Genomic Characterization and Genetic Profiles of Salmonella Gallinarum Strains Isolated from Layers with Fowl Typhoid in Colombia

Salmonella Gallinarum (SG) is the causative agent of fowl typhoid (FT), a disease that is harmful to the poultry industry. Despite sanitation and prophylactic measures, this pathogen is associated with frequent disease outbreaks in developing countries, causing high morbidity and mortality. We characterized the complete genome sequence of Colombian SG strains and then performed a comparative genome analysis with other SG strains found in different regions worldwide. Eight field strains of SG plus a 9R-derived vaccine were subjected to whole-genome sequencing (WGS) and bioinformatics analysis, and the results were used for subsequent molecular typing; virulome, resistome, and mobilome characterization; and a comparative genome study. We identified 26 chromosome-located resistance genes that mostly encode efflux pumps, and point mutations were found in gyrase genes (gyrA and gyrB), with the gyrB mutation S464T frequently found in the Colombian strains. Moreover, we detected 135 virulence genes, mainly in 15 different Salmonella pathogenicity islands (SPIs). We generated an SPI profile for SG, including C63PI, CS54, ssaD, SPI-1, SPI-2, SPI-3, SPI-4, SPI-5, SPI-6, SPI-9, SPI-10, SPI-11, SPI-12, SPI-13, and SPI-14. Regarding mobile genetic elements, we found the plasmids Col(pHAD28) and IncFII(S) in most of the strains and 13 different prophage sequences, indicating a frequently obtained profile that included the complete phage Gifsy_2 and incomplete phage sequences resembling Escher_500465_2, Shigel_SfIV, Entero_mEp237, and Salmon_SJ46. This study presents, for the first time, the genomic content of Colombian SG strains and a profile of the genetic elements frequently found in SG, which can be further studied to clarify the pathogenicity and evolutionary characteristics of this serotype.


Introduction
The genus Salmonella belongs to the family Enterobacteriaceae and includes more than 2600 different serotypes (or serovars) [1]. These Gram-negative bacilli are facultatively anaerobic and are critical for dangerous disease outbreaks in humans and animals [2]. In domestic birds, the serovar Gallinarum is host-specific and nonmotile and can be subclassified to biochemical testing using an API 20E kit (Biomerieux, Marcy l'Etoile, France). The farms, types, sources, and ages of the studied birds are indicated in Table 1.

DNA Extraction and PCR Identification
After biochemical identification, pure colonies were cultured in LB broth (Invitrogen, Carlsbad, CA, USA) for 18 h at 37 • C for DNA extraction using a ChargeSwitch genomic DNA (gDNA) mini bacteria kit (Thermo Fisher Scientific, Carlsbad, CA, USA). The DNA concentration was diluted to 0.5 ng/L based on the Qubit fluorometric quantitation and with a double-stranded DNA (dsDNA) broad-range (BR) assay kit (Life Technologies Corporation, Eugene, OR, USA). A multiplex PCR assay was performed to differentiate Salmonella biovars Gallinarum (SG), Pullorum (SP), and vaccine strain 9R [28].

Whole-Genome Sequencing, Assembly, and Annotation
Whole-genome sequencing of one isolate per farm plus an available commercial 9R-based SG vaccine (9R-Col) was performed. Libraries were prepared using a Nextera XT DNA library preparation kit (Illumina, Inc., San Diego, CA, USA), and sequencing was performed using a MiSeq sequencing system (Illumina, Inc.), with 500 cycles of amplification followed by paired-end sequencing (2 × 250-bp reads).
