Clonality and Recombination in Genetically Differentiated Subgroups of Cryptococcus gattii

All strains used in this study are listed in the supplemental material (see Table S1 in the supplemental material). Isolates from PNG and the NT were from human clinical infections (Fig. 1). The majority of the NT isolates were from indigenous patients. The Sydney isolates were from a variety of veterinary sources including domestic cats and dogs and some native wildlife (30). Mating type reference strains CBS5757 (serotype B, MATα) and CBS6998 (serotype B, MATa) were obtained from the Centraalbureau voor Schimmelcultures culture collection, and molecular type standard reference strains WM179 (VGI/AFLP 4), Ram2 (VGII/AFLP 6), WM161 (VGIII/AFLP 5), and WM779 (VGIV/AFLP 7) were from the Molecular Mycology Research Laboratory culture collection at Westmead Hospital. Isolates were obtained from source laboratories as either freeze-dried cells or on plates and were cultured on Sabouraud dextrose agar.

Chromosomal DNA extraction was based on the Novozyme 234 dodecyltrimethylammonium bromide and hexadecyltrimethylammonium bromide method described previously by Wen et al. (44), with the following modifications: approximately 0.75 g of cells (wet weight) grown on Sabouraud dextrose agar was collected, the protoplasting solution was made with 10 mg/ml of Novozyme 234 in SCE buffer (100 mM sodium citrate, 1 M sorbitol, 10 mM EDTA), and all centrifugation steps were performed at 13,000 × g. The DNA pellet was resuspended in 100 μl of Tris-EDTA buffer (100 mM Tris-HCl, 0.1 mM EDTA [pH 8]). DNA was diluted 1:10 for PCR amplification.

Mating type analysis was carried out by PCR amplification using MFα primers (upper and lower sets) and the STE20_SFa primer sets (15, 17) which are specific to the mating type regions of α and a mating type cells, respectively. PCR amplifications were performed in 50-μl reaction mixtures containing 1× PCR buffer (10 mM Tris-HCl [pH 8.3], 50 mM KCl, 1.5 mM MgCl₂, 0.001% gelatin), 5% glycerol, 6.25 μM concentrations of each deoxynucleoside triphosphate, 0.2 μM concentrations of each primer, 2.5 U of Taq DNA polymerase, and 1 μl of diluted template DNA. Amplification conditions for PCR were 94°C for 5 min followed by 30 cycles of 94°C for 1 min, 40°C for 1 min, and 72°C for 1 min and a final extension step at 72°C for 7 min. All amplifications were carried out in a Perkin-Elmer 2400 thermal cycler. Ten microliters of each amplification product was electrophoresed at 10 V/cm in 2% agarose gels containing 0.5 ng/ml ethidium bromide. The gels were visualized by UV transillumination and photographed. Culture collection strains of known mating type were used as controls (see Table S1 in the supplemental material).

Molecular type was determined by DNA fingerprinting and restriction fragment length polymorphism (RFLP) analysis. Fingerprinting was performed using the simple repetitive sequence primer (GACA)₄ (2, 29). Amplification conditions were 94°C for 5 min followed by 35 cycles of 94°C for 20 seconds, 50°C for 30 seconds, and 72°C for 20 seconds and a final extension step at 72°C for 7 min. Ten microliters of each amplification product was electrophoresed on 2% agarose Tris-acetate-EDTA gels and visualized by UV transillumination. The protocol used to determine molecular type via RFLP was modified from a method described previously by Meyer et al. (28). The URA5 gene was amplified in 50 μl containing 50 ng DNA, 1× PCR buffer (10 mM Tris-HCl, [pH 8.3], 50 mM KCl, 1.5 mM MgCl₂), 0.2 mM of deoxynucleoside triphosphate, 3 mM MgAc, 1.5 units AmpliTaq DNA polymerase, and 50 ng of each of the primers URA5 and SJO1. PCR amplification was 1 cycle of 94°C for 3 min and 35 cycles of 94°C for 45 s, 61°C for 1 min, and 72°C for 2 min followed by 1 cycle of 72°C for 10 min. PCR products were digested with Sau96I and HhaI for 3 h and separated on 3% agarose gels by electrophoresis at 10 V/cm for 2 h. Standard strains of known molecular type were used as controls (see Table S1 in the supplemental material).

The AFLP fingerprinting protocol was based on the technique described previously by Vos et al. (43) AFLP primers EcoRI-TG-MseI-CA and EcoRI-CA-MseI-GT were used with minor modifications to the procedure described previously by Halliday and Carter (16). All amplifications were carried out in a Perkin-Elmer 2400 thermal cycler.

