Diversities of antimicrobial resistance patterns across Staphylococcus aureus and Coagulase-Negative Staphylococci isolated from dairy herbs in Jiangsu province, China

As mastitis major causing agents, Coagulase-Negative Staphylococci (CNS) and Staphylococcus aureus (SA), are important and their connections are special and worth comparing. The overall aim of this study is to investigate antimicrobial resistance patterns of CNS and SA. Understanding the special characters of staphylococci is essential for nding the precise strategies or directions against them.

resistance gene transfer among them [12]. Although some features of CNS and SA have been often reported separately [6,[8][9][10], the similarities and differences of their resistant models still are covered. Only after understanding the special characters of staphylococci, we can nd the precise strategies and research directions to defeat them.

Phenotypic Antimicrobial Resistance Characteristics
All staphylococci were evaluated for the susceptibility to 15 antimicrobials within nine categories, then the results of SA and CNS were compared by the chi-square testing ( Table 1). The majority of staphylococci were sensitive to chloramphenicol, nitrofurantoin and clarithromycin, but showed resistance to penicillin, ceftazidime and nor oxacin. Chloramphenicol was also the most effective drug against CNS (86.67%), but the highest proportion of SA is susceptible to clarithromycin (92.26%). Staphylococcus aureus were mainly resistant to penicillin (94.05%), nor oxacin (58.33%), levo oxacin (54.76%), cipro oxacin (57.74%), ceftazidime(55.95%), and co-trimoxazole(41.07%). CNS were mainly resistant to penicillin (77.78%) ceftazidime (55.95%) tetracycline (36.89%), and erythromycin (24.44%). According to the chi-square test results, it shows that SA was signi cantly better to resist antibiotics interfering in the metabolism of nucleotides than CNS; however, CNS strains were signi cantly better to resist antibiotics interfering in bacterial protein synthesis, except chloramphenicol which was forbidden to use in China. Table 2&3 shows the panorama of multidrug resistance patterns of SA and CNS isolates in this study. For SA, 62.52% of them were multiple drug-resistant (MDR) resisting more than 3 antimicrobial categories. Staphylococcus aureus were mainly resistant to four antimicrobial categories, and the most common multidrug resistance patterns within them were resistant to β-lactamase, cephalosporin, quinolone, and sulfonamide antibiotics. 73.34% Of CNS were extensively drug-resistant (XDR), which were resistant to ≥ 2 antimicrobial categories, and 45.78% were MDR strains. CNS were often resistant to two antimicrobial categories; the most common multidrug resistance pattern within them was resistant to β-lactamase and lincomycin antibiotics. However, what is worth mentioning, there is 4 CNS were pandrug-resistant (PDR) bacteria which were non-susceptible to all antimicrobial agents listed in this study.

Association of antimicrobial resistance phenotype with resistanceassociated genes
PCR was used to detect the major resistance genes for tetracyclines (tetK, tetM, tetL, tetO), macrolides (ermC), and β-lactamase (blaZ). Amond the 83 strains, tetracycline-resistant CNS group, 66.27% of them carried tet-type genes, and the rest appeared resistant to tetracycline without detecting out any of the tet-type genes (Table 4). Conversely, there were also 8 isolates carrying the tetK gene in the group of 92 CNS strains that were susceptible to tetracycline. Besides, only two SA isolates carrying tetK and they both were resistant to tetracycline; other tet-type genes were not found in SA in this research. Figure 1a, 1b and 1c show the PCR products of the tet-type genes from staphylococci. The blaZ (β-lactamase) gene was detected in 59 out of 175 CNS strains that were resistant to penicillin, and 7 out of 50 CNS isolates that were susceptive, but 42 SA isolates with blaZ gene all could resist penicillin. Figure 1d shows the PCR products of the blaZ gene from staphylococci. The ermC was found in seven CNS strains and two SA strains, and all of them were resistant to erythromycin. Figure 1e shows the PCR products from the ermC gene from staphylococci. All the above information about the differences between SA and CNS strains was summarized in Table 4. Notably, we found that all SA strains carrying resistance genes were resistant to the related antibiotics, but CNS were not.

