Molecular Genetic Analyses of Mating Pheromones Reveal Intervariety Mating or Hybridization in Cryptococcus neoformans

Abstract

The sexual mating of the pathogenic yeast Cryptococcus neoformans is important for pathogenesis studies because the fungal virulence is linked to the α mating type (MATα). We characterized C. neoformans mating pheromones (MFα 1 and MFa1) from 122 strains to understand intervariety hybridization or mating and intervariety virulence. MFα 1 in three C. neoformans varieties showed (a) specific nucleotide polymorphisms, (b) different copy numbers and chromosomal localizations, and (c) unique deduced amino acids in two geographic populations of C. neoformans var. gattii. MFα 1 of different varieties cross-hybridized in Southern hybridizations. Their phylogenetic analyses showed purifying selection (neutral evolution). These observations suggested that MATα strains from any of the three C. neoformans varieties could mate or hybridize in nature with MATa strains of C. neoformans var. neoformans. A few serotype A/D diploid strains provided evidence for mating or hybridization, while a majority of A/D strains tested positive for haploid MFα 1 identical to that of C. neoformans var. grubii. MFα 1 sequence and copy numbers in diploids were identical to those of C. neoformans var. grubii, while their MFa1 sequences were identical to those of C. neoformans var. neoformans; thus, these strains were hybrids. The mice survival curves and histological lesions revealed A/D diploids to be highly pathogenic, with pathogenicity levels similar to that of the C. neoformans var. grubii type strain and unlike the low pathogenicity levels of C. neoformans var. neoformans strains. In contrast to MFα 1 in three varieties, MFa1 amplicons and hybridization signals could be obtained only from two C. neoformans var. neoformans reference strains and eight A/D diploids. This suggested that a yet undiscovered MFa pheromone(s) in C. neoformans var. gattii and C. neoformans var. grubii is unrelated to, highly divergent from, or rarer than that in C. neoformans var. neoformans. These observations could form the basis for future studies on the role of intervariety mating in C. neoformans biology and virulence.

Cryptococcus neoformans causes cryptococcal meningoencephalitis in healthy and immunocompromised individuals. The fungus has two mating types (mating type α [MATα] and MATa), three varieties (C. neoformans var. neoformans, C. neoformans var. gattii, and C. neoformans var. grubii), and five serotypes (A, B, C, D, and A/D). C. neoformans var. grubii is most common in cryptococcosis among immunocompromised patients; the isolates are usually serotype A, MATα. This variety apparently reproduces asexually in nature, as indicated by the rarity of fertile MATa isolates (9, 19, 29). Sexual mating is seen in the laboratory both in C. neoformans var. neoformans (serotype D) and C. neoformans var. gattii (serotype B or C). The latter two varieties also show laboratory cross-hybridization; thus, their teleomorphs are classified as Filobasidiella neoformans var. neoformans and F. neoformans var. bacillispora (30). Currently, C. neoformans sexual mating is being intensively studied to define the observed linkage between fungal virulence and MATα and to understand the signaling cascades that regulate mating, sporulation, and virulence (31, 33).

Fungal pheromones are lipopeptides, which, along with their receptors, act as the primary determinants of mating specificity (4, 8, 45). Available evidence suggests that MATα pheromones share nucleotide homologies with sister fungal taxa and can functionally replace one another in transgene experiments (37, 48). MATa, on the other hand, is quite unique in Saccharomyces species, with no nucleotide or functional homology even among the most closely related taxa (45). C. neoformans MFα1 contains a CAAX prenylation signal motif at the carboxy terminus; this is the only common feature shared among most fungal pheromone sequences characterized to date (16, 41). Recent mapping of a MATα locus from C. neoformans var. neoformans revealed three copies of the MATα pheromone (renamed MFα1, MFα2, and MFα3) along with a mating-type specific mitogen-activated protein kinase cascade (16, 27). Lengeler et al. (32) identified STE20a (a protein kinase homolog) in the MATa mating-type locus from a serotype A strain. One of our laboratories recently characterized a C. neoformans var. neoformans MATa pheromone (36). Thus, focused studies have begun to define the genetic elements constituting C. neoformans MAT loci. However, the pheromones from three varieties have not been compared with respect to their structural and evolutionary relationships, cross-variety mating, and/or hybridization or with respect to the relevance of the latter to pathogenesis. Such studies are imperative for understanding the pathogenic differences among the three C. neoformans varieties (11, 17).

The biology and virulence of C. neoformans A/D diploid isolates were not well defined until recently, even though they were occasionally reported from clinical and environmental sources (3, 29). A number of recent publications of studies using various investigative approaches focused on the origin of A/D isolates (15, 34, 47, 52, 55). Notably, Lengeler et al. (34) found A/D isolates to have a much lower pathogenic potential than those of a control C. neoformans var. grubii strain. Cogliati et al. (15) were the only investigators among this group to use pheromone genes for analysis of A/D diploids, albeit the study relied on the use of slot blots, which yield limited information. Our laboratories reported unique restriction fragment length polymorphisms in partial MATα pheromone sequences from three C. neoformans varieties, which allowed rapid determination of variety, mating type, and ploidy (10). The present report is a comprehensive characterization of MFα1 and MFa1 pheromone sequences from 122 clinical and reference isolates designed to determine C. neoformans intervariety mating or hybridization and intervariety virulence.

MATERIALS AND METHODS

Fungal cultures.

