal 3398 bp from G2583_0245 with the 3’-terminal 817 bp of G2583_0247 (S2C Fig). The total 2565 bp length of DNA between both fragments was lost during evolution (S2C Fig). For Z0269, six copies of the repeat nucleotide sequence (GGTGAA) were inserted into the coding region compared to aEPEC G2583_0248, leading to the insertion of six copies of amino acids (Gly-Glu) between Glu₄₉ and Pro₅₀ to generate Z0269 of EHEC (S2C Fig). Our results revealed that the EHEC T6SS gene cluster has various mutations and might have its degenerative functions compared to aEPEC.

EHEC T6SS was also composed of two vgr orphan loci: vgrG-2 (Z0701-Z0707) and vgrE (Z2253-Z2262), evolutionarily derived from ancestral aEPEC vgrG’-2 (G2583_0724-G2583_0730) and vgrE’ (G2583_1818-G2583_1822) loci, respectively (Fig 4A). While vgrG-2 locus was similar to vgrG’-2 locus, the ancestral effector gene G2583_1819 (vgrE’ locus) underwent fragmentation into three genes (Z2257, Z2259, and Z2261) during EHEC evolution (S2D Fig). Similarly, G2583_1822 partitioned into two coding sequences (Z2253 and Z2254) and an intergenic spacer (S2D Fig).

Although a previous study demonstrated a functional EHEC T6SS capable of secreting the effector Hcp2 [27], our study revealed that the system had degenerated and appeared to have lost the capacity to secrete another effector, Hcp1, under laboratory conditions (Fig 4G). This conclusion is supported by the observation that the ancestral aEPEC strain, which possesses a more intact T6SS, secretes low but detectable levels of Hcp1 under the same conditions (S3 Fig).

Translational suppression of ehtssM via evolutionary mutation in EHEC

Comparative genomic analysis indicated that the EHEC-derived tssM (ehtssM), a key T6SS structural protein-coding gene, was mutated during evolution. The ancestral aEPEC-derived tssM (eptssM) underwent significant modification through a 28-bp tandem repeat insertion at its 5’ terminus, shifting the start codon from ATG to TTG and truncating the N-terminal 10 amino acids (Fig 5A and 5B). We hypothesized that the TTG start codon would exhibit substantially lower translational efficiency than ATG, potentially attenuating T6SS functionality by impairing tssM expression. To test this hypothesis, we engineered C-terminal His-tag fusions to the N-terminal 10-amino-acid peptides of both ehTssM and epTssM for comparative expression analysis using western blotting (Fig 5C). Immunoblot quantification demonstrated a 7.7-fold greater accumulation of epTssM compared to that of ehTssM (Fig 5D), indicating that EHEC attenuates T6SS activity through translational suppression of tssM. These findings confirm that the 28-bp tandem repeat insertion in the 5’ terminus of eptssM disrupts translational initiation efficiency (Fig 5E).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Analysis of the tssM gene and evaluation of its translational activity in EHEC and ancestral aEPEC.

(A) Comparison of the DNA sequences at 5’ region of tssM between EHEC and aEPEC. (B) Comparison of the amino acid sequences in the N-terminus region of the ehTssM and epTssM proteins. (C) Schematic diagram of translation fusion constructs for detecting tssM expression. A His-tag was fused to the C-terminus of the first 10 amino acids of ehTssM and epTssM, respectively. P_T5: T5 promoter. (D) Detection of TssM expression via Western blot analysis. Control, empty vector pQE80L; EDL933, the expression level of ehtssM; CB9615, the expression level of eptssM. EF-Tu (elongation factor thermo unstable) served as a reference. (E) Schematic illustration of the proposed mechanism for reduced tssM expression in EHEC due to gene mutation.

https://doi.org/10.1371/journal.ppat.1014039.g005

tssM evolutionary mutation attenuated T6SS activity in EHEC

TssM serves as a core structural component required for T6SS functionality. We hypothesized that tssM evolutionary mutation would impair T6SS-mediated effector secretion in EHEC. To validate this, we generated a restored TssM strain (RSTM) by excising the 28‑bp tandem repeat insertion in tssM of EHEC, thereby restoring the ancestral eptssM genotype derived from aEPEC. T6SS effector Hcp2 was inducibly overexpressed in EHEC using plasmid pQE-hcp2. Western blot analysis revealed basal Hcp2 secretion in EHEC WT supernatants, whereas no detectable secretion occurred in ΔtssM or ΔT6SS mutants (Fig 6A). Strikingly, RSTM exhibited 8.6-fold enhanced Hcp2 secretion relative to the WT (Fig 6A), confirming that the 28-bp repeat insertion in tssM evolutionarily suppressed T6SS secretory capacity in EHEC. Quantitative analysis revealed that upon host cell infection, the RSTM strain exhibited a 12.9-fold reduction in excised Φstx2 genomes (7.76% of WT levels) and a 5.9-fold decrease in mature Stx2A production (17.01% of WT levels) (Fig 6B and 6C). Furthermore, under H₂O₂ stress in vitro, the high-functioning T6SS RSTM strain displayed a stronger growth (OD₆₀₀ = 2.242 ± 0.164) and a substantial reduction in Φstx2 genome release (only 2.34% of the WT level) (Fig 6D and 6E). This demonstrates that an intact T6SS enhances oxidative stress resistance and attenuates Φstx2 prophage induction. Collectively, these results indicate that the 28‑bp repeat insertion in tssM evolved to render EHEC more susceptible to ROS, thereby predisposing it to premature prophage activation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Characterization of the RSTM strain.

