New Variant of Multidrug-Resistant Salmonella enterica Serovar Typhimurium Associated with Invasive Disease in Immunocompromised Patients in Vietnam

Nontyphoidal Salmonella (NTS) is a common cause of bacterial enterocolitis (diarrheal disease) in humans and animals (1). Additionally, subsets of NTS organisms are also associated with an aggressive invasive disease in susceptible humans (2) and have been shown to cause invasive disease in animals (3, 4). Invasive NTS (iNTS) infections principally occur in sub-Saharan Africa, are life-threatening, and are commonly associated with malnourished infants and the immunocompromised, particularly those infected with HIV. Notably, iNTS disease is generally uncommon outside sub-Saharan Africa, but the disease has recently been described in a comparatively small patient cohort in Southeast Asia (5). The microbiological reservoirs of these two NTS disease presentations are distinct, with organisms causing enterocolitis in humans in industrialized countries primarily thought to arise through the food chain (1). In contrast, the principal source of the organisms causing iNTS in sub-Saharan Africa is thought to be the human population (6).

One of the most common NTS serovars associated with both enterocolitis and iNTS in humans is Salmonella enterica subsp. enterica serovar Typhimurium (S. Typhimurium). This serovar is globally ubiquitous and can be isolated from a range of other animal species. Successive human epidemics of S. Typhimurium have been described over the past several decades, many of which have been caused by variants that exhibit resistance to multiple antimicrobials, including those recommended for clinical care. These epidemic variants are of great concern, as antimicrobial-resistant Salmonella infections are associated with a higher probability of hospitalization and treatment failure, leading to a prolonged infection and increased likelihood of onward transmission (7).

An accurate understanding of how antimicrobial-resistant Salmonella variants emerge and spread is essential for controlling their geographic scope and limiting their potential public health impact. The advent of high-throughput whole-genome sequencing (WGS) and phylogenetics has enabled detailed investigations of the sources and potential transmission routes of S. Typhimurium variants (2, 8). These studies permitted the identification of distinct S. Typhimurium populations found in colocated animals and humans in an industrialized country (8) and outlined the complex phylogeography of an epidemic iNTS-causing S. Typhimurium across sub-Saharan Africa (2).

While there is a good understanding of the S. Typhimurium genomic landscape in Europe and sub-Saharan Africa, such an investigation has not yet been performed extensively for S. Typhimurium originating from Southeast Asia, a known global hot spot for zoonotic disease. Vietnam is a low-middle-income country (LMIC) in Southeast Asia, characterized by widespread human-animal interaction and the excessive use of antimicrobials in humans and agriculture (9, 10). Here, we exploited WGS, phylogenetic reconstruction, and genomic analysis to provide insight into the epidemiology and potential impact of zoonotic transfer and antimicrobial resistance (AMR) in S. Typhimurium and its monophasic variant Salmonella I:4,[5],12:i:− in Vietnam. Additionally, we aimed to define the genetic characteristics of the recently isolated iNTS S. Typhimurium/S. I:4,[5],12:i:− in Vietnam, representing the first such investigation of a novel collection of iNTS organisms isolated outside sub-Saharan Africa.

The genomes of 85 human S. Typhimurium and S. I:4,[5],12:i:− isolates collected between 2008 and 2013 (36 associated with enterocolitis, 41 associated with iNTS, and 8 from asymptomatic carriage) and 113 animal S. Typhimurium and S. I:4,[5],12:i:− isolates collected between 2011 and 2013 (chickens [n = 14], ducks [n = 70], and pigs [n = 29]) (see Table S1 in the supplemental material) in southern Vietnam were sequenced using an Illumina HiSeq2000 sequencer. By mapping the genome sequence reads against the S. Typhimurium SL1344 reference sequence, we were able to reconstruct the phylogenetic relationship between these contemporary Vietnamese isolates from different sources (Fig. 1). Using hierBAPS (11), the isolates clustered into five distinct clades (Table S3). The most striking observation within this initial phylogeny was an apparent lack of mixing between animal and human isolates (P = 0.0005). Notably, organisms in three of the five clades were predominantly associated with a single host species (clade 3, 80% [12/15] of the isolates from humans, D = 0.78, P > 0.10; clades 1 and 5, 95% [56/59] and 75% [12/16] of isolates from ducks, D = 0.73/P > 0.10 and D = 0.44/P > 0.10, respectively). Clades 2 and 4 contained a more comparable number of human and animal S. Typhimurium/S. I:4,[5],12:i:− isolates than clades 1, 3, and 5 (Table S3). There was a significant phylogenetic association with the host species of origin (animal or human) in clades 2 and 4, indicating nonrandom clustering of isolates by host species (clade 2, D = 0.18, P = 0; clade 4, D = −0.06, P = 0.002).