Fifty-five C. neoformans cultures were described previously (10). Briefly, the isolates used in the present study include five type strains from three varieties and clinical and environmental isolates from around the world. The type strains were from the American Type Culture Collection (ATCC), Manassas, Va., and the New York State Herbarium (NYS), Albany, N.Y. The type strains were H99 (NYS 1649, C. neoformans var. grubii, serotype A, MATα), NIH 12 (ATCC 28957, C. neoformans var. neoformans, serotype D, MATα), NIH 430 (ATCC 28958, C. neoformans var. neoformans, serotype D, MATa), NIH 433 (ATCC 34875, C. neoformans var. neoformans, serotype D, MATa), NIH 191 (ATCC 32608, C. neoformans var. gattii, serotype C, MATa), and NIH 444 (ATCC 32609, C. neoformans var. gattii, serotype B, MATα). We also included 33 C. neoformans var. gattii (serotypes B and C) isolates from our ongoing investigation of C. neoformans molecular epidemiology in Southern California (S. Chaturvedi, R. A. Larsen, and V. Chaturvedi, unpublished data). Additionally, we examined 34 C. neoformans A/D isolates from clinical and environmental sources previously characterized using isolates from the United States by Brandt et al. (6) and from Spain by Baró et al. (3).

Morphology, serotype, and variety determination.

The gross morphology and occurrence of phenotypic switching were examined on Sabouraud dextrose agar plates (22). The growth curves of test strains were determined in yeast extract-peptone-dextrose (YEPD) broth as described previously (12). The serotypes were determined with the Crypto Check kit (Iatron Laboratories, Inc., Tokyo, Japan). The varietal status of the isolates was determined using canavanine-glycine-bromthymol blue agar, India ink preparations of C. neoformans cells were used for capsule size measurement with an ocular micrometer, and melanin production was compared on caffeic acid agar and niger seed agar (12, 29).

Mating induction.

The diploid A/D strains were investigated for self-fertility on V8 juice agar by examining wet mounts for typical hyphae, basidia, and basidiospores (12). The standard mating pair strains of F. neoformans var. neoformans (NIH 12 × NIH 430) were used as positive controls and as mating partners for test strains (29).

Pathogenicity test.

The pathogenicity of diploid A/D strains was compared in mice, along with those of the type strain of C. neoformans var. grubii (19) and congenic C. neoformans var. neoformans MATa (JEC20) and MATα (JEC21) strains received from J. C. Edman (25). Briefly, C. neoformans strains were grown exponentially in yeast nitrogen base and glucose at 30°C, washed, and resuspended in sterile 0.85% NaCl (12). Cells were counted by trypan blue exclusion using a hemacytometer and by CFU counts on YEPD agar. Stock dilutions were made to obtain 10⁶ cells in 0.1 ml of cell suspension. Five male BALB/c mice (approximately 4 weeks old, 15 to 20 g body weight) received 0.1 ml of cell suspension intravenously. The injected animals were observed for 30 days for any overt sign of illness, and all sick animals were promptly sacrificed by CO₂ inhalation. Brain tissue from dead mice was cultured on niger seed agar for C. neoformans recovery.

Histological lesions were examined in detail in mice infected as described for the survival curve experiments, except that all infected animals were euthanized at day 7. Brains from two or three mice were fixed whole in 10% neutral buffered formalin. The brains were then sliced at 2- to 3-mm intervals in the transverse plane. The brain slices were processed using routine histologic methods. The tissue sections were stained with hematoxylin and eosin and with mucicarmine (49).

SEM and image analysis.

C. neoformans isolates were grown on YEPD agar plates at 30°C for 1 week. The petri dishes were flooded with 2% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) and allowed to stand at 4°C. The fixed cells were resuspended in 0.2 M sodium cacodylate buffer and applied to a 0.4-μm-pore-diameter Nuclepore filter. The cells were washed with the same buffer, 1% osmium tetroxide was added for 20 min, washings were repeated with buffer and water, and the cells were dehydrated using a graded ethanol series. A piece of the filter was critical-point dried using liquid carbon dioxide. The dried filters were sputter coated with pure gold and viewed in an enterotoxigenic Escherichia coli (ETEC) Autoscan scanning electron microscope (SEM). All micrographs were taken at ×2,000 magnification. C. neoformans cell sizes on SEM photomicrographs were measured with a Power Macintosh G4 computer, using the public domain National Institutes of Health (NIH) Image program (http://rsb.info.nih.gov/nih-image/). A total of 50 cells were included per analysis, with all budding cells and visibly damaged yeasts excluded. The data were analyzed using SigmaStat and SigmaPlot software for Windows (SPISS Inc.). The cell sizes of A/D diploids were compared to those of the control strains, using Student's t test.

Genomic DNA and PCR.