Discussion
Staphylococci are the bacteria most commonly isolated from bovine mastitis [13]. The discussions and comparisons between CNS and SA have often been reported [6,14]. Although SA causes more server mastitis than CNS, CNS are now predominant over SA in most countries [6]. Staphylococcus aureus mastitis easily remains persistent [15], but CNS are more able to persist in the mammary gland with causing a moderate increase of milk SCC [16]. In this study, 168 isolates of SA were recovered from subclinical cases in the 6 farms. For 225 CNS isolates, Staphylococcus sciuri (22.67%) and Staphylococcus xylosus (17.33%) are identi ed most frequently. Staphylococcus sciuri is also reported as the commonest CNS species isolated from cows in Poland [17,18]; Staphylococcus chromogenes, however, is typically the most commonly identi ed species in Columbus (USA) or Helsinki (Finland) [14,19].
On the other hand, vitro drug sensitivity tests were conducted on staphylococcal isolates. We found that the most effective antibiotics against CNS and SA were different. Most SA strains were susceptible to clarithromycin (92.26%), followed by nitrofurantoin (88.69%) and chloramphenicol (85.71%). The majority of CNS were sensitive to chloramphenicol (86.67%), cotrimoxazole (83.56%) and nitrofurantoin (82.22%). Therefore, chloramphenicol can defend against most staphylococci; this might due to that the use of chloramphenicol has been banned in food animals in China because of its inhibition of bone marrow hematopoietic function [20]. Fortunately, chloramphenicol now can be replaced by orfenicol, which is also a member of the amide alcohols. Florfenicol was developed in the late 1980s as a new kind of broad-spectrum antibiotic for veterinary use.
The concepts of MDR, XDR and PDR are very important, characterizing the different patterns of resistance found in bacteria. But there were many different de nitions for them until 2012, Magiorakos reported and de ned them [21]. According to the report, we exercised statistic data; from Table 2 and Table 3, it can be seen that most of SA were quadruple-drug resistances, and most CNS resist double. However, CNS seem to outperform SA in resisting six or more classes of drugs; there were even 4 strains of CNS are PDR strains that were non-susceptible to all of the 15 kinds of antibiotics used in this study. This point agrees with the view that CNS tend to be more resistant to antimicrobials than SA and easily develop multiresistance [11]. Therefore, CNS are more likely represented as a reservoir which transmitted antibiotic-resistance to other staphylococci like SA [11].
According to the chi-square test results, we summary and nd that quinolones (cipro oxacin, levo oxacin, and nor oxacin) and co-trimoxazole antibiotics, both interfere with the metabolism of nucleotides. They affected CNS signi cantly more than SA. Quinolones antibiotics can trap the bacterial topoisomerases on DNA [22] and co-trimoxazole includes trimethoprim work as a tetrahydrofolate reductase inhibitor, and sulfamethoxazole provides a second blockage of the folate biosynthetic pathway [23]. Therefore, whatever the results in Jiangsu, Finland [13], or Portuguese [24], there might be mechanisms or compounds that tend to protect nucleotide metabolism in SA from these antibiotics under pressure from these kinds of antibiotics. Proctor proposes that the special thermonuclease in SA could release extra thymidine from bacterial inoculum and mammalian tissues, so they can bypass the metabolic blockades caused by co-trimoxazole [25]. We remained unknown whether this release process could also affect the working mechanism of quinolones or there are other defense mechanisms in SA.
Equally, We also compare the resistance of CNS and SA to lincomycins (clindamycin), macrolides (including erythromycin and clarithromycin), tetracycline, and nitrofurantoin antibiotics; those antibiotics all interfere the bacterial protein synthesis by inhibiting the prolonging of peptide chains [26][27][28]. We summary and found that the antimicrobial resistance rates of CNS to them were signi cantly higher than that of SA; that is to say, they affected CNS signi cantly less than SA in general. Moreover, it can be found that resistance genes were detected much more frequently in CNS than SA, which means resistance genes may have more contribution in CNS than SA. Taponen also reported that CNS were more resistant to macrolides, lincosamides, oxytetracycline and fusidic acid in Finland [29]. These are partly in line with the view CNS tend to be more resistant to antimicrobials than SA [6,11].
