Comparative genomics reveals high genetic similarity among strains of Salmonella enterica serovar Infantis isolated from multiple sources in Brazil

Background

Non-typhoidal Salmonella enterica (NTS) serovars are one of the four major bacterial pathogens related to human foodborne diseases worldwide, with estimated 93.8 million gastroenteritis cases and 155 thousand deaths per year (Majowicz et al., 2010; WHO, 2022). Salmonella enterica subspecies enterica serovar Infantis (Salmonella Infantis) is NTS, ubiquitous, zoonotic and globally distributed serovar. Despite its presence in several types of isolation sources, the main reservoirs of Salmonella Infantis are food-producing animals, with a higher prevalence in poultry (Crim et al., 2015; EFSA & ECDC, 2019). In humans, the main clinical manifestation of Salmonella Infantis infection are gastroenteritis cases developed through the consumption of contaminated raw or undercooked meat products. The increasing resistance rates to antimicrobial drugs of clinical and non-clinical use also raised an alert over this serovar, in special due to the global dissemination of the pESI plasmid, that harbours genes related to antimicrobial resistance, heavy metal tolerance, virulence and stress adaptation (Alvarez et al., 2023).

In Brazil, strains of this serovar may be an important issue for the food safety and public health fields. The country currently ranks among the world’s leading meat exporters (ABPA Brazilian Association of Animal Protein, 2021), and previous reports have demonstrated high frequencies of this serovar among food, environmental, animal and human sources (Medeiros et al., 2011; Voss-Rech et al., 2015; Monte et al., 2019; Vilela et al., 2022a).

Over the years, methods such as pulsed-field gel electrophoresis (PFGE) and multi-locus sequence typing (MLST) have been broadly employed for studying NTS serovars, including Salmonella Infantis strains in Brazil (Almeida et al., 2013; Monte et al., 2019; Vilela et al., 2022a). Currently, whole-genome sequencing (WGS) has been providing significant advances in monitoring genomic relationships and antimicrobial resistance development among bacterial pathogens due to the development of novel methods of analysis, broader access and cost reductions (Gilmour et al., 2013; Allard et al., 2018). As a result, WGS is now an essential tool in the global tracking of zoonotic and foodborne pathogens of public health importance, thanks to its discriminative nature, speed, and cost-effectiveness (Gilmour et al., 2013; Allard et al., 2018). Such advances in WGS have been driven by international efforts to integrate human, animal, and environmental health through the One Health philosophy (Gilmour et al., 2013; Allard et al., 2018).

Different phylogenetic approaches can be applied in the analysis of WGS data, such as single-nucleotide polymorphisms (SNPs), gene-by-gene analysis such as core and whole genome MLST, fragmented genomes, circular comparisons, gene synteny and pangenome calculations (Darling, Mau & Perna, 2010; Alikhan et al., 2011; Agren et al., 2012; Liu, Chiou & Chen, 2016). Genetic markers, such as antimicrobial resistance and virulence genes, pathogenicity islands, orthologous proteins, plasmids and prophages, can also be searched using WGS-based tools. In a global context, several published studies have analyzed Salmonella Infantis strains employing WGS-based methods, such as SNP analyses, MLST and core genome MLST. Investigators also have searched for plasmids, antimicrobial resistance and virulence genes (Brown et al., 2018; Acar et al., 2019; Alba et al., 2020; Nagy et al., 2020; Egorova et al., 2021; Kürekci et al., 2021; Szmolka, Wami & Dobrindt, 2021; Tyson et al., 2021; Papić et al., 2022). However, in Brazil, genomic information is still scarce for this serovar (Monte et al., 2019; Melo et al., 2021; Santos et al., 2021; Bertani et al., 2022; Vilela et al., 2022a; Vilela et al., 2022b).

Considering the limited genomic information available about this serovar in Brazil, the aims of this study were to characterize Salmonella Infantis strains isolated from food, farm and industry environments, humans, animals, and animal feed from 2013 to 2018 in Brazil using comparative genomic analyses, and to establish correlations between these strains and isolates from other countries.