Before library assembly, the quality of raw reads was evaluated using FastQC v0.11.9 [29], and low-quality reads were excluded by Trimmomatic v0.36 [30]. The genomes were assembled de novo by using SPAdes v3.15 [31], excluding contigs smaller than 500 bp and coverage means smaller than 2. QUAST v4.1 [32] was used to determine the quality of the assembly stats (genome size, number of contigs, N50 value, and G+C content). Finally, we annotated the assembled genomes using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP v. 6.3) [33]. The raw sequenced reads were deposited into the National Center for Biotechnology Information Sequence Read Archive database (NCBI SRA). The assembled genome sequences were deposited in GenBank under BioProject number PR-JNA483415. We also retrieved 11 additional SG genomes with related available information on the country and year of isolation from the SRA and GenBank databases. Three of these retrieved genomes (SA68, 287/91, and SG9-1955) correspond to complete genomes. The information on the SG strains used for this study is indicated in Table 2.  Comparative genomics analyses were performed with the 20 SG strains mentioned above. Initially, we used Snippy v4.6.0 (https://github.com/tseemann/snippy, accessed on 1 November 2022) to produce an alignment of "core SNPs (single nucleotide polymorphisms)" by setting the strain 287/91 (GenBank: NC_011274) as the reference. We used this core SNP genome alignment to build a phylogenic tree. The best-fit substitution model for the phylogenetic analysis was estimated with ModelTest-NG v0.1.7 [34] based on Akaike information criteria (AIC). To infer phylogenetic relationships among the sequences, a maximum likelihood tree was constructed with RaxML-NG v1.1.0 [35] using the TVM nucleotide substitution model, and nodal supports were estimated with 1000 bootstrap replicates. The tree was visualized and annotated using ITOL v6 [36].
We also used the available online tools in the Center for Genomic Epidemiology (CGE) platform (https://www.genomicepidemiology.org/, accessed on 1 November 2022), namely, by SPIFinder v2.0, to identify Salmonella pathogenicity islands [42] and to identify chromosomal point mutations associated with acquired antimicrobial resistance by Res-Finder v4.1 [43]. For SPI identification, we chose a threshold of 95% for a minimum identity and three different thresholds for the minimum coverage percentage of an SPI. We defined the results based on either complete coverage (=100%) or incomplete coverage (100% > coverage > 60%, and 60% ≥ coverage > 20%). We performed additional manual BLASTN searches to reconfirm the presence of SPIs.
The detection of prophages was performed with the Phaster search tool [44]. Genomes were mapped against the reference genome strain 287/91 and visualized using BRIG v0.95 [45].

Genome Assembly, Annotation, and Molecular Typing
Colombian SG strains were sequenced, and the resultant assemblies were comprised between~4.70 and 4.77 Mb with 34 to 175 contigs, the genome depth ranged from 58× to 77×, and the GC content ranged from 50.1 to 53.1% (Table 3). Genome annotation was performed, and between 4505 and 4593 CDSs were thus identified. These results were in accordance with those obtained from an analysis of the genomic data obtained for other SG strains. In silico serotyping using SISTR was performed to classify all the strains in the D1 serogroup and serovar Gallinarum, with O antigens 1, 9, and 12. The molecular typing of these sequenced and analyzed strains revealed a high frequency of ST 78 among the samples, except for the reference strain 287/91, which was typed as ST 331 (Table 3).

Antimicrobial Resistance Genes
We used an alignment of the core SNPs involving 1570 base pairs to cluster the SG strains. Because the raw data of three retrieved genomes (SA68, 287/91, and SG9-1955) were not available for assembly, we selected core SNP alignment to build the phylogeny. The phylogenetic tree inferred from this alignment showed clustering partially related to geographic origin. Specifically, all eight Colombian field strains were grouped into one cluster ( Figure 1). The 9R-Col vaccine was clustered, as expected, with the other SG9 strains and with two field strains from France and Brazil. The human-origin strains were clustered together, and the African and Indian strains were isolated in separate basal branches.
We used the Comprehensive Antibiotic Resistance Database (CARD) [39] to search for the genes associated with antimicrobial resistance (AMR). We chose the CARD because it included both chromosomal and acquired genes. On the other hand, we used the ResFinder v4.1 tool [43] exclusively to look for chromosomal point mutations associated with AMR. The explored AMR genes included parC, parE, gyrA, gyrB, pmrA, pmrB, acrB, and 16S_rrsD.
