Assessing the mechanisms of multi-drug resistant non-typhoidal Salmonella (NTS) serovars isolated from layer chicken farms in Nigeria

Background In Nigeria, there have been reports of widespread multiple antimicrobial resistance (AMR) amongst Salmonella isolated from poultry. To mitigate the impact of mortality associated with Salmonella on their farms, farmers resort to the use of antimicrobials without sound diagnostic advice. We conducted this study to describe the AMR patterns, mechanisms and genetic similarities within some Salmonella serovars isolated from different layer farms. Method We determine the AMR profiles of two hundred Salmonella isolates, selected based on frequency, serovar, and geographical and sample type distribution. We also assessed the mechanisms of multi-drug resistance for specific genetic determinants by using PCR protocols and gene sequence analysis. Pulsed-field gel electrophoresis (PFGE) was conducted on seven selected serovars to determine their genetic relatedness. Results Of 200 isolates, 97 (48.5%) revealed various AMR profiles, with the multiple antibiotic resistance (MAR) index ranging from 0.07–0.5. Resistance to ciprofloxacin was common in all the multi-drug resistant isolates, while all the isolates were susceptible to cefotaxime, ceftazidime, and meropenem. Genotypic characterization showed the presence of resistance genes as well as mutations in the nucleotide genes with subsequent amino acid substitutions. Fifteen isolates (43%) of S. Kentucky were indistinguishable, but were isolated from four different states in Nigeria (Ogun, n = 9; Kaduna, n = 6; Plateau, n = 3, and: Bauchi, n = 2). PFGE revealed 40 pulsotype patterns (Kentucky, n = 12; Larochelle, n = 9; Virchow, n = 5; Saintpaul, n = 4; Poona, n = 3; Isangi, n = 2, and; Nigeria, n = 2). Conclusion This study recorded strictly related but diversely distributed Salmonella serovars with high AMR rates in poultry. We recommend strict regulation on antimicrobial use and regular monitoring of AMR trends among bacteria isolated from animals and humans to inform public policy.


Introduction
Salmonellosis as a public health problem cannot be overemphasized.It remains one of the leading bacterial infections in humans, accounting for 93.8 million foodborne illnesses and over 155,000 annual deaths globally [1].
Most Salmonella infections are self-limiting, especially in humans; however, in the case of invasive infections or severe diarrhea, antimicrobial treatment is required.Fluoroquinolones and extended-spectrum β-lactams (particularly cephalosporins) are frequently the first-line antimicrobials used to treat salmonellosis.Fluoroquinolones have the potential to cause arthropathy and, hence, are not recommended in the treatment of infections in pediatric patients.However, fluoroquinolones are still considered one of the last treatment options for life-threatening Salmonella infections caused by multi-drug resistant (MDR) isolates [2].The massive use of antimicrobial agents in livestock production as well as in humans favors the survival of MDR pathogens.Resistant Salmonella isolates recovered from different sources have been frequently reported worldwide [3,4].The increasing occurrence of MDR Salmonella, with resistance not only against the first-line antimicrobials, such as trimethoprim/sulphamethoxazole, chloramphenicol, and ampicillin, but also against WHO-listed critically important antimicrobials for human medicine, such as fluoroquinolones and third-generation cephalosporins (which have been extensively used in human and veterinary medicine), is a significant emerging public health concern [5,6].
With the growing population and the increasing demand for animal protein in Nigeria, commercial farmers entering the poultry section are primed for higher profitability.In a bid to mitigate the impact of mortality associated with Salmonella on their farms, farmers resort to the use of antimicrobials without sound diagnostic advice, which results in the development of resistant isolates [7] and drug residues in the tissue of treated animals [8,9].Despite this practice, information about Salmonella's genetic determinants of resistance to these antibiotics is still scanty in Nigeria.
Very little data is available concerning Salmonella serovars circulating in poultry production in Nigeria [10].Moreover, there is a knowledge gap concerning the variability of the isolates within each serovar.This study was carried out (both from the phenotypic and genotypic points of view) to explore the spread of antimicrobial resistance (AMR) among Salmonella isolated from laying hen farms in Nigeria.Moreover, we investigated the similarities (epidemiological correlation) among the isolates.These data are essential to identify putative sources for defining effective control measures.

