Antimicrobial resistance and virulence genes of non-typhoidal Salmonella isolates in The Gambia and Senegal

Introduction: The prevalence of virulence genes in non-typhoidal Salmonella (NTS) and its association with commonly used antibiotics in West Africa is unknown. Methodology: We tested 185 NTS isolates from children, animals, and food products for the presence of twelve virulence genes by PCR. Ten of the virulence genes tested belonged to the five Salmonella pathogenicity islands implicated in its pathogenesis. Results: Ten of twelve virulence genes except sopE and pefA were present in at least 70% of the isolates tested; sopE and pefA were observed in 33% and 44% of the isolates, respectively. The most prevalent gene was invA (99.5%), which is an invasion gene conserved within the Salmonella enterica. pipD and sopB genes, which were associated with serovar Enteritidis, were detected in 92.4% and 94.1% of isolates respectively. S. Istanbul and S. Javiana, which were isolated from chicken-serving restaurants, carried all the virulence genes of the five pathogenicity islands. There was significant association between sopB, sitC, orfLC, pipD and pefA virulence genes and resistance to commonly used antibiotics in Senegal and The Gambia, namely amoxicillin, ticarcillin, trimethoprim plus sulfamethoxazole, tetracycline, trimethoprim, spectinomycin, streptomycin, sulfonamides and nitrofurantoin. Conclusions: This study shows that virulence genes are present in NTS strains isolated from various sources. The significant association between some virulence genes and antibiotic resistance may have important implications with regard to the spread and persistence of resistance and virulence genes in Salmonella and to the prudent use of antimicrobial agents in humans and animals in West Africa.


Introduction
In sub-Saharan Africa including The Gambia and Senegal, non-typhoidal Salmonella (NTS) are a major cause of invasive disease in infants, young children, and HIV-infected adults and are associated with substantial mortality.For example, a retrospective study in Senegal revealed a prevalence of NTS in patients with HIV to be between 0.15% and 2.13% during 1996 to 2005.In addition, Salmonella serovar Enteritidis and Salmonella serovar Typhimurium were the most common serotypes [1,2].In rural Gambia, NTS represents the second most common blood culture isolate after Streptococcus pneumoniae in children with invasive bacterial disease [3].However, data on the prevalence of the animal origin of NTS in sub-Saharan Africa is limited.In Senegal, NTS infections are known to occur in semi-industrial poultry farms and contaminations were found at various levels of the poultry supply chain (retail chicken and street restaurants): 32% at sale place, 28.6% at farm level, and 20.1% at consumption level (street restaurant) [4][5][6].
The ability of NTS to cause invasive disease is attributed to arrays of virulence genes defined in the Salmonella pathogenicity islands (SPIs) [7].There are at least 60 genes associated with SPIs [8] and the majority of these determinants are located on the chromosome or on large virulence-associated plasmids [9,10].Five major SPIs are well characterized in S. Typhimurium and S. Typhi, namely, SPI1, SPI2, SPI3, SPI4 and SPI5 [8,11], and these SPIs are the basis of pathogenicity in the host.
The emergence and spread of antimicrobial resistance among bacterial species including Salmonella has become a public health threat.Antibiotic resistance, especially to the most commonly used antimicrobials in humans and in animal production systems, is of critical concern in African countries where multidrug-resistance NTS strains are among the most frequent causes of bacteraemia in children.The fact that genetic determinants for both antibiotic resistance and virulence genes could be harboured by the same transferable element [12] implies that there is an association between antibiotic resistance and virulence.Such association could have an impact on the spread of resistance clones of Salmonella.Consequently, assessing the presence of virulence genes as well as the antibiotic resistance status in Salmonella serovars would be useful to better understand its pathogenicity.Our study aimed to investigate the presence of twelve virulence genes in Salmonella serotypes and their association with antibiotic resistance.