Regarding the presence of genes, we consistently found the same 26 chromosomal genes in almost all SG strains, except strain 4295/02, which lacked the marA gene, a global activator protein that, in addition to inducing the MDR efflux pump AcrAB, downregulates the synthesis of the porin OmpF ( Figure 1). Most of these genes encode efflux pump complexes and belong to the following AMR gene families: the major facilitator superfamily-MFS (emrAB, emR, tolC, and hns), the resistance-nodulation-division superfamily-RND (acrAB, sdiA, marA, acrD, baeR, cpxR, mdtBC, mdsAB, golS, cpxA, and crp), the multidrug and toxic compound extrusion family-MATE (mdtK), and the ATP-binding cassette family-ABC (mbsA and YojI). Additionally, the two-component regulatory system kdpDE, general bacterial porin ompA, undecaprenyl pyrophosphate-related protein bacA, and aac(6 )-Iaa, which belongs to the AAC(6 ) aminoglycoside acetyltransferase family, were identified.
Moreover, the ResFinder tool revealed genes with aminoglycoside resistance. The aac(6 )-Iaa gene, which potentially confers resistance to amikacin and tobramycin, was found in all the strains. In addition, the strain CDC 4801/72 carried the plasmid-encoded genes aph(3")-Ib and aph(6)-Id, which confers resistance to streptomycin ( Figure 2). Regarding point mutations, we found resistance to quinolones in gyrase genes. The mutation S464T in gyrB, which confers resistance to ciprofloxacin, was found only in six Colombian field strains. In the gyrA gene, we found four variant mutations that confer resistance to nalidixic acid and ciprofloxacin. These mutations were S83F (in the Brazilian strain 287/91), D87G (in the French strain 11CEB2315SAL), D87N (in the Tanzanian strain ST 78), and D87Y (in the Indian strain VTCCBAA614) (Figure 2). Regarding point mutations, we found resistance to quinolones in gyrase genes. The mutation S464T in gyrB, which confers resistance to ciprofloxacin, was found only in six Colombian field strains. In the gyrA gene, we found four variant mutations that confer resistance to nalidixic acid and ciprofloxacin. These mutations were S83F (in the Brazilian strain 287/91), D87G (in the French strain 11CEB2315SAL), D87N (in the Tanzanian strain ST 78), and D87Y (in the Indian strain VTCCBAA614) ( Figure 2).

Virulence Genes
We identified fifteen different SPIs among the analyzed SG strains through SPIFinder v2.0 ( Figure 2, Table S2). The number of SPIs varied from 13 to 15 (15 in all Colombian strains). Regarding the completeness of SPIs, we identified the following seven complete SPIs (in ≥90% of the studied strains): SPI-1, SPI-2, SPI-3, SPI-13, SPI-14, C63PI, and the putative SPI protein ssaD ( Figure 3). In addition, we identified eight incomplete SPIs in the studied strains: SPI-4, SPI-5, SPI-6, SPI-9, SPI-10, SPI-11, SPI-12, and CS54. Surprisingly, the strain with the fewest SPIs (=13), the Brazilian strain 4295/02 isolated in 2002, lacked SPI-1, one of the most important pathogenicity islands in Salmonella, as well as C63PI.  A detailed search for specific virulence factors led to the identification of between 130 and 135 genes in most of the strains, except for the Brazilian strain 4295/02, which presented only 94 genes (this was mainly due to the absence of SPI-1 genes) (Figure 4, Table S3). All the SG strains carried the tssJLM gene, which belongs to the type VI secretion system (T6SS). However, the strain CDC 4801/72 isolated from humans in 1972 carried three additional genes (IDs: STM0276, STM0284, and STM0287) related to the T6SS. When comparing the Colombian strains with 9R-Col (a 9R-based vaccine), we identified some differences in the gene content. The 9R-Col lacked three genes that were present in all the field strains: invJ (an SPI-1-encoded T3SS needle-length regulator), spvD (a plasmid-encoded T3SS effector cysteine hydrolase), and ssaK (an SPI-2-encoded T3SS stator). Moreover, 9R-Col carried the gene shdA (an adhesin and AIDA autotransporter-like protein), which was absent in all the Colombian field strains but was found in other strains. Finally, the strains Buga, FTA303, GDX-03965-17, and GDX-03999-17 lacked the gene ratB (an adhesin and CS54-encoded putative outer membrane protein), while the strains FTA303, GDX-2, GDX-7, and GDX-03999-17 lacked the gene spvB (a plasmid-encoded toxin and T3SS effector with ADP-ribosylation activity). In addition, the strain GDX-03999-17 lacked the gene sseL (a T3SS effector with deubiquitinase activity) (Figure 4).
