Introduction

The rapid emergence of antimicrobial resistance (AMR) in the global ecosystem has become a threat to human, animal and environmental health [1,2,3]. Human, animal and environmental reservoirs contribute to AMR. Aquaculture production has been identified as a hot spot for the development of AMR, and transfer of drug-resistant microorganisms between food producing animals and humans [4, 5]. In particular, fish are reservoirs of zoonotic disease, infecting both the host and humans through food-borne disease or direct contact at the aquaculture facility [6].

Aquaculture sector is one of the fastest-growing sectors in the food industry, providing fish for human consumption as a source of protein and fatty acids [7]. Nile tilapia (Oreochromis niloticus Linnaeus, 1958) is the most popular fish in Kenya due to its palatability and economic price and as a resultthere has been rapid expansion of aquaculture by most farmers [8, 9]. In addition, the demand for Nile tilapia in local and international markets has intensified its production by farmers in Kenya. Nile tilapia was first cultured in Kenya in 1924 to boost the livelihoods of communities and to improve nutrition [10, 11] and is currently the most cultured fish speciesrepresenting about 90% of the national aquaculture production [12]. However, contamination is one of the main challenging factors either in the ponds or during harvesting or marketing, which can be a source of pathogens and may be a potential source of infection to humans [13]. Contaminated fish are unsuitable for human consumption since they can be a source of pathogenic bacteria.

Despite the high nutritional quality that links fish consumption to positive health effects in humans, the unsanitary conditions at fish farms and the occurrence of superbug bacteria in fish products have been reported as worrisome observations. This could pose a threat to human health, especially at these times when the demand of Nile tilapia as a source of animal protein in Kenya seems to be on the increase [12]. Aquatic ecosystems are vulnerable to many contaminants such as chemicals and drug residues as they are the recipients of run-offs from agricultural or livestock farms and healthcare facilities [14, 15]. Intensive fish farming has increased the uncontrolled use of antibiotics in the treatment of infections and as growth promoters resulting in the emergence of resistant bacteria strains [16]. This is of importance to human health as fish and fish products may be an important vehicle for the dissemination antibiotic-resistant pathogenic bacteria to other bacteria or directly to humans. Fish and fish products contaminated with human pathogens have been reported in many countries with Salmonella spp., E. coli, Staphylococcus species, Vibrio species, and Pseudomonas aeruginosa being the most important pathogens [17, 18].

Salmonella spp. and E. coli are the most important zoonotic bacterial pathogens that cause foodborne illnesses worldwide, and deaths due to the consumption of contaminated animal products [19]. Fish and fish products contaminated with human pathogens have been reported in many countries, with Salmonella spp. and E. coli, being the most common pathogens [17]. Salmonella spp. is not part of the healthy fish microbiota and its presence indicates fecal contamination either from polluted water or cross contamination during the production chain (unhygienic fish handling, fish processing or marketing) [20, 21]. Pathogenic E. coli can cause severe foodborne disease and is a pathogen of concern because of the prospective challenges of treatment when humans are infected. Studies targeting the contamination of fish with Salmonella spp. and E. coli are therefore important as there is limited information particularly in the study area, despite the increased popularity of fish consumption.

Studies have reported the presence of antibiotic-resistant Salmonella spp. and E. coli in fish and fish products [18, 22, 23]. Antibiotic resistance is growing and has affected critically important classes of bactericidal antibiotics used to treat bacterial infections in humans [24]. One of the most important resistance mechanisms in Gram-negative bacteria against beta-lactam antibiotics is induced by the production of beta-lactamases [24]. Extended β-lactamases (ESBL) are plasmid-mediated β–lactamase enzyme recognized for their remarkable ability to hydrolyze penicillin, 3rd and 4th generation cephalosporins and monobactams except for carbapenem and cephamycin [25]. These enzymes emerged from blaTEM-1, blaTEM-2 and blaSHV as narrow-spectrum parents. Recently, blaCTX-M, a new class of ESBL genes, appeared to have gained global attention. The rates of CTX-M producing bacteria have increased worldwide and the situation is more complicated as these enzymes confer co-resistance to other commonly used antibiotic classes [26,27,28]. In aquaculture, a variety of these antibiotics are authorized for use, resulting in the emergence of ESBL-producing Gram-negative bacteria.