Sampling procedure and bacterial isolates
NTS isolates from Senegal were recovered from 57 broiler farms that were randomly selected from a total of 70 farms producing chickens all year round in urban and peri-urban settings.From one flock of each farm, five fresh fecal samples were collected in the chicken coop and pooled.Five chicken carcasses were also bought after slaughtering from each of the 57 farms or sale points, for a total of 285 carcasses.Skin (from the neck) and meat cuts were prepared under aseptic conditions.In addition, forty-two street restaurants that regularly serve chicken dishes were randomly chosen from a group of 50 restaurants in the same area.From each of these 42 restaurants, one pooled sample of three whole servings of chicken meat was collected.The collected samples (feces, skin, muscle, and chicken servings) were put into sterile plastic bags and stored in cool boxes during transportation to the microbiological laboratory where analysis was performed within 24hours.
In The Gambia, NTS isolates were obtained through an active population-based case-control surveillance study designated to determine the aetiology and epidemiology of enteric infections in Gambian children less than five years of age as part of the Gates Enteric Multicentre Study (GEMS).Children under five years of age who presented with severe diarrhoea (i.e., diarrhoea with dehydration, dysentery, or requiring hospitalization) within three days of onset were eligible to participate.For each enrolled child with diarrhoea, one healthy control child without diarrhoea was randomly selected from the community in which the case resided, matching the case by age, gender, and time of presentation.After receiving informed consent from the parent/guardian of each case and control, a single fresh, whole stool specimen was collected from cases and controls and cultured to detect Salmonella.The latter was isolated in the feces of 14 children (eight cases and six controls) enrolled in this study.The enrolled children were traced back to their family compounds and anal swabs were collected from apparently healthy animals (chicken, sheep and goats) residing in the same household within a week of isolating strains of NTS from humans.Five animals per species (chicken, sheep and goats) per compound were enrolled in the study, resulting in a total of 210 domestic animals (chicken, sheep and goats) or 70 per species.Thirty-five strains of NTS were isolated from the feces of 210 contact animals.
Salmonella isolates were characterised using standard protocols [13].Stool specimens were transported in buffered glycerol saline (BGS) to the laboratory and processed within six hours of collection.Stools were plated on Xylose Lactose Desoxycholate (XLD) and MacConkey (MAC) agar and incubated at 36°C for 24 hours.Suspected non lactose fermenter colonies were subjected to biochemical reactions using Analytical Profile Index 20 Enteric (API 20E) according to manufactures' instructions (BioMerieux SA, REF 20 100/20 160).Serotyping was done at the Pasteur Institute Laboratory (Dakar, Senegal) by slide agglutination using Salmonella polyvalent and monovalent O and H antisera (Diagnostic Pasteur, Paris, France) according to the Kauffmann-White classification scheme [14].Due to the large number of Salmonella serovar Brancaster, random selection using the simple random sampling procedure with Excel (Microsoft, Reading, UK) was performed.In the end, 40/148 isolates were considered for the study.The final analysis was conducted on 150 isolates from Senegal and 35 isolates from The Gambia, making a total of 185 isolates.
Amplification was performed in a thermocycler (TECHNE) as follows: one cycle of 94ºC (3 minutes); 30 cycles of 94ºC (2 minutes), 55ºC (1 minute) and 72ºC (1 minute); and the final cycle at 72ºC for 5minutes.The PCR product (10 µl each) was loaded onto a 2% agarose gel containing 500 ng/µl ethidium bromide at 100V for 1 hour in 1X TBE buffer together with two lines of a 100 bp and a 1 kb ladder (ABgene, Epsom, UK) ).The PCR electrophoresis bands were photographed using UV illumination with a gel documentation system (gel Doc 2000; Bio-rad).DNA-free water was used as the negative control and S. Typhimurium ST19 was used as the positive control.

Statistical analysis
Stata Statistical Software: Release 11 (StataCorp LP, College Station, TX, USA) and MATLAB software (MathWorks, Cambridge, MA, USA) were used for data management and statistical analysis.Binary logistic regression was used to study the association between the presence of certain Salmonella virulence genes and resistance to antibiotics with an odds ratios (95% CI) indicating the strength of the associations.We also applied the Fisher's exact test whenever there were not enough numbers of observations to fit logistic regression for certain groups of gene-antibiotics resistance combinations.Thus for such cases p-values reported would be from the Fisher's exact test.We further checked our results for sensitivity to multiple tests and present adjusted p-values using Holm's procedure.Stata 11 (StataCorp. 2009.

Ethical Approval
This research was approved by the Gambia Government /MRC Laboratories joint Ethics Committee under the reference "SCC 1144vs01 26 th Nov 2008 Epidemiology and control of Salmonella in animals and humans in The Gambia (Upper River Region) and Senegal (Casamance)".Written consent forms were explained to and signed by the participants prior the study.

Results
Among the eight diarrhoea cases, four were codiagnosed with malaria, two with fever of more than 39°C, and one with dysentery and respiratory infection (Table 2).The prevalence of the 12 different virulence genes studied was high in all serovars ( > 70%) except for the sopE and pefA genes which occurred in 33% and 44% of the isolates respectively.The invA gene was most frequently (99.5%) observed.All serovars contained at least two genes of the SPI1 with the majority containing simultaneously all five genes (prgH, sopB, sopE, invA and sitC) tested (Table 3).sopE gene was not detected in the most prevalent serovars isolated from animals in Senegal; namely, S. Brancaster, S. Agona, S. Goelzau and S. Kentucky.However, the sopE gene was present in all serovars isolated from children with diarrhoea in The Gambia except in one case from which an undetermined serovar was isolated.All S. Javiana and S. Istanbul which were found in ready-to-eat chicken in restaurants in Senegal contained all the genes except pefA.

