Characterisation of novel strains of multiply antibiotic-resistant Salmonella recovered from poultry in Southern Senegal

Introduction: Non-typhoidal Salmonella (NTS) contamination in poultry and poultry products is a major cause of food-borne disease in humans. This study presents the molecular epidemiology of NTS isolated from poultry in Senegal. Methodology: A total of 261 NTS recovered from broiler farms, chicken carcasses and street vendors were characterized using random amplification of polymorphic DNA (RAPD) and multilocus sequence typing (MLST) techniques. Results: We observed 20 distinct RAPD profiles corresponding to 18 different serotypes. Strains from each of these 20 groups were further analysed using MLST. Consequently, 12 new MLST alleles and 17 new sequence types were discovered. Three sequence types (S. Kentucky ST198, S. Agona ST13 and S. Istanbul ST33) have previously been described in Senegal and other countries, suggesting that these clones are geographically widely distributed and are circulating in a wide range of hosts. Nine clones showed multi-resistance to the most commonly used antibiotics in both humans and animals. However, a novel multi-resistant clone of S. Kentucky ST832 was found. Conclusion: This study gives new insights into the genetic diversity of NTS in Senegal. Molecular tools remain essential to improve our understanding of the epidemiology of NTS by tracking the sources of infection and/or contamination.


Introduction
Enteric diseases caused by Salmonella in chickens are of great concern because they are an important cause of mortality and morbidity [1].Poultry and its products, in particular chicken meat and eggs, are a major source of Salmonella infection [2].In Senegal, several studies have reported the presence of Salmonella along the poultry production supply chain and its public health impact [3][4][5][6][7], withserotyping being used to characterise the Salmonella [8].Due to the ubiquitous nature of Salmonella, a typing scheme capable of a more detailed strain identification is essential for epidemiological studies [9], because the ability to distinguish isolates of Salmonella is very important to trace the source of infection and outbreaks.Several methods have been used for deciphering the relatedness among isolates but some have low discriminating power, whereas others demand a considerable amount of expertise, time and equipment [10].Multilocus sequence typing (MLST) was found to have a low discriminative power, but is easier to interpret and to compare between laboratories and provides the best phylogeneticrelationship inferences [11].Our study reports for the first time in Senegal the characterization of the most prevalent NTS strains isolated from poultry and poultry products by using the random amplification of polymorphic DNA (RAPD) and MLST techniques.

Bacterial isolates
From October 2007 to July 2008, a total of 261 strains of Salmonella were collected during a crosssectional study in 57 broiler farms, 285 chicken carcasses from farms/sale points, and 42 dishes of chicken meat from street restaurants in southern Senegal [7].Salmonella was cultured according to the standard technique ISO 6579 (1998, International Organization for Standardization, Geneva, Switzerland).Serotyping was done by slide agglutination using Salmonella polyvalent and monovalent O and H antisera (Diagnostic Pasteur, Paris, France) according to the Kauffmann-White classification scheme [8].The data of the antimicrobial resistance tests were extracted from a previous study [7].

RAPD-PCR reaction conditions
The RAPD reaction was performed on all 261 isolates using P1254 primer as previously described [12].The reaction was as follows: 3 µl of primer (20 mM) was added to 10 µl of DNA free water (Invitrogen, Paisley, UK), 5 µl of 5x Buffer (Qiagen, Crawley, UK), 2 µl of dNTPS (2 mM), and 3.5 µl of MgCl2 (25 mM), 0.5 GoTaq polymerase (Promega, Madisson, UK) and 1 µl of template (cells) to get the final volume of 25 µl.Amplification was performed in a PCR machine (Techne, Hatboro, USA) as follows: one cycle at 94ºC (2 minutes); a series of 3 cycles at 94ºC (2 minutes), 35ºC (1 minute) and 72ºC (2 minutes); a second series of 34 cycles at 94ºC (10 seconds), 40ºC (20 seconds) and 72ºC (2 minutes); with final extension at 72ºC for 5 minutes.The amplified product was stored at 4ºC until required.The PCR product (10 µl each) was loaded on a 1.5% agarose gel containing 0.5ul/ml ethidium bromide at 25 V overnight in 1X (SIGMA, Steinheim, Germany).Two lanes of 50 bp and 100 bp of ladder were included in each gel for reference.The RAPD electrophoresis bands were photographed using UV illumination with a gel documentation system (gel Doc 2000, Bio-Rad, Hertfordshire, UK).Gel pictures were analysed with Bionumerics software (version 4.0; Applied Maths, Saint-Martens-Latem, Belgium) and finally similar profiles resolving into similar clones were selected.

Multi locus sequence typing
MLST was performed on 20 randomly selected Salmonella serotypes with unique RAPD profiles.The protocol was followed as previously described [13] .Briefly, the seven genes were targeted were aroC, dnaN, hemD, hisD, purE, sucE and thrA.Amplification of all genes was performed with a 25 µl reaction mixture of the following: 10x Buffer (Applied Biosystems, Foster City, USA) with 1.5mM MgCl 2 (2.5 µl); 2 mM dNTP (0.5 µl); 12.5 mM forward primer (1 µl); 12.5 mM reverse primer (1 µl); 5U/µl Qiagen Hotstart Taq Polymerase (Qiagen, Crawley, UK) (0.25µl); template (cell lysate) (2 µl) and 17.75 µl sterile DNA free water.PCR cycling conditions were 10 minutes at 94°C, followed by 32 cycles of 94°C for 1 minute, 55°C for 1 minute and 72°C for 1 minute, with a final extension at 72°C for 5 minutes.From each PCR, 2 µl aliquots were separated by 1% agarose gel electrophoresis and visualized with ethidium bromide staining and UV illumination, and using a gel documentation system (gel Doc 2000, Bio-Rad, Hertfordshire, UK).PCR products were purified using Qiagen kit (Qiagen, Crawley, UK).Sequencing was done on both strands with BigDye Terminator Cycle Sequencing kit (Applied Biosystems, Warrington, UK).The labelled fragments were separated by size using a 3130xl Genetic Analyser (Applied Biosystems, UK).Sequences were edited and complementary sense antisense fragments were aligned using the Laser Gene DNA star 7.1 software.Finally, the sequences were submitted to the MLST database website [14] and assigned to existing or novel allele type numbers defined by the database.New sequence types were arbitrarily assigned a number for the purpose of the analysis.

