Antimicrobial Resistance Profiles and Genes of Staphylococci Isolated from Mastitic Cow’s Milk in Kenya

Increasing numbers of potentially zoonotic multidrug-resistant (MDR) staphylococci strains, associated with mastitis in dairy cows, are being reported globally and threaten disease management in both animal and human health. However, the prevalence and antimicrobial resistance profiles of these strains, including methicillin-resistant staphylococci (MRS), in Kenya is not well known. This study investigated the drug resistance profiles and genes carried by 183 staphylococci isolates from 142 dairy cows representing 93 farms recovered from mastitis milk of dairy cows in two selected counties in Kenya. Staphylococci isolates were characterized by phenotypic characteristics, polymerase chain reaction (PCR) amplification, partial sequencing and susceptibility testing for 10 antimicrobial drugs. Detection of seven resistance genes to the various antimicrobial drugs was conducted using PCR. Overall, phenotypic resistance among the staphylococci ranged between 66.1% for ampicillin and 3.5% for fluoroquinolones. Twenty-five percent (25%) of S. aureus and 10.8% of the coagulase-negative staphylococci (CoNS) isolates, were methicillin-resistant staphylococci phenotypically (defined as resistance to cefoxitin disk diffusion). The most common genes found in S. aureus and CoNS were blaZ and strB at 44.3% and 26%, and 78% and 50%, respectively. MDR was observed in 29.67% and 16.3% of S. aureus and CoNS, respectively. These findings pose a threat to bovine mastitis treatment and management as well as human health.


Introduction
Staphylococcus aureus (S. aureus) and coagulase-negative staphylococci (CoNS) are economically significant and potentially zoonotic agents of dairy cow mastitis worldwide [1,2]. Economic losses due to mastitis have been estimated at over USD 2 billion per year in USA [3]. In humans, these Gram-positive bacteria have been isolated from a variety of infections including mastitis, soft tissue infections, food poisoning, endocarditis and septicemia [4,5]. Diseases due to multi-drug resistant (MDR) staphylococci are often severe, challenging to treat, and threaten animal and human healthcare [6].
Among the S. aureus isolates, 75.8% (n = 69) of the isolates were resistant to at least one the antimicrobial agents tested. As shown in Table 1 ampicillin was the most frequent resistant phenotype (71.4%) followed by erythromycin at 25.2%, while a lower resistance rate was reported in fluoroquinolones (3-4%). Further, 25% of the isolates were phenotypically methicillin-resistant S. aureus (MRSA) based on the cefoxitin disc diffusion test. In addition, all MRSA isolates showed 100% phenotypic resistance to ampicillin followed by 52% for erythromycin, 48% for tetracycline, streptomycin 39% and 30% for trimethoprim-sulfamethoxazole. In this study, significantly higher resistance to various antimicrobials was observed in MRSA compared to MSSA (p < 0.05), except for ciprofloxacin and trimethoprim-sulfamethoxazole ( Table 2). Among the CoNS, 68.5% (n = 63) of the isolates were resistant to at least one antimicrobial agent tested. Ampicillin resistance was the most frequent phenotype at 57.6% followed by resistance to tetracycline (22.8%), while lower resistance to fluoroquinolones (3%) was noted among the isolates. In this study, cefoxitin resistance was reported in 10.8% of the isolates and were termed as methicillin-resistant coagulase-negative staphylococci (MRCoNS) ( Table 1).

Prevalence of Multidrug Resistance in Staphylococci Species
Multidrug resistance (MDR), defined as isolates showing resistance to three or more classes of antibiotics, was detected in 22.9% (n = 42) of the staphylococci isolates. Significantly, a higher proportion of MDR isolates were reported among S. aureus at 29.67% (n = 27) compared to CoNS at 16.3% (n = 15; p = 0.032). Among the S. aureus, MRSA and MSSA showed MDR at 56.52% (n = 13) and 20.5% (n = 14), respectively. For CoNS, 10 isolates classified as MRCoNS, where seven were also MDR. Further, 3.2% (n = 6) staphylococci isolates showed resistance to more than seven antimicrobial agents tested, out of which five of them were MRSA and two were MRCoNS.