(A) Detection of Hcp2 secretion in EHEC wild type and derived strains by Western blotting. (B) Quantification of Stx2 phage particles in the co-culture supernatant of RSTM and Caco-2 cells by qPCR. (C) Immunoblot analysis of the secreted Stx2A subunit from RSTM using an anti-Stx2A monoclonal antibody. (D) Bacterial susceptibility to exogenous oxidative stress. (E) Oxidative stress-induced Stx2 phage liberation. (F) Expression levels of recA and key T3SS genes in RSTM compared to the wild type and ΔtssM mutant. (G) Cytotoxicity in Caco-2 cells co-cultured with EHEC wild type or RSTM, analyzed by flow cytometry. (H) Statistical comparison of cytotoxic effects between EHEC wild type and RSTM strains. (I) Attenuation of T6SS-deficient cytotoxicity by RSTM co-infection, assessed via flow cytometry. (J) Statistical analysis of the cytotoxic effects shown in (I). Data in (B, D-F, H, J) are presented as mean ± SD from three independent experiments. P values were determined by one-way ANOVA with Dunnett’s or Tukey’s multiple comparisons test. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant.

https://doi.org/10.1371/journal.ppat.1014039.g006

Our findings establish that T6SS dysfunction drives Φstx2 lytic induction and enhances cytotoxicity mediated by the T3SS. Therefore, we examined the impact of the restored tssM gene on prophage induction dynamics and T3SS expression. Correspondingly, qPCR showed significant downregulation of the SOS response activator recA in RSTM (15.7% of the WT level) (Fig 6F). These results collectively indicate that RecA-dependent Φstx2 activation and subsequent Stx2 liberation were strongly suppressed in the RSTM strain during host interaction. Furthermore, the expression of key T3SS genes (ler, espA, and tir) was significantly reduced in RSTM to 25.2%, 10.6%, and 34.7% of the WT levels, respectively, confirming the coordinated inhibition of the T3SS during infection (Fig 6F). Together, these data established that the evolutionary tssM mutation significantly upregulates T3SS expression by promoting Φstx2 activation in EHEC.

Next, we assessed the virulence of the RSTM strain by measuring Caco-2 cell mortality. Notably, infection with the EHEC WT strain resulted in 89.0% cytotoxicity, which was significantly higher than the 14.6% cytotoxicity observed with the RSTM strain. This corresponded to a 6.1-fold reduction in the virulence of the T6SS-restored mutant (Fig 6G and 6H). Based on our hypothesis that a functional T6SS can counteract the cytotoxic effects associated with its loss, we co-infected intestinal epithelial cells with a 0.5:1 mixture of RSTM and ΔT6SS strains. This mixture significantly reduced cytotoxicity compared to infection with ΔT6SS mutant alone, even at half the dose (Fig 6I and 6J). In contrast, a control co-infection with an equivalent amount of the ΔT6SS mutant (50% dose) increased cytotoxicity relative to the ΔT6SS-alone infection (Fig 6I and 6J). This result provides direct evidence that functional T6SS can attenuate hypercytotoxicity caused by T6SS deficiency. Collectively, our work identified a 28-bp tandem repeat insertion in tssM as an evolutionary adaptation that drives T6SS degeneration and pathoadaptive hypercytotoxicity in EHEC.

EHEC tssM mutation is widely distributed in the Φstx2-associated E. coli

A previous study reported the evolutionary path from aEPEC O55:H7 to EHEC O157:H7 [7]. This transitional series was represented by six key evolutionary intermediates: E. coli strains CB9615 (GCA_000025165.1), DEC5A (CP038394.1), USDA 5905 (GCA_002775135.1), 493/89 (CP038412.1), CDC G5101 (GCA_000187285.4), and ATCC 35150 (CP038405.1). Phylogenetically constrained analysis of T6SS loci identified a 28-bp tandem repeat insertion in tssM emerging between the USDA 5905 and 493/89 evolutionary nodes (Fig 7A). This suggests that tssM may have been mutated after Φstx2 integration (between strains DEC5A and USDA 5905) in the bacterial genome during evolution (Fig 7A). We subsequently conducted a pan-genomic survey of tssM architecture across 3,780 E. coli genomes using NCBI RefSeq. Among these, 492 genomes (13.02%) harbored Φstx2 prophages (E. coli_Φstx2) (Fig 7B, S2 Dataset). In the E. coli_Φstx2 group, 358 strains were categorized as tssM mutations, accounting for 72.76% of total E. coli_Φstx2 (Fig 7C, S3 Dataset). In the tssM-mutated strains, 294 E. coli_Φstx2 strains were mutated in the same manner as ehtssM, evolving the 28-bp tandem repeat insertion (S4 Dataset). The proportion of the ehtssM sequence in E. coli_Φstx2 strains with potential tssM mutations was 82.12% (Fig 7D).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Evolutionary pathway of EHEC and the key mutation in tssM.

