Genetic Diversity among Clinical Isolates of Candida glabrata Analyzed by Randomly Amplified Polymorphic DNA and Multilocus Enzyme Electrophoresis Analyses

The commensal yeast Candida glabrata is frequently isolated from the human digestive tract but is also considered an opportunistic pathogen. Depending on the site of infection or the disease, C. glabrata is often the second- or third-most-common cause of candidiasis after Candida albicans. C. glabrata is associated with a high mortality rate and is innately resistant to azole antifungal agents (8, 26, 36, 39).

Multilocus enzyme electrophoresis (MLEE) has been used to assess genetic diversity, gene flow, and population or genotypic structure, besides being useful for typing, systematic, and epidemiological studies of fungi (35, 38). Randomly amplified polymorphism DNA (RAPD) analysis is a genetic method extensively used to analyze genetic relationships and intra- and interspecific genetic diversity (12). RAPD and MLEE data can provide valid measures of genetic distances or genetic similitude, gene flow, and mode of reproduction (27). Although there are several studies based on MLEE (3, 4, 17, 27, 28) and on genotypic methods (5, 10, 14, 20, 27, 31) describing intraspecific diversity, population structure, and mode of reproduction for C. albicans, few reports exist for other Candida species.

Several studies of molecular typing to relate epidemiological information with molecular fingerprints have been made with C. glabrata (2, 8, 40). For C. glabrata only two papers exist that report quantitative measurements of intraspecific genetic diversity, genetic distance between strains, and mode of reproduction in nature (6, 19).

We report here the diversity and genetic relationship of 47 clinical and reference strains of C. glabrata assessed through MLEE and RAPD techniques. Particularly, we analyze whether the isolates are associated with their geographical origin, the clinical disorder, and their antifungal susceptibility.

A total of 47 Candida glabrata isolates were obtained from four microbiological laboratories from the cities of Mexico City, Guadalajara, Monterrey, and Guanajuato in Mexico. Strains from the vagina (n = 23), expectorate (n = 9), oral cavity (n = 8), urine (n = 3), and pharynx (n = 1) were included. The strains were obtained from patients with vaginitis (n = 19), respiratory infections (n = 9), cancer (n = 5), AIDS (n = 3), diabetes (n = 2), or transplant (n = 1) or from healthy individuals (n = 5). No clinical data were obtained for three C. glabrata isolates. Strains J931010, 1480.41, LA817, and RA732 were used as a reference (19). The strains were identified with an ID 32 C kit (BioMerieux SA, Marcy-l'Etoile, France) and a RAPD-based procedure previously described (1).

All isolates were tested for their susceptibilities to fluconazole, itraconazole, amphotericin B, and nystatin and for minimal inhibition concentrations (MICs) of these drugs by the microdilution assay according to the National Committee for Clinical Laboratory Standards protocol, using M27-A with RPMI-1640 medium (Sigma Chemical Co., St. Louis, Mo.) and incubation at 35°C for 48 h (22). Endpoint readings were recorded with a microplate reader (Labsystem, Helsinki, Finland).

For DNA extraction, the yeasts were grown overnight in YPD broth (1% yeast extract, 2% peptone, 2% dextrose) at 37°C under shaking at 200 rpm. DNA was extracted according to a previously described protocol (18). DNA concentrations and A₂₆₀/A₂₈₀ ratios were determined with a spectrophotometer (Perkin-Elmer Lambda 1A). An A₂₆₀/A₂₈₀ ratio of 1.8 to 2.1 was considered acceptable.

The primers OPE-18, OPE-04, and OPA-18 used for RAPD analysis were previously proposed (19), and two additional primers, EDP-4 (5′-AAGAGCCCTT-3′) and EDP-8 (5′-AAGCCTCGCT-3′), were included (Gibco BRL, Gaithersburg, Md.). DNA fragments were amplified with a DNA Thermal Cycler (Perkin-Elmer 9600). RAPD analysis was performed using a previously described method (19). Only one modification was done to the PCR protocol, which was to amplify by 40 cycles rather than 45 cycles. Reaction products were analyzed by electrophoresis through 1.3% (wt/vol) agarose (Gibco BRL) gel slabs (14 cm by 11 cm by 6 mm), and the relative molecular weight of each band was estimated by Sigma Gel v1.0 software (Jandel Scientific, Chicago, Ill.). Multiple samples were tested on different days to confirm repeatability of the method.