We detected four plasmid-encoded virulence factors: (1) the gene mig-5 (a putative carbonic anhydrase with antimicrobial activity) was found in several strains, including all the Colombian SG strains, (2) the gene spvB encoding a toxin was found in some field strains, (3) the gene spvC, a T3SS effector encoding a phosphothreonine lyase, and (4) the gene spvD, a T3SS effector encoding a cysteine hydrolase (Figure 4).
Other genes exclusively absent in some strains were avrA, sptP, and sipABCD, and all the SPI-1-encoded genes were absent in the Brazilian strain 4295/02. The genes sopD2 and sseK1 were absent in the Indian strain VTCCBAA614. The sseL gene was absent in the strains GDX-03399-17 and 4295/02. The sspH2 gene was absent in the strains ST572 and 4295/02 ( Figure 4).
We explored the prophage diversity and profiles using the PHAge Search Tool Enhanced Release (PHASTER). According to the related database, thirteen most common phage (MCP) references (phages with the highest number of proteins most similar to those found in the query genome) were identified (Figure 2). The number of prophages by strain varied from three to six (Table S4) To gain a deeper insight, we evaluated the genetic content of the most frequent MCPs and then searched for the genes associated with virulence factors or antimicrobial resistance in the SG strain GDX58 as it presented the most prophages among the Colombian strains ( Figure 5). The complete prophage Gifsy_2 carried the virulence factors sopE (encoding an SPI-1 type III secretion system guanine nucleotide exchange factor), sodC1 (encoding superoxide dismutase [Cu-Zn]), ail (encoding the attachment invasion locus protein), clpP (encoding an ATP-dependent Clp protease proteolytic subunit), and yjcS (encoding the putative alkyl/arylsulfatase YjcS), a gene associated with resistance to sodium dodecyl sulfate (SDS). In addition, incomplete MCP Salmon_SJ46 presented the virulence factor virB (encoding the virulence regulon transcriptional activator VirB) and the genes umuC (encoding protein UmuC) and umuD (encoding protein UmuD) associated with UV protection. MCP Shigel_SfIV presented the virulence factors yfdH (encoding a prophage bactoprenol glucosyl transferase) and yfdG (encoding a prophage bactoprenol-linked glucose translocase/flippase), as well as genes potentially conferring resistance to copper pcoE (encoding the putative copper-binding protein PcoE) and reactive chlorine species (RCS) stress resistance genes cusS (encoding the sensor histidine kinase CusS) and rclR (encoding the RCS-specific HTH-type transcriptional activator RclR). MCP Entero_mEp237 presented the gene yjcS (encoding the putative alkyl/arylsulfatase YjcS) with associated resistance to SDS. MCP Escher_500465_2 presented the gene yfcJ (encoding the putative MFS-type transporter YfcJ).

Discussion
According to the U.S. Department of Agriculture, Latin American countries, especially Brazil, Mexico, Argentina, Colombia, and Peru, contribute significantly to global chicken meat production. In fact, Colombia was ranked among the top ten world producers in 2022, with 1.88 million tons [46]. Despite evidence showing that intermittent outbreaks of FT in Colombia reported in recent years caused significant economic impacts, in-depth genetic and genomic characterization studies of SG have been scarce [27,47]. Thus, our aim for this study was to describe the genomic characteristics of eight SG Colombian field strains and compare them against a 9R-based vaccine and strains from other countries. We performed molecular typing and explored the resistome, virulome, mobilome, and additional features that may lead to phenotypic and pathological differences among strains.
