点击显示 收起
Department of Molecular Genetics and Microbiology, Duke University Medical Center
Department of Biology, Duke University, Durham, North Carolina
ABSTRACT
Cryptococcus neoformans is a major, global cause of meningoencephalitis in immunocompromised patients. Despite advances in the molecular epidemiology of C. neoformans, its population structure and mode of reproduction are not well understood. In the environment, it is associated with avian guano or vegetation. We collected nearly 800 environmental isolates from three locations in the United States (viz., North Carolina, California, and Texas) and compared them with one another and with clinical isolates from North Carolina. As expected, they consisted of the most prevalent serotypes, serotypes A and D, as well as serotype AD hybrids. The majority of environmental isolates were obtained from pigeon excreta. All environmental and clinical isolates of serotype A or D had the MAT mating-type allele. However, the AD hybrids included MATa alleles typical of serotypes A and D. Using an amplified fragment length polymorphism fingerprinting technique with two primer pairs, we identified 12 genotypes among the isolates of serotype A. Six of these genotypes were present in both the clinical and the environmental populations. However, one of the most prevalent environmental genotypes was absent from the clinical samples, and two other genotypes were isolated only from patients. The combined molecular data suggest that this environmental population of C. neoformans is predominantly clonal, although there was evidence for recent or past recombination.
INTRODUCTION
Cryptococcus neoformans is an exogenous, opportunistic pathogen capable of causing life-threatening infections, especially in persons with cellular immunodeficiencies, such as patients with AIDS, transplants, or hematologic malignancies (9). This yeast is ubiquitous in the environment, where it is usually associated with avian guano or vegetative debris (9, 11, 29, 40). C. neoformans is a haploid basidiomycetous species with a bipolar mating system and two alternative mating alleles, MATa and MAT. Strains of the opposite mating type are capable of sexual reproduction in the laboratory. However, most clinical populations are dominated by isolates of only one mating type, MAT, and it is unknown whether this fungus undergoes sexual recombination in nature (20, 26, 41, 42).
On the basis of differences in ecology, physiology, capsular polysaccharide, and clinical manifestations, three varieties and five serotypes of C. neoformans have been recognized. The most common variety, C. neoformans var. grubii, represents isolates of serotype A, primarily infects immunocompromised individuals, and causes more than 90% of all cryptococcal infections and more than 99% of the cases of cryptococcosis in patients with AIDS (9, 24). Strains of serotype D, C. neoformans var. neoformans, also infect immunocompromised patients; however, they cause fewer cases of disease and are considered less pathogenic (9, 24). Hybrid serotype AD strains have been isolated from the environment and patients in North America and Europe (6, 18, 43, 47). Isolates of the other two serotypes, serotypes B and C, are represented by C. neoformans var. gattii, and they tend to infect immunocompetent individuals. This variety is usually confined to tropical regions, where it is associated with decaying eucalyptus and other trees (9).
Strains of serotypes A and D are commonly isolated from aged pigeon excreta (11, 15, 27, 46) or soil contaminated with weathered bird excrement (41). In addition, several reports describe the isolation of strains of serotype A from decaying trees (27, 29) and household dust (9, 27, 34).
A few environmental isolates of C. neoformans have been compared with regional clinical isolates by a variety of fingerprinting methods (4-6, 11, 14, 16, 23, 31, 39, 46). Environmental and clinical isolates within the same geographical areas have often proven to be closely related, if not identical (11, 31, 46). However, most of these reports involved studies of a small number of environmental isolates and did not fully analyze the population structure of C. neoformans. Indeed, the diversity of environmental isolates has never been assessed, and several important issues relating to the population structure of C. neoformans remain unanswered. For example, why are clinical samples in the United States dominated by isolates of serotype A with the MAT allele Are particular genotypes more likely to cause disease Are there any spatial or temporal aspects to the population structure of environmental isolates Are natural populations invariably clonal, or is there evidence for genetic recombination
To address these issues, we collected more than 700 environmental isolates of C. neoformans from 24 locations in North Carolina, genotyped each isolate by generating amplified fragment length polymorphisms (AFLPs) with two independent primer pairs (20), and analyzed the genetic structure of this population. We compared this population with clinical isolates from North Carolina, Texas, and California. Although there was definitive evidence of recent or historical recombination among the isolates of serotype A, the data revealed little genetic diversity and supported the hypothesis that clonal expansion is the predominant mode of reproduction among clinical and environmental populations of C. neoformans.
MATERIALS AND METHODS
Isolates of C. neoformans. A total of 762 environmental isolates of C. neoformans were obtained from a single location in San Francisco, Calif. (n = 22); two locations in San Antonio, Tex. (n = 60); and 24 discrete locations in the central, mountainous, and coastal regions of North Carolina (n = 680). To investigate whether any seasonal, temporal, or spatial variation existed, five of eight locations in Durham, N.C., were sampled repeatedly over a period of 1 year: in April, July, and November of 2002 and in February and July of 2003. For comparison, we analyzed 51 clinical isolates from 42 patients who resided in North Carolina or proximal regions of South Carolina and Virginia and who were admitted to the Duke University Medical Center in Durham, N.C., between 2000 and 2002. These isolates were recovered from spinal fluid (n = 32), respiratory specimens (n = 12), tissue (n = 3), blood (n = 2), and bone marrow and skin (1 isolate each). An additional 136 clinical isolates of serotype A from patients residing elsewhere in the United States were donated by the Centers for Disease Control and Prevention (CDC) (5, 6, 43). These isolates were obtained from patients in four metropolitan areas: Atlanta, Ga.; Birmingham, Ala.; Houston, Tex.; and San Francisco, Calif. Most of these isolates were cultured from spinal fluid and blood specimens.
