Phenotypic and genotypic characterization of Salmonella Typhimurium isolates from humans and foods in Brazil

Salmonella enterica subsp. enterica serovar Typhimurium (S. Typhimurium) causes gastroenteritis in many countries. However, in Brazil there are few studies that have conducted a virulence characterization of this serovar. The aim of this study was to evaluate the virulence potential of S. Typhimurium strains isolated in Brazil. Forty S. Typhimurium strains isolated from humans (n = 20) and food (n = 20) from Brazil were studied regarding their invasion and survival in human epithelial cells (Caco-2) and macrophages (U937). Their virulence potential was determined using the Galleria mellonella larvae model combined with the analysis of virulence genes by whole genome sequencing (WGS). A total of 67.5% of the S. Typhimurium studied (32.5% isolated from humans and 35% isolated from food) invaded Caco-2 epithelial cells at levels similar to or greater than the S. Typhimurium SL1344 prototype strain. In addition, 37.5% of the studied strains (25% isolated from humans and 12.5% isolated from food) survived in U937 human macrophages at levels similar to or greater than SL1344. S. Typhimurium strains isolated from humans (40%) and food (25%) showed high or intermediate virulence in G. mellonella larvae after seven days exposure. Approximately, 153 virulence genes of chromosomal and plasmidial origin were detected in the strains studied. In conclusion, the ability of the S. Typhimurium to invade Caco-2 epithelial cells was strain dependent and was not related to the source or the year of isolation. However, S. Typhimurium strains isolated from humans showed greater survival rates in U937 human macrophages, and presented higher proportion of isolates with a virulent profile in G. mellonella in comparison to strains isolated from food suggesting that this difference may be related to the higher frequency of human isolates which contained plasmid genes, such as spvABCDR operon, pefABCD operon, rck and mig-5.


PLOS ONE
Invasion and survival of S. Typhimurium isolates Initially, the Caco-2 epithelial cells were cultured in DMEM medium (Dulbecco's Modified Eagle Medium-Gibco-low glucose) supplemented with 10% fetal bovine serum (Life Technologies) and antibiotic in 5% CO 2 at 37˚C. In addition, 1x10 5 cells were added to each well of a 12-well microplate. The assay was performed after 12 days of incubation until the cells were polarized and differentiated.
The monocytes were cultured in suspension in RPMI medium (1640-powder-Gibco) supplemented with 10% fetal bovine serum (Life Technologies) and antibiotic in 5% CO 2 at 37˚C. In addition, 1x10 5 cells were added to each well of a 24-well microplate. For the differentiation of monocytes into macrophages, 1μl of phorbol 12-myristate-13-acetate (PMA) (Sigma-Aldrich) was used in 50 mL of the RPMI medium (50 ng/mL) and maintained in 5% CO 2 at 37˚C for 48h. All the S. Typhimurium strains were opsonized with 20% mouse serum (Sigma-Aldrich) at 37˚C for 15 min after three washes with PBS (centrifuged at 12000 × rpm for 1 min).
The invasiveness in Caco-2 epithelial cells of these strains was performed after 90 minutes of bacteria-cell interaction with intraepithelial survival during 3 hours. In addition, the survival in U937 human macrophages was performed after 30 minutes of bacteria-cell interaction with intramacrophage survival during 3 hours and in both assays the (multiplicity of infection) MOI was 100:1.
Serial dilutions were performed and plated in LB agar medium plates with incubation during 18-24 hours at 37˚C for later counting of colony forming units (CFU). The experiments were carried out in biological triplicate and in all plates there was a negative control with only cells in the wells.

