Impact Factor 4.259 | CiteScore 4.30
More on impact ›

Correction ARTICLE Provisionally accepted The full-text will be published soon. Notify me

Front. Microbiol. | doi: 10.3389/fmicb.2019.02218

Whole Genome Sequencing of Australian Candida glabrata Isolates Reveals Genetic Diversity and Novel Sequence Types

 CHAYANIKA BISWAS1, Vanessa Marcelino2,  Sebastiaan Van Hal3,  Catriona Halliday4, Elena Martinez1,  Qinning Wang1,  Sarah Kidd5, Karina Kennedy6, Deborah Marriott7, C O. Morrissey8, Ian Arthur9, 10, Kerry Weeks11, Monica Slavin12,  Tania C. Sorrell2, Vitali Sintchenko1,  Wieland Meyer13 and  Sharon C. Chen1*
  • 1Centre for Infectious Diseases and Microbiology Public Health, Western Sydney Local Health District, Australia
  • 2Marie Bashir Institute, University of Sydney, Australia
  • 3Royal Prince Alfred Hospital, Australia
  • 4Westmead Hospital, Australia
  • 5South Australia Pathology, Australia
  • 6The Canberra Hospital, Australia
  • 7St Vincent’s Hospital Sydney, Australia
  • 8Monash University, Australia
  • 9Pathwest Laboratory Medicine, Australia
  • 10Queen Elizabeth II Medical Centre, Australia
  • 11Royal North Shore Hospital, Australia
  • 12Peter MacCallum Cancer Centre, Australia
  • 13Westmead Institute for Medical Research, Australia

1 Introduction

The opportunistic yeast Candida glabrata is the second most common cause of candidemia and invasive candidiasis (IC) in many countries (Guinea, 2014; Arendrup et al., 2013; Pfaller et al., 2014; Chapman et al., 2017). Its clinical importance as a species lies in its reduced susceptibility to azole antifungal agents; more recently, resistance to the echinocandins as well as resistance to both these drug classes (Pfaller et al, 2012; Wisplinghoff et al., 2014; Shields et al., 2015). This has prompted much investigation of the epidemiology and biological properties of C. glabrata infections (Vale-Silva and Sanglard, 2015).

Because the prevalence of C. glabrata candidiasis and drug resistance rates varies both between and within geographical region, local epidemiological data are essential to inform management (Guinea, 2014; Pfaller et al., 2012). The reasons for this variation are uncertain but likely include prior exposure to azoles, patient factors and geographic location-specific determinants. Fungus-specific factors such as genetic strain variation within species are also pertinent. Delineation of intraspecies variation is useful not only to elucidate the molecular epidemiology of C. glabrata infections but also to assess potential transmission routes, biological niches and population structure. Yet relatively little is known about the genomic variation between isolates from different regions or the clinical significance of such differences.

Genetic typing methods e.g. pulsed-field gel electrophoresis, microsatellite analysis and multilocus sequence typing (MLST) have been used to determine genetic relatedness of C. glabrata (Dodgson et al., 2003; Abbes et al., 2012; Lin et al., 2007). In particular, the use of a standardised 6-locus MLST system (Dodgson et al., 2003) has improved discrimination between isolates with good reproducibility and portability of data via internet-accessible databases. Major findings from MLST analyses highlight that despite description of a broad range of MLST sequence types (STs), C. glabrata appears to be highly clonal with infrequent emergence of novel STs, which may be restricted to various geographical regions (Hou et al., 2017; Amanloo et al., 2018, Lott et al., 2012). This observed clonality however, may be fluid with temporal shifts of the major C. glabrata subtypes documented over time in one study (Lott et al., 2010). More discriminatory methods for pathogen discrimination such as next generation sequencing (NGS) offer new insights into C. glabrata genetics including its molecular epidemiology and population dynamics. Global spread of previously isolated populations was inferred from genomic data in a recent study (Carrete et al., 2018). In addition, NGS has been utilized to elucidate mechanisms of drug resistance in this species from the diagnostic laboratory perspective (Biswas et al., 2017; Singh-Babak et al., 2012).

