A total of 586 S. aureus blood isolates collected between May 2016 and April 2017 from six general hospitals in South Korea (Lee et al., 2018) were screened. Among the 319 cefoxitin-resistant isolates, E16SA093 and F17SA003, whose SCCmec elements could not be typified, were selected for further study.

Antimicrobial susceptibility to antimicrobials used for staphylococci infection was determined by disk diffusion tests on Mueller-Hinton agar (Difco Laboratories, Detroit, MI, USA), following the CLSI guidelines (CLSI, 2018). S. aureus ATCC 25923 was simultaneously tested in each batch for quality control. MRSA isolates were subjected to polymerase chain reaction (PCR) for mecA gene and SCCmec typing, as previously described (Oliveira and de Lencastre, 2002).

Multilocus sequence typing (MLST) was carried out by PCR and sequencing of the seven housekeeping genes, arcC, aroE, glpF, gmk, pta, tpi, and yqiL. Allelic numbers and sequence types (STs) were determined by comparing the obtained sequences to the database for S. aureus¹. The agr type was determined by multiplex PCR (Gilot et al., 2002), and spa typing was conducted by comparing the PCR-amplified nucleotide sequence of the variable repeat region of the spa gene against the Ridom SpaServer².

Bacterial whole genomes were sequenced with single-molecule real-time (SMRT) sequencing on a PacBio RSII instrument (Pacific Biosciences, Menlo Park, CA, USA) using genomic DNA from the S. aureus isolates extracted by a Wizard Genomic DNA Purification kit (Promega, Madison, WI, USA). SMRTbell template libraries were prepared, and adapter ligation was performed. Acquired sequencing data were de novo assembled by PacBio SMRT, read with the PacBio SMRT analysis software suite (version 2.3.0). Coding sequences (CDS), including tRNA and rRNA, were annotated using the NCBI Prokaryotic Genome Annotation Pipeline³. Nucleic acid sequences were compared using Basic Local Alignment Search Tool,⁴ and resistance and virulence determinants were searched for using ResFinder⁵ and VirulenceFinder⁶, respectively. Prophages were searched for using the PHAge Search Tool Enhanced Release⁷. For any putative ccr gene, the site-specific serine recombinase motif (Wang and Archer, 2010) and a modified motif by the consensus pattern (Perreten et al., 2013) were searched for against the coding sequences of both genomes.

The nucleotide sequences of the entire genomes of S. aureus E16SA093 and F17SA003 were deposited in GenBank under accession numbers CP031130 and CP031131 for F17SA003 and E16SA093, respectively.

Following the one-year collection of the 586 S. aureus blood isolates, a total of 319 isolates (54.4%) were MRSA conferred by the mecA gene. A total of 176 (30.0%) isolates belonged to ST72; 112 of those isolates (63.6%) were MRSA, 65 were healthcare-associated (HA) MRSA, and 47 were community-associated MRSA (CA-MRSA). All but three MRSA ST72 isolates (97.3%, 109/112) carried SCCmec type IV, while one possessed SCCmec type II and the remaining two isolates (E16SA093 and F17SA003) had non-typeable elements.

The MRSA ST72 harboring SCCmec IV (ST72-MRSA-IV) was one of the top three CA-MRSA clones in the USA until 2002 as a pulse-field type USA700; however, the clone was suddenly eliminated from the country in 2004 (Tenover et al., 2008). In South Korea, the ST72-MRSA-IV acceded a major CA-MRSA clone by 2005 (Kim et al., 2007), and the ST72-MRSA-IV subsequently grew to be a major HA-MRSA clone. This finding supports the idea that the ST72-MRSA-IV clone was disseminating from communities to hospitals (Song et al., 2011).

The CA-MRSA E16SA093 was recovered in September 2016 from an 86-year-old female patient with acute infective endocarditis and infective spondylopathy. The patient was transferred from an acute care hospital to a general hospital located in Gwangju city, and blood cultures were carried straightaway. The bacteremia originated from a bone infection, and empirical treatment was started with cefazoline. Definitive treatment was followed with teicoplanin within 72 h after the initial blood culture, and the patient was cured. The HA-MRSA 17SA003 was recovered in January 2017 from a 63-year-old male patient with diabetes mellitus and stage 4 chronic kidney disease hospitalized in a general hospital in Busan city. An initial blood culture was performed on the 13th day of hospitalization, and the origin of S. aureus bacteremia was unidentified. Empirical treatment with cefazoline was replaced to vancomycin within 72 h, and the patient was cured.

