Molecular typing and antimicrobial resistance profiling of 33 mastitis-related Staphylococcus aureus isolates from cows in the Comarca Lagunera region of Mexico

Mastitis in cows is a major cause of economic losses and it is commonly associated with Staphylococcus aureus. Little is known about the S. aureus lineages causing mastitis in Mexican cattle. The aim of this study was to type S. aureus isolates causing mastitis in cows from the Comarca Lagunera region in Mexico in 2015–2016. Multi-locus variable number tandem repeat fingerprinting (MLVF) of 33 S. aureus isolates obtained from 210 milk samples revealed the MLVF clusters A (n = 1), B (n = 26), C (n = 5) and D (n = 1). Spa-typing showed that clusters A and B represent the spa-type t224, cluster C includes spa-types t3196 and t416, and cluster D represents spa-type t114. The different spa-types were mirrored by the masses of protein A bands as detected by Western blotting. Antimicrobial susceptibility testing showed that one isolate was susceptible to all antimicrobials tested, whereas all other strains were resistant only to benzylpenicillin. These findings show that only four S. aureus lineages, susceptible to most antimicrobials, were responsible for causing mastitis at the time of sampling. Lastly, many isolates carried the same small plasmid, designated pSAM1. The high prevalence of pSAM1 amongst the antimicrobial-susceptible isolates suggests an association with bovine colonization or mastitis rather than antimicrobial resistance.


Results
Isolation of S. aureus mastitis strains. In total, 331 samples collected from milk of Holstein dairy cows (n = 226) and nasal swabs (n = 105) collected from their calves were collected from 14 farms in the region called Comarca Lagunera in Mexico (Fig. 1). The milk samples were taken from single quarters with clinical mastitis, where the criteria for mastitis were a somatic cell count > 400,000 SCC/mL. After a combination of Gramstaining, coagulase testing and MALDI-TOF analysis, 33 isolates of S. aureus obtained from individual cows at seven different farms (designated A-G) were identified (Table 1). To determine whether the potential presence of S. aureus strains resulted in transmission to calves, nasal samples (n = 105) were collected from calves that had contact with the cows with mastitis. All of the nasal swabs from calves tested negative for S. aureus, which suggests that transmission events between cows and calves were infrequent or remained undetected.
MLVF and PCR analysis. The 33 mastitis-related S. aureus isolates were subjected to PCR analysis for the presence of genes of the micrococcal nuclease MN (nuc) and the methicillin resistance gene mecA. The nuc gene was detected in 27 isolates (Fig. 2). None of the isolates contained the mecA resistance gene. This matched with the antimicrobial susceptibility testing, because none of the isolates were resistant against cefoxitin or oxacillin. It was therefore concluded that all isolates were methicillin-susceptible S. aureus (MSSA) 22 .
For typing of the isolates, the total DNA was subjected to MLVF. The data obtained with the Bioanalyzer was used to create the MLVF dendrogram shown in Fig. 2. Altogether, the MLVF analysis of S. aureus isolates from milk samples revealed 4 different clusters, which were designated A (n = 1), B (n = 26), C (n = 5) and D (n = 1).
Judged by the expected lengths of the PCR fragments for the genes sspA (110-bp), spa (195-bp), sdrE (603bp), sdrC (651-bp), sdrD (735-bp), clfB (850-bp), clfA (1014-bp) from the control strain S. aureus USA300, a clear variation in the lengths of several PCR fragments was observed among the four identified mastitis isolate Further, for all mastitis isolates the spa product was smaller compared to that of the control strain USA300. The amplified spa fragments from isolates belonging to clusters A and B appeared to be similar in length, while those for isolates in Cluster C seemed to be even smaller. Based on the band intensity, we assume that the spa and sspA PCR products had an identical size for isolates in cluster C. Lastly, the PCR products tentatively attributed to sdrD and sdrE seemed to be smallest for isolates from cluster B.
