Allelic variation contributes to bacterial host specificity

Abstract

Understanding the molecular parameters that regulate cross-species transmission and host adaptation of potential pathogens is crucial to control emerging infectious disease. Although microbial pathotype diversity is conventionally associated with gene gain or loss, the role of pathoadaptive nonsynonymous single-nucleotide polymorphisms (nsSNPs) has not been systematically evaluated. Here, our genome-wide analysis of core genes within Salmonella enterica serovar Typhimurium genomes reveals a high degree of allelic variation in surface-exposed molecules, including adhesins that promote host colonization. Subsequent multinomial logistic regression, MultiPhen and Random Forest analyses of known/suspected adhesins from 580 independent Typhimurium isolates identifies distinct host-specific nsSNP signatures. Moreover, population and functional analyses of host-associated nsSNPs for FimH, the type 1 fimbrial adhesin, highlights the role of key allelic residues in host-specific adherence in vitro. Together, our data provide the first concrete evidence that functional differences between allelic variants of bacterial proteins likely contribute to pathoadaption to diverse hosts.

Figure 1. Comparative analysis and host origin association for S. Typhimurium genomes.

(a) Functional distribution of core genes in 12 isolates; at left total numbers of SNPs with proportions of nonsynonymous substitutions (NS, blue) and synonymous substitutions (SS, red), and at right, the number of genes with (black) and without SNPs (grey). P-values for associations of sets of genes with SNPs and a given functional category (right-tailed Fisher exact test): *P<0.05; **P<0.01; ***P<0.001. (b) 3D scaling plot from a Random Forest proximity matrix of the SNPs from 15 adhesins (using the first 3 principal components) for human (light blue), bovine (green), equine (dark blue) and chicken (red) isolates; the analysis identified host-specific DNA signatures by separating subpopulations of isolates from the same host.

Figure 2. Residue 223 variation in FimH of S. Typhimurium and Typhi swaps host specificity.

(a) Host origin distribution for 580 S. Typhimurium isolates that have either a valine (Val) or an alanine (Ala) at position 223 of FimH. (b) Binding to three human (in red) and four bovine (in black) intestinal epithelial cells of recombinant E. coli expressing Salmonella type 1 fimbriae with the FimH1 or FimH7 alleles that have valine or alanine at position 223, respectively. (c) The different binding properties of fimH80 with valine and engineered fimH80 with alaline at position 223 for three human and four bovine enterocytes. The data in b and c are expressed as mean percentages of bacterial binding relative to the difference between fimH2 (100% binding for FimH of S. Typhimurium strain AJB3, not shown) and ΔfimH (0% binding) with±s.e.m. of three experiments. P-values were calculated by using a one-sided paired t-test: *P<0.05; **P<0.01; ***P<0.001.

Figure 3. FimH protein sequence variants.

Variant residue positions for the unprocessed (top line) and matured (second line) FimH proteins, with the FimH of strain SL1344 (FimH1) used as the comparative standard (third line). The star for Typhimurium fimH6 is a stop codon. The signal peptide (22 residues), lectin (residues 1–173 of the mature protein, yellow background) and pilin domains (residues 177–315) are framed (second line). Substitutions and their corresponding positions are shown for each listed FimH. Variant residues that are predicted to participate in the mannose-binding pocket are highlighted in blue. At left, phylogenetic tree of the major fimH alleles (found in at least ten isolates per serovar; two, two and four isolates for serovars Typhisuis, Abortusovis and Abortusequi, respectively) based on nucleotide sequences and built by using the Maximum-likelihood method with a bootstrap value of 1,000. The mannose-binding properties of FimH are indicated as high binding (black square), low binding (black triangle) and nonbinding (white circle). On the right highlighted in green are the numbers of isolates studied for each listed allele and their origin (host, environment or unknown).

Figure 4. Predicted structure of Salmonella FimH1 (S. Typhimurium SL1344).

The balls highlight the substituted residues. (a) All the amino-acid residues found to be substituted, as listed in Fig. 3; (b) all the substituted residues in the predicted binding pocket (Fig. 3); (c) all the residues predicted to be involved in the mannose-binding pocket, including position 3, 4, 10, 11, 12, 13, 14, 15, 47, 48, 49, 50, 56, 59, 116, 120, 122, 125, 149, 151, 152, 153, 158, 159, 160 and 162. α-Helices were highlighted in dark blue, β-sheets in yellow, β-turns in grey and γ-turns in light blue in the predicted tertiary structure of FimH1.

Figure 5. FimH-mediated host-specific bacterial binding to enterocytes or hepato-epithelial cells.

(a) Recombinant E. coli binding mediated by different Salmonella fimH-encoded allelic proteins with human (black), porcine (red) and bovine (green) enterocytes, as well as chicken hepatoepithelial cells (blue). The data are expressed as mean percentages of bacterial binding relative to the difference between fimH2 (100% binding for FimH of S. Typhimurium strain AJB3, not shown) and ΔfimH (0% binding) with±s.e.m. of five experiments. P-values were calculated by the t-test that compared groups between each individual alleles and all the rest of data. The null hypothesis assumes a common binding affinity for all variants. The threshold of significance for the P-values indicated by grey dash-lines is as follows: 3.58 for P≤0.0001; 3.16 for P≤0.001; 2.65 for P<0.01; 2.19 for P≤0.05; and not shown, 1.94 for P≤0.1. (b) The bars for each fimH allele represent the proportion (1 representing 100%) of Salmonella isolates from each corresponding host.

Figure 6. Substituted sites on FimH alleles visualized on the Salmonella FimH model (S. Typhimurium SL1344).

The balls highlight the substituted residues. (a) Serovar Typhisuis FimH102 or FimH103; (b) serovar Choleraesuis FimH105; (c) serovar Typhi FimH80; (d) serovar Newport FimH44; (e) serovar Dublin FimH98; (f) serovar Gallinarum FimH95; (g) serovar Typhimurium FimH7; (h) serovar Typhimurium FimH2; (i) serovar Abortusequi FimH100. α-Helices were highlighted in dark blue, the β-sheets in yellow, the β-turns in grey and the γ-turns in light blue in the predicted tertiary structure of FimH1.