A total of 0.5 μl of Genescan-400HD ROX standard size marker (Applied Biosystems) was mixed with 10 μl of Hi-Di Formamide (Applied Biosystems) and aliquoted into 0.5-ml Eppendorf tubes. Selective amplification products were diluted 100-fold in distilled H₂O, and 1 μl of this was added to the ROX standard. Samples were then denatured at 94°C for 5 min and placed on ice for 5 min before being loaded onto an ABI 3700 analyzer. Samples were autoanalyzed using the GS400HD analysis module. Data collection, fragment sizing, and pattern analyses were done with GeneScan, version 3.1, analysis software (Applied Biosystems). Polymorphic loci were defined as bands of the same mobility present in some isolates and absent in others. For each polymorphic locus, there were two possible alleles, which were scored as 1 when the amplified fragment was present and 0 when the fragment was absent. Data matrices were produced using LecPCR (40). A polymorphic locus was included in the analysis only if it was present or absent in at least five of the isolates, was strongly amplified, and was greater than 40 bp and less than 500 bp in size.

The index of association (I_A) (32), RbarD (1), and tree length (T_L) tests (5) were used to distinguish between recombining and clonal modes of reproduction. I_A and RbarD are statistical tests that measure the degree of nonrandom association between alleles at different loci (linkage disequilibrium), with the RbarD algorithm adjusting for limited numbers of loci, and were calculated by using the Multilocus, version 1.0b, software (1). The T_L test uses the permutation test in PAUP* (version 4.0b4a) to calculate the length of phylogenetic trees by treating the isolates as taxa and the alleles at each locus as phylogenetic characters with two character states. Both analyses involved comparing the values for the observed data set with the values for 1,000 artificially recombining data sets. Artificially recombining data sets were constructed by randomly shuffling the alleles for each locus between members of the population while keeping the proportions of alleles at each locus constant. The inability to distinguish between the observed data set and the artificially recombined data sets supports the null hypothesis of sexual recombination, whereas a significant difference between the data sets supports clonality (5, 6, 16). Genetic differentiation between populations, assessed as theta (θ), which is an estimate of Wright's F_st, the standard measure of population subdivision, was also calculated using the Multilocus, version 1.0b, software (1). This compares loci present in different populations to give a measure of the amount of gene flow between populations. If different populations have the same allele frequencies at all loci, then θ will be zero, indicating no genetic differentiation. Conversely, if the populations are fixed for different alleles at all loci, then θ will be 1, indicating total genetic differentiation. The significance of θ is again assessed by comparing the observed data set with 1,000 artificially produced data sets in which the alleles are randomized among populations.

DNA fingerprinting using (GACA)₄ primers and RFLP analysis of the URA5 gene were used to generate digestion profiles for the 81 isolates which were compared with those produced for each molecular type standard (Fig. 2). Of the 81 isolates, 70% were molecular type VGI, 25% were VGII, and 5% were VGIII. Within the PNG population, the vast majority of isolates (84%) were molecular type VGI, 5% were VGII, and 11% were VGIII. In the NT population, VGI isolates were outnumbered by VGII by 43% to 57%. The Sydney population contained 74% VGI and 26% VGII isolates (Table 1). No VGIII isolates were found in the Australian populations.

MFα primers amplified a 109-bp fragment in the culture collection strain of a known α mating type (CBS5757) but failed to amplify any fragment in the control a mating type strain (CBS6998). The STE20a_SF primers were similarly used to amplify a 219-bp fragment from the a mating type only. Mating type analysis of the 81 isolates indicated a significant bias towards the α mating type, with a 77:4 MATα:MATa ratio. Within population groups, the ratio of MATα:MATa varied. In the PNG population, this was 35:2; in the NT population, the ratio was 19:2; and the Sydney veterinary population did not include any MATa isolates among the 23 isolates tested (Table 1). All MATa strains were molecular type VGI.

A total of 55 polymorphic loci were identified across the isolates from the three populations with allele frequencies varying in each population (see Table S2 in the supplemental material). Seventy-six percent of loci were polymorphic in all populations. Five loci were polymorphic in the PNG and NT populations but were fixed in the Sydney population, four were restricted to the NT and Sydney populations, and two were restricted to the PNG and Sydney populations. The PNG and NT populations each had a single polymorphic locus that was fixed in the other two populations. A comparison of molecular types revealed nine loci that were polymorphic in only the VGI isolates and four that were unique to VGII.

