Salmonella enterica serovar Typhimurium ST313 responsible for gastroenteritis in the UK are genetically distinct from isolates causing bloodstream infections in Africa

The ST313 sequence type of Salmonella enterica serovar Typhimurium causes invasive non-typhoidal salmonellosis amongst immunocompromised people in sub-Saharan Africa (sSA). Previously, two distinct phylogenetic lineages of ST313 have been described which have rarely been found outside sSA. Following the introduction of routine whole genome sequencing of Salmonella enterica by Public Health England in 2014, we have discovered that 2.7% (79/2888) of S. Typhimurium from patients in England and Wales are ST313. Of these isolates, 59/72 originated from stool and 13/72 were from extra-intestinal sites. The isolation of ST313 from extra-intestinal sites was significantly associated with travel to Africa (OR 12 [95% CI: 3,53]). Phylogenetic analysis revealed previously unsampled diversity of ST313, and distinguished UK-linked isolates causing gastroenteritis from African-associated isolates causing invasive disease. Bayesian evolutionary investigation suggested that the two African lineages diverged from their most recent common ancestors independently, circa 1796 and 1903. The majority of genome degradation of African ST313 lineage 2 is conserved in the UK ST313 lineages and only 10/44 pseudogenes were lineage 2-specific. The African lineages carried a characteristic prophage and antibiotic resistance gene repertoire, suggesting a strong selection pressure for these horizontally-acquired genetic elements in the sSA setting. We identified an ST313 isolate associated with travel to Kenya that carried a chromosomally-located blaCTX-M-15, demonstrating the continual evolution of this sequence type in Africa in response to selection pressure exerted by antibiotic usage. The S. Typhimurium ST313 sequence type has been primarily associated with invasive disease in Africa. Here, we highlight the power of routine whole-genome-sequencing by public health agencies to make epidemiologically-significant deductions that would be missed by conventional microbiological methods. The discovery of ST313 isolates responsible for gastroenteritis in the UK reveals new diversity in this important sequence type. We speculate that the niche specialization of sub-Saharan African ST313 lineages is driven in part by the acquisition of accessory genome elements.

The ST313 sequence type of Salmonella enterica serovar Typhimurium causes 1 invasive non-typhoidal salmonellosis amongst immunocompromised people in sub-2 Saharan Africa (sSA). Previously, two distinct phylogenetic lineages of ST313 have 3 been described which have rarely been found outside sSA. Following the 4 introduction of routine whole genome sequencing of Salmonella enterica by Public 5 Health England in 2014, we have discovered that 2.7% (79/2888) of S. Typhimurium 6 from patients in England and Wales are ST313. Of these isolates, 59/72 originated 7 from stool and 13/72 were from extra-intestinal sites. The isolation of ST313 from 8 extra-intestinal sites was significantly associated with travel to Africa (OR 12 [95% 9 CI: 3,53]). Phylogenetic analysis revealed previously unsampled diversity of ST313, 10 and distinguished UK-linked isolates causing gastroenteritis from African-associated 11 isolates causing invasive disease. Bayesian evolutionary investigation suggested 12 that the two African lineages diverged from their most recent common ancestors 13 independently, circa 1796 and 1903. The majority of genome degradation of African 14 ST313 lineage 2 is conserved in the UK ST313 lineages and only 10/44 15 pseudogenes were lineage 2-specific. The African lineages carried a characteristic 16 prophage and antibiotic resistance gene repertoire, suggesting a strong selection 17 pressure for these horizontally-acquired genetic elements in the sSA setting. We 18 identified an ST313 isolate associated with travel to Kenya that carried a 19 chromosomally-located blaCTX-M-15, demonstrating the continual evolution of this 20 sequence type in Africa in response to selection pressure exerted by antibiotic 21 usage. 22 23 The S. Typhimurium ST313 sequence type has been primarily associated with 24 invasive disease in Africa. Here, we highlight the power of routine whole-genome-25 sequencing by public health agencies to make epidemiologically-significant 26 deductions that would be missed by conventional microbiological methods. The 27 discovery of ST313 isolates responsible for gastroenteritis in the UK reveals new 28 diversity in this important sequence type. We speculate that the niche specialization 29 of sub-Saharan African ST313 lineages is driven in part by the acquisition of 30 accessory genome elements. 31 Introduction 32 33 Serovars of Salmonella enterica cause infections in a diverse range of hosts. In 34 humans, Salmonellae are responsible for a broad range of clinical presentations, from 35 gastroenteritis to invasion of normally sterile compartments such as the bloodstream 36 or brain. Two serovars, Salmonella Typhi and Salmonella Paratyphi A are particularly 37 associated with both human-restricted, and invasive disease. The clinical syndrome 38 caused by these serovars is known as Typhoid or Enteric fever, and this has led to the 39 remaining 2,600 serovars being loosely described as non-typhoidal. By inference, 40 "non-typhoidal" serovars have been considered to be non-invasive in 41 immunocompetent individuals; this crude clinical distinction is misleading (1). 42 43 These "non-typhoidal" Salmonella (NTS) serovars typically have a broad host-range, 44 and the majority of human cases are foodborne, often originating from zoonotic 45 reservoirs (2). Whilst most infections are typically associated with self-limiting 46 gastroenteritis, in a minority (~5%) invasive disease occurs, frequently due to human 47 host immunosuppression, for example advanced HIV infection (3 The clinical distinction between typhoidal and nontyphoidal disease is particularly 54 unhelpful in sub-Saharan Africa (sSA), where non-typhoidal serovars are amongst the 55 most common cause of bloodstream infection, a clinical condition known as invasive 56 nontyphoidal Salmonella (iNTS) disease (1). Whilst the high prevalence of 57 immunosuppressive illness such as HIV and malaria in sSA are clear predisposing 58 factors for the emergence of iNTS disease as a major public health problem, the huge 59 burden of disease has led to further investigation into the serovars responsible for 60 iNTS disease. S. Typhimurium is the serovar most commonly associated with this 61 condition (4).

