Clonal Strain Persistence of Candida albicans Isolates from Chronic Mucocutaneous Candidiasis Patients

Chronic mucocutaneous candidiasis (CMC) is a primary immunodeficiency disorder characterised by susceptibility to chronic Candida and fungal dermatophyte infections of the skin, nails and mucous membranes. Molecular epidemiology studies of CMC infection are limited in number and scope and it is not clear whether single or multiple strains inducing CMC persist stably or are exchanged and replaced. We subjected 42 C. albicans individual single colony isolates from 6 unrelated CMC patients to multilocus sequence typing (MLST). Multiple colonies were typed from swabs taken from multiple body sites across multiple time points over a 17-month period. Among isolates from each individual patient, our data show clonal and persistent diploid sequence types (DSTs) that were stable over time, identical between multiple infection sites and exhibit azole resistant phenotypes. No shared origin or common source of infection was identified among isolates from these patients. Additionally, we performed C. albicans MLST SNP genotype frequency analysis to identify signatures of past loss of heterozygosity (LOH) events among persistent and azole resistant isolates retrieved from patients with autoimmune disorders including CMC.


Introduction
Chronic mucocutaneous candidiasis (CMC) is a primary immunodeficiency disorder characterised by susceptibility to Candida spp. infection of the skin, nails and mucous membranes. It manifests separately to vulvovaginal candidiasis and rarely as invasive candidiasis. Generally presenting as chronic localised lesions, CMC persists due to isolated primary immune defects or secondary to predisposing immune compromising conditions such as HIV. It can also present as a heterogeneous disorder such as autoimmune polyendocrinopathy candidiasis ectodermal dystrophy (APECED), as a multisystem syndrome with hyper immunoglobin E syndrome (HEIS) or hypothyroidism, and can be complicated by additional autoimmune conditions such as inflammatory bowel disease (IBD) [1][2][3].
In healthy individuals, immunological surveillance of potential invasive microorganisms involves immune recognition of the pathogen and activation of innate and adaptive immune taken from infected sites (mouth, genitals, skin) for diagnostic culture and sensitivity testing to confirm Candida presence. Patients P1-P5 had a primary immune deficiency (PID) manifesting as CMC. Patient P6 had secondary oral candidiasis as a consequence of using steroid inhalers for severe asthma. We previously assessed all PID CMC patients by whole exome sequencing as part of a separate study [3] in which we identified gain-of-function STAT1 mutations in P1, P3 and P5, no mutation was identified in P4 or P6. GOF-STAT1 CMC patients and P6 had disease onset in early childhood/adolescence while P4 who had a candida granuloma of the soft palate and has since developed squamous cancer of the same area first presented at 50 years of age. PID CMC patients were followed for at least 5 years in this clinic, during which time all of them were on antifungal oral treatments. Antifungals were selected based on sensitivities and were either azoles (fluconazole and itraconazole) or amphotericin suspension 100 mg/ml (produced in our local hospital Pharmacy), diluted 1:10 before taking into mouth, swirled and swallowed 4x daily. (See Table 1).
Multiple colonies were typed from swabs taken over a 17 month period. Patients 1 and 3 were swabbed 4 times, patient 6 was swabbed twice and patients 2, 4 and 5 were swabbed once. Oral swabs were taken at all sampling time points, swabs of the groin and buttocks were taken for the penultimate time point of patient 3 (Table 1 and

Isolation and species identification
Patient samples were plated and purified on Sabouraud dextrose (Sabdex) agar plates containing 1% mycological peptone (w/v), 4% (w/v) glucose and 2% (w/v) agar and CAN2 agar (Biomerieux) to detect the presence of mixed isolates. Isolates of yeasts were also inoculated into Sabdex broth. Species identification was carried out using MALDI TOF mass spectrometry. All swabs were taken for routine diagnostics purposes.

Antifungal susceptibility testing
Minimum inhibitory concentrations were determined by broth micro-dilution testing using a modification of the CLSI (formerly NCCLS) guidelines M27-A2 [48]. Drug concentrations ranged from 0.032 mg/L to 16 mg/L for caspofungin and a non-liposomal amphotericin B suspension, and 0.13 mg/L to 64 mg/L for fluconazole and voriconazole. Each drug was serially diluted with sterile water to 50 μL in flat bottomed 96 well plates. Cultures were grown in NGY medium (0.1% (w/v) neopeptone, 0.4% (w/v) glucose, 0.1% (w/v) yeast extract) for 12 h at 30°C to exponential phase and 20 μL transferred to 11 mL RPMI inoculation medium (2x RPMI-1640, 2% (w/v) glucose, buffered to pH 7.0 with MOPS). Aliquots of 50 μL inocula in RPMI were added to drug plates (total volume 100 μL), incubated for 24 h at 37°C and analysed using VERSAmax microplate reader at 405 nm. Each of 42 sequenced colonies were assayed in triplicate, and MIC90 and MIC50 values reported.