Virulence analysis in Galleria mellonella
The analysis it was performed according to [19]-adapted in Galleria mellonella. The larvae were maintained at 28˚C in the dark in glass containers (30 cm height-20 cm wide-2 L capacity) with appropriate oxygen and access to food until reaching the sixth instar, whose weight is between 200 mg and 250 mg. After complete development, the larvae were deprived of food and separated into groups of 10 units in glass Petri dishes for each bacterial isolate and controls.
A Hamilton micro-syringe (model 7000.5KH of 10 μL) was used for artificial inoculation of G. mellonella larvae into the center of the last right pro-leg with 5 μL of S. Typhimurium (10 5 CFU/mL) for each of the 40 isolates studied and for the positive control infected with S. Typhimurium strain ATCC14028. The negative control was inoculated with PBS. After inoculation, the larvae were incubated at 37˚C, deprived of food and direct light. During the experimental period of 7 days the larvae were removed every 24 hours of the pre-pupae, in order to delay their metamorphosis with data recorded daily.
The presence of plasmids were determined using PlasmidFinder (Center for Genomic Epidemiology, https://cge.cbs.dtu.dk/services/PlasmidFinder/) with a threshold set for a minimum of 95% identity, and minimum coverage of 60% [21]. BLASTn (https://blast.ncbi.nlm. nih.gov/Blast.cgi) was used to confirm the location, and percentages of identity, and coverage between the S. enterica subsp. enterica serovar Typhimurium str. LT2 pSLT, 93939 bp, NC_003277 reference sequences of the pSLT plasmid, and the spvABCDR, pefABCD, rck and mig-5 genes of the positive and negative genomes for IncFIIs.

Statistical analyses
The comparisons between the means of the virulence tests were performed using Student's ttest for two means and two-way analysis of variance (ANOVA) with post hoc Tukey test for more than two means in the Minitab1 statistical software (version 18.1). The analysis of the virulence genes identification was performed using two proportion z-test in the Minitab1 statistical software (version 18.1) comparing the difference in isolates proportion which contained genes of interest. The graphics and statistical analyses of the virulence assay in Galleria mellonella were performed using the Log-rank (Mantel-Cox) method, both in the Prism5 program for Windows of the GraphPad1 software (version 5.01). For all analyses, the level of significance was α = 5%.

Caco-2 epithelial cells invasion assay
A total of 67.5% of the 40 S. Typhimurium isolates studied (32.5% isolated from humans and 35% isolated from food) invaded the epithelial cells at various levels compared to the SL1344 reference (Fig 1). Considering all the isolates studied the invasion in Caco-2 cells ranged from 1 × 10 5 to 1 × 10 7 CFU/mL. From the statistical analysis of the Student's t-test, three subgroups were found comprising: isolates that invaded more (black) (35%), equal (grey) (32.5%) or less (white) (32.5%) than the S. Typhimurium SL1344 isolate. The bidirectional analysis of variance (ANOVA) categorized the 14 S. Typhimurium isolates that invaded more than the SL1344 reference in groups from A to G. In addition, ANOVA also categorized the 13 S. Typhimurium isolates that invaded less than the SL1344 reference in groups from A to C (Fig 1).

Survival assay in U937 human macrophages
A total of 37.5% of the 40 S. Typhimurium isolates studied (25% isolated from humans and 12.5% isolated from food) survived in the human macrophages at various levels compared to the SL1344 reference (Fig 2). All the isolates studied survived in U937 human macrophages and ranged from 1 × 10 6 to 1 × 10 7 CFU/mL. From the statistical analysis of Student's t-test, three subgroups were formed comprising: isolates that survived more (black) (27.5%), equal (grey) (10%) or less (white) (62.5%) than the S. Typhimurium SL1344 reference. The bidirectional analysis of variance (ANOVA) categorized the 11 S. Typhimurium isolates that survived more than the SL1344 reference in groups from A to E. In addition, ANOVA also categorized the 25 S. Typhimurium isolates that survived less than the SL134 reference in groups from A to K (Fig 2).

Comparison between sources after invasion assay in Caco-2 epithelial cells and survival assay in U937 human macrophages
In both assays, S. Typhimurium isolates were classified with invasion and survival levels similar, greater or lower than the SL1344 reference with the mean of CFU/well (Log 10 ). Invasion in Caco-2 cells of 6.46 and 6.54 CFU/well (Log 10 ) were observed for isolates from humans and foods, respectively ( Fig 3A). Furthermore, the survival in U937 human macrophages was 7.36 and 7.12 CFU/well (Log 10 ) for humans and food isolates, respectively ( Fig 3B). Therefore, no evidence of statistically significant differences were found between isolates from humans and foods for invasion in Caco-2 cells. However, statistically significant differences were found between isolates from humans and foods for survival in U937 human macrophages.