The phylogenetic analysis program PAUP* (version 4.0b4a) (38) was used to produce a parsimony phylogram from PAUP*-compatible files generated by Multilocus (version 1.0b) software (1) (Fig. 3). Only bootstrap values greater than 60 were included on the phylogram. Most isolates clustered first according to molecular type and then according to geographic location. Three PNG VGI isolates and one VGII isolate from the NT branched with the two VGII and four VGIII isolates from PNG (Fig. 3). All these isolates had relatively distinct genetic profiles, and long-branch attraction may be responsible for this apparent clustering.

I_A, RbarD, and T_L values were generated for each molecular type containing five or more isolates within each of the three populations. The analysis was confined to molecular type groups, as the significant level of genetic variation between molecular types would skew the results in favor of clonality (13). I_A, RbarD, and T_L values for all VGI isolates and for each geographically separate VGI population were significantly different from their associated recombining data sets, indicating clonality (Table 2). Tests of subgroups, evident on the phylogram (indicated by branches labeled C1 to C4 in Fig. 3), also gave strongly clonal results. In contrast, the Sydney VGII veterinary population returned values for each test that lay well within the range of values obtained for the recombined data sets, and the null hypothesis of recombination could not be rejected (P = 0.428 to 0.488). When isolate V21, which grouped more closely with two NT isolates (NT14 and H16), was removed from the analysis, the P value increased significantly (P = 0.974 to 0.999). Analysis of the complete NT VGII population gave a significant difference between the observed and the randomly generated data sets indicating clonality. Analysis was then conducted on subgroups evident on the phylogram (NT1 and NT2) within this population. The eight NT VGII isolates grouping on branch NT1 (Fig. 3) still indicated clonality (P = 0.001); however, five isolates within this group, on branch NT2 (Fig. 3), gave values that strongly suggested recombination (P = 0.849 to 0.989) (Table 2).

The θ values obtained for VGI groups in Sydney, the NT, and PNG were all highly significant, indicating strong genetic differentiation between each of these regions (Table 3). When Sydney veterinary VGII isolates were compared with all VGII isolates from the NT, significant differentiation was also seen. However, when the comparison was confined to NT isolates that branched with the veterinary isolates on the phylogram (Fig. 3), significance dropped to a P value of 0.039, indicating the possibility of allele sharing between these populations.

We report here evidence of recombination occurring in two VGII populations of C. gattii from distinct geographic regions of Australia. In Australia, VGII is a less common cause of C. gattii cryptococcosis than VGI, and there have been few isolations of VGII from the environment (3, 7, 22, 33, 34). However, a high proportion of VGII isolates from clinical samples in the Northern Territory has been reported, which was also found in the isolates analyzed in the current study. Most interestingly, VGII isolates have recently been found to be the cause of the first documented “outbreak” of cryptococcosis on Vancouver Island, British Colombia (12, 23). To date, this outbreak has resulted in over 60 human infections including 4 deaths and in excess of 250 animal infections (19, 25, 36). However, while most of the Vancouver Island VGII isolates were fertile in mating studies, where typeable, all were of the α mating type and genetic diversity was extremely restricted, suggesting a clonal epidemic structure (12, 23).

In contrast to VGII, all VGI populations had highly clonal structures. Clonality was particularly strong in the largest VGI group from Papua New Guinea. PNG has a relatively high prevalence of cryptococcal disease. This, coupled with the absence of the usual Eucalyptus host (24), made it a promising candidate as a center of origin of C. gattii in which the fungus might complete its life cycle in connection with its true ecological niche. Instead, the appearance of the PNG VGI cluster on the phylogram (Fig. 3) is more suggestive of a clonal bloom with all isolates very closely related and appearing to stem from a recent common ancestor. C. gattii has never been isolated from the environment in PNG despite concerted attempts and targeting of likely environmental niches (24). Our data suggest, as with Eucalyptus species in Australia, that a favorable substrate on which extensive clonal propagation occurs might exist. Interestingly, two a mating type strains occurred within the PNG cluster, and it is possible that some inbreeding occurs but with such closely related strains that clonality is not disrupted.