In Australia, we observed a 1.7-fold increase in the proportion of Candida bloodstream infections caused by C. glabrata over a decade (2004-2006 vs. 2014-2015) (Chapman et al., 2017; Chen et al., 2006). Our laboratory is increasingly using whole genome sequencing (WGS) approaches, in line with international trends, in public health practice and investigations of nosocomial infections (Besser et al., 2018). Here we applied WGS to investigate the genetic diversity of Australian C. glabrata strains across more than a decade and sought associations between the frequency of sequence types and two time periods and with drug susceptibility to fluconazole.


2 Materials and Methods
2.1 Ethics Statement
All isolates were obtained from our culture collection spanning 10-20 years and represent previous surveillance isolates for which research ethics approval had been obtained. The present study was a laboratory-based epidemiological study. No identifiable patient data or medical records were accessed.

2.2 Isolates and Identification
Fifty-two C. glabrata (sensu stricto) isolates were studied. These comprised C. glabrata ATCC 90030 and 51 C. glabrata isolates from Australia obtained through the culture collection at the Clinical Mycology Laboratory, Westmead Hospital and the Molecular Mycology Research Laboratory, Westmead Millennium Institute for Research, Sydney. With the exception of two isolates recovered from the same patient three weeks apart, all isolates represented single patient episodes of IC. The majority (>90%) of isolates were from the jurisdictions of New South Wales and Victoria. All isolates were re-confirmed as C. glabrata sensu stricto by matrix-assisted laser desorption ionisation-time of flight technique (MALDI-TOF MS) supplemented by ITS sequencing as required (White, 1990).

2.3 Susceptibility Testing
Susceptibility to antifungal agents were determined using the Sensititre® YeastOneTM YO10 methodology (TREK Diagnostics, Cleveland, OH, USA) according to Clinical and Laboratory standards Institute (CLSI) methodology (Clinical and Laboratory Standards Institute [CLSI], 2017a). Candida parapsilosis ATCC 22019 and Candida krusei ATCC 6258 were the quality control strains. MIC values were interpreted according to CLSI M60 guidelines for fluconazole and the echinocandins (Clinical and Laboratory Standards Institute [CLSI], 2017b); where there are no clinical breakpoints (CBPs) (for voriconazole, posaconazole and amphotericin B), species-specific epidemiological cut-off values (ECVs) defined isolates as wild-type (WT) or non-WT (Clinical and Laboratory Standards Institute, 2018). There are neither CBPs or ECVs for 5-fluorocytosine.

2.4 DNA Extraction and Library Preparation for Whole Genome Sequencing
C. glabrata ATCC 90030 and the 51 clinical isolates were subcultured on Sabouraud's dextrose agar for 48 h at 35°C prior to testing to ensure purity. Genomic DNA was extracted using the Wizard® Genomic DNA Purification kit (Promega, Alexandria, NSW, Australia) and the concentration was quantified by Quant-iT™ PicoGreen® dsDNA Assay Kit (Life Technologies, Mulgrave, VIC). The Nextera XT kit (Illumina, San Diego, CA) was used to construct genomic libraries. Tagmentation, PCR amplification and cleanup, library normalisation and pooling, and sequencing on the NextSeq 500 platform (Illumina) were carried out with 2 X 150-bp paired-end chemistry as previously described (Biswas et al., 2017).

2.5 Data and Genome Analysis
Sequence reads were deposited in the NCBI Sequence Read Archive (SRA: project number PRJNA480138) and were mapped against the reference genome of C. glabrata CBS138 (GenBank Accession 4 No. GCA_0002545.2; http://www.candidagenome.org).

Obtained sequence reads were mapped to each chromosome independently (using C glabrata CBS 138 chromosomes A to M as the reference) employing Stampy v1.0.23 (Lunter et al., 2011) with pre-BWA alignment. Analysis of the mitochondria was not included. Variants were called using FreeBayes v1.1.0-dirty and filtered for read depth (minimum 20), read base quality (minimum Phred score 30), mapping quality (minimum 30) and proportion of reads supporting the variant (>0.9). All indels were excluded from the mapping-based analysis. An aligned mapped file for each chromosome was constructed for all isolates using an in-house script. All probable recombination events were identified using Gubbins (Croucher et al., 2015) and subsequently masked prior to concatenating all chromosome sequences into a single SNP alignment.

