Urban rats as carriers of invasive Salmonella Typhimurium sequence type 313, Kisangani, Democratic Republic of Congo

Background Invasive non-typhoidal Salmonella (iNTS–mainly serotypes Enteritidis and Typhimurium) are major causes of bloodstream infections in children in sub-Saharan Africa, but their reservoir remains unknown. We assessed iNTS carriage in rats in an urban setting endemic for iNTS carriage and compared genetic profiles of iNTS from rats with those isolated from humans. Methodology/Principal findings From April 2016 to December 2018, rats were trapped in five marketplaces and a slaughterhouse in Kisangani, Democratic Republic of the Congo. After euthanasia, blood, liver, spleen, and rectal content were cultured for Salmonella. Genetic relatedness between iNTS from rats and humans—obtained from blood cultures at Kisangani University Hospital—was assessed with multilocus variable-number tandem repeat (VNTR) analysis (MLVA), multilocus sequence typing (MLST) and core-genome MLST (cgMLST). 1650 live-capture traps yielded 566 (34.3%) rats (95.6% Rattus norvegicus, 4.4% Rattus rattus); 46 (8.1%) of them carried Salmonella, of which 13 had more than one serotype. The most common serotypes were II.42:r:- (n = 18 rats), Kapemba (n = 12), Weltevreden and Typhimurium (n = 10, each), and Dublin (n = 8). Salmonella Typhimurium belonged to MLST ST19 (n = 7 rats) and the invasive ST313 (n = 3, isolated from deep organs but not from rectal content). Sixteen human S. Typhimurium isolates (all ST313) were available for comparison: MLVA and cgMLST revealed two distinct rat-human clusters involving both six human isolates, respectively, i.e. in total 12/16 human ST313 isolates. All ST313 Typhimurium isolates from rats and humans clustered with the ST313 Lineage 2 isolates and most were multidrug resistant; the remaining isolates from rats including S. Typhimurium ST19 were pan-susceptible. Conclusion The present study provides evidence of urban rats as potential reservoirs of S. Typhimurium ST313 in an iNTS endemic area in sub-Saharan Africa.

a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 human S. Typhimurium isolates (all ST313) were available for comparison: MLVA and cgMLST revealed two distinct rat-human clusters involving both six human isolates, respectively, i.e. in total 12/16 human ST313 isolates. All ST313 Typhimurium isolates from rats and humans clustered with the ST313 Lineage 2 isolates and most were multidrug resistant; the remaining isolates from rats including S. Typhimurium ST19 were pan-susceptible.

Conclusion
The present study provides evidence of urban rats as potential reservoirs of S. Typhimurium ST313 in an iNTS endemic area in sub-Saharan Africa.

