Stability of Allelic Frequencies and Distributions of Candida albicans Microsatellite Loci from U.S. Population-Based Surveillance Isolates

Abstract

Allelic distributions and frequencies of five Candida albicans microsatellite loci have been determined for strains isolated from the bloodstream and obtained through active population-based surveillance in two U.S. metropolitan areas between 1998 and 2000. These data were compared to data for isolates obtained from two other U.S. regions in 1992 to 1993. In a majority of pairwise combinations between sites, no evidence was seen for shifts in microsatellite allelic frequencies. One to three alleles were highly predominant and correlated with major genotypes. These data both support the concepts of allelic stability and genetic equilibria and suggest that, in the United States, strains of C. albicans isolated from the bloodstream may form a defined, genetically homogeneous population across geographical distance and time.

Candida albicans is a commensal yeast and a normal part of the human microflora. Often an agent of mucosal infections in healthy individuals, this organism is also capable of causing invasive disease, especially when the host is debilitated or immunocompromised. Bloodstream infections due to Candida spp. are associated with significant morbidity and mortality in hospitalized patients (5, 21-24). Their importance has resulted in the initiation of sentinel and population-based surveillance programs to determine the burden of disease and the incidence of antifungal resistance (1, 13, 22-24). Corollary genetic analyses of the most common cause of fungemia, C. albicans, have also been undertaken (30-32).

This species is diploid with a primarily clonal mode of reproduction. However, there is evidence for a low level of recombination (perhaps meiotic) between strains, and within a single strain, there are relatively high rates of nonhomologous mitotic recombination, translocation, and aneuploidy (3, 12). Nevertheless, population structure studies, based on cladistic analyses using randomly amplified DNA polymorphisms (25), single-nucleotide polymorphisms (31), and variable nucleotide repeats, such as microsatellites (4), have demonstrated a tripartite division among U.S. strains. In addition, there is recent evidence to suggest that, on a global level, there are additional strain types that are genetically distinct from these three previously described groups (7, 28).

Between 1992 and 1993, the Centers for Disease Control and Prevention (CDC) conducted a population-based surveillance for candidemia in the Atlanta, Ga., metropolitan area and in the San Francisco, Calif., area (13). The underlying illnesses varied, with approximately 8 to 10% of patients infected with human immunodeficiency virus (13). A total of 100 isolates of C. albicans collected during this surveillance study (one isolate per patient) were randomly selected for a substudy in which they were initially characterized by a limited number of genetic markers to identify possible genetic variation by location and/or time (15). From this set, 25 to 30 isolates from each of the two sites were selected to represent the observed genetic diversity in the larger population and further characterized using a larger number of microsatellite and single-nucleotide polymorphism markers (16).

The five microsatellites that have been employed are briefly described below. The CEF3 locus (named for Candida elongation factor 3) was originally described by Bretagne et al. (2) and is a compound microsatellite consisting of TTC/TTTC repeats. For all five loci, alleles are assigned by the size of the amplified product (in base pairs). Because of the compound nature of the repeat, CEF3 alleles can vary by 1 bp (2). The ERK locus (located in a protein kinase gene) has been analyzed by Field et al. (6) and is another compound repeat of the motif (CAGGCT)_a(CAAGCT)_b. Like CEF3, there are a relatively large number of alleles that vary by one nucleotide (6). ZNF1 is located in the coding region of the zinc finger (transcriptional factor) gene and is a simple CAA repeat (ZNF1 alleles vary by 3 bp) (14). Likewise, A3 and A4 are anonymous loci that were found by searching the C. albicans genome database (www.stanford.edu) and are simple repeats of TAA and GAA, respectively (16; Candida albicans genome sequencing project of the Stanford DNA Sequencing and Technology Center [http://www-sequence.stanford.edu/group/candida]).

Between October 1998 and September 2000, CDC conducted a second population-based active surveillance in Baltimore, Md., and in the state of Connecticut (R. A. Hajjeh, A. Sofair, L. Harrison, G. M. Lyon, M. E. Brandt, B. A. Arthington-Skaggs, J. Morgan, S. Mirza, M. Phelan, L. Thomson-Sanza, S. Huie, L. Yeo, M. Pass, and D. W. Warnock, Abstr. 39th Annu. Meet. Infect. Dis. Soc. Am., abstr. 641, 2001). The five microsatellite loci described above (16) were used to analyze 93 C. albicans isolates from this study. In this paper, we present a comparison of genetic diversity data for C. albicans isolates from two time periods and four geographical locations in the United States.