To infer the phylogenetic relationship of the Australian isolates, the best-fitting substitution model (TVM+F+ASC+R2) was selected with the Bayesian Information Criterion using ModelFinder implemented in IQ-Tree v.1.6.2 (Nguyen et al., 2015; Kalyaanamoorthy et al., 2017). A maximum likelihood tree was then reconstructed using IQ-Tree using 1000 ultrafast bootstrap replicates (Hoang et al., 2018).

To place our data into global context, all 38 publicly-available C. glabrata Illumina short read sequence data from seven countries other than Australia were downloaded and included in the analysis (Carrete et al., 2018; Havelsrud and Gaustad, 2017; Table 2). A network approach using SplitsTree4 (Huson and Bryant, 2006) was employed to examine the relationships between our isolates and the isolates from other countries.

2.6 Single Polynucleotide Polymorphism of Genes Associated with Antifungal Resistance
All SNPs in genes known for their role in drug resistance in C. glabrata (Berila et al., 2009; Arendrup and Perlin D., 2014; Sanguinetti et al., 2005) were manually curated in CLC Genomics Workbench (CLC Bio version 7.0, Arrhus, Denmark) with only non-synonymous SNPs reported. Genes examined included FKS1, FKS2, FKS3 (for echinocandin resistance) FCY1, FCY2, CgFPS1, CgFPS2 (5-fluorocytosine resistance), ERG9, ERG11, CgCDR1 and CgPDR1 (azole resistance) and MSH2 (for multi-drug resistance). Only non-synonymous SNPs with a minimum read depth coverage of 20, defined as high-quality (hq), were included in the analysis.

2.7 MLST Sequence Types and Principal Component Analysis
In silico MLST sequence types (STs), inferred from whole genome sequence data (genome types) were obtained from assembled contigs using SPAdes v3.1.1.1 (Bankevich et al., 2012) and MLST software (Seeman, T., https://github.com/tseemann/mlst). All obtained STs were subsequently confirmed using a read based approach implemented through SRST2. Four novel C. glabrata MLST types (See Results) were submitted to the C. glabrata MLST database (Dr. Andrew Dodgson, http://glabrata.mlst.net; accessed September 11, 2018; and now Professor Oliver Bader, accessed November 5, 2018), and designated as ST123, ST124, ST126 and ST127. Clustering of isolates by genetic similarity according to time of isolation (2002-2004 vs. 2010-2017) and drug susceptibility (susceptible-dose-dependent (S-DD) or resistant (R) to fluconazole (CLSI, 2017b)) was examined by Principal Component Analysis (PCA). PCA was based on a pairwise SNP distance matrix calculated under the K80 model with pairwise deletion, using the native stats R v3.4.1 package (R Core Team., 2013), and visualized with ggplot2 (Wickham, 2009).


3 Results
Of 51 Australian clinical isolates, 49 (94%) were cultured from blood. Thirteen (25%) isolates were from the time period 2002-2004, and 38 (75%) from 2010-2017 (Table 1).

3.1 Susceptibility Data
Table 1 summarizes the MIC values of the isolates against five antifungal agents All clinical isolates tested had low MICs against 5-fluorocytosine (≤0.12 μg/ml) and WT MICs (< 2 μg/ml) against amphotericin B. Four (7.8%) isolates (strains WM_18.26, WM_18.24, WM_18.64, WM_18.63) were resistant to caspofungin (MIC range 2- > 8 µg/ml; Table 1), and cross resistant to micafungin and anidulafungin (results not shown) (CLSI, 2017). Thirteen isolates (Strains WM_03.419, WM_03.449, WM_04.113, WM_05.113, WM_18.33, WM_18.40, WM_18.27, WM_18.42, WM_18.49, WM_18.51, WM_18.53, WM_18.65, WM_18.66) were classified as resistant to fluconazole with MICs of ≥64 ug/ml (CLSI, 2017b); all also had non-WT MICs to voriconazole whilst posaconazole MICs ranged from 2 to >8 μg/ml (Table1). However, six additional isolates (strains WM_18.44, WM_18.48, WM_18.56, WM_18.57, WM_18.59 and WM_18.62) had non-WT MICs to voriconazole and posaconazole although were susceptible-dose dependent to fluconazole, and yet 11 other isolates had non-WT MICs only to voriconazole (Table 1).

