Species distribution and in vitro antimicrobial susceptibility of coagulase-negative staphylococci isolated from bovine mastitic milk

Background Coagulase-negative staphylococci (CoNS) are the most common bovine mastitis causing bacteria in many countries. It is known that resistance for antimicrobials is in general more common in CoNS than in Staphylococcus aureus but little is known about the antimicrobial resistance of specific CoNS species. In this study, 400 CoNS isolates from bovine mastitic milk samples were identified to species level using ribotyping and MALDI-TOF MS, and their antimicrobial susceptibility was determined using a commercially available microdilution system. The results were interpreted according to the epidemiological cut-off values by the European Committee on Antimicrobial Susceptibility testing. Results The most common CoNS species were S. simulans, S. epidermidis, S. chromogenes and S. haemolyticus. Penicillin resistance was the most common type of antimicrobial resistance. Staphylococcus epidermidis was the most resistant among the four major species. Almost one-third of our S. epidermidis isolates were resistant to >2 antimicrobials and close to 7 % were multidrug resistant. The majority of S. epidermidis isolates were resistant to benzylpenicillin. On the contrary, only few S. simulans isolates were penicillin-resistant. Phenotypic oxacillin resistance was found in all four main species, and 34 % of the isolates were oxacillin resistant. However, only 21 isolates (5 %) were positive for the mecA gene. Of these, 20 were S. epidermidis and one S. sciuri. mecC positive isolates were not found. Conclusion Staphylococcus epidermidis differed from the three other major CoNS species as resistance to the tested antimicrobials was common, several isolates were multidrug resistant, and 19 % of the isolates carried the mecA gene encoding methicillin resistance. Electronic supplementary material The online version of this article (doi:10.1186/s13028-016-0193-8) contains supplementary material, which is available to authorized users.

of the risk of spreading the mec genes [7,8]. Furthermore, emergence of resistance among CoNS is a concern because resistance determinants may be transferred between staphylococcal species and form a risk for public health [9,10].
Phenotypic identification methods for bovine CoNS have proven to be unsatisfactory [11][12][13]. Molecular methods have become available for identification of CoNS to species level, which has made species determination more reliable. Reliable genotypic identification has enabled studying frequency of different CoNS species and species-specific antimicrobial susceptibility. Reports on mastitis causing CoNS species and their antimicrobial susceptibility have since been published by some authors [5,6,14]. Unfortunately, only few studies have used epidemiological cut-off values (ECOFF) of the European Committee on Antimicrobial Susceptibility testing to determine proportions of resistant isolates [15], which has made comparisons difficult. Studies on genetic mechanisms for resistance of bovine CoNS species have also been published, with different panels of resistance genes [5,16,17]. In Finland, antimicrobial susceptibility of CoNS has been reported for bovine CoNS as a group only [2,18]. It is likely that in the future CoNS will no more be considered as one homogenous group, but species-specific approaches become possible in mastitis control [3]. Knowledge on the antimicrobial susceptibility of different CoNS species is then also necessary.
The aim of this study was to explore the distribution of CoNS species isolated from mastitic milk samples in Finland and to determine the antimicrobial susceptibility of different CoNS species.

Methods
The material consists of CoNS isolates from two studies, Pitkälä et al. [2] (dataset 1) and Finnish veterinary antimicrobial resistance monitoring program 2010-2012 [18] (dataset 2). The number of CoNS isolates was 312 and 88 in dataset 1 and 2, respectively.