Detection of Resistant Genes from the Staphylococci Species
Overall, 57.2% (95/166) of isolates harbored at least one resistance gene of the seven different genes detected in different combination. As shown in Table S1, one isolate carried four genes, eight isolates carried three genes and 20 isolates carried two genes. In addition, 66 isolates carried one gene, of which 97% (64/66) of these isolates carried blaZ genes. Notably, 22 of the isolates carrying more than ≥2 genes were MDR. All the isolates carrying ≥2 antimicrobial resistance genes showed high resistance to ampicillin (100%) and tetracycline (75%).
Overall, more resistance genes were reported in CoNS isolates at a prevalence of 70.2% (n = 59) compared to S. aureus-42.3% (n = 36; p < 0.001). The most prevalent resistance gene was the β-lactamase gene blaZ-59.2% (n = 90). The prevalence of the blaZ gene was higher in CoNS (65.4%; n = 55) compared to S. aureus (41.1%; n = 35; p < 0.002). Strikingly, we noted that 20.5% (n = 14) of the MSSA strain carried the blaZ gene. Further, one of the MRCoNS isolates carried three resistance genes; namely blaZ, msrA, strB conferring resistance β-lactams, aminoglycosides and macrolides (Table 3 and Table S1). Tetracycline resistance to tetM and tetK genes was detected at 16.2% and 6.9%, respectively. We noted, however, that all tetK detected in this study were from CoNS, with none from S. aureus. The streptomycin-resistant strB gene was present in 35.7% of staphylococci isolates. A Higher occurrence of strB genes was reported in CoNS at 50% compared to S. aureus at 13%. Among the erythromycin-resistant isolates, msrA (24.3%) and ermB (21.6%) were the most prevalent among the isolates. The frequencies of msrA and ermB in S. aureus and CoNS were similar (Table 3).

