Isolates

A total of 40 S. epidermidis (SE) isolates, 28 from humans (Hu) and 12 from companion animals (CpA), and 21 S. haemolyticus (SH) isolates, eight Hu and 13 CpA, were investigated (Table 1 and S1 Table). All Hu isolates were recovered from patients in two separate acute hospitals in Dublin, Ireland; eight SE-Hu isolates were associated with neurosurgical meningitis and were recovered from either external ventricular drains (EVDs) in patients with device-related meningitis or from non-EVD cerebrospinal fluid specimens taken by lumbar puncture between 2004 and 2006 [34]. The remaining 20 SE-Hu and the eight SH-Hu isolates were recovered from blood cultures of patients attending a separate acute hospital between 2010 and 2011. The 12 SE-CpA isolates examined were recovered from a cat (n = 1), dogs (n = 10) and a horse (n = 1). The 13 SH-CpA isolates were recovered from a cat (n = 1), dogs (n = 3) and horses (n = 9). These CpA isolates were recovered primarily in animals with wounds or infections attending a tertiary referral veterinary hospital in Dublin between 2004 and 2011 (S1 Table). Isolates were stored on commercially available cryobeads (Microbank, Pro-lab Diagnostics, Cheshire, UK) at -70°C.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. MLST clonal complexes and sequence types and SCCmec- associated genes detected in S. epidermidis and S. haemolyticus isolates recovered from infections in humans and companion animals.

https://doi.org/10.1371/journal.pone.0138079.t001

Confirmation of isolates as S. epidermidis and S. haemolyticus

Isolates were confirmed as either S. epidermidis or S. haemolyticus by PCR amplification and sequencing of the 16S rRNA gene using previously described primers [35]. Sequence analysis was performed using the BioNumerics (version 7.1; Applied Maths, Ghent, Belgium), and ApE (v1.17) software packages. Homology searches were performed using BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) [36].

Antimicrobial susceptibility testing

All isolates were investigated for MR either as described previously [37] using 10 μg and 30 μg cefoxitin disks (Oxoid Ltd., Basingstoke United Kingdom) according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) methodology and interpretive criteria for disk diffusion tests or using oxacillin broth microdilution assays according to the Clinical Laboratory Standards Institute (CLSI) methodology for broth microdilution [38, 39]. All isolates underwent antimicrobial susceptibility testing against a panel of 23 antimicrobial agents used for antibiogram-resistogram (AR) typing according to EUCAST methodology and a combination of the interpretive criteria by EUCAST [39], CLSI [38] and Rossney et al. [37]. The 23 agents tested were amikacin, ampicillin, cadmium acetate, chloramphenicol, ciprofloxacin, erythromycin, ethidium bromide, fusidic acid, gentamicin, kanamycin, lincomycin, mercuric chloride, mupirocin, neomycin, phenyl mercuric acetate, rifampicin, spectinomycin, streptomycin, sulphonamide, tetracycline, tobramycin, trimethoprim, and vancomycin. All isolates were also tested for clindamycin and linezolid resistance using EUCAST methodology and interpretive criteria. All disc concentrations and interpretive criteria used are listed in S2 Table. The EUCAST and CLSI recommended S. aureus reference strains ATCC29213 and ATCC25923 were used as quality control strains for antimicrobial susceptibility testing. Multidrug-resistance (MDR) was defined as resistance to three or more classes of antimicrobial agents.

DNA isolation, PCR and sequencing

Total genomic DNA for use in 16S rDNA sequencing, DNA microarray profiling, MLST, SCCmec typing and dru typing was extracted using the StaphyType kit (Alere Technologies GmbH, Jena, Germany) according to the manufacturer’s instructions. Apart from PCR for DNA microarray profiling, all PCRs were performed using GoTaq DNA polymerase (Promega, WI, USA). PCR products were purified using the GenElute PCR clean-up kit (Sigma, Wicklow, Republic of Ireland) or, for MLST, the QIAquick 96 well PCR purification kit (Qiagen, Crawley, UK). All DNA sequencing reactions were carried out commercially by Source BioScience LifeSciences (Waterford, Republic of Ireland).

