Multiple-Locus Variable-Number Tandem-Repeat Analysis for Rapid Typing of Candida glabrata

Abstract

A multiple-locus variable-number tandem-repeat analysis (MLVA) using six microsatellite markers was assessed in 127 Candida glabrata isolates. Thirty-seven different genotypes, stable both in vitro and in vivo, were observed. The highest discriminatory power (D = 0.902) was reached by using only four markers. MLVA seems to be relevant for C. glabrata typing.

Candida glabrata has recently emerged as a major pathogen, causing mucosal and systemic infections (10, 17). Several methods, such as electrophoretic karyotyping, restriction enzyme analysis, Southern blotting with probes, randomly amplified polymorphic DNA, and multilocus sequence typing (MLST), have been used to distinguish and type C. glabrata isolates (2, 7, 8, 14, 21, 28). Microsatellite polymorphism analysis has been widely used for typing fungi (3, 4, 11, 15, 18, 25), and this could be an alternative, easy-to-perform, reproducible method suitable for large-scale studies of C. glabrata epidemiology. Recently, Foulet et al. described three polymorphic microsatellite markers to investigate the delineation of clinical C. glabrata isolates (12). The discriminatory power of this method was good but not optimal. The aim of our study was to assess a microsatellite-based multiple-locus variable-number tandem-repeat analysis (MLVA) using new markers for C. glabrata typing.

One hundred twenty-seven C. glabrata strains were analyzed, including four reference strains, 98 independent clinical isolates, and 25 epidemiologically related isolates. These 25 were from the blood cultures and peripheral site isolates of eight patients with C. glabrata candidemia. Genomic DNA was extracted by boiling with Chelex resin as previously described (6, 23). Six microsatellite markers were selected from the C. glabrata DNA sequences available in the GenBank database (29). Primer sequences were designed with Primer3 software (24), and locations in the C. glabrata genome were determined with the Genolevures database (http://cbi.labri.fr/Genolevures/) (Table 1). For each primer set, PCRs were performed with a 20-μl final volume containing 1 μl of DNA, each deoxynucleoside triphosphate at 200 μM, a forward primer and a 5′-dye-labeled reverse primer at 0.25 μM each, and 1 U of Taq DNA polymerase (Promega, Madison, WI). The amplification conditions were 5 min at 95°C, followed by 35 cycles of 95°C for 30 s, 52°C for 30 s, and 72°C for 45 s and then a final step of 72°C for 10 min. For any given isolate, amplicons of each PCR were pooled before multiplex fragment sizing with a Ceq 8000 Genetic Analyzer (Beckman Coulter, Fullerton, CA). Strain IHEM 9670 was run as a control in each experiment. As allele sizing depends on dyes and the analyzer used for electrophoresis, results were expressed as the exact size of the sequence (determined by sequencing of each representative allele for each locus) to allow further interlaboratory comparisons of MLVA results (13, 20). The reproducibility and stability of the method were assessed as described elsewhere (25). Primer specificity was checked by studying the 11 non-C. glabrata reference strains Candida albicans IHEM 9559, Candida dubliniensis IHEM 14280, Candida tropicalis CBS 1920, Candida parapsilosis IHEM 9557, Candida krusei IHEM 9560, Candida norvegensis IHEM 5575, Candida lusitaniae IHEM 10293, Candida nivariensis CBS 9983 and CBS 9984, Candida bracarensis CBS 10154, and Saccharomyces cerevisiae IHEM 6036. Primer specificity was also tested against the sequences in the GenBank database by using BLAST searches.

TABLE 1.