Clade 2 was comprised of isolates belonging solely to multilocus sequence type 34 (ST34). This ST encompasses the European clone associated with the present S. Typhimurium variant pandemic, which is the monophasic S. I:4,[5],12:i:− (12). We expanded our data set with WGS of monophasic and biphasic S. Typhimurium and S. I:4,[5],12:i:− accessed from public databases, which included ST34 isolates from other countries (2, 12–14) (Table S2; Fig. S1) and two new sequences from organisms isolated in Scotland. By doing so, we could demonstrate that the ancestral subclade of the Vietnamese ST34 isolates was the European ST34 S. I:4,[5],12:i:− clone (Fig. 2) (12).

Further interrogation of the phylogenetic structure revealed that the Vietnamese ST34 isolates could be further subdivided into three clear subgroups, which we named ancestral, transitional, and multidrug resistant (MDR), all of which had high bootstrap support values within the phylogenetic tree (Fig. S2). We observed that the majority of organisms within the ancestral ST34 subgroup, comprised of mostly European isolates, were genetically monophasic (Fig. 2). In contrast, there was a cluster of 31 Vietnamese isolates that were predominantly biphasic and further characterized by an extensive complement of AMR genes (MDR subgroup) (Tables S4 and S5). This complement of AMR genes was distinct from the classical AMR gene profile associated with resistance to ampicillin, streptomycin, sulfonamide, and tetracycline (ASSuT), which is typically observed in monophasic ST34 isolates.

We investigated the genomic context of the additional AMR genes in the MDR ST34 subgroup using long-read sequencing data generated using a Pacific Biosciences sequencing system; they were found to be located on a large (~246-kb) IncHI2 plasmid. This plasmid was similar in gene content and structure to plasmid pHXY0908 (accession number KM877269.1), which has been previously described in an S. Typhimurium isolate from chicken feces in China in 2009 (15). The pHXY0908 IncHI2 plasmid carried oqxAB, blmS, sul1, ΔaadA2, dfrA12, aph3, sul3, aadA1a, cmlA2, aadA2, floR, sul2, hph, aac(3′)-IVa, aac(6′)-Ib-cr, blaOXA-1, catB3, and arr3. The predicted phenotype of these organisms was resistance to fluoroquinolones, bleomycin, sulfonamides, trimethoprim, kanamycin, streptomycin, chloramphenicol, spectinomycin, florfenicol, hygromycin B, apramycin, beta-lactams, and rifampin. The six transitional subgroup isolates lay between the ancestral ST34 subgroup and the MDR ST34 subgroup and exhibited some characteristics of both of the other subgroups (Fig. 2). Of these six transitional isolates, five carried the same mercury resistance genes found in the archetypal ST34 monophasic clone, and three carried the bla_TEM-1, strAB, tetB, and sul2 genes, also typical of the monophasic European clone. In contrast, and comparable to the MDR ST34 subgroup, the transitional isolates were also genetically biphasic and carried the MDR IncHI2 plasmid (Table S6).

Two additional features distinguished the isolates in the ST34 MDR subgroup. First, these isolates were significantly associated with iNTS disease in HIV-infected Vietnamese individuals; 73% of human-derived isolates in this subgroup were from the blood of HIV-infected patients in comparison to 32% of the human-derived isolates from other clades (P = 0.001, Table 1). Second, while the MDR subgroup has arisen from monophasic ST34 isolates, the majority (26/31) of these isolates had an intact fljBA operon encoding a phase 2 flagellum and were phenotypically confirmed to be biphasic (Fig. 3). S. Typhimurium typically harbors two flagellin genes, fliC and fljB. These genes are regulated by the hin invertase so that only one flagellar antigen is expressed at any given time (16). In the pandemic ST34 monophasic S. I:4,[5],12:i:− ancestral clone, the fljBA operon has been deleted and replaced by a transposon (IS26-associated) carrying the AMR genes bla_TEM-1, strAB, sul2, and tetBR and mercury resistance genes (12, 17). A fragment of this transposon remained in the majority of the ST34 MDR subgroup isolates, which included the tetBR genes, similarly inserted between hin and iroB, leaving fljBA intact (Fig. 3). While we could not reconstruct the precise sequence of events creating this reversion, the phylogenetic structure, bootstrap branch supports, and DNA sequence identity support that this second flagellar operon was reacquired by the MDR ST34 subgroup isolates from an alternative biphasic S. Typhimurium isolate (Fig. S2).