Spa-typing. The  . Spa-type t224 was found to be the most common as it was assigned to 23 isolates. Spa-type t3196 was assigned to 3 isolates, t416 to 2 isolates and t114 was identified for only one isolate. Notably, all isolates belonging to the MLVF clusters A and B had the spa-type t224, whereas spa-type 114 was unique for MLVF cluster D. In contrast, isolates belonging to MLVF cluster C had either spa-type t3196 or t416. The variation in the number of repeats within the spa genes of the different isolates was in agreement with the variations in length that were observed for the respective spa-specific PCR fragments in the MLVF (Figs. 2 and 3). When comparing the composition of the different types of identified repeats within the spa gene, it seems that isolate G-1 (t114) includes two repeat sequences (r16 and r13) that are not present in the spa genes of the other mastitis isolates (Fig. 3).

Western blotting analysis.
To verify the observed size differences of the spa and sdrE PCR products in the MLVF pattern, a Western blotting analysis was performed to detect the respective protein A and SdrE proteins. For this analysis the covalently cell wall-bound proteins of 6 different mastitis isolates were extracted. As controls, the S. aureus strain Newman, which was originally isolated in 1952 from a human infection, and the mastitis-related S. aureus strain RF122 were used. The strain Newman is known to produce protein A (55.6 kDa) and SdrE (126.5 kDa), while the RF122 strain lacks expression of protein A. From the Mexican mastitis isolates, one strain of each MLVF cluster was selected. Specifically, these included isolates F-1 (t3196), B-1 (t224), G-1 (t114) and E-3 (t224). Using a polyclonal antibody against SdrE, both the expression of SdrE and protein A could be determined by Western blotting (Fig. 4). In the case of protein A, this relates to the fact that it efficiently binds the Fc region of rabbit polyclonal antibodies. The results of the blot are congruous with the VNTR repeats of the different spa-types. The VNTR repeats of the isolates B-1 and E-3 with spa-type t224 (MLVF clusters A and B, respectively) were equal in size and slightly smaller than those of the control strain Newman. For isolate G-1 with spa-type 114 (MLVF cluster D), a smaller size for protein A was detected. The smallest form of protein A was detected for isolate F-1 with spa type t3196 and belonging to the MLVF cluster C (Figs. 3 and 4). www.nature.com/scientificreports/ Unexpectedly, no protein band corresponding to SdrE was detected for strain RF122 (Fig. 4). In the sample of strain Newman a clear SdrE signal at the expected height of 126.5 kDa was detected together with multiple degradation bands. Similar patterns were observed for isolates F-1, B-1 and E-3, indicating that these isolates do express SdrE. Yet, no SdrE signal was detectable for isolate G-1 (Fig. 4), indicating that it does not express the SdrE protein, even though a possible sdrE PCR fragment appeared to be detectable in the MLVF analysis (Fig. 2).
Antimicrobial resistance profiling. To determine possible variations in antimicrobial resistance and possible association of resistance phenotypes with the MLVF and/or spa-types of the investigated mastitis isolates, all 33 isolates were examined for antimicrobial susceptibility. The analysis of antimicrobial susceptibility revealed that for the antimicrobials with clinical breakpoints all isolates were susceptible to erythromycin, oxacillin, and tetracycline. Thirty-two isolates were resistant to benzylpenicillin (97%). Only strain G-1 (3%) was susceptible to all tested antimicrobials. Lastly, all isolates tested negative in the screen for cefoxitin resistance.

Plasmid profiling and sequence analysis.
To determine the presence of plasmids, plasmid DNA was isolated from all mastitis isolates. Using agarose gel electrophoresis it was shown that 22 out of the 33 isolates contain a plasmid of identical size (Supplemental Fig. S1). No plasmid DNA was detected in strains with the spatypes t114 and t416. On basis of the electrophoresis results, the plasmid DNA seemed to be of the same size in all strains. The plasmid DNA extracted from the isolates F-1 (t3196), E-3 (t224) and C-11 (t224) was sequenced. Table 1. Bacterial strains used in this study. a The letters A to G in the names of the isolates refer to the different farms from which they were collected. Mexican mastitis-associated isolates; white-yellowish colonies Mexican mastitis-associated isolates; orange-pinkish colonies C-13 Mexican mastitis-associated isolates; white-yellowish colonies Mexican mastitis-associated isolates; orange-pinkish colonies D-1 Mexican mastitis-associated isolates; white-yellowish colonies F-2 www.nature.com/scientificreports/ Interestingly, the respective plasmid sequences were found to be identical, consisting of 1508-bp. Accordingly, the respective plasmid was named pSAM1. Searches for open reading frames (ORF) combined with Blast analyses showed the presence of several ORFs in pSAM1 (Fig. 5).