63
Multi locus sequence typing (MLST) is a molecular approach for typing micro-64 organisms, and uses the allelic varation of seven highly conserved Salmonella 65 housekeeping genes to approximate bacterial phylogeny (5 It has been suggested that the link between S. Typhimurium ST313 and iNTS disease 77 in sSA is that, compared with the generalist S. Typhimurium ST19, ST313 has adapted 78 to an extra-intestinal/invasive lifestyle via genome degradation (6,8). This would be 79 consistent with the finding of an accumulation of pseudogenes in pathways associated 80 with gastrointestinal colonization, as observed in host-restricted Salmonella serovars 81 such as S. Typhi, and in Yersinia pestis, Shigella spp, Mycobacterium leprae and 82 Bordetella pertussis (9)(10)(11)(12)(13)(14). Another observation from comparative genomic studies, 83 which supports the hypothesized enhanced virulence of ST313 includes the detection 84 of novel prophages BTP1 and BTP5 (15) including the reported BTP1 phage-encoded 85 putative virulence gene, st313-td (16). A number of phenotypic characteristics which 86 distinguish ST313 from gastroenteritis-associated ST19 strains have also been 87 described including a reduction in motility, flagellin expression, stationary-phase 88 catalase activity and biofilm formation (17)(18)(19). Despite these phenotypes, it remains 89 to be proven whether ST313 are intrinsically more invasive than ST19. 90 91 Since April 1st 2014, every presumptive Salmonella isolate received by the Salmonella 92 Reference Service (SRS) of Public Health England (PHE) has been whole genome-93 sequenced (WGS) to allow identification, characterization and typing in one laboratory 94 process (5,20).

96
In this study we investigated the prevalence of ST313 in cases of laboratory-confirmed 97 S. Typhimurium infections reported in England and Wales, obtained clinical data 98 regarding the origin of isolates (faeces or blood) and contacted patients to determine 99 whether infection was associated with travel to high-incidence areas such as sSA. We 100 used a phylogenetic approach to place the UK-isolated ST313 into the evolutionary 101 context of the African ST313 lineages. We then investigated the accessory genome, 102 multi-drug resistance (MDR) determinants and the presence of pseudogenes in the 103 UK isolates to shed light on the population structure and evolution of ST313 S. 104 Typhimurium. We also compared key phenotypes of the UK-isolated ST313 with the 105 African ST313 lineage. 106