Isolates and storage.

A total of 439 incident bloodstream isolates of C. albicans were collected between 1992 and 1993, while 429 were collected between 1998 and 2000. Species identification was confirmed at the CDC using standard phenotypic methods including API 20C biochemical profiles and morphological appearance on Dalmau cornmeal agar plates. Isolates were prescreened to exclude Candida dubliniensis by the presence of a nontruncated copy of IS1, using C. albicans IS1 as a size control (15). All strains were stored at −70°C as 30% glycerol stocks in 1× phosphate-buffered saline. For this study, 57 C. albicans isolates (29 from Atlanta, Ga., and 28 from San Francisco, Calif.; one isolate per patient) were randomly selected from the surveillance population collected from 1992 to 1993, and 93 (46 from Connecticut and 47 from Baltimore, Md.) were from the 1998 to 2000 survey.

DNA extractions.

DNA isolation was performed from 10-ml yeast-peptone-dextrose (YPD) cultures as described previously (16). For IS1 sizing, PCR was performed in a volume of 25 μl and consisted of 35 cycles (1 cycle was 30 s at 95°C, 30 s at 55°C, and 30 s at 72°C), followed by a 4-min extension step at 72°C. The entire reaction mixture was loaded and run on 3:1 Nusieve agarose gels (FMC Bioproducts, Rockland, Maine).

Microsatellite analysis.

PCR was performed in a 25-μl volume as previously described (16) using carboxyfluorescine (FAM)- or tetrachloro-6-carboxyfluorescine (TET)-labeled forward primers. All primer sequences have been described previously (16). PCRs were performed individually but pooled for sizing. All PCRs were run for 35 cycles, with 1 cycle consisting of 30 s at 95°C, 30 s at 55°C, and 30 s at 72°C, with a 4-min extension at 72°C. For FAM- and TET-labeled reaction mixtures, 0.5 μl each was placed in 25 μl of formamide (Applied Biosystems Inc. [ABI], Foster City, Calif.) with 0.5 μl of TAMRA 500-labeled size standard (ABI) and denatured prior to analysis. PCR products were sized by an ABI 310 genetic analyzer (ABI) using Genescan software (ABI) with no split-peak correction (16).

Population analysis.

The microsatellite results for 1998 to 2000 were compared to data obtained from the 1992 to 1993 surveillance population. Genetic analysis was performed using GenePop (GenePop website, wbiomed.curtin.edu.au/genepop) (27) and Popgene (33). Data input files were created for diploid populations; each allele was assigned a two-digit number. For genic and genotypic differentiation, algorithms developed for diploid populations undergoing meiotic recombination were used (9, 27). Although there is little direct evidence for mating in C. albicans, this approach was used, assuming that mating does occur and is more conservative in rejecting the null hypothesis of genetic homogeneity than standard χ² statistics (11).

Microsatellite allelic distributions and frequencies.

Microsatellite allelic distributions and frequencies were determined for isolates from Connecticut and Baltimore, Md., and compared to data from the 1992 to 1993 surveillance collection (Atlanta, Ga., and San Francisco, Calif.) for a total of 150 isolates (n_alleles = 300). The results are given in Table 1. For the CEF3 locus, we found one new allele at bp 120. To date, this is the smallest allele observed at this locus. For ERK, we found four new alleles at bp 222, 235, 237, and 246. For locus A4, we found three new alleles at bp 111, 126, and 132. No additional alleles were found for the A3 or ZNF locus. For each locus, we observed from one to three predominant alleles in the population as previously described (16). With the possible exception of the A4 locus, allelic frequencies did not appear normally distributed around a mean allelic size by a visual inspection. This was particularly evident in the compound loci CEF3 and ERK. In each case, we observed predominant allelic combinations (genotypes). As previously shown using coalescence (16), these combinations correspond to subgroups I (IA) and II, comprising the majority (approximately 75%) of the population. The remaining alleles were found in low numbers, and these genotypes were often represented by one strain each.

