Molecular characterization of methicillin-resistant Staphylococcus aureus genotype ST764-SCCmec type II in Thailand

Methicillin-resistant Staphylococcus aureus (MRSA) is a significant causative agent of hospital-acquired infections. We characterized MRSA isolated from August 2012 to July 2015 from Thammasat University Hospital. Genotypic characterization of MRSA SCCmec type II and III isolates were scrutinized by whole genome sequencing (WGS). The WGS data revealed that the MRSA SCCmec type II isolates belonged to ST764 previously reported mainly in Japan. All of tested isolates contained ACME Type II′, SaPIn2, SaPIn3, seb, interrupted SA1320, and had a virulence gene profile similar to Japan MRSA ST764. Rigorous surveillance of MRSA strains is imperative in Thailand to arrest its potential spread.


Results
Five isolates of HA-MRSA SCCmec type II [SATU130 (SRA: SRR16971216), SATU131 (SRA: SRR16971215), SATU132 (SRA: SRR16971214), SATU134 (SRA: SRR16971213) and SATU136 (SRA: SRR16971212)] and two isolates of HA-MRSA SCCmec type III [SATU133 (SRA: SRR17033804) and SATU135 (SRA: SRR17033804) were whole-genome sequenced. All isolates were from patients with invasive diseases, including pneumonia, septicemia, osteomyelitis and endocarditis as described in Table 1. The majority of the patients were over 60 years old. All five SCCmec type II isolates belonged to the ST764 genotype while the SCCmec type III isolates belonged to ST22 and ST239. None of the patients was HIV seropositive, but four patients with ST764 infection had diabetes with chronic kidney diseases. Only one patient with MRSA ST764 infection survived.
All isolates were resistant to cefpirome and clindamycin yet were susceptible to vancomycin, trimethoprimsulfamethoxazole, linezolid, mupirocin and teicoplanin. However, MICs of vancomycin against most ST764 isolates were relatively high at 1.5-2.0 μg/ml. The MICs of the drugs against each isolate were shown in Table 2.
The isolates were characterized genetically by SCCmec typing, spa typing, MLST, and WGS with the results shown in Supplementary Table S1. Isolates belonging to ST239 and ST22 harbored SCCmec type III while all ST764 isolates harbored SCCmec type II, similar to previous reports 14,19,20 . Characterization of single nucleotide variants (SNVs). More than 90% of reads were uniquely mapped to S. aureus N315 (ASM964v1, GenBank accession: BA000018.3), an ST5 isolate. The mean coverage of the sequence was 93% with 72X average depth. A total of 50,021 SNVs were identified based on alignment to the reference genome of S. aureus N315. The number of SNVs identified among ST764 isolates ranged from 237 to 261. The average pairwise SNV distance among ST764 isolates was 49.8.
The maximum likelihood phylogenetic tree of five ST764 isolates, with N315 as the outgroup, was shown in Fig. 1. There were two pairs, SATU130-SATU131 and SATU132-SATU136, with the same SNV distances of 11 (Table 3), suggesting epidemiological linkages between each pair 21 . SATU130 and SATU131 were isolated from the same male surgical ward, 26 days apart, and therefore probably belonged to the same clones. Isolates SATU132 www.nature.com/scientificreports/ and SATU136 were from male and female medicine wards, 9 months apart. Moreover, isolate SATU136 did not contain seb, while SATU132 did. The epidemiological link between the two isolates was therefore uncertain. Al l f ive ST764 is olates share d t hre e delet ions (g.ASM964v1:823521_823580del, g.ASM964v1:2068598_2068612del, g.ASM964v1:2793533_2793544del) and four insertions (g.ASM964v1:8342_8343ins [unknown], g.ASM964v1:941411_941412ins [unknown], g.ASM964v1:1104595_1104596ins [unknown], g.ASM964v1:1527689_1527690ins [unknown]), which were not present in the ST22 and ST239 isolates. There were some indels found in some but not all ST764 isolates. A deletion (g.ASM964v1:1608977_1609107del) was shared by SATU132 and SATU136 while a deletion (g.ASM964v1:383663_383683del) and an insertion (g.ASM964v1:676384_676385ins [unknown]) could be found only among isolates SATU130, SATU131, and SATU134, as shown in Fig. 1 and Supplementary Tables S2 and S3.
Characterization of SCCmec-ACME II′ composite island of ST764. We investigated the presence of various genes and genetic elements previously described in ACME-SCCmec composite island (ACME-SCCmec CI) 12 . All ST764 isolates here contained arcA-D, but not opp3; thus the ACME element was classified as ACME II′. They did not harbor any SNV in arcA-D as compared with N315, which was similar to those Japan ST764 isolates. It should be noted that C4913T mutation in arcB was found in an ST764 isolate in Japan 22 . ACME-SCCmec CI was previously classified into five types. The five ST764 isolates harbored cJR1 DNA segment but carried neither pUB110 nor ccrC. Thus, these isolates belonged to type A3 23 . The ACME-SCCmec CI type A3 is the shortest form of the complex and probably puts less burden on the fitness cost for the bacteria.
The presence of genetic elements specific to ST764. In order to further evaluate similarity between the ST764 isolates in this study and the isolates reported from Japan, we identified other genetic elements previously described for ST764 in Japan apart from the ACME-SCCmec CI.
The five ST764 isolates contained SaPIn2 and SaPIn3 pathogenic islands. The former exotoxin island, SaPIn2, contained set6 to set15. The latter enterotoxin island contained seg, sei, sem, sen, and seo. These isolates did not have SAPIn1, evidenced by the absence of tst, sea, sec, sel or sek. Our ST764 isolates did not contain sed, see, seq, Figure 1. An ML phylogenetic tree of five ST764 isolates with N315 (ST5) as the outgroup. The deletions and insertions were labeled alongside with the branches shared by all isolates that shared the deletions/insertions as indicated with red (common to all isolates), blue and green. The HGVS nomenclature (available from http:// varno men. hgvs. org/) were used to describe the deletion and insertion events. The "unknown" indicated in the nomenclature means unknown inserted sequences. www.nature.com/scientificreports/ or sep either. They did not contain exfoliative toxin genes (eta and etb). These ST764 isolates were characterized as having cna with pvl-negative but also having lukE-lukD. These isolates contained hla, hld and hlgA-C. Such genetic characteristics matched the genetic profile of the ST764 isolates in Japan. The Thai isolates contained adhesin genes, icaA, sdrC-E, ebpS, clfA, clfB,-bbp, fnbA and fnbB (partial sequence). The ST764 isolates in Japan were described as containing three more insertions, SaPInn54-13.5 kb fragment containing seb, prophage carrying SA1320 gene, and tet(M)-carrying Tn5251. All except one ST764 here harbored seb, which was reported missing in some ST764 isolates in Japan 22 . All ST764 isolates here harbored the insertion in SA1320, tet(M) and Tn5251. The similarity between genetic profiles of the isolates studied here and the MRSA ST764 SCCmec type II in Japan suggests that the isolates in Thailand were from the outbreak clone in Japan.