Virulence analysis in Galleria mellonella
The results of the S. Typhimurium studied isolates are presented separately in Figs 4 and 5, respectively. In Fig 4, the ATCC14028 control strain isolated from chickens showed high virulence (black circle) killing 100% of the larvae. For human isolates, four groups were formed after seven days of experiments. The STm48, STm29, STm34 and STm38 isolates killed 70-90% of the larvae, forming the group of virulent isolates (red shapes). The intermediate virulence group (orange shapes) killed between 30-50% of the larvae, comprised of STm39, STm33, STm31 and STm49 isolates. The STm35, STm42, STm06, STm23, ST28, STm30 and STm44 isolates formed the group of low virulence (green shapes) killing between 10-20% of the larvae. Finally, the avirulent group (white circle) comprised of STm02, STm11, STm17, STm27, STm47 isolates and the negative control did not kill any larvae (Fig 4).
In Fig 5, four groups were formed after seven days of experiment for isolates from food. The 583/13 isolate and the positive control ATCC14028 showed a high virulence killing 100%
The STm41 isolate was not classified into either the intermediate or low virulence groups because its measurement fell between the two groups ( Fig 5).
Comparing the mortality of the isolates belonging to the same group (human and food) no significant difference was observed. On the other hand, significant difference was found

PLOS ONE
Invasion and survival of S. Typhimurium isolates among the different groups mentioned above, except for the groups of low virulence (green shapes) and avirulent (white circle). These evidences were found by comparing representatives of each group.

Genetic characterization focused on virulence genes
Genetic characterization focused on discovering virulence genes using the Virulence Factors Database (VFDB) which detected 153 genes related to invasion, survival, colonization, fimbriae and flagella production, among others for all isolates studied ( Table 2).
The similarity rate varied between 87% and 100% of identity for all strains with coverage between 78% and 100% ( Table 2). The spvABCDR operon, pefABCD operon, rck and mig-5 genes were detected in 13 S. Typhimurium isolates from humans and 9 isolates from foods ( Table 3). In order to confirm the location of genes, Plasmid Finder was used to document the presence of the IncFIIs plasmid incompatibility group, which belongs to the pSLT plasmid (Table 4). Independent, BLASTn analysis for all isolates confirmed the presence of the pSLT plasmid carrying the spvABCDR operon, pefABCD operon, rck and mig-5 genes for 22 S. Typhimurium isolates with coverage of 100% and 99-100% of identity (Table 4). The other studied isolates did not present IncFIIs, pSLT plasmid, spvABCDR operon, pefABCD operon, rck and mig-5 genes.

Discussion
In this study, 40 S. Typhimurium isolates from humans (n = 20) and foods (n = 20) in Brazil were compared after invasion assays in Caco-2 epithelial cells, survival assays in U937 human macrophages, Galleria mellonella assays and virulence gene analysis.
Among the 40 S. Typhimurium isolates studied, from humans and foods many invaded the epithelial cells at similar or higher levels compared to the SL1344 reference control. Furthermore, analysis of variance (ANOVA) was performed for all S. Typhimurium isolates that differed from the SL1344 reference (Fig 1). By this analysis three main groups were formed and the source and/or year of isolation did not correlate with the observed virulence profiles. Therefore, these results reinforce that the ability for S. Typhimurium to invade is probably isolate dependent and not related to the source or the year of isolation (Fig 3A).
During the infection process, host neutrophils and macrophages try to control invasion by generating reactive oxygen species (ROS). The degranulation of these cells occurs in a process called respiratory burst [27].
S. Typhimurium has the ability to infect epithelial cells and macrophages in the small intestine, replicating in a niche called Salmonella-containing vacuole (SCV) and consequently triggering an inflammatory process culminating in gastroenteritis [28]. It is important to emphasize that in the present study, among the 40 S. Typhimurium strains studied, from humans and food survived at various levels compared to SL1344 control. In addition, analysis of variance (ANOVA) was performed for all S. Typhimurium strains that differed from the SL1344 control (Fig 2). Three main groups were formed by this analysis and statistical significance was observed among the profiles, suggesting that S. Typhimurium isolates from humans survive more intramacrophage than isolates from food ( Fig 3B).
Among the possible causes of this difference we highlight a lesser dispersion of the survival assay data in U937 human macrophages and subtle differences in the genetic characterization for some important virulence genes.