The occurrence and increase of bacterial strains resistant to routinely used antibiotics in fish hatcheries and their possible human health implications is calling for intensified surveillance systems. There is limited scientific data on the antimicrobial resistance of food pathogens in fish from retail markets in Kenya. Therefore, monitoring the prevalence of antibiotic resistance microorganisms is necessary to provide knowledge about the magnitude of the problem and help government authorities to evaluate the effectiveness of the control measures. In this study, we hypothesized that the Salmonella spp. and E. coli contaminants isolated from Nile tilapia marketed in the Nairobi region of Kenya are highly resistant to multiple antibiotics. The objective of this study was to establish the occurrence, virulence potential and antimicrobial resistance of Salmonella spp. and E. coli isolated from fresh Nile tilapia fish from retail markets. Additionally, the genetic diversity among the multi-drug resistant Salmonella spp. and E. coli isolates from Nile tilapia from retail markets was determined. Furthermore, for every MDR Salmonella spp. and E. coli isolate, ESBL production and ESBL gene presence were determined by double-disc synergy test and polymerase chain reaction (PCR) tests, respectively.

Materials and methods

Study site and fish sample collection

The study was conducted in five sub-Counties (Kasarani, Langata, Westlands, Embakasi and Makadara) under the administration of Nairobi County, Kenya. The five sub-counties and their locations were Kasarani (latitude − 1.227841 and longitude 36.905729), Makadara (latitude − 1.296140 and longitude 36.871042), Westlands (latitude − 1.2683 and longitude 36.8111), Embakasi (latitude − 1.3000 and longitude 36.9167) and Lang’ata (latitude − 1.366667 and longitude 36.733333). This was a cross section study in which a total of 68 fish samples were collected retail markets of the five sub-counties in Nairobi, Kenya. The samples collected from each of the sub-Counties were 14 each for Kasarani, Makadara, Westlands sub-Counties and 13 each for Embakasi and Langata. The number of fish collected in every market depended on the availability of fish vendors. The fish samples were collected in sterile zip-lock polypropylene bags and transported in cool boxes with ice packs to the Microbiology section at the Department of Biochemistry, University of Nairobi.

Isolation and identification of presumptive Salmonella spp. and Escherichia coli on selective and differential media

The media used were xylose lysine deoxychocolate agar (HIMedia Laboratories Pvt. Mumbai, India), triple sugar iron (TSI) agar (HIMedia Laboratories Pvt. Mumbai, India) (for Salmonella spp.), brain-heart infusion (BHI) agar (HIMedia Laboratories Pvt. Mumbai, India) (for both Salmonella spp. and E. coli), and MacConkey Agar (HIMedia Laboratories Pvt. Mumbai, India) and Eosin Methylene Blue (EMB) agar (HIMedia Laboratories Pvt. Mumbai, India) (for E. coli). Fish were aseptically dissected to obtain 15 g of the sample (the gills and flesh of fish) which were added to 50 ml buffered peptone water (HIMedia Laboratories Pvt. Mumbai, India) and homogenized using a stomacher 400 circulator (Seward Ltd, England). Five milliliters of each tissue homogenate was analyzed for any enterobacteriaceae. The homogenate was inoculated on Brain-heart infusion agar. The inoculated plates were incubated at 37 °C for 24 h. Sub-culturing was done on MacConkey, Eosin Methylene Blue (EMB) and Xylose Lysine Deoxychocolate agar plates to obtain pure cultures of the respective bacteria isolates [29]. The bacterial isolates were confirmed by standard morphological characteristics (shape, size, surface texture, margins and elevation, colour and opacity) and biochemical tests as described by Cheesbrough [29]. The biochemical tests performed were triple sugar iron test (TSI), methyl red test (MR), Voges-Proskauer test (VP), citrate utilization test (CIT), urease test (URE), catalase test (CAT), oxidase test (OX) and hydrogen sulphide (H2S) production test.

Enumeration of total bacterial load

The homogenized samples were serially diluted (10− 1 to 10− 3) and cultured onto their respective media [30]. The plates were inverted and incubated at 37 °C for 18 to 24 h [29]. The bacterial load was determined by the counting the number of discrete colonies using the viable plate count method [31].

Antimicrobial drug susceptibility testing

Antimicrobial drug susceptibility testing of Salmonella spp. and E.coli isolates to antibiotic discs (HiMedia, India) of penicillin (10 µg), vancomycin (30 µg), rifampicin (5 µg), ampicillin (10 µg), cefepime (30 µg), cefpodoxime (10 µg), chloramphenicol (30 µg), nitrofurantoin (300 µg), ceftazidime (30 µg), meropenem (10 µg), and streptomycin (10 µg) was determined on Mueller Hinton agar plates using the Kirby–Bauer agar disc diffusion method [32]. These antimicrobials were selected based on the availability and upon the recommendation of World Health Organization and World Organization for Animal Health (OIE) for use in both human and food-producing animals [33, 34]. The test organism was uniformly seeded over the Muller Hinton Agar (MHA) (Oxoid Basingstoke, England) surface and exposed to the concentration of the antibiotic and incubated at 37 °C for 24 h. Inhibition zone diametersaround the discs were measured to the nearest millimeters and classified as resistant, intermediate or susceptible as per the criteria of Clinical Laboratory Standards Institute 2021 [35]. Salmonella spp. and E. coli isolates resistant to antibiotics from three or more antimicrobial classes were defined as multi-drug resistant (MDR) isolates.