Bacteria engage a variety of mechanisms to resist the action of antibiotics. For some bacteria, the balance in membrane permeability/impermeability plays a key role in limiting the uptake of antimicrobial compounds and the subsequent diffusion and efflux that reduces the intracellular concentration [48]. We found 26 chromosomal genes in all the SG strains, with most encoding efflux pumps, associated with resistance to 16 classes of antimicrobial agents as well as antiseptics and disinfectants. These results are similar to previous WGS-based studies of chromosome-encoded genes [49][50][51]. The bacterial chromosome-encoded systems involved in AMR typically confer a low level of resistance, but they are important as possible 'stepping stones' to higher levels of resistance [52]. Several resistance mechanisms may lead to high-level antimicrobial resistance via the action of multidrug efflux pumps, mutations in gyrases and topoisomerases, and decreased outer membrane permeability [53,54]. Our results are in concordance with those of a previous report [51] and show an SG profile comprising conserved chromosomal genes associated with AMR. The impact of mutations and polymorphisms on the expression of these genes needs to be explored.
Gyrases are type II topoisomerases that modulate dsDNA topology and maintain chromosomes in a loosely wound state. They form heterotetramers composed of two gyrA and two gyrB chains. GyrA carries an active tyrosine site through which a transient covalent intermediate with DNA is formed, while gyrB binds cofactors and catalyzes ATP hydrolysis. In Salmonella, quinolone resistance has been attributed to point mutations in gyrase genes, mainly in a region termed the quinolone resistance-determining region (QRDR), which is located between amino acids 67 and 106 in gyrA [54]. We detected the gyrA mutations S83F, D87G, D87N, and D87Y in SG strains isolated from different continents. These mutations have been associated with quinolone resistance in Salmonella [54] and have been reported in SGs showing AMR to quinolones [55,56]. In addition, we detected the gyrB S464T mutation in several SG Colombian strains. This mutation has been reported only in a clinical S. Typhimurium strain, which shows resistance to ciprofloxacin [57].
Aminoglycoside resistance requires careful consideration. AMR genes against aminoglycosides are the most diverse and the most frequently identified in Salmonella field strains [58]. Moreover, the aac(6 )-Iaa gene has been frequently detected, and we confirmed that it was carried in all the analyzed SG strains. However, this gene has been reported to be a cryptic gene, with a mutation required to activate it [59]. On the other hand, horizontally acquired AMR genes were found in only one strain, CDC 4801/72, in the plasmid-encoded genes aph(3")-Ib and aph(6)-Id, which confers resistance to streptomycin. This finding is interesting, as this strain was isolated from humans. The possibility of these genes being transferred to SGs should not be disregarded. In recent years, an increase in AMR against aminoglycosides and quinolones has been observed in SG [60]. The indiscriminate use of antimicrobial drugs in poultry farms, still being practiced in some countries, is considered to be responsible for this resistance and is a concern with respect to animal and human health [4,61].
SPIs are horizontally acquired loci encoding genes that facilitate several virulence mechanisms. While some of these SPIs are ubiquitous and distributed among all Salmonella species, others are associated with specific serotypes, revealing fitness advantages and host specificities [7]. We identified 15 different SPIs (C63PI, CS54, ssaD, SPI-1, SPI-2, SPI-3, SPI-4, SPI-5, SPI-7, SPI-9, SPI-10, SPI-11, SPI-12, SPI-13, and SPI-14) identified in most of the SG strains, which indicate that they are representative of the SG biovar profile. SPI-1 and SPI-2 are the most extensively studied islands, and they encode the T3SS and are fundamental to host cell invasion and intracellular survival, respectively [5,7,9]. Interestingly, the Brazilian strain 4295/02 lacked SPI-1, one of Salmonella's most important pathogenicity islands. The genome sequence of this strain was retrieved from GenBank, and the isolation source was indicated as a sample of layer chicken litter via drag swabbing. An SPI-1 mutant SG showed decreased invasiveness of avian cells in vitro, but its ability to persist within chicken macrophages did not differ from that of the wild type [62].
