Salmonella is a very important zoonotic pathogen that has been reported to cause over 200 million human clinical infections with an estimated mortality of 3 million annually (Coburn et al., 2007; Graham et al., 2018; Moussa et al., 2021). Humans easily acquire infection from contact with animal or environmental reservoirs. The emergence of antimicrobial-resistant Salmonella enterica subspecies enterica serovar Enteritidis constitutes a serious global health problem (Pan et al., 2018). This phenomenon is believed to be due to the unregulated use and abuse of antimicrobials especially veterinary drugs including the World Health Organization’s critically important antimicrobials for prophylaxis, during the management of diseases or as growth promoters (Azmi et al., 2018; Sharma et al., 2018; Yu et al., 2021). S. Enteritidis is one of the most important serovars shared between humans and animals in most parts of the world (Mohan et al., 2019). It is notorious for its ability to colonize livestock; wild animals as well as the reproductive system of chickens where they persist without any apparent clinical manifestation, thus continuously contaminate eggs and the immediate environment (Campioni et al., 2018; Abdulhaleem et al., 2019). The sustained and undetected presence of these bacteria in food-producing animals makes it possible to cause prolonged epidemics globally, especially where the consumption of poorly cooked poultry egg and meat is common (Salihu et al., 2015; Judd et al., 2019).

Salmonella Enteritidis is one of the major cause of invasive salmonellosis globally, particularly among malnourished children and adults with debilitating illnesses (Stanaway et al., 2019). Population-based estimates of the burden of enteric Salmonella infection in Southeast Asia range from 2.2 to 10.7 cases/100,000 people annually (Whistler et al., 2018). Similarly, S. Enteritidis is reported to account for 38% of all clinical cases in the Southeast Asia region (Eng et al., 2015). Recent investigations in Thailand on the causes of bacteremia discovered that S. Enteritidis accounts for 51.6% of all cases of invasive non-typhoidal Salmonella infections (Phu Huong Lan et al., 2016; Whistler et al., 2018). In a related investigation to determine the major food pathogens isolated in food materials in China, Salmonella, alongside Vibrio parahaemolyticus, and Campylobacter were among the most prevalent (Paudyal et al., 2018). These observations points to the continued risk humans face from these infectious disease pathogens, particularly where they exhibit antimicrobial resistance (AMR).

Gastro-enteric infection as a result of S. Enteritidis has persisted in Malaysia and continues to be a serious public health problem (Thung et al., 2016; Zakaria et al., 2020). The incidence rate has steadily increased in recent years, and this may be attributed to the unhygienic mode of handling and processing of food, especially poultry and poultry products (Thung et al., 2018; Mohan et al., 2019). Although considerable measures to improve food safety has been put in place by the relevant public health authorities, pockets of outbreaks involving large populations continue to be recorded (Balkis et al., 2017; Packierisamy et al., 2018). Other possible reasons why S. Enteritidis continues to be a health burden in Malaysia include the proliferation of rodent reservoirs in poultry farms, households, and public food restaurants, evolutionary selection, and the emergence of antimicrobial-resistance among other evolutionary traits (Porwollik et al., 2005; Betancor et al., 2009; Campioni et al., 2018; Azmi et al., 2021). In Malaysia like many parts of the world, it is believed that resistant S. Enteritidis had been introduced into the country through the importation of poultry meats and poultry products from other endemic countries (Elemile et al., 2019). In light of these explanations, and the need to understand the phenotypic and genomic diversity of circulating S. Enteritidis strains in Malaysia, a Whole-genome sequencing approach was used to assess the phylogenetic relationships, virulence factor determinants, and the antimicrobial-resistance profile of a collection of S. Enteritidis isolated from the Central region of Peninsular Malaysia from 2016 to 2018, with a view to understand the genetic relationship of this important pathogen across different sources (human clinical samples, live birds, and chicken meat).