(A) The evolutionary pathway from aEPEC O55:H7 to EHEC O157:H7, with the predicted step of tssM mutation. Blue boxes represent O55 strains; red boxes represent O157 strains. EHEC O157:H7 evolved from the less virulent strain aEPEC O55:H7 through several genetic events. Critical tssM mutation occurred between the divergence points of reference strains USDA 5905 and 493/89, positioned between integration events of Stx2- and Stx1-converting phages (Shiga toxin-encoding bacteriophages). (B) Proportion of E. coli_Φstx2 among all E. coli strains. (C) Proportion of tssM mutations in E. coli_Φstx2 strains. (D) Proportion of the 28-bp tandem repeat nucleotide insertion variant among all tssM mutation variants.

https://doi.org/10.1371/journal.ppat.1014039.g007

Phylogenomic analysis revealed the E. coli_Φstx2 group tended to be segregated into two distinct clusters: one comprising tssM-mutated strains and the other containing ancestral eptssM-harboring strains (Fig 8A). Notably, strains carrying the ehtssM form a group that was phylogenetically distinct from the evolutionary clade of the other tssM-mutated variants (Fig 8B). To control for co-evolving traits, non-O157 STEC strains were integrated into a building of phylogenetic tree. The inclusion subtly changed the evolutionary relationships, suggesting that the pleiotropic effects were negligible (S4A and S4B Fig). This phylogenetic segregation indicates that TssM functional decay represents an evolutionarily distinct trajectory that has emerged not only in EHEC but also widely distributed in the Φstx2-associated E. coli.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. Phylogenetic analysis of tssM mutation diversity in E. coli_Φstx2.

Phylogenetic trees were constructed using genomic sequences of E. coli_Φstx2 strains, annotated to highlight both tssM-mutated variants and strains carrying the 28-bp tandem repeat insertion mutation in tssM. (A) Distribution of tssM-mutated strains in the phylogenetic tree of the E. coli_Φstx2 cluster. (B) Distribution of strains carrying the 28-bp tandem repeat insertion among all tssM-mutated E. coli_Φstx2 strains. eptssM denotes the intact ancestral tssM gene from aEPEC.

https://doi.org/10.1371/journal.ppat.1014039.g008

T6SS Inactivation enhances EHEC cytotoxicity

Generally, T6SS, which is widely present in Gram-negative bacteria, is important for bacterial competition and is considered a bacterial virulence factor [22,26,39–41]. Contrary to most of the previously established functions, our study revealed that T6SS inactivation markedly accelerated intestinal epithelial cell death in EHEC. Further investigation demonstrated that T6SS deficiency significantly enhances multiple EHEC virulence traits: host cell adherence, Φstx2 prophage liberation, Shiga toxin 2 secretion, and T3SS expression. Genetic analysis identified the Φstx2 prophage as an essential component and the T3SS as the primary driver within the cytotoxic network of T6SS-deficient EHEC.

In exploring the host-derived signals that activate this hyper-virulent state, we found that the antioxidant vitamin C significantly attenuated the cytotoxicity of both the WT and ΔT6SS strains to a comparable extent. Therefore, we hypothesized that ROS are the key mediators. Consistent with this, in vitro assays showed that H₂O₂ dramatically inhibited the growth of T6SS-deficient mutants while enhancing Φstx2 liberation and T3SS expression—a phenotype that was reversed by restoring T6SS function. Collectively, these results indicate that host ROS are critical signals that trigger the hypervirulent state in EHEC. Previous studies have established that T6SS contributes to stress adaptation; in EHEC, it secretes the catalase KatN, which is vital for oxidative stress defense and survival within macrophages by neutralizing host-derived ROS [27,42]. Therefore, we propose that T6SS loss compromises this defense, leading to increased intracellular ROS levels. Oxidative stress may trigger a conserved bacterial SOS response via RecA, ultimately resulting in prophage induction [37]. In EHEC, lysogenic Stx2-converting prophages repress T3SS activity [19]. Consequently, SOS-induced prophage activation derepresses T3SS expression. This cascade, initiated by T6SS loss and impaired ROS detoxification, culminates in SOS-mediated prophage induction and subsequent T3SS derepression, providing a coherent mechanism for the accelerated epithelial cell death observed during infection with T6SS-deficient EHEC.

Current animal models for EHEC research rely primarily on murine systems [43]. Although previous murine studies have suggested that T6SS deficiency reduces pathogenicity [27], this model is limited by the inherent resistance of mice to EHEC intestinal colonization and their divergent receptor profiles (e.g., Gb3), which prevent the development of human-like severe diseases such as HUS [43–45]. To address this, we evaluated ΔT6SS cytotoxicity using MC-38 murine colon adenocarcinoma cells. Intriguingly, T6SS inactivation accelerated MC-38 cell death compared to that in the WT strain, indicating that the effect of EHEC cytotoxicity involving the T6SS is not entirely consistent with pathogenicity in the whole-animal model, and the underlying mechanism is likely more complex.