3.2 Sequence Analysis
Overall, an average of 95% of sequencing reads were mapped to the C. glabrata reference genome with a median read depth coverage of 75-fold. After mapping to each reference chromosome independently, a number of recombination events were identified (range, n=320-776, lowest for chromosome B and highest for chromosome I - data not shown).

3.3 In silico MLST and Global Phylogeny
By WGS, there were 18 distinct STs defined based on the alleles from six genetic loci (FKS, LEU2, NMT1, TRP1, UGP1, URA3) among 52 isolates (Table 1, Table S1), including four new STs (referred to as ST123, ST124, ST126 and ST127) not previously recognised by the C. glabrata MLST database (http://cglabrata.mlst.net; see Table S1 for the allele numbers of the new STs). Strain ATCC 90030 typed as ST10.

Of the known STs, the commonest ST amongst the Australian isolates was ST3 (8/51, 15.7% of isolates) followed by ST83 (7/51, 13.7%), ST7 and ST26 (each n=5, 9.8%). Collectively, these four STs were responsible for almost half (n=25; 49%) of the isolates. The most common new ST was ST123 (n=4 isolates). Eight (17.6%) STs (ST6, ST18, ST36, ST45, ST59, ST124, ST126, ST127) were represented by only a single isolate (Table 1). Despite a relatively small number of isolates sequenced, the number of ST types was considerable compared to previous studies suggesting a relatively high genetic diversity within Australian C. glabrata isolates. Three isolates cultured from body sites other than blood were of ST10, ST16 and ST26 (Table 1). The two isolates from the same patient (WM_18.63 and WM_18.64) were both ST8.

In general, the whole genome data clustered broadly within determined MLST types but with greater intra-cluster resolution. The split tree network analysis (Figure 1) resolved isolates to at least nine terminal branches containing four or more isolates; many of these comprise multiple isolates clustering together, and with isolates representing disparate geographic locations. However, four branches comprised of Australian isolates only (Figure 1). To place local isolates into a more global perspective, the STs of isolates reported from seven other countries are shown in Table 2. Certain STs were common to isolates from the regions studied herein e.g. ST3 (Belgium, France, Germany, USA, Australia), whilst others were either more restricted, or were more prevalent to one or two countries e.g. ST6 in Norway and France, ST8 in the USA and continental Europe. The two isolates from Taiwan showed new STs. Of note, isolates of ST19 were absent from Australia (vs. 4/12 US isolates; Table 2).

3.4 Genomic Similarity According to Period of Isolation and Susceptibility to Fluconazole
On analysis of the sequenced genomes by PCA, there was no temporal association between the two periods of isolation (2002-2004 and 2010-2017) and genomic similarity (as represented by ST distribution), or between genomic similarity and fluconazole susceptibility as measured by MICs (fluconazole S-DD: n=39 isolates vs. fluconazole-resistant: n=12) (Figures 2A and 2B).
The phylogenetic relationship of the 51 Australian isolates was also reconstructed with high bootstrap support (Figure 3). There was no association between drug susceptibility to fluconazole, voriconazole, posaconazole or caspofungin (results were similar for anidulafungin and micafungin) or resistance, and phylogenetic clustering. Rather the analysis illustrated that resistance or non-WT MICs emerged at several time points along the phylogeny.