Genomic DNA extraction and sequencing
Genomic DNA was extracted from yeast grown from single colonies in overnight YPD broth cultures (1% yeast extract, 2% bacto-peptone and 1% glucose). Pelleted cultures were subject to mechanical lysis in 2 mL Eppendorf tubes with 200 μl extraction buffer (100mM Tris-HCl [pH 8.0] containing 2% Triton X-100, 1% sodium dodecyl sulfate, and 1 mM EDTA), 200 μl 1:1 (vol/vol) phenol chloroform and 0.3g of acid washed glass beads (0.45 to 0.52 mm in diameter) using a bead beater. Addition of 200 μL TE (1 mM EDTA, 10 mM Tris-HCl, pH 8.0) preceded phase separation by centrifugation and ethanol precipitation of DNA containing upper phase, as previously described [18]. For each single colony isolate the internal region of seven housekeeping genes were amplified using seven pairs of primers (S1 Table) in separate PCR reactions with a first cycle of denaturation for 2 min at 94°C, followed by 25 cycles of denaturation at 94°C for 1 min, annealing at 52°C for 1 min, elongation at 72°C for 1 min, and a final extension step of 10 min at 72°C. PCR reactions were performed in 20 μl volumes using high-fidelity DNA polymerase (Finnzymes Phusion F-530L), with thermal cycler conditions as advised by the manufacturer and as previously described [16]. Aliquots of 5 μl of PCR products were run on agarose gels to confirm successful generation of products at expected sizes and 5 μl aliquots were purified using Mag-Bind E-Z magnetic beads as per manufacturer's instructions. Chainterminating dideoxynucleotide Sanger sequencing reactions were prepared from purified PCR products separately for forward and reverse strands and precipitated with addition of 15 μl sterile H2O to prevent dye blobs, 50 μl 3 M NaOAC (pH 5.2) followed by 70% ethanol rinse as previously described [15,49]. PCR products containing purified and dried sequencing reactions were sent to the University of Oxford, Department of Zoology and sequenced on ABI 3730 instruments. DNA sequence results were analysed and SNPs determined using DNASTAR

SNP frequency distribution
At all SNP positions across MLST sequences, implementing the IUPAC ambiguity nomenclature for heterozygous SNPs, the frequency of each SNP genotype were calculated for Dataset A and the frequency distribution matrix was used to prepare normalised frequency distribution matrices of the 6 further datasets (B-G). Further data reduction was performed by tallying within non-overlapping 10% ranges and multiplying these counts by the range upper bound value to produce a normalised frequency distribution (NFD) of normalised frequency presence (NFP), and absence (NFA) in respect of heterozygous (NFPhet, NFAhet) and homozygous (NFPhom and NFAhom) classes of SNPs. (see also S5 and S6 Figs).

Results
MLST analysis was performed on 42 C. albicans isolates from 6 CMC patients collected over a period of 17 months. All isolates retrieved from the same patient had the same DST and no LOH, point mutations or larger scale genetic differences were observed between colonies obtained from the same patient. Of the 6 DSTs observed, only one (DST 392) had been previously reported. These data show C. albicans DSTs persisting across time, between different anatomical sites as identical DSTs mutually exclusive for each patient in this cohort (

Phylogenetic analysis
Phylogenetic neighbour net analysis of DST profiles together with multiple sequence alignment and UPGMA tree of concatenated SNP positions demonstrated the distinct genetic backgrounds of these isolates. Clade membership and eBurst analysis highlighted 4 identical genotypes (MP1b, ZWF1b, AAT1a, AAC1) for DST 392 (P4) and DST 2130 (P6) in clade 4, and 4 identical genotypes (ADP1a, MP1b, SYA1, ZWF1b) for DST 2125 (P2) and DST 2129 (P5) in clade 2 with both pairs of DSTs identical at a single genotype (MP1b). Together these genetic divisions and groupings strongly confirm independent origins for these isolates with no shared source of infection for these patients ( Table 2, Fig 1 and S3 Fig).