Determination of multiple antibiotic resistance (MAR)

Multiple antibiotic resistance (MAR) index was analyzed as described by Krumperman [36]. MAR index was calculated by dividing the number of antibiotics to which the test isolate depicted resistance to the number of antibiotics to which the test isolate was evaluated for susceptibility. Multiple antibiotic-resistant phenotypes (MARPs) for each sampling site were generated for isolates that showed resistance to more than three antibiotics following the method described by Wose et al. [37]. The antibiotic resistance pattern, number of antibiotics to which the isolates were resistant, frequencies and percentages were obtained from Kirby Bauer tests.

Phenotypic detection of ESBL production

Phenotypic testing of ESBL production was tested by the Modified Double Disc Synergy Test (DDST) [38] by using a disc of amoxicillin with clavulanate (20/10 µg) along with cefotaxime (30 µg) and ceftazidime (30 µg). A standard inoculum (0.5 McF) of the Salmonella spp. and E. coli isolates were swapped on the surface of Mueller-Hinton II (MH II) agar plates (Biolife, IT). An amoxicillin with clavulanate (20/10 µg) disc was placed at the center of MH II agar plates while discs of cefotaxime (30 µg) and ceftazidime (30 µg) were placed in close proximity of 15 mm distance. Any distortion or increase in the inhibition zone of cephalosporin antibiotics towards the disc of amoxicillin-clavulanate was considered as positive for the ESBL production.

Molecular identification and phylogenetic analysis of multidrug resistant (MDR) bacteria

The 16S ribosomal RNA (16S rRNA) PCR amplification and sequencing of the amplicons was used in the identification of MDR Salmonella spp. and E. coli isolates. Multidrug resistant bacteria were sub-cultured by picking a single colony followed by genomic DNA extraction (Qiagen kit, Hilden, Germany), as per manufacturer’s instructions. The DNA was used for polymerase chain reaction (PCR) using primers targeting the 16S rRNA gene (Table 1). The PCR mixture consisted of 25 µl GoTaq Green MasterMix (Promega), 2 µl DNA, 1 µl of each forward and reverse primer, and nuclease-free water up to 50 µl. Recycling conditions and time of the primers during PCR are shown in Table 1. The PCR products were resolved by 1.5% (w/v) agarose gel electrophoresis (Qiagen, Hombrechtikon, Switzerland). The Agarose gels stained with gelred were viewed under a Gel imager (Biorad Gel Doc XR System, USA). QIA-quick kit (Qiagen, Hilden, Germany) was used to purify PCR products (amplicons) according to the manufacturer’s instructions. Sanger method (Dideoxy sequencing of DNA) was used to obtain the nucleotide sequences of the amplicons (Macrogen, Europe).

Table 1 16S rRNA, virulence and antibiotic resistance genes primer sequences, expected amplicons sizes and PCR cycling conditions
Full size table

The obtained sequences were edited manually using BioEdit v7.0.5.3 [39] to remove gaps and minimize insertions and aligned in MUSCLE [40]. BLASTn searches were done using target sequences from the study isolates and compared with those obtained from the GenBank database. The homologous sequences to the queries were selected based on the Expectation value (E value) as well as query coverage and percent identity. Phylogenetic analysis was done using the Bayesian inference method by MrBayes software v3.2.7 (https://nbisweden.github.io/MrBayes/).The resulting phylogenetic trees constructed by MrBayes were visualized on FigTree software v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/).

Molecular detection of virulence-associated and antibiotic resistance genes

Genes encoding resistance to antimicrobialswereidentifiedfrom the antibiotic resistant bacterial isolates. The PCR mixture consisted of 12.5 µl GoTaq Green MasterMix (Promega, United States), 1 µl DNA, 0.5 µl of eachof the forward and reverse primers, and nuclease-free water up to 25 µl.The primer sequences and PCR cycling conditions are shown in Table 1. The reaction was done using a Proflex PCR system (Applied Biosystems™, United States).The primers used were; uidA, invA, hilA for virulence genes (Table 1) and blaTEM-1, blaCMY-2, blaCTX-M, blaZ (β-lactamases-encoding genes), catI (chloramphenicol resistant gene), tetA, tetC (tetracycline resistant genes), sul2 (sulfonamide-resistant genes), dfrA7 (trimethoprim-resistant genes), strA, aadA (aminoglycosides resistant genes) for antibiotic resistance genes (Table 1).