It is not possible to rule out the circulation of a less virulent SG lineage potentially adapted to the poultry environment. Other pathogenicity islands consistently detected in SG were C63PI, which carries an iron uptake system, and the T3SS protein ssaD [63]. The CS54 island (also recognized as SPI-24), encodes intestinal colonization and persistence determinants [7]. As evolutionarily conserved elements, SPIs are susceptible to horizontal transfer and mutations [64]. Incomplete SPIs may be a consequence of constant evolution and host adaptation, including insertions and deletions that impair function or become pseudogenes. We have identified some incomplete SPIs in SG. Variations in the SPI content have been previously described, and SPIs may be differentially distributed among Salmonella serovars [65,66]. Our findings regarding virulence factors and SPIs are in concordance with previous studies [2,[21][22][23]. The functional analysis of each SPI in SG should be performed to better understand the evolutionary processes and impacts of this serovar.
Protection against SG is globally covered with 9R vaccines. This strain is defective in O-side chain repeats in lipopolysaccharide (LPS) because of a mutation in the rfaJ gene, and the attenuated expression of this gene was initially attributed to this mutation [67]. In this study, we identified a 9R-Col vaccine lacking three virulence genes (spvD, invJ, and ssaK). The spvD gene has been previously reported to be deleted in a 9R strain [67]. Our results agreed with this finding, as all the 9R-related strains analyzed in this study (9R-Col, SG9-1955, andSG9-1995) lacked this gene (Figure 4). This plasmid-encoded gene is a T3SS effector of SPI-1 and SPI-2 and contributes to systemic infection in mice by inhibiting the NF-κB pathway and the secretion of proinflammatory cytokines [68]. The invJ gene is a T3SS encoded by SPI-1 and is essential for the projection of the needle length of the injectisome complex, which in turn is relevant to the protein secretion and bacterial invasion of epithelial cells [69,70]. The ssaK gene is a T3SS encoded by SPI-2 and is required to inject effector proteins into host cells to facilitate intramacrophage survival and virulence levels in an animal host [71]. Interestingly, a comparative proteomic study with wild-type and live 9R vaccines identified 42 virulence genes, including invJ and ssaK, that were downregulated in the vaccine strain [72]. Our results reinforce the notion that the attenuation of SG 9R strains may be a result of combined defects in virulence factors [72]. Additional studies exploring point mutations in 9R-based vaccines may further clarify the picture.
Plasmid replicons are mobile genetic elements capable of conferring or increasing bacterial virulence and resistance to antimicrobials. We identified two frequent plasmids distributed among SG strains. The most frequently distributed, the small plasmid Col(pHAD28) (GenBank: KU674895), was discovered in avian-origin S. Hadar and was associated with quinolone resistance because it carries the qnrB19 gene [73]. However, the SG Col(pHAD28) variants found in this study lacked this gene. Interplasmid recombination is presumed to be the event underlying the formation of these variants [73]. The second most frequently distributed plasmid was identified as IncFII(S). According to the reference genome (strain 287/91), this plasmid has been named pSG (GenBank: HG970001), which is a virulence plasmid specific to the S. Gallinarum serotype. This plasmid has been characterized based on its Salmonella plasmid virulence locus (spv), the expression of which contributes to systemic virulence and intramacrophage survival [66].