Isolation, identification, maintenance, and serotyping of environmental isolates. Small quantities of desiccated avian excreta or decayed wood were collected from several sites at each location and placed in separate plastic bags, which were tightly secured with flexible wire closures. Two grams of material from each sample were suspended in 10 ml of sterile distilled water, and the suspension was incubated for 2 h at room temperature with constant agitation at 150 rpm. Two 10-fold dilutions were prepared from each suspension, and 50-μl portions of undiluted sample and 1:10 and 1:100 dilutions of each sample were evenly spread with a sterile glass rod on the surface of nigerseed agar plates (32) supplemented with 0.4 g of chloramphenicol (Sigma-Aldrich, St. Louis, Mo.) per liter, 0.025 g of gentamicin (EM Science, Gibbstown, N.J.) per liter, and 0.1 g of biphenyl (Alfa Aesar, Ward Hill, Mass.) (0.1 g/10 ml of 95% ethanol) per liter and incubated at either 30 or 35°C for 3 to 5 days (we observed no differences in the yields between the two incubation temperatures). Brown yeast colonies were selected, grown in pure culture on nigerseed agar plates without antibiotics, confirmed to be C. neoformans by standard morphological and physiological criteria (9), and maintained on yeast extract-peptone-dextrose (YPD) agar medium (Difco, Detroit, Mich.) at 30°C. All environmental samples were processed within 1 to 3 days after collection. Of the many colonies of C. neoformans obtained from each environmental sample, approximately 30 isolates were randomly selected for further analysis. Isolates were serotyped with commercial monoclonal antibodies (Iatron, Tokyo, Japan) and by PCR amplification with serotype-specific primers and AFLP analysis (see below). The ploidy of the serotype AD hybrids was evaluated by flow cytometry (20).
DNA extraction. Genomic DNA was extracted from each isolate as described elsewhere (http://www.fhcrc.org/labs/breeden/Methods_BreedenLab.html), with the following modifications: each yeast isolate was grown in pure culture on a YPD agar plate for 48 h. Approximately 50 μl of yeast cells was scraped from the plate with a sterile toothpick, suspended in a microcentrifuge tube containing 500 μl of lysis buffer (100 mM Tris [pH 8.0], 50 mM EDTA, 1% sodium dodecyl sulfate) and approximately 100 μl of sterile glass beads (diameter, 425 to 600 μm; Sigma-Aldrich), and vortexed for 5 min. Then, 275 μl of 7 M ammonium acetate (pH 7.0) was added; this mixture was incubated at 70°C for 5 min and chilled on ice for 5 min. We added 500 μl of chloroform, the mixture was vortexed for 1 min and centrifuged at 16,000 x g for 5 min, and the clear supernatant was withdrawn. DNA was precipitated by adding 1 ml of isopropyl alcohol, washed with 70% (vol/vol) ethanol, air dried, and resuspended in 50 μl of 10 mM Tris (pH 8.0) supplemented with 0.25 μg of RNase A (Sigma-Aldrich) per ml.
AFLP analysis. The generation of AFLP markers was performed as described previously (20). Two primer combinations, primers EcoRI-AC-FAM and MseI-G and primers EcoRI-TG-FAM and MseI-G, labeled at the 5' end with 6-carboxyfluorescein (FAM), were used for the selective PCR. The selective PCR products were electrophoresed in an automated sequencer (ABI 3700; Applied Biosystems, Foster City, Calif.) under the default run module for Genescan (version 3.7) analysis software, according to the instructions of the manufacturer. The fragment sizes were determined with the Genescan (version 3.7) analysis software and the 500 TAMRA 6-carboxytetramethylrhodamine internal size standard (Applied Biosystems). The data were analyzed with Genographer (version 1.6) software (3) and scored manually. Polymorphic AFLP bands were defined as bands of the same size that were present in some but not all isolates. To assess the reproducibility of the AFLP method, DNA was extracted and the AFLP reactions and analyses were performed on at least two separate occasions for each isolate. By comparison of the results from the replicate analyses, 92% of the AFLP bands were identical (data not shown). Only intense and reproducible bands were scored for analysis of the population structure. AFLP genotypes that differed by two or more bands were considered unique.
Mating-type identification. The mating type of each strain was determined by PCR with mating type- and serotype-specific primers that amplify portions of the STE20a or STE20 allele of serotype A or D strains and were designated primers STE20aA and STE20A, respectively, for serotype A, and STE20aD and STE20D, respectively, for serotype D (26). For isolates of serotype D or serotype AD hybrid strains that yielded ambiguous results, PCR primers specific to portions of the STE3a or STE3 allele were designed: primers STE3a-f (5'-ACCTTTGCGGTTTCATCAAC) and STE3a-r (AAGGTCGCATGGGTAATGAG) for STE3a and primers STE3-f (5'-TAACATTGGACATCCCAGCA) and STE3-r (5'-GAAGACGCAGGGTACAGCTC) for STE3. The PCR conditions for amplification of portions of the STE3 alleles were as follows: for STE3a, 94°C for 5 min, followed by 30 cycles of 94°C for 30 s, 56°C for 30 s, and 72°C for 1 min and then 7 min of extension at 72°C; for STE3, 94°C for 5 min, followed by 30 cycles of 94°C for 30 s, 62°C for 30 s, and 72°C for 1 min and then 7 min of extension at 72°C. Each PCR mixture contained 10 μl of 1x PCR buffer, 2 mM MgCl2, 0.2 mM deoxynucleoside triphosphates, 1 μM each primer, 0.065 μl of Taq DNA polymerase (Invitrogen), and approximately 1 ng of genomic DNA. To determine the mating type of the AD1 hybrid strain, we performed Southern hybridizations with probes specific to the portions of the STE20a and STE20 genes of serotypes A and D. Briefly, 300- to 500-bp fragments of the STE20 genes were amplified and labeled by using the PCR DIG probe synthesis kit (Roche Diagnostics, Indianapolis, Ind.), according to the instructions of the manufacturer. A sample of approximately 10 ng of genomic DNA was digested with two restriction endonucleases, XhoI and XbaI, and electrophoresed in a 0.8% agarose gel in 0.5x TBE (Tris-borate-EDTA). DNA fragments were transferred by a capillary tube from the gel to a positively charged nylon membrane (Roche Diagnostics) by standard procedures (30). Southern blot hybridizations were performed under the conditions recommended by the manufacturer (Roche), and the blots were hybridized and washed under stringent conditions (65°C).