TABLE 1.

Allelic frequencies and distributions for five microsatellite loci

Locus and allelea	No. of isolates (frequency)b				Mean frequency ± SD
	CT	MD	SF	GA
ERK
222	1 (0.01)	0 (0.00)	0 (0.00)	0 (0.00)	0.0025 ± 0.0005
231	25 (0.27)	29 (0.30)	19 (0.32)	18 (0.30)	0.297 ± 0.02
235	3 (0.03)	0 (0.00)	0 (0.00)	0 (0.00)	0.0075 ± 0.015
237	2 (0.02)	0 (0.00)	0 (0.00)	0 (0.00)	0.005 ± 0.01
239	51 (0.54)	61 (0.63)	27 (0.46)	26 (0.43)	0.515 ± 0.08
242	7 (0.07)	4 (0.04)	6 (0.10)	12 (0.20)	0.102 ± 0.07
244	1 (0.01)	1 (0.01)	0 (0.00)	1 (0.02)	0.010 ± 0.008
246	1 (0.01)	1 (0.01)	0 (0.00)	0 (0.00)	0.005 ± 0.005
248	1 (0.01)	0 (0.00)	3 (0.05)	1 (0.02)	0.020 ± 0.020
262	1 (0.01)	0 (0.00)	0 (0.00)	2 (0.03)	0.010 ± 0.010
265	0 (0.00)	0 (0.00)	2 (0.03)	0 (0.00)	0.007 ± 0.010
268	0 (0.00)	0 (0.00)	0 (0.00)	1 (0.02)	0.005 ± 0.010
271	0 (0.00)	0 (0.00)	1 (0.02)	0 (0.00)	0.005 ± 0.010
CEF3
120	0 (0.00)	2 (0.02)	0 (0.00)	0 (0.00)	0.005 ± 0.01
126	42 (0.44)	47 (0.49)	12 (0.21)	17 (0.38)	0.38 ± 0.12
127	0 (0.00)	0 (0.00)	1 (0.02)	1 (0.02)	0.01 ± 0.01
130	1 (0.01)	1 (0.01)	1 (0.01)	6 (0.10)	0.01 ± 0.005
131	2 (0.02)	3 (0.03)	6 (0.10)	5 (0.08)	0.06 ± 0.03
132	2 (0.02)	4 (0.04)	0 (0.00)	1 (0.02)	0.02 ± 0.02
133	0 (0.00)	0 (0.00)	3 (0.05)	1 (0.02)	0.02 ± 0.02
135	32 (0.34)	24 (0.25)	19 (0.33)	15 (0.25)	0.29 ± 0.05
136	1 (0.01)	0 (0.00)	6 (0.10)	3 (0.05)	0.04 ± 0.04
137	0 (0.00)	0 (0.00)	1 (0.02)	0 (0.00)	0.005 ± 0.01
139	0 (0.00)	1 (0.01)	6 (0.10)	3 (0.05)	0.04 ± 0.04
142	3 (0.03)	6 (0.06)	0 (0.00)	2 (0.03)	0.03 ± 0.02
144	8 (0.08)	5 (0.05)	3 (0.05)	5 (0.08)	0.06 ± 0.02
150	3 (0.03)	0 (0.00)	0 (0.00)	1 (0.02)	0.01 ± 0.01
152	0 (0.00)	3 (0.03)	1 (0.02)	0 (0.00)	0.01 ± 0.01
ZNF
147	0 (0.00)	0 (0.00)	1 (0.02)	0 (0.00)	0.005 ± 0.01
156	0 (0.00)	0 (0.00)	0 (0.00)	1 (0.02)	0.005 ± 0.01
165	4 (0.04)	2 (0.02)	1 (0.02)	1 (0.02)	0.02 ± 0.01
168	77 (0.82)	80 (0.83)	39 (0.67)	53 (0.88)	0.80 ± 0.09
171	8 (0.08)	10 (0.10)	10 (0.17)	3 (0.05)	0.10 ± 0.05
174	4 (0.04)	4 (0.04)	4 (0.07)	1 (0.02)	0.04 ± 0.02
180	1 (0.01)	0 (0.00)	3 (0.05)	1 (0.02)	0.02 ± 0.02
A3
133	48 (0.51)	46 (0.48)	19 (0.33)	28 (0.46)	0.44 ± 0.08
135	17 (0.18)	23 (0.24)	16 (0.27)	12 (0.20)	0.22 ± 0.04
138	3 (0.03)	3 (0.03)	3 (0.05)	2 (0.03)	0.03 ± 0.01
141	26 (0.27)	24 (0.25)	19 (0.33)	18 (0.30)	0.29 ± 0.03
144	0 (0.07)	0 (0.00)	1 (0.02)	0 (0.00)	0.005 ± 0.01
A4
111	1 (0.01)	0 (0.00)	0 (0.00)	0 (0.00)	0.002 ± 0.005
114	18 (0.19)	13 (0.13)	7 (0.12)	1 (0.02)	0.11 ± 0.07
117	45 (0.48)	50 (0.52)	28 (0.48)	31 (0.52)	0.50 ± 0.02
120	23 (0.24)	19 (0.20)	21 (0.36)	23 (0.38)	0.30 ± 0.09
126	0 (0.00)	1 (0.01)	0 (0.00)	0 (0.00)	0.002 ± 0.001
129	6 (0.06)	13 (0.13)	2 (0.03)	5 (0.08)	0.07 ± 0.04
132	1 (0.01)	0 (0.00)	0 (0.00)	0 (0.00)	0.002 ± 0.001