Last, hypothesizing that these novel iNTS-associated ST34 biphasic organisms were adapted to cause an iNTS disease phenotype in humans (as has been reported for S. Typhimurium associated with iNTS in sub-Saharan Africa), we investigated potential genomic degradation. In a systematic scan for putative pseudogenes, we observed almost no evidence of genomic degradation in the ST34 MDR subgroup (Table S7), unlike iNTS-associated S. Typhimurium in sub-Saharan Africa. The only exception found in all 31 isolates was a frameshift in pduT, which was not present in the remaining Vietnamese ST34 isolates. This gene encodes a hypothetical propanediol utilization protein, and the frameshift would theoretically render this gene inactive.

Our study outlines some important observations for Salmonella epidemiology in a global health context. Vietnam is a rapidly industrializing LMIC where the separation between humans and livestock species is less demarcated than in more developed countries. In this context, we observed either that distinct clades of S. Typhimurium and/or S. I:4,[5],12:i:− were composed predominantly of isolates from a single host species or that isolates from different host species were nonrandomly distributed within the clade. This observation suggests restricted interspecies transmission but should be interpreted with caution due to the limited overlap between sampling periods (humans, 2008 to 2013, and animals, 2011 to 2013). However, we found no significant association between sampling date and phylogeny in a tree containing only human isolates (lambda = 0.092, P =0.21), indicating that isolates did not cluster by year of isolation. Therefore, the observed clustering of isolates by host population was not confounded by the year of isolation.

Previously, it was assumed that the majority of NTS infections in humans arise through the food chain and are ultimately derived from animals (1). However, an increasing number of studies have shown that this scenario is more nuanced. In an industrialized setting, WGS of S. Typhimurium definitive type 104 (DT104) revealed that the major source of human NTS DT104 infections and the AMR of those infections was unlikely to be the local animal population (8). Additionally, a study from Kenya that compared NTS isolates from patients and from animals found that the organisms associated with the environment or food from within or near the homes of the patients were not significantly related to human isolates (18). A further investigation found that NTS isolates from patients were more comparable to NTS organisms obtained from asymptomatic household members than from the environmental or animal samples from the homes of index cases (6). Our work provides insight into the previously limited understanding of the transmission of iNTS and enterocolitis-causing S. Typhimurium and S. I:4,[5],12:i:− in Southeast Asia and identifies trends similar to those observed in other parts of the world.

There have been successive pandemics of S. Typhimurium, including both DT204c and DT104, in recent decades. The current pandemic is caused by the European clone of the monophasic ST34 S. Typhimurium variant, S. I:4,[5],12:i:−. This clone rose to prominence as a major cause of NTS disease in humans in Europe in the 2000s; pigs were identified as the most likely reservoir (19, 20). Since then, this organism has spread globally. ST34 was the second most common ST found within our Vietnamese isolates and is the most common S. Typhimurium ST currently isolated from humans and animals in China (21, 22). The pHXY0908-like plasmid found here within the MDR and transitional subgroups is also epidemic in China (15) and also associated with S. Typhimurium ST34 (21, 22). pHXY0908 additionally carries various metal tolerance genes conferring resistance to tellurite, and IncHI2 plasmids are facilitating the spread of oqxAB and aac(6′)-Ib-cr plasmid-mediated quinolone resistance determinants in NTS organisms. Here, we infrequently found IncHI2 plasmids outside the MDR and transitional ST34 subgroups. The exceptions to this IncHI2 plasmid distribution were a duck isolate from clade 1, a human isolate from clade 4, two Chinese ST19 isolates, and several isolates in the monophasic ST34 subgroup: nine Vietnamese isolates (human bloodstream infections, n = 3; pigs, n = 5; ducks, n = 1), five European isolates, and a Chinese ST34 isolate. However, in comparison to those in the ST34 MDR and the transitional subgroups, the plasmids identified in isolates in other clades carried fewer and/or a different complement of AMR genes (see Tables S5 and S6 in the supplemental material).