3.5 Analysis for SNPS Related to Drug Resistance and Their Relationship to STs
Of two isolates with caspofungin MICs of ≥ 8 μg/ml, one (strain WM_18.24; ST16) contained the FKS1 mutation leading to the amino acid substitution Ser629Pro, whilst strain WM_18.26 (ST10) harbored the FKS2 mutation Ser663Pro. There were no other SNPs in any of FKS1, FKS2 or FKS3 in both strains. Isolates WM_18.63, and WM-18.64 (both ST8) recovered from the same patient harbored the FKS1 mutation Ser629Pro as well as a FKS2 mutation Glu784Gly. The Glu784Gly mutation was not present in any echinocandin susceptible isolates. Other SNPs were present in FKS1 (Gly14Ser) and FKS2 (Thr926Pro) but only in isolates of ST3. Several SNPs in FKS3 (Table S2) were present in both echinocandin-susceptible and echinocandin-resistant isolates.

The presence of SNPs in genes linked to 5-fluorocytosine resistance e.g. CgFPS1, CgFPS2 and CgFCY1 and CgFCY2 broadly, varied with ST with no SNPs observed in isolates of ST6, ST22, ST55, ST59 and ST123 (Table S2). Fluconazole-resistant isolates (Table 1) in general harbored mutations in CgPDR1 and to a lesser extent in CgCDR1, but overall, SNPs in these genes and in other efflux pump genes, CgFLR1 and CgSNQ2 (data not shown) were also present in azole-susceptible/WT isolates (Table S2). However, although no particular SNP was definitively linked to the resistance phenotype, 5/13 fluconazole-resistant isolates harbored Pdr1 amino acid substitutions in the region beyond the first 200-250 amino acids with no fluconazole-susceptible isolate containing such changes. In addition, 2/6 isolates with non-WT MICs to voriconazole and posaconazole (strains WM_18.48 and WM_ 18.62; fluconazole MIC both 32 μg/ml) also harbored mutations resulting in amino acid substitutions outside this region. Conversely, substitutions within the first 250 amino acid positions in Pdr1 were common to both azole resistant and susceptible isolates (Table S2). Eight fluconazole resistant isolates however, did not demonstrate a PDR1 mutation. Isolates exhibiting pan-azole resistance, or which had non-WT MICs also had carried mutation CgCDR1 His58Tyr (6/10 isolates), but the last was also present in azole-susceptible isolates.

SNPs occurred in isolates of diverse ST. There were no SNPs in ERG11 and the few SNPs observed in ERG9 were predominantly in isolates of ST3 and ST26.

SNPs in the MSH2 gene were observed in 19 of 51 (37%) isolates with three main locations of mutations – Val239Leu (9 isolates), Glu456Asp (7 isolates) and Leu269Phe (3 isolates), with two isolates (strains WM_18.63 and WM_18.64) having two mutations at Val239Leu and Ala942Thr (Table S2). The same MSH2 mutations were found in azole susceptible as well as azole-resistant/non-WT isolates. Overall, SNPs were identified in isolates of diverse STs including in isolates of the new STs, ST123 and ST127 (Table S2). Whilst the mutation Glu456Asp was found in isolates of 5 different STs and that of Val239Leu in 4 different STs, the Leu269Phe mutation was found only in isolates of ST16. The combination of Val239Leu and Ala942Thr were only identified in ST8 isolates, both of which were pan-echinocandin resistant.

4 Discussion
Understanding the genomic diversity of C. glabrata and its antifungal susceptibility patterns is key to optimal management of infections caused by this problematic pathogen. The few studies that have examined the genetic variation of large culture collections have employed traditional MLST and indicate a predominantly clonal population structure with infrequent recombination (Dodgson et al., 2005). Prevalence of circulating STs also showed geographical bias (Amanloo et al., 2018; Dodgson et al., 2003, Hou et al., 2017). Hence, genetic variation amongst isolates from one region cannot be generalised to another. Here, we determined for the first time using a WGS approach, the relative frequency of endemic STs among 51 Australian C. glabrata isolates from two time periods, and verified the applicability of WGS to determine STs, STs by WGS clustered isolates within similar “ST” groupings as in silico MLST with good intra-cluster resolution (Figure 1).

