Fig 1.
Summary of reference genomes used in this study.
(a) Midpoint rooted maximum likelihood core-gene phylogeny of all Shigella and E. coli reference genomes used in this study (tips coloured as per legend) and a set of 100 non-redundant E. coli genomes (grey tips). E. coli phylogroups are labelled in grey, with Shigella clades labelled as per Yang et al 2007 (C designations) and Sahl et al 2015 (S designations). Grey boxes indicate the three Shigella lineages that are analysed in this study. (b) Bar plot showing total number of IS in each reference genome (names coloured as panel a), broken by IS type as per legend. Final bar shows median number of each IS found amongst the other E. coli (grey tips in panel a). (c) Bar plots showing total genome size for each reference genome, broken down by number of bases belonging to IS sequence (red) vs not IS sequence (black).
Fig 2.
IS found in each Shigella species.
(a) Time-calibrated Bayesian (BEAST) phylogeny of 125 S. dysenteriae genomes next to bar plots showing IS copy number in each genome. Inset, PCA of IS insertion site matrix, with points coloured by lineage as indicated by tree branch colours. (b) Same for 343 S. flexneri genomes but with a midpoint rooted maximum-likelihood phylogeny. (c) Same for 126 S. sonnei genomes (time-calibrated Bayesian phylogeny). (d) Scatterplot of IS copy number (inferred using ISMapper) on year of isolation, for 126 the S. sonnei genomes. Points are coloured by lineage, as per tree branch colours in (c). Fitted lines show linear regression of IS copy number against year for each lineage, with a single slope fit to all lineages. Dashed horizontal lines indicate the total IS copy number estimated in each lineage’s MRCA using ancestral state reconstruction; grey boxes show 95% HPD intervals for the date of each lineage MRCA estimated from the BEAST analysis. (e) Phenogram of S. sonnei time-calibrated tree from panel (c), mapped to y axis to indicate IS copy number inferred at each node on the tree based on ancestral state reconstruction. Branches are coloured by lineage as per legend. Dashed lines indicate two possible reconstructions for the lower and upper bound of IS burden at the root.
Fig 3.
Comparison of nucleotide and IS-profile similarities across Shigella species.
a, Pairwise nucleotide divergence for genomes in each Shigella population, estimated from mapping-based SNV counts. b-c, Pairwise comparisons of shared (b) and non-shared (c) IS insertion sites for genomes in each Shigella population, based on ISMapper analysis.
Fig 4.
Activity of five common IS in each Shigella species.
Bars indicate the ratio of strain-specific IS sites (estimated by ISMapper) to strain-specific SNVs, for each of the five common IS in each Shigella species population dataset. Error bars indicate 95% confidence intervals for the mean ratio across all terminal branches. Significant differences between species are indicated by asterisks, * indicates p < 0.05, *** indicates p < 10−4.
Fig 5.
Comparison of IS burden in Shigella species and E. coli.
a, Boxplots showing the distribution of number of IS in each Shigella species and the six pathogenic E. coli lineages, using ISMapper. b, Boxplots showing the distribution of number of unique insertion sites for each of the five common IS in each Shigella species and the six pathogenic E. coli lineages, estimated using ISMapper. c, Histogram of the total number of IS (as estimated by BLAST) for the non-redundant set of E. coli shown in Fig 1A. Coloured lines indicate median number of IS found in each Shigella species and the six pathogenic E. coli lineages (as measured by ISMapper).
Fig 6.
Patterns of IS, pseudogene and metabolic diversification within each Shigella species.
a, Histogram summarises the population prevalence of gene interruptions detected in each Shigella species, coloured to indicate the mechanism/s of interruption as per inset legend. b, Linear regressions of pairwise counts of non-shared IS (red), pseudogenes (green) and metabolic phenotypes (purple) against pairwise SNV distance, for each Shigella species. Fitted lines are labelled with their slope. The full data to which each line was fit can be found in S10 Fig. All correlations were significant (p<0.001) using Mantel test to compare the pairwise distance matrices.
Fig 7.
Genome-scale modelling illustrates convergent evolution of Shigella species.
a, Shared and strain-specific metabolic reactions present in GEMS constructed for hypothetical IS-free reference sequences of four selected Shigella species. b, Shared and strain-specific reactions are distributed across metabolic subsystems with the most strain-specific reactions found in inner membrane transport and carbohydrate metabolism. c, Percentage of Shigella, EIEC and E. coli isolates predicted to be capable of growth on 386 different growth supporting nutrients. x-axis is a list of substrates order by their percentage growth across all species. d, Blue boxplots show the number of metabolic phenotypes (growth capabilities) lost in each E. coli, EIEC and Shigella strain, compared to the core metabolic capability of E. coli (n = 316 substrates). Yellow boxplots indicate the overlap between these lost phenotypes and the set of 178 core E. coli phenotypes lost by S. dysenteriae. e, Heatmap of model-predicted growth capabilities for subsets of Shigella strains, including IS-free reference genomes (R) and lineages within each species (lineages defined and ordered by tree structure as illustrated on left and in Fig 2); three EIEC reference genomes (from Fig 1A, all grouped together); and 47 E. coli genomes, grouped by disease type (C–commensal; EI–extra-intestinal; I–intestinal). Proportions of isolates in each group capable of growth on each nutrient source are coloured according to legend. f, Summary of convergent degradation in taurine (sulfur) and maltodextrin (carbon) metabolism pathways. Arrows indicate reactions present in the intact E. coli pathways and are coloured by their frequency of loss in the three main Shigella species as per inset legend. Gene inactivation events resulting in loss of reactions are indicated with symbols above or below the arrow; shapes indicate the genetic mechanism (IS, mutation or deletion, see inset legend), colours indicate the species (as per panel c), black-outlined shapes indicate mutations that are fixed within the species.