The first study (dataset 1) was a nationwide prevalence study carried out in 2001, in which milk samples were collected from all four quarters of all lactating cows in herds randomly allocated into the study. Conventional microbiological methods were used to identify bacteria isolated from the milk samples [2,19]. The milk samples were classified as mastitic when the milk somatic cell count, measured with an electric counter (Fossomatic Milk Analysis, Foss Electric, Hillerød, Denmark), exceeded 300 000 cells/ml. Of the total of 2103 CoNS isolated in that study, 335 were randomly chosen for in vitro antimicrobial susceptibility testing [2]. Of these isolates, 318 were successfully identified to species level in the present study. Six isolates appeared to belong to S. aureus species and were withdrawn from this study. In the second study (dataset 2) the samples were routine mastitis samples (based on elevated milk somatic cell count and/ or clinical signs of mastitis) submitted during 2012 to the laboratory of Valio Ltd by veterinarians and dairy farmers. Bacteriological etiology of mastitis was determined by a real-time PCR assay (Thermo Scientific PathoProof ™ Mastitis Complete-12 Kit, Thermo Fisher Scientific Ltd.,Vantaa, Finland). During the study, the preservative was left out from the milk tubes. Samples positive for CoNS in the PCR test were selected for culture. From the cultured pure samples the first cultured 88 CoNS isolates, but only one isolate per herd, were selected for determination of antimicrobial susceptibility. Staphylococcus species identification was performed during the present study.
The CoNS isolates from both datasets were identified to species level with the 16S and 23S rRNA gene restriction fragment length polymorphism method (ribotyping) as described previously [20]. The CoNS species were determined by comparison in a numerical similarity analysis of the ribotype patterns with a ribotype library using BioNumerics 5.1 software package (Applied Maths, St.-Martens-Latem, Belgium). For some isolates ribotyping failed, and they were later analyzed with MALDI-TOF MS (Microflex LT, Bruker Daltonic Gmbh, Bremen, Germany). The correctness of species identification based on ribotyping was confirmed by analyzing a representative sample of all different ribotype patterns by MALDI-TOF MS [21]. The agreement between the methods was excellent.
Antimicrobial susceptibility of CoNS isolates from both datasets was determined in the previous studies [2,18] using a commercially available microdilution system (VetMIC ™ ; SVA, Uppsala, Sweden). Minimum inhibitory concentrations (MIC) in both datasets were determined for penicillin, cephalothin, oxacillin, erythromycin, chloramphenicol, clindamycin, tetracycline, gentamicin, neomycin, streptomycin, and trimethoprim/ sulfamethoxazole. In addition, MICs for virginiamycin, vancomycin and avilamycin in dataset 1, and for fusidic acid, kanamycin, ciprofloxacin, trimethoprim, florfenicol, and cefoxitin in dataset 2, were determined. Results from the susceptibility testing were interpreted according to the epidemiological cut-off values (ECOFFs) by the European Committee on Antimicrobial Susceptibility testing (EUCAST) [22] as non-wild type (from now on referred as resistant) or wild type (sensitive). If a specific ECOFF was not available for the specific species or for CoNS as a group, ECOFF of Staphylococcus aureus was used. Production of beta-lactamase was tested using nitrocefin discs (dataset 1: AB Biodisk, Solna Sweden; dataset 2: Becton-Dickinson, NJ, USA). The isolates with MIC values for oxacillin >1 mg/l were tested for presence of the mecA gene in dataset 1 and for presence of the mecA and mecC genes in dataset 2, using PCR and primers reported previously [23,24].

Distribution of CoNS species
The numbers and proportions of isolates of different CoNS species are shown in Table 1. In dataset 1, a total of 14 staphylococcal species were identified. The most common CoNS species were S. simulans (25.0 %), S. epidermidis (25.0 %), S. chromogenes (15.4 %), S. haemolyticus (11.9 %), and S. warneri (10.3 %). Three isolates could not be identified by ribotyping or by MALDI-TOF MS and were grouped as Staphylococcus sp. In dataset 2, similarly as in the dataset 1, S. simulans (34.1 %) and S. epidermidis (30.7 %) were the most common species. Staphylococcus chromogenes was the third most common species (6.8 %). The proportion of both S. haemolyticus and S. cohnii isolates was 5.7 %. Three isolates could not be identified by ribotyping or MALDI-TOF MS and were grouped as Staphylococcus sp. The unidentified isolates may represent a new Staphylococcus species or one of the few CoNS species, like S. devriesei, which were not included in the ribotype and MALDI-TOF MS comparison databases at the time of the analyses. The four most common CoNS species represented 77.3 % of all 400 isolates.

