Collaborative Consensus for Optimized Multilocus Sequence Typing of Candida albicans

Abstract

A panel of 86 different Candida albicans isolates was subjected to multilocus sequence typing (MLST) in two laboratories to obtain sequence data for 10 published housekeeping gene fragments. Analysis of data for all possible combinations of five, six, seven, eight, and nine of the fragments showed that a set comprising the fragments AAT1a, ACC1, ADP1, MPIb, SYA1, VPS13, and ZWF1b was the smallest that yielded 86 unique diploid sequence types for the 86 isolates. This set is recommended for future MLST with C. albicans.

Multilocus sequence typing (MLST) is becoming a widely used approach to microbial isolate differentiation for epidemiological purposes (6). For Candida albicans, the species most often involved in deep-organ fungal diseases, MLST was introduced in 2002 (1), and a central internet database has been set up for deposition and analysis of C. albicans MLST data from any global source (http://calbicans.mlst.net). This MLST system is based on fragments of six C. albicans genes (in alphabetical order): ACC1, ADP1, GLN4, RPN2, SYA1, and VPS13.

A second study of C. albicans MLST involved four of these genes plus four other gene fragments, AAT1a, AAT1b, MPIb, and ZWF1b (5). Both sets of gene fragments provided highly discriminatory typing systems that gave stable, reproducible results and could distinguish even closely related strains. While in principle the use of as many gene sequences as possible should enhance the discriminatory power of an MLST scheme, in practical and technical terms a compromise is required to provide the maximum level of isolate differentiation with the minimum set of fragments. Our two groups therefore agreed to collaborate by exchanging C. albicans isolates that had already been described in the published MLST papers and compiling a data set of sequences based on all 10 fragments that have been used for MLST. Analysis of these data allows us to propose an optimized gene set for routine use in C. albicans MLST research.

A total of 92 C. albicans isolates (1, 5) were shared between the laboratories. These included duplicate cultures of isolate SC5314 and one culture of CAF2, derived from SC5314. As expected, identical MLST results were obtained for these three isolates, so two of the three data sets were excluded from analysis. Among the remaining 90 unique isolates, incomplete sequence data were obtained for 4, so the results for 86 isolates were analyzed to determine the optimal set of gene fragments for MLST.

The method used for MLST was as previously described (1, 5). Both DNA strands in this diploid fungus were sequenced for each of 10 fragments (Table 1), and the sequences were recorded by the one-letter code for nucleotides from the International Union of Pure and Applied Chemistry nomenclature. For each fragment, each different genotype was assigned a unique number. Diploid sequence types (DSTs) are the numbers assigned to each unique combination of genotypes.

TABLE 1.

Properties of the 10 DNA fragments used for C. albicans MLST

Gene fragment	C. albicans chromo- somea	No. of bases analyzed	No. of variable bases		dN/dSc	No. of Genotypes
			This study	Previous study (4)b		Found	Per vari -able base
AAT1a	2	373	10	7	0.17	23	2.3
AAT1b	2	339	6	6	0.08	15	2.5
ACC1	3	407	7	6	0.20	15	2.1
ADP1	1	443	16	15	0.41	23	1.4
GLN4	3	404	11	11	1.00	20	1.8
MPIb	2	375	11	11	0.33	20	1.8
RPN2	1	306	13	11	0.11	20	1.5
SYA1	6	391	13	13	0.34	26	2.0
VPS13	4	403	17	16	0.70	38	2.2
ZWF1b	1	491	9	8	0.12	36	4.0

^aTentative assignment of the chromosome on which the fragment is located, from http://cbr-rbc.nrc-cnrc.gc.ca/biovis/candida/.

^bThe number of variable bases has increased because of the larger panel of isolates sequenced.

^cCalculated according to Nei and Gojobori (3).

Table 1 summarizes the characteristics of the 10 DNA fragments used for MLST with the 86 C. albicans isolates. The sizes of the fragments were similar, ranging from 306 to 491 bases. An internet database, http://cbr-rbc.nrc-cnrc.gc.ca/biovis/candida/, compiled by Whiteway and colleagues with the input of unpublished data from the Stanford Candida genome project (http://alces.med.umn.edu/candida/), allows tentative assignment of C. albicans genes to individual chromosomes (3). From this database our 10 DNA fragments are probably distributed over five of the eight C. albicans chromosomes. The lowest proportion of bases that varied between isolates for any fragment was 3.7%, for ACC1, and the highest was 9.4%, for VPS13. The ratio of nonsynonymous to synonymous base changes (dN/dS), calculated by the method of Nei and Gojobori (4), was <1 for all fragments except GLN4 (Table 1), indicating neutral effects of selective pressure for most of the fragments.