Cluster analysis and mapping
To perform the cluster analysis of the serotypes, MLST data were analysed with Bionumerics software (version 6.5; Applied Maths, Sint-Martens-Latem, Belgium).Analysis using a hierarchic unweighted pair group method (UPGMA) with averaging was used to generate a dendrogram describing the relationship among Salmonella serotypes.A minimum spanning tree was used to compare Senegalese clones to Gambian and other African clones.The mapping of the locations of the cases was performed using Arc Gis 9.3 software.

Results
In this study, we analyzed the genomic relatedness between 261 NTS belonging to 18 different serovars, initially using RAPD profiling.All RAPD fragments amplified were between 200 bp and 3000 bp.Twenty different RAPD fingerprints were observed among the 261 NTS (Figure 1).Each serovar generated a unique RAPD fingerprints, with the exception of two serovars, S. Javiana (Figure 1, lanes 14 and line 15) and S. Kentucky (Figure 1 1).The strains characterised belong to three clusters showing a DNA identity of 98.8 to 100% at MLST loci.Most of the antibiotic sensitive strains belonged to a different cluster from that containing predominantly resistant strains (Figure 2).The Senegalese and Gambian serovars cluster together and these are derived from the ST19, ST626 and ST937 clonal complexes (Figure 3).Interestingly, both S. Javiana RAPD types belonged to the same novel sequence type (ST830) while both S. Kentucky RAPD types had different ST (ST198 and ST832).Ten of the strains tested were resistant to multiple antibiotics, with two resistant just to spectinomycin and eight strains were fully susceptible to all tested antibiotics (Table 1).

Discussion
Salmonella is frequently isolated from poultry products in Senegal as shown by several studies [3][4]6] and confirmed in our previous study [7], with serotypes S. Brancaster, S. Hadar, S. Kentucky, and S. Agona found most commonly in poultry and poultry products.
RAPD profiling has been widely used to assess the genetic relatedness between Salmonella isolates indicating the reliable discriminatory power of this technique in subdividing strains within serotypes [1,10,12,[15][16].MLST has also been considered a method of choice for detailed characterization of a number of pathogens [14,[17][18].Although some authors have reported that MLST is not suitable to distinguish between closely related strains within the same serotype [19][20],in this study we showed that MLST is able to distinguish strains belonging to the same serotype, as in the case with the two S. Kentucky serotypes.Furthermore, both MLST and RAPD provided subtypes for S. Kentucky, meaning that MLST and RAPD techniques are more distinctive than serotyping for S. Kentucky; this result concurs with those of other studies that showed that MLST was more sensitive than serotyping [21][22].The S. Kentucky ST198 and S. Kentucky ST832 did not share any alleles and while the former was fully susceptible to all antibiotics, the latter was multi-resistant to four antibiotics.
The high genetic similarity among serovars is not surprising as the clones were sampled from poultry farms in the same geographic urban area in Southern Senegal.The antibiotic sensitive strains cluster together and have been shown to belong to a new set of strains which is not frequently isolated in poultry in Senegal [7], Their high rate of sensitivity could be explained by the fact that they have not yet been under stress (antibiotic treatment) and thus have not adapted themselves to the environment by acquiring resistance determinants [24].
The currently accepted "gold standard" for the characterization of NTS is PFGE, and this, along with other methods such as plasmid profiling, could also be used to determine the diversity of isolates within this study.S. Kentucky ST198 has been   previously isolated from humans in Senegal and from food and cattle in the United States (USA).Our study revealed its presence in chicken skin in southern Senegal.S. Agona ST13 found in chicken skin in our study was also previously described in Denmark from both human and veterinary sources, from milk in Ireland, from humans in Scotland, and from undetermined hosts in Australia, Germany, the USA, Canada and Spain [23], while S. Hadar ST33 was isolated from various sources in several European and American countries and South Africa [23].However, the ST33 found in southern Senegal in our study belonged to serovar S. Istanbul.Collectively, ST13, ST33 and ST198 clones are widely distributed throughout the world and all three clones have been simultaneously identified in both human and animal sources, suggesting that the same clones are circulating in both hosts.S. Istanbul ST33 was multiresistant to most commonly used antibiotics in humans and animals in southern Senegal, contrary to S. Kentucky ST198 and S. Agona ST13, which were fully sensitive to all antibiotics tested.However, a novel multi-resistant clone of S. Kentucky ST832 was found in poultry.

Conclusion
Our study showed that a variety of serovars and clones is circulating in southern Senegal; furthermore, 17 out of the 20 clones characterised in this study have never been reported elsewhere in the world.This is the first report describing the use of MLST to characterise the most common NTS in poultry and poultry products in Senegal.The discovery of new alleles and sequence types is very useful for a better understanding of the epidemiology of NTS in Senegal.

Figure 2 .
Figure 2. UPGMA dendrogram based on the similarity matrix showing the relationship between the concatenated seven housekeeping gene fragments from 20 Salmonella strains

Figure 3 :
Figure 3: Clustering of STs using minimum spanning tree