Flow cytometry. To establish the ploidy of certain isolates, the relative DNA content was measured by flow cytometry and was compared to that of C. neoformans strain H99, the haploid reference strain of serotype A (10, 20, 36).
Data analysis. The genetic relatedness among the isolates was evaluated by nonmetric multidimensional scaling (MDS) analysis by using Euclidian distance measures with the Community Analysis Package (version 2.4; PISCES Conservation Ltd., Hampshire, United Kingdom). In addition, genetic relatedness among the isolates was also by evaluated by using Nei and Li genetic distances for restriction fragment data (25), followed by cluster analysis with the neighbor-joining algorithm from the PAUP software package (35).
Genotypic diversity was defined as the probability that two individuals taken at random have different genotypes and was equivalent to
where pi is the frequency of the ith genotype, and n is the number of individuals sampled (1). Genotypic diversity was calculated by using Multilocus (version 1.2) software (1). Phylogenetic analyses of the AFLP data were also performed with the PAUP package. Maximum-parsimony trees were identified by using Dollo parsimony criteria (13) with heuristic searches based on 100 random sequence additions for each data set. A majority-rule consensus tree and strict consensus trees were generated. The parsimony tree length was tested by using 1,000 permutations (1). To evaluate the association among loci in each sample, we used the index of association (IA), the new unbiased estimate of multilocus linkage disequilibrium (rd) (8, 22, 44), and the parsimony tree length test (2, 8). IA and rd values were calculated with Multilocus (version 1.2) software, and 1,000 artificially recombined data sets were used to determine the statistical values of the tests (2).
RESULTS
C. neoformans was recovered from pigeon habitats. In the environment of North Carolina, C. neoformans was primarily associated with excreta at the habitats of feral pigeons (Columba livia). C. neoformans was isolated from 38 of 61 (63%) sites containing weathered and desiccated pigeon excreta. However, despite persistent efforts, C. neoformans was not isolated from the excreta of other species of birds, including the domestic sparrow, swallow, robin, and owl. Around San Francisco, Calif., we sampled 17 sites with abundant, dry pigeon excreta, but C. neoformans was isolated from only one location (6%). In San Antonio, Tex., 10 similar sites were sampled, but only 2 (20%) yielded C. neoformans.
Isolation of C. neoformans from an arboreal site. In several reports, C. neoformans was cultured from the hollows of various species of decaying trees (27, 29, 39). In Durham, N.C., we sampled the hollows of 30 trees representing a variety of species (viz., red, white, and willow oaks; maple; sweet gum; tulip poplar; loblolly pine; and magnolia). Only one site, a white oak tree hollow, was positive; and multiple isolates of serotype D with the same genotype (D5) from that site (Fig. 1) were cultured repeatedly over a period of 3 months.
Distributions of isolates of serotypes A, D, and AD differ in the local environment and patients. Environmental and clinical isolates were serotyped with commercial monoclonal antibodies (Iatron), and their serotypes were confirmed by AFLP analysis (4, 20). Among 762 environmental isolates from North Carolina, 650 (85.3%) were serotype A, 58 (7.6%) were serotype D, and 54 (7.1%) were serotype AD hybrids. No isolates of serotype B or C were recovered. The clinical isolates from 42 patients in North Carolina consisted of 40 isolates of serotype A (95.2%) and 1 isolate each of serotype D (2.4%) and serotype AD (2.4%) (Table 1).
AFLP genotypes reveal greater diversity among the clinical isolates than the environmental isolates. We used AFLP genotyping with two different primer pairs to determine the genetic diversity of the environmental and clinical populations from North Carolina, Texas, and California (4, 20). Sixty-four polymorphic bands were generated; and these allowed us to distinguish isolates of serotypes A, D, and AD (Fig. 1). Among the isolates of serotype A, 21 polymorphic bands defined eight distinct AFLP genotypes that differed by 2 or more bands; for the serotype D isolates, 13 polymorphic bands yielded five AFLP genotypes. The AD hybrids differed in 18 polymorphic bands, and six AFLP genotypes were resolved. In addition, four serotype A isolates had genotypes that varied by a single band; one of these isolates was from a patient in North Carolina (genotype A8"), another one was isolated from the environment in North Carolina (genotype A6"), and the other two were from California (genotypes A1CA1 and A1CA2). Genotypic diversity, which is the probability that two random individuals have different genotypes, was higher for the clinical isolates of serotype A (genotypic diversity, 0.87) than the environmental samples of serotypes A, D, and AD (genotypic diversities, 0.62, 0.85, and 0.67, respectively) (Table 1).
The AFLP genotypes were plotted on the MDS graph (Fig. 1A) to illustrate their genetic relationships. This analysis indicates that populations of serotypes A and D and the serotype AD hybrid were genetically distinct. Furthermore, the serotype A sample consisted of two well-separated subgroups.
Phylogenetic relationships among the isolates were determined by the maximum-parsimony approach with the Dollo parsimony criteria (13). From the 108 most parsimonious trees, the majority-rule consensus phylogram was generated (Fig. 1B). This phylogram contains three major clades that correspond to serotype D and the two subgroups of serotype A. Most of the serotype AD hybrids clustered with one of these clades, depending upon the genetic background of their serotype A counterparts (Fig. 1B). Representatives from both subgroups of serotype A were isolated from the environment and patients. However, the proportion of subgroup I isolates was significantly greater (P < 0.0001) in the clinical sample than in the environmental sample (Table 1).
Lack of geographic isolation among the environmental strains of serotype A in North Carolina. Isolates of serotype A from the coastal, central, and mountainous regions of North Carolina shared many of the same AFLP genotypes (Fig. 2A). Isolates of serotypes D and AD were found only in the central region (Durham and Raleigh); however, most samples were collected from this region. There was no apparent spatial structure or geographic distribution in the environmental population (Fig. 2).
Lack of temporal changes in the isolation of environmental strains of serotype A at five sites in Durham, N.C. Repeated sampling of five sites in Durham, N.C., revealed little change in the genotypic composition over time (Table 2). The same AFLP genotypes were associated with specific locations for at least several months (Table 2).
