Genetic exchanges are more frequent in bacteria encoding capsules

Abstract

Capsules allow bacteria to colonize novel environments, to withstand numerous stresses, and to resist antibiotics. Yet, even though genetic exchanges with other cells should be adaptive under such circumstances, it has been suggested that capsules lower the rates of homologous recombination and horizontal gene transfer. We analysed over one hundred pan-genomes and thousands of bacterial genomes for the evidence of an association between genetic exchanges (or lack thereof) and the presence of a capsule system. We found that bacteria encoding capsules have larger pan-genomes, higher rates of horizontal gene transfer, and higher rates of homologous recombination in their core genomes. Accordingly, genomes encoding capsules have more plasmids, conjugative elements, transposases, prophages, and integrons. Furthermore, capsular loci are frequent in plasmids, and can be found in prophages. These results are valid for Bacteria, independently of their ability to be naturally transformable. Since we have shown previously that capsules are commonly present in nosocomial pathogens, we analysed their co-occurrence with antibiotic resistance genes. Genomes encoding capsules have more antibiotic resistance genes, especially those encoding efflux pumps, and they constitute the majority of the most worrisome nosocomial bacteria. We conclude that bacteria with capsule systems are more genetically diverse and have fast-evolving gene repertoires, which may further contribute to their success in colonizing novel niches such as humans under antibiotic therapy.

Fig 1. Gene exchange in bacterial species is higher in species coding for capsules.

A. Percentage of genes for which the null hypothesis of no homologous recombination was refuted by PHI program as measured by excess polymorphism (CHI), by phylogenetic incongruence (PHI) and by neighbour similarity score (NSS). Species with capsules are designated as C_sp+, N = 68; species without capsules are grouped as C_sp-, N = 59. Percentage at the bottom of the panel indicates the average percentage of recombining genes. The median is highlighted by the boxplot. B. Number of recombination events as inferred by ClonalFrameML. C. Comparison between species with and without capsule in pan-genome size expressed as the number of gene families. D. Horizontal gene transfer events as inferred by Count. Events are log₁₀-transformed for visual purposes. See S1 Table for the details on the statistical tests. * P < 0.05, GLM. Points represent individual species, and dispersion along the x-axis was done for visualization purposes.

Fig 2

Co-occurrence between capsule systems and mobile genetic elements; Prophages (A), Transposases (B), Integrons (C) and Plasmids (D). Stars inside bars represent the result of two-tailed binomial tests to measure the difference between the observed over the expected events, indicated by the dashed line corresponding to the database bias (57%). The stars on top of the bars are the result of dependence tests (χ2 test). All statistics were corrected for genome size and phylogeny (see S3 and S4 Tables for details). *** P < 0.001.

Fig 3. Capsule systems in MGEs.

A. Number of plasmids encoding capsule systems in function of the type of plasmid (classed in terms of mobility by conjugation). Plasmids lacking MOB may be mobilized by conjugation if they have a compatible oriT or mobilizable by other unknown means (e.g., natural transformation in competent species). B. Details of the nine capsules found in prophages. The arrows indicate the relative position and span of the capsule system in each prophage. Right panel indicates the size of each prophage. Dashed line indicated the average size of prophages in the database (40 kb). C and D. Details of the prophages and capsule systems from S. enterica and B. selenitireducens. Genetic schemes are drawn to scale (kb). In the drawing of the genetic locus of the prophage; genes associated to prophage biology are highlighted in green and capsule genes in dark blue. Circular diagrams represent the genomic localization of all the prophages in both species. The capsule-coding prophage is highlighted in dark blue. In the drawing of the locus of the capsule system, proteins in red-pink tones are associated to sugar modifications and may determine capsular serotype. Gene names are indicated below the arrows. GT1: glycosyl transferase.

Fig 4. Co-occurrence between capsule systems and RMS systems.

A. Presence of RMS systems in genomes with and without capsule system (χ² test, and corrected for genome size with glm). RMS systems were identified using the highly specific and publicly available HMM profiles in https://gitlab.pasteur.fr/erocha/RMS_scripts. To control for phylogeny we made a complementary analysis restricted to Firmicutes and Proteobacteria. This analysis gave similar results (using BayesTraits, Bayes Factor of 41.3 and 17.2 for Proteobacteria and Firmicutes respectively). Dashed line indicates the ratio of genomes encoding at least one capsule system in the database (57%). B. Number of RMS systems in genomes with and without capsule systems. Correction for genome size was performed as above. Phylogeny was taken into account using GEE (P < 0.0001 and P = 0.1 for Proteobacteria and Firmicutes).

Fig 5. Antibiotic resistance proteins in genomes.

Results displayed correspond to those hits with at least 50% of protein identity for all databases (except RESFAM, which is based on HMM profiles). A. Venn Diagram showing the total number of genes associated with antibiotic resistance in all C_g+ according to five different ARG databases. B. Mean number of ARGs per genome, for all five databases and the intersect between them (Identified by all DB). P-value corresponds to the difference between the mean MGE in C_g+ and in C_g- genomes (all corrected for genome size). C. Distribution of resistance proteins per genome in function of capsule content classified by resistance mechanisms. These results are based on protein hits from the RESFAM database. Stars indicate significant difference in the median number of resistance proteins, *** P < 0.001.