The PCR products were resolved by 1.5% (w/v) agarose gel electrophoresis in 1× TAE buffer for 1 h at 100 V. Agarose gels stained with gelred were viewed under a gel imager (Biorad Gel Doc XR System, USA) and photographed.

Statistical analysis

Statistical analysis was performed using Minitabv17.1 statistical software (Minitab, LLC). In order to compare bacterial counts in five sampling locations, chi-square (X2) was used. P-value of ≤ 0.05 was regarded significant. A Microsoft Excel spreadsheet software package was also used to generate graphs and charts.

Results

Phenotypic and biochemical characteristics of isolated bacteria

All the 42 recovered bacterial isolates had round colonies ranging from 2 to 4 mm in diameter and were opaque on their respective selective media. The colonies were round with entire margins while the texture was either firm or mucoid (Supplementary Table S1). Salmonella spp. colonies on xylose lysine deoxycholate (XLD) agar appeared red with black centres (Supplementary Fig. S1). Salmonella spp. on triple sugar iron (TSI) agar had a characteristic appearance of red slant surface, yellow butt, blackish growth on the media with gas production (cracks were observed in the media) (Supplementary Fig. S2). The colonies with black centres in XLD and blackish growth on TSI were considered as presumptive Salmonella spp. positive. E. coli colonies on MacConkey agar revealed the growth of bacteria after 24 h of incubation 37 °C and appeared bright pink colour colonies. Eosin Methylene Blue (EMB) agar streaked with E. coli isolates revealed the growth of bacteria after 24 h of incubation 37 °C and was indicated by growth of green-colored metallic-sheen colonies (Supplementary Fig. S3A). All E. coli isolates were confirmed positive to cherry color red formation after adding Kovac’s reagent which gave positive result of indole test (Supplementary Fig. S3B). Based on the Gram staining reaction, all the presumptive Salmonella and E. coli isolates were Gram negative and rod-shaped (Supplementary Table S1; Supplementary Fig. S4).

All the isolates of E. coli were confirmed negative for oxidase and urease tests with no purple colour formation. E. coli isolates were positive to MR test and a bright red coloration was produced. Supplementary Table S1 shows the biochemical reactions done on the isolates and the identities of the isolates. Based on morpho-cultural characteristics and biochemical tests, 18 and 24 bacterial isolates were identified as Salmonella spp. and E. coli, respectively.

Enumeration of total bacterial load/count in retail fish

Salmonella spp. counts ranged from 3.50 Log CFU/ml to 3.72 Log CFU/ml while Escherichia coli counts ranged from 3.51 Log CFU/ml to 3.88 Log CFU/ml across the different markets from five sampled locations (Fig. 1). There was no significant difference (p = 0.192) in the Salmonella spp. and E. coli counts in the five sampling locations.

Fig. 1
figure 1

Salmonella spp. and E. coli counts in fish from five locations of Nairobi County, Kenya. No significant difference (p = 0.192) was observed in both Salmonella spp. and E. coli counts in the five sampling locations

Full size image

Occurrence of Salmonella spp. and E. coli in retail Nile tilapia samples

The occurrence and contamination rates of Salmonella spp. and E. coli ranged from 16.7 to 22.2% and 16.7–20.8%, respectively (Fig. 2). The number of Salmonella spp. in retail markets within each of the five sampling locations were 4 (22.2%), 3 (16.7%), 3 (16.7%), 4 (22.2%), and 4 (22.2%), for Kasarani, Makadara, Westlands, Embakasi and Lang’ata, respectively. For E. coli, the number of isolates were 5 (20.8%) for Kasarani, Makadara, Westlands and Embakasi and 4 (16.7%) for Lang’ata. Overall, the prevalence of contamination of Nile tilapia fish with Salmonella spp. and E. coli was 26.47% (18/68) and 35.29% (24/68), respectively.