Temperate phages (prophages) are viruses that infect bacteria and persist benignly within the chromosomal or plasmid DNA of a bacterial host. Complete or cryptic (defective) prophages can carry genes that contribute to the bacterial acquisition of antimicrobial resistance and the adaptation to new environments or hosts, increasing the virulence and pathogenicity of the bacteria [44]. We found 13 different MCPs distributed heterogeneously among the studied SG strains. No geographical or temporal relationship was observed concerning the composition or distribution of these MCPs. However, one complete prophage (Gifsy_2) and four cryptic prophages (Escher_500465_2, Shigel_SfIV, Entero_mEp237, and Salmon_SJ46) were frequently found in SG. The composition and diversity of the prophage sequence profiles and identified cargo genes may be indicative of specific serotypes, and therefore may contribute to subtyping approaches [74,75]. Interestingly, the abovementioned MCPs carried certain virulence factors as well as genes with the potential to confer resistance to chemical and physical stressors. The abundant diversity and insertion of phages into the SG genome may contribute to the pathogenicity and host adaptation of this chicken-associated serovar [66].
The genomic characteristics of eight SG Colombian field strains showed that they belong to a different genetic cluster than SG vaccine strains and field strains from other countries. Specific differences were also observed in the gyrase B gene which carried an infrequent point mutation (S464T) which is associated with resistance to ciprofloxacin [57]. Some virulence-associated genes were irregularly distributed in Colombian strains compared to other SG strains. We detected the plasmid-encoded virulence genes spvC, spvD, and mig-5 in all Colombian strains, while they were less frequent in the other SG strains. SpvCD genes, which are T3SS effectors, were found to be expressed by Salmonella when ingested by macrophages, contributing to their survival and proliferation [10]. The mig-5 gene has been associated with the resistance of Salmonella to the host complement system [10]. Conversely, the genes ratB and shdA were poorly detected or absent in the Colombian strains of GS, while they were more frequent in the other strains. These genes encode non-fimbrial adhesins located in the pathogenicity island CS54 and are involved in intestinal colonization and persistence in the host [76]. A particular case was the absence of the gene sseL in one Colombian strain. SseL is a T3SS effector with deubiquitinase activity required for the Salmonella-induced delayed cytotoxicity of the macrophages [77]. Salmonella gain and lose virulence factors over time as adaptive mechanisms to new hosts or environments [78,79]. On the other hand, the genomic profiles of SPIs, chromosomal genes associated with antimicrobial resistance, plasmids, and prophage sequences revealed relative commonalities in these important genetic features among Colombian strains and the SG isolated from other countries. These findings highlight the importance of these elements for the survival and pathogenicity of SG [6][7][8]10].
WGS is a powerful tool for detecting many diverse virulence factors. The use of bioinformatic tools such as ABRicate and SPIFinder [42] integrated into platforms and databases such as the Center for Genomic Epidemiology (CGE) and the Virulence Factor Database (VFDB) [40] in this study helped us to understand the mechanisms underlying SG virulence better. Defining the most important genetic traits that confer virulence and resistance to SG is essential. The knowledge of these features can allow us to predict the level of virulence and resistance, and thus develop strategies to reduce it.

Conclusions
The genomic characterization of Colombian SG isolates shows a high genetic similarity between them, but differences with other SG vaccine or field strains isolated in other countries. However, all the strains belonging to the SG serotype maintain a relatively conserved profile of pathogenicity islands, virulence factors, efflux pumps, and prophages. On the other hand, the comparison of the presence or absence of virulence factors between field strains and a 9R vaccine strain allows new perspectives to emerge on the pathogenicity of SG. Finally, the information obtained in this study may help to better understand the mechanisms underlying SG virulence and resistance.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/genes14040823/s1, Table S1: Chromosomal genes associated with antimicrobial resistance (AMR). Genes were categorized by AMR family, resistance mechanism, efflux component, and antimicrobial classes. Table S2: Salmonella pathogenicity islands identified by SPIFinder v2.0. Genetic features such as origin, percent identity, position, and insertion location are indicated. Table S3: Virulence factors. The indicated values represent the percentage of coverage of identified genes in accordance with the VFDB. Table S4: Prophages sequences identified by Phaster search tool. Features such as localization, length, completeness, score, number of proteins, most common phage, and GC% are indicated.