Discussion
Our study investigated the antibiotic susceptibility patterns and antimicrobial resistance genes in staphylococci recovered from mastitic cow milk in Kenya in order to improve mastitis therapy and the control emergence and spread of AMR.
S. aureus and CoNS are recognized as the leading causative agents of mastitis in dairy cows in Kenya [21,30]. However, further molecular characterization of the CoNS to species level is required in Kenya in order to confirm the circulating species and their specific antibiogram.
Overall, our study showed a high proportion of resistance (71.5%) to at least one antibiotic tested, with no differences between S. aureus (75%) and CoNS (68%). This was slightly lower than reported in other studies in Malaysia (96.15%) and South Africa (90%) in S. aureus and CoNS, respectively [31,32] In contrast, a slightly lower resistance proportion in S. aureus (50%) and CoNS (50%) were reported in Uganda [33]. High resistance levels observed in this study could be linked to indiscriminate use of antimicrobials for treatment of mastitis in dairy cows by farmers and veterinarians [34]. As with many developing countries, in Kenya most of these antimicrobials are cheap and readily available as overthe-counter drugs and can be bought without any veterinary prescription [33]. However, further studies investigating antibiotic use and practices in dairy farms in Kenya are necessary to provide conclusive evidence on the extent which such practices contribute to the spread of antimicrobial resistance.
This study reported much higher resistance rates in penicillins (ampicillin) than those reported previously in dairy cows in Kenya by Gitau et al. [21] and Shitandi et al. [27] of about 30% [21,27]. However, our findings were comparable to a recent report by Mureithi et al. [28] in Kenya, who found a prevalence rate of 64% in ampicillin [28]. These results indicate an increase in resistance in staphylococci over the years. High resistance levels to penicillin and other βlactams among mastitis-causing staphylococci has been described [35][36][37]. In contrast, a lower resistance to penicillin in staphylococci (0-20%) has also been reported in some European countries [38,39] and Canada [7]. Higher resistance rates are likely due to the wide-spread use of penicillin in the treatment of mastitis in dairy cows as observed in the study farms and as evidenced by other previous studies in Kenya [27,40] Further, changes overtime, spatial sampling, differences in antibiotic use and practices might explain discrepancies in resistance levels between regions [33,41].
We observed that S. aureus had a significantly higher proportion of ampicillin resistance when compared to CoNS species. S. aureus has been described as a common cause of bovine mastitis in the study region [21]. This higher resistance could be due to the fact that penicillin is still the first-line drug of choice for treatment of mastitis in Kenya [27,34]. Routine culture and identification, coupled with antibiotic sensitivity testing, should be adopted before treatment with antimicrobials to avoid selection pressure of antimicrobial resistant S. aureus [37,39].
S. aureus had a significantly higher proportion of methicillin resistance compared to CoNS in this study. A study in Korea found a slightly higher prevalence of MRSA compared to MRCoNS [42]. However, in contrast, Schnitt and Tenhagen [43] in their review highlighted several studies that have reported higher MRCoNS compared to MRSA in mastitic milk samples [43]. Noteworthy, in all these studies, MRS was defined based on the presence of the mecA gene unlike in our study and could explain the discrepancy. However, although the reason for the higher prevalence of MRSA in our findings is unclear; lower virulence observed in MRCoNS might be a contributing factor [43].
Low resistance levels to quinolones and chloramphenicol were reported in this study among the staphylococci. Authors in Ethiopia [44], South Africa [32], Canada [7] and Uganda [33] reported similar findings. These critically important human-medicine antibiotics are restricted for use in treatment of animal diseases in many countries, including Kenya [32]. However, even the low resistance rates reported are of public health significance and control measures should be implemented to curb further spread [7].
In the present study, 25% of S. aureus and 10.8% of CoNS isolates were phenotypically resistant to cefoxitin, and were consequently classified as MRSA and MRCoNS, respectively. Cefoxitin disk tests have shown in several studies to be a reliable marker for methicillinresistant S. aureus and CoNS not identified to species levels [45,46]. It is worth noting that all MRSA and MRCoNS showed resistance to ampicillin. Our findings on MRSA were in close agreement with reports by Liu et al. [37] in China. A relatively higher prevalence of cefoxitin-resistant MRSA has been reported in Malaysia (38.6%) and Ethiopia (58.1%) [31,45]. Similarly, a higher prevalence of phenotypic MRCoNS has been reported in South Korea (21.2%) [47], Tunisia (29.41%) [48] and Switzerland (47%) [36]. Detection of methicillin-resistant staphylococci (MRS) in mastitic milk is a public health concern and should be further investigated as most of these organisms have shown to be potentially zoonotic in addition to multidrug-resistant, reducing the role of therapy in the control of staphylococcal mastitis [29]. The culling of infected cows to avoid further transmission would be of highest importance [49].