Author summary
Dadi (˚1974, DR Congo) is a Medical Doctor (Kisangani University 2005) with a master in pediatrics (Kisangani University 2015) with special interest in infectious diseases and tropical medicine. He has 11 years of field research experience. He was team member of scientific expedition "Boyekoli Ebale Congo" in 2010 as medical support for the researchers. He conducted work field in a multidisciplinary framework with biologists from faculty of sciences (university of Kisangani) exploring zoonotic diseases in several places in the Congo. Currently, he is doing his PhD research at KU Leuven and Institute of Tropical Medicine Antwerp (ITM), Belgium. Passionate about transmissible diseases, Dadi is exploring the potential reservoirs of non-typhoidal Salmonella in Kisangani, DR Congo.
In high-income countries, the reservoir of non-typhoidal Salmonella is zoonotic and transmission is mainly foodborne [10,13,14]. In contrast, iNTS infections in sub-Saharan Africa are believed to be anthroponotic [7,15,16]. In line with Salmonella Typhi and Paratyphi A causing enteric fever in humans [17], invasive S. Typhimurium and Enteritidis have genetically adapted to humans [2,11]. These invasive NTS are clinically specific: they are mainly causing bloodstream infections (and not diarrhea) and are associated with high mortality [18,19]. Further, household studies of index patients with NTS bloodstream infections in Kenya and Burkina-Faso have demonstrated NTS stool excretion in household members but not in household livestock [15,20]. However, reviewing the evidence of source attribution, several authors stated that an animal reservoir for NTS in sub-Saharan Africa is not excluded and needs further study [21][22][23].
The Democratic Republic of Congo (DR Congo) is highly endemic for iNTS disease. The National Institute of Biomedical Research (INRB) in Kinshasa coordinates, in collaboration with the Institute of Tropical Medicine (ITM) in Antwerp, Belgium, a nationwide microbiological surveillance network processing blood cultures in patients suspected of bloodstream infections. Kisangani University Hospital participates-through hospitals and health centers within and around Kisangani-in this network [24]. Over the past decade, NTS consistently represented the most frequent blood culture isolates in children attending the hospitals and health centers in this network, with S. Typhimurium and Enteritidis isolates representing respectively 63.1% and 36.4% of iNTS isolates of which 87.4% displaying MDR [19].
As a fast-growing densely populated metropolis, Kisangani provides a wide array of ecological niches attractive for rodents. Previously, Kisangani University had assessed rats trapped in marketplaces in Kisangani as reservoirs of Rickettsia typhi and Bartonella spp. [25]. Rodents have been described as a reservoir and vector of many pathogens including Salmonella [26] but to our knowledge, rats have not yet been assessed as a potential reservoir of NTS. Therefore, we aimed to assess rats present in marketplaces in Kisangani for carriage of NTS. Our objectives were to estimate the proportion of S. Typhimurium and Enteritidis carriers (rats with S. Typhimurium or Enteritidis isolates from liver, spleen, blood, and/or intestinal content) among rats, and to assess genetic relatedness between rat and human NTS isolates. A secondary objective was to assess the antimicrobial resistance profile of the Salmonella isolates.

Ethics statement
Ethical approval for the blood culture surveillance study was granted by the Institute of Tropical Medicine Antwerp Institutional Review Board (approval number: 613/08), the Ethics committee of Antwerp University (approval number: 8/20/96) and the Ministry of Health of the Democratic Republic of the Congo (approval number: 8/20/96). For the rat carrier study, authorization by the health authorities of Tshopo Province (Division Provincial de la Santé) was obtained on 12 May 2016 and procedures were carried out according to the 2016 Guidelines of the American Society of Mammologists for the use of wild mammals in research and education [27]. Before starting rat trapping, the responsible in charge of security at the trapping site was contacted for permission and surveillance of the traps.

Study design
In an exploratory study, rats trapped alive in marketplaces in the Kisangani area were sampled for Salmonella in the microbiological laboratory at the University Hospital of Kisangani. Salmonella isolates were tested for serotype and antimicrobial susceptibility. S. Typhimurium and Enteritidis isolates were compared by molecular typing for genetic relatedness with clinical isolates obtained from blood cultures.

Rat sampling
2.3.1 Sample size calculation. Based on CLSI guidelines, a minimum sample size of 30 Salmonella spp. isolates was set for reliable data about serotype distribution and antimicrobial susceptibility profiles [28]. At an assumed NTS carrier ratio of 5-10% [29][30][31][32] we targeted to capture a total of 600 rats to have sample size of 30 rats carrying Salmonella. sterile instruments) were collected from the spleen and subsequently from the liver. From May 2018 onwards, additional blood (0.5-1ml) was sampled by intracardiac puncture, before sampling of spleen and liver. At the end of the procedure, the rectal content was collected. Internal organs collected were transported to the lab for preliminary analysis within 4 hours.
The samples of spleen, liver and rectal content were placed in a sterile Petri dish, sliced, and transferred into 10ml Selenite broth (Becton Dickinson and Company, Franklin Lakes, NJ, US) followed by vortex mixing. Blood was inoculated directly into the Selenite broth tube. Selenite broth tubes were transported to the microbiology laboratory within 4 hours after inoculation.