108
Epidemiology of ST313 in the UK 109 Between January  ST313 isolates submitted to PHE formed a monophyletic group that clustered with 135 previously described African ST313 isolates (7) (Figure 1). A second maximum 136 likelihood ST313 phylogeny was generated with a closely-related ST19 strain also 137 received by PHE (strain U21) as an outlier, to study phylogenetic relationships 138 ( Figure 2A). Of the 76 isolates from distinct patients/sources 12 belonged to the 139 previously described lineage 2, and 64 did not fall within any ST313 lineage that has 140 been reported to date. No lineage 1 isolates were identified. Furthermore, the African 141 associated lineages 1 and 2 did not form a monophyletic group within the novel 142 diversity observed. Both of the African lineages share more recent common 143 ancestors with UK-associated strains than with each other. Neither the food nor the 144 animal isolate belonged to lineage 2. All the UK-derived isolates in this study 145 originated from diagnostic laboratories in England and Wales. To simplify the 146 categorization/differentiation of the lineages for discussion purposes, isolates 147 belonging to lineage 1 and 2 (including those isolated in the UK) will be referred to as 148 African lineages, and the non-lineage 1 and 2 isolates will be referred to as UK 149 ST313. The UK ST313 isolates do not themselves form a coherent monophyletic 150 cluster, revealing an unappreciated level of genetic diversity within ST313 ( Figure 1).

152
To examine the evolutionary history of ST313, a maximum clade credibility tree was 153 inferred using BEAST (Figure 3), and the topology was largely congruent with respect 154 to the Maximum Likelihood tree ( Figure 1, Figure 2 and blaOXA-1, and genes conferring resistance to aminoglycosides, trimethoprim and 201 tetracycline; aac(6')-Ib-cr, dfrA-14, tet(A)-1 (Figure 4a). Isolate U60 also encoded the 202 tellurium heavy metal resistance operon (terBCDEF). Comparison to lineage 2 203 reference strain D23580 identified a putative 29kb deletion in the pSLT-BT virulence 204 plasmid (6), which corresponded to the conjugal transfer region. Additionally, we 205 identified sequence reads that mapped to 97% of the IncHI2 pKST313 plasmid, a 206 novel plasmid which has recently been reported in lineage 2 isolates from Kenya and 207 is known to encode ESBL resistance loci (21). 208 209 More detailed analysis of the genome of isolate U60 showed that a copy of the blaCTX-210 M-15 gene was inserted into the chromosome (location 1648104-1648109 on the 211 D23580 reference genome), disrupting the ompD locus (Figure 4b Supplementary 212 Figure S1). ESBL resistance genes have been reported previously in African ST313 213 isolates carrying plasmids such as pKST313 (21,22) but this is the first report of a 214 chromosomally-encoded ESBL resistance gene in S. Typhimurium ST313.

216
The assembled genomes of the UK-ST313 isolates were compared to the African 217 ST313 reference strain D23580 using BLAST ( Figure 2B). In agreement with 218 published data (4), the majority of the core genome, including the Salmonella 219 Pathogenicity Island (SPI) repertoire was conserved in the ST313 isolates in this study 220 and in 3 ST19 gastroenteritis isolates ( Figure 2B). The African ST313 lineages carry 221 two prophages, BTP1 and BTP5, that are absent from ST19 strains (15). The entire 222 BTP1 and BTP5 prophages were found in most ST313 isolates that belonged to 223 African lineage 2 (12/13), but one UK-isolated lineage 2 strain, U68, lacked both 224 prophages. The complete BTP1 and BTP5 prophages were not identified in any of the 225 UK-ST313 isolates ( Figure 2B), though some isolates contained partial and 226 fragmented identity to BTP1 and BTP5, indicating the presence of related prophages 227 (23, 24) which may not occupy the same attachment site. As expected, the st313-td 228 gene (25) was carried by all twelve lineage 2 strains isolated from the UK that 229 contained prophage BTP1. Only 1/51 UK-ST313 isolates contained the st313-td gene 230 (isolate U76), where it was located on a prophage with 90% identity to BTP1. 231 232 To confirm the conservation of chromosomal organization in the UK ST313 isolates, 233 representative isolate U2 was re-sequenced by PacBio long-read sequencing. The 234 assembly produced two closed circular contigs representing the chromosome and the 235 pSLT virulence plasmid (Supplementary Figure S2). Comparison with the ST313 236 lineage 2 reference genome D23580 identified no large chromosomal re-237 arrangements, deletions or duplications, and confirmed that the BTP1 and BTP5 238 attachment sites were unoccupied and did not contain additional prophages. No 239 additional plasmids larger than the detection limit of the PacBio sequencing (~10kb) 240 were detected in isolate U2.