PLOS ONE
The virulence assay in Galleria mellonella larvae divided the 40 S. Typhimurium isolates into four groups according to their virulence profiles (Figs 4 and 5). It is important to emphasize that there was a higher proportion (60%) of strains isolated from humans that demonstrated a virulent profile in comparison to strains isolated from foods. Therefore, this result may suggest that the S. Typhimurium studied isolated from humans were more virulent than strains isolated from food in Brazil according to the G. mellonella infection model.
G. mellonella larvae are easily grown in large numbers at low costs and produce innate immune response components very similar to humans composed by hemocytes and opsonins [9]. A limitation of this infection model is that insects do not present the second line of defense characterized by an adaptive immune response formed by antibody-producing and memory cells [9].
It is important to mention that the melanization of insects such as G. mellonella occurs when it is infected by a pathogen, followed by melanin synthesis and deposition of this substance to encapsulate the infectious agents at the inoculation site [29]. Therefore, this process is stimulated by the presence of bacteria and fungi, initiating a serine protease cascade responsible for the activation of phenoloxidase that catalyzes the formation of melanin [9,29].
The genetic repertoire research focused in virulence genes by WGS revealed that the isolates regardless of the source were very similar. Moreover, several essential genes for the pathogenesis of salmonellosis were identified in this study with high identity and coverage reinforcing the pathogenic potential of these strains. It is important to mention that the ability of S. Typhimurium strains to invade and to survive in the host cells is closely linked to virulence genes present in the bacterium [30].
The invG, invH, prgH, prgK, prgI, prgJ and iagB genes are found in the SPI-1 of Salmonella spp. and are involved with the formation of the basal body of the T3SS [31]. The export apparatus of the T3SS are encoded by the spaS, spaP, spaQ, spaR and invA genes present in the SPI-1. Furthermore, the spaO, invC, invI, orgB, invJ, invE, sipC, sipB and sipD genes also are present in the SPI-1 and related to cytoplasmic ring, ATPase complex, regulation and translocation of the T3SS [31].
All the genes mentioned above were found in the S. Typhimurium genomes of this study, suggesting that although these strains belong to collections from different years the essential genes for the injection of T3SS effectors proteins have been preserved (Table 2).
Moreover, genes present in the SPI-3, SPI4 and SPI-5 are related to virulence of Salmonella spp. However, more studies are needed in this area, due to the lack of theoretical and scientific information [30]. For example, the pip (pipA, pipB, pipC and pipD) genes may be related to the rate of fluid secretion and inflammatory response during salmonellosis, suggesting that such genes are related to the bacterial enteropathogenicity, but the exact mechanism has not been yet fully elucidated [30,32].
Plasmids are known to be essential for the resistance and virulence of different bacteria, the IncF plasmid incompatibility group is heterogeneous and often described in enterobacteria [33]. The virulence plasmid (pSLT) which belongs to the IncFIIs plasmid incompatibility group has been reported in S. Typhimurium and carries important genes for the pathogenesis of this serovar [34]. Among the content of the plasmid pSLT can highlight the following genes: spvABCDR, pefABCD, rck and mig-5.
The spv operon (Salmonella plasmid virulence) which is formed of five genes (spvA, spvB, spvC, spvD and spvR) has been associated with Salmonella spp. survival and multiplication in macrophages [35]. In addition, the pef fimbrial operon (plasmid encoded fimbriae) is responsible for the adhesion of Salmonella spp. to intestinal epithelium in infant mouse resulting in fluid accumulation and is consisted by four genes (pefA, pefB, pefC and pefD) [36,37].
It is important to emphasize that the rck (resistance to complement killing) and mig-5 (macrophage inducible gene coding for putative carbonic anhydrase) plasmid genes have been associated to the resistance of S. Typhimurium strains to the host complement system which would cause bacterial cell disruption and related to neutralization of toxic compounds produced by macrophages after phagocytosis, respectively [35].