The Kauffmann White Scheme serotyping tests indicated that all the isolates from the three different sources (human, poultry, and food) belonged to the S. Enteritidis serogroup. However, the overall results suggested that SeqSero2 (SeqSero2 v1.1.0) improved the serotype prediction compared to the Kauffmann White Scheme serotyping scheme. Although 93% of the serotype prediction according to the Kauffmann White Scheme was concordant with the in silico WGS prediction, six isolates were found to conflict as they were predicted to be Brancaster (S6-human), Mbandaka (S18-human), Ohio (S72-poultry), Weltevreden (S77-poultry), and Kentucky (S81-poultry), while S63 (poultry) was found to be either Albany or Duesseldorf because they share the same antigenic formula (8: z4, z24). Additionally, the SeqSero2 also identified one unique isolate with antigenic formula I 4:b:- (S87-poultry), which is not listed in the Kauffmann White Scheme. The antimicrobial susceptibility assay, according to the Clinical and Laboratory Standards Institute (CLSI) standards (Patel et al., 2016), showed that thirty (30) isolates were resistant to at least one of the tested antimicrobial drugs, while twenty-four (24) were resistant to multiple drugs tested (MDR: resistance to three or more antimicrobial classes), and the remaining isolates were all susceptible to the tested antimicrobials. However, only one isolate showed resistance to each of ceftiofur (1 human isolate), cefotaxime (1 human isolate), sulfamethazine/trimethoprim (1 food isolate), gentamicin (1 poultry isolate), and chloramphenicol (1 human isolate), while none of the isolates were resistant to ciprofloxacin. The phenotypic and genotypic antimicrobial susceptibility results are summarized in Tables 1, 2.

The pattern of resistance showed that tetracycline had the highest frequency (37/82, 45.12%), and isolates from human samples showed the highest rate with 17/35 followed by poultry 11/29 and then food 9/18 (Figure 1). The second drug with the highest resistance rate is ampicillin with 4/35 (11.43%) for human, 5/29 (17.24%) for poultry, and 0/18 (0.00%) for food isolates respectively (Figure 2). However, the chi-square statistic to check for any association between the resistances observed and the source of the samples (human, poultry, or food) was found to be 1.772, and the p-value was 0.412. This signifies that the association is not significant at p < 0.05. Similarly, no statistically significant difference was observed based on the ANOVA with a p-value of 0.704 (p > 0.05).

A total of 19 antimicrobial resistance (A.R) genes corresponding to the nine drugs used in the disc diffusion assay were evaluated from the whole genome sequence data. The aminoglycoside resistance gene aac(6′)-ly was detected in 79 of the 82 isolates (96.34%). This observation is in contrast to the phenotypic analysis, which showed that only one gentamicin-resistant isolate (1.22%) was found among all the isolates. Although, the detection of the tetracycline-resistant genes (tetA, tetC, and tetD) 38/82 (46.34%) was found to correspond with the phenotypic result for tetracycline 37/82 (45.12%), albeit with the genotypic detection being higher by one isolate (Figure 2). However, the predominant tetracycline resistance encoding gene was tetC, with only one isolate possessing the tetA gene.

Ten (10) isolates were resistant to beta-lactams (10.98%), and the two genes encoding beta-lactamases identified in these isolates were TEM33 (10/82), and TEM4 (10/82). Not a single isolate was found to show resistance to the quinolone ciprofloxacin based on the disc diffusion method. However, the genes qnrS, and qnrD1 were detected from two human isolates and one poultry isolate. Other resistance genes identified by the whole-genome AMR analysis are dfrA14 (3/82), dfrA15 (3/82), and sul2 (2/82) for trimethoprim and sulfonamide resistance as well as one floR gene (1/82) that codes for chloramphenicol resistance.