242
Genome degradation and pseudogenes in UK and African ST313 243 The Several studies have associated the ability of ST313 to cause iNTS disease with 253 particular phenotypic characteristics, such as the lack of RDAR morphotype formation, 254 reduced swimming motility and the inability to produce catalase at stationary 255 phase (18,19,26). We investigated these phenotypic characteristics in the context of 256 the UK-ST313 strains, using a subset of 16 UK-isolated ST313, consisting of 13 UK-257 ST313 isolates and 3 lineage 2 isolates. The phylogenetic context of these 16 isolates 258 is shown in Supplementary Figure S3. Lineage 2 representative strain D23580 and 259 ST19 representative strain 4/74 were included as positive and negative controls.

261
The swimming motility of UK-isolated ST313 was highly variable between isolates 262 (Figure 6a). One lineage 2 strain, U1, showed low levels of motility. However, this 263 strain was observed to have a growth defect (data not shown). The ST313 lineage 2 264 representative strain D23580 was less motile than ST19 strain 4/74, consistent with 265 previous reports (18). However, there was no apparent association between motility 266 (as measured by migration diameter) and phylogenetic context of the lineage 2 strains 267 and the UK-ST313 strains.

269
The katE pseudogene was reported to be responsible for the lack of catalase activity 270 in ST313 lineage 2 (19). All 16 UK-isolated strains were shown to be negative for 271 stationary phase catalase activity, as was the lineage 2 representative strain D23580 272 (Figure 6b). In contrast, ST19 strains 4/74 showed considerable stationary phase 273 catalase activity, consistent with previous findings (19).

275
The RDAR morphotype of Salmonella enterica is linked to resistance to desiccation 276 and exogenous stresses (27). African lineage 2 ST313 are reported to be defective in 277 RDAR morphotype formation due to a truncated BcsG protein due to the introduction 278 of a premature stop codon (19). All the UK-isolated strains and the African lineage 2 279 reference strain D23580 did not exhibit the RDAR morphotype. In our experiments, 280 only the ST19 strain 4/74 exhibited the RDAR morphotype (Figure 6c).

282
These experiments did not identify any phenotypic differences between the UK-ST313 283 isolates and ST313 lineage 2, and future work is needed to identify African ST313-284 specific phenotypes. 285

287
Recent reports of iNTS disease have been associated with the novel Salmonella 288 Typhimurium ST313 in sSA (6,7,16), and suggested that ST313 was geographically 289 restricted to sSA. This prompted the investigation of the presence of this sequence 290 type amongst S. Typhimurium isolates from the UK.

292
We discovered that 2.7% of Salmonella Typhimurium isolates referred to Public Health 293 England are of MLST type ST313, and that this sequence type is heterogeneous in 294 terms of clinical presentation, genomic characteristics and epidemiology. The UK-295 isolated ST313 strains are predominantly fully antimicrobial-susceptible and cause 296 gastroenteritis. We identified a significant association between travel to Africa and 297 infection with the previously described African-associated, ST313 lineage 2. We found 298 two cases of lineage 2 infection where the patient did not report travel to Africa, 299 although one of these patients reported consumption of food from West Africa in 300 London. This indicates that African lineage 2 is predominantly circulating in Africa but 301 may also be circulating in the UK and other countries -potentially via person to person 302 transmission or through exposure to food imported from Africa. People infected with 303 ST313 lineage 2 in the UK were significantly more likely to suffer from invasive disease 304 than patients infected with isolates belonging to UK-ST313 lineages. There is also an 305 increasing number of immune-compromised people in the UK with increasing organ 306 transplants and immune-modulating cancer therapies.

308
This study revealed novel diversity within ST313, which was previously restricted to 309 two African lineages that had exhibited recent clonal expansion (7). Here we place 310 these lineages into an evolutionary context by showing that lineage 1 and 2 do not 311 form a monophyletic group within ST313, which is suggestive of two separate 312 introductions of ST313 into sSA. African lineages 1 and 2 diverged from their MRCA 313 with UK-lineages around 1796 and 1903 respectively. These findings reflect the 314 limitations of classifying bacterial pathogens simply on the basis of sequence type, 315 and show that in the post-genomic era, the resolution offered by MLST may not be 316 sufficient to describe epidemiologically relevant population structures.