A rapid DNA extraction procedure based on boiled yeast cells was described previously in a report from our laboratory (10). Additionally, C. neoformans genomic DNA was prepared using the DNeasy Plant Mini kit (Qiagen Inc., Valencia, Calif.). The PCR primers, DNA amplification protocol, and electrophoresis of PCR amplicons were recently described (10). We also used duplex PCR for simultaneous amplification of both MFα1 and MFa1 in two individual 50-μl reaction volumes, each of which contained 5.0 μl of PCR buffer with 15 mM MgCl₂, 2.5 μl each of two primer pairs (10 mM stock), 3.0 μl of deoxynucleoside (dNTP) mix (10 mM), and 2.0 U of Taq DNA polymerase (Perkin-Elmer). The MATα-specific 5′ oligonucleotide primer was 5′-CTTCACTGCCATCTTCACCA-3′, and the 3′ oligonucleotide primer was 5′-GACACAAAGGGTCATGCCA-3′. The MATa-specific 5′ oligonucleotide primer was 5′-CGCCTTCACTGCTACCTTCT-3′, and the 3′ oligonucleotide primer was 5′-AACGCAAGAGTAAGTCGGGC-3′. The template DNA was 5.0 μl of either a boiled cell suspension or a solution with 50 ng of genomic DNA. Initial denaturation was at 95°C for 3 min, followed by 30 cycles of denaturation at 94°C for 1 min, annealing at 57.5°C for 1 min, amplification at 72°C for 1 min, and final extension at 72°C for 7 min in a GeneAmp PCR System 9600 (Perkin-Elmer). The PCR amplicons were electrophoresed on 2.0% Nusieve-1.0% Seakem agarose (FMC BioProducts, Rockland, Maine) in TBE buffer, pH 8.3 (0.1 M Tris, 0.1 M boric acid, 0.002 M EDTA).

Nucleotide sequencing and analysis.

All sequencing was done on both strands of the DNA by using an ABI PRISM 310 or 377 sequencer with the BigDye terminator cycle sequencing ready reaction kit (Applied Biosystems, Foster City, Calif.). The sequences generated from both strands were edited and initially aligned using the GAP and PILEUP programs and other Genetics Computer Group (GCG) programs (Wisconsin Package Version 10.0-UNIX, 1999; GCG, Madison, Wis.), and the alignments were then optimized visually, based on predicted amino acid alignments.

Southern blotting and contour-clamped homogeneous electric field hybridization.

MFα1 and MFa1 probes for Southern blotting were purified by gel extraction of the respective PCR amplicons with the Qiaex II gel extraction kit (Qiagen Inc.). Restriction digestion of genomic DNA, electrophoretic transfer of nucleic acids, and hybridization to radiolabeled probes were performed according to standard procedures (2). The blots were hybridized and washed at both high (60°C) and moderate (50°C) stringencies. Pulsed-field gel electrophoresis of chromosomal DNA and hybridization were performed according to a published protocol (53).

Flow cytometry.

Flow cytometry was done with a FACScan flow cytometer (Becton Dickinson) to measure the total DNA content of diploid strains (10). The data acquisition and analysis were done with Cell Quest software. More than 20,000 cells were used to measure fluorescence intensity at each data point.

Phylogenetic analysis.

The PHYLIP program package was used to construct phylogenetic trees (18). Distance matrixes were computed using the DNADIST program with Kimura's two-parameter model, and phylogenetic trees were calculated using the additive tree model of the neighbor-joining method of Saitou and Nei (44). One thousand bootstrap replicates of the aligned sequences were generated using SEQBOOT software.

Nucleotide sequence accession numbers.

The MFα1 and MFa1 sequences have been deposited in the GenBank (National Center for Biotechnology Information, Bethesda, Md.) under the accession numbers AF226880 to AF226941 and AF501671 to AF501682; MFα1 sequences are designated MFα1A, MFα1B, MFα1C, MFα1D, and MFα1A/D, corresponding to the serotype of the parent strain.

RESULTS

Variety and specificity in MFα1.

Of the 101 nucleotides, 61 nucleotides not fixed by the PCR primers exhibited variety-specific polymorphisms (Fig. 1), thus substantiating our earlier report on the occurrence of unique restriction fragment length polymorphism in MFα1. C. neoformans var. grubii had six polymorphisms: 24T→C, 27C→T, 32G→A, 34C→T, 36C→T, and 55T→C. C. neoformans var. gattii had three polymorphisms: 38A→G, 61C→T, and 73C→T. C. neoformans var. neoformans had two polymorphisms: 54T→C and 79C→T. Additional random changes were also seen in serotype B and C strains (data not shown). An optimal alignment of MFα1-deduced amino acid sequences, and a comparison with its complete sequence as previously deposited in the database, was carried out. These data revealed unique codon signatures. C. neoformans var. grubii had ^Phe11^Ser, ^Thr12^Ile, ^Ala14^Thr, and ^Ala15^Val amino acid polymorphisms. The unique codon signature for C. neoformans var. neoformans was ^Val21^Ala, and for C. neoformans var. gattii it was ^Thr16^Ala. Another interesting observation was that of the ^His28^Asn polymorphisms, seen only in C. neoformans var. gattii isolates from California, which distinguished them from C. neoformans var. gatti isolates from Australia.

FIG. 1.

Characterization of MFα1 nucleotide sequences. The consensus sequences representing three different varieties were obtained from the alignment of MFα1 sequences of all test isolates by using a GCG software package. C. neoformans var. gattii sequences of the isolates from California (CA) and Australia (AU) revealed population-specific substitutions.

The copy numbers and genetic relatedness of MFα1 from test strains were examined by Southern hybridization (Fig. 2). HaeII-digested genomic DNAs were prepared from mating pairs of C. neoformans var. neoformans (NIH12 MATα and NIH430 MATa) and C. neoformans var. gattii (NIH444 MATα and NIH191 MATa), as well as from the haplotype of C. neoformans var. grubii (H99 MATα). The probe used was C. neoformans var. grubii partial MFα1, obtained by PCR amplification, gel purification, and ³²P labeling. All MATα isolates from reference strains hybridized with the probe. Three bands seen with C. neoformans var. neoformans were consistent with the description of Karos and coworkers (27). However, four bands were observed for C. neoformans var. grubii and only two bands were observed for C. neoformans var. gattii. The specificity of the probe was absolute for MATα isolates, since the three control MATa strains showed no hybridization in blots processed at high or moderate stringencies.