Salmonella serotyping using the traditional phenotypic method was the first analysis conducted in this study because of its importance in the characterization of Salmonella, especially during epidemiological investigations and surveillance programs (Diep et al., 2019). However, because of the ease and superior predictability of WGS based in silico methods, SeqSero2 was used to confirm the phenotypic serotyping (Zhang et al., 2019). The results show that the SeqSero2 correctly identified S. Enteritidis in concordance with the Kauffmann White Scheme result with 93% similarity. However, the SeqSero2 was able to identify six serovars other than S. Enteritidis (Brancaster, Mbandaka, Ohio, Weltevreden, and Kentucky), including Albany and Dusseldorf which had the same antigenic formula (I 4:b:-), while one was not available in the database. This finding agrees with earlier studies that indicated that the same antigenic formula could be shared by different subspecies (Diep et al., 2019). In silico WGS-based serotyping has been shown to perform better in identifying Salmonella serotype diversity compared to the traditional method and has become popular among molecular epidemiologists and public health scientists (Diep et al., 2019; Elnekave et al., 2020).

The present investigation also evaluated AMR determinants, virulence factor genes, and phylogenetic relationships of the isolates using a whole-genome sequencing approach. This is because comparative genomics studies have indicated that host specificity (humans, animals, birds, and environment) is a very important factor driving the evolution of new lineages of enteric Salmonella species (Tasmin et al., 2017; Deblais et al., 2018). Secondly, the choice of WGS is because its application permits a broader inference of pathogen characterization, including the prediction of antibiotic resistance and virulence determinants from the sequence (Gupta et al., 2019). The 82 S. Enteritidis, S. Brancaster, S. Mbandaka, S. Albany, S. Ohio, S. Weltevreden, and S. Kentucky isolates identified in this study were cultured from multiple sources, including human clinical samples (blood and feces), poultry (cloacal swab), and food (chicken meat). Upon obtaining the whole genome sequence, EPInod pipeline was used for the analysis of AMR profile, virulence determinants, and the SNP-based phylogenetic analysis.

Antimicrobial usage represents a very critical factor in the emerging public health crisis due to antibiotic resistance. Its extensive application in food animal production to manage clinical infections, as prophylaxis to prevent and control livestock diseases, as well as its use as an additive in feed to enhance growth and productivity have all played roles in the emergence of resistance (Salihu et al., 2013; Wang et al., 2019; Van et al., 2020; Ma et al., 2021). In this study, tetracycline (45.12%) and ampicillin (11.43%) were identified as drugs with the highest phenotypic resistance rate. Similarly, resistance to these drugs was highest among the human isolates (17/35), followed by food isolates (9/18) and then poultry (11/29). These antimicrobials are among those that have been declared critical by the WHO and their use in animals has been restricted. This finding implies that some of these legislation is not being observed and this could lead to the loss of potency of the drugs and progressive emergenc of antimicrobial resistant pathogens. Also, 29.27% of the isolates comprising of S. Enteritidis, S. Mbandaka, S. Ohio, S. Kentucky, and S. Brancaster exhibited multiple resistances to 2 or more drugs. According to studies sanctioned by the Department of Veterinary Services Malaysia under the Livestock Farm Practices Scheme (SALT) program, for Salmonella species isolated from chicken cloacal swabs revealed that 13.5% of the Salmonella isolates were resistant to tetracycline, 5.4% to polymyxin and erythromycin. This is similar to the investigation conducted in China where tetracycline resistance was found to predominate among Salmonella isolates obtained from slaughtered pigs (Wu et al., 2021). In comparison, 2.7% were resistant to chloramphenicol and trimethoprim (Azmi et al., 2018). Despite the resistance observed in this study, the association between the resistance rate and the source of the sample as well as statistical significance based on ANOVA between their means was found not to be statistically significant (p > 0.05). Other drugs that were found to exhibit resistance were chloramphenicol among human isolates (2.86%), gentamicin in poultry (3.45%), sulfamethazine/trimethoprim in food (5.56%), ceftiofur and cefotaxime both in human (2.86%). In a related study conducted in the East Coast of Peninsula Malaysia, Salmonella isolates from broiler chicken were found to exhibit resistance to chloramphenicol (76.2%), sulfamethoxazole/trimethoprim (42.9%) (Ibrahim et al., 2021). This further highlights the crucial role of poultry and poultry products in the dissemination of resistant Salmonella pathogens to humans and other susceptible animal species.

