Capsules and their traits shape phage susceptibility and plasmid conjugation efficiency

Abstract

Bacterial evolution is affected by mobile genetic elements like phages and conjugative plasmids, offering new adaptive traits while incurring fitness costs. Their infection is affected by the bacterial capsule. Yet, its importance has been difficult to quantify because of the high diversity of confounding mechanisms in bacterial genomes such as anti-viral systems and surface receptor modifications. Swapping capsule loci between Klebsiella pneumoniae strains allowed us to quantify their impact on plasmid and phage infection independently of genetic background. Capsule swaps systematically invert phage susceptibility, revealing serotypes as key determinants of phage infection. Capsule types also influence conjugation efficiency in both donor and recipient cells, a mechanism shaped by capsule volume and conjugative pilus structure. Comparative genomics confirmed that more permissive serotypes in the lab correspond to the strains acquiring more conjugative plasmids in nature. The least capsule-sensitive pili (F-like) are the most frequent in the species' plasmids, and are the only ones associated with both antibiotic resistance and virulence factors, driving the convergence between virulence and antibiotics resistance in the population. These results show how traits of cellular envelopes define slow and fast lanes of infection by mobile genetic elements, with implications for population dynamics and horizontal gene transfer.

Fig. 1. Capsule serotype determines phage infectivity.

A Overview of the genetic loci encoding the four different capsule serotypes included in this study and their chemical composition. Arrows represent the different genes, galF and ugd in blue corresponding to the regions involved in homologous recombination to generate the swaps. Conserved genes involved in assembly and export of the capsule (wza, wzb, wzc, wzi) and initiating glycosyl-transferase (wcaJ, wbaP) are labelled. The chemical composition of the capsule (monomers and their organisation), is displayed on the right of each locus (predicted by K-PAM). B Matrix of phage infection. Infection assay for each of the three phages (panels), three swapped strains (y-axis), and different genotypes (x-axis). White tiles correspond to non-productive infection, i.e., no plaque could be identified. Coloured tiles correspond to the average PFU/mL normalized by the lysate titre for productive infections (Supplementary Fig. S2). Grey tiles correspond to non-productive infection with significant adsorption, while white tiles correspond to non-adsorptive pairs (Supplementary Fig. S3). Values are the mean of three independent replicates after log₁₀-transformation. Strain-specific defence systems identified by DefenseFinder as of 02/2023 are displayed on the right. Source data are provided as a Source Data file 1 (Adsorption) and Source Data file 2 (Infection).

Fig. 2. Recipient’s capsule and serotype influences conjugation efficiency.

A Conjugative plasmids included in the analysis. The plasmids are presented on a cladogram representing the evolutionary relations between the MPF types. Note that p580 encodes two separate MPF systems of type F and T, but the F-type locus is interrupted by a transposon. Plasmid names colours match the colours of the points in the other panels. The three first columns correspond to the mean conjugation efficiency (n = 3) measured from E. coli to each of the three wild type strains. Additionally, we indicate the predicted incompatibility (Inc) groups, the antibiotic used for selection (R), either Ertapenem (E) or Kanamycin (K), and the TraN allele for F-type plasmids. B Log₁₀-transformed conjugation efficiency (CE) relative to the associated Δcps mutant (by subtraction of the log₁₀-transformed values) by capsule serotype of the recipient, from E. coli DH10B donors. Points represent the mean of independent triplicates after log₁₀-transformation and subtraction, with colours and shapes corresponding respectively to plasmids (A) and strains. Solid line at y = 0 represents the conjugation efficiency of Δcps mutant. We used paired Wilcoxon tests (two-sided) to assess statistically significant differences. C Same as (B), but with K. pneumoniae strains as donors. The shapes of the data points represent the genotype of the donor strain. Source data are provided as a Source Data file 3 (Conjugation from E. coli) and Source Data file 4 (Conjugation from K. pneumoniae). Panels showing data of the individual biological replicates are presented in Supplementary Figs. S4 and S5. ***p < 0.001; ns p > 0.05.

Fig. 3. Donor serotype influences conjugation efficiency.