Salmonella identification, serotyping, and antibiotic susceptibility testing in rats
Laboratory processing was done as described previously for stool samples [35]. Selenite broth tubes were incubated at 35˚C for 12-18 hours, whereafter 10 μl broth was inoculated on two Salmonella-Shigella (SS) agar plates (Lab M Limited, Lancashire, UK). These were incubated at 35˚C and read after 18-24h. In case of no growth, the plates were evaluated after another 24 hours of incubation. In case of growth, up to five (to detect the presence of multiple seroptypes) colonies suspected of Salmonella were transferred to Kligler Iron Agar (KIA) tubes (Lab M Limited) and incubated for 18-24 hours at 35˚C. Bacterial growth in the KIA tubes suggestive of Salmonella was biochemically confirmed by a panel of disk-based biochemical tests (DiaTabs, Rosco, Taastrup, Denmark). Isolates with a reaction pattern compatible with Salmonella were stored in 2 ml tubes of Trypticase Soy Agar (Oxoid, Basingstoke, UK) and shipped to the Institute of Tropical Medicine (ITM, Antwerp, Belgium) for serotyping and antibiotic susceptibility testing (AST).

Human sampling
Salmonella isolates were obtained and stored as part of this INRB-ITM microbiological blood culture surveillance network at Kisangani University Hospital as previously described [24]. Basic clinical and demographic data were recorded. To assess the genetic relatedness between Salmonella isolates from rats and humans, blood culture isolates obtained during the rat trapping period and six months before and 6 months after this period (01 October 2015-30 June 2019) were selected. Serotyping and AST of blood culture Salmonella isolates were done as described previously [19].
2.6.1 Genetic relatedness between Salmonella isolates from rats and humans. Genetic relatedness between the iNTS isolates (in case those belonging to S. Typhimurium serotype, see Results) was assessed by three different approaches: multilocus variable-number tandem repeat (VNTR) analysis (MLVA), MLST, and cgMLST.
For WGS, the 7-genes MLST scheme was applied, based on the sequences of the seven housekeeping genes: aroC, dnaN, hemD, hisD, purE, sucA, thrA [39,40]. Hierarchical clustering on cgMLST (HierCC) and minimum spanning trees (MS-trees) was performed on EnteroBase based on the sequences of 3002 loci for Salmonella [41]. On the HierCC scheme, HC2850 values allow to deduct the Salmonella subspecies (i.e. difference of � 150 or less alleles from the 3002 loci in the cgMLST scheme); HC200 values provide a similar discrimination as the 7-genes MLST typing (� 200 alleles difference between genomes); and HC5 values can be considered a strong indicator of epidemiological relatedness (i.e. if no more than five alleles are different) [41,42].
2.6.2 Genetic relatedness between isolates from this study and the literature. We compared some of the isolates of S. Typhimurium and S. Enteritidis identified in this study with previous studies on iNTS in Africa [7,43] using EnteroBase. This database recovers regularly Salmonella genomes from the GenBank Sequence Read Archive (SRA). The "Custom View" utility allows to add metadata for genomes already present on the site. EnteroBase has recovered 564 genomes from the study on S. Typhimurium ST313 by Pulford et al. [7] and 503 genomes from the study on S. Enteritidis ST11 by Feasey et al. [43]. A "Custom View" was created to add the information on lineages and plasmid repertoire for the S. Typhimurium ST313 from the Pulford et al. study [7], that was next compared to the HierCC clustering on cgMLST by EnteroBase [41]. A "Custom View" previously created, including the hierBAPS clade/cluster data described by Feasey et al. [37] was used to compare the hierBAPS clustering with the EnteroBase HierCC clustering. A minimum spanning (MS) tree (MStree V2 or GrapeTree) based on the EnteroBase "cgMLST V2 + HierCC V1" scheme was produced for each serotype to estimate the allelic distances between isolates from this study and the genomes from Pulford et al. [7] and Feasey et al. [43].
For the S. Enteritidis ST11 in this study, plasmid pSEN-BT (accession: LN879484.1) described by Feasey et al. [43] as characteristic of the hierBAPS clade 9, also denominated Central/Eastern African clade was also searched and the information was included in the S. Enteritidis ST11 "Custom View".