Materials and methods
Two hundred Salmonella isolates from a previous study [10] were selected (based on the frequency, serovar, and geographical and sample type distribution) and analyzed to determine their antimicrobial resistance profiles (see also the S1 File).PCR protocols were used to assess the presence of specific genetic determinants of resistance, while for isolates whose mode of resistance to certain antimicrobials (fluoroquinolones and olymyxins) was by gene mutation, gene sequence analysis was carried out.Pulsed-field gel electrophoresis (PFGE) was conducted on the seven most common serovars to determine the genetic relatedness among the isolates.
The EUCAST epidemiological cut-offs were used (11), except for SMX and KAN, where the CLSI breakpoints were considered [12].Escherichia coli ATCC 25922 served as the control strain in this study.We defined AMR as resistance to one or more classes of antimicrobials (there were 8 classes among the 14 antimicrobials) [13], whereas MDR isolates were defined as having resistance to three or more classes of antimicrobials.The formula: multiple antibiotic resistance (MAR) index = (number of antibiotics that an isolate is resistant to)/(total number of antibiotics tested) was calculated for all the isolates [14].

Sequencing of gyrA, parC, pmrA and tem genes
Genetic determinants were identified by DNA sequencing of the amplicons using a Genetic Analyzer 3130XL (Applied Biosystems, Life Technologies Corporation, Carlsbad, California).The sequences obtained were compared using the BLAST program (http://blast.ncbi.nlm.nih.gov/Blast.cgi)to determine point mutations.

Pulsed field gel electrophoresis
For the PFGE analysis, 7 different serovars (S.Kentucky n = 35, S. Poona n = 15, S. Isangi n = 8, S. Saintpaul n = 9, S. Nigeria n = 5, S. Virchow n = 9, and S. Larochelle n = 13) were selected based on the frequency of isolation of the serovar, antimicrobial susceptibility pattern, as well as the geographical and sample type distributions.All 94 selected Salmonella isolates were characterized by PFGE, which was carried out after the digestion of genomic DNA with the restriction enzyme XbaI [19].Gel images were analyzed by BioNumerics version 7.6 (Applied Maths, Sint-Martens-Latem, Belgium) and compared by cluster analysis using the Dice coefficient for similarity and the unweighted pair group method with arithmetic means (UPGMA), with a position tolerance of 2% and optimization of 1%.

Ethical considerations
We obtained written informed consent from farm owners before samples were collected from their farms.We guaranteed the respondents that the data we collected would be kept confidential.Permission was also obtained from the management of the national Veterinary Research Institute (NVRI) and the Federal Ministry of Agriculture and Rural Development.The Research Ethics Committee of NVRI gave the ethical approval for the project.