Discussion
In this study, we found that the sopE gene was common in serovars isolated from diarrhoeic children, chicken, sheep and goats in The Gambia.Previous studies have shown that the sopE gene was present in strains of S. Typhimurium associated with epidemic disease in both humans and animals [16] and therefore the sopE gene if expressed may be implicated in diseases in both children and animals in The Gambia.In addition, the sopB and pipD genes, which are associated with enteritis in birds [11,15], were present in 94.1% and 92.4% of all serovars tested, respectively, including those isolated from diarrhoeic children.Furthermore, the invA gene was detected in 99.5% of Salmonella serovars in this study.The invA gene encodes for a protein in the inner and outer membrane, which is essential for the invasion of epithelial cells [17].The invA gene is conserved among Salmonella serovars and is a useful marker for molecular detection of Salmonella by PCR [18][19][20][21].Serovars Istanbul and Javiana, which were isolated from chicken meat in Senegal, harboured all twelve virulence genes tested except pefA.Serovar Istanbul is known to be resistant to commonly used antibiotics in Southern Senegal [13].High levels of resistance, especially to trimethoprimsulfamethoxazole, tetracycline, trimethoprim, streptomycin, and sulfonamides, were observed in Salmonella serovars isolated in animals from Senegal [13].Significant associations were found between the presence of some virulence genes, most notably sopB, sitC, orfLC, pipD and pefA, and resistance to commonly used antibiotics in Senegal and The Gambia.These associations could be explained based on the mechanisms involved in the pathogenicity and the acquisition of resistance genes by Salmonella.In fact, in bacteria, the majority of the molecular pathogenicity determinants are located on the chromosome or large virulence-associated plasmids [10,22], whereas antibiotic resistance genes are also often located on extrachromosomal genetic elements or in segments inserted within the chromosome that originate from other genomes [12].The simultaneous presence in the same bacterial cell of a resident virulence plasmid and resistance gene has been frequently reported in Salmonella [12].It has been reported that resistance plasmids carry genes encoding virulence factors [23].Also, antibiotic resistance and virulence genes can be linked in the same replicon, or eventually a single determinant can be involved in both virulence and resistance [23].Although the direct association of virulence and resistance is not determined in this study, much epidemiological data support the idea that, in some cases, antibiotic-resistant organisms may show a decrease in pathogenicity [24][25][26].The opposite situation might also occur [27][28][29].Two of the seven genes, sopB and pipD associated with resistance in our study, were also associated with enteritis or epidemic disease in both children and animals [11,15,16].This result shows that our isolates could also exhibit such epidemic potential since these virulence genes were detected in almost all diarrhoeic children.

Conclusion
This study provides a starting point for the development of appropriate molecular tools to study the epidemiology of NTS in Africa.The limitations of our study were mostly related to insufficient numbers of serovars, which could explain the relatively low occurrence of certain genes.Nevertheless, these findings were worrisome, considering that most serovars detected from food and animal sources harboured virulence genes and thus cause diarrhoea in children.Although transmission from animal to man has not yet been elucidated, these results suggest that animals could play an important role in the epidemiology of NTS.The invA gene, as shown by several studies, remains a suitable molecular tool to diagnose Salmonella in humans as well as in animals and animal products.More detailed studies should focus on various host factors, including age, sex, nutritional status, genetic constitution (including association of virulence and resistance), and the status of the immune system, which affects the outcome of the parasite-host interaction.These further studies would better the understanding of the full mechanism of the pathogenicity of Salmonella.Furthermore, these results have shown the important implications of Salmonella with regard to the spread and persistence of resistance and virulence genes in Salmonella and to the judicious use of antimicrobial agents in humans and animals.

Serovar
TIC CF FOX CTX CAZ TM AN NA PEF NOR STX TE C GM TMP CIP SPT

Table 1 .
[15]lence genes and PCR primers used to screen Salmonella isolates in this study[15]

Table 2 .
Clinical symptoms present in 14 Gambian children with NTS

Table 3 .
Virulence genes detected in Salmonella serovars in The Gambia and Senegal 0

Table 3 .
Virulence genes detected in Salmonella serovars in The Gambia and Senegal (continued)