Data management and analysis, definitions
Geographic coordinates, morphometric measurements of rats and microbiological data were entered in an Excel database (Microsoft, Redmond, WA, USA). Rats with at least one isolate confirmed as Salmonella were categorized as Salmonella carriers. Several rats carried more than one Salmonella serotype. Only the first isolate per serotype per rat was used for the analysis of the distribution of serotypes and AST. For the display of MLVA typing results, all Salmonella Typhimurium isolates were assessed but only the first isolate per MLVA type per rat was presented. All genomes obtained by WGS and analyzed on EnteroBase are available at the ENA depository under project number PRJEB54047 (S1 Table).

Rats sampling
During 55 outreach visits, 566 (34.3%) of 1,650 traps yielded 566 rats including 541 (95.6%) Rattus norvegicus (Norwegian or brown rat) and 25 (4.4%) Rattus rattus (roof rat or black rat). The highest trapping success was seen in Central and Yaté marketplaces (30.4% and 48.7% respectively). There were no incidents (e.g. tripping) with the traps reported and apart from one cat, no other non-rat species was captured. A total of 46 (8.1%) rats carried Salmonella spp. (Table 1). All captured rats were alert and active, and none showed signs of illness and weakness. Most rats (89.0%, 504/566) were sexually mature, with a male-to-female ratio of 0.91; median (Interquartile Range, IQR) length (without tail) and mass were 22.1 cm (8.3-28.4 cm) and 280 g (30-515 g) respectively; showing no difference between Salmonella carriers and non-carriers (S2 Table).

Distribution of serotypes and specimens in rats
The 46 Salmonella rat carriers yielded a total of 254 Salmonella isolates, of which 253 were available for further analysis, one isolate was not viable after transport. Table 2 lists the Salmonella serotypes isolated per body specimen and rat. Thirteen rats yielded multiple serotypes and a total of 73 non-duplicate serotypes isolates were recovered. Most common serotypes were serotype II.42:r:-(found in 18 rats), Kapemba (12 rats), Weltevreden and Typhimurium (10 rats each), and Dublin (found in 8 rats). All Salmonella Dublin non-duplicate serotype isolates (n = 11) and all but one S. Typhimurium isolates (n = 12 in 10 rats) were obtained exclusively from deep organ samples (spleen or liver). S. Enteritidis was not isolated from rats.

S. Typhimurium isolates
Results of MLVA and MLST typing of S. Typhimurium isolates from rats (12 isolates from 10 rats) and humans (18 isolates from 18 patients available for MLVA typing, 16 isolates from 16 patients available for MLST typing) are listed in Table 3. Fig 2A and 2B depicts the clustering of those 32 isolates through a minimum spanning tree on cgMLST. The 18 human S. Typhimurium isolates comprised nine different MLVA types and all 16 sequenced isolates belonged to MLST ST313. Among the 12 S. Typhimurium isolates recovered from 10 rats, four isolates (from three rats) belonged to ST313. The rat ST313 isolates represented 6.5% of Salmonella-carrying rats and 0.5% of all rats captured. Rat ST313 isolates belonged to three MLVA types. The remaining eight S. Typhimurium isolates from rats belonged to ST19 (four MLVA types, seven rats). The three rats carrying Salmonella ST313 did not show signs of illness and had length and mass in line with all the other rats in the study (S2 Table).
Two human-rat clusters were identified among the ST313 isolates, involving the three ST313 carrying rats and 12 out of 16 (75.0%) human isolates (Table 3, Fig 2A and 2B). In the first cluster, two rat isolates (Yate Market, October 2018) grouped with six human isolates obtained between May 2016 and September 2018 (28 months) from patients living between in a perimeter of 3.5 to 3.7 km from the marketplace. In the second cluster, two isolates from one rat (Central Market, October 2016) grouped with six human isolates obtained between August 2016 and May 2017 (8 months) from patients living within a perimeter of 1.2 to 2.9 km from the marketplace.
The ST313 S. Typhimurium human and rat isolates clustered with the ST313 Lineage 2 isolates described by Pulford et al. [7], at less than 50 alleles (by the EnteroBase cgMLST scheme), and at more than 200 alleles to the Lineage 1 and Lineage 3 isolates in their study (Fig 3). As to the plasmid repertoire described by Pulford et al. [7], there were two distinct profiles: pBT2 and pBT3 (n = 6 humans and n = 1 rat) and pBT3 alone (n = 7 humans and n = 2 rats) (Fig 4). The ten ST19 S. Typhimurium isolates (all eight rats captured at Yaté and Central markets) clustered both by MLVA and HC5 (Fig 5).  clustered at less than 50 alleles to the hierBAPS clade 9, also denominated Central/Eastern African clade (S2 Fig), and carried the expected plasmid pSENT-BT.