Determination of antimicrobial resistance profiles
The MIC analysis of the 200 isolates showed that 97 (48.5%) were resistant to at least one antimicrobial agent.Five isolates of S. Isangi, three of S. Kentucky, two of S. Poona and one of S. Saintpaul were resistant to 7 (50%) of the antibiotics tested.Resistance to ciprofloxacin was common in all the MDR isolates (Table 1), while all the tested isolates were susceptible to cefotaxime, ceftazidime, and meropenem (Table 2).The MAR index ranged from 0.07-0.5 (Table 1).
The PCR and sequencing results of resistance genes showed the presence of targeted genes (Table 3).The detection of pmrA, gyrA, parC, qnr, tem, catA1, cmlA1, floR, dfrA5-14, sul2, aac (3)-le, and tetA genes as well as mutations in the nucleotides of parC, gyrA, and pmrA with subsequent amino acid substitution, as shown in Table 4, provides an insight to the AMR of these isolates (see also the S2-S4 Files).
All S. Kentucky isolates were resistant to at least three classes of antibiotics.In particular, the predominant resistance pattern, CIP/GEN/NA/SMX/TET, was observed in 73% (32/44) of the isolates.All S. Kentucky isolates showed resistance to fluoroquinolones and quinolones (ciprofloxacin and nalidixic acid), and this resistance was due to a double mutation of gyrA (Ser83 and Asp87) genes in 95% of the isolates (19/20).One isolate showed only one mutation in the gyrA (Ser83) gene and simultaneously harbored qnrS and qnrA plasmid genes.Only two isolates of S. Kentucky showed the presence of qnrB genes in addition to the chromosomal mutations occurring on the gyrA gene.The tetA and aac(3)-le genes were present in all 20 isolates.Most (95%) of the S. Kentucky isolates showed the sul1 gene (19/20), while only one isolate was positive for the sul2 gene.Eight isolates showed phenotypic resistance to ampicillin with the subsequent detection of the bla TEM gene.The TEM variants observed were TEM1, TEM57, and TEM30 (Table 3).
Two of the S. Poona isolates (2/6) were resistant to ampicillin, chloramphenicol, tetracycline, ciprofloxacin, tigecycline, and sulphomethoxazole.Resistance to ciprofloxacin was seen in most (4/6) of the S. Poona isolates, and this resistance was related to the presence of a mutation in the parC (Thr53) gene in addition to the presence of the plasmid-mediated resistance gene, qnrB.Four isolates showed phenotypic resistances to chloramphenicol, ampicillin, and sulphamethoxazole, with the subsequent detection of cmlA1, tem57, and sul3 genes, respectively.One isolate showed a double mutation in pmrA (T89 and V93) for colistin resistance (Table 4).All (10/10) of the S. Isangi isolates were resistant to the quinolones (ciprofloxacin and nalidixic acid) and in addition, five of the 10 isolates were resistant to azithromycin, chloramphenicol, sulphamethoxazole, tetracycline, and trimethoprim.All but one of the isolates had both qnrB and qnrS genes.One of the isolates had a mutation in gyrA (Leu55) in addition to having the plasmid genes, qnrB and qnrS.Resistance genes sul2 (sulphomethoxazole), tetA (tetracycline), dfrA5-14 (trimethoprim), and catA1 (chloramphenicol) were detected in two of the isolates (Table 3).

Comparison and cluster analysis of seven Salmonella serovars by PFGE
The PFGE results of S. Kentucky, S. Poona, S. Isangi, S. Saintpaul, S. Nigeria, S. Virchow, and S. Larochelle showed varying degrees of similarity/difference between the molecular weight (band) patterns of the restricted segments (pulsotypes).The comparison of PFGE profiles obtained from 94 isolates examined in this study identified PFGE cluster patterns for each of the seven serovars investigated, with significant ranges of genetic similarities (see also the S4 File).S. Kentucky strains were classified into three clusters with different genetic similarities (Fig 1).Cluster 1 in particular, contained five out of 35 strains (three of them were indistinguishable, Dice coefficient = 100%), that were grouped into two different subgroups (Dice coefficient = 97.4%).The remaining 30 strains were grouped into two clusters with 10 different pulsotypes that were characterized by a high degree of genetic similarity (Dice coefficient � 87.7%).Among the strains in cluster 2, 43% (15/35) of the total number of strains analyzed were indistinguishable (Dice coefficient = 100%).Most of the strains in each cluster were from different sample types and geographical locations.
Salmonella Poona strains were grouped into two main clusters with the intra-cluster strains probably being related (Dice coefficient = 85.1%).Each cluster contained pulsotypes with genetic similarity of 96% and 100%, meaning that they were strictly related genetically (in cluster 1) or indistinguishable (in cluster 2) (Fig 2).The strains belonging to cluster 2 were isolated from different sample types (water = 5, feed = 2, feces = 1, bootswab = 1, and dust = 1) and different geographical locations (states in Nigeria: Ogun and Lagos = Southwest, Kaduna = Northwest, Imo = Southeast).Cluster 1 also contained four indistinguishable strains isolated from different sample types and geographical locations.
S. Isangi isolates were classified into two different clusters, probably related (Dice coefficient = 82.8%).Each of the clusters had pulsotypes that were indistinguishable (Fig 3).Each cluster had strains that were from different sample types, but the same geographical locations (states in Nigeria: cluster 1 = Plateau and cluster 2 = Ogun).Salmonella Saintpaul strains were classified into two main clusters, with the intra-cluster strains probably being related genetically (Dice coefficient = 82.1%).In the first cluster, each pulsotype was characterized by subgroups with strains that were genetically similar (89.7%), meaning that they were strictly related.Moreover, within each subgroup, there were indistinguishable strains (dice coefficient = 100%).The most prevalent pulsotype was shared by five of the nine S. Saintpaul strains investigated (Fig 4).
Salmonella Virchow strains were classified into three unrelated clusters (Dice coefficient = 63.3%).Cluster 1 had two pulsotypes with three subgroups that were strictly related (93.3% and 98.3%), and two of the subgroups were characterized by indistinguishable strains (Dice coefficient = 100%).Clusters 1 and 2 were characterized by strains that were probably related (Dice coefficient � 78.9%).Cluster 3 had two isolates that were indistinguishable (Fig 6).Salmonella Larochelle was classified into two main clusters that are probably related (Dice coefficient = 85.4%) (Fig 7).The two clusters contained different pulsotypes with genetic similarity of 91.6% to 91.9%, respectively, for clusters 1 and 2, meaning that the intra-cluster pulsotypes were strictly related genetically.