Fig. 2
figure 2

Prevalence of Salmonella spp. and E. coli isolated from Nile tilapia collected from five locations in Nairobi, Kenya. For Salmonella isolates, n = 4, 3, 3, 4, and 4 for Kasarani, Makadara, Westlands, Embakasi and Lang’ata, respectively. For E. coli, n = 5, 5, 5, 5 and 4 for Kasarani, Makadara, Westlands, Embakasi and Lang’ata, respectively

Full size image

Occurrence of antimicrobial resistant Salmonella species and E. coli isolates

The distribution of antimicrobial susceptibility profile of Salmonella spp. and E. coli is presented in Table 2. Salmonella spp. isolates showed resistance to all the antibiotics tested except to cefepime. The highest resistance of Salmonella spp. isolates from samples collected from all the locations were to penicillin (22.2%) followed by vancomycin (16.7%). Resistance to ampicillin, cefpodoxime and rifampicin was 11.1%, whereas resistance to ceftazidime, chloramphenicol, meropenem, nitrofurantoin and streptomycin was 5.5%. None of the Salmonella spp. isolates was resistant to cefepime. Intermediates were common in all the antibiotics tested with the range of 11.1–61.1% (Table 2).

Table 2 Antibiotic resistance patterns of Salmonella spp. and E. coli isolates
Full size table

The percentage of resistance to the antibiotics differed among the E. coli isolates, of which 25% were resistant to rifampicin. Resistant to penicillin-G, vancomycin, and meropenem was 20.8%, 16.7% and 12.5%, respectively (Table 2). None of the E. coli isolates were resistant to cefepime, nitrofurantoin and streptomycin (Table 2). A higher percentage of intermediate isolates 58.3% was observed in meropenem (Table 2). For both Salmonella spp. and E. coli isolates, the new generation cephalosporins such as cefepime (fourth generation) were completely effective against all the isolates.

Multidrug resistant patterns of the S. typhimurium and E. coli isolates

The different Salmonella spp. and E. coli isolates exhibited a diverse pattern of resistance to a minimum of one class and a maximum of 5 classes of antimicrobials, among the 9 classes tested. The distribution of MDR and MAR index of S. typhimurium and E. coli isolates is presented in Table 3. Isolates were classified as multi-drug resistant (MDR) if they were resistant to at least three different classes of antibiotics. Based on this classification, 5 isolates for both S. typhimurium and E. coli were MDR. In this study we found that 3 out of 18 (16.7%) S. typhimurium isolates from fresh Nile tilapia fish were resistant to at least three different classes of antibiotics and were considered to be MDR isolates (Table 3). The three S. typhimurium isolates were resistant to five antibiotics (AX + NIT + P + RIF + VA and AX + CAZ + C + MRP + VA) which belonged to 5 different classes of antibiotics with a MAR index of 0.45. Two E. coli isolates were resistant to five antibiotics (AX + MRP + RIF + VA) which belonged to four different classes of antibiotics with a MAR index of 0.36 (Table 3).

Table 3 Distribution of multiple antibiotic resistant characterizations of the Salmonella spp. and E. coli isolated from fresh Nile tilapia sold in retail markets in Nairobi
Full size table

ESBL production in MDR Salmonella spp. and E. coli isolates

All MDR Salmonella spp. and E. coli isolates were ESBL producers (Table 4), as determined by the Modified Double Disc Synergy Test method.

Table 4 Resistance and ESBL production test results of MDR Salmonella spp. and E. coli isolates
Full size table

Molecular identification of MDR Salmonella spp. and E. coli isolates

For the molecular identification of the recovered MDR isolates, DNA were extracted and 1500 bp size of 16S rRNA genes were amplified and sequenced. The sequences of 16S rRNA gene of five MDR isolates were deposited in the NCBI database under accession numbers OP293362.1, OP293363.1 and OP293364.1 for S. typhimurium and OP293365.1 and OP293366.1 for Escherichia coli. BLASTn results revealed that the three isolates WES-09, MAK-22 and EMB-32 were closely related to S. typhimuriumNR_074910.1 with similarity of 93%, 91% and 93%. The two isolates MAK-26 and LAN-35 were closely related to E. coli NR_114042.1 with identity percentage of 90% and 93%, respectively (Supplementary Table S2).

16S rRNA phylogenetic analysis

The phylogenetic analysis (Fig. 3) of 16S rRNA sequence of the three MDR Salmonella spp. isolates (OP293362.1, OP293363.1 and OP293364.1) showed distinct clustering and had the same node showing that they both evolved from the same ancestor. The three isolates clustered together in the same cladograph and had 100% homology.