Susceptibility to antifungal drugs
In general, few changes in drug resistance profiles of isolates were noted over the time course of the study. Most isolates were resistant to both azoles tested from the start of the study period with MIC values above published breakpoints. The MIC testing revealed similar susceptibility profiles within a doubling dilution range across multiple isolates from the same swab site and across multiple swabs at multiple body sites and time points with the exception of two sets of voriconazole values. A fourfold increase was recoded in voriconazole MIC for P1 and a fourfold decrease in voriconazole MIC50 occurred between the penultimate and final swab for P3 (data not shown), the MIC90 values for these isolates remained stable across all swabs for the patients. Overall these susceptibility data show MIC values for strains of a consistent and relatively highly resistant phenotype with no notable acquisition of increased resistance during sustained chronic carriage over the study period (Table 2).

SNP frequency distribution
To assess the distribution of SNPs across MLST sequences, SNP genotype frequencies were computed for a non-redundant dataset of 2867 isolates taken from the C. albicans MLST database (Dataset A), and used as a normalising dataset for 6 further datasets with distinct epidemiological characteristics (Datasets B-G), including 6 individual DSTs identified herein (Dataset C). Patterns of SNP frequencies for datasets B-G normalised against Dataset A were visualised in NFD plots to identify hotspots of frequency differences (Fig 2 and S5 Fig). Of particular interest were increased heterozygous normalised frequency absence (hetNFA) accompanied by increased homozygous normalised frequency presence (homNFP), as these SNPs are indicative of LOH events. Interestingly, Dataset C and Dataset E, both derived from isolates obtained from patients with autoimmune conditions, displayed the most pronounced elevated hetNFA and homNFP (Fig 2) resulting in the most negative LOHscore of these datasets (S2 Table). Conversely, isolates obtained from patients with oral and vaginal carriage (Dataset D and Dataset G respectively) exhibit levels of greater heterozygosity, while those of blood stream origin and those of a range of clinical origins (Fig 2, S5 and S6 Figs) exhibited a more balanced distribution (Fig 2 and S5 and  S6 Figs). We conclude therefore that these isolates were genetically stable and consistent across different body sites, although a number of LOH events had occurred in the past.

Discussion
The underlying CMC condition promotes susceptibility to Candida carriage and infection and we wished to determine whether such patients commonly exchanged Candida isolates or were   subject to serial colonisation. We therefore sequenced 42 C. albicans colonies for MLST from 6 unrelated CMC patients treated at the same hospital, and multiple colonies were typed from swabs taken from multiple body sites across multiple time points over a 17 month period (Table 1 and S1 Fig). A single stable DST was identified for all isolates of each individual patient and no patient yielded isolates with the same DST. These data demonstrate clonal and persistent DSTs, stable over time and identical between multiple infection sites with no shared origin or common source of infection for isolates from these patients. Isolates exhibited elevated azole resistance but no acquisition of resistance was notable within serial isolates from individual patients. However, changes in susceptibility for voriconazole observed at MIC50 but not MIC90 among isolates from patients 1 and 3 hint towards the potential presence of differentiated sub-populations genetically indistinguishable through MLST but that may harbour mutations conferring differences in azole resistance, these do not strongly represent an acquired resistance (Table 2). Additionally, we investigated the landscape of MLST genotypes to highlight differential frequency patterning of homozygous and heterozygous SNPs, these data suggest increased levels of past LOH events for persistent and azole resistant colonisers of patients with autoimmune disorders (Fig 2). The general stability and persistence of C. albicans populations has been highlighted previously in MLST studies using isolates taken at serial time points in a range of clinical settings [32][33][34]. The few MLST studies of isolates causative of CMC have taken this approach and have observed over time occasional MLST micro-variations and associated mutations at drug resistant loci conferring acquired azole drug resistant phenotypes [45,46]. Azoles, are the largest family of antifungal drugs and are popular due to their safe oral profiles, yet there are a number of mechanisms underlying antifungal resistance to azoles [52][53][54][55][56][57]. Our data identifies existing elevated azole resistance within persistent and clonal colonisers of CMC patients and is congruent with previous studies of CMC using MLST.
Genetic diversity among C. albicans populations can increase through rare recombination events during cryptic mating and as a result of associated concerted chromosome loss following transient aneuploidy [58,59]. However, more frequently genetic variation is driven within populations by LOH [28][29][30][31]. MLST consistently identifies LOH rather than strain replacement or larger scale micro-evolutionary changes as driving the generation of new strain types [30][31][32][33][34]. Recent genome wide studies have highlighted the importance of LOH in the genetic plasticity of C. albicans, using lab strains to dissect the role of abiotic stressors to induced LOH [29]. Indeed LOH plays a key role along with SNP and aneuploidy variations in the evolution of antifungal resistance among clinical isolates [31], and is coupled with mutations at several mediators of azole resistance including TAC1, CDR1/2, ERG11 and MRR1(MDR) [60,52]. Specifically these molecular mechanisms are the origin of acquired azole resistance among APECED CMC patients in two separate studies that identify MLST LOH mutations along with ERG11 and TAC1 derived overexpression of CDR1/2 [45,46].
It is well recognised that azole resistance develops with long-term use and is often seen in CMC patients who are on anti-fungal treatment for life. Patients with the PID CMC have an underlying genetic immune defect and are very rare, the incidence estimated to be 1:100000. To date, less than 230 patients with GOF-STAT1 mutations have been reported world-wide (Anne Puel, personal communication, manuscript in preparation). As the mutation was identified only recently [3], there is very little longitudinal data, and no data regarding Candida spp. strains and sensitivity in these patients. We report no consistent changes in MIC, but observe that MICs for azoles of most isolates were above the reported breakpoints suggesting they were already resistant at the beginning of the period of survey. Some minor changes in voriconazole were noted; in one case a small increase in MIC was seen whilst in another a decrease was observed. We conclude that no marked changes in antifungal drug sensitivity occurred over the 17 months of the survey.
To our knowledge, analysis of SNP frequencies reported here has not been applied previously to Candida spp. MLST data in this way. We have developed and implemented a novel and scalable analytical strategy that highlights the potential for MLST data to identify hallmarks of past LOH events from large scale MLST datasets. Interestingly these data suggest indicators of elevated levels of past LOH events are found among C. albicans isolates from CMC and autoimmune patients in comparison with oral, vaginal and blood stream carriage. Further studies implementing similar analyses with larger datasets may find these signatures of LOH to be synonymous with the increased azole resistance prevalent among autoimmune infection types.