Discussion
In a sub-Saharan Africa setting endemic for iNTS, the present study showed a Salmonella carrier ratio of 8.1% among rats captured in an urban area. Among the serotypes recovered, there

Rats as reservoirs of non-Typhoidal Salmonella
Rodents-in particular rats-are well-known reservoirs of zoonotic pathogens including Salmonella [26]. Prevalence of Salmonella in rats has mainly been assessed in high-income countries and varied widely: in the natural environment, prevalence was consistently low (0 to 1%) [26,44,45] but is was higher in the proximity of human settlements (5-20%) such as cities and . Minimum spanning tree (MSTree V2) comparing the S. Typhimurium ST313 genomes analyzed by Pulford et al. [7] in 2021 and the genomes in this study. The ST313 genomes in this study (circled in red, filled in white) cluster together with the Lineage 2 strains in the study by Pulford et al. [7], colored in green). In the above-mentioned studies, Salmonella serotypes from commensal rats varied widely but there was a consistent overlap with human epidemiologically relevant serotypes. Examples are Salmonella Enteritidis and Infantis from layer farms [26,47,55] and Salmonella Weltevreden from rats in marketplaces in Thailand [54]. S. Typhimurium and-to a lesser extent-Enteritidis belong to the most observed serotypes [29,31,32,47,48,52,54,56]. Molecular typing allowing for sequence typing of the non-Typhoidal Salmonella was not performed so far. The aforementioned study in Japan showed that S. Typhimurium isolates obtained from urban rats  belonged to the DT104 phage type, i.e., the world-wide distributed zoonotic S. Typhimurium causing diarrhea in humans [32].

Rats carried S. Typhimurium ST313, the predominant S. Typhimurium clone in Africa
The 8.1% Salmonella prevalence among rats in the present study fits into the prevalence ranges described above. Among the serotypes recovered, the main finding was the presence of S. Typhimurium ST313 in three rats. By MLVA as well as by comparison of genomes in Entero-Base, the ST313 isolates grouped in two clusters comprising each six human ST313 isolates recovered from blood cultures. ST313 is the main pathovariant of iNTS in Africa [7]. Both rat and human isolates clustered together with what Pulford et al. describe as Lineage 2, i.e. the most common and recent S. Typhimurium ST313 lineage in the African continent [7]. Close genetic relation was also supported by the identification of the pSLT-BT plasmid, encoding for the different AMR genes that most ST313 isolates from rats and humans in the present study harbored, particularly the cat1 gene encoding for chloramphenicol resistance, which is hallmark for Lineage 2 [7].

Finding of Salmonella Typhimurium ST313 in rats: laboratory animal models
So far, reservoir and transmission of iNTS including ST313 were hypothesized to be human, based on the genetic signature [4,5,22] and household, case-control, and environmental studies [15,16,20]. Finding ST313 in rats indicates that ST313 is not restricted to the human host. Previous laboratory animal models showed that ST313 was able to cause invasive infections in chickens, mice, and rhesus macaque infection models [6,[57][58][59]. Findings differed according to the challenging strain and host animal but overall ST313 caused, in comparison with ST19, a more rapid onset of bacteremia and infestation of liver and spleen combined with less diarrhea and colonization of the intestinal tract. The fact that in the present study ST313 in rats was exclusively recovered from deep organs may point to an invasive disease too, but none of the rats captured was visibly sick and their length and mass were in line with the other rats. Moreover, although S. Typhimurium strains can cause deadly infections, asymptomatic infections of liver and spleen with intermittent fecal shedding can occur too [29,60,61] and may even constitute the majority of infections [62].