Materials and Methods

Results

wgMLST and Gegenees

The phylogenomic trees generated by wgMLST and Gegenees analyses are displayed, respectively, in Figs. 1 and 2. Both methods showed, in overall, a separation among Brazilian and some non-Brazilian genomes.

In wgMLST, five major groups were formed. The two largest groups comprised 72 Brazilian genomes of strains isolated from food, environment, human and animal feed, and the eight non-Brazilian genomes with common SNP clusters, from USA, UK, Bolivia and Paraguay. The remaining seven genomes from Brazil of strains isolated from animals and a single environmental strain were located in a separate group. The 10 non-Brazilian genomes that were assigned to different SNP clusters from those assigned for Brazilian genomes were grouped into two different groups: one composed of genomes from Australia, Mexico and Salmonella Infantis reference SINFA, and another with genomes from Germany, Hungary, Israel, Peru, Russia, South Africa and Turkey (Fig. 1).

Gegenees also showed five main groups, but with some differences in relation to wgMLST (Fig. 2). All 80 genomes from Brazil were distributed in four major groups, with no clear separation regarding the strains’ isolation sources. In addition, the genomes from Australia and Mexico with different SNP clusters and four genomes from Bolivia, Paraguay, UK and USA with common SNP clusters were also grouped along with Brazilian genomes in these same groups. The fifth group formed was comprised by the eight genomes from Germany, Hungary, Israel, Peru, Russia, South Africa and Turkey of different SNP clusters and four genomes from the UK and the USA from common SNP clusters to Brazilian genomes (Fig. 2). Gegenees also showed an overall similarity >91% among all Salmonella Infantis genomes included in the comparison (Fig. 2).

In the wgMLST analysis conducted to evaluate the genomic relationship of the antimicrobial resistance gene profiles of the 80 Brazilian genomes (Fig. S1), it was possible to observe the occurrence of three major profiles/groups. The majority of the strains of Profile 1 were characterized by the presence of genes encoding aminoglycoside resistance (aac(6′)-Iaa), multi-drug efflux pumps (mdsA, mdsB, acrA, acrB, baeR, crp, emrB, emrR, golS, hns, kdpE, kpnF, marA, marR, mdfA, mdtK, msbA, rsmA, sdiA, soxR, soxS), arsenic tolerance (arsR), and gold tolerance (golS and golT). The strains of Profile 2 shared most of the genes of Profile 1, but differed in the absence of mdfA and in the presence of genes encoding resistance to beta-lactams (bla _TEM−1), phenicols (floR), trimethoprim (dfrA8), tetracycline (tet(A)) and silver tolerance (silABCDEFPRS). The strains of Profile 3 differed of Profile 1 in the presence of beta-lactam resistance gene bla _{CMY −2}.

Pangenome analysis

Figure 3 shows the number of pangenome CDSs, the pangenome development calculated using Heap’s law (considering α = 1- γ), and the accessory genome and singletons development, calculated by curve fitting based on least-squares fit of the exponential regression decay, for the five Salmonella Infantis subsets analyzed.

All five Salmonella Infantis subsets analyzed showed the presence of an open pangenome (subset A, α = 0.92; subset B, α = 0.907; subset C, α = 0.936; subset D, α = 0.958; subset E, α = 0.925). The pangenome of subset A (Brazilian genomes) showed the highest proportion and number of genes in the core genome, with 4,066 CDSs identified. The pangenome of subset D (non-Brazilian genomes with common SNP clusters to Brazilian strains) showed the highest proportion and number of genes in the accessory genome, with 668 CDSs identified. The pangenome of subset C (non-Brazilian genomes with different SNP than Brazilian strains) showed the highest proportion and number of singletons, with 209 CDSs identified (Fig. 3).