The enzymatic extracts to MLEE were obtained from yeast isolates grown overnight at 37°C on YPD broth. Yeasts were harvested by centrifugation of broth cultures (20,000 × g for 10 min at 4°C). Harvested cells were washed in cold deionized sterile water. They were resuspended in 2.5 ml of 0.1 M Tris hydrochloride (pH 7.0) buffer, and 7 g of glass beads (diameters, 425 to 600 μm) per each gram of wet cells was added. The suspended yeasts were mechanically disrupted by 1-min Vortex pulses in an ice bath. The resulting lysates were centrifuged for 10 min at 20,000 × g at 4°C. The supernatants were aliquoted in 1.5 microtubes and stored at −80°C. Starch gel electrophoresis and enzymatic assays were performed according to methods previously described (32). The activities of the following enzymes were analyzed: alcohol dehydrogenase (ADH) (EC 1.1.1.1), malate dehydrogenase (MDH) (EC 1.1.1.37), isocitrate dehydrogenase (IDH) (EC 1.1.1.42), glucose 6-phosphate dehydrogenase (G6P) (EC 1.1.1.49), hexokinase (HEX) (EC 2.7.1.1), peptidase 1 (PEP1; substrate, Val-Leu) (EC 3.4.13), peptidase 2 (PEP2; substrate, Leu-Gly-Gly) (EC 3.4.13), peptidase 3 (PEP3; substrate, Phe-Pro) (EC 3.4.13), mannose-6-phosphate isomerase (MPI) (EC 5.3.1.8), and glucose-6-phosphate isomerase (PGI) (EC 5.3.1.9). All enzymes were electrophoresed in buffer system A, and enzyme mobilities were determined by staining for specific enzyme activity (32). All enzymes migrated anodally, and each unique combination of electrophoretic migration patterns was designated as the electrophoretic type (ET).

For RAPD data analysis, the amplification products were recorded as binary data, presence (1) and absence (0). Pairwise genetic similarities among isolates were calculated from these data using Jaccard's coefficient (16). The genetic relationships among isolates were established by cluster and ordination analyses performed on the matrix of genetic similarities. Cluster analysis was performed by mean of the unweighted paired group method using arithmetic average (UPGMA) (33). The distortion of the inferred tree was estimated by means of the cophenetic correlation coefficient (CCCr) (34) with the nonparametric Mantel test (21) and the best-cut test (37). The genetic diversity by primer and isolate was estimated using Shannon's index, H_o = −Σ P_i · log₂P_i, where P_i is the frequency of RAPD phenotypes (9). All analyses were made with the NTSYS-PC software, version 2.02j (29), and PHYLIP version 3.6 software (phylogeny inference package, version 3.6a3; J. Felsenstein, Department of Genome Sciences, University of Washington, Seattle, Wash.).

The genetic relationships of only 44 isolates (strains with complete information) were measured using Euclidean metric distances calculated among all possible pairwise combinations (11, 15). The matrix was used with analysis of molecular variance (AMOVA) (7) to estimate variance components for RAPD phenotypes of the isolates grouped according to their geographic origin, body location, and type of patient. The interpopulation distance matrix for all populations was used to calculate a dendrogram based on the UPGMA method.

The MLEE data analysis was done based on an allelic frequencies matrix, and then a pairwise genetic distance matrix was estimated through Nei's coefficient (25). Cluster analysis was performed on the genetic distance matrix with UPGMA. Genetic diversity measured as heterozygosity for a locus was calculated as h = 1 − Σ x_i² [n/(n-1)], where x_i is the frequency of the ith allele at the locus, n is the number of isolates or ETs in the sample, and n/(n-1) is a factor correction for small samples (24). The distortion of the dendrogram was estimated as RAPD data.

