Genomic epidemiology reveals multidrug resistant plasmid spread between Vibrio cholerae lineages in Yemen

Abstract

Since 2016, Yemen has been experiencing the largest cholera outbreak in modern history. Multidrug resistance (MDR) emerged among Vibrio cholerae isolates from cholera patients in 2018. Here, to characterize circulating genotypes, we analysed 260 isolates sampled in Yemen between 2018 and 2019. Eighty-four percent of V. cholerae isolates were serogroup O1 belonging to the seventh pandemic El Tor (7PET) lineage, sub-lineage T13, whereas 16% were non-toxigenic, from divergent non-7PET lineages. Treatment of severe cholera with macrolides between 2016 and 2019 coincided with the emergence and dominance of T13 subclones carrying an incompatibility type C (IncC) plasmid harbouring an MDR pseudo-compound transposon. MDR plasmid detection also in endemic non-7PET V. cholerae lineages suggested genetic exchange with 7PET epidemic strains. Stable co-occurrence of the IncC plasmid with the SXT family of integrative and conjugative element in the 7PET background has major implications for cholera control, highlighting the importance of genomic epidemiological surveillance to limit MDR spread.

Fig. 1. Phylogenetic diversity of V. cholerae isolates from Yemen.

ML phylogeny of 882 assembled V. cholerae genomes based on the 37,170 SNP sites from the concatenated alignments of 291 core genes. Low-diversity clades (VcH and part of VcK) are collapsed and marked by black stars. Clades are highlighted with background colours (legend key 1). Coloured rings outside the tree depict the match with previously described lineages (ring 2), the geographical origin of isolates at the level of continents (ring 3) and their year of isolation when from Yemen (ring 4). The presence of parts of the plasmid pCNRVC190243 are indicated by coloured circles (ring 5 in A): IncC plasmid backbone (light brown) and the MDR PCT YemVchMDRI (dark brown); full circles indicate more than 70% coverage in assemblies of the reference length, hollow circles indicate 30%–70% coverage in assemblies and confirmed presence based on mapped reads, with even coverage over the MGE reference sequence, whereas half-circles represent heterogeneous presence in a collapsed clade. The scale bar represents the number of nucleotide substitutions per site.

Fig. 2. Phylogenetic diversity and spatio-temporal distribution of V. cholerae 7PET-T13 isolates (VcH.9) from Yemen.

a, Subtree of the ML phylogeny of 456 7PET genomes mapped to reference VcH.9 strain CNRVC190243 genome, including 335/456 genomes covering VcH.9 (as defined in Supplementary Fig. 5), which corresponds to the 7PET-T13 sub-lineage and close South Asian relatives. The full tree containing the 456 genomes is available as supplementary material on figshare (10.6084/m9.figshare.16595999) and was obtained based on 2,092 SNP sites from concatenated whole-chromosome alignments. Brown branches indicate the clade grouping all Yemeni 7PET-T13 isolates. Bootstrap support greater than 70% is indicated by white circles. Phylogenetic clusters within VcH.9 are highlighted with background colours (legend key 1). Coded tracks outside the tree depict the serotype of isolates (ring 2) as predicted from genomic data, year of isolation when isolated in 2012 or later (ring 3) and the governorate of isolation if in Yemen (ring 4). The presence of MGEs is indicated by coloured circles in the outermost track (ring 5): ICP1-like phage (pink), SXT ICE ICEVchInd5 (blue), ICEVchInd5^Δ that is featuring the characteristic 10-kb deletion in the variable region III (green), IncC plasmid backbone (light brown) and the MDR PCT YemVchMDRI (dark brown); filled and unfilled circles indicate different levels of coverage in assemblies (as in Fig. 1 legend). The position of the reference sequence to which all other genomes were mapped to generate the alignment is labelled. The scale bar represents the number of nucleotide substitutions per site. b, Frequency of each phylogenetic subcluster among Yemen isolates per month since the onset of the Yemen outbreak. Where relevant, the cluster group is subdivided by the presence or absence of the IncC plasmid as indicated by the filled brown (present) or open (absence) circle on the right of the chart. The contribution of each governorate of isolation is indicated by the coloured portion of each bar. c,d, A map of Yemen governorates (c) and a focus on the Sana’a and Amanat Al Asimah governorates (inner and outer capital city; d), with dots corresponding to isolates, coloured by phylogenetic subcluster.