Discussion
High rates of resistance to ciprofloxacin, nalidixic acid, sulphamethoxazole, tetracycline, and gentamicin were found in this study.Resistance rates to chloramphenicol, ampicillin, trimethoprim, azithromycin, tigecycline, and colistin were low.A comparative study in Ibadan, Nigeria, reported a high frequency of resistance to ciprofloxacin, tetracycline, nalidixic acid, and sulphamethoxazole among chicken isolates, supporting our findings [20].Another study, [21], reported 100% resistance to fluoroquinolones for clinical isolates from chickens in the northern part of Nigeria, while [22] reported only one isolate from a chicken that showed ciprofloxacin resistance out of the thirty investigated isolates.Twelve isolates (S.Poona n = 2; S. Saintpaul n = 2; S. Larochelle n = 3; S. Chomedey n = 2; S. Telekebir n = 2, and; S. Kingston n = 1) in this study showed low susceptibility to ciprofloxacin and high susceptibility to nalidixic acid as was earlier reported by Fashae et al. [20].
All the S. Kentucky isolates were resistant to ciprofloxacin and nalidixic acid, while most were resistant to tetracycline and sulphamethoxazole.The multi-drug resistance displayed by Salmonella isolates in this study could suggest the emergence of co-resistance and/or crossresistance.For example, low levels of resistance to colistin were seen, and this may be due to cross-resistance from ciprofloxacin, tetracycline, and sulphonamide resistances [23,24].Coresistance to fluoroquinolones and third-and fourth-generation cephalosporins has also been identified in Salmonella isolated from humans [25].
Raufu et al. [7] reported significant contamination of poultry with multidrug-resistant S. Kentucky in Nigeria.The association of S. Kentucky with poultry dates back to 1937 in the United States, where it was isolated from chickens.Although multilocus sequence typing (MLST) was not carried out in this study, the patterns of resistance to antimicrobials we tested reaffirmed the initial speculation of a circulating, poultry-associated STI98-X1 CIP S. Kentucky clone in Africa, including in Nigeria [26].The lack of policy in Nigeria to control the use of antimicrobials in poultry, especially fluoroquinolones, including ciprofloxacin, enrofloxacin, and ofloxacin [27], could have contributed to the rapid spread of AMR in poultry farms.In Salmonella, the acquisition of resistance genes likely occurred by conjugation, usually with other Enterobacteriaceae, or through the transfer of plasmids [28].Bacteria that develop resistance via extended-spectrum β-lactamase genes could become a reservoir of resistance genes, which can enter the food chain [29].Screening of resistance genes has proved to be a useful tool in the epidemiological tracing of genes between bacterial species, prediction of future outbreaks in different regions, and treatment and control strategies of bacterial diseases [30].
Resistance to gentamicin involves enzymatic modification of aminoglycosides [31].These are acetylation, phosphorylation, and adenylation of hydroxyl and amine groups.Although aminoglycoside modifying enzymes are the leading cause of resistance, plasmid-mediated 16S rRNA-methylases (16S-RMTases) have been increasingly reported in recent years [32].However, 16S-RMT-ases were not investigated in this work.There are scarce reports on aminoglycoside resistance genes in Salmonella species [30,33], but we detected the gene for aminoglycoside modifying enzyme, aac(3)-le, in less than half of the isolates screened for gentamicin resistance.There is a similarity between the aminoglycoside gene detected in this study with that from E. coli in China [34].Aminoglycoside-resistant isolates from foods of animal origin are of public health importance in developing countries because this class of antimicrobial is used to treat a wide variety of infections [35].
Tetracycline-resistance genes were the most frequently occurring resistance genes in this study, and all of the tetracycline-resistant isolates yielded targeted amplicons with tetA primers, but there were no amplification products detected by tetB and tetG primers.In this study, both tetracycline-susceptible and -resistant isolates harbored the tetracycline-resistance genes.These present results agree with those reported by both Deekshit et al [36] and Maynard et al [37], who showed that some antimicrobial-resistance genes of bacteria can be silent when resistance testing is carried out in-vitro.These silent genes, under the selection pressure of antibiotic use, can be transferred to other bacteria or be turned on in-vivo [38].All Salmonella isolates that were resistant to sulphamethoxazole carried either the sul1, sul2, or sul3 genes, with sul1 being the most predominant.Most isolates resistant to sulphamethoxazole also showed resistance to ciprofloxacin, nalidixic acid, and tetracycline.The dfrA5-14 gene was found in isolates that were resistant to trimethoprim.The catA1, cmlA1, and floR genes were detected among the chloramphenicol-resistant isolates tested.In Nigeria, chloramphenicol was previously the antibiotic of first choice for the treatment of Salmonella-related illnesses; however, a report found that chloramphenicol resistance increased by 72.4-89.2% between 1997 and 2007, limiting this drug's therapeutic usefulness [22].
