Multilocus sequence analyses reveal extensive diversity and multiple origins of fluconazole resistance in Candida tropicalis from tropical China

Candida tropicalis is among the most prevalent human pathogenic yeast species, second only to C. albicans in certain geographic regions such as East Asia and Brazil. However, compared to C. albicans, relatively little is known about the patterns of genetic variation in C. tropicalis. This study analyzed the genetic diversity and relationships among isolates of C. tropicalis from the southern Chinese island of Hainan. A total of 116 isolates were obtained from seven geographic regions located across the Island. For each isolate, a total of 2677 bp from six gene loci were sequenced and 79 (2.96%) polymorphic nucleotide sites were found in our sample. Comparisons with strains reported from other parts of the world identified significant novel diversities in Hainan, including an average of six novel sequences (with a range 1 to 14) per locus and 80 novel diploid sequence types. Most of the genetic variation was found within individual strains and there was abundant evidence for gene flow among the seven geographic locations within Hainan. Interestingly, our analyses identified no significant correlation between the diploid sequence types at the six loci and fluconazole susceptibility, consistent with multiple origins of fluconazole resistance in the Hainan population of C. tropicalis.

Given the high prevalence of C. tropicalis in tropical regions and its increasing medical significance, it's important to understand the patterns of genetic variation of this yeast species in the tropics. Here in this study, we analyzed strains of C. tropicalis from the tropical island of Hainan in southern China. Hainan Island is located in China's southernmost province, Hainan Province (latitude 3°30′ -20°10′ N; longitude 108°15′ -120°15′ E). The island is separated from Mainland China by Qiongzhou Strait, ~40 km in width. Similar to those found in several other tropical regions, recent epidemiological analyses of yeasts in Hainan identified that C. tropicalis is among the most common yeast species from the oral cavities of asymptomatic hosts, second only to C. albicans 14,25 . Interestingly, a number of C. tropicalis strains from hosts in Hainan not exposed to fluconazole showed resistance and tolerance to the most common antifungal drug fluconazole 25 . At present, the patterns of genetic variation among strains and geographic populations and the genetic relationships between fluconazole-resistant and fluconazole-susceptible isolates from Hainan are unknown.
The objective of this study is to analyze the patterns of genetic variation of C. tropicalis from Hainan Island. Because Hainan is a tropical island with abundant organic matter conducive for the growth of C. tropicalis, we hypothesize that the favorable environmental conditions for C. tropicalis in Hainan may allow the generation and maintenance of abundant genetic variation of this species within Hainan. Furthermore, since human colonization of the Island is relatively recent in evolutionary terms and that there is frequent migration of people among different regions within the island, especially since the 1980s, we further hypothesize that populations of C. tropicalis from different geographic regions of the Island should be highly similar to each other. Finally, our recent study identified unexpected resistance and tolerance to fluconazole among isolates of C. tropicalis from this island, we were interested in whether there is any association between strain MLST genotype and drug resistance pattern.

Results
We successfully obtained DNA sequences from all six loci (ICL1, MDR1, SAPT2, SAPT4, XYR1 and ZWFa1) for all 116 isolates from 7 regions in Hainan (Tables 1 and 2). The genomic locations of these six loci are presented in Supplementary Table 1. As shown in this table, these six loci are situated on six different supercontigs corresponding to six chromosomal scaffolds. A total of 2677 bp from the six gene loci were sequenced for each of the isolates and 79 (2.96%) polymorphic nucleotide sites were found among our strains. All six gene fragments were found to be polymorphic among isolates within the Hainan population of C. tropicalis. The number of sequence types at each gene fragment ranged from 11 to 37, with a mean of 26.6 sequence types per gene fragment among the 116 isolates. Among the combined total of 124 sequence types at the six gene fragments, 36 were found to be new and had never been reported from other geographic regions. The combined analyses of sequence information from the six gene fragments identified a total of 94 diploid sequence types (DSTs) ( Table 2). Among these 94 DSTs, only 14 have been reported previously and the remaining 80 DSTs were completely new to the database. The details about the genetic variation at each of the six gene fragments are briefly described below.

ICL1.