Attribution of rats to Salmonella transmission
Despite being acknowledged as a reservoir for Salmonella, there is only circumstantial and anecdotal evidence as to the attribution of rats to the transmission of Salmonella [29,61,63]. Rather than being an original reservoir for Salmonella, rats are considered to be a "sponge" reflecting the environment they are living in [26,45,51]. This has been demonstrated in layer farms, where proportions of Salmonella carrying rats were related to the percentage of Salmonella contaminated eggs and the intensity of the environmental contamination [47].
Rats shed Salmonella for 2-4 months after infection and have up to 40 droppings a day [29]; Salmonella survives long (up to 86 days) in rat droppings [30]. In this way, rats contaminate the layer hen's food and thereby maintain and even amplify Salmonella infection in the food production chain [26,47,55]. Conducive to this is the social life of rats (allowing transmission among rats [64] and the long (> 1 year) environmental survival of Salmonella [65], although ST313 may have lower environmental persistence compared to the diarrhea-causing ST19 [11]. In addition, rat carcasses may also constitute a Salmonella reservoir on which other animals (stray cats, lizards, cockroaches, ants, etc.) can feed and further contribute to the spread of Salmonella [26].
A similar mechanism of indirect ("sponge") reservoir may be postulated for wet markets. In the present study, this is supported by the observation that the marketplaces where ST313 carrying rats were captured (Central Market and Yaté market) were those with the highest rat density. Further, the market setting in a low resource setting such as Kisangani offers exposure of food and shelter in the absence of pest control. By consequence, these are high density populations of rats which was demonstrated by the high trapping success in the present study (one third of non-pre-baited traps) [45]. The present rat population consisted of nearly 90% sexually mature rats versus 63.9% in a rat population in urban Canada; also, length and mass were higher in Kisangani versus Canada (medians of 22.1 cm and 280 g versus 17.5 cm and 142.2 g respectively). This means that the Kisangani rats were probably older, which may increase the probability of rat associated zoonosis [45].
At the human side, socioeconomic factors fueling transmission in Kisangani are numerous: poor sanitary conditions, crowding, overnight stay of vendors, unsafe drinking water and cooking practices. The marketplaces sampled in the present study are easy to reach by the Kisangani population, either by walking or by motorcycle, and human mobility is compatible with the geographical perimeters of the rat-human clusters in the present study. Of note, rats can move too: urban rats are territorial [32,45] but may walk up to 3 km and further in one night [66,67]

The present findings point to the evidence of an animal reservoir for ST313
As only 6.5% of Salmonella carrying rats and 0.5% of all rats captured harbored ST313, it is unlikely that rats constitute a major or preferred reservoir of ST313. However, the rat ST313 isolates clustered with the majority (75.0%) of human isolates, pointing to a close interaction between both. Moreover, the occurrence of ST313 in rat populations postulates also other animals as potential reservoirs of ST313. Previous household studies assessing the iNTS reservoirs in sub-Saharan Africa sampled livestock and poultry but no indoor rodents [15,20,68]. Recently, two studies assessed the meat pathway as a potential source of Salmonella. A slaughterhouse study from Kenya and Malawi found a 12.7% Salmonella prevalence in pig carcasses but failed to detect iNTS clades [66]. A study from East Africa showed that S. Enteritidis ST11 was present in the meat pathway and clustered with human isolates. Other non-typhoidal Salmonella serotypes were observed in the meat pathway too, but not ST313 [23]. Furthermore, a study in DR Congo showed a direct association of rainfall with iNTS, pointing to a possible waterborne environmental reservoir [69].
The potential existence of an environmental reservoir and transmission of iNTS may have an impact on ongoing efforts on iNTS vaccine development, most of which target S. Typhimurium [70] and put more emphasis on water, sanitation, and hygiene as control measures. Moreover, ongoing deforestation and climate change are expected to favor both rodent populations and salmonellosis [65,71] with a potential increasing impact of the rat-Salmonella reservoir and transmission.