Sequence typing demonstrated relatively “large” genetic diversity amongst Australian C. glabrata isolates, with just under half of the isolates represented by only four STs. The remaining STs, not only represented Australian specific STs (Figure 1, Figure 3) but suggested an overall diversity within this pathogen that is greater than previously appreciated. In a nationwide Chinese study (Hou et al., 2017) of 411 isolates, a “new” ST sequence type was encountered approximately every 11 isolates compared to our study, which observed a “new” ST every 5 isolates. These, and our data emphasise the regional differences, with 75.9% of Chinese isolates comprised of ST7 and ST3 compared to only 25.5% of Australian isolates. Another recent report noted a predominance of ST3 and ST7 (70% isolates) in Korea (Byun et al., 2018) whilst in Iran, three STs (ST59, ST74 and ST7) accounted for 50% of isolates further supporting the notion of low intraspecies diversity within C. glabrata (Amanloo et al., 2018). Isolates belonging to ST3 have been reported with relative high frequency from Europe and Asia and now, from Australia (Dodgson et al., 2003; this study). The presence of strains with the same ST on different continents demonstrates that clones may have arisen from the same ancestor and disseminated globally followed by local adaptation.
Whilst ST5 isolates were reportedly common in Europe (Dodgson et al., 2003), this ST was not found amongst our Australian isolates nor amongst those from a more recent study of European and US isolates (Tables 1 and 2). Conversely, isolates of ST7 appear uncommon in Europe and the USA but are more prevalent in Japan, Korea and China (Dodgson et al., 2003; Carrete at al., 2018, Hou et al., 2017; Byun et al., 2018). Strains of ST8, ST18, and ST19 were the commonest types in the USA (Dodgson et al., 2003) whereas we identified only one ST18 isolate, no ST19 isolates and two ST8 isolates (from the same patient) in Australia. Broadly, there are more common STs between Australian and Asian isolates than between Australian and US/European isolates likely reflecting the geographical proximity between Asia and Australia. Further studies involving a larger number of C. glabrata isolates to test this hypothesis would be of interest. The observed geographic variation among STs highlights the importance of acquiring local data.

MLST analysis of US C. glabrata isolates collected over three time periods between 1992-2009 revealed a relatively small number of STs with little genetic differentiation (Lott et al., 2010). In the present study, the results indicate that there is no evidence of genomic similarity or ST distribution among the sequences isolated in the two timeframes studied (Figure 2A), suggesting that there have been no marked shifts in the present STs. We further found no association between WGS STs and susceptibility or /resistance to fluconazole (Figure 2B), consistent with that reported by traditional MLST studies (Amanloo et al., 2018; Dodgson et al., 2003). The phylogenetic tree (Figure 3) suggests that both azole and echinocandin resistance may arise at multiple time points, independent of strain clustering. The small numbers of isolates are acknowledged as a study limitation.

Strain typing is essential for epidemiological investigation. MLST has the advantage of providing easily comparable results via internet-accessible databases (Dodgson et al., 2003), but does not adequately capture the breadth of genetic diversity and is not readily available in diagnostic laboratories. The present study has illustrated the utility of WGS to delineate genome variability in C. glabrata and importantly offers both superior discriminatory power and convenience. The costs of the two techniques are nearly identical – approximately AUD 50/sample for MLST and AUD 80/sample for WGS. With decreasing footprint and technological advances however, cost reduction for WGS is anticipated.

The echinocandin resistance rate of 7.8% (4/51 isolates) in the present study is influenced by sampling bias, being <2% across Australia (Chapman et al., 2017), lower than that in the USA (Alexander et al., 2013, Shields et al., 2015). The FKS1 mutation Ser629Pro (in one isolate) and the mutation FKS2 Ser663Pro (three isolates) identified are among the most common in C. glabrata strains with high-level resistance phenotypes (Garcia-Effron et al., 2009; Arendrup and Perlin, 2014). None of the isolates harboured other well-known mutations that confer echinocandin resistance e.g. R665G, R636S and F659Y (Zimbeck et al., 2010; Shields et al., 2015). Conversely, the role of FKS2 Glu748Gly in isolates WM_18.63 and WM_18.64 in echinocandin resistance remains uncertain as this SNP has not been previously described. Interestingly, the SNP was absent in genomes of all ST8 isolates from other countries in that cluster (Italy2, France9, USA3, USA4 and USA11)