All strains showed high susceptibility to fluconazole (93.6%), only three strains were susceptible-dose dependent (S-DD) with a MIC between 16 and 32 μg/ml. An extended itraconazole resistance (66%) and S-DD (34%) was found in all strains, with no clearly susceptible strains. The amphotericin B and nystatin MICs were 0.25 to 2 μg/ml and 2 to 8 μg/ml, respectively.

A total of 49 reproducible RAPD markers, 19 monomorphic and 30 polymorphic, in the range of 400 to 2,500 bp, were generated (Fig. 1). The genetic diversity among isolates, estimated by Shannon's index (H_o), varied from 0.233 to 0.437; the average genetic diversity was 0.372. The observed genetic similarity varied from 0.56 (isolates CGL19-CGL9, and CGL28-CGL19) to 1.0 (many isolates); the average genetic similarity was 0.76. Three groups were recognized on the dendrogram based on the best-cut test (0.83) (Fig. 2). The dendrogram (CCCr = 0.923; P = 0.0020) showed no relationship among the clustered isolates and their geographical origin, antifungal susceptibility, body location, or the clinical disorder of the patients (Fig. 2).

The AMOVA Φ_ST values estimated for each pair of populations revealed no differentiation among them. The highest Φ_ST value estimated was 0.130 between Nuevo León and Jalisco; however, neither pair was significantly different from zero, which indicates that no subdivision existed among populations, rejecting the hypothesis of Φ_ST ≠ 0. Furthermore, the results indicate that most of the genetic variation was found within populations (86 to 100%). Similar results were found when body location and type of patient were analyzed. Alternatively, an AMOVA was performed that included all strains to verify the lack of subdivisions among populations of different geographic and clinical (type of patient, body location, and antifungal susceptibilities) origins. For this, all strains were considered as one big population, and clinical and geographic locations were included as subpopulations. The genetic variation was significantly higher within populations than among populations; however, the Φ_ST values found were not significantly different from zero (Table 1). The dendrogram, built with the interpopulation distance (pairwise Φ_ST matrix), did not reveal any relationship between populations and their geographic origins (Fig. 3). We expected populations to be grouped by their geographic proximity; however, no relationships were found.

A total of 11 loci were obtained, two polymorphic (HEX-2 and PEP3) and nine monomorphic (ADH, MDH, IDH, G6P, HEX-1, PEP1, PEP2, MPI, and PGI). Three distinct ETs were identified (1, 2, and 3), containing 34, 8, and 5 strains, respectively, and representing all the multilocus genotypes found (Table 2). The genetic diversity measured as average heterozygosity (h) was 0.055. The genetic distance observed varied from 0.0 to 0.05 among isolates; the average genetic distance was 0.01. Only one group was recognized on the dendrogram (CCCr = 0.985; P = 0.0020) based on the best-cut test (0.13), and no relationship among the dendrogram subgroups and the geographical origin, antifungal susceptibility, body location, or clinical disorder was observed (data not shown).

Genetic variation analysis based on RAPD allows proper genetic diversity estimation due to its capacity to generate random markers from the entire genome. The estimation of genetic variation by two different methods, as was done in this study with MLEE and RAPD, permits better estimations of genetic diversity from any species.

Recently, the genetic diversity of C. glabrata isolates from France was estimated through 33 enzymatic loci using MLEE (6). The value obtained was 0.11, which was twice that obtained in this work (0.055). However, genetic diversity is influenced by the number of isolates and of loci examined (23, 24); particularly, genetic diversity has been negatively correlated with the number of loci investigated (13). Thus, we thought that the average genetic diversity of C. glabrata might have been overestimated because we studied only 11 loci; however, in spite of loci examined, it was lower than that previously reported for C. glabrata (6). This difference indicates that the genetic diversity of C. glabrata is broad and that it could be a consequence of its adaptability to particular clinical characteristics. The genetic diversity value obtained for C. glabrata in this study was low compared with those reported by other authors for the diploid species C. albicans (0.13 [4], 0.17 [28], 0.35 [3], 0.132 and 0.155 [41]). This could be attributed to the haploid state of C. glabrata, as has been suggested (6).

Regarding the genetic diversity estimated by RAPD data, there are no Candida spp. data reported in the scientific literature. However, genetic diversity data obtained by our group from work done with C. tropicalis (0.351), C. guilliermondii (0.274), and C. lusitaniae (0.433) strains (unpublished data) were similar to the present results with C. glabrata (0.372). In this study, it was not possible to associate the genetic diversity values with the mode of reproduction.