Evidence of clonality among environmental isolates of serotype A. All 22 environmental isolates from one location in San Francisco, Calif., were of AFLP genotype A1, isolates of which were also present in several regions in North Carolina. However, the California isolates included two AFLP genotypes that were closely related to genotype A1, designated genotypes A1CA1 and A1CA2, but they had unique bands that were missing from strains from North Carolina and Texas (Fig. 1). The 60 environmental isolates from San Antonio, Tex., were represented by four AFLP genotypes (genotypes A2, A3, A5, and A6), which were also found in various areas of North Carolina. None of the Texas isolates had unique AFLP banding patterns.
Unequal distribution of strains of serotype A among environmental and clinical populations. The proportions of different AFLP genotypes in the environmental and clinical samples are shown in Fig. 3. Among the 12 AFLP genotypes, 6 (genotypes A1, A3, A4, A5, A6, and A7) were present in both clinical and environmental samples. Genotype A2 was quite prevalent in the environment, as it was isolated from Texas and from the central, mountainous, and coastal areas of North Carolina; however, genotype A2 was not detected among the patient isolates from central North Carolina, and it was rare in the CDC collection of clinical isolates of serotype A (Fig. 3). Conversely, strains of genotype A8 and closely related genotype A8" were isolated only from the clinical samples and were never found in the environment, and genotype A7 was common in the clinical samples but was rarely isolated from the environment. Since genotypes A1CA1 and A1CA2 were isolated only in California, where few clinical isolates were available, it was difficult to assess their prevalence.
We also analyzed a sampling of the CDC collection of clinical isolates from the metropolitan areas of four U.S. cities (5, 6). These isolates were previously genotyped by multilocus enzyme electrophoresis (MLEE) and random amplified polymorphic DNA analysis. Among these isolates of serotype A, Brandt et al. (5, 6) identified 10 distinct MLEE types that clustered in two distinct groups, designated ET1 and ET2. We determined the AFLP genotypes of 136 of these serotype A strains and compared their MLEE and AFLP genotypes. With one exception, the CDC strains of MLEE types ET1 (n = 109) and ET2 (n = 27) corresponded to AFLP subgroups IIA and IA, respectively. The AFLP genotyping method was more discriminating than MLEE; at least 10 different AFLP genotypes have been identified among the otherwise homogeneous ET1 group (6) (data not shown). The CDC collection includes representatives of each AFLP genotype: A1 (23 CDC strains), A2 (1 CDC strain), A3 (24 CDC strains), A4 (25 CDC strains), A5 (35 CDC strains), A6 (11 CDC strains), A7 (7 CDC strains), and A8 (10 CDC strains). These AFLP genotypes were equally distributed among isolates from the four cities (Atlanta, Birmingham, Houston, and San Francisco). Consistent with the results in Table 3, the IA and rd values for the clone-corrected AFLP genotypes of these CDC isolates were 6.9 and 0.21, respectively (P < 0.01). Similarly, the IA and rd values for the AFLP genotypes of ET1 group strains were 0.93 and 0.13, respectively (P < 0.01), and those for ET2 group strains were 0.13 and 0.009, respectively (P = 0.18).
Molecular epidemiology of clinical isolates in North Carolina. Among the 42 patients treated for cryptococcosis at Duke University Hospital (Durham, N.C.) between 2000 and 2002, 16 (38%) had AIDS, 10 (24%) had undergone solid-organ transplantation, 5 (12%) had hematological malignancies, 10 (24%) had other conditions (e.g., diabetes, lupus, or sarcoidosis), and 1 had no defined underlying disease. Twenty-six (62%) of the patients were men. Of the 39 patients whose ethnicities were known, 22 (56%) were African American and the others were Caucasian. The median age of the patients was 47 years, and the age range was 25 to 82 years. There was no obvious correlation between any particular genotype and the underlying disease, gender, or age of the patient (data not shown). Multiple isolates were obtained from six patients, and with one exception, isolates from each patient had the same genotype. The exception was patient A, who was admitted to Duke Hospital in February 2002, when a strain of genotype A1 was isolated from the bone marrow; the patient was subsequently readmitted in August 2002, and a strain of genotype A3 was isolated from the patient's spinal fluid.
Environmental and clinical isolates of serotypes A and D have the MAT mating-type allele. The mating-type allele of each isolate was determined by PCR amplification with primers specific to portions of the mating type-specific genes (STE20a and STE20 or STE3a and STE3) of each serotype. All the environmental and clinical isolates of serotype A produced the signature amplicon of STE20 (20, 26), indicating they had the MAT mating-type allele (data not shown). All the environmental and clinical isolates of serotype D produced a 600-bp amplicon characteristic of a fragment of STE3, indicating that they also possessed the MAT mating-type allele (data not shown). Several serotype D strains failed to generate any amplicon from PCRs with the primers specific for STE20a and STE20, but characteristic amplicons were produced by the primer pairs specific for STE3a and STE3. The distribution of the mating-type alleles among the AD hybrids is presented in Table 4.
Both serotype A MATa and serotype D MATa alleles were detected among the hybrids. The AD1 hybrid genotype, which was isolated from two different locations in Durham, N.C., was the most abundant hybrid genotype in the environment (Table 4). Analysis by flow cytometry indicated that all the AD hybrids possessed approximately diploid amounts of DNA. When the mating types of the AD1 isolates were tested, serotype A STE20-specific primers produced the characteristic amplicon (data not shown); however, both STE20a- and STE3a-specific primers failed to produce any products for the serotype D component of the AD1 hybrid. Moreover, Southern blot hybridizations with a STE20a-specific probe confirmed the absence of ether serotype A or serotype D MATa alleles (data not shown). Unexpectedly, hybridization with the STE20-specific probes revealed the presence of both serotype A and serotype D STE20 alleles (data not shown), suggesting that the genotype AD1 strain has an unusual serotype A MAT and serotype D MAT mating type. However, more research is necessary to parse the structure and origin of the mating-type loci of this unusual strain.