318
It has been estimated that 9.2% of cases of Salmonellosis in the EU can be attributed 319 to international travel, and therefore sequencing Salmonella isolated in Europe can 320 provide valuable information regarding the global diversity of Salmonellae associated 321 with human disease (28,29 only been found to be plasmid-associated in Salmonellae. We discovered that the 327 blaCTX-M-15 gene was chromosomally encoded in isolate U60, causing disruption of the 328 ompD locus. This is notable for two reasons. Firstly, chromosomal integration ensures 329 stability of ESBL-resistance even if the plasmid were lost. Secondly, ompD encodes 330 an outer membrane porin that is absent from Salmonella Typhi. OmpD has previously 331 been identified as a highly immunogenic protein (30) and so the disruption of ompD 332 could enhance the reported "stealth" phenotype of ST313 lineage 2 infection (17). We 333 note that OmpD is a potential vaccine target for iNTS (31) and the absence of OmpD 334 from African ST313 populations could have implications for future iNTS vaccine 335 development.

337
Our discovery of UK-ST313 isolates that were not associated with invasive disease 338 provides an excellent opportunity to use comparative genomics to relate genetic 339 findings that have been linked the pathology of lineage 2 ST313 into the context of 340 closely related, gastrointestinal-associated strains. We found that the only genetic 341 characteristics common to both lineages 1 and 2 and absent from the UK-ST313 342 genomes were the BTP1 and BTP5 prophages and plasmid-associated MDR loci. 343 The two African lineages do not share a common ancestor that carried either 344 prophage, suggesting independent acquisition of BTP1 and BTP5 by lineage 1 and 345 2, and whilst the MDR loci confer similar patterns of AMR they are genetically 346 distinct. This implies a strong selection pressure on ST313 in Africa to acquire and 347 maintain these mobile elements, resulting in convergent evolution of the two African 348 lineages. In contrast, there is evidence of an assortment of distinct prophage 349 repertoires in the UK-ST313 isolates, indicating an absence of selection for the 350 conservation of any particular mobile element.

352
Aside from the addition of mobile genetic elements and virulence factors, genome 353 degradation by the accumulation of pseudogenes and deletion events is known to 354 accompany adaption to a more invasive lifestyle (32,33). Initial analysis of the African 355 ST313 representative strain D23580 genome reported 23 pseudogenes compared to 356 6 present in ST19 strain SL1344 (34). Here, we found that the majority of genome 357 degradation found in lineage 2 strain D23580 was conserved in UK representative 358 strain U2. The only pseudogenes associated with characterized genes that were found 359 to be specific to African lineage 2 ST313 were the SPI2-secreted effector gene sseI, 360 lipid A modification gene lpxO and macrolide efflux pump gene macB, each of which 361 could play a role in infection dynamics (35)(36)(37).

363
A number of in vitro phenotypes have been reported for lineage 2 ST313, that could 364 contribute to a host-adapted lifestyle (17)(18)(19)26) and were examined in the UK-ST313. 365 Swimming motility was highly variable amongst the strains tested, and UK-ST313 366 isolates behaved identically to African lineage 2 isolates in the catalase and RDAR 367 morphotype assays. We detected no African-lineage-specific phenotypic 368 characteristics and speculate that reduced motility, defective catalase activity and loss 369 of RDAR formation are not be directly linked to iNTS disease.

371
A key contributing factor to iNTS disease is host immunosuppression and one 372 limitation of this retrospective study was that the underlying health status of the 373 patients was unknown. This study does highlight the extraordinary epidemiological 374 insights that routine genomic surveillance of pathogens by public health agencies can 375 offer, and the ability to understand the pathogenesis of novel pathovars. The 376 knowledge that an immune-compromised patient was infected with lineage 2 ST313 377 could impact clinical decision-making.