Other Salmonella serotypes and antimicrobial resistance
The other serotypes obtained from rats mainly comprised II:42:r:-, Kapemba, Weltevreden and Dublin. They were susceptible to all antibiotics tested (= pan-susceptible), as was also the case for the S. Typhimurium ST19 isolates. This pan-susceptibility might be explained by the fact that the non-ST313 Salmonella isolates most probably represent animal-confined isolates. Agriculture and livestock raising in the Kisangani area are mainly organized as small-scale subsistence farming with expected a lesser use of antibiotics compared to human medicine. A similar observation was noted in the aforementioned study about Salmonella in pigs in Kenya and Malawi: most isolates were susceptible to all antibiotics tested, in line with presumed low antibiotic exposure [72].
Salmonella Dublin is host-adapted to cattle but has been recognized as iNTS too in Mali [73]; it has not yet been detected as part of the microbiological surveillance in DR Congo [22,71]. Salmonella Weltevreden is a major cause of intestinal infections in Asia and was cultured from rats in marketplaces in Thailand and India in different environments as poultry farm, feed store and around residential quarters [48,54]. To our knowledge, Salmonella serotypes II:42:r:-, Kapemba, Mikawasima and Orion so far have not yet been described in rats.

Limitations and strengths
A limitation of the present study was the fact that only outdoor sampling was conducted. Indoor sampling would have been a valuable adjunct but in view of the public perception and acceptability, it was presently not done. Likewise, traps were not pre-baited whereas doing so could have increased trapping success as rats are neophobic [45,67]. Trapping success was however high and in line with a previous study [25]. Last, the molecular subtyping alone (as presently used) cannot provide information about the pathways of transmission (rat to human or vice-versa) [13].
Among the strengths were the multiple sampling sites which precluded rat territorial bias [74]. Further, trapping was not impacted by non-target species and, thanks to excellent communication with the city and market authorities, there was no tripping or stealing by the public [74]. The present sample size (n = 566) was large compared to other studies [48,51,72] and human clinical isolates from the same period and area were available for comparison. In addition, we collected physical data of the rats (which may relate to the presence of zoonosis [74] and sampled year-round, precluding seasonal variation. The co-located laboratory facilities allowed for expertized staff and short and regular transport. Deep organs were sampled, and they accounted for three quarters of the non-duplicate serotype isolates. If only stool/rectal swab or content would have been sampled (as done in many studies [32,45,54,75]), carriage would have been as low as 2.6% (15 rats) and ST313 would have been missed. Furthermore, as confirmed here, rats may carry more than one Salmonella serotypes [44,45], which is the reason why we processed multiple colonies per sample.
Further research should confirm and extend the present findings to other iNTS-endemic areas. Field sampling should include wet marketplaces but also indoor rodent sampling in households of iNTS index patients. In conjunction with rat studies, market retail products should be assessed for contamination with iNTS [76]. Laboratory experiments should establish the pathophysiology of iNTS in rats in particular intestinal colonization and fecal shedding. At a larger scope, further environmental research into the reservoir of iNTS should be conducted [6,21,23]. Apart from rats, small rodents such as mice should be assessed, as well as lizards and geckos which are notable carriers of Salmonella too [77]. Food and water should be further assessed as a vehicle for transmission, including fork-to-farm studies along the meat pathway [23,78], crops and green leaves [13,69]. Source attribution studies such as case-control and outbreak investigation studies should complement the picture of iNTS transmission [13,78].
In conclusion, the present study provided evidence of rats as carriers of S. Typhimurium ST313 clustering with human blood culture isolates in an iNTS endemic area in sub-Saharan Africa.
Supporting information S1