Genome plasticity and gene synteny

In the circular map generated with BRIG for the evaluation of genome plasticity (Fig. 4), it is possible to observe few conserved deletion fields (blank spaces) among the 20 Brazilian and ten non-Brazilian Salmonella Infantis analyzed, aligned with Salmonella Infantis reference genome SINFA.

In the gene synteny analysis of 20 Brazilian and ten non-Brazilian Salmonella Infantis draft genomes, it is possible to observe the presence of large, conserved and less numerous Locally Collinear Blocks (LCBs), with few areas of deletions, inclusions and inversions, in comparison to the Salmonella Infantis reference SINFA (Fig. 5).

Discussion

Salmonella Infantis is a NTS, ubiquitous and zoonotic serovar, with widespread global distribution over different isolation sources, and mainly associated with gastroenteritis in humans through the consumption of raw or undercooked poultry meat (Crim et al., 2015; EuropeanFoodSafetyAuthority, 2019). Considering Brazil’s relevance in the global meat exportation market (ABPA Brazilian Association of Animal Protein, 2021), the use of genomic tools and analyses could greatly improve the understanding of specific traits of Salmonella Infantis strains in this country. In this work, different genomic approaches were employed in order to expand the characterization of strains of this serovar in the country.

In the present study, it was possible to observe a closer relationship of Brazilian and non-Brazilian Salmonella Infantis strains of the same SNP clusters assigned by NCBI Pathogen Detection in comparison to non-Brazilian genomes of different SNP clusters. Previously, we have also demonstrated this clear genomic distinction among the same 80 Salmonella Infantis genomes from Brazil analyzed here in comparison to other 40 additional genomes from several countries and distinct SNP clusters using cgMLST (Vilela et al., 2022b).

Regarding the distribution of the Brazilian genomes of Salmonella Infantis observed in wgMLST and Gegenees phylogenies, no clear correlation was observed between any specific lineage and the isolation source or date in the majority of the groups formed by both methods. (Figs. 1 and 2). However, wgMLST interestingly clustered all Salmonella Infantis genomes of strains isolated from animals in a group apart from Brazilian genomes of other sources (Fig. 1), which could indicate the circulation of this specific genomic profile mostly among animals in the country.

When wgMLST was combined to the genotypic antimicrobial resistance profile of the 80 Brazilian strains, it revealed the formation of the three distinct genetic groups showing different resistance gene profiles (Fig. S1). While Profiles 1 and 3 grouped strains predicted to be less resistant, Profile 2 contained strains with genetic markers predicting multi-drug resistance.

In comparison to NCBI Pathogen Detection, while wgMLST accurately clustered Brazilian and non-Brazilian Salmonella Infantis genomes with common SNP clusters in comparison to non-Brazilian genomes of different SNP clusters (Fig. 1), Gegenees produced similar results but with less resolution (Fig. 2). In this analysis, a high overall relatedness was also observed among all Salmonella Infantis genomes (>91%; Fig. 2). While wgMLST employs a gene-by-gene approach, Gegenees analysis is based on the similarities found through the alignment of fragments of entire genomes. Eventual differences in the genome sizes and presence of duplicated genes, genomic islands and repetitive regions, for example, could result in small biases in the Gegenees analysis that would not allow the identification of small nucleotide variations, detected by wgMLST. However, as demonstrated in the present study, although with lower resolution, Gegenees was able to generate some similar results to wgMLST and the SNP clusters defined by NCBI Pathogen Detection.

These results indicated that the majority of the Salmonella Infantis strains from Brazil studied were allocated into different genomic groups comprising strains of diverse origins, suggesting that these might be widespread among human and non-human sources and sharing correlated profiles of antimicrobial resistance genes. Moreover, the results demonstrated genetically related groups comprising Brazilian Salmonella Infantis and non-Brazilian genomes from the United Kingdom and North and South America, suggesting their global circulation, while other profiles are distinct and present exclusively within the country.