FIG. 2.

Copy numbers of MFα1 in type strains of three C. neoformans varieties. Southern blots of genomic DNA digested with HaeII and probed with ³²P-labeled, 101-bp PCR fragments of MFα1 from C. neoformans var. grubii (H99) are shown. Lambda DNA digested with HindIII was used as a size marker. NIH 12 (C. neoformans var. neoformans, serotype D, MATα) showed three bands and H99 (C. neoformans var. grubii, serotype A, MATα) revealed four bands, while NIH 444 (C. neoformans var. gattii, serotype B, MATα) yielded two bands. The MATa strains (NIH 430, NIH 433, and NIH 191) showed no hybridization signal, confirming the specificity of the MFα1 probe.

Neutral evolution and purifying selection in MFα1.

The MFα1 ratio of the number of nonsynonymous nucleotide substitutions per nonsynonymous site (Ka) to the number of synonymous nucleotide substitutions per synonymous site (Ks) was calculated (Table 1). Ka/Ks ratio values equal to 1 indicate neutrality; purifying selection occurs when Ka/Ks ratio values are substantially less than 1. The lower the Ka/Ks ratio between sequences, the more purifying selection is demonstrated. During purifying or neutral selection, detrimental amino acid changes are eliminated (46). Our data revealed that the Ka/Ks ratio values were 0.4462 and 0.6128 when C. neoformans var. grubii was compared to C. neoformans var. gattii and C. neoformans var. neoformans, respectively. Thus, this fragment of the gene was subject to relatively strong conditions of purifying selection. In contrast, the Ka/Ks ratio value for C. neoformans var. gattii and C. neoformans var. neoformans was 0.2355; thus, this fragment was also subject to strong conditions of purifying selection, although to a greater extent than that for C. neoformans var. grubii. Overall, the result strongly suggests that a part of the MFα1 sequence in the three varieties is evolutionarily conserved.

TABLE 1.

Comparison of the numbers of nonsynonymous (Ka) and synonymous (Ks) substitutions in MFα1 from three C. neoformans varieties

Varieties compared	Ka	Ks	Ka/Ks ratio
C. neoformans var. grubii vs C. neoformans var. gattii	0.0747	0.1674	0.4462
C. neoformans var. grubii vs C. neoformans var. neoformans	0.0747	0.1219	0.6128
C. neoformans var. gattii vs C. neoformans var. neoformans	0.0289	0.1227	0.2355

A representative neighbor-joining tree for MFα1 is shown in Fig. 3. C. neoformans var. grubii, C. neoformans var. gattii, and C. neoformans var. neoformans formed three separate clades strongly supported by high bootstrap values. Branch length comparisons indicated that C. neoformans var. grubii was greatly separated from the other two varieties. Both C. neoformans var. grubii and C. neoformans var. neoformans were monophyletic. Interestingly, C. neoformans var. gattii showed three sister groups, with California isolates in one cluster and the Australian isolates in the other two clusters. However, low bootstrap values precluded any definitive interpretation of this subgrouping in C. neoformans var. gattii.

FIG. 3.

Neighbor-joining phylogenetic tree of MFα1. C. neoformans var. grubii, C. neoformans var. gattii, and C. neoformans var. neoformans formed three separate clades, which were strongly supported by high bootstrap values. MFα1A/D segregated with C. neoformans var. grubii. Branch length comparisons indicated that C. neoformans var. grubii was greatly separated from the other two varieties. Both C. neoformans var. grubii and C. neoformans var. neoformans were monophyletic. Interestingly, C. neoformans var. gattii showed three sister groups, with California isolates in one cluster and the Australian isolates in the other two clusters. However, low bootstrap values precluded any definitive interpretation of this subgrouping in C. neoformans var. gattii.

MFa1 appears to be unique to C. neoformans var. neoformans.

We found MF1a amplicons in two serotype D tester strains and eight diploid A/D isolates, while no amplicon was obtained from the serotype B MATa tester strain (NIH 191). None of the other C. neoformans strains tested in this study were MATa positive. A total of 117 MATa nucleotides in the amplicon were found to be identical in all isolates.

The BamHI genomic DNA digests from the panel of strains described for Fig. 2 were probed with ³²P-labeled MFa1 from NIH 430 (Fig. 4). The probe was specific for MATa, as MATα strains showed no hybridization. BamHI digests revealed three bands from C. neoformans var. neoformans (serotype D) MATa strains NIH 430 and NIH 444. Not surprisingly, the serotype B MATa strain (NIH 191) did not hybridize with this probe, which was consistent with the negative results obtained in PCR with MFa1-specific PCR primers.

FIG. 4.

Copy numbers of MFa1 in type strains of three C. neoformans varieties. Southern blots with ³²P-labeled, 117-bp PCR-amplified MFa1 fragments from C. neoformans var. neoformans (NIH 430). Genomic DNA was digested with BamHI; lambda DNA digested with HindIII was used as a size marker. MFa1 hybridization of restriction-digested genomic DNA showed three distinct bands in NIH 430 and NIH 433. All MATα strains were negative, thus confirming the specificity of the probe. Interestingly, NIH 191 (MATa, serotype C) was also negative, suggesting a widely divergent MATa in C. neoformans var. gattii.

Similarity of A/D strains to those of C. neoformans var. grubii.