Analyzed as a whole, the NT VGII population returned a clonal population structure, and it was not until subgroups on the phylogram were identified that evidence of recombination was found. This shows the value of an initial phylogenetic analysis in which substructure within populations can be identified. Assessing recombination in a population that is genetically subdivided will skew results in favor of clonality, as any alleles that are fixed across the division will appear linked (13). Clonality can also be erroneously concluded when loci are physically linked or when clone mates are oversampled from a population (32). It is more difficult to incorrectly conclude that recombination has occurred, although this can happen if high levels of homoplasy, due, for example, to hypermutation, occur (20).

Our major concern in the current study was that the recombining populations found were small, which in turn reduced the number of polymorphic loci. I_A and RbarD values were generated for several small (n = 5) randomly chosen groups of VGI isolates selected from clonal populations; these consistently remained clonal, indicating that population size was not affecting our results (data not shown). Homoplasy was also considered unlikely, as the level of genetic diversity observed in the recombining populations was not particularly high, with conserved loci outnumbering polymorphic loci by 28 to 6 in the NT population and 40 to 10 in the Sydney population.

The most significant evidence supporting recombination in these populations comes from the concurrent study by us and colleagues investigating mating in Australian C. gattii strains (5a). That study identified 27 fertile Australian isolates capable of mating, and of these isolates, 18 have been included in the current analysis. The overwhelming majority of strains showing any sign of fertility and all “robust maters” were of molecular type VGII. When superimposed on the results of the indirect analysis used in this study, the recombining populations and the isolates shown to be capable of mating in the laboratory were strongly correlated (Fig. 3). However, several NT isolates outside the recombining group were also fertile. It is possible that these were sampled from separate, genetically unconnected populations in which recombination occurs. The Northern Territory is a very large region (almost twice the size of the state of Texas), and individual infections might be acquired from very geographically separate areas or even outside this region, given the propensity of humans to travel.

Previous studies have indicated widespread homogeneity of C. gattii throughout Australia (34, 35). The development of more highly discriminating molecular techniques, combined with powerful statistical analyses, revealed significant genetic differentiation between VGI isolates from Sydney, NT, and PNG. This result was not surprising considering the geographic distance between these populations. Genetic differentiation between the VGII populations was considerably lower, and only borderline differentiation was seen between the Sydney population and NT isolates that branched with it on the phylogram (Fig. 3) (P = 0.039). Again, the number of isolates in this group is small, and further analysis with additional isolates will help clarify whether any genetic exchange is occurring. Aerial dispersal over hundreds of kilometers is presumed to have occurred with some plant pathogens (4), and it is possible that sexual spores could allow long-distance dispersal of C. gattii.

C. gattii populations in Australia appear to reproduce clonally or sexually, and in the current study, together with data in the accompanying paper on mating, this segregates strongly with molecular type. These two different reproductive modes influence the ecology of C. gattii in the environment, which may in turn have implications for the acquisition of cryptococcal disease. First, our recombination analysis together with studies of C. neoformans suggest that although the pathogenic cryptococci are sexual species, extensive clonal propagation occurs on certain favorable substrates: pigeon guano for C. neoformans, decaying eucalyptus detritus in tree hollows for C. gattii VGI in Australia, and unknown substrates for C. gattii VGI in PNG and VGII on Vancouver Island. The outbreak on Vancouver Island and the high incidence of C. gattii cryptococcosis in PNG indicate that asexually reproducing C. gattii isolates can be an important source of cryptococcal infection. A pertinent question remains regarding the nature of the infective propagule, which is considered unlikely to be a vegetatively produced encapsulated yeast cell (45). Basidiospores are produced via monokaryotic fruiting from some C. neoformans var. neoformans strains, and interestingly, this process has recently been shown to involve diploidization of cells of the same mating type (α) and the development of recombinant progeny (26); however, basidiospore development via monokaryotic fruiting has never been seen in C. gattii (46), but it is possible that particular environmental conditions are required to initiate this process.

Second, our study has identified two recombining populations, and given the data on fertility, it is likely that all of the Australian VGII isolates in this study have a sexual origin. The two geographic VGII populations appear quite different in their association with cryptococcosis. In Sydney, VGII isolates have rarely been reported from human infections (34), yet this molecular type was relatively common among the veterinary isolates. This might reflect a difference in host habits, as animals generally spend more time outdoors and in closer proximity to vegetation than humans. In contrast, the NT has one of the highest incidences of C. gattii cryptococcosis in the world (10, 21). It is tempting to speculate that this might be related to the presence of fertile and sexually active VGII isolates and the associated production of infectious basidiospores. If C. gattii VGII appears to be adapting and migrating into more temperate areas, such as Vancouver Island, this could have implications for the further emergence of C. gattii cryptococcosis.