Population genetics analysis of AFLP data. The existence of identical AFLP genotypes in both clinical and environmental samples from North Carolina and other geographic areas (Fig. 2 and 3), as well as the absence or rarity of MATa isolates in the environment, suggest that these populations of C. neoformans are predominantly clonal. To assess the possibility of recombination in the population, we performed statistical tests for linkage disequilibrium among the AFLP loci. Since asexual reproduction is well documented in C. neoformans, linkage disequilibrium tests were conducted with the clone-corrected data that included only one representative of each AFLP genotype (20, 22). IA and rd values were estimated separately for genotypes of serotype A, D, and AD and for each subgroup of serotype A (Fig. 1) (20, 22).
When all 12 serotype A AFLP genotypes were included in the analysis, IA and rd values were significantly greater than 0, rejecting the null hypothesis of recombination and suggesting clonality (Table 4). However, when both statistics were calculated for the two genetically isolated subgroups (Fig. 1) (20), the null hypothesis of recombination was rejected for subgroup I (P < 0.01) but not for subgroup II (P = 0.27). The null hypothesis of recombination was also rejected for the serotype AD hybrids, but it was not rejected for the population of serotype D (P = 0.1) (Table 3).
Nonrandom association between alleles in the population was also evaluated by phylogenetic methods (8, 13). As mentioned previously, the majority-rule consensus tree constructed from the AFLP data was well resolved (Fig. 1A). Comparison of the tree lengths observed for the individual serotypes with the tree lengths of 500 artificially recombined data sets determined that the tree lengths observed for the serotype A populations and for the AD hybrids were significantly shorter than the tree lengths observed for the recombined data sets (32 to 65 [P = 0.01] and 20 to 24 [P = 0.01], respectively) (Table 3). However, neither the tree lengths observed for subgroups I and II of serotype A nor those observed for the serotype D population were significantly different from the tree lengths of the randomized data, indicating that the null hypothesis of recombination cannot be rejected for these groups.
DISCUSSION
Although serotypes A, D, and AD of C. neoformans can be isolated from the environment worldwide, neither the natural reservoir nor the source of human infection has been determined. This investigation confirmed that C. neoformans is common in the environment in North Carolina, where it is clearly associated with weathered pigeon excreta, implying that pigeon habitats may provide a source of human infection.
The association between C. neoformans and pigeons has been reported frequently (9, 11, 27, 28, 46). However, pigeons do not acquire cryptococcosis, most likely because C. neoformans cannot grow at the pigeon's normal body temperature of 42°C. Nevertheless, the yeasts can survive passage through the pigeon intestinal tract (19), and yeast cells of C. neoformans were found to survive for 2 years in moist or dry pigeon excreta that was protected from the sun (12). In soil, C. neoformans may compete with certain microorganisms and can be inhibited by others, such as amoeba, which can devour the yeast (7). Thus, a plausible explanation for the intimate association between serotypes A, D, and AD of C. neoformans and pigeons is that pigeon excreta offer an advantageous ecological niche for C. neoformans. Both pigeons and C. neoformans are widely distributed and indigenous to the same areas in nature (21). Conversely, C. neoformans may have a different ecological niche that has not been defined, and although the yeast is unquestionably enriched by growth on pigeon excreta, its main reservoir may not depend upon the pigeon habitat. Perhaps the strongest circumstantial evidence for a major alternative ecological site is that subgroup I isolates of serotype A were prevalent in the clinical populations, but they were rarely isolated from the environmental samples (Fig. 1 and 3).
We investigated many potential, alternative ecological niches that have been reported to harbor C. neoformans, including the excreta of other birds, several species of decaying trees, soil, and house dust. The only one of these sites that was positive was the hollow of a white oak tree, from which a single serotype D genotype isolate was cultured. Isolates of serotypes A, B, and C have been associated with wood debris in Australia, south Asia, South America, and North America (15, 29, 33, 39, 40); however, to our knowledge, this is the first report of the isolation of a strain other then serotype B from wood debris in the United States.
The relative abundance of C. neoformans in the environment varied among the geographical regions: it was more common in North Carolina than in California or Texas. Both serotype A and D isolates, as well as serotype AD hybrids, were present in environmental and clinical samples. Isolates of serotype A dominated both types of samples, but the fraction of isolates that were serotype D or AD was the same in the environmental and clinical samples.
All of the serotype A isolates were genetically separated into two isolated subgroups. Most of the serotype AD hybrid strains could be ascribed to one of these subgroups, depending on the genetic background of their serotype A counterparts. Preferential clustering of AD hybrids with either one of the serotype A clades supports the previously proposed hypothesis of the multiple origins of the AD hybrid strains (4, 43) and suggests that isolates from both subgroups are potentially capable of hybridization with serotype D strains.
The existence of genetically distinct subgroups within the serotype A clade has been reported elsewhere (4, 6, 23). The mechanisms that promote genetic isolation of these two subgroups are not well understood. Isolates of subgroup I were quite rare in the environment (Fig. 1, 2, and 3), but they were common in both the North Carolina and the CDC clinical populations (Fig. 3). Several hypotheses could explain the relative scarcity of subgroup I isolates in the environment. First, sampling error may have produced a spurious, biased distribution of the subgroups in the environment. In the laboratory, isolates of both subgroups produce pigmented colonies with comparable rapidities and intensities on nigerseed agar; therefore, the isolation method was unlikely to be more selective for subgroup II. Second, isolates from subgroup I may have a different ecological niche that was inadvertently not sampled in this study. As noted above, several reports have implicated other potential environmental sources of C. neoformans, such as other birds, domestic dust (27, 34), and different tree species (27, 29, 39). Although we did not isolate C. neoformans serotype A from such sites, subgroup I strains may be associated with these or other sites. Third, because a small number of isolates of AFLP genotypes A6 and A7 from subgroup I were isolated from pigeon excreta, perhaps strains of both subgroups coexist in the same ecological niche but subgroup I strains may be overrepresented in the clinical samples because they are more likely to cause infection in humans. These hypotheses can be tested by evaluation of larger clinical samples, more extensive sampling of other potential environmental niches, and comparison of the virulence of different strains from both subgroups in animals.