Evolutionary degeneration of EHEC T6SS enhances cytotoxic effects

The findings above prompt a critical question: What evolutionary advantage might the ΔT6SS-associated cytotoxicity confer to EHEC? EHEC harbors E. coli T6SS-2, a system inherited from the ancestral enteropathogenic E. coli O55:H7 lineage [7,32]. However, this apparatus is likely to have undergone progressive degeneration through evolutionary mutations that compromised the integrity of T6SS coding genes and suppressed their functional expression. Although the T6SS in most bacteria typically mediates interbacterial competition via antibacterial effector delivery [21,22,24,26], EHEC T6SS-2 appears to retain only vestigial functionality. Etse4 is an EHEC antibacterial effector that specifically targets the cell wall in the periplasmic space to suppress the growth of rival bacteria. Etsi4, the cognate immune protein of Etse4, counteracts Etse4 toxicity and protects EHEC from self-intoxication. Etsi4 is a DUF1311 domain protein that shares structural homology with Mycobacterium tuberculosis lysozyme inhibitor, LprI, which has been implicated in toxin antagonism and T6SS immunity [46,47]. Additionally, its ability to bind and inhibit macrophage lysozymes may facilitate bacterial survival [47]. Paradoxically, despite bile salt induction being a condition previously shown to activate T6SS in related species [27,48], no detectable antibacterial activity was observed in interbacterial competition assays. Our study revealed that a single-nucleotide mutation fragmented the ancestral etse4_ori into two ORFs, effectively abolishing the antibacterial capability of EHEC. It is noteworthy that the mutation was confined to the effector coding gene etse4 while preserving the integrity of its cognate immunity protein-coding gene etsi4. This suggests that although EHEC has lost its antibacterial activity, it can still counteract the toxicity of similar effectors from competing bacteria. Thus, it might have gained a defensive advantage against bacterial competition. Furthermore, Hcp1, positioned directly upstream of the etse4 locus and hypothesized to facilitate Etes4 secretion, was not detected in the extracellular fractions. To date, no experimental evidence has confirmed the secretion of Hcp1 itself by EHEC. This may be another reason why EHEC failed to exhibit antibacterial activity during interbacterial competition. Other genes encoding EHEC Rhs-type T6SS effectors (Z0268, Z2257-Z2261 and Z2253-Z2254) were also found to be truncated or disrupted by mutations. These genetic alterations are likely to diminish T6SS activity or its associated functions. A similar disruption of a T6SS effector was found in Acinetobacter baylyi, where the tse1 gene was split into non-functional fragments by an IS1236 insertion. Removal of this insertion restores tse1 functionality, enabling the strain to significantly reduce the recovery of E. coli during bacterial competition [49].

Unlike bacterial pathogens with strong T6SS activity (e.g., V. cholerae, A. baylyi, Pseudomonas aeruginosa) or its ancestor aEPEC O55:H7, which possesses a more intact T6SS, EHEC colonizes the large intestine, a niche with an exceptionally high microbial density [26,49–51]. In such an environment, the deployment of a potent antibacterial T6SS to attack numerous competitors may not be an optimal survival strategy. Conversely, many adaptive traits, such as antibiotic resistance, are often plasmid-encoded [52]. Both T6SS activity and plasmid conjugation require cell-cell contact; however, successful conjugation necessitates recipient survival, a paradox given that the recipient is also a potential T6SS target [52,53]. Therefore, in addition to providing defense by inactivating the antibacterial effectors while preserving immunity, this mechanism in EHEC may serve a dual purpose. It can also facilitate the acquisition of exogenous plasmids, thereby promoting horizontal gene transfer and trait acquisition without eliminating potential conjugation partners.

During bacterial evolution, two genes, tssM and clpV, which encode the T6SS core structural proteins, were mutated by the insertion or deletion of repeated nucleotide sequences. These mutations arise from the insertions or deletions of repetitive nucleotide sequences, resulting in mature TssM and ClpV proteins lacking several amino acid residues. Both proteins are endowed with ATPase activity, which serves as a molecular engine for T6SS assembly (TssM) and disassembly (ClpV), and their structural integrity is essential for T6SS functionality [31,54]. Notably, tssM exhibited more pronounced mutational damage compared to clpV. A key event in tssM degeneration was the insertion of a 28-bp repeated sequence, which disrupted both its 5’ untranslated region (5’ UTR) and coding sequence. This mutation likely impairs transcriptional and translational regulation and compromises the structural integrity of TssM. Consistent with this, ehtssM displayed markedly reduced translational efficiency relative to its ancestral counterpart eptssM, leading to a severe decline in T6SS-mediated secretion activity. Restoration of the intact tssM gene in EHEC rescued T6SS function. Interestingly, the RSTM strain exhibited significantly lower cytotoxicity toward intestinal epithelial cells than the WT strain. This attenuation occurred because the restored T6SS efficiently neutralized host-derived ROS during co-culture, which in turn reduced the activation of the Φstx2 prophage and the expression of the T3SS. These findings indicate that the tssM mutation is a crucial factor driving T6SS degeneration in EHEC.

The expression of all genes required for T6SS assembly and firing imposes a significant energetic burden on bacterial cells. Various strategies have been developed to mitigate these costs [33]. These include intrinsic mechanisms, such as ClpV-mediated disassembly and recycling of sheath components for reuse, as well as behavioral adaptations [33]. For example, P. aeruginosa can sense the direction of incoming attacks. This enables a targeted T6SS counterstrike, thereby avoiding the wastefulness of random firing [50]. Despite such energy-saving adaptations, ecological contexts may arise in which maintaining a functional T6SS is prohibitively costly, favoring its loss. Compared with Shigella sonnei, S. flexneri has undergone extensive T6SS degeneration, retaining only four core genes and completely losing its T6SS activity [22]. Supporting this, a recent study used experimental evolution to show that enteroaggregative E. coli (EAEC) T6SS activity was nearly abolished when co-cultured with immune recipients, but was unchanged in susceptible cells [55]. This demonstrates that the T6SS can be significantly attenuated or entirely lost. Furthermore, loss-of-function T6SS mutant ΔtssM may enhance certain virulence traits. Although ΔtssM shows reduced epiphytic survival and transmission, it can display higher aggression, cause more severe symptoms, and achieve larger population sizes than the wild-type strain [56,57]. S. flexneri, which arises from different ancestral E. coli isolates [58], also has a complete loss of the T6SS, which may allow energy conservation to support its high-risk virulence strategy.