Of the 447 aligned nucleotides of the ICL1 locus, 6 were found to be variable in the Hainan population of C. tropicalis. The 6 SNP sites generated a total of 8 genotypes (Table 3) among the 116 isolates from Hainan. Among these 8 genotypes, 7 (representing 114 isolates) have been reported from outside of Hainan while the remaining one (representing 2 isolates Ct_C14 and Ct_C20) is new, so far found only in Hainan. However, this novel genotype was clustered with known genotypes in the database from other geographic locations. The relationships among our genotypes and representatives of the unique genotypes at the ICL1 locus in the MLST database are shown in Supplementary Figure 1. The most frequent genotype at this locus, genotype 1, was found in 86 of the 116 isolates (74.1%).

SAPT2.
Of the 525 aligned nucleotides of the SAPT2 locus, 7 were found to be variable in the Hainan population of C. tropicalis. These 7 SNP sites generated a total of 11 genotypes at this locus (Table 3)

SAPT4.
Of the 390 aligned nucleotides of the SAPT4 locus, 18 were found to be variable in the Hainan population of C. tropicalis. These 18 SNP sites generated a total of 21 genotypes at this locus (Table 3) among the 116 isolates from Hainan. Among these 21 types, 16 (representing 107 isolates) have been previously reported from outside of Hainan while the remaining 5 (representing 9 isolates) are so far found only in Hainan. The 5 novel genotypes at this locus were all closely related to those present in the existing MLST database. The relationships among our genotypes and representatives of the unique genotypes at the SAPT4 locus in the MLST database are shown in Supplementary Figure 4. The most frequent genotype, genotype 7, was found in 28 isolates (24.1%) and the second most frequent genotype 17 was found in 26 isolates (22.4%).

XYR1.
Of the 370 aligned nucleotides of the XYR1 locus, 16 were found to be variable in the Hainan population of C. tropicalis. These 16 SNP sites generated a total of 33 genotypes at the locus (Table 3) among the 116 isolates from Hainan. Among these 33 types, 25 (representing 106 isolates) have been previously reported from outside of Hainan while the remaining 8 (representing 10 isolates) are so far found only in Hainan. The 8 novel genotypes contained both closely related (e.g. strain Ct_BT107) and moderately related (e.g. Ct_NK211 and Ct_1_HNHK) ones to those present in the existing MLST database. The relationships among our genotypes and representatives of the unique genotypes at the XYR1 locus in the MLST database are shown in Supplementary Figure 5. The most frequent genotype, genotype 60, was found in 26 isolates (22.4%) and the second most frequent genotype 9 was found in 15 isolates (12.9%).
The combined diploid sequence types based on all six gene fragments. The combined analyses of all six gene fragments identified that the 116 isolates from Hainan contained 94 diploid sequence types (DSTs). Even though a significant proportion of the alleles at each locus were shared between Hainan and outside of Hainan, relatively few combined DSTs at the six loci (14/94) were shared between the Hainan population and those from other parts of the world. The shared DSTs and the specific numbers of strains from within Hainan for each of these DSTs are DST149 (5 strains), DST331 (4 strains), DST346 (3 strains), DST394 (7 strains), DST427 (2 strains), DST430 (4 strains), DST432 (2 strains), DST 465 (2 strains), and DST490 (2 strains). The remaining 85 known DSTs have one strain each in our analyzed Hainan sample. Among the 14 shared DSTs between Hainan and outside of Hainan, nine were shared with those from Mainland China, two were shared with those from Taiwan, one each from Korea and the Netherlands. The remaining one shared DST was found in multiple countries/regions. The genetic relationships among the 116 isolates based on sequences at all six loci are shown in Fig. 1.
Evidence for extensive gene flow among geographic populations. The population genetic analyses of our samples based on nucleotide information at the 79 polymorphic nucleotide sites from the six gene fragments revealed that the majority (75%) of the genetic variation was found within individual strains ( Table 4). The second most important contributor was the differences among individuals within individual geographic populations that contributed 23% of the total genetic variation. In contrast, the geographic separations among local and regional populations contributed relatively little to the overall patterns of genetic variation (Table 4). In addition, multiple DSTs were shared among regions within Hainan (Table 2). Together, the presence of shared DSTs and the lack of genetic differentiation among geographic populations of C. tropicalis within Hainan are consistent with frequent gene flow among these regions in Hainan.