In vitro antimicrobial susceptibility
The four major CoNS species differed in their in vitro antimicrobial susceptibility. Among them, antimicrobial resistance was most common in S. epidermidis (Tables 2  and 3). MIC distributions of the four major CoNS species for the tested antimicrobials in 2001 and 2012 are shown in Tables 2 and 3. The MIC results of all 400 isolates by species are shown in the Additional file 1: Table S1 and as a CoNS group in the Additional file 2: Table S2.
The majority, 74.4 and 74.1 % of S. epidermidis isolates in years 2001 and 2012, respectively, were resistant to benzylpenicillin (ECOFF 0.125 mg/l). Resistance to benzylpenicillin was also common in S. haemolyticus of which 64.9/40.0 % were resistant. Prevalence of isolates producing betalactamase, i.e. positive in the nitrocefin test, varied between CoNS species, and was lower in S. chromogenes (20.8/16.7 %) and S. simulans          (Tables 1, 2). Oxacillin resistance (conferring resistance also to methicillin), using ECOFF of 1.0 mg/l, was found in all the four main species (Tables 2, 3 Isolates harboring mecC were not found. Resistance to trimethoprim, measured only in dataset 2, was most common in S. simulans and S. haemolyticus, but the number of isolates of the latter species was low (Table 3). Staphylococcus epidermidis was the only species showing resistance to cefoxitin using the ECOFF of S. aureus (4 mg/l) ( Table 3). Resistance to fusidic acid was common in all other species but S. haemolyticus (Table 3). One S. simulans isolate was resistant to vancomycin, using the specific ECOFF (4 mg/l) for that species (Table 2).
Combining datasets 1 and 2, a total of 44.4 % of S. chromogenes, 23.8 % of S. haemolyticus, 45.4 % of S. simulans and 16.2 % of S. epidermidis isolates were susceptible to all antimicrobials tested. Resistance to more than one antimicrobial was most common in S. epidermidis isolates. Close to one-third (28.6 %) of them were resistant to >2 antimicrobials. Among the other three major species, one S. chromogenes isolate, one S. haemolyticus isolate, and five S. simulans isolates (4.6 %) were resistant to >2 substances. Seven S. epidermidis isolates (6.7 %) were multidrug-resistant (MDR = resistant to 3 or more classes of antimicrobials); one isolate was resistant to five different antimicrobial classes. Among the other species, only one MDR S. simulans isolate was found.

Discussion
The most common CoNS species in our data, S. simulans, S. epidermidis, S. chromogenes, and S. haemolyticus, belong to the CoNS species reported most frequently in numerous studies on bovine intramammary infection (IMI) or mastitis. Staphylococcus chromogenes has been isolated most commonly in almost all studies [11][12][13][25][26][27][28][29]. It is much more common in primiparous than multiparous cows [11,28,30], with peak occurrence around the first calving [4]. Staphylococcus chromogenes seems to be present in all herds [26][27][28][29] and has been frequently isolated not only from milk but also from bovine teat skin and orifice and from other body sites of heifers and cows [20,31]. Staphylococcus simulans and S. epidermidis are common causes of IMI in some herds but not found or only occasionally found in some other herds [26,27,29,30,32]. Both S. epidermidis and S. simulans are reported to be more common in IMIs of multiparous than primiparous cows [11,28,30]. For some reason, S. simulans is common in the Nordic countries [11,12,33] but not so much in Middle European countries [12,25,26]. Staphylococcus xylosus is commonly reported in Dutch and Belgian studies [13,26,27] but is rare in Finland, Norway and Sweden [6,11,28]. Staphylococcus haemolyticus is a fairly common finding in many studies [11,13,[26][27][28][29]. The reasons for variable proportions of CoNS species isolated from dairy cattle in different countries and individual herds are not fully elucidated but are likely related to different environmental conditions and herd management [10,31].