Discussion
HA-MRSA is a global problem, especially in large tertiary hospitals. Previous WGS studies in Thailand focused mainly on ST239 SCCmec type III, which was the most common MLST type [24][25][26] . It is, therefore, intriguing to find that SCCmec type II MRSA was common in TUH, unlike in other tertiary hospitals in Thailand 10,11 . Moreover, all belonged to ST764-MRSA-II and were from patients with severe hospital-acquired invasive diseases. The apparently high incidence remains to be confirmed.
ST764-MRSA-II, is a recently recognized genotype, reported as a common MRSA genotype in many settings in Japan and might be spreading 16 . Such hospital settings encompass outpatient departments and healthcare professionals 27 . This spreading may be attributed to the presence of ACME type II, which may provide the bacteria a better ability to colonize on skin and mucous membrane which could consequently lead to better transmission. It has recently been reported as a common genotype in Shanghai 28 and Eastern China 29 but not yet in Southern China 30 .
ST764 is a hybrid variant of the globally disseminated ST5 lineage of HA-MRSA, carrying a few virulence traits similar to community-acquired MRSA (CA-MRSA). ST764 has evolved via multiple steps, including the acquisition of ACME arcA and the staphylococcal enterotoxin B gene (seb) in SaPInn54 12 .
Here we reported, for the first time, the WGS of the five MRSA SCCmec type II isolates identified from 2013 to 2015 and demonstrated that all ST764 isolates in this study carried ACME-SCCmec CI type A3 similar to some ST764 isolates in Japan 23 . The ST764 strains from Thailand and those from Japan also shared other genetic characteristics, such as the lack of pvl, tst, sec, sel and sep with the presence of SaPIn2, SaPIn3, seb, tet(M) 12 . This implies that the ST764-MRSA-II clone in Thailand was similar to and probably originated from the clone in Japan.
The genetic similarity of ST764-MRSA-II in Thailand to ST764-MRSA-II in Japan suggested a possibility that the ST764-MRSA-II strain was brought to Thailand by a healthy carrier, and later was transmitted into the hospital settings. There is also a possibility that the ST764-MRSA-II strain may continue to spread wider via colonization of healthy carriers.
Analysis of whole genome sequencing data suggested potential transmission between a pair of isolates. The isolates had a SNP-distance of 11 and were isolated from the same ward within a month. This finding of MRSA genotype ST764-SCCmec type II is the first HA occurrence reported in Southeast Asia. Two isolates of CA ST764-SCCmec type II from Thai pig farm workers were also reported 31 . The potential danger of spreading necessitates rigorous surveillance of emerging MRSA strains.

