Shigella sonnei utilises colicins during inter-bacterial competition

Abstract

The mammalian colon is one of the most densely populated habitats currently recognised, with 10¹¹-10¹³ commensal bacteria per gram of colonic contents. Enteric pathogens must compete with the resident intestinal microbiota to cause infection. Among these enteric pathogens are Shigella species which cause approximately 125 million infections annually, of which over 90 % are caused by Shigella flexneri and Shigella sonnei. Shigella sonnei was previously reported to use a Type VI Secretion System (T6SS) to outcompete E. coli and S. flexneri in in vitro and in vivo experiments. S. sonnei strains have also been reported to harbour colicinogenic plasmids, which are an alternative anti-bacterial mechanism that could provide a competitive advantage against the intestinal microbiota. We sought to determine the contribution of both T6SS and colicins to the anti-bacterial killing activity of S. sonnei. We reveal that whilst the T6SS operon is present in S. sonnei, there is evidence of functional degradation of the system through SNPs, indels and IS within key components of the system. We created strains with synthetically inducible T6SS operons but were still unable to demonstrate anti-bacterial activity of the T6SS. We demonstrate that the anti-bacterial activity observed in our in vitro assays was due to colicin activity. We show that S. sonnei no longer displayed anti-bacterial activity against bacteria that were resistant to colicins, and removal of the colicin plasmid from S. sonnei abrogated anti-bacterial activity of S. sonnei. We propose that the anti-bacterial activity demonstrated by colicins may be sufficient for niche competition by S. sonnei within the gastrointestinal environment.

Keywords: Interbacterial competition; Shigella; T6SS; anti-bacterial activity; colicins; polymicrobial.

Fig. 1.

S. sonnei has potent anti-bacterial activity against E. coli but not S. typhimurium or K. pneumoniae. (a and b) Competition assays of S. sonnei and E. coli were conducted to assess the anti-bacterial activity of S. sonnei. Two representative isolates a) SS381 and b) CIP106347 and their isogenic T6SS mutants were competed against E. coli BZB1011-Km (a) or E. coli BZB1011 (b). The two S. sonnei isolates were also competed against c) S. typhimurium SL1344 d) and K. pneumoniae 43 816. The amount of E. coli, S. typhimurium and K. pneumoniae remaining after a competition with S. sonnei was enumerated to measure the anti-bacterial activity of S. sonnei. Each graph represents the mean±SEM of at least three biological replicates. A one-way ANOVA with Tukey’s correction was applied. **** = p<0.00001, ns=non-significant. LOD=Limit of detection. (e and f) Inducible Hcp1 and Hcp2 with C-terminal HA fusions were expressed from plasmids in (e) SS381 and (f) SS381ΔT6. Following Hcp induction, bacteria were centrifuged to separate the secreted fraction (supernatant) and the cellular fraction. Proteins were then separated by SDS-PAGE. The cytoplasmic protein DnaK was used as a loading control for the cellular faction. Hcp-HA fusion proteins were detected in all induced cellular factions but not the supernatant factions.

Fig. 2.

An in silico analysis of the T6SS cluster across three isolates of S. sonnei reveal mutations within the T6SS operon. (a) The T6SS operons were identified and aligned using BlastN and visualised in EasyFig. The reference genome in this synteny analysis was the T6SS operon from EHEC EDL933. Nucleotide percentage identity is denoted by the strength of the orange shading as indicated by the legend in the bottom right. EHEC EDL933 has a T6SS operon that is very similar to the published S. sonnei T6SS operon and has been shown to be functionally active. Core genes are labelled for EDL933 and S. sonnei 53G. Genes predicted to be non-functional are denoted in red. Genes of unknown function or unable to be annotated are coloured in black. Insertion sequence elements or transposons are coloured in purple. TssC is annotated as TssC1 (182 bp) and 2 (296 bp). (b) In silico analysis of 20 S. sonnei T6SS clusters reveal predicted non-functional proteins in red and predicted functional proteins in grey. The strain name is labelled on the left of the graph and the genotype of each isolate is indicated on the right of the graph. The non-functional status of TssM is acquired through different mutations; 53G, AR0426, KCCM 41282 have a point mutation leading to a premature stop codon while all others have an IS integrated. (c) A graphical representation of the S. sonnei T6SS with predicted non-functional components highlighted in bold.

Fig. 3.

The S. sonnei T6SS is expressed at very low levels when under the control of its native promoters. (a) In silico analysis of the T6SS operon in S. sonnei reveals a bi-directional regulatory region. For both promoters the predicted transcriptional start sites (TSS) is indicated with an arrow. Putative binding sites for transcriptional regulators, H-NS and CRP identified by BProm are also indicated. (b-e) qRT-PCR was performed to determine the expression level of genes within both operons. Relative gene expression was determined by normalisation to rhoB as the bacterial house-keeping gene. Expression of (b) tssB, (c) tssH and (d) tssA1 from T6SS operon 1 and (e) hcp2 from T6SS operon 2 were quantified. For comparison highly expressed rpoS (f) and the lowly expressed aat (g) were measured. (h-k) The intergenic region between the two operons was inserted in both orientations into a promoter-less GFP vector to create a reporter construct to determine promoter activity. The promoter of ompA was used as a positive control. Transcriptional activity was measured by determining the relative fluorescence compared to bacteria without a reporter construct. Conditions tested included: (h) osmolarity, by culturing in low [8 mM] or high [595 mM] salt concentrations; (i) nutrient availability, by culturing in low (M9 minimal media) and high (TSB) nutrient conditions; (j) temperature, by culturing at 30 °C or 37 °C; and (k) membrane stress and permeability by adding polymyxin B (10 µg ml⁻¹). A two-way ANOVA with Sidak’s correction was applied to test for significance. **= p<0.01, NS=non-significant.