In contrast, EHEC appears to have adopted a distinct, intermediate evolutionary strategy. Instead of a complete loss, it retained a degenerated, low-activity T6SS. This compromise suggests that EHEC may derive a selective benefit from the residual T6SS function, potentially in stress response or ion acquisition, while avoiding the high energy cost of a fully active system. The conserved energy could be redirected to critical processes, such as adherence, expression of other virulence factors, and precise regulatory control of prophage induction and Shiga toxin production, which further promotes T3SS expression in a cascade. Thus, degenerated T6SS in EHECs may represent a finely tuned adaptation, balancing limited functionality against metabolic costs to optimize fitness within the complex intestinal niche. Our study elucidates the mechanistic basis of tssM-driven T6SS decay in EHEC and its implications for bacterial adaptation during evolution.

T6SS degeneration may be promoted by Φstx2 integration

A previous study revealed that the evolutionary trajectory of aEPEC O55:H7 to EHEC O157:H7 involves multistep genomic remodeling [7]. The time point of tssM mutation in EHEC likely occurred subsequent to Φstx2 integration during pathogen evolution. Our observation suggests a potential association between Φstx2 acquisition and subsequent T6SS genetic alterations in evolving pathogenic E. coli lineages. Bioinformatic analysis of E. coli_Φstx2 strains revealed nearly three-quarters of the isolates had mutated tssM, and the proportion of 28-bp tandem repeat insertion was more than 80%. This finding indicated that tssM mutation is the principal driver for T6SS functional degeneration, and the mutation through 28-bp insertional mutagenesis is the dominant mutation type in E. coli_Φstx2 tssM mutants. The observed co-occurrence of Φstx2 integration and T6SS degeneration suggests that this mobile genetic element may drive evolutionary adaptation through genome reorganization. While T6SS itself functions as a virulence determinant, the integration of novel high-risk virulence factors such as Φstx2 may trigger the pathogen genomic adjustments (e.g., T6SS mutation). This genomic rebalancing may enable pathogens to optimize virulence expression, driving the emergence of highly virulent and pathogenic strains through evolution.

Based on these findings, we proposed a mechanistic model to elucidate how the degeneration of the T6SS enhances EHEC cytotoxicity during host infection (Fig 9). This model posits that upon infecting intestinal epithelial cells, EHEC triggers a host immune response that includes the production of ROS to combat the infection. In response to this immune pressure, EHEC activates the T6SS to secrete effectors that mitigate the bactericidal effects of host-derived ROS. Concurrently, EHEC has acquired, through evolution, a critical virulence factor: the Φstx2 prophage, which potentiates virulence through the production and release of Shiga toxin. Host immune factors such as ROS can induce a RecA-dependent SOS response in EHEC. This, in turn, activates the Φstx2 prophage into its lytic cycle, leading to phage excision, replication, and ultimately the liberation of Shiga toxin. Once released, Shiga toxin can enter the systemic circulation via the bloodstream, causing distant tissue damage, most notably to the kidneys, resulting in conditions such as HUS [59]. Importantly, Φstx2 lysis further activates the T3SS, which plays a major role in damaging intestinal epithelial cells.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 9. Proposed model for enhanced cytotoxicity in intestinal epithelial cells infected with T6SS-deficient EHEC.

Upon colonizing the intestinal epithelium, host cells initiate an immune response to counteract EHEC infection. In response to host-derived ROS, EHEC activates its T6SS to secrete effectors that protect the bacterium from immune clearance. During evolution, EHEC acquired the high-risk virulence determinant Φstx2, which enhances pathogenicity through Shiga toxin production. A degenerate T6SS may compromise bacterial resistance to host immunity, increase susceptibility to Φstx2 induction, and promote more efficient Shiga toxin expression. Furthermore, excision of Φstx2 triggers a cascade that activates other key EHEC virulence factors, such as the T3SS, ultimately amplifying host cell damage. Created in BioRender. Sun, Z. (2026) https://BioRender.com/a494isq.

https://doi.org/10.1371/journal.ppat.1014039.g009

Our results indicate that a degenerate T6SS in EHEC reduces bacterial resistance to host-derived ROS, increases sensitivity to Φstx2 induction, and improves the efficiency of prophage excision from the bacterial genome—thereby facilitating Shiga toxin expression and release. Moreover, Φstx2 excision activates the key EHEC virulence factor T3SS, exacerbating damage to the intestinal epithelial cells. This intricate interplay between the degenerate T6SS, prophage induction, and T3SS activation may explain why EHEC exhibits higher virulence than most other pathogenic E. coli strains. To our knowledge, this study provides the first description of how T6SS degeneration arises and enhances the pathogenic potential of EHEC, while clarifying its biological significance in host-pathogen interplay. Future studies should focus on determining the evolutionary or adaptive significance of T6SS degeneration in EHEC and related pathogenic lysogens, and on dissecting the mechanisms by which this degeneration modulates virulence outcomes within the tripartite interaction of the bacterium, prophage, and host cell.