In the present study, genotypic identification of CoNS species and species-specific or CoNS-specific EUCAST ECOFFs when available were used. In only few other studies species-specific identification and the same cutoffs than here have been used [5,6]. In a Swedish study, prevalence of resistant isolates in the four major CoNS species was substantially lower than here [6]. Our results agree with them in that the most common resistance was to benzylpenicillin, but proportion of resistant isolates was clearly lower in the Swedish study. Only one out of 34 Swedish S. epidermidis isolates was resistant to oxacillin and harboured the mecA gene. In a Swiss study [5], oxacillin resistance was the most common resistance phenotype. They found as much as 47.0 % of the isolates (all CoNS together) to be oxacillin resistant, but the cutoff used was two dilutions lower (0.25 mg/l) than the current EUCAST ECOFF, which explains the discrepant results. The mecA gene was present in 9.7 % of the isolates classified as oxacillin-resistant. In the study by Frey et al. [5] the total proportion of penicillin-resistant isolates was lower (23.3 %) than in our study, using the same ECOFF, but the selection of CoNS species was different from ours. Results from the nitrocefin test were compared with results based on the penicillin MIC values of the isolates. The nitrocefin test performed better in detecting penicillin susceptible CoNS isolates; of isolates with MIC for benzylpenicillin >0.125 mg/l, considered as resistant, 23 % were negative in the test. These results agree with the study by Pitkälä et al. [34] comparing different betalactamase tests, who found no false positive but some false negative results for blaZ positive CoNS using nitrocefin disk test. In testing bovine S. aureus isolates, nitrocefin test has been found to be very reliable [34,35]. Frey et al. [5] carried out in vitro betalactamase testing with pre-incubation with benzylpenicillin but did not report the results, so we cannot compare them with ours. Interpreting results of bovine CoNS isolates from nitrocefin disk assay is challenging, as the color change is sometimes slow and not clear (Suvi Nykäsenoja, Evira, personal communication). The performance of the nitrocefin tests may not be fully satisfactory in detecting penicillin resistance of bovine CoNS. This is of practical importance because nitrocefin tests are widely used to predict betalactamase production of bovine staphylococci.
Staphylococcus epidermidis was the most resistant among the four major species identified. Almost onethird of our S. epidermidis isolates were resistant to >2 antimicrobials and close to 7 % were MDR, which were not found among the other species. This agrees with previous studies which also have reported S. epidermidis being frequently resistant to several antimicrobials [5,36]. The most common combination was resistance to penicillin and tetracycline. All mecA positive isolates, except one mecA positive S. sciuri isolate, were S. epidermidis. However, phenotypic methicillin (oxacillin) resistance was also common in the other CoNS species. Among the other three major species, MIC values to oxacillin in the methicillin-resistant isolates were mainly only one step higher than ECOFF and did not form a distinctly different population with clearly higher MIC values. According to many studies, methicillin-resistance is much more common in S. epidermidis than in other CoNS species [8,[36][37][38]. Among clinical S. epidermidis isolates from humans, 75-90 % are resistant to methicillin (reviewed by Otto [39]). Staphylococcus epidermidis differs from other mastitis causing CoNS in many aspects. It is a well-known human pathogen which causes nosocomial infections often associated with medical devices [39,40]. Staphylococcus epidermidis has a selection of virulence characteristics which include biofilm formation and antimicrobial resistance [3,40]. It has been suggested that bovine S. epidermidis may originate from humans [38,41]. A Finnish study did not find bovine methicillin-resistant S. epidermidis strains being closely related to human isolates [8].
