Whole Genome Sequence and Comparative Genomics Analysis of Multi-drug Resistant Environmental Staphylococcus epidermidis ST59

Staphylococcus epidermidis is a major opportunistic pathogen primarily recovered from device-associated healthcare associated infections (DA-HAIs). Although S. epidermidis and other coagulase-negative staphylococci (CoNS) are less virulent than Staphylococcus aureus, these bacteria are an important reservoir of antimicrobial resistance genes and resistance-associated mobile genetic elements that can be transferred between staphylococcal species. We report a whole genome sequence of a multidrug resistant S. epidermidis (strain G6_2) representing multilocus sequence type (ST) 59 and isolated from an environmental sampling of a hotel room in London, UK. The genome of S. epidermidis G6_2 comprises of a 2408357 bp chromosome and six plasmids, with an average G+C content of 32%. The strain displayed a multi-drug resistance phenotype which was associated with carriage of 7 antibiotic resistance genes (blaZ, mecA, msrA, mphC, fosB, aacA-aphD, tetK) as well as resistance-conferring mutations in fusA and ileS. Antibiotic resistance genes were located on plasmids and chromosome. Comparative genomic analysis revealed that antibiotic resistance gene composition found in G6_2 was partly preserved across the ST59 lineage.


Staphylococcus
epidermidis antibiotic resistance whole genome sequence comparative analysis Staphylococcus epidermidis is a common human skin commensal, but also the most frequent pathogen among coagulase-negative staphylococci (CoNS), causing primarily device-associated healthcare associated infections (DA-HAIs). Compared with more virulent S. aureus, CoNS rarely produce toxins and less is known on whether the toxin genes contribute to strain virulence (Otto 2013a). S. epidermidis forms biofilms on medical devices and implants, from which single cells dissociate and disseminate via the bloodstream to start colonization at a different site, which might lead to sepsis, meningitis and endocar-ditis (Becker et al. 2014). In addition, S. epidermidis and other CoNS are believed to act as a reservoir of resistance and virulence genes for S. aureus, contributing to the evolution and emergence of successful clones of methicillin-resistant S. aureus (MRSA) (Otto 2013b). Together with S. aureus and other CoNS, S. epidermidis accounts for 30% of hospital associated infections (Conlan et al. 2012). These nosocomial pathogens have developed an arsenal of strategies contributing to colonization and infection of the hosts (Becker et al. 2014), while often being resistant to multiple antibiotics. Emergence of antibiotic resistant bacteria has been mostly attributed to the healthcare-associated settings (Oliveira and Tomasz 2002). However, more recently, selection of antibiotic resistance has been also associated with the community which has been linked to the misuse of antibiotics (DeLeo et al. 2010). A typical example of this is the community-acquired MRSA (CA-MRSA) which, in addition to acquiring methicillin resistance, has gradually increased the frequency of resistance determinants similarly to hospital-acquired MRSA (HA-MRSA) (Chambers 2005). There is an increasing evidence that horizontal gene transfer between closely related species may contribute to this (Otto 2013a).
Recently, Méric et al. showed that S. aureus and S. epidermidis share half of the genome and while homologous recombination between the two species was rare, there was an evidence of extensive MGE sharing, in particular SCCmec, metal resistance and SaPIn1 elements (Méric et al. 2015). As a result, attention is now focusing on the multidrugresistant coagulase-negative staphylococci and their rapid spread as opportunistic pathogens particularly in relation to patients with an immuno-compromised status (Morfin-Otero et al. 2012). Multidrugresistant coagulase-negative staphylococci (MDR-CoNS) are primarily recovered from healthcare-associated medical devices, ambulatory patients and healthy animals (Becker et al. 2014).
Molecular approaches such as pulse field gel electrophoresis and multi-locus sequence typing have been widely used to evaluate the dissemination of resistant clones of bacteria (Miragaia et al. 2008). Recently, complete genome sequencing of S. epidermidis strains have been reported, however these are limited to commensal and nosocomial strains (Conlan et al. 2012;Gill et al. 2005;Zhang et al. 2003). Only one study has compared whole genome sequences of four S. epidermidis isolated from rice seeds with that of type strain (Chaudhry and Patil 2016). To our knowledge this is the first whole genome based study looking at MDR-CoNS isolated from general public settings.
In this study, we present the genetic features of this multidrug resistant S. epidermidis (strain G6_2) and compare it with six S. epidermidis reference genomes and 133 previously published genomes of clinical S. epidermidis.