Methods
Bacterial strains. One-hundred and one clinical isolates of MRSA, confirmed by the detection of mecA, at TUH from August 2012 to July 2015 9 , were characterized for SCCmec types, pulsed-field gel electrophoresis (PFGE) patterns, virulence genes patterns, and antibiotic resistance phenotypes. There were 41 samples belonging to SCCmec type II that revealed either PFGE pattern A or P. Five isolates with both A-and P-PFGE patterns were randomly selected for sequencing, together with two isolates belonging to SCCmec type III. Bacterial collection, demographic data, clinical and microbiological data retrieval from Microbiology laboratory, Thammasat University Hospital (TUH) were carried out in accordance with relevant guidelines and regulations after approval of the Human Research Ethics Committee of Thammasat University (approval no. MTU-EC-DS-1-193/63 and MTU-EC-DS-4-197/63) and TUH.
Microbiological and clinical data. Demographic data, clinical and microbiological data were retrieved from MRSA-infected patients' record files from the TUH database. Examples of patients' clinical data include lengths of stay, clinical outcomes and underlying diseases including diabetes mellitus (DM) and chronic kidney diseases (CKD). Successful treatment outcomes referred to events when the patients were cured or improved and MRSA was eradicated. Outcomes of persistent-MRSA, recurrent-MRSA or death due to MRSA were regarded as failure of treatments. The patients who were diagnosed with MRSA infection at an outpatient clinic or within 48 h of hospitalization without history of exposure to any healthcare facilities during the previous two months were defined as community-acquired MRSA infection (CA-MRSA). Hospital-acquired MRSA infection (HA-MRSA) referred to those patients who became infected after 48 h of hospitalization.

Antimicrobial susceptibility testing. Minimum inhibitory concentration (MIC) of vancomycin (VAN)
was carried out by both broth microdilution method 32,33 and E-test for determining the high end MICs. Other antibiotics including cefpirome (CPR), clindamycin (CLI), trimethoprim (TMP), trimethoprim-sulfamethoxazole (SXT), rifampicin (RIF), mupirocin (MUP), fusidic acid (FUS), teicoplanin (TEC) and linezolid (LZD) were tested by E-test (Liofilchem ® , Italy).  -TCC AGA TTA CAA CTT CAC CAG G-3′ and 5′-CCA CTT CAT ATC TTG TAA CG-3′) 34 and pvl (5′-ATC ATT AGG TAA AAT GTC TGG ACA TGA TCC A-3′ and 5′-GCA TCA AST GTA TTG GAT  AGC AAA AGC-3′) 35  Whole-genome sequencing. DNA purity was assessed using a spectrophotometer, NanoDrop™ 2000 (Thermo Fisher, USA) and DNA concentration was determined by measuring fluorescence in a fluorometer, Qubit™ 3 Fluorometer (Invitrogen™, USA). DNA integrity was visualized by running the isolated samples in agarose gel electrophoresis to assess extent of DNA shearing. A DNA input of 1 ng was used to prepare a 150-base read library, using the Nextera XT DNA Library Preparation Kit (Illumina, USA). The library preparation protocol includes tagmentation of genomic DNA, library amplification, library clean up, library normalization and library pooling. Library denaturation and dilution were performed, and then diluted PhiX control V3 was spiked into the libraries to obtain 1% PhiX as an external control. The prepared library pool was loaded into the reagent cartridge and then sequenced by 150 bp pairedend NextSeq 500 sequencing (Illumina, USA). Sequences of the five isolates of HA-MRSA SCCmec type II were submitted to SRA. The accession numbers of each isolate were listed as follows.  39 . Low quality variants that have read depth lower than 10 were discarded. Variant annotation was performed using SnpEff version 4.3 40 . Structural variants (SVs) were identified by running the program Manta 41 on an input BAM file (sorted and mapped reads) that has aligned to the reference sequence. Resulting SV events flagged with "IMPRECISE" were discarded. The remaining SV events were confirmed by visualizing them using integrative genomics viewer (IGV) 42 . De novo Genome assembly was done by SPAdes3.9 (available from http:// cab. spbu. ru/ softw are/ spades/) ( Supplementary Fig. S1).

Phylogenetic tree construction.
To prepare an input sequence for building a phylogenetic tree, SNVs present on Mobile Genetic Element (MGE) were discarded and only homozygous and biallelic SNVs were concatenated into the SNV input sequence (in-house script). Multiple sequence alignment of five ST764 isolates and N315, representing an outgroup, were tested for recombination signals by using Recombination Detection Program v4.10 (RDP4) 46 that includes seven well-known recombination checking routines, namely RDP, GENECONV, Chimera, MaxChi, Bootscan, SiScan, and 3Seq. If more than four utilities reported recombination regions, these regions would be removed. These RDP-corrected sequences were used as the input for phylogenetic tree construction by the maximum likelihood (ML) method. PhyML version 3.0 was used to construct an ML-based phylogeny which can be visualized by the SeaView program 47,48  www.nature.com/scientificreports/ more than four utilities reported recombination regions, these regions would be removed. These RDP-corrected sequences were used as the input for phylogenetic tree construction by the maximum likelihood (ML) method. PhyML version 3.0 was used to construct an ML-based phylogeny which can be visualized by the SeaView program 49 . ACME/SCCmec type II. The structure of ACME/SCCmec type II was determined by identification of arcA, arcB, arcC, arcD, opp3 and ccrC using the sequence of MRSA USA300, (GenBank Accession: CP000255.1) as reference. Identification of pUB110 and cJR1 was done using N315 and NCTC10442 (SCCmec type I, GenBank Accession: UHCH01000003.1) as reference, respectively.