Genomic analysis of ST88 community-acquired methicillin resistant Staphylococcus aureus in Ghana

Bacterial isolates and antibiogram analysis

The 17 ST88 S. aureus isolates analysed from Ghana are listed in Table 1 and comprised five strains isolated in the Akwapim South District (Eastern Region) of Ghana with previously published genome data (GenBank accession numbers: LFNJ00000000, LFNI00000000, LFNH00000000, LFMH00000000, LFMG00000000) (Amissah et al., 2015) and 12 isolates recovered from wounds of 11 patients attending Buruli ulcer (BU) treatment centres in the Ga-South and Ga-West municipalities (Greater Accra Region) of Ghana with two isolates from one patient; one a PVL positive isolate and the other PVL negative. Patients were outpatients, nine of whom had laboratory confirmed BU. Initial isolate identification was made using colony and microscopic morphology, catalase and coagulase biochemical reactions and a Staphylase kit BD BBL™ Staphyloslide Latex Test (Becton, Dickinson and Company, NJ, USA) for further confirmation. Antibiograms were determined using the Kirby Bauer disc diffusion method according to CLSI guidelines (CLSI, 2014) and PCR targeting the mecA gene (Oliveira & de Lencastre, 2002) for identification of MRSA. Ethical clearance was obtained from the institutional review board of the Noguchi Memorial Institute for Medical Research (NMIMR) (Federal-wide Assurance number FWA00001824). All study participants were well informed of the study objectives and written informed consent was obtained either from the patient or from the guardian of the patient.

DNA extraction, whole genome sequencing and analysis

Genomic DNA was extracted from isolates using the Qiagen DNeasy kit and protocol (Qiagen, Hilden, Germany). DNA libraries were prepared using Nextera XT (Illumina, San Diego, CA, USA) and whole genome sequencing was performed using the Illumina MiSeq with 2 × 300 bp chemistry. Small Molecular Real Time sequencing was performed on the RS-II (Pacific Biosciences, California, United States) using P6-C4 chemistry, and reference genome assembly was performed as described (Baines et al., 2016).

Read mapping, variant calling and phylogenomic analysis

The sequence reads were processed using Nullarbor (nullarbor.pl 0.6, https://github.com/tseemann/nullarbor), a recently developed bioinformatics pipeline for public health microbial genomics as described previously (Kwong et al., 2016). S. aureus ST88 raw sequence reads with accession numbers ERS1354589-600 have been deposited in the European Nucleotide Archive (ENA), Project PRJEB15428. Ortholog clustering was performed using Roary (Page et al., 2015) and was visualized with Fripan (http://drpowell.github.io/FriPan/). Recombination within the core genome was inferred using ClonalFrameML v1.7 (Didelot & Wilson, 2015) with the whole genome alignment generated by Nullarbor. Using FastTree v2.1.8 (Price, Dehal & Arkin, 2010), a ML tree was generated and used as a guide tree for ClonalFrameML. Positions in the reference genome that were not present in at least one genome (non-core) were omitted from the analysis using the “ignore_incomplete_sitestrue” option and providing ClonalFrameML with a list of all non-core positions. Maximum likelihood trees were constructed using a recombination free SNP alignment using FastTree. Bootstrap support was derived from comparisons between the original trees against 1,000 trees that were built upon pseudo-alignments (sampled from the original alignment with replacement).

ST88 complete reference genome

A prerequisite for high-resolution comparative genomics by read-mapping is a high-quality, complete reference genome, closely related to the bacterial population under investigation (Kwong et al., 2016). There were no fully assembled ST88 S. aureus genomes publicly available.

Hence, to address this issue, we selected the methicillin-susceptible, penicillin-resistant ST88 S. aureus isolate AUS0325. This clinical isolate was obtained in 2013 from a patient in Melbourne, Australia who had a persistent infection of a prosthetic joint, and was part of a separate, unpublished study. The AUS0325 genome comprised a 2,771,577 bp circular chromosome with 32.9% GC content. There were no plasmids; the beta-lactamase operon (bla) was carried by the Tn552 transposon and integrated into the chromosome. The overall chromosome architecture of AUS0325 was like representative S. aureus genomes from other community-associated lineages (ST1, ST8 and ST93) but with five distinct regions of difference, discussed in more detail below (Fig. 1A). We took advantage of the PacBio data to define the Sa_aus0325 methylome. Motif analysis and inspection of the AUS0325 annotation identified two active type I restriction modification hsdMS loci. Protein alignment of the two hsdS alleles with previously characterised hsdS proteins allowed the attribution of target recognition sequences to either allele (Monk et al., 2015) (Table 2). The first hsdS recognized a motif not previously described, while the second hsdS contained an identical sequence to the target recognition domain-2 of CC30-2 and ST93-2, which recognises TCG (Table 2).