Isolates analyzed in this study
Between October 2012 and April 2013, we sampled different sites in three hotels in London, UK. Permission to carry out sampling was granted by the manager/owner of each hotel and the results from each hotel were reported to each manager/owner for their information. Inanimate objects in 32 hotel rooms were sampled using COPAN dry swabs (Copan Diagnostics Inc., USA). All specimens were inoculated onto Nutrient Agar (Oxoid, Basingstoke, UK) and Mannitol Salt Agar plates (Oxoid Basingstoke, UK). These cultures were incubated aerobically at 37°for 24-72 h.
The S. epidermidis G6_2 was recovered from one of the hotel rooms in April 2013 in London, UK. Preliminary identification was achieved by using Matrix-assisted laser desorption ionization time-flight massspectroscopy (Microflex LT, MALDI-TOF-MS, Bruker Daltonics, Coventry, UK) as described previously (Mkrtchyan et al. 2013).
16S rRNA gene sequencing Genomic DNA of S. epidermidis G6_2 was prepared using a Qiagen DNA extraction kit (Qiagen, Crawley, UK). 16S rRNA amplification was performed as described previously (Okazaki et al. 2009), PCR products were sequenced by Eurofins MWG GmBH (Ebersberg, Germany) using ABI 3730 L DNA analyzer.

Molecular characterization of S. epidermidis G6_2
Carriage of the mecA gene was determined with PCR as described previously (Hanssen et al. 2004). SCCmec typing was carried out by determination of mec and ccr complexes (Kondo et al. 2007). Multi locus sequence tying (MLST) has been used to determine seven housekeeping genes as describe previously (Thomas et al. 2006). Sequence types were determined using MLST V1.8 software (https://cge.cbs.dtu.dk/ services/MLST/).
Whole genome sequencing, assembly and comparative genomics Genomic DNA was extracted using the MasterPure Gram Positive DNA Purification Kit (Cambio, Dry Drayton, UK) from overnight cultures grown from single colonies in 5 ml of tryptic soy broth overnight at 37°. Illumina library preparation was carried out as described previously (Quail et al. 2008), and genome sequencing using Hi-Sequation 2000 performed following the manufacturer's standard protocols (Illumina, Little Chesterfield, UK). The raw fastq data were quality trimmed using trimmomatic, (version 0.35) default settings, specifying a phred cutoff of Q20. Read quality was assessed using FastQC  and Kraken (version 0.10.5beta) metagenomic pipeline (Wood and Salzberg 2014), including KronaTools (version 2.5) (Ondov et al. 2011) was used to assess library purity, that is, it was not a mixed sample and ensure the species was S. epidermidis. De novo assemblies were performed using assembler, SPAdes (version 3.5.0) (Bankevich et al. 2012), default PE settings, from which only contigs greater than 500 bp in length were taken for further analysis. Using the program, Andi (version 0.9.4-beta) (Haubold et al. 2015) the de novo assembled G6_2 genome along with 108 assembled Staphylococci genomes were aligned, clustered and visualized using PHYLIP (http://evolution. genetics.washington.edu/phylip.html) and FigTree (http://tree.bio.ed.ac. uk/software/figtree/). Annotations were performed using the pipeline Prokka (version 1.11) (Seemann 2014). The resultant annotated genome was used for all subsequent comparative genomic studies. Carriage of antimicrobial resistance and virulence genes was assessed using the SRST 2 software (Inouye et al. 2014) and the ARG-ANNOT (Gupta et al. 2014) and VF-DB databases (Chen et al. 2016). Pangenome analysis was performed using the Roary pipeline (version 3.4.2) . To reconstruct phylogenetic tree, short reads were mapped against the S. epidermidis ATCC12228 reference genome (Zhang et al. 2003), using SMALT version 0.5.8 (http://www. sanger.ac.uk/science/tools/ smalt-0). A core genome alignment was created after excluding MGE regions, variable sites associated with recombination (detected with Gubbins (Croucher et al. 2015) and sites with more than 5% proportion of gaps (i.e., sites with an ambiguous base). A maximum likelihood (ML) phylogenetic tree was generated with RAxML v8.2.8 (Stamatakis 2014) based on generalized time reversible (GTR) model with GAMMA method of correction for among site rate variation and 100 bootstrap (BS) replications. The phylogenetic tree was annotated using Evolview (Zhang et al. 2012).