Macrolides (MLS) and lincomycin antibiotics have similar binding sites in the 50S ribosomal subunit. It can irreversibly bind to bacterial ribosomal 23SrRNA and selectively inhibit protein synthesis by blocking tranlspeptide action and mRNA shift [26]. The erm gene can methylate speci c nucleotide residues of 23SrRNA, resulting in their resistance to macrolides, lincosamide, and streptomycin B with the same target position, clinically known as MLS resistance. Chi-square test results show that CNS were signi cantly more resistant to macrolides (erythromycin and clarithromycin) or lincomycin (clindamycin) than SA in this area. At the genetic level, the detection rate of ermC in the CNS is also higher than that of SA. MLS resistance can be divided into constitutive (cMLS) and induced (iMLS) resistance [30]. The resistance of cMLS type could resist erythromycin and clindamycin, which could be detected routinely in vitro drug sensitivity tests. However, iMLS resistance means that erythromycin, as an inducer, can induce erm gene expression through the attenuating mechanism to produce resistance toward clindamycin; hence some bacteria with iMLS could be resistant to erythromycin and sensitive to clindamycin in vitro drug sensitivity assay, but clindamycin might be ineffective to them in clinical treatment [31]. Therefore, ermC is often added to the drug sensitivity test to prevent clinical drug failure. The detection rate of ermC in this study was 3.1% in CNS and 1.2% in SA; both were low, indicating that the most clinical susceptibility results should be valid and the probability of clinical clindamycin failure is low. It also shows that the resistant models of CNS and SA to macrolides and lincomycin drugs were little dominated by the ermC mechanism in Jiangsu province. The resistance rate of staphylococci to erythromycin in Jiangsu area (15.78%) is still somewhat high, and CNS were signi cantly more resistant to macrolides (erythromycin and clarithromycin) or lincomycin (clindamycin) than SA Besides, tetK, tetM, tetL and tetO, the major resistance genes that defend against tetracycline antibiotics in staphylococci, were detected in this study. We found that staphylococcal resistance to tetracycline is mainly mediated by energy-dependent e ux of tetracycline in this area; the tetK and tetL genes encode proteins that pump out tetracycline, but this cannot remove minocycline [32]. Another resistance mechanism, ribosomal protection proteins (RPPs), plays a minor role in this area. RPPs are encoded by tetO and tetM; together of them were only detected out a tiny proportion, 1.2%, in all isolated staphylococci. RPPs result in an allosteric disruption of primary tetracycline binding and consequently release of the drug. This mechanism confers resistance to tetracyclines and minocycline [33]. Therefore, although 26.6% of CNS can resist tetracycline theoretically due to the energydependent e ux of tetracycline, most of them should be susceptible to minocycline. Phenotypic antimicrobial-resistant observations of this study were also supported this hypothetical pattern.
A related point to consider is that, for minocycline and chloramphenicol, they are not used in these 6 farms and chloramphenicol is forbidden to be used in China. Few CNS and SA isolates were resistant to these two in this study, and the chi-square test results of them didn't show a signi cant difference between SA and CNS. Therefore, we speculate that only after its involvement, the stress of antibiotics interfering with protein synthesis, CNS strains might be more likely to become the mutations to survive than SA. Interestingly, recently Fernando also makes a very similar interpretation on another kind of bacteria; he assumed that the evolutionary landscapes of Pseudomonas aeruginosa towards ribosome-targeting antibiotic resistance depend on selection strength [34]. We further proposed that this evolutionary also relies on the classes of antibiotics and the species of bacteria.