On the other hand, the WGS analysis of AMR genes correlated significantly by identifying some genes that encode resistance to the majority of the phenotypic resistance observed. These included tetracycline, ampicillin, streptomycin, sulfamethazine/trimethoprim, cefotaxime, and ciprofloxacin resistance genes such as tetA, tetC, TEM33, TEM4, floR, aac(6′)-ly, strA, sul1, dfrA14, dfrA15, blaCTX-M, and qnrS1, respectively (Table 2). Although no phenotypic resistance was observed for gentamicin, the AMR genotype result showed that the gentamicin gene (aac(6′)-ly) was the most frequent (96.34%) with isolates from a human source having the highest percentage (97.14%). This observation goes to show WGS as an excellent tool for the accurate prediction of antimicrobial-resistant phenotype in human, animal, and environment samples (Pornsukarom et al., 2018; Liu et al., 2021). The gene aac(6′)-ly is one of the important chromosomal genes responsible for the enzymatic modification of aminoglycoside leading to the development of resistance (Odumosu et al., 2015; El-Badawy et al., 2017). However, it is important to note that mechanism for aminoglycoside resistance include among others decreased uptake and accumulation of the drug in the bacterial pathogen, as well as the expression of aminoglycoside modifying enzymes (AMEs) that causes inactivation of the drug (Mir et al., 2016). Even though, a strong correlation between the phenotypic resistance by disc diffusion and the AMR gene prediction by WGS was observed (sensitivity 20.12% and specificity 91.78%), the result for gentamicin was discordant (Table 3). The most common genes conferring resistance to β-lactam antibiotics in this study were TEM 33 and TEM 4 without corresponding phenotypic consequence. Qualitative determination of antimicrobial susceptibility based on the zone of inhibition is prone to error due to the instability of antibiotic-containing discs under varying temperature and humidity, interference of the functionality of some antimicrobials by the components of growth agar, and the possibility that an isolate may be falsely classified as susceptible (Neuert et al., 2018; Srivastava et al., 2020). This may explain the relative number of mismatches observed in the present study. Moreover, many of the resistance genes detected by the WGS algorithm are plasmid-encoded. In other words, bacterial plasmids are known to commonly cure during storage and sub-culture of the isolates, such that genes detected during sequencing might have become altered (Neuert et al., 2018). Furthermore, Salmonella has been shown to possess silent resistance genes, as well as variants (aac(6′)), which only become transcriptionally active in rare cases (Heider et al., 2009; Frye and Jackson, 2013; Adesiji et al., 2014).