Characteristics of microsatellite markers of C. glabrata

Marker	GenBank accession no.	Microsatellite sequencea	Primer sequence (5′ to 3′)	No. of alleles	Allele size range (bp)	Marker locationb (noncoding region or name of gene if coding)
Cg4	BZ298249	(GT)19	AATGCGTGTGTGTGCGTAGT	9	228-252	Chromosome E (noncoding)
			DyeD2-AAAAATTTAGGCCCCATCG
Cg5	BZ295911	(GT)3(CT)2(GT)15	CTCAACAGAACCAATCTCTGC	9	122-146	Chromosome M (noncoding)
			DyeD3-TCTTCACGCTGCCAATCTTA
Cg6	BZ298679	(CA)12	DyeD2-AGCAAGAGGGAGGAGGAAACT	11	301-345	Chromosome E (noncoding)
			AAATCCGGGGATAGATGAGG
Cg7	BZ298409	BD2BFBD2BFB4D2B3DBDB4D9B3	GATGATTCTGCCCGTTAGGA	7	179-215	Chromosome J (CAGL0J01595g)
			DyeD3-AAGAGTTCCCTGGTGGAATG
Cg10	BZ297776	B3DB11DBD2B4DB5D2B11D	TGCCTACGATGAAGAAATCG	12	249-307	Chromosome E (CAGL0E00561g)
			DyeD4-CTGGTAAGCACCGTTTTGGT
Cg11	BZ298943	DBD2B2D2B4D6(CCA)BD	CTGGTCTACCCAGCACCAAT	5	141-169	Chromosome A (CAGL0A03872g)
			DyeD4-TGCTGATACTGGTAGTTTTGTTG

^aNotation of trinucleotides as used by Shemer et al. (26): B, CAA; D, CAG; F, CTG.

^bMicrosatellite location on C. glabrata genome according to the Genolevure database (http://cbi.labri.fr/Genolevures). Microsatellites Cg7, Cg10, and Cg11 were each located in separate genes coding for a protein with no unidentified function.

Sequencing of microsatellites was performed for 30 strains, including reference strains and at least 1 strain for each allele of a given locus by using the CEQ DTSC Quick Start Master Kit (Beckman Coulter, Fullerton, CA).

The discriminatory power (D) of MLVA was calculated with Hunter and Gaston's formula (16). To group the unrelated isolates according to their genetic distance, hierarchical clustering analysis was performed with R software (http://www.r-project.org) and the pvclust package (27). Potential relationships between the genotypes and origins of isolates (clinical data, sex, ward, and anatomical sites) were assessed by hierarchical clustering analysis with canonical discriminant analysis with Tanagra software (http://eric.univ-lyon2.fr/∼ricco/tanagra/en/tanagra.html).

Thirty-seven different MLVA profiles of C. glabrata strains were observed (Fig. 1). Three clusters of 23, 20, and 10 genotypically similar isolates were identified (respectively, 22.5%, 19.6%, and 9.8% of the isolates). Twenty-six isolates each gave their own unique MLVA patterns. The discriminatory power (D) of each locus alone varied from 0.64 for Cg7 to 0.79 for Cg10. A D value of 0.902 was reached by combined use of the six markers, but MLVA data obtained by using only four of the six (Cg4, Cg5, Cg6, and Cg10) achieved the same results.

FIG. 1.

MLVA-based dendrogram and genotype scores derived from results of the six markers for the 102 unrelated isolates. Hierarchical analysis by the Euclidean distance and the Ward clustering method was performed to classify the 37 genotypes. Allele size is expressed as the exact size of the sequence. The approximately unbiased P values (values on nodes, in percent) were calculated with a multiscale bootstrap (B = 1,000) and are shown only for P values of ≥80% and genetic distances of ≥0.2. Reference strain data are identified by the following superscript letters: a, IHEM 19154 and IHEM 19221; b, CBS 138; c, IHEM 9670. Abbreviations: Dig., digestive tract; Gen.Ur., genitourinary tract; Pul., pulmonary tract; Oro., oropharynx; Blood, blood cultures and heart valves; Oth., other sites (skin, nails).

Sequencing data showed that fragment size polymorphism was explainable by the variation of microsatellite repeats, with the exception of additional polymorphisms observed in the flanking regions for locus Cg6 (poly-T [8 or 9 T residues] and a 16-bp deletion for isolate B473) and Cg4 (a 2-bp deletion at allele 238). Homoplasy (i.e., identical lengths but different DNA sequences) was identified for three loci, Cg7, Cg10 (differences in alternations of CAA and CAG tags in the microsatellite sequence), and Cg6 (mutations in the flanking region). In our investigation, sequencing did not, however, increase the discriminatory power of our MLVA method based on fragment size analysis and thus no further studies were undertaken. Except for loci Cg5 (700-bp amplicon with C. lusitaniae) and Cg11 (120-bp amplicon for C. albicans, multiple bands for C. parapsilosis and C. norvegensis), primers failed to amplify the DNA of non-C. glabrata yeast species. No other potential cross-amplifications were detected with the BLAST software. MLVA markers were reproducible, and none showed any variations or microevolutions even after 300 generations in vitro. This stability was also confirmed after at least 40 days of in vivo maintenance in two patients with C. glabrata candidemia (patients 5 and 6, Table 2). Statistical analysis failed to show any correlations between the epidemiological characteristics of strains and MLVA data. Lastly, for each of the eight patients with C. glabrata candidemia, isolates from peripheral sites and blood culture showed the same MLVA genotype, confirming the endogenous origin of C. glabrata candidemia (Table 2).