Presence of the mecA gene is considered the gold standard of defining MRSA, in addition to the new resistance genes mecC and mecB which are homologue to mecA [14,46,50,51]. However, in this study, screening for mec genes in the phenotypically methicillin-resistant strains were not carried out which is a limitation in this study. Further investigation targeting mec genes and other mechanisms should be carried out in future studies in order to broaden our understanding on the genetic basis of antimicrobial resistance of these isolates [52]. Molecular typing to assess the clonality of the isolates is also recommended.
Strikingly, the trimethoprim-sulfamethoxazole resistance rate in MRSA reported in this study was quite high (30%). Higher resistance rates in Escherichia coli (E. coli) to trimethoprim-sulfamethoxazole have been reported in livestock (54.2%), especially in poultry and pigs in Kenya [53]. Although data supporting the use of trimethoprimsulfamethoxazole directly in cows are limited, sulfonamides have been reported as the second commonly used antibiotic in food animals after tetracycline [34,40]. Moreover, Muloi et al. [54] in their study on antibiotic practices and knowledge among antibiotics retailers in Nairobi, reported that sulfonamides (63%) were amongst the most purchased class of drug by dairy farmers from Agrovet in Nairobi, Kenya, which neighbors the two study counties [54]. Further, according to Mitema et al. [40] sulfonamides are extensively used in the poultry industry and treatment of calf scours and pneumonia in Kenya [40]. The role of horizontal transfer of AMR genetic determinant between different bacterial species and among humans and animal species has been described [20]. This high resistance rate reported in this study is very concerning since trimethoprim-sulfamethoxazole is heavily used for prophylaxis in HIV-infected patients especially in a highly infectious disease setting, including Kenya [55].
This study reported 20.5% (n = 14) of methicillin-susceptible S. aureus (MSSA) with an effective blaZ variant. A significantly higher prevalence of blaZ (91%) in MSSA has been reported in a hospital setting in Kuwait [56]. Qu et al. [8] also found that 76 of S. aureus isolates carried the blaZ gene but lacked the mecA gene [8]. Presence of blaZ genes in S. aureus has been shown to play a significant role in promoting the acquisition and stabilization of the mecA gene [57]. However, according to Vali et al. [56] and Milheiriço et al. [57] the presence of the blaZ gene in MRSA and MSSA may be responsible for encoding for resistance to penicillin only [56,57]. Penicillins are extensively used in food animals and are still the first-line drug of choice for the treatment of mastitis in Kenya [27]. Further, African MSSA isolated from hospital settings in urban areas has shown to have a significantly higher resistance to penicillin, ranging between 73.7 and 100% compared to other MSSAs as described by [58] in their review. Diversities between blaZ allotypes due to non-clonal evolutions in MRSA and MSSA isolates have also been observed in different geographical regions [57]. Therefore, evaluation of the blaZ allele between MRSA and MSSA isolates in the study should be investigated in future.
Similar to other studies, we observed a significantly higher multidrug resistance (MDR) in MRSA strains compared to MSSA [29,37]. MRSA have shown to have the potential to develop resistance to nearly all the antimicrobial agents [2]. This evolving trend, and the rapid emergence of antimicrobial resistance in S. aureus, threatens disease management in both animal and human health [10].
A higher proportion of MDR isolates were reported among S. aureus compared to CoNS. In contrast, Dorneles et al. [59] and Cheng et al. [60] reported higher MDR rates in CoNS compared to S. aureus in a similar study in Brazil and China, respectively. Significantly higher MDR rates in S. aureus have been described [29,37]. The widespread resistance of cattle-derived S. aureus presents a serious challenge to bovine mastitis therapy and a potential public health risk to humans in Kenya.
Knowledge of distribution of antimicrobial resistance genes among pathogenic udder microbes is key to understanding the evolution of multidrug-resistant agents in dairy cattle. Higher levels of resistance genes were reported in CoNS (70.2%) compared to S. aureus (42.3%) in this study. These findings support the hypothesis that CoNS are considered a main reservoir of genetic elements transferrable to other species of bacteria, including S. aureus [61].
In agreement with other studies, genes encoding for the beta-lactamases blaZ gene were identified in staphylococci at 97% [8,62]. This likely contributed to the high resistance level to penicillin (ampicillin 66.1%) recorded in this study. High blaZ genes might indicate an increased use, and possibly misuse, of β lactams in the study farms [62].
We observed a low to moderate prevalence of antimicrobial resistance genes (ARGs) to tetracycline (tetM and tetK), macrolides (msrA, emrB, ermC) and streptomycins (strB) compared to the phenotypic resistance. Low prevalence of ARGs has been reported by Pekana et al. [63] who found low expression of genotypic resistance in S. aureus in South Africa. Other studies by Gao et al. [64] and Yang et al. [62]-both in China-also reported low genotypic resistance compared to phenotypic resistance in staphylococci isolates. Resistance mediated by other independent mechanisms, such as point mutations, biofilm formation or antibiotic tolerance, could explain these findings [36,65]. Moreover, resistance genes not included in this study may account for phenotypic resistance observed [63]. Further, use of human disc diffusion interpretative criteria may have contributed to the misalignment between phenotypic and genotypic resistance observed in the isolates. Whole genome sequencing is needed to expand our knowledge on staphylococci and their genetic basis of antimicrobial resistance [66].