The WGS analysis revealed multiple virulence genes among the Salmonella isolates across the different sources (Supplementary Table 2). These genes included the type III secretion system (T3SS) located within the Salmonella Pathogenicity Sequence I and II, intimin-like proteins, plasmid-encoded fimbriae chaperone protein, host recognition/invasion determinants, as well as magnesium uptake, iron acquisition factors. These genes are responsible for Salmonella infectivity, transmission, and survival (Pornsukarom et al., 2018). The result of the present study indicates that the S. Enteritidis isolated from the poultry cloacal swabs and food sources possessed the same virulence genes as the human clinical isolates. Virulence factors such as adhesins, toxins, and iron ion transport systems are responsible for the adhesion of pathogens to epithelial cells of intestines, which constitute an essential colonization factor during intestinal infections (Sarowska et al., 2019). The detection of a high number of plasmid encoding virulence genes in food, and especially poultry, may serve as a means of transmission of these resistance factors to humans, as observed in this study. The detection of fimbrial genes and a host of other plasmid-encoded genes within the pathogenicity sequences 1–2 among the isolates in this study may signify the potentials of these isolates to cause severe humans infection. Hence, this report will provide valuable reference data that will assist in future investigations of human Salmonella infection. In the same vein, the phylogenetic tree obtained from SNPs exhibited diverse clustering of the isolates with an initial one singleton and two clades that subsequently bifurcated. Phylogenetic analysis based on WGS-derived has proven to be a superior cluster resolution method compared to many of the standard subtyping methods (Mohammed and Thapa, 2020; Ford et al., 2021). In order to determine the genetic relatedness among the isolates and infer their evolutionary history based on a genome-wide scale, across the genomes of all S. Enteritidis and closely related serovars were identified. Phylogenetic analysis based on the maximum likelihood tree produced strong support for the monophyly of the Salmonella isolates with S. Enteritidis being the predominant (Figure 3). Within the S. Enteritidis isolates from the different sources, a split that delineates two sister lineages, Clade B1 and B2 was observed. At the same time, Clade A consisted of 1 branch occupied by the E. coli K12 outgroup, Clade B comprised of the S. bongori reference strain and the other members of the S. enterica serogroup. Sub-clade B2-I which is a branched form the B1 clade, clustered with five of the isolates (S77, S6, S81, S63, and S18) which shared same branches with S. Mbandaka, S. Weltevreden, S. Brancaster, S. Kentucky, and S. Ohio implying close genetic relationship as determined via the SeqSero serotyping. The major bifurcating clade B2-II, on the other hand, contained one monophyletic clade occupied by S. Bandaka and another highly branched sub-clades containing isolates from all the three sources. The fact that samples from poultry, human and food as well as the control S. Typhimurium, and S. Typhi were located in this well-supported sub-clade may indicate similarity in their gene sequence that can be attributed to convergent evolution due to adaptation to the various sources as earlier reported by Didelot et al. (2007). As deduced from the phylogenetic tree in Figure 3, though the split of clade B is suggestive of allele sharing, possibly via horizontal gene transfer and homologous recombination between the clades, which is a common phenomenon in enteric bacteria (Lawrence and Retchless, 2009; Hall et al., 2020). However, the gene transfer may not have occurred to the extent that divergence between the two clades is eliminated. This is in concordance to divergent lineages observed in E. coli following high recombination (Touchon et al., 2009).

An insight into the genetic profile of enteric S. Enteritidis to assess the evolutionary and genetic diversity based on SNPs phylogenomics, as well as understand the virulence and resistance pattern using WGS, was undertaken. The aim was to understand further the genomic diversity existing among S. Enteritidis from different host species. The potential transmissibility of genetic elements was evident due to the clustering of multiple isolates from all the three sources. Most of the virulence factors were also detected from the genome of all the isolates across the human, poultry, and food isolates. However, while the genotypic AMR was mostly congruent with the phenotypic profiles of the isolates, slight mismatches were observed with particular reference to gentamicin. This discrepancy was attributed to their silent nature of the aac(6′)-ly gene whose transcription occurs intermittently. As a whole, the WGS analysis produced a comprehensive understanding of the resistance profile of S. Enteritidis and the ability of the bacteria to adapt to different host species. However, a more in-depth study, including other Salmonella serovars, will be of immense benefit to the understanding of the adaptability and survival mechanism in Salmonella serovars.

This investigation was not able to ascertain the geographical details of the various samples/sources because the isolates were donated as part of stock cultures covering the study period by the agencies involved with the Malaysian National Antimicrobial Resistance Surveillance Program.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

ZZ, LH, SAH, and NA: conceptualization. ZS, NMS, and RA: methodology. ZZ, NMS, and BG: software sequence analysis and interpretation. ZZ and BG: writing—original draft preparation. ZZ, BG, LH, NA, SAH, ZS, NMS, and RA: writing—review and editing. ZZ: funding acquisition. ZZ and LH: supervision.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.