Fig. 3
figure 3

Phylogenetic tree built using eighteen16S rRNA sequences of Salmonella spp. New isolates S. typhimurium strains WES-09, MAK-22 and EMB-31 are shown in red. Numbers indicated on the nodes are percent posterior probabilities showing statistical support for each node. Branches are coloured based on percent posterior probabilities. The scale bar below the tree indicates the number of expected changes (or substitutions) per site. The outgroup Bacillus altitudinis strain GOES12 (OL851791.1) was used in rooting the tree

Full size image

The phylogenetic analysis (Fig. 4) of 16S rRNA sequence of the two MDR E. coli isolates (OP293365.1 and OP293366.1) showed distinct clustering and had the same node showing that they both evolved from the same ancestor.

Fig. 4
figure 4

Phylogenetic treebuilt using 16S rRNA sequences of Escherichia coli species. New isolates of E. coli strains MAK-26 and LAN-35 are shown in red. Numbers indicated on the nodes are percent posterior probabilities showing statistical support for each node. Branches are coloured based on percent posterior probabilities. The scale bar below the tree indicates the number of expected changes (or substitutions) per site. The outgroup Bacillus altitudinis strain GOES12 (OL851791.1) was used in rooting the tree

Full size image

Detection of virulence-associated genes

The distribution of virulence genes among the multidrug resistant (MDR) S. typhimurium and E. coli are presented in Table 5. All the three MDR S. typhimurium harbored invA (Salmonella invasion gene) and hilA, whereas the two MDR E. coli isolates contained uidA (Fig. 5).

Table 5 Multidrug resistance patterns, virulence genes, and drug resistance-associated genes of MDR S. typhimurium and E. coli isolates
Full size table
Fig. 5
figure 5

Agarose gel image showing virulence-associated genes hilA and uidA for S. typhimurium and E. coli, respectively, and antibiotic resistance gene (sul2) for S. typhimurium and E. coli. Lane L: Molecular weight marker, Lane + ve: positive control, Lane -ve: negative control, Lanes 9, 22 and 32: MDR Salmonella spp. isolates, Lanes 26 and 35: MDR E. coli isolates

Full size image

Detection of antimicrobial resistance genes

The distribution of antibiotic-resistant elements among the multidrug resistant (MDR) S. typhimurium and E. coli are presented in Table 5. Among the S. typhimurium screened for resistant genes blaTEM-1, blaCMY-2, blaCTX-M, and blaZ (beta-lactamase resistance genes), catI (chloramphenicol resistant genes) in EMB-32 only, sul2 (sulphonamide resistant gene), strA (streptomycin inactivating gene), aadA (aminoglycoside resistant gene) and tetC (tetracycline resistance gene) were present. No amplification of tetA and dfrA7 was observed in S. typhimurium WES-09 and S. typhimurium EMB-32 isolates, respectively.

The two E. coli isolates tested positive for antimicrobial resistant genes blaTEM-1, blaCMY-2, blaCTX-M, blaZ (beta–lactamase resistance gene), sul2 (sulphonamide resistant gene), strA (streptomycin inactivating gene), aadA, tetC (tetracycline resistance gene) and dfrA7. The strain E. coli (LAN-35) showed no amplification for tetA gene. Figure 5 shows a representative agarose gel of the amplification of tested antibiotic resistant genes in MDR S. typhimurium and E. coli. The presence of beta-lactamase genes was detected in all tested isolates (Table 5), confirming the phenotypic results of ESBL production tests.

Discussion

Salmonella spp. and pathogenic E. coli are the most frequent causes of foodborne illnesses worldwide and deaths due to consumption of contaminated food and food products. In the present study, Salmonella spp. and E. coli were isolated from 26.47% to 35.29% of the raw Nile tilapia fish samples, respectively. The prevalence of Salmonella spp. in this study was lower than that reported by Budiati et al. [22] in tilapia samples from Malaysia. Also our findings inconsistent with Lerma-Fierro et al. [53] who detected Salmonella spp. in 41.7% of fresh Nile tilapia fillets marketed in Tepic Nayarit City, Mexico. Higher prevalence of Salmonella spp. of 64.9% and 75% were also reported previously in tilapia imported from Thailand [54] and fresh tilapia samples collected from different markets in Bangladesh [55], respectively. Various countries have different hygienic control and management programs, which may explain the differences in Salmonella spp. contamination rates of Nile tilapia fish. Another reason, for these differences may be due to sample size and sample types. Salmonella spp. and pathogenic E. coli are found in animal or human reservoir and their presence in raw Nile tilapia fish from retail markets in this study suggests poor hygienic practices during production, handling, processing and marketing, which could be as a result of direct or indirect fecal contamination, posing risk to people consuming raw or undercooked contaminated fish.