The population structure of a sample of environmental isolates of C. neoformans from North Carolina was determined, and the results suggest significant clonality and a low level of genotypic diversity in this population. Clonality is supported by the overrepresentation of a few genotypes within the population (8, 22, 37, 38). The genotypic diversities were low in both environmental and clinical samples (Table 1), indicating an excess of particular genotypes in the environment. Only 12 different genotypes were identified from a sample of more than 800 serotype A isolates, and only 7 unique genotypes were isolated from the environment in North Carolina. Similarly, only five different genotypes were identified among serotype D and serotype AD hybrid isolates (Table 1). These results also confirm the findings of Brandt et al. (5, 6), who found widespread clonality and little correlation of the genotype with the underlying disease or specimen; however, one genotype was overly represented in cutaneous specimens. For comparison, greater genotypic diversity (0.95) and 34 unique genotypes were found in a much smaller clinical population of serotype A isolates from Botswana (20).
IA and rd were calculated for the AFLP multilocus genotypes in the North Carolina populations, and the results provided evidence for both clonality and recombination. Both statistics were significantly greater than 0 for the combined serotype A sample, as well as for the serotype AD hybrid data set; however, when IA and rd were calculated independently for serotype D or serotype A subgroups I and II, they were nearly 0, indicating linkage equilibrium among the loci in these samples. Similar results were obtained by determination of the length of the most parsimonious tree (Table 3) (8).
Burt et al. (8) calculated IA and rd values for the 222 isolates of serotype A from the CDC collection of clinical strains (6), and they found no statistically significant linkage disequilibrium among the loci, which provides some evidence of recombination in this population (8, 38). We used AFLP genotypes to calculate IA and rd for 136 of the CDC serotype A isolates (5, 6, 8, 38). In that computation, both IA and rd were significantly greater than 0 for the combined subgroup I and subgroup II sample (P < 0.01), as well as for the subgroup I sample alone (P < 0.01), but they were nearly 0 for subgroup II (P = 0.18), which was consistent with the data obtained for the North Carolina sample (Table 3) as well as the results of other investigators (8, 37).
Linkage equilibrium among the loci in the individual subgroups provides evidence for recombination within these populations. However, as little as 1% recombination can contribute to the apparent linkage equilibrium among the loci in a population (22, 38). Moreover, in most cases, it is impossible to distinguish between contemporary recombinational events and recombination that occurred in the evolutionarily distant past (20, 38). Therefore, considering the apparent dearth of MATa mating alleles in the population and the occurrence of strains with identical genotypes at geographically disparate locations, it is unlikely that sexual recombination occurs to any large extent in the population of C. neoformans in the mideastern region of the United States.
The environmental isolates from North Carolina were compared with the clinical populations from the same area and the CDC collection of clinical isolates of serotype A from patients in other U.S. locations. The genotypic diversity and the total number of genotypes were higher in the clinical sample than in the environmental sample (Table 1). However, the patients treated at Duke University Hospital actually resided in a broad area that covered a large portion of the southeastern United States, well beyond the region that we sampled environmentally. Six of seven genotypes that were isolated from the North Carolina environment were also isolated from the patients, suggesting that most of the environmental strains are capable of infecting humans. Yet, one genotype (genotype A2) was very common in the environment but was not detected in patients at Duke University Hospital, and only one clinical isolate of this genotype was present in the CDC collection (Fig. 3). These findings may indicate that this particularly prevalent environmental strain is less pathogenic. Conversely, two other genotypes (genotypes A8 and A8") were common among both patients at Duke University Hospital and the CDC clinical sample, but they were not isolated from the environment. The apparent absence of these genotypes from the environment can be attributed to sampling error; however, this finding may also indicate that certain strains are relatively rare but more pathogenic. Additional experiments will investigate these alternatives.
All clinical and environmental isolates of serotypes A and D in this study possessed the MAT mating-type allele, which confirms the observations of others (42). Historically, the MAT allele has been linked to enhanced virulence in strains of serotype D (17). However, recently constructed congenic MATa and MAT strains of serotype A were equally virulent in mice (26). Although not one MATa strain of serotype A or D was found among more then 300 clinical isolates analyzed by Yan et al. (47) or more than 800 clinical and environmental isolates analyzed in this study, both serotype A and D MATa alleles were shown to exist in the environment within the serotype AD hybrid strains. Two hypotheses may explain these data. The AD hybrid strains may be the products of recent hybridization between serotypes A and D, which suggests that both serotype A and D MATa isolates are extant in the environment, even though they were not recovered. Alternatively, these strains may have persisted clonally for long periods and now represent the historical existence of MATa mating types among environmental strains. It is not possible to determine the approximate time at which the hybrids were generated, even though the phylogenetic evidence presented here and in other studies (43, 45) indicates that multiple hybridization events have occurred. The most abundant environmental AD strain (genotype AD1; Table 4), which was isolated in large numbers from two locations in Durham, N.C., possessed both serotype A and serotype D MAT alleles at the STE20 locus and had neither MATa allele at the same locus (Table 4). The origin of this obscure mating locus structure is not entirely clear. It may have resulted from a vegetative fusion between two MAT isolates of different serotypes; however, to our knowledge, such fusions have not been reported in C. neoformans. Conversely, it may be a product of hybridization between two hybrid strains or between a hybrid and a strain of serotype A or D. Such a hybridization event would have produced a tetraploid or triploid strain, which subsequently lost its extraneous chromosomes. More detailed investigation of the structure of the mating-type locus and ploidy may elucidate the derivation of this unusual strain.
This study applied molecular biology-based genotyping and population biology to investigate a large sample of environmental isolates of C. neoformans. Our investigation addressed several relevant issues regarding the diversity, prevalence, and mode of reproduction of this pathogen in the environment. The results indicated the following: (i) C. neoformans is common in the environment in North Carolina, where it is associated with desiccated pigeon excreta and decayed wood; (ii) although both serotype D and serotype AD hybrids were present in the environment, isolates of serotype A predominated in both environmental and clinical populations; (iii) all the isolates of serotypes A and D had the MAT mating-type allele; (iv) the population of serotype A in North Carolina consists of two genetically isolated subgroups, designated subgroups I and II; (v) isolates of subgroup I were common in patients but were rarely isolated from the environment; (vi) isolates of subgroup II were common in both the environment and patients; (vii) six of the seven genotypes in the environment in North Carolina were also found in the patients; however, the most prevalent environmental genotype (genotype A2) was absent from the clinical population; (viii) although the numbers were small, there was no correlation between the clinical genotypes from patients in North Carolina and their underlying disease (e.g., human immunodeficiency virus infection) or specimen type (e.g., cerebrospinal fluid); and (ix) there was clear but limited evidence for linkage equilibrium among the alleles in the clinical subgroups, but the environmental population of C. neoformans in North Carolina is predominantly clonal.