Bacterial strains, cell lines, and growth conditions

The bacterial strains and cell lines used in this study are listed in S1 Table, provided as Supplemental Material. Enterohemorrhagic E. coli O157:H7 strain EDL933 was used as the EHEC wild type strain. The E. coli O157:H7 precursor, atypical enteropathogenic E. coli O55:H7 strain CB9615, was kindly provided by Prof. Bin Liu (Nankai University) [30]. E. coli DH5α was used for cloning and for the tssM translation assay. The Caco-2 and MC-38 cell lines were obtained from the Cell Resource Center of Shanghai Academy of Sciences, Chinese Academy of Sciences. Cells were cultured in DMEM (Gibco, #11965118) supplemented with 10% FBS (Gibco, #10100147) at 37°C under 5% CO₂. Bacterial strains were routinely grown in LB broth (pH 7.4) at 37°C with shaking at 250 rpm, unless otherwise indicated. Antibiotics were used at the following concentrations: ampicillin, 100 μg/mL; chloramphenicol, 30 μg/mL; kanamycin, 50 μg/mL; tetracycline, 10 μg/mL; streptomycin, 20 μg/mL. All culture-related chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA).

Construction of plasmids and mutant strains

All oligonucleotide primers are listed in S2 Table and were synthesized by Sangon Biotech Co., Ltd. (Shanghai, China). The EHEC tssM-restored strain RSTM was constructed using the CRISPR-Cas12a-assisted recombineering system [60]. The Stx2-converting prophage deletion mutant (ΔΦstx2) was generated and validated by Abiocenter Biotech Co., Ltd. (Wuxi, China). The EHEC T6SS main cluster deletion mutant (ΔT6SS, spanning Z0244–Z0275) was generated in the EDL933 background via the λ-Red recombinase system [61]. Using the same method, we also constructed the clpV deletion mutant (ΔclpV), the tssM in-frame deletion mutant (ΔtssM), the T3SS-inactivating escN deletion mutant (ΔescN), and other mutants in both EHEC EDL933 and aEPEC CB9615 strains. For complementation of clpV, a 500-bp upstream regulatory region together with the clpV coding sequence was cloned into pACYC184 using NcoI/ScaI restriction sites, yielding plasmid pclpV. To overexpress T6SS effectors Hcp1 (Z0248) and Hcp2 (Z0266), the corresponding genes were PCR-amplified with primers 80-hcp1-F/R and 80-hcp2-F/R, then ligated into pQE80-YX1 via AfeI/FseI sites, yielding plasmids pQE-hcp1 and pQE-hcp2. All plasmids were constructed using either the ClonExpress II One Step Cloning Kit (Vazyme, #C112; Nanjing, China) or by restriction-ligation with enzymes from New England Biolabs (Ipswich, MA, USA), following standard protocols.

In silico analysis of the EHEC T6SS

Sequence alignment of the T6SS gene clusters from EHEC strain EDL933 and aEPEC strain CB9615 was performed using BioEdit and Clustal Omega [62,63]. Putative T6SS proteins were analyzed using CD-Search [64], HHpred [65], SecReT6 [66], Bastion6 [67], and SignalP 4.1 [68]. All mutations generated for T6SS functional assays were confirmed by Sanger sequencing.

Bacterial infection and cytotoxicity assay

Caco-2 cells were seeded in 12-well plates at a density of 2 × 10⁵ cells per well and incubated for 48 h. Monolayers were gently rinsed twice with phosphate-buffered saline (PBS, pH 7.4). The EHEC and its derivative strains were grown overnight in LB broth at 37°C, harvested, and added to the cells that had been replenished with fresh DMEM at a concentration of 1 × 10⁷ bacteria per well, followed by incubation at 37°C under 5% CO₂ for the indicated time. After infection, cells were rinsed with PBS and stained using the LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Thermo Fisher Scientific, #L34957, USA) according to the manufacturer’s instructions. Following three washes with PBS, cells were detached with 0.25% trypsin-EDTA (Gibco, #25300–054, USA) and gently resuspended. Cell suspensions were fixed in 4% paraformaldehyde (PFA) for 15 min at 25°C before flow cytometric analysis. Data were acquired on a BD LSRFortessa X-20 flow cytometer and analyzed using FlowJo software (v10.4, BD Biosciences).

Measurement of bacterial growth and adherence in host cell co-culture

Bacterial growth and adherence to host cells were assessed using a co-culture model of EHEC strains with Caco-2 cells for specified durations. For growth quantification, co-cultures were treated with 0.1% (v/v) Triton X-100 to lyse the host cells. The lysate was then centrifuged (4,500 × g for 10 min at 4°C), washed twice with PBS (pH 7.4), and the bacterial pellet was resuspended for quantification by measuring the OD₆₀₀ or by serial dilution plating on LB agar. For adherence quantification, the supernatant from co-cultures was first removed. The remaining Caco-2 cells were gently washed twice with PBS to remove non-adherent bacteria. Adherent bacteria were then released by lysing the host cells with 0.1% Triton X-100 in PBS and quantified as described for the growth assay.

