Figure 1

Figure 1. Demographic reconstruction and phylogenetic analysis of Salmonella enterica subspecies enterica4,[5],12:i:- sequence type 34 isolates. A) Demographic reconstruction (subset 2) shows the population exponential growth...

Figure 2

Figure 2. Summary of posterior estimates of all 10 subsets of sequences of Salmonella enterica subspecies enterica4,[5],12:i:- sequence type (ST) 34 collected from multiple sources in...

We used publicly available whole-genome sequences of 1,431 Salmonella 4,[5],12:i:- ST34 isolates from the United States and Europe from 2008–2017, including sequences from 690 isolates from Europe (mainly from the United Kingdom and Denmark) and 741 isolates from multiple US states (Appendix 1). We used BEAST version 1.8.4 (5) to estimate divergence times, mutation rates, and location trait transitions. We applied the modeling approach to 10 subsets of 112 sequences selected from the study population. These sequences represented 33% (474/1,431) of the study population and included 242 sequences from Europe (76% from humans, 8% from food products, 8% from livestock, and 8% from other sources) and 232 from the United States (62% from humans, 13% from food products, 21% from livestock, and 3% from other sources). Time-scaled phylogenies of each subset were reconstructed by using a general time-reversible nucleotide substitution model, an uncorrelated lognormal relaxed molecular clock, and an exponential growth coalescent model with asymmetric trait transitions (Figure 1; Appendix 2 Figures 1–10). All time-scaled phylogenies presented similar topology to a maximum-likelihood phylogeny constructed by using all 1,431 study isolates (based on visual inspection; Appendix 2 Figure 11). Overall, averaged estimates from all subsets were in agreement as follows (Figure 2): the evolutionary rate was 3.64 × 10⁻⁷ substitutions/site/year (95% highest posterior density [HPD] 2.65–4.64 × 10⁻⁷), which corresponds to an accumulation of »1–2 single-nucleotide polymorphisms/genome/year; the time to most recent common ancestor was 1994 (95% HPD 1988–2000); the number of collection location state transitions (Markov jumps) from Europe to the United States was 7.7 (95% HPD 5.9–9.3) and from the United States to Europe was 0.8 (95% HPD 0–2.2); and the waiting times (in years; Markov rewards) were 519.9 (95% HPD 393.1–667.8) in Europe and 318.6 (95% HPD 234.0–417.6) in the United States. The exponential growth rate of the population was estimated at 0.316/year (averaged across all subset means; Figure 1). In addition, the estimated (averaged) main introduction into the United States was 2004 (95% HPD 2000–2006; Appendix 2 Table 1). The occurrence of several additional smaller introductions was suggested by 48 sequences (6 from Europe and 42 from the United States). The 6 sequences from Europe were recovered from human sources; among sequences from the United States, 28 were from humans, 5 from food products, 6 from livestock, and 3 from other sources. Information on recent international travel was obtained for 22/28 of US isolates from humans, 2 of whom had traveled (1 to the Philippines and the other to France [S. Meyer et al., Minnesota Department of Health, pers. comm., 2019 Sep 23]).

Among the 1,431 Salmonella 4,[5],12:i:- ST34 sequences, 978 (68.34%) had genetic determinants contributing to the ASSuT profile, 108 (7.55%) conferred resistance to quinolones, and 82 (5.73%) conferred resistance to extended-spectrum cephalosporins. The probability of harboring most predominant acquired antimicrobial resistance genes (AARGs) conferring the resistance phenotypes described above was significantly higher for sequences of US isolates (odds ratio 2.37–26.05; Table). Yet associations between the collection location and the presence of bla_CTX-M or qnrS1 genes were not significant (Table). In addition, AARGs conferring resistance to colistin (mcr-1/mcr-3/mcr-5; Appendix 1) were detected in isolates from Europe only (n = 5).

Salmonella 4,[5],12:i:- ST34 was introduced into the United States from Europe on multiple occasions since the beginning of the 21st century. The main introduction occurred in 2004; additional independent introductions resulted in small clades for which the predominant sources were human travelers and imported food products. Human travelers (6) and imported food products (7) have been described as potential vehicles for introduction of salmonellae.

The date of introduction of the main clade into the United States is later than the first peer-reviewed report of a Salmonella 4,[5],12:i:- infection in the country in 1998 (2). However, given the antimicrobial susceptibility profile of isolates from that report (mostly not ASSuT) (2), they most likely belonged to the nonemerging ST19 clade, which was described elsewhere (4). In addition, the incidence of Salmonella 4,[5],12:i:- in humans increased only modestly (9.5%) during 2006–2011 but increased dramatically (78.3%) during 2011–2016 (8). A similar increase in detection after 2011 was described for clinical cases in swine from the midwestern United States (9). The difference between the date of main introduction into the United States found in this study and the later sharp increase in its prevalence in animals and humans may in part result from changes in reporting practices and increasing awareness (8). However, the increase since 2011 can be the result of rapid propagation of the ST34 population, possibly associated with swine (4). Moreover, White et al. (10) recently reported that according to the National Antibiotic Resistance Monitoring System, the percentage of ASSuT-resistant Salmonella 4,[5],12:i:- from humans increased from 17% in 2009 to 59.1% in 2015 (out of all Salmonella 4,[5],12:i:- clinical isolates from humans). This increase probably resulted to a large extent from ST34 strains, in which this phenotype is predominant. The estimated exponential yearly growth rate determined in our model (0.316/year), which corresponds to a population doubling time of 2.2 years, is in agreement with this dramatic increase of the ST34 population.

The presence of AARGs conferring resistance to quinolones and extended-spectrum cephalosporins has mainly been observed since 2014 and may be biased by the lack of sequences before 2013 (Appendix 2 Figure 12). Yet AARGs conferring resistance to quinolones were not found in Salmonella 4,[5],12:i:- ST34 strains from Europe collected before 2010 (1), and therefore our findings may reflect an increasing prevalence of these resistance determinants. Given time and overall unidirectionality of Salmonella 4,[5],12:i:- ST34 transmission from Europe to the United States, it is likely that the acquisition of AARGs to quinolones occurred independently in the United States and in Europe. Yet introduction of resistant strains from the United States to Europe is also possible. Contributors to the acquisition of resistance in the United States might be the approval for enrofloxacin use in swine in the United States since 2008 (11) and the potential dissemination of plasmids harboring AARGs to quinolones between Salmonella serotypes (12). Independent acquisition of resistance to quinolones by Salmonella in Asia has also been suggested (13). The presence of mcr resistance genes conferring resistance to colistin in sequences from Europe (n = 5) is alarming, given their recent worldwide spread (14). However, further investigation of the travel history associated with these cases may be required because the acquisition of mcr genes may be travel associated (15). The spread of Salmonella 4,[5],12:i:- ST34 from Europe to the United States and the presence of plasmid-mediated resistance genes to key antimicrobial classes such as quinolones, extended-spectrum cephalosporins, and colistin in this clade further highlights its potential risk to public health and emphasizes the need for robust surveillance and mitigation programs for such transboundary pathogens.