ACKNOWLEDGMENTS
We thank Barbara D. Alexander, Rachel Addison, and L. Barth Reller for the clinical isolates from Duke Hospital, Mary E. Brandt and Jianping Xu for the CDC strains, David E. Padgett for assistance with environmental sampling in coastal North Carolina, Chris Mankoff for help with the mapping software, Robert E. Marra for helpful discussions, and Lisa Bukovnik for running the AFLP reactions.
This investigation was supported by the following grants from National Institute of Health: AI 25783 and AI 44975. A.L. was supported by training grant AI 07392.
REFERENCES
Agapow, P. M., and A. Burt. 2001. Indices of multilocus linkage disequilibrium. Mol. Ecol. Notes 1:101.
Archie, J. W. 1989. A randomization test for phylogenetic information in systematic data. Syst. Zool. 38:239-252.
Benham, J., J.-U. Jeung, M. Jasieniuk, V. Kanazin, and T. Blake.1999. Genographer: a graphical tool for automated fluorescent AFLP and microsatellite analysis. J. Agric. Genomics 4:15-19. [Online.] http://www.ncgr.org/jag. (First published 1 May 1999.)
Boekhout, T., B. Theelen, M. R. Diaz, J. W. Fell, W. C. Hop, E. C. Abeln, F. Dromer, and W. Meyer. 2001. Hybrid genotypes in the pathogenic yeast Cryptococcus neoformans. Microbiology 147:891-907.
Brandt, M. E., L. C. Hutwagner, L. A. Klug, W. S. Baughman, D. Rimland, E. A. Graviss, R. J. Hamill, C. Thomas, P. G. Pappas, A. L. Reingold, R. W. Pinner, and the Cryptococcal Disease Active Surveillance Group. 1996. Molecular subtype distribution of Cryptococcus neoformans in four areas of the United States. J. Clin. Microbiol. 34:912-917.
Brandt, M. E., L. C. Hutwagner, R. J. Kuykendall, R. W. Pinner, and the Cryptococcal Disease Active Surveillance Group. 1995. Comparison of multilocus enzyme electrophoresis and random amplified polymorphic DNA analysis for molecular subtyping of Cryptococcus neoformans. J. Clin. Microbiol. 33:1890-1895.
Bunting, L. A., L. B. Neilson, and G. S. Bulmer. 1979. Cryptococcus neoformans: a gastronomic delight of a soil ameba. Sabouraudia 17:225-232.
Burt, A., V. Koufopanou, and J. W. Taylor. 2000. Population genetics of human-pathogenic fungi. In R. A. Thompson (ed.), Molecular epidemiology of infectious diseases. Edward Arnold, London, United Kingdom.
Casadevall, A., and J. R. Perfect. 1998. Cryptococcus neoformans. ASM Press, Washington, D.C.
Cogliati, M., M. C. Esposto, D. L. Clarke, B. L. Wickes, and M. A. Viviani. 2001. Origin of Cryptococcus neoformans diploid strains. J. Clin. Microbiol. 39:3889-3894.
Currie, B. P., L. F. Freundlich, and A. Casadevall. 1994. Restriction fragment length polymorphism analysis of Cryptococcus neoformans isolates from environmental (pigeon excreta) and clinical sources in New York City. J. Clin. Microbiol. 32:1188-1192.
Denton, J. F., and A. Di Salvo. 1968. The prevalence of Cryptococcus neoformans in various natural habitats. Sabouraudia 6:213-217.
Farris, J. S. 1977. Phylogenetic analysis under Dollo's law. Syst. Zool. 26:77-88.
Franzot, S. P., J. S. Hamdan, B. P. Currie, and A. Casadevall. 1997. Molecular epidemiology of Cryptococcus neoformans in Brazil and the United States: evidence for both local genetic differences and a global clonal population structure. J. Clin. Microbiol. 35:2243-2251.
Halliday, C. L., and D. A. Carter. 2003. Clonal reproduction and limited dispersal in an environmental population of Cryptococcus neoformans var. gattii isolates from Australia. J. Clin. Microbiol. 41:703-711.
Haynes, K. A., D. J. Sullivan, D. C. Coleman, J. C. K. Clarke, R. Emilianus, J. Atkinson, and K. J. Cann. 1995. Involvement of multiple Cryptococcus neoformans strains in a single episode of cryptococcisis and reinfection with novel strains in recurrent infection demonstrated by random amplification of polymorphic DNA and DNA fingerprinting. J. Clin. Microbiol. 33:99-102.
Kwon-Chung, K. J., J. C. Edman, and B. L. Wickes. 1992. Genetic association of mating types and virulence in Cryptococcus neoformans. Infect. Immun. 60:602-605.
Lengeler, K., G. M. Cox, and J. Heitman. 2001. Serotype AD strains of Cryptococcus neoformans are diploid or aneuploid and heterozygous at the mating-type locus. Infect. Immun. 69:115-122.
Littman, M. L., and R. Borok. 1968. Relation of pigeon to cryptococcosis: natural carrier state, heat resistance and survival of Cryptococcus neoformans. Mycopathol. Mycol. Appl. 35:329-345.
Litvintseva, A. P., R. E. Marra, K. Nielsen, J. Heitman, R. Vilgalys, and T. Mitchell. 2003. Evidence of sexual recombination among Cryptococcus neoformans serotype A isolates in sub-Saharan Africa. Eukaryot. Cell 2:1162-1168.
Long, J. L. 1981. Introduced birds of the world: the worldwide history, distribution and influence of birds introduced to new environments. David & Charles, London, United Kingdom.