Conclusion
This study revealed the characters of antibiotic-resistant patterns of CNS and SA in Jiangsu Province of China. Staphylococcus aureus might be more likely to survive from antibiotics that interfere the metabolism of nucleotides, and CNS strains might be more likely to become the mutations to survive than SA, responding to the stress from antimicrobial interfering in bacterial protein synthesis. The evolutionary landscapes of staphylococci towards antibiotic resistance are affected by the classes of antibiotics and the species of bacteria.

Sampling
Milk samples, which were positive by the California Mastitis Test (CMT), were collected from six dairy farms in Jiangsu Province, China, from March 2015 to March 2016. The experimental scheme of this study is outlined in Fig. 2. Before sampling 10 mL milk aseptically as described [35], the teat ends were disinfected with moist cotton swabs (10 to 15 s) soaked in 70% alcohol then dried with a paper towel, and the rst streams of milk were discarded. The milk tubes were transported on ice to the Key Laboratory of Jiangsu Preventive Veterinary Medicine.
To obtain a representative sample of the herds on Jiangsu Province, we selected the farms for this study according to the following criteria: (i) the farm had a minimum of 100 lactating cows, (ii) the average milk production was above 5500 kg of milk per cow per year, (iii) the cows were milked twice daily at 7:00 a.m. and 4:00 p.m., (iv) the cows all were Holstein. Milk samples from dairy cows with mastitis were obtained from six herds in total from four districts (2 herds in Yangzhou, 2 herds in Suzhou, and 2 herds in Changzhou and Nantong); the mean herd size of them was 196; the antibiotic list they used in farms are penicillin, ceftriaxone, levo oxacin, kanamycin, amikacin, tetracycline, nor oxacin, ceftazidime, and streptomycin. Due to that the tetracycline, macrolide, and β-lactamase are most used in these farms, we choose the corresponding and common resistance genes (tetK, tetM, tetL, tetO, ermC, and blaZ) to explore their general resistance mechanisms.
For this study, all of the staphylococci were isolated from milk samples from dairy cows with subclinical mastitis. All the udder quarter milk from subclinical mastitic cows was analyzed by the CMT by veterinarians. The CMT-reaction is graded 1 to 5. The scores are ranked according to an increase in viscosity, where the highest viscosity (CMT 5) is roughly correlated to the highest SCC. If the udder quarter milk had a CMT score ≥ 3, it would be determined as test positive.

Laboratory Bacteriological Culture
Milk samples were examined by procedures as Laboratory Handbook on Bovine Mastitis recommended by the American National Mastitis Council [36]. Brie y, milk samples (100 µL) from mammary quarters were plated onto trypticase soy agar (TSA) plates supplemented with 5% (vol/vol) de brinated sheep blood (Hangzhou Tianhe Microorganism Reagent Ltd., Hangzhou, China). Inoculated plates were incubated at 37 °C. After incubation for 18 to 24 h, all plates were observed for microbial growth. If bacteria did not grow, the plate was returned to the incubator for an additional 18-24 h and then reevaluated for microbial growth. Those plates showing growth were recorded, and species identi cation started. If a quarter milk sample resulted in the culture of 3 or more dissimilar colony types, the milk sample would be recorded as contaminated and excluded from the study. Isolates were Gram-stained to assist in organism identi cation.
Staphylococci were identi ed by means of typical colony morphology, gram-positive coccal staining and by being tube catalasepositive; if the above staphylococci were with hemolysis, pigmented colonies and tube coagulase-positive then they were classi ed as SA potential, or they were classi ed as CNS potential. What calls for special attention is that, although both Staphylococcus intermedius and Staphylococcus hyicus colonies are both tube coagulase-positive like SA, they are nonpigmented; Staphylococcus hyicus do not produce hemolysis. Therefore, they should be distinguishable in the experiment. The obtained 393 staphylococci were selected for further identi ed by Vitek 2 system with Gram-positive identi cation card (GPI; bioMerieux Vitek) and detecting drug sensitivity and resistant genes. For the non-de nitive strains, 16S rRNA genes were ampli ed for sequencing by BGI (China), and the primers were shown in Table 5 [37]. ATCTTTTAGCAAACCCGTATTC On the other hand, colonies suspected as being streptococci by colony morphology, Gram-positive cocci, as well as catalasenegative; bacillus were determined by Gram-staining bacillus under the oil mirror. Both of them were not further characterized in this paper.