Sequencing of the bla TEM genes revealed three variants.These were TEM 57 (substitution of amino acid G with D at position 90), TEM 30 (substitution of amino acid R with S at position 241), and TEM 1 (the reference sequence).Bae et al. [39] reported that ampicillin-and/or cephalosporin-resistant Salmonella strains can carry the bla TEM gene, with or without the bla CTX-M or bla OXA gene.However, only the bla TEM gene was detected in this study.This further confirms the low rates of resistance by these Salmonella isolates to the extended-spectrum β-lactam antibiotics, especially the cephalosporins.Mutations were observed in the pmrA gene (T89!S and V93!G) for colistin resistance.This amino acid substitution in the pmrA gene has also been reported by Hong and Ko [40].
Resistance to fluoroquinolones typically arises as a result of modifications in the target enzymes (DNA gyrase and topoisomerase IV), changes in drug entry/efflux, and plasmidmediated quinolone (qnr) resistance, which protects the quinolone target proteins from inhibition [41].Genes regulating the influx and efflux of quinolones into/from cells were not investigated in this study.The genetic mutation in fluoroquinolone resistance is a step-wise phenomenon, affecting both gyrA and parC genes [42].Most investigators have reported various substitutions of amino acid at different positions in the gyrA gene [43], but the substitution in this study, which was Leu55!Phe in the gyrA subunit, is extremely rare.Contrary to the report of Wang et al. [44], the mutation at position 87 has no comparative advantage for ciprofloxacin resistance over the mutation at position 83.Previous studies have not been able to detect gyrB or parE genes [42,45].Although gyrB and parE gene mutations were not detected in this study either, we did detect particular substitutions in gyrA and parC, which are considered major quinolone resistance mechanisms [42,45].
Different plasmid-mediated quinolone resistance (PMQR) genes (qnrA, qnrB, qnrC, qnrD, qnrS, qepA, aac(60)-Ib-cr) have been described [46], but only qnrA, qnrB, and qnrS genes were detected in this study.While S. Kentucky, S. Saintpaul, and S. Isangi harbored the qnrS gene, only S. Kentucky and S. Muenchen had the qnrA gene.The PCR and sequencing results showed that plasmid-mediated quinolone resistance (qnr) genes were responsible for the resistances seen in four of the isolates that were resistant to ciprofloxacin but susceptible to nalidixic acid, even though two of the isolates had a single mutation in the parC gene.
In previous studies, researchers have demonstrated the relationship between the increased prevalence of antimicrobial-resistant bacteria and (i) the increased use of antimicrobials in human and veterinary medicine, (ii) greater movement of people and animals, and (iii) increased industrialization [47].Ogun State, located in the southern part of Nigeria has a large poultry industry.This area not only meets the protein needs of its people through the production of day-old chicks and poultry products, but it also supplies other areas and imports animal feeds and ingredients from other regions [48].This movement is a potential contributor to the spread of antibiotic-resistant bacteria [49].The environment itself also contains a variety of bacteria that potentially present an immense pool of AMR genes, which if transferred between bacteria, could cause human and animal disease [50].The high level of resistance observed in this study reaffirms the great importance of strengthening collaboration between veterinary and public health sectors on appropriate detection and reporting of zoonotic foodborne pathogens [51].Despite bans and legislation, some antibiotics are still fed to livestock prophylactically and routinely to increase profits and to limit potential bacterial infections in stressed and crowded livestock [52].Antimicrobial-resistant bacteria will inevitably follow wherever antimicrobials are used; therefore, a coordinated, multi-disciplinary approach will be required to address this issue [53].Monitoring AMR trends among bacteria isolated from food, animals, and humans is necessary to inform public policy regarding the appropriate use of antimicrobial agents in veterinary and human medicine [54].
Many studies have reported the successful use of PFGE for assessing the genetic diversity of Salmonella isolates [55] and the clonal transmission of isolates in a country, a food industry, or a herd [56][57][58].Veterinarians, food producers, and risk managers can utilize the PFGE-based database to trace the sources of contamination on farms and in food processing facilities.In the present study, fewer than half of the S. Kentucky isolates were indistinguishable or were possibly epidemiologically related, but they were isolated from four different states.PFGE has been successfully used for the sub-typing of several Salmonella serovars and other bacterial pathogens [59][60][61] as well as in the context of foodborne disease investigations.The results of the present study provided a baseline molecular PFGE database of the most prevalent Salmonella serovars in poultry in Nigeria, upon which other research works can be built.
Some Salmonella serovars are known to be associated with specific sources [62].S. Derby, S. Montevideo and S. Kottbus have often been linked to pig, cattle, and poultry production systems, respectively [63].Other serovars may also be linked to various sources.There was no specific association between most of the PFGE pulsotypes in this study and the sample types.
For example, all 20 indistinguishable S. Kentucky isolates were from the five different sample types analyzed.

