Genomic and phenotypic diversity of coastal Vibrio cholerae strains is linked to environmental factors

Abstract

Studies of Vibrio cholerae diversity have focused primarily on pathogenic isolates of the O1 and O139 serotypes. However, autochthonous environmental isolates of this species routinely display more extensive genetic diversity than the primarily clonal pathogenic strains. In this study, genomic and metabolic profiles of 41 non-O1/O139 environmental isolates from central California coastal waters and four clinical strains are used to characterize the core genome and metabolome of V. cholerae. Comparative genome hybridization using microarrays constructed from the fully sequenced V. cholerae O1 El Tor N16961 genome identified 2,787 core genes that approximated the projected species core genome within 1.6%. Core genes are almost universally present in strains with widely different niches, suggesting that these genes are essential for persistence in diverse aquatic environments. In contrast, the dispensable genes and phenotypic traits identified in this study should provide increased fitness for certain niche environments. Environmental parameters, measured in situ during sample collection, are correlated to the presence of specific dispensable genes and metabolic capabilities, including utilization of mannose, sialic acid, citrate, and chitosan oligosaccharides. These results identify gene content and metabolic pathways that are likely selected for in certain coastal environments and may influence V. cholerae population structure in aquatic environments.

FIG. 1.

Locations of sampling sites in and around San Francisco Bay, California. Sites where Vibrio cholerae strains were isolated are displayed in bold followed by an asterisk. Darker colored areas along the coast indicate dense vegetative cover. (Map background data available from U.S. Geological Survey/EROS, Sioux Falls, SD.)

FIG. 2.

Estimation of Vibrio cholerae core genome size by regression analysis. Open circles with 95% confidence limits represent the mean number of core genes with increasing numbers of genomes sampled for 10,000 random permutations of sampling order. A power law regression fit [y = a × (◯b) + c] with an R-squared value of 0.9998 is included. Regression coefficients with 95% confidence limits (CL) are as follows: a, 906.1 (CL, 894.1, 918.0); b, −0.8215 (CL, −0.8348, −0.8083); and c, 2,741 (CL, 2,739, 2,744). The horizontal dashed line represents the extrapolated core genome size for Vibrio cholerae, which is equal to 2,741 genes for a threshold of genes shared among 95% of sampled genomes. (Inset) Closed squares show the reduction in projected core genome size with increased stringency for gene ubiquity from 95% to 100% of strains.

FIG. 3.

Division of strains into clades based on CGH profile. The UPGMA tree was generated using Jukes-Cantor distances and 1,000 bootstrap replicates, which provide 100% support for the five genotype clusters. Clades A, B, C, and D and clinical strains are shown in cyan, green, yellow, red, and brown, respectively. Bootstrap scores greater than 50 are displayed above the respective nodes.

FIG. 4.

Distribution of genotype groups illustrates relationship with changes in the environment. Clades A, B, C, and D are shown in cyan, green, yellow, and red, respectively. (A) Number of unique genotypes in each clade isolated, plotted for each month throughout the year. (B) Diversity of genotypes isolated from individual sampling sites over the entire sampling period listed, from north to south. Total numbers of strains sampled are listed below each pie graph. (Bottom) Canonical correspondence analysis ordination plots for (C) water temperature and (D) log ammonium. Each spot represents a genotype colored by clade, with the size of the spot proportional to the magnitude of the parameter when the strain was isolated. R values for axes 1 and 2 for water temperature are −0.742 and −0.294, respectively, confirming that clade C, followed by D, is most likely to be found in warmer water. R values for axes 1 and 2 for log ammonium are −0.470 and 0.779, respectively, confirming that clade D, followed by B, is the most likely to be found in nutrient-enriched waters.