Aside from conducting the overall AMOVA, we also obtained the F ST values between pairs of geographical populations. However, to ensure the robustness of the results, the populations of less than five samples were removed in the pairwise comparisons. This analysis identified no statistically significant differentiation between any pairs of geographic populations. The lowest F ST value (0.011) was found between Haikou and Lingshui while the highest (0.053) was between Wenchang and Sanya ( Table 5). The result from the Mantel test is shown in Fig. 2. The test showed that there was little correlation between genetic distance and geographical distance among the analyzed populations (P = 0.390), consistent with extensive gene flow among the geographic populations. eBURST analysis. We used the eBURST program to identify genotype clusters 26 . In this analysis, we applied the default setting of identical alleles at five of the six loci for genotype cluster identification and 1000 re-samplings for confidence estimates through bootstrapping. Among the 502 DSTs from the C. tropicalis MLST database, the eBURST analysis found 55 clusters and 214 solitary DSTs, known as singletons. Of the 94 DSTs representing the 116 isolates in our sample from Hainan, 50 DSTs were grouped into 20 clusters (clusters 1, 2, 4, 5, 7, 10,11,14,18,23,32,34,44,46,47,48,49, 50, 51 and 53), and 44 DSTs were classified as singletons (i.e. not belonging to any obvious clusters) ( Table 2). Specifically, cluster 1 had a total of 31 DSTs, including 2 from this study. Cluster 2 had a total of 21 DSTs, with 1 from this study. Cluster 3 had a total of 14 DSTs, including one from this study. Cluster 4 had a total of 17 DSTs, including 14 from this study. The 14 Hainan DSTs in cluster 4 contained a total of 28 isolates distributed in all seven geographic regions. Among these 14 DSTs, DST 394 was represented by 7 isolates from three regions in Hainan (Haikou, Lingshui, and Wenchang) and occupied the central position of this cluster. DSTs 331 and 430 had four strains each with each DST distributed in three regions respectively (Table 2).
Another cluster with multiple DSTs and multiple strains from Hainan was cluster 7. This cluster had a total of 11 DSTs, including 8 from this study. Among these eight DSTs, one (DST 149) was at the center of this cluster and it contained five isolates from four regions in Hainan (Baoting, Dongfang, Haikou, and Lingshui). Two isolates in this cluster have intermediate resistance to fluconazole (see also below). Cluster 10 had a total of 8 DSTs, including 3 from this study with isolate C30 representing DST 333 at the center of this cluster. Cluster 11 had a total of 7 DSTs, including one from this study. Cluster 14 had a total of 5 DSTs, including one from this study. Clusters 23 and 34 each had 3 DSTs while Clusters 44, 46 and 53 each had 2 DSTs. Each of these five clusters (i.e. clusters 23, 34, 44, 46, and 53) contained only one strain each from Hainan. Clusters 18, 47, 48, 49, 50 and 51 are new clusters added from this study and they comprised DSTs identified only in the present work. Among the singletons, isolates Ct_HY8, Ct_C26P, and Ct_C27 showed the largest genetic distances from the other DSTs. Overall, results from the eBURST analyses were consistent with the UPGMA tree generated from multilocus sequence data and demonstrated that the genotypes from Hainan were dispersed in most of the known clusters that included strains from other geographic areas (Fig. 1).  Fig. 1). Four (DST 149, 331, 441, and 500) of the 9 DSTs belong to 3 genotype clusters while the remaining five were singletons (Table 2). Among these nine DSTs, DST 331 and DST 149 also contained fluconazole susceptible strains. Specifically, fluconazole-susceptible isolates Ct_BT20, Ct_C2, Ct_C29, and Ct_DFR77 shared the same DST 149 with fluconazole dose-dependent isolate Ct_30_HNHK (Fig. 1). Similarly, fluconazole-susceptible isolate Ct_15_HNHK, Ct_19_HNSY and Ct_DFG118 shared the same DST331 with the fluconazole-resistant isolate Ct_33_HNHK (Fig. 1). These dose-dependent/resistant strains were distributed broadly among the clusters and across the UPGMA tree. Taken together, our results suggest that fluconazole resistance among Hainan C. tropicalis isolates most likely originated multiple times through independent mutations.

Fluconazole susceptibility analysis.