DNA microarray profiling

All isolates underwent DNA microarray profiling using the StaphyType kit (Alere) according to the manufacturer’s instructions. The DNA microarray detects 333 gene targets including staphylococcal antimicrobial-resistance, virulence, SCCmec and ACME-arc genes [40, 41]. All isolates harboring the mecA gene were subjected to additional DNA microarray profiling using separate mecA allele typing arrays (Alere) designed to identify 15 different mecA alleles as previously described [42]. Using this method, the mecA alleles were designated according to their GenBank accession numbers [42]. The sequences of mecA alleles were compared using the GenBank sequences and the alignment software program Mega 6.0 [43].

SCCmec typing

Any isolates found to carry unusual combinations of SCC or SCCmec genes using the StaphyType DNA microarray underwent multiplex PCRs to confirm the presence or absence of particular genes. This included previously described multiplex SCCmec typing PCRs to detect the mec gene complexes A-C, the ccr gene complexes ccrAB1-AB4 and ccrC and the joining regions of SCCmec types I-IV [44, 45]. The following S. aureus strains were used as positive controls for SCCmec typing PCRs: phenotype II 43.2 (SCCmec I, ccrAB1) [46], CA05 (SCCmec IV, class B mec, ccrAB2) [47], WIS (class C mec) [48], 07.4/0237 (SCCmec II) [46], JCSC 4744 (IVA) [44], M00/0005.2 (ccrAB4) [49], and E0898 (SCCmec III, class A mec, ccrAB3 ccrC) [49]. All isolates found to carry ccrC underwent multiplex PCR for the ccrC allotypes ccrC2 and ccrC8 to differentiate between SCCmec V (ccrC2) and V_T (ccrC2 and ccrC8), as described previously [50], using the S. aureus clinical isolate M06/0318 (SCCmec V_T) as a positive control strain [51]. All isolates found to harbor SCCmec IV or possible novel SCCmec types with mec and/or ccr genes indicative of SCCmec IV, underwent SCCmec IV subtyping PCR as previously described [52], using the following S. aureus strains as positive controls: CA05 (SCCmec IVa) [47], 8/63P (SCCmec IVb) [47], JCSC4788 (SCCmec IVc) [53], JCSC4469 (SCCmec IVd) [53], M04/0177 (SCCmec IVg) [49] and E1749 (SCCmec IVh) [49].

PCR-based detection of antimicrobial resistance genes

Isolates were subjected to PCR-based detection of antimicrobial resistance genes to confirm (i) the absence of a resistance gene(s) if an isolate exhibited phenotypic resistance to an antimicrobial agent and no corresponding resistance gene was detected using the StaphyType DNA microarray, or (ii) the presence of a resistance gene(s) detected in an isolate using the DNA microarray which did not exhibit phenotypic resistance to the corresponding antimicrobial agent(s). This included PCRs to detect the presence of aacA-aphD, aadD, aphA3, cat-pC221, dfrS1, erm(A), erm(C), ileS2, merA, merB, qacA and qacC. Lastly, PCRs were also performed to detect additional trimethoprim resistance genes (dfrG and dfrK) not detected using the DNA microarray in isolates that exhibited phenotypic resistance to trimethoprim but lacked dfrS1. The oligonucleotide primers used for these PCRs are detailed in S3 Table.