Nearly all S. simulans isolates were resistant to trimethoprim. No specific ECOFF is available for this antimicrobial, and we used that of S. aureus. The MICs of trimethoprim of most S. simulans isolates were several steps above the ECOFF used, which indicates true resistance. More than one-third of S. epidermidis and S. simulans were resistant to fusidic acid according to ECOFF for CoNS. Looking at the MIC distributions of our isolates, this ECOFF seemed not optimal for this group of CoNS. In a Norwegian study, 10 % of bovine CoNS isolates were resistant to fusidic acid using the same ECOFF, but all isolates classified as resistant had MIC values several steps over the cut-off [42]. In some CoNS species resistance is more common ( Table 2). Almost half of S. epidermidis isolates of human origin have been resistant to fusidic acid and harbored the same horizontally acquired resistance determinants than reported in S. aureus [43].
Antimicrobial resistance among bovine CoNS isolates causes two types of concerns. First, it decreases options for antimicrobial treatment of mastitis as well as response to treatment. Mastitis caused by CoNS is mostly subclinical or mild clinical and routine treatment is not recommended [4]. In cases where treatment is warranted, resistance to benzylpenicillin is an issue at least in countries where penicillin or aminopenicillins are the drugs of choice [4]. Penicillin resistance of three of the four major CoNS species was here at so high level that penicillin can no more be considered as the first treatment option in mastitis caused by these species. Another option could be macrolides to which some degree of resistance was also found, in particular among S. epidermidis isolates. Prevalences of oxacillin resistant isolates were alarmingly high in all four major species. Cloxacillin is commonly used to treat mastitis, also mastitis caused by CoNS, so this is of practical relevance. If the causing CoNS strain harbors a mec gene, treatment with any betalactam antimicrobial is inefficient and only increases selection pressure.
The second concern is related to public health: cows can pass resistant CoNS to humans via direct contact or indirectly [37]. Bovine CoNS can also act as a reservoir for resistance determinants [3,44]. The greatest concern is methicillin-resistance which was common in bovine CoNS and presents a relevant risk for public health [38,40]. There is evidence for transfer of resistance determinants between staphylococcal species and also from CoNS to the more pathogenic species S. aureus [9,[44][45][46]. Regarding critically important antimicrobials [47] included in the present study, the situation was good as only one CoNS isolate (S. simulans) was resistant to vancomycin and no isolates to ciprofloxacin.
Unfortunately specific ECOFFs are not yet available for all CoNS species, and we had to use those of S. aureus for several antimicrobials. EUCAST ECOFFs are not veterinary specific but isolates originate from multiple sources and perhaps mainly from humans. Most studies on antimicrobial resistance of CoNS have used CLSI (Clinical and Laboratory Standards Institute) [48] breakpoints for veterinary pathogens. CLSI breakpoints are aimed for clinical purposes only [48]. They are derived from animal specific microbiological, pharmacokinetic and pharmacodynamic data, which is not relevant for studies on in vitro susceptibility of epidemiological data sets [15]. Furthermore, animal specific breakpoints are not available for all antimicrobials and those based on human data have been used [48]. CLSI documents do not give specific breakpoints for bovine CoNS, but some breakpoints are available for S. aureus. In general, CLSI breakpoints are higher than EUCAST ECOFFs. Studies which have used different breakpoints for resistance cannot be compared. After phenotypic screening of antimicrobial resistance, as done here, the next step would be to study the genetic mechanisms for resistance. For many antimicrobials several genes can code for resistance in staphylococci, and new genes are discovered [44,45]. Selection of a representative panel of resistance genes for genotypic studies is challenging.

Conclusions
The most common CoNS species were S. simulans, S. epidermidis, S. chromogenes, and S. haemolyticus. Staphylococcus epidermidis differed from the three other most common CoNS species isolated from mastitic bovine milk samples as resistance to most tested antimicrobials was more common in S. epidermidis than in S. chromogenes, S. haemolyticus or S. simulans. Except one S. sciuri isolate, all mecA gene positive isolates were S. epidermidis. Resistance to more than two antimicrobials was also common in S. epidermidis.