Fig. 3. Genetic organization of the MDR PCT YemVchMDRI.

AMR genes are filled in black and labelled in bold; genes encoding endonucleases transposases and other genes involved in genetic mobility are filled in grey. Genomic position is indicated by tick marks every kilobase, in reference to the pCNRVC190243 plasmid coordinates.

Extended Data Fig. 1. Culture confirmation of samples derived from suspected cholera cases in Yemen, 2017–2019.

Distribution over time of V. cholerae culture result samples received at the NCPHL, broken down by governorate. Data are derived from Electronic Disease Early Warning System (eDEWS) sample lists (Supplementary Table 1). NG, no growth; NA, not available.

Extended Data Fig. 2. Antibiotic susceptibility phenotypes of all V. cholerae isolates collected in Yemen, 2017 and 2019.

Distribution over time of resistance and sensitivity to broad antibiotic classes among culture-confirmed V. cholerae isolates received at the NCPHL. Data are derived from eDEWS sample lists (Supplementary Table 1).

Extended Data Fig. 3. Flowchart of sample collection, management and use in protocols and analyses.

Experiments and analyses are itemized and grouped according to the different locations of the collaborative consortium where they were undertaken: NCPHL, The National Centre of Public Health Laboratories; IP, Institut Pasteur; WSI, Wellcome Sanger Institute.

Extended Data Fig. 4. Phylogenetic diversity of Vibrio cholerae isolates from Yemen and contextual samples.

Expanded version of phylogentic tree shown in Fig. 1 (no clades are collapsed).

Extended Data Fig. 5. Phylogenetic diversity of Vibrio cholerae VcH isolates from Yemen and contextual samples.

Expanded version of phylogentic tree shown in Fig. 1, focusing on the subtree of clade VcH (collapsed in Fig. 1), with details of its phylogenetic substructure.

Extended Data Fig. 6. Timed phylogeny of Vibrio cholerae VcH.9 isolates from Yemen and contextual samples.

Timed phylogeny of 335 VcH.9 genomes obtained by estimating the dates of nodes using BactDating from a recombination-free phylogeny computed with ClonalFrameML as input; the 335 genomes correspond to the VcH.9 clade within the 456 mapped 7PET genome tree (as presented in Fig. 2a). Subclusters are labelled and coloured as per previous figures. X axis represents time in years. Horizontal blue bars indicate 95% confidence intervals around the node dates. The vertical purple line marks the emergence of the clade of Yemen isolates.

Extended Data Fig. 7. In-silico prediction of the O-antigen diversity among Vibrio cholerae from Yemen and contextual samples.

Prediction of lipopolysaccharide (LPS) O-antigen serogroup and O1 serotypes projected on the isolate trees corresponding to (A) the core-genome tree as presented in Fig. 1a,b the subtree of the mapped 7PET genome tree as presented in Fig. 2a. Serogroup prediction are based on a best normalized BlastN score in a search against the reference LPS O-antigen biosynthetic cluster sequence from Murase et al. 2022; predictions where the best hit had a normalized percent nucleotide identity below 98% are indicated in grey font.

Extended Data Fig. 8. Comparison of IncC plasmids pCNRVC190243 and pYA00120881.

BlastN alignment of the differing regions of IncC plasmids pCNRVC190243 and pYA00120881. Regions of similarity are highlighted by interleaving bands; band colour indicate similarity intensity and orientation of the alignment (blue, direct; pink, reverse).

Extended Data Fig. 9. Comparison of ISCR1 elements.

BlastN alignments of ISCR1 regions of (a) A. baumanii str. AB154 plasmid pAB154 integron (JQ639792.1), (b) pCNRVC190243/YemVchMDRI and (c) A. baumanii str. AP2 integron (HQ713678.1).

Extended Data Fig. 10. Recombination-free phylogeny of Vibrio cholerae VcH.9 isolates from Yemen and contextual samples.

Tree computed from the same alignment as in Fig. 2a, but using ClonalFrameML to infer a recombination-free phylogeny reflecting the clonal propagation of the organism.