Study Areas and Design
Study areas and design have been previously described [30]. Briefly, this study was a cross-sectional survey undertaken between November 2018 and June 2019 in Embu and Kajiado-counties in Kenya. The counties were selected purposefully based on the high populations of cows, the human population's demand for milk and dairy products, and diversity of cattle breeds and farming practices. Lactating dairy cows of local and exotic origin in different stages of lactation and parity were randomly selected and sampled.

Herd and Sampling
The isolates utilized in this study were obtained from a previous study on prevalence of mastitis in dairy cows in Kenya [30]. Out of the 595 available staphylococci isolates, a total of 183 colonies, including 91 Staphylococcus aureus (S. aureus) and 92 coagulasenegative staphylococci (CoNS) from 142 dairy cows representing 93 smallholder farms, were randomly selected and analyzed. Distribution of the isolates among the counties were as follows: For S. aureus, n = 58 isolates from 31 cows from Embu and n = 43 were isolated from 29 cows from Kajiado County. For CoNS, n = 53 isolates from 48 cows in Embu and n = 39 were isolated from 34 cows in Kajiado. Detection of clinical and subclinical mastitis was conducted as previously described by [67].

Milk Sample Collection
Milk samples were collected following standardized procedures described by the National Mastitis Council [68]. The samples were transported at 4 • C to the University of Nairobi, Department of Veterinary Pathology, Microbiology and Parasitology bacteriology laboratory for microbiological examination.

Isolation and Phenotypic Characterization of the Isolates
Initial bacterial culture and identification was performed according to standard methods described in [68,69]. Briefly, a 0.01 mL aliquot of each milk was aseptically streaked onto the surface of 5% sheep blood agar (Oxoid, Basingstoke, England) and incubated aerobically at 37 • C for 24-48 h. Where growth occurred, colony morphology, Gram stain reaction, catalase, coagulase tube testing and mannitol salt agar (Oxoid, Basingstoke, England, UK) were used to presumptively identify S. aureus and CoNS. Single colonies from respective isolates were sub-cultured on nutrient agar slants, and the slants were stored at 4 • C for further use.

Bacterial Genomic DNA Extraction
The boiling method was used to extract bacterial genomic DNA [70]. Briefly, a loopful of a bacterial colony grown overnight on tryptone soy agar (TSA) was added to 1.5 mL Eppendorf tubes containing 100 µL of nuclease-free water. The tubes were boiled in water bath at 100 • C for 25 min. After centrifugation at 30,000× g for 5 min in a microcentrifuge (Eppendorf Hamburg, Germany), the supernatant was transferred to a new 1.5 microcentrifuge tube. The extracted DNA was stored in a freezer at −20 • C until used for PCR analysis.

Staphylococcus Aureus nuc Gene Amplification
PCR amplification of the staphylococcal thermonuclease (nuc) gene was used to confirm ninety-one biochemically identified S. aureus. The oligonucleotide primers described by [71] in Table 4 were used and the PCR was performed using Taq Polymerase (QIAGEN, GmbH, Hilden, Germany) following the manufacturers' instruction. For positive and negative controls, S. aureus ATCC 29,213 (nuc positive strains) [50] and DNase deionized water, respectively, were used. Electrophoresis for each PCR amplicon was performed using ethidium bromide-stained 1.5% agarose gel in the Tris-Borate-EDTA (TBE) buffer and visualized using UV illuminator (GelMax ® Imager, Cambridge, UK). A molecular ladder was used to determine the sizes of the amplicon (GelPilot 1 kb Plus Ladder (100), QIAGEN, GmbH, Hilden Germany) with an expected amplicon size of 276 bp.

