Alternate typhoid toxin assembly evolved independently in the two Salmonella species

ABSTRACT AB5-type toxins are a diverse family of protein toxins composed of an enzymatic active (A) subunit and a pentameric delivery (B) subunit. Salmonella enterica serovar Typhi’s typhoid toxin features two A subunits, CdtB and PltA, in complex with the B subunit PltB. Recently, it was shown that S. Typhi encodes a horizontally acquired B subunit, PltC, that also assembles with PltA/CdtB to produce a second form of typhoid toxin. S. Typhi therefore produces two AB5 toxins with the same A subunits but distinct B subunits, an evolutionary twist that is unique to typhoid toxin. Here, we show that, remarkably, the Salmonella bongori species independently evolved an analogous capacity to produce two typhoid toxins with distinct B subunits. S. bongori’s alternate B subunit, PltD, is evolutionarily distant from both PltB and PltC and outcompetes PltB to form the predominant toxin. We show that, surprisingly, S. bongori elicits similar levels of CdtB-mediated intoxication as S. Typhi during infection of cultured human epithelial cells. This toxicity is exclusively due to the PltB toxin, and strains lacking pltD produce increased amounts of PltB toxin and exhibit increased toxicity compared to the wild type, suggesting that the acquisition of the PltD subunit potentially made S. bongori less virulent toward humans. Collectively, this study unveils a striking example of convergent evolution that highlights the importance of the poorly understood “two-toxin” paradigm for typhoid toxin biology and, more broadly, illustrates how the flexibility of A-B interactions has fueled the evolutionary diversification and expansion of AB5-type toxins. IMPORTANCE Typhoid toxin is an important Salmonella Typhi virulence factor and an attractive target for therapeutic interventions to combat typhoid fever. The recent discovery of a second version of this toxin has substantial implications for understanding S. Typhi pathogenesis and combating typhoid fever. In this study, we discover that a remarkably similar two-toxin paradigm evolved independently in Salmonella bongori, which strongly suggests that this is a critical aspect of typhoid toxin biology. We observe significant parallels between how the two toxins assemble and their capacity to intoxicate host cells during infection in S. Typhi and S. bongori, which provides clues to the biological significance of this unusual toxin arrangement. More broadly, AB5 toxins with diverse activities and mechanisms are essential virulence factors for numerous important bacterial pathogens. This study illustrates the capacity for novel A-B interactions to evolve and thus provides insight into how such a diverse arsenal of toxins might have emerged.

WT and DpltB S. bongori strains were grown TTIM for 6 hours at which point RNA was isolated and transcript levels were analysed using RT-qPCR.pltA transcript levels were normalized to a constitutively expressed control gene (dsbC).Bars represent the average relative transcript levels for pltA over three biological replicates from two separate experiments (n=6).Two-tailed ttests were used to investigate the statistical significance of any differences in pltA levels observed between the WT and DpltB strains.*** = p, 0.001.

Fig. S5: Example of the gating strategy used for analysis of flow cytometry data for cellular intoxication assays.
Gating to analyze cell cycle using PI-stained cells was accomplished using a four-step strategy, as shown above.Particles were first gated for size using FSC area / SSC area to omit cellular debris (1), followed by selection for single cells (doublet omission) using FSC height / FSC area (2), followed by gating on fluorescence intensity in the channel used to detect PI fluorescence in order to omit any remaining particles with a fluorescence intensity outside the minimum (G1 population) and maximum (G2-M population) expected of single cells (3), followed by the final step of identifying the G2-M peak used for intoxication analysis (4).Results shown are from a typical unintoxicated/uninfected control sample.Note that somewhat larger particles were included in our gating to account for CdtB-mediated cellular distension, which is known to yield cells with an atypically large volume.Two cdtB-His6 S. bongori strains, pltB-3F (pltBSb) as well as a strain in which the native pltB gene was replaced with S. Typhi pltB-3F (pltBTy), were grown in TTIM for 24 hours.The bacteria were then lysed, and clarified lysates were immunoprecipitated using an a-FLAG antibody.Samples from clarified lysates (input) and elutions from the immunoprecipitation (Post-IP) were then analyzed by western blot using both a-FLAG and a-His6 antibodies.The red box highlights the lanes where interactions are investigated (detecting the protein in the elution samples that lacks the 3F tag targeted by the IP).This experiment was conducted independently two times with equivalent results.Fig. S8: Replicate experiment comparing the levels of toxin subunits in the infection inoculum to levels 24 hpi.S. bongori strains encoding the indicated 3F-tagged genes were used to infect HeLa cells at an MOI of 50.Bacteria were extracted 24 hpi, pelleted, and whole cell lysates were analyzed by western blot using a-FLAG and a-RpoB (loading control) antibodies.Despite higher levels of the inoculum being loaded (RpoB blot), the typhoid toxin subunits PltD and CdtB are only detectable in the infection samples, demonstrating their infection-specific expression pattern.This is a second independent experiment conducted in the same manner as that presented in Fig 5B .