The independent origin hypothesis for fluconazole resistance is also supported by the Mantel test results (Fig. 3). Specifically, our analyses showed that pairwise strain genetic distance and fluconazole susceptibility differences were not correlated with each other (p = 0.238). This result is consistent with not only the independent     origins of fluconazole resistance among the strains analyzed here but also the hypothesis that many, if not all, genotypes or genotype clusters are capable of developing fluconazole resistance.

Discussion
This study analyzed the patterns of DNA sequence variation at six nuclear gene loci and compared the patterns of variation with those in the MLST database representing strains from other geographic regions. Our analyses revealed extensive novel sequence polymorphisms not only at the individual locus level but also more noticeably at the combined genotype level. Interestingly, most genetic variations were found within individual strains and among strains within the same geographic populations. Despite the extensive genetic variations within individual populations, we found no evidence of genetic differentiations among the analyzed geographic populations within Hainan, consistent with frequent gene flow among these geographic regions. In addition, even though evidence for clonal dispersal and expansion were found in our samples for fluconazole-susceptible genotypes, there was no evidence of clonal dispersal for fluconazole-resistant isolates. Each of the nine fluconazole-resistant or dose-dependent isolates belonged to a different multilocus DST. Among these nine DSTs, three also had representatives of fluconazole-susceptible strains, consistent with the independent mutations causing fluconazole resistance among our nine strains. However, the lack of evidence for clonal expansion of fluconazole-resistant genotypes in Hainan doesn't mean that such clonal expansions do not exist at all in Hainan. Indeed, the inclusion of more strains from Hainan, especially those that are resistant to fluconazole, might reveal clonal expansion of fluconazole-resistant C. tropicalis in Hainan. Below we discuss the implications of our results. Our study is the first genetic analysis of C. tropicalis from tropical Asia. The genetic variations observed here expand our understanding of this organism in nature. However, even though we identified abundant genetic variations, we believe that there are likely additional genetic variations in Hainan and in other parts of tropical Asia. For example, there might be bias in the efficiency of PCR amplification between the two alleles at each heterozygous locus (e.g. due to mutations in primer sequences between the two alleles) that could have resulted in underestimates of heterozygosity within individual strains. Furthermore, our samples were all from one ecological niche, the oral cavities of humans. C. tropicalis has been found in a diversity of other environments, including organically enriched soil and aquatic environments 17,27 , and animals such as wild birds 28 , horses 29 , rheas 30 as well as in tortoises and sea turtles 31 . Hainan Island and the tropics in general are rich in organic compounds and wild animals. Thus, it's possible that these ecological niches contain additional genetic diversity of C. tropicalis and one or several of these niches may represent the natural reservoirs of C. tropicalis for humans. A population genetic comparison of C. tropicalis from these environments with our data here could reveal the relationships between these populations and help identify the environmental reservoir(s) of C. tropicalis in Hainan (as well as elsewhere).
The potential existence of an environmental reservoir of C. tropicalis for humans is also supported by our data. Specifically, none of the hosts had taken any fluconazole or other triazole drugs. However, nine C. tropicalis isolates from nine different hosts showed intermediate susceptibility or were resistant to fluconazole. We believe the likely source(s) for the observed fluconazole resistance in C. tropicalis is natural or human-made environments in Hainan. As shown recently in another opportunistic human fungal pathogen Aspergillus fumigatus, the application of agricultural fungicides was most likely responsible for the emergence of drug-resistant strains in human populations for that filamentous fungus [32][33][34] . A similar process could have happened here whereby drug-resistant strains selected in agricultural fields with heavy applications of triazole fungicide were passed onto human hosts 35 . Alternatively, other types of settings, e.g. human-made products such as paint or human-associated environments such as house stuff, where antifungal agents are applied could also select for drug-resistant fungal strains. Targeted samplings of agriculture fields or other environments where triazole fungicides are commonly used could help reveal the potential sources of fluconazole-resistance in C. tropicalis in Hainan.
Regardless of the potential sources, such environmentally induced drug-resistance isolates pose a significant threat to human and animal health. This can be especially troublesome for tropical regions where C. tropicalis candidaemia are of particular concern 3 . Patients infected with triazole-resistant C. tropicalis are often associated with high mortality 3,4,9,17,29 . In this study, all the C. tropicalis strains were isolated from oral cavities of local healthy people or in-patients in hospitals. The oral cavity is a significant niche of the human microbiome and a gateway for the microbiota in many other human body sites. A drug-resistant strain from the oral cavity could be passed on to other body sites, potentially causing untreatable invasive infections.