Our work describes a novel evolutionary pathway by which an emergent Salmonella, typically associated with noninvasive disease, has exploited an alternative human niche in HIV-infected individuals. This linkage of an enterocolitis-associated Salmonella in developed countries to one causing invasive disease in predominantly immunocompromised individuals in developing countries has been previously observed with Salmonella enterica serovar Enteritidis and non-ST34 S. Typhimurium subtypes (primarily ST313) in sub-Saharan Africa (2, 23). In the case of ST313, these biphasic organisms have undergone genome degradation comparable to that of Salmonella enterica serovar Typhi, acquired AMR genes on a virulence plasmid, and spread systemically in susceptible individuals (24). Here, a monophasic ST34 S. Typhimurium variant, S. I:4,[5],12:i:−, with an international distribution has reacquired a secondary flagellin gene and become associated with invasive disease in HIV-infected individuals in an industrializing country in Southeast Asia. Notably, and unlike ST313 in sub-Saharan Africa, this Vietnamese ST34 variant does not exhibit extensive evidence of genome degradation. The additional flagellin gene may possibly confer a virulence advantage in immunocompromised individuals, but further work exploiting suitable experimental systems, as has been performed for ST313 (3, 4), is required to test this hypothesis robustly. It is likely that the acquisition and maintenance of a broad-range MDR plasmid confer an advantage, due to the sustained prescribing of broad-spectrum antimicrobials to HIV-infected individuals in Vietnam. This series of events combines an evolving Salmonella clone with a global distribution, a pervasive Asian MDR plasmid, and a primary burden of disease in HIV-infected individuals. Although we have identified this new sublineage in Vietnam, it is likely not restricted to Southeast Asia. Recently, two S. Typhimurium ST34 isolates with similar IncHI2 plasmids were identified from two patients in Portugal, with no travel history to Asia and no report of foodborne outbreaks or recent contact with animals (25).

Our data suggest that conditions in Vietnam likely influenced the emergence of a new sublineage of S. Typhimurium. We predict that these conditions may be replicated in comparable LMICs, which may facilitate the emergence of new variants of pathogenic bacteria into human populations. These results demonstrate the incredible genomic plasticity, global mobility, and virulence potential of S. Typhimurium. The international circulation of these organisms combined with their ability to acquire AMR genes and to cause invasive disease in HIV-infected humans highlights the need for improved surveillance of bacterial pathogens in a One Health context. Our study highlights the impact of the global AMR crisis and adds a unique insight into the international epidemiology and emergent variants within Salmonella enterica.

This work was supported by a Wellcome Trust Strategic award (WT/093724); a Sir Henry Dale Fellowship, jointly funded by the Wellcome Trust and the Royal Society (100087/Z/12/Z) to S.B.; Biotechnology and Biological Sciences Research Council Anniversary Future Leader fellowship BB/M014088/1 to A.E.M. at the University of Cambridge and the Quadram Institute Bioscience BBSRC-funded Strategic Programme: Microbes in the Food Chain (project number BB/R012504/1; current funding supporting A.E.M.); an OAK Foundation fellowship through the University of Oxford (OCAY-15-547) to D.P.T.; Wellcome Trust grant 098051 supporting S.C., S.M., D.A.G., J.P., N.R.T., and G.D.; VA5IOBX-000372, NIH AI123640, and a Research Centre of Excellence Grant in Mechanobiology from the Singapore Ministry of Education to L.J.K.; and a Wellcome Trust Intermediate Clinical Fellowship (110085/Z/15/Z) to J.C.-M. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Author contributions were as follows: study design, A.E.M. and S.B.; data analysis, A.E.M., S.B., and L.J.K.; isolate acquisition and processing and clinical data collection: T.L.T.P., Y.G., S.C., S.M., D.A.G., N.T.D.H., H.T.T., N.P.H.L., C.N.T., N.H.T.T., J.C.-M., N.T.T., J.I.C., M.A.R., D.P.T., K.H., N.T.H., N.V.T., C.S., G.G.P., J.E.C., D.J.B., C.O., J.P., N.R.T., N.V.V.C., G.E.T., D.J.M., G.D., L.J.K., and S.B.; manuscript writing, A.E.M. and S.B. All authors contributed to manuscript editing.