379
We have uncovered previously un-sampled diversity in the ST313 clone reflecting the 380 convergent evolution towards niche specialization that has occurred in the African 381 lineages. The routine genomic surveillance of pathogens continues to be adopted 382 internationally and will bring an unprecedented ability to monitor emerging threats. 383 Whole genome sequencing of clinical isolates represents a new window to view the 384 epidemiology and microbiology of infectious diseases. 385

386
Strains and Metadata 387 Genome sequences from a total of 363 Salmonella Typhimurium isolates dating from  388  2012 and 3,014 Salmonella Typhimurium from January 1 st 2014 to March 14th 2016  389 were analyzed for this study. 390 For simplicity, isolates derived from blood culture, the infection was classed as extra-391 intestinal. If only a faecal isolate was received for a patient, the infection was classed 392 as gastrointestinal (though this is only suggestive, not conclusive, data that the 393 infection was restricted to the gastrointestinal tract). Of these, 7/363 (1.9%) and 394 79/3,014 (2.6%) were Salmonella Typhimurium ST313, respectively. Full strain 395 metadata can be found in Table S1. Sequence data (FASTQs) from 23 representative 396 ST313 sequenced by Okoro et al. (7) were downloaded from the European Nucleotide 397 Archive (accessions available in Table S1) and analyzed in the same way as sequence 398 data generated from UK-isolated ST313 isolates. 399 400 Sequencing 401 DNA extraction for Illumina sequencing of Salmonella isolates was carried out using a 402 modified protocol of the Qiasymphony DSP DNA midi kit (Qiagen). In brief, 0.7 ml of 403 an overnight Salmonella culture in a 96 deep well plate was harvested. Bacterial cells 404 were pre-lysed in 220 μl of ATL buffer (Qiagen) and 20 μl Proteinase K (Qiagen), and 405 incubated with shaking for 30 mins at 56°C. Four microliters of RNase (100 mg/ml; 406 Qiagen) was added to the lysed cells and re-incubated for a further 15 minutes at 407 37°C. This step increased the purity of the DNA for downstream sequencing. DNA 408 from the treated cells was then extracted on the Qiasymphony SP platform (Qiagen) 409 and eluted in 100 μl of sterile water. DNA concentration was derived using the GloMax 410 system (Promega) and quality (optimal OD260/230 = 1.8 -2.0) was determined using 411 the LabChip DX system (Perkin Elmer). Extracted DNA was prepared using the 412 NexteraXT sample preparation method, and sequenced with a standard 2x100 base 413 pair protocol on a HiSeq 2500 instrument (Illumina, San Diego). Raw FASTQs were 414 processed with Trimmomatic (38) and bases with a PHRED score of less than 30 415 removed from the trailing end.

417
PacBio sequencing was performed on the PacBio RS II instrument at the Centre for 418 Genomic Research, University of Liverpool. DNA was extracted from strain U2 using 419 the Zymo Research Quick-DNA TM Universal Kit (cat# D4069) as per the Biological 420 Fluids & Cells protocol. Extracted DNA was purified with Ampure beads (Agencourt) 421 and the quantity and quality was assessed by Nanodrop and Qubit assays. In addition, 422 the Fragment analyzer (VH Bio), was used to determine the average size of the DNA, 423 using a high sensitivity genomic kit. DNA was sheared to approximately 10kb using a 424 Covaris g-tube and spinning at 5400rpm in an Eppendorf centrifuge. The size range 425 was checked on the Fragment Analyzer. DNA was treated with exonuclease V11 at 426 37°C for 15 minutes. The ends of the DNA were repaired as described by Pacific 427 Biosciences. Samples were incubated for 20 minutes at 37°C with damage repair mix 428 supplied in the SMRTbell library kit (Pacific Biosciences). This was followed by a 5-429 minute incubation at 25°C with end-repair mix. DNA was cleaned using 1:1 volume 430 ratio of Ampure beads and 70% ethanol washes. DNA was ligated to adapters 431 overnight at 25°C. Ligation was terminated by incubation at 65°C for 10 minutes 432 followed by exonuclease treatment for 1 hour at 37°C. The SMRTbell library was 433 purified with 1:1 volume ratio of Ampure beads. The library was size-selected on the 434 Blue Pippin (Sage) in the range 7kb-20kb. The DNA was recovered and the quantity 435 of library and therefore the recovery was determined by Qubit assay and the average 436 fragment size determined by Fragment Analyser. SMRTbell libraries were annealed 437 to sequencing primer at values predetermined by the Binding Calculator (Pacific 438 Biosciences) and complexes made with the DNA polymerase (P6/C4 chemistry). The 439 complexes were bound to Magbeads and loaded onto 3 SMRT cells. Sequencing was 440 done using 360-minute movie times. Sequence data from the 3 SMRT cells was 441 assembled using the HGAP3/Quiver assembler. This resulted in 2 contigs 442 representing the chromosome and the pSLT virulence plasmid. Terminal repeats were 443 manually trimmed to represent circular molecules, and the chromosome assembly was 444 reordered so that the sequence started at the thrL locus in accordance with convention 445 for Salmonella finished genomes. The closed sequences for the U2 chromosome and 446 pSLT virulence plasmid were 4,811,399bp and 93,862bp respectively. Prokka (39) 447 was used to annotate the two sequences, using the -force flag to preferentially 448 annotate CDS from reference databases FN424405 for the chromosome and 449 AE006471 for the virulence plasmid. The finished U2 genome and annotation were 450 submitted to Genbank and can be accessed using the study number PRJEB20926.