The plasmid sequences showed a G + C content of 32.6 mol%, which is in agreement with the G + C content determined for S. aureus (33 mol%) 23 . Based on DNA sequence similarity, plasmid pSAM1 can be classified as a rolling circle replicating (RCR) plasmid that belongs to the pSN2 Family 21 . In total, six ORFs were identified on pSAM1 (Fig. 5). The largest identified orf of pSAM1 encodes a protein of 154 amino acid (AA) residues which, on the basis of sequence similarity, corresponds with the replication protein RepL. Downstream of the repL gene, the orf2 gene is located, which encodes a protein of 62 residues with unknown function. An inverted repeat with a ∆G° of − 12.6 kcal/mol that may function as a rho-independent terminator is located downstream of orf2. 136-bp   www.nature.com/scientificreports/ downstream of this stem-loop another stem-loop with a ∆G° of − 10.2 kcal/mol is located, which could function as a terminator for the orf3. Blast analysis shows that the sequence similarity with other plasmids stops directly after either of these two stem loops, suggesting that they are potential sites for recombination. Blast analyses also showed that in S. capitis a nucleotide sequence is present encoding a protein of 73 residues with unknown function, which overlaps with orfs 3, 4 and 5 24 . Also, in other S. aureus plasmids like pSK3, pSK6 and pUR2355, mutations have been identified for this ORF (Fig. 5), suggesting that the presently identified orfs 3, 4 and 5 do not encode a functional protein in S. aureus. The last orf, orf6, has a G + C content of 46%, which is much higher than the average G + C content of 32.6% observed for the rest of the plasmid. Overall, plasmid pSAM1 shows a high degree of sequence similarity with the plasmids P2D1C1, P2D8C1 and P2D15C1 that were previously identified in other S. aureus strains 25 . Compared to these plasmids small mutations are located in the repL gene. The overall homology with a plasmid from S. aureus strain SAP104B (1552-bp) was also very high although the SAP104B-derived plasmid contained some extra nucleotide sequences in between orf3 and the downstream located stem loop. On basis of its repL gene plasmid pSAM1 was grouped in rep family 10b 26 .

Discussion
In the present study, we describe the diversity of 33 S. aureus isolates causing mastitis in Mexican cows by MLVF and spa-typing, as well as their susceptibility for a range of commonly used antimicrobials.
Our typing analysis shows that the S. aureus isolates associated with mastitis group into four MLVF clusters, which overlap with the determined spa-types. The predominantly encountered spa-type t224 was determined for all isolates of the MLVF clusters A and B. In recent studies, this spa-type has been reported for S. aureus isolates from milk of cows with intra-mammary infections in European countries, China, Japan and Egypt 9,27,28 . The present identification of mastitis-associated S. aureus isolates with spa-type t224 in Mexico implies that this staphylococcal type has now been detected on all continents except Australia. However, in other studies, t224 was not the predominantly encountered spa-type. Of note, the spa-type t224 is frequently observed for isolates belonging to the ST97 group that is known to be carried by cows, humans and swine. In swine, such isolates are occasionally MRSA 29 . Whether the isolates used in this study also belong to the ST97 remains to be investigated. The MLVF cluster C includes the spa-types t416 and t3196. S. aureus isolates with spa-type t416 have been reported in Austria and China, mainly for human MRSA isolates 30,31 . According to the Ridom Spa Server (https:// spase rver. ridom. de/), S. aureus isolates with the spa-type t3196 have been identified in Belgium, Denmark, and the Netherlands. However, no information is currently available concerning the respective host species, or whether these t3196 isolates were associated with bovine mastitis. The MLVF cluster D is formed by a singleton with spa-type t114, which has been described for S. aureus isolates from the milk of cows with mastitis in Brazil and China 32,33 .