^aAlleles given as number of base pairs.

^bFrequency values are percentages. Site abbreviations: CT, Connecticut; MD, Baltimore, Md.; SF, San Francisco, Calif.; GA, Atlanta, Ga.

We then calculated the probability of each locus being in Hardy-Weinberg equilibrium based on χ² and G² (G is log likelihood) likelihood ratio tests (11). In these cases, the null hypothesis is that the allelic frequencies are in equilibrium and is rejected when P ≤ 0.05 (27). We observed no differences in the significance levels for χ² from those previously published; all were significantly less than 0.05 (16). For the more-conservative G² test, the ZNF and ERK values that were previously >0.05 (16) were found to be 0.008 and 0.01, respectively. Thus, there is no evidence for Hardy-Weinberg equilibrium in any of the five loci tested.

Likelihood tests (9, 27) were performed to determine if the allelic distribution is identical across populations (genic differentiation) and if the genotypic distribution is identical across populations (genotypic differentiation). These tests estimate the P value of a G-based exact test and in principle are the same as a probability test (or Fisher exact test) (9, 27). In each case, the null hypothesis is that the distribution is the same across populations. For genic differentiation, we observed that the A3, CEF3, and ERK loci exhibited P values of <0.05 (i.e., populations were significantly different), while the analysis of A4 and ZNF loci suggested that the populations were the same. Thus, from allelic frequency alone, we could not conclude that the populations were homogeneous. For genotypic differentiation, the results of a pairwise analysis across the four populations are shown in Fig. 1. For a total of 30 pairwise combinations, we observed that 23 of 30 (76%) of the comparisons suggested that the populations are the same (P ≤ 0.05). Using Bonferroni's correction for multiple tests on the same data, with P = 0.05/30 = 0.0016, all but two pairwise combinations were the same. For comparisons of geographical locations studied during the same time period (e.g., comparing Connecticut to Maryland data and comparing Georgia to San Francisco, Calif., data), we found that 100% of the comparisons could not reject the null hypothesis. Therefore, we conclude that there is no evidence for a shift in allelic frequencies between the two studies and that they effectively represent a single U.S. population of this type. The combined total frequencies (± standard deviations) are given in Table 1. In a calculation of the number of migrants (N_m) using private alleles (32), the calculated N_m is 13.9. This is a relatively large value, suggesting high rates of migration between the four cities sampled.

FIG. 1.

Genotypic differentiation (G-based) for all pairs of populations. For each microsatellite, P values (based on 1,000 iterations) are given, and shaded, broken lines are values not supporting the null hypothesis. The two cities represented in the 1992 to 1993 surveillance group are in the top boxes, and the cities in the 1998 to 2000 surveillance are in the bottom boxes. CT, Connecticut; MD, Baltimore, Md.; GA, Atlanta, Ga.; SF, San Francisco, Calif.