Nucleotide sequence accession numbers
Reads for S. epidermidis G6_2 were submitted to the European Bioinformatics Institute Sequence Read Archive, accession ERR387168.

Data availability
The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article and its tables and figures. Supplemental material available at Figshare: https:// doi.org/10.25387/g3.6133946.

RESULTS AND DISCUSSION
S. epidermidis has become a leading hospital-associated pathogen due to the increased use of medical devices (Vuong et al. 2004). Treatment of S. epidermidis infections is challenging as the bacteria are commonly resistant to methicillin and might also display multi-drug resistance phenotype, which presents a serious public health challenge (Xu et al. 2015) . S. epidermidis, represents an important reservoir of mobilizable genes that can be horizontally transferred between staphylococci species, which has likely contributed to the development of antibiotic resistance in S. aureus (Otto 2013a).
S. epidermidis G6_2 was isolated from a hotel room in London, UK in 2013, and the species were determined by MALDI-TOF MS and 16S rRNA sequencing. Initial molecular analysis revealed that the S. epidermidis G6_2 strain was mecA positive, carrying SCCmec type IV, and represented ST59.
A draft genome was assembled, comprising of 53 contigs (48 .= 1kb) for the isolated S. epidermidis G6_2 genome (Table S1; Table S2 and Figure S1). The assembly comprised of one chromosome (2408357 bp in length) and six plasmids, annotated as pG6_2_1 to pG6_2_6 (the largest, pG6_2_1, is 10570 and the smallest, pG6_2_6, is 3426 bp in length), with an average G+C content of 32.02%. It has a total (chromosome and plasmids) of 2213 predicted protein coding sequences, of which 21.5% were annotated as hypothetical proteins and 14.3% were annotated as putative functions (Table 1).
Phylogenetic relationship with other S. epidermidis isolates A previously described collection of 129 whole genome-sequenced S. epidermidis isolates together with 6 reference strains was used to determine the phylogenetic relationship between the G6_2 strain and other S. epidermidis lineages. After removal of variable sequence regions corresponsing to mobile genetic elements (MGE), recombination blocks as well as sites with more than 5% proportion of gaps, the core genome alignment contained 4262 SNP sites. Seven ST59 isolates clustered and formed a distinct clade with S. epidermidis G6_2 (Figure 1).
Genotypic and phenotypic characterization of antibiotic resistance S. epidermidis G6_2, revealed 9 antibiotic resistance determinates across the chromosome and plasmids (Table 2). This included aminoglycoside resistance gene aac(6')aph(2"), beta-lactam resistance genes mecA and blaZ, fosfomycin resistance gene fosB, macrolide resistance genes mphH and msrA (the latter also conferring resistance to lincosamide and streptogramin B) and tetracycline resistance gene tet(K). This correlated with the results of antibiotic susceptibility testing as the strain was found resistant to 11 out of 13 antibiotics tested, demonstrating susceptibility to vancomycin and chloramphenicol only. Resistance to mupirocin and fusidic acid was associated with point mutations in chromosomally located genes, ileS and fusA, respectively. In addition to antimicrobial resistance genes, the G6_2 strain also carried plasmid-associated qacC gene, which encodes the multidrug resistance efflux protein and mediates resistance to biocides, and a chromosomally-inserted copper resistance operon composed of copZ-copA-csoR genes together with an additional copy of cobalt-zinc-cadium efflux pump gene czcD. The latter was distinct from the conserved chromosomal copy of czcD gene, and was previously identified on a number of CoNS plasmids.
The G6_2 strain carried a 47-kb composite island composed of the SCCmec IV and a SCC element that contained plasmin-sensitive surface protein gene pls, spermidine N-acetyltransferase gene speG and a coppertranslocating ATPase gene copA. The full sequence of this composite island was unique and did not match previously described reference genomes. However, the SCCmec IV sequence shared 99% identity with SCCmec IVa from various MRSA strains including the MRSA M1 isolated in Denmark (Larner-Svensson et al. 2013). The SCC element matched most closely the MRSA UCI62 strain representing ST5 (GenBank: CP018766). Carriage of blaZ, tetK and qacC genes was associated with plasmid sequences whereas other genes were inserted chromosomally. Elements carrying tetK and qacC matched previously reported S. aureus plasmids. Méric et al. showed that hospital associated S. aureus and S. epidermidis share genes involved in pathogenicity, metal toxicity resistance and antibiotic resistance. In addition they have demonstrated that high levels of recombination of genes that n a Chromosome section includes: the length of the chromosome, G+C content of the chromosome, protein coding region, ribosomal RNA and transfer RNAs numbers.
b '-' No data available in Genbank file. Draft assembly. c Plasmids section includes: the length of each plasmid and the number of plasmids. P -Plasmid. Numbers -the number of plasmids. might be successful in healthcare settings contribute to proliferation of subpopulations of two species (Méric et al. 2015).
Comparison of resistance determinant distribution revealed that the S. epidermidis G6_2 strain shared a common antibiotic resistance gene composition with other ST59 isolates, suggesting that the particular combination of antibiotic resistance genes found in the G6_2 strain is preserved across the ST59 lineage ( Figure 1). All ST59 isolates harbored aac-aph, blaZ and mecA genes, and majority contained mphC and msrA genes, whereas tetK was uniquely found in S. epidermidis G6_2. The G6_2 strain also shared the qacC plasmid with other ST59 isolates as well as the SCCmec IV sequence but not full SCCmec-SCC composite island, which was not detected in any other analyzed S. epidermidis genome.
Functional genes uniquely found in S. epidermidis G6_2 compared with reference strains Pan-genome analysis of the G6_2 strain and six S. epidermidis reference genomes revealed that 78 genes were unique to G6_2. After excluding genes found on plasmids, 64 chromosomally located genes were unique to G6_2 strain. This included a number of SCCmec-and SCC-associated genes as well as some of the chromosomally inserted resistance genes such as mphC, msrA, copZ-copA-csoR operon and the additional copy of czcD genes.