Of the 34 A/D strains sampled in this study, 26 tested as haploids by PCR and flow cytometry. The MFα1 amplicons from these strains were sequenced, and multiple alignment of these sequences showed them to be identical to those of the C. neoformans var. grubii strains (data not shown). Isolates testing positive for both pheromones and showing 2n ploidy by flow cytometry were identified tentatively as diploids awaiting contour-clamped homogeneous electric field analysis (12). A total of eight A/D diploids were confirmed in our sample strains. These included two strains from France (UM2 and UM4), five from Spain (69C, 87C, 92C, 190C, and 195C), and one from the United States (92-793).

Histograms of DNA fluorescent intensities from test strains were compared with those from standard strains (data not shown). The relative distribution of fluorescence and the positions of two DNA peaks allowed accurate estimations of diploidy among A/D isolates relative to that of haploid C. neoformans type strains. The serotype A/D MFα1 sequences were identical to those of the serotype A (C. neoformans var. grubii) isolates, and both differed from the serotype D (C. neoformans var. neoformans) sequences by eight nucleotide polymorphisms (data not shown). Moreover, MFα1A/D segregated with C. neoformans var. grubii in the phylogenetic tree, indicating that the MFα1 in the serotype A strains and that in the serotype A/D strains are identical (Fig. 3). Strikingly, MFa1 sequences from A/D isolates showed 100% identity with those from serotype D (C. neoformans var. neoformans) type strain NIH 430 (data not shown).

The hybridization probe for diploid strains was C. neoformans var. grubii partial MFα1. HaeII digests of genomic DNA revealed four bands among diploids, as seen with the type strain for C. neoformans var. grubii (Fig. 5A). All diploids shared an approximate 1.0-kb band with type strain H99. Four A/D strains (UM2, UM4, 69C, and 92-793) had two other MFα1 bands similar to those of C. neoformans var. grubii type strain H99. A >5.0-kb band was unique to some diploid strains. The same genomic DNA digested with BamHI and probed with MFa1 revealed two bands, in contrast to the three bands seen with the two type strains (Fig. 5B).

FIG. 5.

A/D diploid MFα1 and MFa1 copy numbers. Genomic DNA was digested with HaeII or BamHI and hybridized with labeled MFα1 or MFa1 probes as described for Fig. 2 and 4. (A) Three distinct MFα1 bands in strain NIH 12 (MATα, serotype D) and four MFα1 bands in C. neoformans var. grubii type strain H99 (MATα, serotype A) as well as in all of the eight A/D isolates are shown. (B) Three MFa1 bands for the two C. neoformans var. neoformans strains NIH 430 and NIH 433 (MATa, serotype D) are shown; A/D isolates showed two bands.

Karyotype analyses showed that all suspected diploid isolates had chromosome patterns similar to that observed for C. neoformans var. grubii type strain H99 (data not shown). Southern hybridization of chromosomes with MFα1 revealed one intense band in each strain, which localized to the MAT chromosome of 2.5 Mb except in NIH 444, in which the localization was to a chromosome of approximately 1.0 Mb (data not shown). Identical single hybridization bands were obtained by MFa1 probe of C. neoformans var. neoformans reference strains NIH 430 and NIH433 and the two representative diploid strains (data not shown).

None of the diploid strains showed any atypical colony morphology compared to standard strains, nor did they exhibit any evidence of phenotypic switching or appreciable differences in capsule size or melanin production (data not shown). The morphological differences were noticeable with diploid strain 195C, whose cells appeared to be larger and highly mucoid. Cells without buds were considered spheres, and using SEM images, linear measurements along the long axes were taken to estimate cell size (data not shown). A comparison of C. neoformans cell sizes is summarized in Fig. 6. All A/D diploid cells were significantly larger than those of either of the two type strains (P < 0.001).

FIG. 6.

Morphometry of SEM images was used to compare the cell sizes of A/D isolates with those of C. neoformans var. grubii MATα type strain H99 or C. neoformans var. neoformans MATa type strain NIH 430 (details in Materials and Methods). The box graph illustrates average cell sizes, with median values marked by horizontal lines; positive and negative standard deviations are also shown along with outlier values for 50 cell sizes. Cells of all diploid strains were significantly larger than those of the two control strains (Student t test; P < 0.001).

V8 juice agar cultures revealed that only strains 69C and 92C produced basidia and basidiospores, indicative of self-fertility. The remaining six A/D diploid strains produced hyphae but no clamp connections on the mating medium.

High pathogenicity of A/D diploids.

The survival curves for mice infected intravenously with 10⁶ cells of eight A/D isolates along with the three control strains are shown in Fig. 7. Type strain H99 (C. neoformans var. grubii, MATα, serotype A) was most lethal, with all infected mice dead or moribund by day 8. C. neoformans caused all the deaths, as evidenced by its recovery from brain necropsies. The isogenic pair of C. neoformans var. neoformans tester strains (JEC20, MATa and JEC21, MATα) did not cause any deaths or gross disease symptoms during the 30-day observation period, even though we recovered the fungus from the brains of infected mice sacrificed after day 30. Mice infected with any one of the eight A/D isolates showed a survival pattern more similar to that for H99 infection. Thus, A/D strains 69C, 87C, 92-793, UM2, UM4, and 190C caused mortality within 9 to 13 days, while strain 92C caused mortality on day 18. The only exception was strain 195C, with no morbidity or mortality for the entire observation period. Thus, a majority of the A/D isolates were highly pathogenic for mice, and they differed in this attribute from the low virulence of an isogenic pair of C. neoformans var. neoformans isolates.