The six gene fragments analyzed here showed abundant genetic variation within and among strains from Hainan ( Table 3). The number of polymorphic sites (79 of 116 isolates) in the present study is slightly higher than that (28 of 58 isolates) from Beijing, China, but lower than those in other places. For example, the number of polymorphic sites from Brazil was 154 among 61 isolates (Table 3). In contrast, except for the XYR1 locus, the ratios for the number of genotypes per polymorphic nucleotide site in our samples are higher than previously reported for geographic populations including the US and European countries and Brazil (Table 3). These ratios were slightly different for the Beijing sample where extremely large ratios were found, e.g. one polymorphic nucleotide site allowed the identification of 15 genotypes at the XYR1 locus 22 . Specifically, if this polymorphic site is biallelic (i.e. containing two alternative bases), a maximum of 3 genotypes should be found (two homozygotes and one heterozygote). With three alternative bases, a maximum of 9 genotypes would be expected in a diploid organism at this site. Only with all four alternative bases at this site in the Beijing sample would we expect a maximum of 16 genotypes based on one polymorphic nucleotide site and assume all possible associations among the four bases at this site in this diploid organism.
Interesting, the most frequent genotypes in our sample at ICL1 (genotype 1), SAPT2 (genotype 3) and SAPT4 (genotype 7) were also the most frequent in the global population analyzed so far. In contrast, the most frequent genotype at the other three loci MDR1 (genotype 9), XYR1 (genotype 60), and ZWFa1 (genotype 22) in Hainan were not the most frequent in the global population analyzed so far. Furthermore, the most frequent genotype at Scientific RepoRts | 7:42537 | DOI: 10.1038/srep42537 five of the six loci (except ICL1) in our samples also differed from that reported from Beijing. China. Specifically, the most frequent genotypes at MDR1, SAPT2, SAPT4, XYR1, and ZWFa1 from the Beijing sample were genotype 7, 4, 17, 2 and 7 respectively 22 . Together, these data suggested that the Hainan C. tropicalis population contained abundant and novel genetic variation at the assayed loci.
The observed novel genetic variation was found not only at the individual locus level. At the combined DST level from all six sequenced gene fragments, only 14 of the 94 DSTs from the 116 strains were shared with those from other geographic areas while the remaining 80 were novel to the C. tropicalis MLST community database. Among the 14 shared DSTs between Hainan and those from outside of Hainan, 13 19 , and one (DST394) from Korea. Only one DST (DST 23) was shared only with a strain from outside of Asia (the Netherlands) 18 . DST 23 is a singleton genotype in the MLST database and the carrier of this strain in Hainan was an 11 year-old schoolboy in Lingshui along the east coast of Hainan. He had no travel history to the Netherlands. Interestingly, DST203 found on both Hainan Island and Taiwan Island has also been found in Brazil 21 . Together, these results suggest the potential of long distance dispersal for C. tropicalis among geographic regions, likely through humans or human activities, including importing and exporting of foods colonized by C. tropicalis.
Among these shared DSTs, one (DST149) is worthy of special mention. DST 149 was represented by fluconazole resistant strains in both Taiwan Island and Hainan Island 36,37 . Furthermore, DST 149 was the main fluconazole-resistant DST in Taiwan from 1999-2006 36 . Thus, instead of independent origins, it's possible that the Hainan fluconazole-resistant strain of C. tropicalis could have originated in Taiwan and dispersed to Hainan (or vice versa). However, more strains need to be investigated from Southeast Asia, including from southern China and the Philippines, before the conclusion about a Taiwan-Hainan transfer of this specific fluconazole-resistant genotype could be made.