ST88 population structure

To understand the genomic diversity and evolutionary origin of the ST88 isolates, a core genome phylogeny was inferred by mapping reads from the 17 ST88 isolates (Table 1; Fig. 1B), two published ST88 MRSA genomes from Lebanon and USA and 15 other geographically and genetically distinct S. aureus clones to AUS0325 (Table 3; Fig. 1B). To assess the clonal ancestry, SNPs within inferred regions of recombination (71,862 clonal SNPs; 26,570 recombinogenic SNPs) (Fig. S1) were removed and a maximum likelihood phylogenomic tree was established using the clonal core SNP alignment (71,862 SNPs). All 20 ST88 genomes formed a discrete, closely related lineage, defined by only 1,759 core genome SNPs, compared with 71,862 SNPs among all 35 S. aureus genomes (Figs. 1B and 2A). The global tree was rooted using Bacillus_subtilis_B4068 (GenBank ID: JXHK00000000) (Berendsen et al., 2016) as an out-group and this phylogeny indicated ST88 shares a most recent common ancestor (MRCA) with ST72 (Fig. 1A).

Five distinct genomic regions were identified by ortholog comparisons, enriched among the ST88 genomes compared to the 15 other diverse S. aureus genomes. These regions included νSAα (GI-3, Fig. 1) that harboured 10 staphylococcal superantigen-like (ssl) genes, of which four were uniquely present in the ST88 isolates. Upregulation of SSLs has been reported in some CA-MRSA strains and may be involved in neutrophil and complement activation (Foster, 2005; Voyich et al., 2005). GI-3 also harboured the first of the two functional type I restriction modification hsdMS loci (see above, Table 2). GI-1 and GI-4 may be mobile integrative elements of unknown function with the presence of putative integrases and 4 and 12 CDS, respectively, all encoding hypothetical proteins. GI-1 also harbours elements of a putative restriction modification system (Fig. 1; Table S1). GI-2 contains 13 CDS. Most of unknown function although three CDS may encode membrane proteins (Fig. 1; Table S1). GI-5 had 14 CDS that included the second of the type I restriction modification hsdMS loci and seven CDS encoding putative proteases (Table 2; Table S1).

Evolution and molecular epidemiology of ST88 in Ghana

To assess the evolutionary relationships among the ST88 genomes, a phylogenomic tree comprised exclusively of ST88 genomes was established using clonal, core SNPs (1,759 clonal SNPs; 207 recombinogenic SNPs) (Fig. S2; Fig. 2A). The tree was rooted using an ST93 genome (Sa_JKD6159) as an out-group. The phylogeny and the restricted genomic diversity (658 core SNPs) suggests that the spread of ST88 MRSA in Ghana is a recent phenomenon, with isolates from the United States, Australia and Lebanon ancestral to the spread of these isolates in Ghana. Five specific clusters of CDS were also found to be exclusively present with the African ST88 genomes (Fig. 2C). These CDS were different to the five genomic regions identified in all ST88 relative to other S. aureus clones (Fig. 1B) suggesting that they were horizontally acquired by an ST88 MRCA that has since spread in Ghana (although a significantly larger sampling effort would be required to confirm this hypothesis). These regions harbour CDS suggestive of plasmid, phage and other mobile DNA elements (Table S2). We conducted a phylogeographic analysis to formally assess the relationship between the Ghanaian ST88 phylogeny and the specific geographic origin of the isolates, based on patient villages. However, there was no correlation between geography and phylogeny, suggesting again that the spread of ST88 in Ghana has been recent and rapid (Fig. 3).

Phenotypic and genotypic antibiotic resistance

All 17 Ghanaian ST88 isolates harboured a SCCmec-IV [2B] cassette, and displayed phenotypic resistance to β-lactams, tetracycline and chloramphenicol (Table 1). Isolates were variably resistant to erythromycin, clindamycin, trimethoprim, amikacin and streptomycin (Table 1). There was agreement between phenotypic and inferred genotypic resistance (Fig. 1B). For the four genes (blaZ, mecA, tetL and tetM) detected in all 12 ST88 isolates from the Greater Accra Region, resistance correlated with phenotypic resistance to all β-lactams and tetracyclines. Six isolates showed phenotypic and genotypic resistance to chloramphenicol (Table 1; Fig. 1B). Five of these isolates were from the same health centre; however, the time of isolation and the geographic origins of the patients were different, suggesting that these isolates are widespread across the region and were not acquired from a common source.