Interestingly, plasmid genes (spvABCDR, pefABCD, rck and mig-5) were detected in the present work in equal numbers and its location was confirmed in the plasmid pSLT, being that the S. Typhimurium strains isolated from humans and food studied had all the plasmid genes mentioned previously or none (Tables 3 and 4). According to Kuijpers and collaborators (2019), the spvABCDR, pefABCD, rck and mig-5 plasmids genes were detected only in the human cases associated with serovars Typhimurium and Enteritidis [38].
The predominant sequence type (ST) of S. Typhimurium from fecal samples has been ST19 worldwide. However, ST313 has caused a significant mortality rate in sub-Saharan Africa and has been described in recent years in Brazil [39]. In the present study, all the ST313 isolates presented the plasmid pSLT carrying the spvABCDR operon, pefABCD operon, rck and mig-5 genes showing a high genomic similarity among each other regardless of the isolation source as previously described [40].
Considering that the S. Typhimurium strains isolated from humans survived more in U937 human macrophages than strains isolated from food in the present study, it is important to emphasize that a complex response like survival in human macrophages probably is not triggered for one or a few genes, highlighting specifically the importance of plasmidial gene groups. In addition, despite the high similarity of the genes present among the isolates studied, a difference was observed in the proportion of human isolates which contained the plasmid gene group spvA, spvB, spvC, spvD, spvR, pefA, pefB, pefC, pefD, rck and mig-5 between S. Typhimurium strains isolated from humans and foods.
Therefore, a statistically significant difference was found (p-value = 0,001) suggesting that a higher proportion of human isolates which contained these genes may contribute to their greater survival in U937 human macrophages. It is important to mention that this plasmid gene group was the only one that showed the same pattern of presence/absence among the isolates. Furthermore, it was also the only group in which p-value differed statistically, calculated using both the normal approximation method and the Fisher's exact method, that latter tends to be more conservative, reinforcing the statistical significant difference.
Similarly, among 9 (69%) of the 13 S. Typhimurium isolates from humans and foods that had a virulent profile (high and intermediate virulence) in G. mellonella, the plasmid gene group was detected reinforcing their importance in the virulence of these isolates.
Finally, many other genes were detected in the S. Typhimurium genomes in the present study, specifically genes related to the fimbriae production, such as the fim (fimA, fimC, fimD, fimF, fimH, fimI, fimW, fimY and fimZ) genes, which encode structural subunits and fimbrial proteins. The lpf (lpfA, lpfB, lpfC, lpfD and lpfE) fimbrial genes have been described as important in the intestinal colonization in murine mucosa [36,37]. Moreover, no relationship was found between the isolates proportion that presented these researched chromosomal genes suggesting that the different profiles found in the invasion assay in Caco-2 epithelial cells and in the survival assay in U937 human macrophages can be linked to the expression of such genes.
In conclusion, a significant percentage of S. Typhimurium isolates from humans and foods showed high invasion in Caco-2 epithelial cells regardless of the source suggesting that the invasiveness to Caco-2 cells is probably isolated dependent and not related to the source or the year of isolation. However, S. Typhimurium isolates from humans showed greater survival rate in U937 human macrophages and higher proportion of isolates with a virulent profile in Galleria mellonella than isolates from foods suggesting that this difference may be related to the higher proportion of human isolates which contained plasmid genes, such as spvABCDR, pefABCD, rck and mig-5. Moreover, several virulence genes present in the pathogenicity islands 1, 2, 3, 4, 5 were detected in the S. Typhimurium isolates from humans and foods studied reinforcing the virulence of this important serovar independent of their clinical or nonhuman origin.
Altogether, the results obtained in this work contributed for a better characterization of S. Typhimurium isolates from humans and foods in Brazil over decades regarding itsability to invade Caco-2 epithelial cells, to survive in U937 human macrophages, virulence, and pathogenic potential.