Conclusions
MLST studies of Candida spp. have targeted specific clinical groupings such as nosocomial infections [35,36], periodontitis [40], recurrent bloodstream infection [41,42], cystic fibrosis [39] and genital candidiasis [44]. Although C. albicans MLST studies have repeatedly reported clade 1 isolates as the most commonly encountered clade and one associated with an increased ability to evolve and to cause infection, repeated attempts to draw further clade associations with clinical classes of infection and with virulence properties have been unsuccessful [43,61]. Similarly, although some geographical enrichment for C. albicans clades has been observed, MLST is limited in its ability to describe an in-depth evolutionary history of C. albicans geographic population structure [30,[62][63][64]. This is partly because MLST clades consider only a few hundred mutations in a reductive manner and use phylogenies that ignore the potentially informative properties of mutation frequency dynamics within and between populations. C. albicans exhibits largely clonal populations yet is capable of large scale genetic plasticity that can rapidly confer selective advantages such as resistance phenotypes. Population distribution analyses for C. albicans is also made more challenging in an era where individuals commonly travel internationally and may acquire strains that are not normally indigenous.
Our data demonstrates the frequency distribution of SNPs among C. albicans MLST datasets can provide useful biomarkers such as indicators of lost heterozygosity as a proxy for azole resistance. We detect this signal among isolates colonising CMC patients and suggest this is as a hallmark of selection for resistance in the absence of functional mucosal immunity. Similarly, genetic signatures of conserved heterozygosity may be evinced among commensal isolates of oral, vaginal and GI mucosae where mating and recombination may be more frequent. Using these approaches to molecular epidemiology of C. albicans isolates may provide improved methods to distinguish between clinical and geographical cohorts beyond using clades alone. Future studies implementing higher throughput genome wide sequencing, together with more sophisticated models for population dynamics of medically relevant yeasts [65][66][67], will improve our understanding of the population dynamics of fungal infection in diverse epidemiological arrangements and inform our approach to preventative and therapeutic treatments and strategies.  Table 2. Autoimmune datasets C and E (green), oral and vaginal carriage datasets D and G (gold), and larger individual isolate datasets B and F (blue) are included (left), and together with individual DSTs from data set C identified in the present study (black and numbered by DST). (TIF) S1  Table. NFD Summary statistics. Table of NFD summary statistics for individual DSTs identified in the present study (dataset C) and for datasets B-G. Loss of heterozygosity score is calculated as (LOHscore) = (hetNFA + homNFP)-(homNFA + homNFP). (DOCX)