Direct repeat unit (dru) typing

All methicillin-resistant staphylococcal isolates investigated (n = 55) were subjected to dru typing using previously described primers and thermal cycling conditions [16]. The BioNumerics tandem-repeat sequence typing (TRST) plug-in was used for dru sequence analysis and assignment of dru types (dts). The dru region of five MRSE isolates could not be amplified by the originally described dru typing primers. For these isolates, the dru region was amplified using previously described primers mecAF and ISmecR that extend from mecA to IS431 (S3 Table) [46]. Minimum spanning trees (MSTs) were constructed based on the dts identified as previously described [54]. Due to the increased likelihood of recombination amongst S. epidermidis and S. haemolyticus populations, the bin distance was set to 1%, i.e., the distance between two entries with >99% similarity was 0 (a distance interval of 99 to 100% similarity equals a distance of 0) on the MST, and the distance between two entries with 98 to 99% similarity was 1 (a distance interval of 98 to 99% similarity equals a distance of 1).

MLST

All isolates were subjected to MLST. A previously described species-specific scheme, including primers and thermal cycling conditions, was used for MLST of S. epidermidis isolates [55]. A S. haemolyticus-specific scheme was used for MLST of S. haemolyticus isolates [32] but primer SH1200R was substituted with a novel primer (SH1200R2 5′-ACCAGGCTTGTCACCATGA-3′) and SH1431F was substituted with a novel primer (SH1431F2 5′-TCAGACCAACAATTCCCACC -3′) to increase amplimer yields. For S. haemolyticus isolates, thermal cycling conditions consisted of an initial denaturation step of 94°C for two min, followed by 35 repeated cycles of 94°C for one min, 51°C for 30 s and 72°C for 30 s, and a final elongation step of 72°C for five min. Sequence analysis was performed using the ABI prism Seqscape (version 2.6, Applied Biosystems, Foster City, CA) or BioNumerics software. Staphylococcus epidermidis alleles and sequence types (STs) were identified using the S. epidermidis-specific MLST database (http://sepidermidis.mlst.net/) [56]. As there is no publicly available S. haemolyticus MLST database, alleles and STs were assigned identification numbers using our own in-house database (S4 and S5 Tables). For both species, assignment of STs to CCs was performed using the eBURST algorithm, where an ST was only assigned to a CC if it shared at least 6/7 MLST loci with at least one other ST within a CC [57].

Statistical analyses

In order to determine if the differences in the prevalence of antimicrobial resistance genes and phenotypes and ACME were significant between SH and SE isolates or between isolates recovered from Hu and CpA hosts, two-tailed Fisher’s exact tests were utilized. These analyses were carried out using GraphPad QuickCalcs (http://www.graphpad.com/quickcalcs/index.cfm).

Nucleotide accession numbers

The nucleotide sequences of the mecA-IS431mec amplimers for MRSE isolates 23767, 28427, 31169, 408 996.1, and BM11 that lacked the dru region have been deposited in the GenBank database under accession numbers KP265311, KP265312, KP265313, KP265314 and KP265315, respectively.

Methicillin resistance, genotypes and SCC-associated genes among SE isolates

In total, 33/40 (82.5%) SE isolates exhibited MR and carried mecA (Table 1). Twelve STs were identified amongst the MRSE isolates with 22/33 (66.7%) and 8/33 (24.2%) isolates belonging to CC2 clusters I and II, respectively (Table 1) [33]. ST2 was common to both MRSE-Hu and-CpA isolates, and this was the predominant ST identified amongst both groups (9/22, 40.9% MRSE-Hu and 9/11, 81.8% MRSE-CpA) (Table 1).

Overall 25/33 (75.8%) MRSE isolates were assigned to SCCmec types III, IV and VI but a single SCCmec type, SCCmec IV (most commonly subtype IVc), predominated amongst both MRSE-Hu (13/22, 59.1%) and MRSE-CpA (7/11, 63.6%) (Table 1). Based on DNA microarray analysis and PCR, non-typeable (NT) SCCmec elements, tentatively designated NTs 1–8, were detected among 8/33 (24.2%) MRSE (Hu and CpA) isolates, as these lacked, contained additional, or had unusual combinations of mec and/or ccr genes (Table 1). However, half of these NT SCCmec elements consisted of class B mec with ccrAB2 indicative of SCCmec IV but they also carried additional ccr genes (NTs 1–4, Table 1) with NTs 1–3 also harboring ACME-arc (Table 1). Three further NTs carried class A mec with unusual combinations of ccr genes or ACME-arc genes (NTs 5, 6 & 8, Table 1). The final NT SCCmec element carried class C mec with ccrAB2, SCCmec IV subtype IVh and ACME-arc genes (NT7, Table 1).