TABLE 2.

MLVA patterns of isolates from patients with C. glabrata candidemia

Patient no. and anatomical origin of isolate	Day of isolation	Cg4	Cg6	Cg10	Cg5	Cg7	Cg11
1
Expectoration	0	236	323	268	134	209	141
Blood culture	1	236	323	268	134	209	141
2a
Oral mucosab	0	234	323	292	138	206	147
Rectal swab	0	236	323	283	146	194	163
Surgical wound	0	236	323	283	146	194	163
Abdominal drain effluent	0	236	323	283	146	194	163
Gastric juice	0	236	323	283	146	194	163
Blood culture	1	236	323	283	146	194	163
3
Rectal swab	0	238	321	274	134	182	163
Blood culture	2	238	321	274	134	182	163
4
Rectal swab	0	236	323	268	134	209	141
Abdominal collection	0	236	323	268	134	209	141
Blood culture	1	236	323	268	134	209	141
5
Rectal swab	0	236	321	268	134	209	141
Blood culture	1	236	321	268	134	209	141
Expectoration	41	236	321	268	134	209	141
Blood culture	41	236	321	268	134	209	141
6
Aortic hematoma	0	234	321	262	136	182	169
Rectal swab	11	234	321	262	136	182	169
Urine	15	234	321	262	136	182	169
Blood culture	52	234	321	262	136	182	169
7
Abdominal drain effluent	0	246	321	283	126	209	160
Blood culture	2	246	321	283	126	209	160
8
Perineal swab	0	230	316	283	138	182	160
Blood culture	2	230	316	283	138	182	160

^aPatient 2 harbored two genotypically distinct strains (only one was isolated from a blood culture).

^bMLVA genotype not previously described among epidemiologically unrelated isolates.

In the present study, we investigated six new microsatellite markers for C. glabrata. By combining four of them, a discriminatory power (D) of 0.902, higher than those previously published, was achieved (12). As D values higher than 0.90 are needed for the accurate typing of epidemiologically related isolates (12, 16), microsatellite-based MLVA seems to be an attractive method for large epidemiological surveys and appears to be as discriminant as MLST (D = 0.898 computed from data obtained by Dodgson et al. [9]). Because of their specificity and typeability, microsatellite markers can also be used to distinguish phenotypically and genetically related yeast species, i.e., C. parapsilosis, C. metapsilosis, and C. orthopsilosis in the study by Lasker et al. and C. glabrata, C. nivariensis, and C. bracarensis in our study (1, 5, 18).

At a given locus, most of the possible numbers of repeats were present, demonstrating a continuum for increasing or decreasing the numbers of repeats. However, the distribution of alleles was irregular and several alleles were prominent, since three multilocus genotypes represented 52% of the isolates studied. Other studies using microsatellite or MLST data have highlighted the fact that distinct genetic clades of C. glabrata prevail in different geographical regions (8, 9, 12). Our unrelated isolates were collected from patients from a restricted geographical area, which could partially explain the unequal distribution of genotypes in the population studied. The other hypothesis is that some genotypes could have an ecological advantage (12).

Another finding of our study is that no microsatellite genotypes were associated with any of the clinical data recorded, as previously reported (8, 12, 19). Additionally, in the eight infected patients, we have shown that C. glabrata candidemia originated from colonizing isolates, as previously described for both C. glabrata (12) and C. albicans (4, 22).

The MLVA method, based on fragment size analysis of four microsatellite markers (Cg4, Cg5, Cg6, and Cg10), is easy to perform, discriminatory, and highly reproducible. It appears to be a powerful method for distinguishing epidemiologically related isolates and could be especially useful for studying nosocomial cross-transmission and the kinetics of the colonization-to-infection process. Use of additional microsatellite markers, such as those described elsewhere (12), could perhaps further improve the discriminatory power of our MLVA primer set.