The number of different genotypes determined for 86 C. albicans isolates from the 10 DNA fragments varied from 15 to 38, with ZWF1b giving the highest number of genotypes per variable base (Table 1). On the basis of this ratio, fragments ADP1, RPN2, GLN4, and MPIb gave the poorest level of isolate discrimination.

When the genotypes for the 86 C. albicans isolates were analyzed across all 10 DNA fragments, 86 unique DSTs were found, indicating perfect discrimination for MLST based on all the fragments. To deduce the minimum set of fragments that could differentiate all 86 isolates by MLST, the number of DSTs for the isolates was determined for every possible combination of five, six, seven, eight, and nine gene fragments. Results are shown in Table 2. With combinations of five fragments, 94% of results showed fewer than 80 DSTs for the isolates, and no combination resulted in 86 unique DSTs for the 86 isolates. The most discriminatory set of five fragments was AAT1a + ACC1 + ADP1 + VPS13 + ZWF1b, yielding 84 different DSTs (Table 2). The two most discriminatory sets of six fragments added either MPIb or SYA1 to the best five-fragment set, in both cases increasing the number of DSTs obtained to 85. The minimum fragment set of seven—the smallest set that gave 86 unique DSTs for the 86 isolates—was AAT1a + ACC1 + ADP1 + MPIb + SYA1 + VPS13 + ZWF1b.

TABLE 2.

Discrimination of 86 C. albicans isolates by MLST with all possible combinations of five, six, seven, eight, and nine DNA fragments

No. of:		No. (%) of DSTs determined for indicated no. of C. albicans isolates (n = 86)
Fragments	Combinations	<60	60-70	70-74	75-79	80-82	83	84	85	86
5	252	2 (0.8)	60 (23.8)	90 (35.7)	85 (33.7)	13 (5.2)	1 (0.4)	1 (0.4)	0 (0.0)	0 (0.0)
6	210	0 (0.0)	5 (2.4)	42 (20.0)	108 (51.4)	39 (18.6)	8 (3.8)	6 (2.9)	2 (1.0)	0 (0.0)
7	120	0 (0.0)	0 (0.0)	2 (1.7)	46 (38.3)	43 (35.8)	8 (6.7)	12 (10.0)	8 (6.7)	1 (0.8)
8	45	0 (0.0)	0 (0.0)	0 (0.0)	4 (8.9)	21 (46.7)	0 (0.0)	7 (15.6)	10 (22.2)	3 (6.7)
9	10	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	3 (30.0)	0 (0.0)	0 (0.0)	4 (40.0)	3 (30.0)

A number of considerations influence the choice of the set of gene fragments used for MLST. In principle, the fragments used for typing should be from housekeeping genes that are under stabilizing selective pressure for conservation of function. This property is indicated by the ratio of nonsynonymous to synonymous amino acid changes that result from base variation, which should ideally be less than 1.0 (2). Among our 10 fragments, GLN4, with a dN/dS of 1.0 (Table 1), was least suitable for MLST on the basis of this criterion. Four of our 10 fragments yielded fewer than two genotypes per base variation: ADP1, GLN4, MPIb, and RPN2 (Table 1). In both our laboratories, the greatest number of technical problems with reliable PCR amplification were encountered with GLN4. Based on these three considerations, fragment GLN4 is the least suitable of the 10 to be used for MLST; its absence from the minimum fragment set capable of discriminating all 86 C. albicans isolates confirms this.

The two other fragments not included in the minimum seven-fragment, fully discriminatory set were AAT1b and RPN2. AAT1b yielded only six variable bases in its 339-base sequence. Although the differentiating power of these variable bases was the second highest, at 2.5 genotypes per variation (Table 1), its contribution to the DST was poor; it was represented in only one of the three eight-fragment sets that discriminated all 86 isolates. RPN2 sequence variability represented a low ratio of nonsynonymous to synonymous amino acid changes but also one of the lowest numbers of genotypes per variable base (Table 1). Like AAT1b, RPN2 contributed to a full 86-DST discrimination between the 86 test isolates only as part of sets of eight or more fragments.

On the basis of this study we propose following gene set as an international standard for C. albicans MLST: AAT1a + ACC1 + ADP1 + MPIb + SYA1 + VPS13 + ZWF1b.