Foodborne pathogens transmitted from fish, fish products and seafood products can lead to serious infections or even death. It is estimated that more than 12% of food-borne outbreaks related to the consumption of fish are caused by bacteria pathogens [56, 57].Total bacteria count in fish is an important parameter in assessing the level of contamination, qualityand public health concern. Several indicator bacteria are not pathogenic themselves but their abundance represents potential risks of contamination [58]. In this study, high counts of Salmonella spp. and E. coli were obtained from the fish samples collected from all the five different locations/sites in Nairobi County, and this confirms previous reports that Salmonella spp. and E. coli are important foodborne pathogens of animal-derived foods. The findings from this study are similar to the high total bacterial counts in fish observed in other studies [58, 59].The presence of Salmonella spp. and E. coli in fish is due to the fact that aquatic environment is tremendously vulnerable to pollution and run-off from anthropogenic sources which contaminate fish products representing a potential hazard to humans [60]. E. coli is one of the major pathogenic microorganisms that may reach animal-derived foods and is an indication of faecal contamination from warm-blooded animals [61]. E. coli are commensal bacteria and E. coli pathotypes can cause zoonotic disease that poses a public health risk. The presence of E. coli in thefish samples sold in open-air markets could be due to poor handling of fish by traders as well as unhygienic handling during transportation and storage. The useof contaminated water for cleaning and processing of fish in the markets mayalso contribute to the secondary contamination. The lack of proper drainage facilities and heavy flyinfestation in these fish markets also promotes tertiary contamination to a great extent [59].

The examination of Salmonella spp. and E. coli isolated from fish for resistance to 11 antibiotics from 9 different classes of antibiotics revealed the existence of antibiotic resistant phenotypes. The isolated Salmonella spp. and E. coli showed resistance to ampicillin/Cloxacillin, ceftazidime, chloramphenicol, nitrofurantoin, rifampicin, streptomycin, penicillin-G, rifampicin, cefpodoxime, meropenem and vancomycin which is in agreement with a study by Sifuna et al. [62]. This antibiotic resistant profile can be due to the frequent use of antibiotics in fish for therapeutic and growth promotion [63, 64]. The presence of antibiotic-resistant Salmonella spp. and E. coli in fresh Nile tilapia fish indicates the role of these fish as spreaders of resistant microorganisms in aquatic environments. The low resistance observed in chloramphenicol by Salmonella spp. and E. coli could be as a result of its ban on usage, since it inhibits protein translation causing aplastic anaemia in some patients [65]. Of specific concern is the high rate of resistance seen to streptomycin, as this is one of the watch group antibiotics in the WHO Access, Watch and Reserve (AWaRe) classification of antibiotics for the evaluation and monitoring of use [66]. Resistance to carbapenems (meropenem) may be due to the transmission of bacteria from human sources, especially that carbapenems are not approved for use in food-animals [67]. According to the WHO, carbapenem-resistant Salmonella spp. and E. coli are considered to be among the most critical pathogens [68]. The detection of carbapenem-resistant Salmonella spp. and E. coli in fish has to be treated as an urgent public health problem. Vancomycin is an antibiotic of the last resort in bacterial infections [69], so presence of vancomycin resistant Salmonella spp. and E. coli isolates in this study is a concern for consumer health.

The morphological characteristics observed in this study support the previous observation contained in the WHO Global Salm-Surv as described by Henderiksen et al. [70] for Salmonella spp. and also in the manual for identification of medical bacteria as described by Phillips [71] for E. coli. The 16S rRNA sequencing further confirmed the Salmonella spp. and E. coli isolates, and clustered them into closely related phylogenetic clades. The 16S rRNA gene is an important landmark in the study of the evolution and classification of bacteria, and has served as base molecular identification tool for study of evolutionary relationships among groups of bacteria [72]. Based on the results from this study, all the MDR Salmonella spp. and E. coli were identified as S. typhymurium WES-09, S. typhymurium MAK-22, S. typhimurium EMB-32 and E. coli MAK-26 and E. coli LAN-35. PCR is a robust and rapid detection method with increased sensitivity and specificity for detecting Salmonella spp. in food, environmental, and clinical samples [73]. Detection of virulence factors is a key step to identify the potential pathogenicity of the obtained bacteria isolates. Fish surface and tissue invasion by the bacterial pathogens is considered to be facilitated by the functioning of virulence factors [74]. The invA gene has been the target for many PCR protocols, as it is found in almost all known serovars of Salmonella spp. [75]. This gene encodes an inner membrane protein necessary for invasion of epithelial cells by Salmonella spp. [76]. The occurrence of the MDR invasive Salmonella spp. isolates among the fish samples suggests that consumers and other stakeholders within the food and value chain might be at a risk of Salmonella-borne infections.