Drug Susceptibility Testing
The disk diffusion method was used for antibiotic susceptibility tests and the results were read based on the recommendations of the Clinical and Laboratory Standards Institute (CLSI, 2010a) of the USA. The antibiotics penicillin (β-lactamase), co-trimoxazole (sulfonamide), cefotaxime (cephalosporin), clarithromycin (macrolide), ceftriaxone (cephalosporin), chloramphenicol (amphenicol), ceftazidime (cephalosporin), cipro oxacin (quinolone), nitrofurantoin, tetracycline (tetracycline), erythromycin (macrolide), levo oxacin ( uoroquinolone), minocycline (tetracycline), nor oxacin ( uoroquinolone), and clindamycin (lincomycin) were used in this study; they were universal therapeutic agents against Gram-positive bacteria and purchased from Hangzhou Tianhe Microorganism Reagent Co., Ltd. Antimicrobial agent concentrations on each absorbent paper and the interpretative criteria are shown in Table 6, and the tests were performed on Mueller-Hinton agar (Hangzhou Tianhe Microorganism Reagent Ltd.) with incubation at 37 °C for 24 h. SA ATCC 25923 was used as a quality control strain tested in parallel with each batch of isolates, and on all occasions, the results were within acceptable ranges according to the speci cations of the CLSI. Each diffusion was examined carefully by using incident light to examine the plate.

DNA Extraction and PCR Conditions
Bacterial DNA was prepared as previously described [38] and used in PCR. Table 5 shows base sequences and predicted sizes of ampli ed products for the speci c oligonucleotide primers used in this study. The oligonucleotide primer for the tetM gene was designed according to the nucleotide sequence of the tetM gene (GenBank: X56353.1), and other sequences have been described elsewhere [39,40]. The PCR reactions were performed in a total volume of 20 µL, including 1 µL of template DNA, 0.5 µM primers (Invitrogen, China), and 10 µL 2 × PCR Master Mix (Thermo, China). Ampli cation reactions were carried out using a DNA thermal cycler (GeneAmp PCR system 9700, Applied Biosystems, USA) as follows: thermal denaturation at 94 °C for 5 min followed by 32 cycles of denaturation at 95 °C for 1 min, primer annealing (temperature as in Table 5

Statistical Analysis
Chi-square tests were performed using SPSS Software version 20 (IBM SPSS Statistics for Windows, Armonk, NY, USA) to compare the number of antibiotic-resistant strains and the total number of susceptible strains and intermediate strains between CNS and SA within the same antibiotic. In the test, P < 0.05 was considered statistically signi cant. Moreover, the occupation rate, also called sensitivity according to Martin's described [43], was used to estimate the contribution rate of each resistant gene in bacteria with resistance to corresponding drugs; the true positive rate of each resistant gene, also called effective rate, was used to evaluate the proportion of each resistant gene in bacteria with the real effective to resist relative drugs. Both were used to investigate the connections between phenotypic and genotypic resistance. The true positive sample number means the number of strains containing resistance genes and also resisting related antibiotics. The formulae for calculating them are shown below: Abbreviations   The experimental scheme of the study. Milk samples were aseptically collected from bovine with subclinical mastitis and plated onto trypticase soy agar (TSA) plates supplemented with 5% (vol/vol) de brinated sheep blood. The staphylococci were then selected for further identi ed by Vitek 2 system with Gram positive identi cation card (GPI; bioMerieux Vitek); their antibiotic susceptibilities were detected by the disk diffusion method, and the resistance genes were revealed by PCR assay.