FIG. 7.

Survival curves of mice infected with C. neoformans strains. Groups of five mice received 10⁶ C. neoformans cells intravenously, and the course of infection was followed over 30 days. C. neoformans H99 (C. neoformans var. neoformans, MATα) was the most lethal strain, while the C. neoformans var. neoformans isogenic mating pair (JEC20 and JEC21) caused no mortality. Most A/D strains caused infections similar to that caused by H99.

Representative histological sections of mouse brains collected on day 7 postinfection are shown in Fig. 8. Cellular reaction to the yeast was minimal, as is typical in mice (49). The foci with yeasts were scattered in the white and grey matter of the cerebrum, cerebellum, and brain stem. A large numbers of yeasts filled these cyst-like, expanding perivascular spaces. In the more severely affected brains, yeasts were also seen in the meninges, hippocampus, or choroid plexus. The numbers of cysts were counted to arrive at a quantitative comparison among various strains. The isogenic pair of C. neoformans var. neoformans strains (JEC20, MATa and JEC21, MATα) showed moderate numbers of cysts, with 3 to 10 in the cortex and thalamus or mesencephalon and 0 to 2 in the cerebellum with or without the brain stem (Fig. 8A and a). Type strain H99 (C. neoformans var. grubii, MATα, serotype A) caused 25 to 45 cysts in the cortex and thalamus or mesencephalon and 25 cysts in the cerebellum with or without the brain stem (Fig. 8B and b). Most diploid A/D strains also caused large numbers of cysts to form (16 to 54 in the cortex and 5 to 25 in the cerebellum), except for strain 195C, which caused the formation of 0 to 2 cysts. A representative section from diploid 190C is shown (Fig. 8C and c). These data indicated that the numbers and the distribution of cysts with C. neoformans yeasts in various parts of the brain were a good predictor of the virulence of a C. neoformans strain. These results also showed good correlation with the mouse survival curves of the A/D diploids and the control strains.

FIG.8.

Brain sections from infected mice. In groups of two or three, mice were infected with 10⁶ C. neoformans cells intravenously and euthanized on day 7, and the brain sections were stained with Mayers mucicarmine. Few, scattered, and small foci with yeasts of control strain JEC20 are seen in the grey and white matter of cortex and diencephalons (A; ×20 magnification). A close-up of small thalamic cysts shows expansion in closely apposed perivascular spaces by yeasts (a; ×200). A section from another control strain (H99) shows numerous large cysts scattered in many parts of the brain (B; ×20), and higher magnification of a large thalamic perivascular cyst shows it to be full of yeasts, with scant inflammatory reaction (b; ×200). The cerebrum, with thalamus and lateral ventricle, of a mouse inoculated with a representative diploid 190C strain shows scattered cysts containing numerous yeasts (C; ×20). A close-up of cyst-like, confluent perivascular spaces filled with numerous yeasts is shown (c; ×200).

DISCUSSION

There were several novel findings related to MFα1 and serotype A/D isolates. In three C. neoformans varieties, MFα1 showed (a) specific nucleotide polymorphisms, (b) different copy numbers and chromosomal localizations, and (c) unique deduced amino acids in two populations of C. neoformans var. gattii. MFα1 of different varieties cross-hybridized in Southern hybridizations. Their phylogenetic analyses showed purifying selection (neutral evolution). These observations suggested that MATα strains from any of the three varieties can mate or hybridize in nature with MATa strains. A few serotype A/D diploids strains provided evidence for mating or hybridization. MFα1 sequence and copy numbers in diploids were identical to those of C. neoformans var. grubii, while their MFa1 sequences were identical to those of C. neoformans var. neoformans; thus, these strains were hybrids. The mouse survival curves and histological lesions revealed A/D diploids to be highly pathogenic, in similarity to the C. neoformans var. grubii type strain and unlike the C. neoformans var. neoformans strains. In addition to A/D diploids, MFa1 amplicons and hybridization signals were only obtained from two C. neoformans var. neoformans reference strains. This mating type bias suggested that yet undiscovered MATa pheromone(s) in C. neoformans var. gattii and C. neoformans var. grubii are unrelated to, highly divergent from, and/or rarer than those in C. neoformans var. neoformans.

Our results suggest that the MFα1 sequences from three C. neoformans varieties are closely related homologs. Most strains tested as haploid MATα, which was expected in view of the predominance of this mating type in C. neoformans populations. Significantly, three copies of MFα1 were seen in Southern hybridizations of C. neoformans var. neoformans, while four copies were found in C. neoformans var. grubii and only two copies were found in C. neoformans var. gattii. The occurrence of multiple copies of pheromone genes is parallel to that seen for Saccharomyces cerevisiae and unlike that seen with the single-copy pheromone genes found in Neurospora, Podospora, and Yarrowia lipolytica (21, 28). It is not yet known whether the multiple pheromone copies constitute evidence of a silent mating type and/or potential for a mating type switch, as is well known for S. cerevisiae. It is tempting to speculate that the closely related MFα1 has the potential to cross-hybridize with the MATa strain from another variety. We view this possibility as one plausible explanation for the reported mating of F. neoformans var. bacillispora MATα (strain NIH 444) and F. neoformans var. neoformans MATa (strain NIH 430). Several plant pathogenic fungi are also reported to undergo rapid evolution via natural interspecific hybridization (7). Three fertile Saccharomyces species (S. cerevisiae, S. uvarum, and S. douglasii) cross-hybridize to form diploid hybrids which are essentially infertile (24). Interspecific hybridization could also be obtained with several other Saccharomyces species, with stable hybrids resultant only in crosses between members of phylogenetically related taxa (35). Thus, it is reasonable to expect intervariety mating and hybridization in C. neoformans, due to the close phylogenetic relatedness of MFα1 in three varieties.