452
Genomic analysis 453 The multi-locus sequence type (ST) was determined using a modified version of SRST 454 (40). For phylogenetic analysis, processed sequence reads were mapped to the S. 455 Typhimurium LT2 reference genome (GenBank: AE006468) using BWA mem (41). 456 SNPs were called using GATK2 (42) in unified genotyper mode. Core genome 457 positions that had a high quality SNP (>90% consensus, minimum depth 10x, MQ 458 >=30) in at least one strain were extracted and IQ-TREE with parameters -m TEST -459 bb 1000 was used to construct a maximum likelihood phylogeny (43).

461
To examine the evolutionary history of ST313, four timed phylogenies were 462 constructed using BEAST v1.8.0 (44), with varying clock rate models and tree priors.

463
The resulting models were compared in terms of their tree likelihood and posterior and 464 the strict exponential and strict constant models were found to be superior. A 465 comparison using AICM calculated with Tracer v1.6.0 showed that the models had 466 very similar values, tree topologies and branching support in terms of posterior 467 probability were similar between the models. The 95% HPD for the exponential growth 468 rate estimate was -0.0026 to 0.0006; the strict, constant growth model was selected 469 as the estimate of growth rate from the exponential model was around 0 (i.e. constant).

478
Microbiology 479 Phenotypic antimicrobial susceptibility testing was carried out for all UK-isolated 480 ST313 strains. The antimicrobial susceptibility testing was done using breakpoint 481 concentrations. Briefly, an agar dilution method involving Iso-sensitest agar or Muller-482 Hinton agar was used to determine if isolates were sensitive or resistant to a set 483 concentration of individual antimicrobials (Supplementary Table S2).

485
Swimming motility assays were performed based on methods previously described 486 (48). A 3µl aliquot of bacteria grown overnight in LB (Lennox Broth; 10 g/L Bacto 487 Tryptone, 5 g/L yeast extract and 5 g/L NaCl, pH7.0) was spotted onto LB (Lennox) 488 plates containing 0.3% Bacto Agar (Difco). Plates were incubated at 37°C. After 489 exactly 5 hours the migration diameter was measured and plates were photographed. 490 491 Catalase activity and RDAR morphotypes were assayed based on methods used by 492 Singletary et al. (19). Food poisoning is a notifiable disease in the UK and diagnostic laboratories are 503 obliged to report the isolation of Salmonella from human clinical diagnostic samples. 504 However, data are frequently incomplete and detailed exposure information for cases 505 is not always available in retrospect. Therefore, targeted surveillance questionnaires 506 were attempted to obtain enhanced information, focusing primarily on collection of 507 information on clinical severity of disease, travel history and consumption of foods of 508 African origin were utilized during telephone interviews for cases reported from 2014-509 2016 to collate relevant epidemiological data. Cases for which enhanced information 510 were available are shown in Table S1. 511 512 Collection of this epidemiological data was not attempted for the 2012 cases, but 513 limited travel data had been recorded on the SRS Salmonella surveillance database 514 or some isolates. It is important to emphasize that the travel information for the 2012 515 isolates is of low quality, and the absence of reported travel does not mean that 516 international travel had not occurred. Odds ratios were calculated using the 517 medcalc.org website https://www.medcalc.org/calc/odds_ratio.php. 518