The bloodstream isolates collected through population-based active surveillance in 1992 to 2000 permitted an investigation of the genetic relatedness of C. albicans populations from four different geographical locations over this period. We conservatively tested the null hypothesis of genetic homogeneity in these four populations. After enumerating the alleles at five microsatellite loci and calculating the allele frequencies, we have found no evidence to support a shift in allelic frequencies in these populations. Our previous study using these microsatellite markers and similar statistical tests to compare the Atlanta and San Francisco populations to populations on other continents was able to detect genetic differentiation (8), so we expect that the tools used here were sensitive enough to detect differentiation if it existed.

In this study, we found one new rare allele in the compound locus CEF3 and four new rare alleles in ERK in the 93 isolates from the 1998 to 2000 survey. In addition, CEF3 and ERK each have two predominant alleles per locus and predominant genotypes that do not vary across the four populations surveyed. Considering the large number of alleles per locus (15 for CEF3 and 13 for ERK) and the discovery of new alleles in this study, we suspect that additional rare alleles will be uncovered as sample size increases. At present, it is unknown whether alleles of identical size are also identical in sequence, and there is evidence in the case of ERK that this may not be true (18). In the case of locus A4, we observed three new alleles and therefore suspect that, like CEF3 and ERK, there might be additional rare alleles uncovered in larger populations. We found no new alleles for A3 and ZNF; because we genotyped 93 isolates from the 1998 to 2000 survey, we suspect that there will be few additional, very rare alleles discovered for A3 and ZNF as sample size increases. These new observations give additional support for a departure from Hardy-Weinberg equilibria for all five loci.

In this study, we found genetic homogeneity among these four populations. The homogeneity found may be caused by several factors: movement of alleles between different genotypes of C. albicans due to recombination factors intrinsic to C. albicans, movement of the humans who are the hosts for the yeast, or selection of genotypes suited to the bloodstream environment. Sexual reproduction would need to be extensive to be the sole factor accounting for the homogeneity observed in these populations. While fusion of different cells is possible (12), C. albicans populations appear to be predominantly clonal (7, 8, 31). C. albicans frequently colonizes the host at or near birth, and individuals typically carry one or a few strains (21), so there is limited opportunity for sexual recombination within any given human host. These strains can persist over time, sometimes lasting years before recolonization of the host by other strains. The species is not free-living, and there is little information on other natural reservoirs (19). In addition, there is no evidence for horizontal transmission of genes.

The majority of genetic differentiation in C. albicans populations may be due to migrations throughout the evolutionary past and recent history of the natural human host population and not through other natural reservoirs. Due to high mutation rates, microsatellites are thought to be more useful in detecting recent migration events. The Atlanta and San Francisco human host populations were previously compared to populations obtained from other continents and found to have a high migration rate between the two U.S. sites, but the populations were genetically different from those from other sites worldwide (8). One possible interpretation of this study is that there has been migration among the four U.S. host populations examined (supported by an N_m of 13.9 for C. albicans). As observed in other systems, such migration leads to genetic homogeneity (29).

Another possibility is that these five loci are in genetic equilibrium by selection-mutation balance and that any other defined bloodstream population in the United States will display the same frequencies. Despite high mutation rates and number of alleles per locus common to microsatellites, a limited number of genotypes were found for CEF3 and ERK. This is consistent with selection at these loci. It will be of interest to compare these data with those derived from commensal isolates obtained from healthy individuals to determine if a characteristic profile of isolates from bloodstream infections exists. Other population genetic studies suggest a difference between “normal” commensal strains and strains from individuals infected with human immunodeficiency virus (17; M. Tibayrenc, Letter, Trends Microbiol. 5:253-254, 1997; R. Vilgalys, Y. Graser, and T. G. Mitchell, Author's Reply, Trends Microbiol. 5:254-257, 1997). Significantly, our data suggest that the isolates obtained from these selected U.S. sites may be representative of the overall U.S. population of isolates from bloodstream infections.