The nuc gene, which encodes a thermostable nuclease, has been described as a specific marker for direct detection of S. aureus involved in infection of humans 34 . Nonetheless, in 4 of the 33 presently investigated S. aureus isolates the nuc gene was apparently not present. Interestingly, Javid et al. reported that nuc could only be detected in a fraction of S. aureus isolates collected from cows with mastitis in India 35 . These observations are intriguing, as it has been proposed that the micrococcal nuclease is involved in the evasion of the human immune system, by preventing the capture and elimination of S. aureus in neutrophil extracellular traps 36 .
It was previously reported that the spa, clfA and sdrC genes of the bovine mastitis-related S. aureus strain RF122 are pseudogenes, because of the presence of premature stop codons 37 . This phenomenon was not observed for the spa genes of the presently investigated S. aureus isolates from Mexico, because protein A was detectable for all isolates by Western blotting. On the other hand, RF122 contains a complete sdrE gene, which was confirmed in a Western blot using an antibody against the region A of the SdrE protein of S. aureus strain Newman 38 . The sequence similarity of region A in the SdrE proteins of strains RF122 and Newman is only 76%. This difference in the sequence could be the reason that SdrE was not detected by Western blotting in the sample for covalently cell wall-bound proteins of S. aureus RF122. Another possible explanation could be the use of whey permeate as a growth medium for S. aureus in the present study, which may have lowered the expression of SdrE in the RF122 strain.
Antimicrobial resistance testing showed that the majority of the isolates (32/33) were resistant to penicillin. This is not surprising, as this antimicrobial is the first option to treat bovine mastitis. In many dairy farms it is common practice to use intramammary infusions of antimicrobials as a prophylactic approach to prevent and control mastitis in all dairy cows during the dry period, primarily with penicillins, cephalosporins, or other beta-lactam drugs 6 .
In 67% of the investigated isolates, the presence of a single plasmid was detected. Based on agarose gel electrophoresis and sequencing data, it seems that all plasmid-bearing strains contained the same plasmid, which was named pSAM1. Sequence analysis of the repL gene showed that pSAM1 belongs to the pSN2 family, which encompasses the smallest RCR plasmids identified in staphylococci 21 . The presence of pSAM1 was evidently not linked to any of the MLVF clusters, a spa-type or a specific antimicrobial resistance. Besides the putative replication protein RepL, no function has so far been assigned to any of the five other proteins encoded by pSAM1. The high prevalence of pSAM1 amongst the presently investigated S. aureus isolates could suggest an association of this plasmid and its encoded proteins with bovine colonization or even mastitis. However, the latter remains to be investigated, preferably in the context of follow-up studies on possible adaptations that allowed the present mastitis isolates to thrive in the bovine mammary gland.

Materials and methods
Sample collection and selection. Cases of mastitis were identified using the California Mastitis Test (CMT), which is a simple on-farm method to indirectly indicate the level of the somatic cell count (SCC) 39 . Two hundred and twenty six individual quarter milk samples were collected in 2015-2016 from 14 different farms in the Comarca Lagunera region in Mexico. Foremilk was discarded after which a little amount of milk was drawn in the CMT paddle. An equal amount of the CMT reagent was added and gently mixed. After a few seconds, the reaction was scored following the scale described by Easterday et al. 39 . The weakly positive samples with 400,000 to 1,000,000 SCC/mL (mixture is slightly mucous, but still without gel formation), positive samples with 800,000 to 5,000,000 SCC/mL (distinct gel formation), and the samples with more than 5,000,000 SCC/mL (strongly positive with strong gel formation that adheres to the paddle) were collected for plating. For transmission analysis single nasal swabs were collected from 105 calves of these cows.
Identification of isolates and culture conditions. All milk samples and nasal swabs collected from calves were directly plated at the farm on mannitol salt agar (Becton Dickinson de Mexico). Subsequently, the plates were incubated at 37 °C overnight to determine the presence of bacteria. From each plate one potential S. aureus colony was selected, depending on the following colors: golden-yellow, creamy-yellow, yellowish, orange, pinkish and white (Table 1). After Gram-staining, all the Gram-positive cocci were subjected to the catalase 40 and Voges Poskauer 41 tests. Finally, all colonies that tested potentially positive as S. aureus in the biochemical tests were collected and shipped in mannitol salt agar to the University Medical Center Groningen, the Netherlands. The potential S. aureus isolates were cultured on Blood Agar (BA) plates containing 5% sheep blood and aztreonam (Media products, Groningen, the Netherlands). Firstly, a quick screen of the strain was done using the Pastorex Staph Plus test (Bio-Rad, Marnes-la-Coquette, France). To select and identify S. aureus strains, single colonies were analyzed by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) with a Microflex LT Biotyper (Bruker Daltonics, Bremen, Germany) as previously described 42 . Only S. aureus strains with log scores ≥ 2 were selected. Thirty-three isolates were positively identified as S. aureus. The isolates were cultured overnight in tryptic soy broth (TSB, Oxoid, Hampshire, UK) at 37 °C with shaking (250 rpm) and, subsequently, stored in 12% glycerol (Sigma-Aldrich, Zwijndrecht, the Netherlands) at −80 °C for further analyses.