While our results showed no genetic clustering of fluconazole resistant isolates, possibly due to the high-level genotype diversity and the relatively limited sampling in our study, several studies of similar or smaller sample sizes found evidence of genetic clustering of azole-resistant isolates. For example, both Chou et al. 19 and Li et al. 35 found clonal cluster 2 (containing DST140 and DST98) was enriched with isolates with resistance or trailing growth in the presence of fluconazole. A recent report by Wang et al. 38 showed that 23 of the 30 azole-resistant isolates of C. tropicalis from Shanghai belonged to four DSTs (DST 376, 505, 506, and 507) of the same genetic cluster. However, though four DSTs (DST 149, 331, 346, and 394) were shared between our sample and those in the recent Wang et al. study, none of these four DSTs had fluconazole-resistant isolates in both the Shanghai and Hainan samples. Similarly, Chen et al. 20 found that DST 164 was associated with a high MIC to flucytosine. Clonal expansion of flucytosine resistance in C. tropicalis has also been reported from Paris, France 39 . Furthermore, Li et al. 36 described several single locus genotypes (genotype #3 of ICL1, #9 of MDR1, #1 of SAPT2, #3,6 and 10 of SAPT4, #48 of XYR1 and #7 of ZWF1a) associated with low MICs to fluconazole. Different from these studies, we found no isolate with these seven DSTs (DST 98,140, 164, 376, 505, 506, and 507) in our sample. In addition, genotype #3 of ICL1, #1 of SAPT2, and #7 of ZWF1a were found associated with both fluconazole-susceptible and fluconazole-resistant isolates in Hainan (Table 2 and Fig. 2). Together, these results suggest both shared and unique features among geographic populations of C. tropicalis.
At present, the molecular mechanisms of resistance among our strains are not known. Though we found evidence for multiple independent origins of the fluconazole-resistant strains in our samples, their mechanisms of resistance could be very similar or even identical. Molecular studies of triazole-resistant strains of Candida have shown three common types of mechanisms: (i) mutation of the target gene ERG11 (or CYP51) leading to reduced affinity of the drugs to the target enzyme; (ii) over-expression of ERG11; and (iii) over-expression of efflux pumps 40 . For example, Barchiesi et al. 41 found that over expressions of the major facilitator gene MDR1 and the ATP-binding cassette transporter CDR1 were responsible for fluconazole resistance among independently selected fluconazole-resistant mutants of C. tropicalis strain ATCC750. We would also like to mention that our conclusion about the lack of evidence for clonal expansion of fluconazole resistant C. tropicalis in Hainan is specific to our sample analyzed here. Indeed, it's entirely possible that samples from patients and clinics where the use of fluconazole or other triazoles may be prevalent would likely show evidence of such clonal expansion, similar to what has been found in other studies.
In conclusion, the establishment of a MLST database for C. tropicalis has facilitated comparisons of strains and populations from different laboratories and different geographic regions around the world. Our analyses identified high and novel genetic diversity of C. tropicalis in Hainan samples and revealed no evidence of genetic differentiation among the regional population. Our combined analyses of MLST genotype and fluconazole resistance suggested multiple independent origins of fluconazole resistant and dose-dependent strains in Hainan. Our genotypic comparisons revealed evidence for genotype sharing between strains from Hainan Island and those from other regions including Mainland China, Taiwan Island, the Netherlands, and Brazil. The results and data presented here not only provide an understanding of C. tropicalis in tropical Asia but also expand the database for future studies of C. tropicalis in other regions. Our study also calls for greater effects in analyzing strains from both clinics and natural environments in the tropics in order to further understand the origins and distributions of fluconazole-resistant genotypes in these regions.

Materials and Methods
Isolates. All the samples included in this study were collected from the oral cavities of either healthy people or hospitalized patients (Table 1). All experimental protocols for sampling were approved by Hainan Medical College and informed consent was obtained from all hosts. The sampling procedures were carried out in accordance with relevant guidelines and regulations. The hosts were from seven different cities/municipalities on Hainan Island. The geographical coordinates for the biggest city in each of the seven regions are presented in Table 1 of 116 isolates of C. tropicalis were obtained in this study, with 23 from health people and 93 from hospitalized patients. However, none of the hosts, including the hospitalized patients, had clinical symptoms of oral thrash at the time of sampling. The procedures for obtaining and identifying the species status of these isolates were described in previous studies 14,22,42 . The detailed information for each isolate is described in Table 1. The yeast pure cultures were maintained on Sabouraud dextrose broth containing 30% glycerol in − 80 °C freezer until use.