In the genome plasticity and gene synteny analyses conducted herein by BRIG and Mauve, respectively, Brazilian and non-Brazilian Salmonella Infantis strains shared a highly conserved genomic structure (Figs. 4 and 5). In the genome plasticity analysis, the selected Salmonella Infantis genomes possessed few conserved deletion areas when aligned to the reference Salmonella Infantis genome SINFA (Fig. 5). Similarly, the gene synteny analysis produced by the Mauve also revealed that the selected Salmonella Infantis genomes analyzed possessed highly similar LCBs and few areas of deletions, inclusions and inversions among each other in comparison to Salmonella Infantis SINFA (Fig. 5).

It is important to address that an equal linear organization of LCBs was not observed for the Brazilian and non-Brazilian genomes in comparison to the reference genome SINFA used in the Mauve analysis (Fig. 5). This characteristic does not necessarily indicate a real change in the genomic ordering of the LCBs. Since the genomes analyzed were in a draft configuration, their genomic sequences do not start in the common expected replication origin as in closed genomes, such as the reference strain SINFA. However, despite the visual difference, the results still corroborate a high similarity observed in the other analyses performed.

Several prophages were found among the 80 Brazilian Salmonella Infantis genomes analyzed (Table S1). In NTS, prophages have been demonstrated to be indicatives of genetic diversity and to contribute to bacterial virulence (Gao et al., 2020). In this study, it is worthy to highlight Fels 1 and Gifsy 1, which are well described phages in NTS and were detected in more than 40% of the strains analyzed. Fels 1 has been reported to act in the production of neuraminidase and superoxide dismutase, while Gifsy 1 has been demonstrated to increase the virulence of Salmonella Typhimurium in mice (Brüssow, Canchaya & Hardt, 2004). Entero BP 4795 was found in 35% of the strains and is related to the production of the Shiga toxin in enterohemorragic Escherichia coli (Creuzburg et al., 2005). Yersin L 413C, detected in 20% of the strains, is a common serotype marker of Yersinia pestis (Garcia et al., 2008). However, little is known about the roles in NTS serovars of other phages detected in the present study, such as Salmon 118970 sal3, Salmon SPN3UB, Escher pro483, Stx2 c 1717, Salmon SW9, Entero P4, Salmon vB SosS Oslo, Entero ES18, Entero YYZ 2008, Entero fiAA91 ss and Salmon SP 004.

Prophages have been previously reported in Salmonella Infantis genomes by Gymoese et al. (2019), whose analysis included isolates from humans and several types of non-human sources from Denmark between 2002 and 2012, and in Pardo-Esté et al. (2021), where strains isolated from chicken producing facilities from Chile between 2018 and 2019 were analyzed. In both studies, a greater diversity of phages was detected in comparison to the results obtained here, including differences in the percentages of phages more commonly found, such as Gifsy 1. Therefore, these results demonstrated that important phages that may contribute in the virulence of NTS were detected in high frequencies among Salmonella Infantis strains from Brazil. In addition, they highlight the necessity for further studies to better investigate the roles of other diverse prophages in Salmonella Infantis and other NTS serovars.

The calculation of the pangenome for NTS and other bacteria has an important role for epidemiological and evolutionary purposes, and even as an important tool for secondary analysis, such as the search in silico for vaccine targets (Benevides et al., 2017; Seribelli et al., 2020; Felice et al., 2021). NTS has been considered as a recombinant bacterial genus with an open pangenome and a growing number of orthologous genes as new genomes are sequenced (Alikhan et al., 2018). In the present study, the five subsets analyzed were confirmed to have an open pangenome, with α values smaller than 1 detected for all comparisons.