We characterized 20 of the 38 amino acids constituting the MFα1 lipopeptide, with four codon-changing (nonsynonymous) substitutions in those of C. neoformans var. grubii and one each in those of C. neoformans var. gattii and C. neoformans var. neoformans. The Ka/Ks ratio values for this region were less than 1, indicating that MFα1 sequences were under purifying or neutral selection to eliminate any deleterious changes (46). C. neoformans var. grubii had relatively higher ratio values than C. neoformans var. gattii and C. neoformans var. neoformans. Perhaps the higher ratio value for C. neoformans var. grubii is suggestive of an ongoing trend towards more rapid evolution of MFα1A, especially if there is no constraint imposed with regard to mating with the opposite mating type. The consensus phylogenetic trees showed MFα1 clustering in accordance with the currently accepted three varieties. This grouping is congruent with similar results obtained from a number of studies using maximum likelihood estimation, AFLP, and gene genealogy (6, 11).

Among notable observations of A/D analyses, a majority of A/D strains were found to be haploids, with MFα1 identical to that of C. neoformans var. grubii (serotype A). The extent to which this is true for other studies on A/D strains is not readily apparent, even though Cogliati et al. (15) have reported observing MFα1 alleles associated with both serotypes A and D and Lengeler et al. (34) reported observing MFα1D in their A/D analyses. It must be also emphasized that not all A/D strains represent diploids, since serotype A/D has been identified mainly by the presence of weak agglutinations to serotype A and D specific antisera rather than by the presence of any unique antigenic factors. Moreover, serotypes are not stable over time and many strains are reported to switch from an A/D serotype to either A or D, although not the other way round (3, 39). We found that eight A/D strains were stable diploids; repeat testing showed 2n DNA content and the presence of both the MATα and MATa pheromone genes. Karyotype analyses of these strains revealed no gross chromosomal abnormalities, nor were there any unique phenotypic features in diploid strains, including colony morphology, capsule size, melanin production, etc. Significantly, we found that all diploid strains have the MFα1 of serotype A and the MATa of serotype D. Thus, we consider our diploid A/D strains to be hybrids between the serotype A MATα strain and the serotype D MATa strain. Our observations now confirm a reasoned speculation by Boekhout et al. (5) on the likely occurrence of MFα1D in A/D diploids. These results also agree in part with the findings of Cogliati et al. (15) and Yan et al. (55), who both reported observations of MFα1A and MFα1D in their samples of A/D diploids. However, these results are contrary to the report of Lengeler et al. (34), who concluded that the pheromone constituents of their A/D diploids were MATaA and MFα1D. Thus, C. neoformans A/D diploids with three different pheromone constituents have now been identified: MFα1A × MFa1D, MFα1D × MFa1D, and MFα1D × MFa1A. We relied exclusively on pheromone sequence analyses, because MFα1 polymorphisms are unique among three varieties and the published literature on polymorphisms in other C. neoformans genes is not exhaustive for the three varieties. A majority of our diploid strains and all of the diploids identified by Cogliati et al. (15) originated in Europe, where C. neoformans var. grubii and C. neoformans var. neoformans have overlapping niches in pigeon manure; by implication, ample opportunities exist for cross-variety hybridization.

We found four copies of MATα in A/D diploid strains, a finding similar to that for haploid MATα C. neoformans var. grubii and unlike those of three copies in C. neoformans var. neoformans and only two copies in C. neoformans var. gattii. Unexpectedly, only two copies of MATa were evident in the diploid strains, in contrast to the three copies seen in haploid strains NIH 430 and NIH 444. Such variable copy numbers could arise from a gene duplication, conversion, or loss event. However, we have no data at this stage to support any particular one of these possibilities. It is difficult to link this finding regarding variable copy numbers to the observed sterility of most diploids, since it is not yet known how many copies of MAT genes are necessary for successful mating. It is apparent that the presence of an opposite pheromone gene in a C. neoformans cell rarely leads to self-fertility (15, 34). This inference is consistent with fungal literature indicating that the transgene expression of the opposite pheromone in a haploid fungal cell could lead to either self-fertility or sterility (1).