Six and four mecA alleles were identified amongst the MRSE-Hu and-CpA isolates, respectively (Table 1), but these shared >99.85% DNA sequence identity and differed by a maximum of three nucleotide bases. All mecA alleles detected amongst the MRSE-CpA were also detected amongst the MRSE-Hu (Table 1). The mecA allele ABSA010000166 previously detected in S. aureus, S. pseudintermedius and SE was detected in 10/22 (45.5%) MRSE-Hu and 4/11 (36.4%) MRSE-CpA (Table 1). With the exception of one MRSE-CpA isolate harboring SCCmec III, all of the MRSE isolates in which this allele was detected harbored SCCmec IV.

Four mecA alleles were detected among isolates of the most prevalent MRSE genotype (ST2-MRSE-IVc) with only one allele (ABSA01000066) common to both hosts (6/9, 66.7% MRSE-Hu and 2/9, 22.2% MRSE-CpA; Table 1), however, all mecA alleles identified among ST2-SCCmec IVc isolates differed by a maximum of three nucleotides indicating their close similarity. The ST2-MRSE-IVc isolates either lacked the dru region or were assigned to one of four dts, with non-typeable dts common to Hu and CpA isolates (Table 1). However, two of the remaining ST2-MRSE-IVc dts, dt10h and dt9g, were deemed to be closely related (MST value of 2.5 i.e. 96.5–97% DNA sequence identity; S1A Fig) and were identified from a Hu and CpA host, respectively.

Among the MSSE, four of five STs identified belonged to CC2 (Table 1). Only one MSSE-CpA isolate was identified and was distinct from the Hu isolates in terms of ST and the presence of ACME-arc genes. Among the MSSE, two possible novel SCCs (tentatively designated SCCs 1 and 2) were detected consisting of ccrAB2 alone or in combination with ccrAB1 and ACME-arc genes (Table 1).

The ACME-arc genes were more common amongst SE-CpA (6/12, 50%) than SE-Hu (4/28, 14.3%) (p = 0.04) (Table 1).

Methicillin resistance, genotypes and SCC-associated genes among SH isolates

Among the SH isolates 95.2% (20/21) were MR and carried mecA (Table 1). Eight STs were identified, seven of which were assigned to a single CC (CC1) (Table 1 and S4 Table). While ST1 was the most common ST among SH isolates and was the only ST identified in both SH-Hu and-CpA isolates, ST2 was more common among the SH-CpA isolates (Table 1).

Previously described SCCmec elements, either SCCmec V or V_T, were detected in only 7/20 (35%) MRSH isolates (Table 1). Four NT SCCmec elements were detected and tentatively described as NTs 9–12. With the exception of NT12, all MRSH SCCmec NTs harbored class C mec and various combinations of ccr genes (Table 1). According to microarray analysis, the NT12 isolate carried class B mec and ccrAB2 indicative of SCCmec IV, as well as ccrA1 and was the only SH isolate that harbored the ACME-arc genes. No SCCmec IV subtype was identified by PCR. Multiplex SCCmec typing PCR and sequencing revealed that this isolate harbored ccrAB4, with 100% DNA sequence identity to ccrAB4 in S. haemolyticus (GenBank accession no. AB587081.1) rather than ccrAB2 [58]. This SH ccrAB4 allele exhibited 91% and 87% DNA sequence identity to ccrAB4 and ccrAB2 in S. aureus, respectively, which the array ccr primers and probes are based on. The ambiguity in the identification of the ccrAB alleles in this SH isolate using the DNA microarray may be linked to this.