Despite of the presence of correlated groups observed in wgMLST (Fig. 1), the high similarity percentages found with Gegenees (Fig. 2) and the close α values obtained in the pangenome calculation, it is interesting to notice how the number of CDSs in the core genome, accessory genome and singletons varied in the five comparisons performed (Fig. 3). For example, the subset of Brazilian genomes showed the highest number of CDSs in their core genome, even when compared to the subset of genomes from the same SNP clusters, which in contrast, showed the biggest accessory genome of all the comparisons (Fig. 3). These results demonstrated that Brazilian isolates of Salmonella Infantis presented a high genomic relatedness in the pangenome analysis, with a high core genome content, which significantly differed from genomes of other countries from common or different SNP clusters. This level of genomic conservation of their core genome could also represent an advantage for the possible development of vaccine targets suited for Salmonella Infantis circulating among animals and humans in Brazil.

The pangenome analysis was employed by Mattock et al. (2022), that characterized 395 genomes isolated from humans from South Africa between 2004 and 2020, and also by the report of Pardo-Esté et al. (2021) mentioned above. While Mattock et al. reported 3,983 genes belonging to the core content in the pangenome, a very similar result to the values obtained here was found. Pardo-Esté et al. (2021) reported a core content of 2,618 genes, which differed from our results. It must also be noticed that both studies have reported much higher values for the accessory genome and singleton genes, demonstrating a greater genomic diversity in comparison to our results. These differences demonstrate how strains from the same serovar but from distinct locations may present different genomic traits, reinforcing the importance of more WGS-based works to unravel the epidemiology of important pathogens such as Salmonella Infantis.

Brazilian and non-Brazilian Salmonella Infantis genomes were also demonstrated to have similar unique orthologous protein clusters related to biological processes, molecular functions and cellular components in relation to Salmonella Infantis reference SINFA. Since SINFA was isolated from chicken in 1973 in the United Kingdom (data available at NCBI’s GenBank BioSample page; accession number SAMEA3106395), these results showed that strains of this serovar, in overall, have evolved in order to acquire novel proteins capable to favor the virulence, survival and adaptation of Salmonella Infantis within its hosts. For example, the most diverse unique orthologous protein clusters observed were those found in the two non-Brazilian genomes positive for the pESI global widespread plasmid (119944 from Israel and VGNKI000011 from Russia). These genomes shared unique orthologous protein clusters related to plasmid maintenance, plasmid partitioning, response to mercury ion and mercury ion transmembrane transporter activity, corroborating to the features already described for Salmonella Infantis harboring these plasmids. Also, such clusters of orthologous protein were not present among pESI-negative genomes from Brazil and other countries, reinforcing the hypothesis that strains of this serovar are still in constant acquisition of novel genetic features.

Finally, it is important to mention some of the limitations of the present study. We must state that the Brazilian sampling of Salmonella Infantis does not comprise an even distribution of isolation sources, years and Brazilian states. Therefore, despite of the corroboration of the results here presented, some groups of strains may be over-represented. The number of non-Brazilian genomes included for comparison purposes (n = 18) was limited when compared to the number of Salmonella Infantis already sequenced. By the date of the present study (August 11th, 2023), more than 27 thousand Salmonella Infantis genomes had been deposited in NCBI Pathogen Detection. This specific number of genomes was selected due to limitations of the platforms employed, and since the main objective was to promote a characterization of Brazilian genomes, the amount of non-Brazilian genomes had to be reduced using specific criteria as mentioned before. Finally, the inability to include the allele differences in the wgMLST tree constructed through PGAdb-builder also complicated the inference of the genetic proximity among the strains within each cluster and between clusters.

Conclusion

The results presented using different genomic approaches emphasized the significant genomic similarity among Brazilian Salmonella Infantis genomes analyzed, suggesting wide distribution of closely related genotypes among diverse sources in Brazil. The data generated contributed to novel information regarding the genomic diversity of Brazilian and non-Brazilian Salmonella Infantis in comparison, indicating that the different genetically related subtypes of Salmonella Infantis from Brazil can either occur exclusively within the country, or also in other countries, suggesting that some exportation of the Brazilian genotypes may have already occurred.

Supplemental Information