Approximately 25% of C. glabrata isolates in our study were fluconazole-resistant, comparable to that in the USA (20-30%) (Castanheira et al., 2014). Genome-wide sequencing revealed mutations in several multidrug resistance transporter genes (Table S2) (e.g. CgPDR1 and CgCDR1) that are associated with resistance through activation of drug efflux pumps (Ferrari et al., 2009; Rodrigues et al., 2014). Although SNPs in these genes were found in both azole-resistant/non wild-type as well as azole-susceptible/wild-type isolates, it is noteworthy that as previously reported, amino acid alterations within the first ~250 amino acids of Pdr1, were common in both susceptible and resistant isolates (Ferrari et al., 2009; Tsai et al., 2010; this study). Conversely amino acid substitution differences between susceptible and resistant isolates had a propensity to be located outside this region (Table S2).
However, the effects of specific SNPs in these genes on the resistance phenotype needs to be confirmed by functional analyses to assess the gene expression level but was beyond the scope of the present study. Through gene deletion studies, it has been recently shown that when functionally active, all of CDR1, PDR1 and SNQ2 contribute to high level resistance to azoles (Whaley et al., 2018). The absence of ERG11 mutations predominant in other Candida species, such as azole-resistant C. albicans is also noteworthy (Morio et al., 2010).

Mutations in the DNA mismatch repair gene MSH2 are reported to be a genetic driver of multi-drug resistance and about 55% of C. glabrata isolates are expected to contain MSH2 gene mutations. Whilst over 35% of our isolates harboured a mutation in MSH2, these were present in both azole-susceptible and azole-resistant (or non-WT) isolates. Hence it is possible that MSH2 mutations are more a marker of “ST” of C. glabrata rather than an indicator of drug resistance, having been linked to isolates of ST16 (Healey et al., 2016; Delliere et al., 2016); similarly, in our study, the mutation Leu269Phe was only present in isolates of ST16 (Table S2). However, the remaining MSH2 mutations were found in several STs including two of the novel STs identified herein. Of note, the mutation combination of Val239Leu and Ala942Thr were only identified in ST8.

Limitations of the present study include the relatively small numbers of isolates analysed, which may have precluded the identification of associations between ST and period of isolation. In addition, the majority of isolates were from blood. However, Lott et al. demonstrated that bloodstream isolates of C. glabrata were genetically indistinguishable from those colonising the host (Lott et al., 20912). By using genome-wide information in 33 strains, Carrete et al. inferred the population structure of C. glabrata where strains were clustered into highly divergent clades but with the structure suggesting recent global spread of previously isolated populations (Carrete et al, 2018). WGS with its superior discrimination is well placed to provide additional clues to evolutionary traits in this species.

In conclusion, we have shown the value of a WGS approach for high resolution sequence typing, discovery of novel STs of C. glabrata, and the potential to monitor trends in genetic diversity. We envisage useful contribution of our data including that of five novel STs to the global sequence repository. Our results suggest that azole, as well as echinocandin, resistance may arise at multiple time points independent of strain clustering or of STs. WGS assessment for echinocandin resistance has good potential to augment phenotypic susceptibility testing methods. Further study by WGS of C. glabrata STs and their evolution over time is warranted.

Keywords: whole genome sequencing, Candida glabrata, mlst, sequence type, Australia

Received: 09 Aug 2019; Accepted: 11 Sep 2019.

Copyright: © 2019 BISWAS, Marcelino, Van Hal, Halliday, Martinez, Wang, Kidd, Kennedy, Marriott, Morrissey, Arthur, Weeks, Slavin, Sorrell, Sintchenko, Meyer and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

* Correspondence: Prof. Sharon C. Chen, Centre for Infectious Diseases and Microbiology Public Health, Western Sydney Local Health District, Sydney, Australia, Sharon.Chen@health.nsw.gov.au