Comparative analysis of virulence genes
Pathogenicity of S. epidermidis has been linked primarily with its capacity for biofilm formation. Biofilm formation occurs by initial attachment of bacteria on both biotic and abiotic surfaces, which further accumulates into multi-layered cell agglomerates. This facilitates the internalization and persistence of S. epidermidis species in the host cells. Strains that facilitate this feature are therefore considered more virulent (Becker et al. 2014). S. epidermidis carries a number of virulence determinants that have been associated with its ability to attach to biotic and abiotic surfaces as well as the various phases of biofilm formation. Analysis of virulence gene composition based on the VF database, revealed a number of such virulence determinants that were detected in all or majority of analyzed S. epidermidis isolates, including the G6_2 strain. This included the autolysin gene atlE (138/140), the cell wall associated fibronectin Figure 1 Core-genome mid-point rooted phylogenetic tree of 136 S. epidermidis isolates. The tree nodes are annotated with bootstrap value ranges based on 100 replicates. The tree is also annotated with the sequence type (ST) assignment and antimicrobial resistance gene (AMR) carriage. Gene names above the annotation are grouped in accordance with the corresponding antimicrobial class mecA;macrolides,lincosamides and streptogramines: ermA,ermC,lsaB,mphC,msrA,msrD,cfrA and vgaA;aadC,aadD,sat4A and spc;tetracyclines: tetK and tetM;trimpethoprim: dfrG). The ST59 cluster that contains the G6_2 strain is highlighted in pink.
In addition to the described biofilm formation-associated virulence determinants, majority of S. epidermidis isolates carried the hemolysinbeta gene hlb (136/140), which was also present in the G6_2 strain. Less common was the delta hemolysin gene hld (41/140), also detected in the G6_2 strain although absent in most other ST59 isolates.
In conclusion, this study is the first analysis of the genome of S. epidermidis isolated from the general public environment and harboring a cassette of resistance genes to an array of antimicrobials. The comparison of S. epidermidis G6_2 genome with clinical reference strains revealed its antibiotic resistance and virulence gene arsenal. Resistance genes were carried on both bacterial chromosome and plasmids. We established that S. epidermidis G6_2 harbors 12 virulence genes, and delta hemolysin gene hld (41/140) is known to be detected in the G6_2 strain but absent in most other ST59 isolates. In addition, 9 antibiotic resistance determinants which are responsible for the resistance to 12 antibiotics, including streptomycin, gentamicin, penicillin, oxacillin, amoxicillin, cefoxitin, cefepime, erythromycin, fosfomycin, tetracycline, fusidic acid, mupirocin, have been identified in S. epidermidis G6_2. Additional whole genome sequence and comparative genomics analysis are warranted to further our understanding of the origin and evaluation of multidrug resistant isolates from different ecological niches.

ACKNOWLEDGMENTS
Genome sequencing was supported by Medical Research Council (MRC) partnership grant G1001787/1.

LITERATURE CITED
n