Fig. S1 :
Fig. S1: Sequence diversity within typhoid toxin islet clades as a function of lineage.The percent DNA sequence identity of typhoid toxin islets for the enterica I clade (A) or the enterica II clade (B) relative to the representative members of that clade, broken down by relevant phylogenetic groups.Clade enterica I (panel A) comparisons are to S. Typhi strain Ty2 and are colour coded by typhoidal strains (sv.Typhi and Paratyphi A), nontyphoidal subsp.enterica strains, and subsp.salamae strains as indicated.Clade enterica II comparisons (panel B) are to S. arizonae strain S499 and are colour coded by subspecies as indicated.These analyses use the same raw data as the analyses in Fig 1C and are intended to highlight phylogenetic relationships within clades.For both panels, the specific lineage of numerous database sequences was not available and thus these sequences were excluded from this analysis.Raw data for these analyses can be found in Supplemental Dataset 1.

Fig. S2 :
Fig. S2: Phylogenetic and sequence analysis of PltD.(A) Amino acid sequence alignment comparing the sequence of PltD to diverse pertussis toxin family B subunits.Sb, S. bongori strain SARC11; Ty, S. Typhi strain Ty2; Az, S. arizonae strain S499; DT104, S. Typhimurium phage type DT104; SubB sequence is from E. coli strain 98NK2 and S2 (pertussis toxin) sequence is from B. pertussis strain Tohama-1.Alignment generated using Clustal Omega.A snippet of this alignment is shown as Fig 1E.(B) Genome diagram showing the regions of significant DNA sequence similarity between the S. bongori pltD locus and the S. bongori artBA locus (top) or the S. Typhi pltC locus (bottom).(C)Phylogenetic tree showing predicted evolutionary relationships for the pertussis family B subunits described in (A).The tree was generated with the MEGA (Molecular Evolutionary Genetics Analysis) V11 software using the maximum likelihood method and a WAG +G +I substitution model.A bootstrap method with 500 total replicates was used, and the numbers at the nodes represent the support values.

Fig. S3 :
Fig. S3: RT-qPCR analysis of S. bongori toxin gene expression indicates that pltD is coregulated with typhoid toxin genes.WT S. bongori (SARC 11) was grown in either TTIM and TTIM+Mg for 24 hours (A), or LB with and without 0.5 µg/ml MMC for 16 hours (B), at which point RNA was isolated and transcript levels were analysed using RT-qPCR.Transcript levels of the indicated genes were normalized to a constitutively expressed control gene (dsbC).Bars represent the average fold increase in RNA levels for the indicated genes over at least five independent samples from two separate experiments in the inducing condition [TTIM in (A), +MMC in (B)] compared to the non-inducing condition [TTIM+Mg 2+ in (A), LB only in (B)], and error bars represent the standard deviation.Two-tailed t-tests were used to determine the significance of the induction observed for each gene under each growth condition: * = p < 0.05, ** = p < 0.01, *** = p , 0.001, **** = p < 0.0001, n.s.s.= not statistically significant.

Fig
Fig S4: RT-qPCR analysis of the effect of the DpltB mutation on pltA transcript levels.WT and DpltB S. bongori strains were grown TTIM for 6 hours at which point RNA was isolated and transcript levels were analysed using RT-qPCR.pltA transcript levels were normalized to a constitutively expressed control gene (dsbC).Bars represent the average relative transcript levels for pltA over three biological replicates from two separate experiments (n=6).Two-tailed ttests were used to investigate the statistical significance of any differences in pltA levels observed between the WT and DpltB strains.*** = p, 0.001.

Fig. S6 :
Fig. S6: Protein levels and toxin formation for S. Typhi PltB encoded in place of the native S. bongori pltB.Two cdtB-His6 S. bongori strains, pltB-3F (pltBSb) as well as a strain in which the native pltB gene was replaced with S. Typhi pltB-3F (pltBTy), were grown in TTIM for 24 hours.The bacteria were then lysed, and clarified lysates were immunoprecipitated using an a-FLAG antibody.Samples from clarified lysates (input) and elutions from the immunoprecipitation (Post-IP) were then analyzed by western blot using both a-FLAG and a-His6 antibodies.The red box highlights the lanes where interactions are investigated (detecting the protein in the elution samples that lacks the 3F tag targeted by the IP).This experiment was conducted independently two times with equivalent results.

Fig. S7 :
Fig. S7: Recovery of S. bongori mutant strains from HeLa cell infections.WT S. bongori and the indicated mutant strains were used to infect HeLa cells at a multiplicity of infection (MOI) of 10.Gentamycin was added to the growth medium to prevent the growth of extracellular bacteria.Cells were collected and bacteria were isolated 48 hours post-infection and plated on LB-agar plates to determine the numbers of CFU recovered.statistical significance of the indicated comparisons was determined by Tukey's test: n.s.s., not statistically significant.Error bars represent one standard deviation.