The pathogenicity experiments were performed on A/D diploids, because few previous studies have focused on the role of ploidy in fungal pathogenesis. Thus, Aspergillus nidulans and Beauveria bassiana and Beauveria sulfurescens hybrids are reported to be more virulent than the two haploid prototrophic or auxotrophic parents (43, 51). Hubbard et al. (26) found that a Candida albicans tetraploid strain was more virulent than its auxotrophic diploid parents, but its virulence does not exceed that of its prototrophic diploid parent. Similarly, Suzuki et al. (50) demonstrated poor germ tube formation in haploid C. albicans versus high frequency germ tube formation in diploids and tetraploids, which could positively influence pathogenicity. More recently, Lengeler et al. (34) reported lower virulence in C. neoformans A/D diploids compared with that of a control strain (C. neoformans var. grubii). We therefore included both C. neoformans var. grubii and C. neoformans var. neoformans type strains in our analyses of A/D diploids. Interestingly, seven of the eight diploid strains studied mimicked the high pathogenicity of C. neoformans var. grubii type strain. Only one diploid failed to cause any morbidity or death in the infected mice; this was similar to the pattern seen with two isogenic strains of C. neoformans var. neoformans. Additional data on virulence was obtained from detailed histological studies on infected brain tissues. We found more C. neoformans lesions and wider distribution of lesions in mice infected with H99 and the seven of eight diploid strains mentioned above, while C. neoformans var. neoformans type strains caused minimal lesions. Thus, mouse histology provided an independent confirmation of the high pathogenic potential of A/D diploids, at a level similar to that of the C. neoformans var. grubii type strain. Overall, these observations are in contrast with the marked low virulence of C. neoformans hybrids observed by Lengeler et al. (34). It is tempting to attribute these differences to the origin of MATα constituents in these strains (i.e., MFα1A versus MFα1D). This hypothesis would be easy to test in experiments with hybrids containing either MFα1A or MFα1D paired with MFa1D, as reported by Cogliati et al. (15). Interestingly, none of the hybrid strains tested exhibited a higher level of pathogenicity than the haploid C. neoformans var. grubii strain, which thereby indicates that duplication of the C. neoformans genome does not enhance virulence, an observation which is similar to those made using C. albicans but unlike that of the increased virulence seen with some other fungi (26, 43, 51).

A link between an increase in ploidy and a larger cell size has been reported for many organisms, including Candida albicans, and has been confirmed recently by elegant experiments with S. cerevisiae (20, 26, 50). This aspect of C. neoformans morphology has not been investigated in the recent studies on C. neoformans diploids (15, 34). Using SEM morphometry and image analysis software, we compared cell sizes of haploid type strains and A/D diploids. These measurements revealed that the A/D cells were significantly larger than cells of C. neoformans var. grubii and C. neoformans var. neoformans. Thus, ploidy content appears to influence the cell size in C. neoformans diploids, a finding similar to the observations with C. albicans and S. cerevisiae (20, 26). A special mention must also be made of diploid strain 195C, cells of which were highly mucoid, significantly larger, and remarkably nonpathogenic. The implications are that 2n DNA content, altered cell surface, and much larger cell size did not positively influence the pathogenicity of this strain.

The negative result for MFa1 in the C. neoformans var. gattii MATa type strain was unexpected, because the presence of MATa in C. neoformans var. gattii is well established (23, 29, 30). Further studies might be needed to identify C. neoformans var. gattii and C. neoformans var. neoformans hybrids (MATαD × MATaB or MATaC), especially in parts of the world where these varieties have overlapping ecological niches (3, 29, 40). We sampled B/C test strains for MATa by PCR and Southern hybridization based on the C. neoformans var. neoformans MFa1 gene, because degenerate primers and genomic hybridizations have been successfully used to identify mating type genes in many fungi (13, 14, 42, 48). Our negative results could possibly be due to the facts that the 117-bp PCR amplicon was a hypervariable region of the MATa locus and that targeting of this sequence may yield negative results with genomic DNA from C. neoformans var. gattii. However, this is unlikely if the MFa1 genes in these varieties are related. Further support for our conclusions comes from the recent data of McClelland et al. (36), who used slot blots to confirm the presence of MFa homologs in C. neoformans var. neoformans alone and its absence in C. neoformans var. grubii and C. neoformans var. gattii. Negative MFa1 results for C. neoformans var. grubii were suggestive of the extreme rarity or divergence of this gene in C. neoformans var. grubii isolates. This finding is consistent with unsuccessful attempts to find MATa strains and the infertility of C. neoformans var. grubii progeny in mating experiments (32, 52, 55). Additional evidence for the failure of C. neoformans var. grubii to undergo meiosis came from crosses between MFα1A and MATaD strains, in which all mitochondrial transmission in the progeny originated from C. neoformans var. neoformans (54). Our results, viewed in conjunction with other published reports, suggest that C. neoformans var. grubii may exist overwhelmingly in the form of an asexual population. Recently, Lengeler et al. (32) reported identification of the MATa mating-type locus (STE20 protein kinase homolog) in a sterile C. neoformans serotype A isolate. Another report described isolation of a serotype A MATa strain from Italy on the basis of STE20- and MATa-positive amplicons, although no sequence information was provided (52). Overall, a clonal population appears to be the norm in C. neoformans var. grubii, a conclusion that should prevail until sexually competent MATa strains or phylogenetic evidence for recombination is discovered in this variety (38).

In summary, a C. neoformans mating pheromone gene (MFα1) consists of variety-specific nucleotides and, perhaps, population-specific amino acids. This gene from different varieties is closely enough related to undergo intervariety mating or hybridization with MATa strains. The evidence for the latter scenario was found in a few serotype A/D diploids that carried both MFα1A and MFa1. All hybrid strains had four copies of MFα1A and two copies of MFa1, and many of these isolates were sterile. A/D diploids largely matched C. neoformans var. grubii strain in their pathogenicity for mice. Incidentally, most serotype A/D isolates were haploid strains with MFα1A. The MFa1 gene appears to be unique to C. neoformans var. neoformans, with no homologs in C. neoformans var. grubii and, more surprisingly, none in the sexually competent C. neoformans var. gattii. These results could form the basis for future studies on the role of intervariety mating in C. neoformans biology and virulence.

ADDENDUM IN PROOF

Shen et al. (W.-C. Shen, R. C. Davidson, G. M. Cox, and J. Heitman, Eukaryot. Cell 1:366-377, 2002) recently reported that Cryptococcus neoformans pheromones stimulate mating and differentiation and that the MFα pheromone is not essential for virulence but contributes to the overall virulence composite.