DNA extraction and genotyping. The total genomic DNA of the isolates was extracted using a Yeast DNA miniprep protocol described previously 43 . The DNA concentrations were estimated with a spectrophotometer absorbance at 260 nm and diluted to 10 ng/ml. Each PCR amplification reaction was carried out in a final volume of 50 μ l that consisted of 25 μ l of 2x Premix Taq (Tiangen), 21 μ l of dH 2 O, 2 μ l of template DNA, and 2 μ l of the forward/reverse primers. The primer sequences and amplification conditions for obtaining sequence information at the six loci (ICL1, MDR1, SAPT2, SAPT4, XYR1 and ZWFa1) followed those described previously 18 . As shown in Supplementary Table 1, these six gene fragments are located on six different chromosomal scaffolds. The amplified fragments were purified using a PCR purification kit (Qiagen) according to the manufacturer's instructions. Both the forward and reverse strands of the purified DNA fragments were sequenced at the Public Research Laboratory of Hainan Medical College using the same primers as those used in the initial PCR amplification. DNA sequencing was performed with Chromas 2.13 software 44 . Sequence type identification at each locus and at the six combined loci. For each locus in each strain, sequence chromatograms from the two directions were aligned using the DNASTAR software (http://www. dnastar.com) to obtain a combined consensus sequence for the locus. Because C. tropicalis is a diploid organism, heterozygous nucleotide sites are expected for our isolates. To ensure that all heterozygous sites are accounted for, all sequence chromatograms were manually inspected. These sequences were then compared with the existing sequences at the C. tropicalis MLST sequence type database (http://pubmlst.org/ctropicalis/) to obtain a sequence profile for each locus for each strain and a combined diploid sequence type (DST) for each strain based on the sequence profiles at all six gene fragments. The new sequence profiles at both the individual locus and the combined six loci that were absent in the original database were respectively assigned new numbers ( Table 1). The sequences for all strains at the six loci have been deposited in the C. tropicalis MLST database.
Relationships among sequence types at each locus and at the combined six loci. To analyze the relationships among our sequences and between ours and those already in the database, we downloaded all the representative sequences for each locus from the C. tropicalis MLST database and aligned them together with ours. The sequence relationships at each locus and strain relationships based on sequences at the combined 6 loci were determined through cluster analysis using UPGMA (unweighted pair group method using their arithmetic averages) of the MEGA software 45 . The putative clonal clusters showing the likely ancestor-descendant relationships among the isolates were identified with the eBURST package, v3.0 (http://eburst.mlst.net) 26 .
Geographic patterns of DNA sequence variation. Since C. tropicalis is a diploid yeast and heterozygous nucleotides sites have been frequently found, our analyses of the geographic patterns of DNA sequence variation followed those for diploids. Here, each polymorphic nucleotide site is treated as an informative site and alternative nucleotides at each locus as different alleles. To infer the patterns of genetic variation, the sequences were imported into the computer program GenAlEx 6.5 46 . The population genetic parameters such as the number of polymorphic nucleotide sites within each gene fragment and allelic diversity in each population were estimated. In addition, GenAlEx 6.5 was used to calculate the pairwise population F ST values and determine the potential correlation between genetic and geographical distances (Mantel test). The analysis of molecular variance (AMOVA) was performed to estimate the relative contributions of geographic separation to the overall genetic variation.
Relationship between fluconazole susceptibility and MLST genotype relatedness. All strains were tested for their susceptibility to fluconazole. The details of antifungal susceptibility testing were described in a previous study 25 . The putative association between genotypes and fluconazole susceptibilities was examined using GenAlEx 6.5. Specifically, we obtained and compared two distance matrices. In one matrix, we obtained the genetic distances between all strain pairs (116 × 115/2 = 6670 pairwise distances) based on the nucleotides at all 79 polymorphic sites. In the second matrix, we obtained the absolute differences in the size (in mm) of the "zone of inhibition" between all strain pairs (also 6670 pairwise distances). A non-parametric Mantel test was used to investigate whether there was a significant correlation between genetic distance and fluconazole susceptibility difference in our sample. In addition, to help visualize the relationships between genetic relationship and fluconazole susceptibility, we also marked S, I, and R respectively beside strains that were susceptible, intermediate, and resistant to fluconazole onto the UPGMA graph representing the genetic relationships among the 116 strains based on sequence information from the six loci.