Fig. 4.

Replacement of the native promoters of the T6SS with an inducible version increases T6SS gene expression but does not show evidence of increased anti-bacterial activity. (a) A divergent promoter containing the P_BAD and P_tac was inserted to replace the native promoters within the regulatory region of the S. sonnei T6SS locus. The inducible promoters allowed for regulated control via the supplementation of arabinose and IPTG during bacterial culture. (b-c) RT-qPCR experiments were conducted using extracted RNA to determine gene expression of (b) tssB and c) hcp2. (d-e) The wild-type and inducible strains of S. sonnei were then utilised in competition assays with E. coli. The amount of E. coli, remaining after competition was enumerated to measure the anti-bacterial activity of S. sonnei. Two-way ANOVAs were performed with uncorrected Fisher’s LSD. ****= p<0.0001, ns=non-significant.

Fig. 5.

The anti-bacterial activity of S. sonnei is due to the activity of E-type colicins encoded on spB. (a) A plasmid map of CIP106347 spB depicting the colicin genes encoded: cea (the colicin gene), cei (the colicin immunity gene) and cel (colicin lysis gene required for colicin release). Also depicted are regions involved in replication (ori and rop), recombination (cer) and plasmid mobilisation genes (mobA, exc2 and exc1). (b-e) Competition assays of an E. coli prey strain against attacker strains of S. sonnei were conducted at 37 °C during their exponential growth phases on agar plates. (b and c) The colicin-susceptible E. coli (or E. coli^Km ) but not the colicin-resistant E. coli ΔbtuB (or E. coli ΔbtuB^Km ) were efficiently killed by both SS381 and SS381ΔT6 (b) and CIP106347 and CIP106347ΔT6 (c). In addition, the anti-bacterial activity of CIP106347 was significantly reduced when the plasmid encoding spB was removed from CIP106347 (ΔspB and ΔspBΔT6). (d and e) The wild-type and inducible strains of S. sonnei were then utilised in competition assays with colicin resistant E. coli ΔbtuB (or E. coli ΔbtuB^Km ). The amount of E. coli ΔbtuB remaining after competition was enumerated to measure the anti-bacterial activity of S. sonnei. Each graph represents the mean±SEM of at least three biological replicates. Mixed-effects models (b and c) or two-way ANOVAs were performed with uncorrected Fisher’s LSD (d and e). *= p<0.05, **= p<0.01, **** = p<0.00001, ns=non-significant.

Fig. 6.

S. sonnei secreted colicins only target closely related bacteria and are dependent on the presence of BtuB. (a) A diffusible indicator assay where a susceptible E. coli or non-susceptible E. coli ΔbtuB are seeded in soft LB-agar. Filtered supernatants from the indicated strains were then spotted onto the E. coli lawns. Zones of clearance indicate the presence of a secreted colicin. (b) A panel of human associated commensal bacteria were also tested for their susceptibility to colicin. Zone of clearance (grey) indicates the presence of a secreted colicin and a sensitive strain. No zone of clearance (red) indicates either no colicin secreted or a resistant strain. Abbreviations; SL: S. typhimurium, KP: K. pneumoniae, KO: K. oxytoca, EC: E. coli, PA: P. aeruginosa, CR: C. rodentium, CF: C. freundii. (c) CIP106347 was grown in LB medium with or without the SOS agent Mitomycin C for 5h, pelleted, the supernatant filtered and concentrated. Concentrated S. sonnei supernatants were subjected to SDS-PAGE and proteins visualised by Coomassie blue staining. The predicted position of ColE1 (57.2 kDa) is indicated by the red arrow. MS analysis of this band indicated it was colicin E1. Molecular weight markers (in kDa) are indicated on the left.

Fig. 7.

Colicin activity is dependent on BtuB and, for Colicin E1, TolA. Colicin-containing supernatant from the indicated strains of S. sonnei or EHEC was added to (a-g) E. coli WT and E. coli ΔbtuB or (h-l) SS381 and SS381ΔtolA. The OD600 of E. coli or S. sonnei was measured every ten minutes for 3 h while shaking at 37 °C. (a-g) Growth inhibition of the WT E. coli but not E. coli ΔbtuB was observed for all colicin-containing supernatants tested. (h-l) No growth inhibition of SS381 and SS381ΔtolA was seen for (h) EHEC control supernatant, (i) CIP106347ΔspB or (j) SS381 E7-containing supernatant. (k) Growth inhibition of the SS381 but not SS381ΔtolA was observed for CIP106347 E1-containing supernatant, while (l) both SS381 and SS381ΔtolA were inhibited by SS595 E2-containing supernatant. Each graph represents the mean±SEM of three biological replicates.