The small genetic differentiation observed among C. glabrata isolates by MLEE indicates that the isolates are genetically homogenous. In this sense, it is very illustrative that the foreign isolates (J931010, LA817, RA732, and 1480.41) had enzymatic profiles identical to those depicted for the Mexican isolates. On the other hand, RAPD clusters were more variable than the MLEE groups; however, RAPD clusters did not show association with the MLEE genotypes. This lack of association was confirmed by the low significant correlation (CCCr = −0.09448, Mantel test) between the two half-matrices of MLEE and RAPD. These results do not agree with those previously reported (6).

The obtained RAPD fingerprints showed many similarities among strains; strain J931010 from Germany and strain LA817 from the United States displayed very similar band patterns. In other cases, identical fingerprints were obtained with strains 1480.41 and CGL29 from the United States and Mexico, respectively. These findings revealed that even when comparing strains from distant geographical origins, C. glabrata strains showed no differentiation at the level of RAPD genotypes.

The observed pairwise Φ_ST values were not significantly different from zero, indicating no differentiation either among populations of the four geographic origins or among populations organized according to clinical characteristics.

These results do not agree with those reporting differentiation among C. glabrata isolates from different geographic origins. For example, Lockhart et al. (19) using RAPD markers found differentiation among different geographic populations; however, the associations were based on the S_AB similarity coefficient, which is not an appropriate index for establishing the differentiation level. In addition, an arbitrary cutoff value of 0.49 was used improperly to support the differences among populations from diverse locations. On the other hand, through an analysis of the mitochondrion-encoded cytochrome c oxidase subunit 2 gene (COX2), Sanson and Briones found differentiation between isolates from Brazil and the United States based on only two nucleotide changes (30). However, this result, based on a gene that evolves approximately 10 times faster than nuclear genes, lacks an appropriate support to validate these changes with statistical significance. Although de Meeûs et al. found a low differentiation level (F_ST = 0.089) between C. glabrata strains from Paris and Montpellier, France, with MLEE (6), there are no other reports with RAPD or MLEE that associate C. glabrata isolates with their geographic origin or the clinical characteristics.

In summary, our MLEE and RAPD results do not show a significant differentiation among the studied C. glabrata isolates, nor do they show an association with their geographic origin or the clinical characteristics.

• Open in new tab
• Download powerpoint

FIG. 1.

RAPD patterns obtained with the five primers employed. (A) Primer OPE-18; (B) primer OPE-04; (C) primer OPA-18; (D) primer EDP-4; (E) primer EDP-8. Lane M, molecular size marker in base pairs (phage λ DNA digested with PstI); lanes CGL20 to CGL3, strains of C. glabrata.

• Open in new tab
• Download powerpoint

FIG. 2.

Dendrogram generated by the UPGMA method from the RAPD data showing the relationships among the clustering isolates and the clinical characteristics. GO, geographic origin; BL, body location; TP, type of patient; SF, susceptibility to fluconazole; SI, susceptibility to itraconazole; SN, susceptibility to nystatin; SAB, susceptibility to amphotericin B; GU, Guanajuato City, Mexico; DF, Mexico City, Mexico; NL, Nuevo León, Mexico; JA, Jalisco, Mexico; DE, Detroit, Mich.; RI, Richmond, Va.; GE, Germany; NA, not available; Resp. inf., respiratory infections; S, susceptible; S-DD, susceptible dose dependent; R, resistant. Solid line, statistical significance (0.83).

• Open in new tab
• Download powerpoint

FIG. 3.

Dendrogram constructed from the interpopulation distance matrix (Φ_ST values) generated from the comparison of each pair of populations. The dendrogram shows the relationships among the populations from different geographic origins. GU, Guanajuato City, Mexico; DE, Detroit, Mich.; RI, Richmond, Va.; DF, Mexico City, Mexico; NL, Nuevo León, Mexico; JA, Jalisco, Mexico; GE, Germany.

• Copyright © 2003 American Society for Microbiology