Quantitative PCR (qPCR) analysis of gene expression

EHEC strains were co-cultured with Caco-2 cells for 11h. Bacteria were then harvested by lysing host cells with 0.1% Triton X-100, followed by centrifugation (5,000 × g, 10 min, 4°C). Total RNA was extracted using the Bacterial RNA Extraction Kit (Vazyme, #R403-01). Subsequent reverse transcription to generate cDNA was performed using the HiScript II 1st Strand cDNA Synthesis Kit (Vazyme; #R212-01), in accordance with the manufacturer’s protocols. The cDNA was diluted 1:10 in nuclease-free water and used as the template for quantitative real-time PCR (qRT-PCR). qPCR was performed in a LightCycler 96 system (Roche) using ChamQ Universal SYBR qPCR Master Mix (Vazyme, Q311-02). Primers used are listed in S2 Table. The relative expression of target genes was calculated using the comparative 2 ^(–ΔΔCt) method [69], with 16S rRNA serving as the internal reference. All experiments included three independent biological replicates, each with three technical replicates.

Quantification of excised phage genomes

Excised phage genomes in the supernatants of EHEC and Caco-2 co-cultures were quantified by qPCR as previously described with minor modifications [44]. Briefly, the cultured supernatant was centrifuged (5,000 × g, 10 min, 4 °C) to remove cells and debris. The supernatant was then filtered through a 0.22 μm membrane and treated with DNase I (Vazyme, #EN401) to digest residual extracellular DNA. The processed phage sample was serially diluted (1:10) in nuclease-free water and used as the qPCR template. During the initial high-temperature step of the PCR, phage capsids were denatured, releasing encapsidated DNA for amplification [70]. qPCR was performed on a LightCycler 96 System (Roche) using ChamQ Universal SYBR qPCR Master Mix (Vazyme, #Q711) and stx2‑specific primers (qPCR‑stx2‑F/R; see S2 Table). Quantification was performed as described by Balasubramanian et al. [44].

Assessment of EHEC susceptibility and Φstx2 induction under H₂O₂-induced oxidative stress

EHEC susceptibility to oxidative stress and the consequent induction of the Φstx2 prophage were assessed as follows. Bacterial cultures (OD₆₀₀ = 0.4) were treated with 1.5 mM H₂O₂ and incubated for 16 h at 37°C with shaking. Bacterial growth was monitored by measuring the optical density at 600 nm (OD₆₀₀). To quantify prophage induction, culture supernatants were collected by centrifugation, filtered through a 0.22 μm membrane, and treated with DNase I to remove unprotected DNA. The protected (encapsidated) phage DNA was then extracted, purified as described above, and used as the template in qPCR with stx2-specific primers to quantify excised Φstx2 genomes.

Assessing growth inhibition by heterologous expression of the T6SS toxin-effector Etse4 and its immunity protein Etsi4

To evaluate the antibacterial activity of the T6SS toxin-effector Etse4 (Z0246), the etse4 gene was subcloned into pET22b and pET28a vectors via NcoI/XhoI sites, generating plasmids pET22b-etse4 and pET28a-etse4. To test the function of its predicted cognate immunity protein Etsi4 (Z0245), the etse4 and etsi4 genes were co-cloned into pET22b, yielding plasmid pET22b-etse-i4. E. coli BL21(λDE3) cells transformed with these constructs were cultured overnight, normalized to an OD₆₀₀ of 1.0, serially diluted (10-fold), and spotted (2 μL per dot) onto LB agar plates containing 0.5 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG) and appropriate antibiotics. Plates were incubated at 37°C for 16 h, after which growth inhibition was quantified.

In vitro competition assay for T6SS-mediated antibacterial activity

Bacterial competition assays were performed as described by Sana et al. with a slight modification [48]. The EHEC EDL933 (predator) and the streptomycin-resistant E. coli K12 str. MG1655 (prey) were grown separately in LB broth, mixed at a ratio of 1:5 (prey:predator). Ten microliter aliquots were spotted onto LB agar supplemented with 50 mg/mL bile salts (Sigma-Aldrich, #48305) and incubated at 37 °C for 48 h. Cells from the competition spots were harvested, serially diluted, and plated on both basic LB agar and LB agar supplemented with streptomycin. Prey (MG1655) CFUs were enumerated on streptomycin plates, while predator (EDL933) CFUs were calculated by subtracting the counts on streptomycin plates from the total counts on basic LB plates. The competitive index was determined as the ratio of prey cells to predator cells.

Hcp secretion assay

Secretion of Hcp was analyzed by Western blot as described [4,27]. Briefly, protein samples were separated on 12% SDS-PAGE gels and transferred to PVDF membranes. Membranes were blocked and then incubated overnight at 4°C with a mouse anti-His tag monoclonal antibody (BBI, #D191001; Shanghai, China; 1:3,000). After washing, membranes were incubated for 1.5 h at room temperature with an HRP-conjugated goat anti-mouse IgG secondary antibody (BBI, #D110087; 1:5,000). Signals were detected using an enhanced chemiluminescence (ECL) substrate, and band intensities were quantified with ImageJ software (National Institutes of Health, NIH). Data were normalized to total protein loading controls.