Antimicrobial susceptibility testing. Antimicrobial susceptibility was measured with the VITEK 2 system (ID. Card: AST-GP69, bioMerieux Corporate, Marcy l'Etoile, France) in accordance with the manufacturer's instructions. For quality control of the card, the strain S. aureus ATCC 29,213 was used as suggested by the supplier. The following antimicrobials were evaluated: benzylpenicillin, cefoxitin, enrofloxacin, erythromycin, kanamycin, oxacillin and tetracycline. The minimum inhibitory concentrations (MIC) obtained from the VITEK analysis were validated using the Advanced Expert System following the Clinical and Laboratory Standards Institute (CLSI) of the veterinary guidelines 43 , which employed the human interpretive data taken from the CLSI M100-S series 44 . The MIC breakpoints are specific for S. aureus udder isolates and are expressed as μg/ml in parentheses. These breakpoints are for benzylpenicillin (Susceptible [S] ≤ 0.12, Resistance [R] ≥ 0.25); erythromycin (S ≤ 0.5, R ≥ 8); oxacillin (S ≤ 2, R ≥ 4) and tetracycline (S ≤ 4, R ≥ 16). For cefoxitin only negative or positive test results were obtained. For the veterinary antimicrobials enrofloxacin and kanamycin, interpretative data are not available for mastitis-associated S. aureus. All results are presented in the Supplementary Table S1. MLVF and PCR analysis. Total DNA was extracted from S. aureus colonies picked from BA plates by beadbeating according to Glasner et al. 14 .
MLVF typing was performed following the adjusted protocol described by Sabat et al. 15 and Glasner et al. 14 . Briefly, 1 μL of total DNA was subjected to a multiplex PCR of seven VNTRs of S. aureus (sdrC, sdrD, sdrE, clfA, clfB, sspA and spa) (Eurogentec, Maastricht, the Netherlands). Subsequently, 1 µL aliquots of the resulting amplicons were separated employing a Bioanalyzer 2100 (Agilent Technologies, Palo Alto, USA) with microfluidic DNA 7500 chips following the manufacturer's instructions. The strain USA 300 was used as the technical control in each Bioanalyzer run to ensure the reproducibility of the data. Data was analyzed with the GelCompar II software (Applied Maths, Kortrijk, Belgium). For this purpose, the data generated in the Bioanalyzer were imported as CSV files. The position tolerance and optimization were set at 0.9% and 0.5% respectively, as with these settings the control isolate (USA300) displayed identical MLVF banding patterns. To calculate the pairwise similarity coefficients, the dice formula was used. A dendrogram was created with the unweighted pair group method using average linkages (UPGMA). A cut-off of 81% was set to allow a differentiation between clusters. Identical banding patterns were assigned to the same subtype; when one or more bands differed in the MLVF pattern, they were assigned to different clusters.
All S. aureus strains were also screened for the presence of the mecA gene encoding methicillin resistance and the nuc gene for the micrococcal nuclease MN produced by S. aureus, as previously described 34,45 . Spa-typing. Spa-typing was performed as described by Harmsen et al. 13 using the Ridom Staph Type software version 2.2.1 (Ridom GmbH, Würzburg, Germany). An ABI Prism 3130 genetic analyzer (Applied Biosystems, Foster City, USA) was used to obtain the DNA sequences.
Plasmid sequencing. Plasmid DNA was extracted using the Innu PREP Plasmid Mini Kit (Analytik Jena, Jena, Germany) following the special protocol of the supplier, where in the second step the cell pellet was resus-