Possible novel SCCmec V and V_T subtypes were detected in two additional MRSH-CpA isolates which carried the kdp and pls genes in addition to the class C mec and ccrAA and ccrC genes (Table 1).

Three mecA alleles were identified among both the MRSH-Hu and-CpA isolates investigated, which shared >99.9% DNA sequence identity and differed by a maximum of two nucleotide bases indicating their close similarity. The mecA alleles AY786579 and GQ92039 were detected in 10/20 (50%) and 7/20 (35.0%) MRSH, respectively, both being detected in MRSH-Hu and–CpA isolates (Table 1).

ST1-MRSH-NT9 was the only common ST and SCCmec type combination detected among both MRSH-Hu (n = 4) and-CpA (n = 1). Two mecA alleles and two dts were detected amongst these five isolates, with only one mecA allele common to Hu and CpA isolates (Table 1). However, the mecA alleles (ABSA01000066 & AY786579; one nucleotide difference) and dts (dt11v & dt11ca; MST value of 2 i.e. 97–98% similarity, S1B Fig) were closely related. All of the other ST and SCCmec type combinations were unique to either MRSH-Hu or MRSH-CpA. The SCCmec types V or V_T were detected in MRSH-CpA (n = 6) and MRSH-Hu (n = 1), but these isolates were assigned to CC1 and as a singleton, respectively, and were assigned to four distinct dts (Table 1 and S1 Fig). However, all SCCmec types V or V_T isolates harbored the GQ902038 mecA allele.

Only one MSSH isolate was identified (Hu). This isolate was identified as ST1 and harbored a NT SCC element, consisting of ccrAA and ccrA4 (SCC3, Table 1).

Antimicrobial susceptibility

The antimicrobial resistance phenotypes and genes detected among the isolates investigated are shown in Table 2 and S1 Table. Multidrug resistance (MDR) was detected among 96.7% (59/61) of isolates and resistance to almost all the classes of antimicrobial agents investigated was detected among both the SE and SH isolates.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Prevalence of antimicrobial resistance genes and phenotypic resistance to antimicrobial agents among S. epidermidis and S. haemolyticus isolates from humans and companion animals^a.

https://doi.org/10.1371/journal.pone.0138079.t002

However, differences were identified in the prevalence of phenotypic antimicrobial resistance and resistance genes detected among Hu and CpA isolates and SE and SH isolates. These differences are described below in more detail.

Comparison of antimicrobial resistance genes amongst SE and SH isolates

Genes encoding resistance to fusidic acid and trimethoprim, primarily encoded by fusB and dfrS1, respectively, were significantly (p < 0.05) more prevalent among SE than SH isolates (Table 2). In contrast, resistance to macrolides encoded by msr(A) and mph(C), and aminoglycosides, encoded by aacA-aphD and aphA3 were significantly (p < 0.05) more common among the SH isolates (Table 2). The trimethoprim resistance genes dfrG and dfrK were significantly more prevalent (p = 0.0001) amongst SH than SE isolates (Table 2). Although tetracycline resistance encoded by tet(K) was detected among both SH and SE isolates, tet(M) was only detected among the SH isolates (n = 3).

Comparison of antimicrobial resistance genes amongst isolates from Hu and CpA hosts

Resistance to aminoglycosides encoded by aadD and aphA3, tetracycline encoded by tet(K) and fusidic acid encoded by fusB, were significantly (p < 0.05) more common among the CpA than Hu isolates (Table 2). Resistance to ciprofloxacin and rifampicin was also significantly more common in the CpA isolates (p < 0.05). Resistance to chloramphenicol was detected in CpA isolates only (p = 0.001) where it was encoded by fexA in SE-CpA isolates and cat-pC221 among SH-CpA isolates (Table 2).