Quantification of secreted Stx2

Secreted Stx2 in the supernatants of EHEC and Caco-2 co-cultures was quantified by Western blot using an antibody specific for the Stx2A subunit. The culture supernatant was centrifuged (4,000 × g, 5 min) to remove cells and debris. Secreted proteins were then precipitated and concentrated as previously described for hemolysin detection [4]. Protein samples were separated by 10% SDS-PAGE and transferred to a membrane for immunoblotting. The membrane was incubated overnight at 4°C with a mouse monoclonal anti-Stx2A antibody (Invitrogen, MA5–33371; 1:2,000), followed by a 1-hour incubation at room temperature with an HRP-conjugated goat anti-mouse IgG secondary antibody (Proteintech, SA00001–1; 1:5,000). Signals were detected by enhanced chemiluminescence (ECL), and band intensities were quantified using ImageJ software (NIH).

tssM translation efficiency assay

To assess tssM translation efficiency, DNA fragments encompassing the first 30 bp of the tssM coding sequence along with 200 bp of its upstream regulatory region were amplified from EHEC O157:H7 and aEPEC O55:H7. These fragments were cloned into the AfeI/FseI sites of the pQE80YX1 vector, generating plasmids pQE80-ehtssM and pQE80-eptssM, respectively. These constructs were designed to express N-terminal TssM fragments (TssM₁₀) containing C-terminal hexahistidine (His₆) tag for immunodetection. The plasmids were transformed into EHEC EDL933, and a positive clone for each was selected. Overnight cultures of the transformants were diluted 1:100 into fresh LB medium and grown at 37°C with shaking (250 rpm) to an OD₆₀₀ of 0.5. Protein expression was then induced with 0.5 mM IPTG for 4 h under the same conditions. Cells were harvested by centrifugation (4°C), and bacterial pellets were lysed by ultrasonication (150 W, 5 min, on ice). After centrifugation (10,000 × g, 30 min, 4°C), the supernatants were analyzed by 15% SDS-PAGE followed by Western blotting. An anti-RpoA antibody served as the loading control for normalization. Band intensities were quantified using ImageJ software, and the TssM₁₀-His₆ signal was normalized to the RpoA signal to determine relative expression levels. The assay was performed based on a published method with modifications [71].

Determining the proportion of TssM evolutionary mutants and ehTssM in the Φstx2-associated E. coli

All complete genomes of E. coli were downloaded from the Genome Resources of the National Center for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov/home/genomes/). Plasmid sequences were subsequently removed from the E. coli genomes. To identify E. coli strains bearing Φstx2 or its homologs (E. coli_Φstx2), the Φstx2 sequence was mapped to all E. coli genomes using the BLAST algorithm (version 2.9.0) with an e-value <1 × 10^-10 and query coverage per subject (qcovs) >30%. Genes of the filtered E. coli strains were identified using Prodigal software (version 2.6.3) [72], and translated into peptide sequences via the transeq command from the EMBOSS package (version 6.5.7.0) (https://emboss.sourceforge.net/). The epTssM sequence was mapped to the translated peptide sequences. E. coli strains with 100% identity in the first 10 amino acids of the epTssM sequence, qcovs >30%, and percentage of identical matches (pident) >70% were kept. Among the E. coli strains, those with qcovs = 100% and fewer than 5% mismatched bases in the epTssM sequence were considered to contain a functional epTssM sequence. Strains failing the above criteria, not mapping to the epTssM sequence, with qcovs ≤30% or with pident ≤70% were listed as potentially having mutated epTssM sequences. To determine the proportion of the ehTssM sequence in this list, the ehTssM sequence was mapped by the blastp algorithm. Only those with an e-value <1 × 10^-10, qcovs >30% and pident >70% were considered to have a functional ehTssM sequence.

Building a phylogenetic tree of E. coli

To disentangle the impacts of TssM mutation on the phylogenetic diversity of E. coli, the comparative genomic analysis was conducted. The genome data of 492, 358 and 294 E. coli strains were collected, and their genes were predicted using Prodigal software (version 2.6.3) [72]. The redundancy of gene sequences was reduced through CD-HIT package (version 4.8.1) [73]. The transeq command from the EMBOSS package (version 6.5.7.0) was adopted to translate the nucleotide sequences into peptide sequences. The phylogenetic orthology of E. coli strains was deduced via incorporating these peptide sequences into OrthoFinder software (version 3.0.1b1) with the following parameters “-S diamond -M msa -T fasttree” [74]. The visualization of the rooted phylogenetic tree was done using online iTOL tool [75].

Controlling for co-evolving traits of the phylogenetic tree

To control for co-evolving traits, comparative genome analysis was further performed to include non-O157 STEC strains. First, we did the selection of non-O157 STEC strains from the whole 3780 E. coli. BLAST algorithm (version 2.9.0) was used to identify those mapped to Stx2A and Stx2B at an e-value <1 × 10^-10, pident >90% and qcovs >90%, yielding 492 strains. Out of them, the strains with the Φstx2 sequence at an e-value <1 × 10^-10 and qcovs >30% were considered as STEC strains. Subsequently, the remaining 400 strains were mapped to the rbfE gene under the threshold of e-value <1 × 10^-10, which is essential for the synthesis of O157 [76,77]. A total of 104 non-O157 STEC strains were included for the following analysis. Second, the non-O157 STEC strains were incorporated into the genome data of 492, 358 and 294 E. coli strains. The new phylogenetic trees were built following the same procedure of “Building phylogenetic tree of E. coli”.