In this study, we report the detection of MDR S. typhimurium and E. coli from Nile tilapia fish marketed for human consumption. The presence of MDR isolates from fresh Nile tilapia fish investigated indicate that consumers are exposed to disease-causing pathogens that make treatment challenging. This is significant to human health due to the zoonotic nature of these pathogens. To avoid the development of MDR, the use of antibiotics should be more strategic and selective. Given that fish harbour multiple bacterial communities living in close proximity to each other, antibiotic resistance in some of these bacteria could lead to easy transfer of resistant genes to others. This could result in increased spread of antibiotic resistance to humans. The spread of MDR bacteria could also be exacerbated by consumption of raw, undercooked or insufficiently heat-treatedfish and fish products. Based on these results, we propose improving sanitary handling and processing of fish to reduce the risk of spread of bacteria pathogens capable of spreading antibiotic resistant genes to humans. Our study highlight the serious issue of S. typhimurium and E. coli multidrug resistance in retail Nile tilapia fish which could result in the evolution of S. typhimurium and E. coli into a super bacteria and risk to public health.

The emergence and spread of ESBL Salmonella spp. and E. coli have become a public health concern because of their association with morbidity and mortality and reduced treatment options. Our results showed that all the MDR S. typhimurium isolates showed resistance to antibiotic classes important in human medicine such as beta-lactamases. Therefore, we conducted a double-disc synergy test (DDST) for ESBL phenotype production and our experiments indicated that all the MDR isolates had an ESBL phenotype [77]. The detection of beta-lactamase genes using PCR was performed to confirm the phenotypic pattern. The beta-lactamase genes blaTEM-1, blaCMY-2, blaCTX–M, and blaZ were present in all the MDR S. typhimurium and E. coli isolates. The blaCTX-M genes encode for ESBLs frequently identified in Gram-negative pathogens. These types of enzymes are active against cephalosporins and monobactams (but not cephamycins or carbapenems), and are of great epidemiological and clinical interest [78].

In the current study, the MDR S. typhimurium and E. coli isolates were examined by PCR to identify the antibiotic resistant genes. PCR analysis revealed the presence of antibiotic-resistance genes belonging to β-lactamases, tetracycline-resistant, sulfonamide-resistant, trimethoprim-resistant and aminoglycosides-resistant genes. Genes like aadA, dfrA7 and sul2 detected in MDR S. typhimurium and E. coli often co-exist as part of gene cassettes on class 1 integrons [79]. The class 1 resistance integrons is located on mobile elements like transposons and plasmids and is widely distributed among clinical and environmental isolates and plays an important role as reservoirs of antimicrobial resistance genes [80, 81]. Amplification of blaTEM-1, blaCMY-2, blaCTX-M, and blaZ is attributed to long term exposure of β-lactam antibiotics in animal and fish farming and for treatment of Gram-negative infections [82]. The presence of antibiotic resistant genes shared across the bacterial isolates reflects active horizontal gene transfer among bacteria in aquaculture. Horizontal gene transfer (HGT) allows bacteria to exchange their genetic materials including antibiotic resistance genes (ARGs) among different species [83], hence promoting multidrug resistance. The use of antimicrobial agents in aquaculture for long periods of time have contributed to increase of antibacterial resistance in fish pathogens, emergence of antimicrobial resistant bacteria in aquatic environments, and also increasing the potential to transfer these resistant genes to pathogenic bacteria of terrestrial animals and humans [84]. The use of antibiotics in aquatic culture in Kenya is not regulated and their indiscriminate use has led to the rise of antibiotic resistant bacteria hence the transfer of the resistance to human bacteria. Therefore there is need for studies to understand the epidemiology of antibiotics in aquaculture in Kenya, as this will be an important step in solving the problem of antibiotic resistance in the aquaculture environment.

Conclusions

The isolation and characterization of S. typhimurium and E. coli, especially those with multiple antimicrobial resistance, in Nile tilapia in retail markets is of public health concern. The occurrence of multidrug-resistant isolates is of specific concern for human and domestic animal health.The potential ability of these MDR bacteria to enter into the food chain can expose humans to serious health risks. This calls for application of more hygienic practices during all stages of fish production and processing for selling. Further, it requires a wide microbiological surveillance and strict governing of the uncontrolled use of antimicrobials either for treatment or as growth promoters, not only in fish production but also in other livestock production systems. Till date, there is insufficient data regarding the role of wildlife and the environment in the complex mechanism of antimicrobial resistance. Therefore, there is a need for research on antibiotic susceptibility surveillance in aquatic environments where fresh fish and marine fish are obtained for human consumption.