Antibiotic Susceptibility Testing
Phenotypic antibiotic susceptibility characterization was carried out using the Kirby-Bauer disc diffusion method following the Clinical & Laboratory Standards Institute (CLSI) guidelines [50]. Briefly, fresh cultures of bacterial isolates were suspended in sterile physiological saline to attain turbidity equivalent to 0.5 McFarland. Sterile cotton wool swabs were then used to inoculate the standardized bacterial suspension onto Mueller-Hinton (Oxoid) agar plates. The antibiotics discs belonging to seven classes of antibiotics at the following concentration were analyzed: aminoglycosides (gentamicin 10 µg, streptomycin 10 µg), fluoroquinolone (ciprofloxacin 5 µg and norfloxacin 10 µg), tetracycline (tetracycline 30 µg), folate pathway inhibitors (sulfonamide + trimethoprim 25 µg), macrolides (erythromycin 15 µg), beta lactams (ampicillin 25 µg and cefoxitin 30 µg) and phenicols (chloramphenicol 10 µg). The choice of antibiotics was guided by drugs that are commonly used in dairy veterinary practice in Kenya, some of which are important to human medicine. The inoculated plates were then incubated at 35-37 • C for 17 h.
Since most of the antibiotics used have no approved bovine breakpoints, interpretive criteria described by [46,50] were used. The inhibition zones were measured and interpreted as either susceptible, intermediate, or resistant to the tested antibiotic agent when appropriate break points were available. However, in this study, all strains read as intermediate were considered as resistant. Due to lack of approved interpretive criteria for streptomycin in staphylococci, isolates were interpreted using the interpretive criteria of another aminoglycoside (gentamicin 10 µg) in this study [50]. Staphylococcus aureus ATCC 25,923 was used as the quality control strain. Isolates resistant to cefoxitin (S. aureus) and (CoNS) were presumptively identified as MRSA and MRCoNS, respectively [46,50]. The isolates were classified as multidrug-resistant (MDR) if they were found resistant to three or more different antimicrobial classes used [75].

Detection of Antimicrobial Resistance Genes
All staphylococci isolates (S. aureus and CoNS) showing phenotypic resistance to beta lactams, erythromycin, tetracyclines and streptomycin were analyzed by PCR for genes that confer resistance to penicillin (blaZ), erythromycin (ermB, ermC, msrA), streptomycin (strB) and tetracycline (tetK, tetM). The details on primers and annealing temperatures used in the study are provided in Table 4. PCR products (10 µL) were electrophoresed using ethidium bromide-stained 1.5% agarose gel in the Tris-Borate-EDTA (TBE) buffer and visualized using the UV illuminator (GelMax ® Imager, Cambridge UK). Molecular ladder was used to determine the sizes of the amplicon (GelPilot 1 kb Plus Ladder (100), QIAGEN, GmbH, Hilden Germany). In all PCR reactions, RNAse-free water was used as negative control.

nuc Gene and Antibiotic Resistance Gene Sequencing and Analysis
A subset of the PCR amplicon of the nuc and the antibiotic resistant genes were sequenced to confirm identities of the detected organisms and genes using Sanger DNA sequencing approaches [76]. Quality control, assembly and editing of nucleic sequence trace files were performed using SnapGene version 5.2.4 [77] and customized UNIX shell scripts. Sequence identities were confirmed using the Basic Local Alignment Search Tool (BLAST) [78].

Statistical Data Analysis
Antibiotic resistance data were analyzed using STATA version 15. Descriptive statistics were used to calculate the proportion and frequencies of all variables. The chi-square test (χ 2 test) or Fisher's exact test were used when applicable to compare categorical variables. Statistical significance level was set at 0.05 (p < 0.05).

Conclusions
This study revealed a high ampicillin resistance rate among bovine mastitis staphylococci. In addition, detection of various antimicrobial resistance genes in these strains signifies a public health concern and a serious challenge to bovine mastitis therapy. Therefore, there is a need to control the emergence and spread of AMR in dairy farms. The presence of phenotypic methicillin-resistant staphylococci (MRS) in this study provides baseline for further monitoring in Kenyan dairy farms. Further screening of the mec genes (A, B, C) and other intrinsic mechanisms encoding for resistance to MRS should be considered in future studies. Molecular characterization of CoNS isolates to species level will be necessary to confirm circulating species in Kenya. We used PCR to determine the genotypic resistance, this technique targets fewer AMR genes, restricting the results to screened elements. Therefore, there is a need for further characterization of the isolates using whole genome sequencing and spa typing in order to assess the clonal diversities of the isolates.