A Log₁₀-transformed conjugation efficiency (CE) relative to the associated Δcps mutant (by subtraction of the log₁₀-transformed value) by capsule serotype of the donor. Points represent the mean of independent triplicates after transformation and subtraction, with colours and shapes corresponding to strains and plasmids in Fig. 2A. Solid line at y = 0 represents the conjugation efficiency of Δcps mutant. All four serotypes are associated with significantly lower conjugation efficiencies than the Δcps mutants (Pairwise Wilcoxon (two-sided), all p < 0.001), but not when considering pKPN4 individually (all p > 0.05). Individual biological replicates are presented in Supplementary Figs. S4 and S5. B Log₁₀-transformed conjugation efficiency between pairs of donors and recipients with similar or different serotypes. Points represent mean conjugation efficiencies (n = 4 biologically independent experiments), colours indicate different plasmids (as in Figs. 2 and 3) and shapes correspond to distinct pairs. Conjugation is not more efficient between donor and recipient expressing the same serotype, paired Wilcoxon test (two-sided), p = 0.9. C Networks of plasmid transfer between capsule states. Data drawn from the K. pneumoniae to K. pneumoniae assays. Nodes represents distinct serotypes, and Δcps mutants. Edges thickness represents the mean conjugation efficiency for all pairs. Edges are coloured according to the donor, indicating the direction of transfer. Source data are provided as a Source Data file 4 (Conjugation from K. pneumoniae) and Source Data file 5 (Conjugation within serotype). ***p < 0.001; ns p > 0.05.

Fig. 4. Effective volume within colonies.

A Differences in effective volume (μm³/CFU) between serotype swaps and Δcps mutants. Individual points represent the mean effective volume for each mutant (n = 4), while the boxplots are drawn from all the observations. We assessed statistical differences between volumes with an ANOVA followed by a Tukey’s honest significant difference (HSD) test. The strain, genotype and their interaction significantly influenced the effective volume. All comparisons between genotypes were significantly different (K3 vs Delta, p < 0.001; K2 vs K24, p = 0.004; K1 vs K2, p = 0.004), except K24 and K3 (p = 0.85). Genotypes were ranked according to the median effective volume. Boxes represent the first quartile (Q1), median (line), third quartile (Q3), while whiskers represent the Q1–IQR*1.5 and Q3 + IQR*1.5 (IQR, inter-quartile range). Effective volume positively correlates with amount of capsule quantity (Supplementary Fig. S1A). B Mean effective volume vs. mean log₁₀-transformed conjugation efficiencies of recipients (from E1 and E2 combined). Points correspond to distinct capsule states in the recipient strain. Colour corresponds to the strains as in (A). Lines represent linear regressions for each chassis strain between the log₁₀-transformed conjugation efficiency and the average effective volume (EV). The linear mixed model of the log₁₀ transformed conjugation efficiency using the effective volume as a fixed effect and the chassis strain identity as a random effect showed a significant effect of the volume (F test, F = 45, p < 0.001, Statistics 6, Text S2). Source data are provided as a Source Data file 6 (Volume) and Source Data file 7 (Volume vs. recipient conjugation). *p < 0.05; ***p < 0.001; ns p > 0.05.

Fig. 5. Distribution of recently acquired plasmids.

Analysis of 623 complete K. pneumoniae genomes. A Number of conjugative plasmids acquired recently in the genomes of our dataset. The colours correspond to the different categories of plasmids. The data correspond to conjugative plasmids acquired in terminal branches. B Number of plasmids (including non-conjugative) recently acquired (terminal branches of the species tree) in the focus serotype. Bars represent the mean of plasmid gains per genome (n = 73 for “Large”, n = 29 for “Small”). Error bars represent the standard deviations. A generalized additive model was fit with the gains as the response variable, log₁₀ branch lengths as a smooth function term (thin-plate regression spline), and serotype (grouped between large and small capsules) and MPF type (I, T, F, Others) as fixed terms (Statistics 9, Text S2). Source data are provided as a Source Data file 8 (Plasmids annotations) and Source Data file 9 (Plasmid acquisitions).