Maynard Smith, J. M., N. H. Smith, M. O'Rourke, and B. Spratt. 1993. How clonal are bacteria Proc. Natl. Acad. Sci. USA 90:4384-4388.
Meyer, W., A. Castaeda, S. Jackson, M. Huynh, and E. Castaeda. 2003. Molecular typing of IberoAmerican Cryptococcus neoformans isolates. Emerg. Infect. Dis. 9:189-195.
Mitchell, T. G., and J. R. Perfect. 1995. Cryptococcosis in the era of AIDS—100 years after the discovery of Cryptococcus neoformans. Clin. Microbiol. Rev. 8:515-548.
Nei, M., and W.-H. Li. 1979. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proc. Natl. Acad. Sci. USA 76:5269-5273.
Nielsen, K., G. Cox, P. Wang, D. Toffaletti, J. R. Perfect, and J. Heitman. 2003. Sexual cycle of Cryptococcus neoformans var. grubii and virulence of congenic a and isolates. Infect. Immun. 71:4831-4841.
Nishikawa, M. M., M. Lazera, G. G. Barbosa, L. Trilles, B. Balassiano, R. Macedo, C. F. Bezerra, M. A. Perez, P. Cardarelli, and B. Wanke. 2003. Serotyping of 467 Cryptococcus neoformans isolates from clinical and environmental sources in Brazil: analysis of host and regional patterns. J. Clin. Microbiol. 41:73-77.
Nosanchuk, J. D., J. Rudolph, A. L. Rosas, and A. Casadevall. 1999. Evidence that Cryptococcus neoformans is melanized in pigeon excreta: implications for pathogenesis. Infect. Immun. 67:5477-5479.
Randhawa, H. S., T. Kowshik, and Z. U. Khan. 2003. Decayed wood of Syzygium cumini and Ficus religiosa living trees in Delhi/New Delhi metropolitan area as natural habitat of Cryptococcus neoformans. Med. Mycol. 41:199-209.
Sambrook, J., and D. W. Russell. 2001. Molecular cloning: a laboratory manual, vol. 1, 3rd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
Sorrell, T. C., S. C. Chen, P. Ruma, W. Meyer, T. J. Pfeiffer, D. H. Ellis, and A. G. Brownlee. 1996. Concordance of clinical and environmental isolates of Cryptococcus neoformans var. gattii by random amplification of polymorphic DNA analysis and PCR fingerprinting. J. Clin. Microbiol. 34:1253-1260.
Staib, F., M. Seibold, E. Antweiler, and B. Frohlich. 1989. Staib agar supplemented with triple antibiotic combination for the detection of Cryptococcus neoformans in clinical specimens. Mycoses 32:448-454.
Stephen, C., S. Lester, W. Black, M. Fyfe, and S. Raverty. 2002. Multispecies outbreak of cryptococcosis on southern Vancouver Island, British Columbia. Can. Vet. J. 43:792-794.
Swinne, D., M. Deppner, R. Laroche, J. J. Floch, and P. Kadende. 1989. Isolation of Cryptococcus neoformans from houses of AIDS-associated cryptococcosis patients in Bujumbura (Burundi). AIDS 3:389-390.
Swofford, D. L. 2002. PAUP: phylogenetic analysis using parsimony (and other methods), version 4.0b. Sinauer Associates, Inc., Sunderland, Mass.
Tanaka, R., H. Taguchi, K. Takeo, M. Miyaji, and K. Nishimura. 1996. Determination of ploidy in Cryptococcus neoformans by flow cytometry. J. Med. Vet. Mycol. 34:299-301.
Taylor, J. W., D. M. Geiser, A. Burt, and V. Koufopanou. 1999. The evolutionary biology and population genetics underlying fungal strain typing. Clin. Microbiol. Rev. 12:126-146.
Taylor, J. W., D. Jacobson, and M. Fisher. 1999. The evolution of asexual fungi: reproduction, speciation, and classification. Annu. Rev. Phytopathol. 37:197-246.
Trilles, L., M. Lazera, B. Wanke, B. Theelen, and T. Boekhout. 2003. Genetic characterization of environmental isolates of the Cryptococcus neoformans species complex from Brazil. Med. Mycol. 41:383-390.
Vilcins, I., M. Krockenberger, H. Agus, and D. A. Carter. 2002. Environmental sampling of Cryptococcus neoformans var. gattii from the Blue Mountains National Park, Sydney, Australia. Med. Mycol. 40:53-60.
Viviani, M. A., M. C. Esposto, M. Cogliati, M. T. Montagna, and B. L. Wickes. 2001. Isolation of a Cryptococcus neoformans serotype A MATa strain from the Italian environment. Med. Mycol. 39:383-386.
Viviani, M. A., R. Nikolova, M. C. Esposto, G. Prinz, and M. Cogliati. 2003. First European case of serotype A MATa Cryptococcus neoformans infection. Emerg. Infect. Dis. 9:1179-1180.
Xu, J., G. Luo, R. J. Vilgalys, M. E. Brandt, and T. G. Mitchell. 2002. Multiple origins of hybrid strains of Cryptococcus neoformans with serotype AD. Microbiology 148:203-212.
Xu, J., and T. G. Mitchell. 2003. Population genetic analyses of medically important fungi, p. 703-722. In D. H. Howard (ed.), Pathogenic fungi in humans and animals, 2nd ed. Marcel Dekker, Inc., New York, N.Y.
Xu, J., R. J. Vilgalys, and T. G. Mitchell. 2000. Multiple gene genealogies reveal recent dispersion and hybridization in the human pathogenic fungus Cryptococcus neoformans. Mol. Ecol. 9:1471-1481.
Yamamoto, Y., S. Kohno, H. Koga, H. Kakea, K. Tomono, M. Kaku, T. Yamazaki, M. Arisawa, and K. Hara. 1995. Random amplified polymorphic DNA analysis of clinically and environmentally isolated Cryptococcus neoformans in Nagasaki. J. Clin. Microbiol. 33:3328-3332.
Yan, Z., X. Li, and J. Xu. 2002. Geographic distribution of mating type alleles of Cryptococcus neoformans in four areas of the United States. J. Clin. Microbiol. 40:965-972.