Both MRSA isolates belonged to ST72 and agr-type I. They were negative for both Panton-Valentine leukocidin and toxic shock syndrome toxin (Table 1). The spa-types of E16SA093 and F17SA003 were t324 and t2460, respectively. Among the 10 antimicrobials tested, the E16SA093 isolate was resistant only to cefoxitin, while F17SA003 was resistant not only to cefoxitin but also to erythromycin and clindamycin.

The de novo assembly of the genome resulted in a 2,767,631,390-bp circularized chromosome, including 2,564 assigned CDSs, 60 tRNAs, and 19 rRNAs for E16SA093, and a 2,849,947,596-bp circularized chromosome, including 2,546 CDSs, 60 tRNAs, and 19 rRNAs for F17SA003. The overall GC contents were 32.9% for both. No plasmid was identified. Acquired genetic elements by both chromosomes were alike, including two intact Staphylococcus prophages (44.1-kb ϕNM3 and 41.2-kb Sap26), 17 virulence factors, and three antimicrobial resistance genes, with an extra copy of blaZ for F17SA003. No known heavy metal resistance genes were identified for either.

For the SCCmec element, a class B mec gene complex lacking the ΨIS1272 upstream from the mecA gene was identified, and neither the ccr gene nor any putative site-specific serine recombinase gene was identified elsewhere in the chromosome (Figure 1). The ΨSCCmec, designated as ΨSCCmec_E16SA093, resembled a SCCmec type IV, which is common in MRSA ST72. When compared with the genome of HL1 that is a CA ST72-MRSA-IV recovered in South Korea before 2010 (Chen et al., 2013), a 12.6-kb region was deleted from the SCCmec type IV element, and 7.0- and 31.9-kb chromosomal DNA region was deleted in the E16SA093 and F17SA003, respectively. The Ccr recombinase involves the site-specific integration/excision of SCCmec elements (International Working Group on the Classification of Staphylococcal Cassette Chromosome, 2009), and the CcrA2/CcrB2 in the SCCmec IV is targeting the attB site at the orfX gene (Wang and Archer, 2010). The ΨSCCmec_E16SA093 was indeed integrated exactly at attB, suggesting the subsequent elimination of the ccrA2/ccrB2 genes from the SCCmec IV element after the integration event. As the ΨIS1272 was absent, IS-associated recombination was suspected. However, no palindromic sequences were observed in either end of the deleted 19.7- and 44.5-kb DNA fragments targetable by IS1272, with an insertion sequence involved in the structure-dependent transposition or stem-loop replacement (Wan et al., 2017).

The fitness benefit of the ccr-gene-loss from SCCmec, resulting in an inherent mecA in the chromosome, has never been assessed, while spontaneous mecA-gene-loss in the absence of selective pressure, driven by the huge biological cost of gene expression, has been demonstrated (Noto et al., 2008). The furnished mecA gene could provide advantages to MRSA in the beta-lactam-abundant habitat, such as the clinical settings, suggesting a course of adaptive evolution for MRSA. While ΨSCCmec is occasionally identified in MRSA (Chen et al., 2010), methicillin-resistant coagulase-negative staphylococci (MRCNS) carrying the element is much more common (Perreten et al., 2013; Shore and Coleman, 2013). The speculation of MRCNS to be a reservoir of SCCmec (Archer et al., 1994), in the MRSA is inspiring.

In this study, we evaluated MRSA ST72 isolates carrying ΨSCCmec_E16SA093, which was likely being fabricated from the SCCmec type IV. Excised portions of the chromosomes differed in size, and the event likely occurred independently, indicating that the clonal dissemination of ST72-MRSA-ΨSCCmec_E16SA093 has not yet been occurred. The immobile mecA gene could make the MRSA fit the antimicrobial-abundant habitat, even though the mecA gene expression is known to be costly. Further study of the molecular mechanisms driving ccr gene loss is needed, and dissemination of the clone should be surveilled.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.