Limitation
This study is not without limitations, as it was impossible to purchase primers for some resistance genes.It was also not possible for us to carry out multilocus sequence typing (MLST) to confirm whether the S. Kentucky isolates were the multi-drug resistant ST198 XI strains that are of public health concern.

Conclusion
This study showed high resistance rates of Salmonella isolates to fluroquinolones, with subsequent detection of relevant resistance genes especially amongst the S. Kentucky isolates.Resistances to other commonly used antibiotics in animal production were also detected.This could be attributed to the unregulated sales and use of antibiotics in the poultry industries in Nigeria.Although the Salmonella isolates were obtained from farms that were far from each other and collected during different time frames, a high level of genetic similarity was evidenced, demonstrating the presence of strictly related clones within the Nigerian layer farms.Based on our findings, we recommended that the antibiotic sensitivity of Salmonella isolates should be routinely verified by clinicians to determine each isolate's susceptibilities before prescribing or administrating antibiotics.Moreover, antimicrobials cannot be considered the solution to control Salmonella in poultry flocks.Some other alternatives, such as the application of biosecurity measures and vaccination of animals, must be considered, as they more effectively limit the spread of Salmonella (Regulation EC N˚1177/2006).There is a need for regulatory authorities to monitor AMR trends among bacteria isolated from food, animals, and humans, which will inform public policy regarding the appropriate use of antimicrobial agents in veterinary and human medicine.The high rate of antibiotic resistance in Salmonella observed in this study reaffirms the great importance of strengthening collaboration between veterinary and public health sectors on appropriate detection and reporting of zoonotic foodborne pathogens.