Sexual Cycle of Cryptococcus neoformans var. grubii and Virulence of Congenic a and {alpha} Isolates

Cryptococcus neoformans is a human-pathogenic fungus that hasevolved into three distinct varieties that infect most prominentlythe central nervous system. A sexual cycle involving haploidcells of a and ${alpha}$ mating types has been reported for two varieties(C. neoformans var. neoformans, serotype D, and C. neoformansvar. gattii, serotypes B and C), yet the vast majority of infectionsinvolve a distinct variety (C. neoformans var. grubii, serotypeA) that has been thought to be clonal and restricted to the ${alpha}$ mating type. We recently identified the first serotype A isolateof the a mating type which had been thought to be extinct (strain125.91). Here we report that this unusual strain can mate witha subset of pathogenic serotype A strains to produce a filamentousdikaryon with fused clamp connections, basidia, and viable recombinantbasidiospores. One meiotic segregant mated poorly with the serotypeA reference strain H99 but robustly with a crg1 mutant thatlacks a regulator of G protein signaling and is hyperresponsiveto mating pheromone. This meiotic segregant was used to createcongenic a and ${alpha}$ mating type serotype A strains. Virulence testswith rabbit and murine models of cryptococcal meningitis showedthat the serotype A congenic a and ${alpha}$ mating type strains hadequivalent virulence in animal models, in contrast to previousstudies linking the ${alpha}$ mating type to increased virulence in congenicserotype D strains. Our studies highlight a role for sexualrecombination in the evolution of a human fungal pathogen andprovide a robust genetic platform to establish the moleculardeterminants of virulence.

INTRODUCTION

Top

Introduction

Sexual reproduction involves the mixing of genetic materialfrom two parents to form progeny that contain genes from bothparents (4). While sex is the most prevalent mode of reproductionin eukaryotes, there is much debate as to why sexual reproductionpredominates (5, 49, 61). Two hypotheses have been proposed.In the first, sex provides variation for natural selection toact upon and promotes adaptation by allowing beneficial mutationsto spread and escape deleterious mutations in other genes (6,37). A second model suggests that sexual reproduction facilitatesDNA repair or dilutes deleterious mutations (7, 65). This modelis consistent with the idea that organisms with low levels ofdetrimental DNA damage would not necessarily require sexualreproduction (3, 21, 58).

A few organisms, such as the bdelloid rotifers, are exclusivelyasexual (3, 7, 21, 58). Asexual reproduction involves copyingof a parent to produce genetically identical progeny (65). Whileexclusively asexual organisms are thought to be rare, organismswith facultative parthenogenesis, the ability to produce offspringsexually or asexually, are more common. Many insect species,such as aphids and cockroaches, as well as nematodes have theability to produce either females asexually or males and femalessexually (2, 11, 43). Many lower eukaryotes reproduce asexually;in the cases of yeasts and filamentous fungi, 20% or more ofspecies have no defined sexual cycle (19).

The role of the sexual cycle in human-pathogenic fungi is unclear.The yeast Candida albicans is the most common cause of humanfungal infections yet has long been thought to be an obligateasexual diploid (8). The recent discovery of the mating type(MAT) locus of C. albicans led to the identification of matingand the characterization of a unique cell type specialized formating (22, 23, 25, 38, 40). While mating has been identifiedin C. albicans, meiosis and ploidy reduction to complete thesexual cycle have not yet been detected, and population geneticstudies reveal a largely clonal population (53, 64). Hence,when and where C. albicans employs its sexual cycle and therole the sexual cycle may play in virulence remain elusive.

Cryptococcus neoformans is an opportunistic human fungal pathogenthat infects the central nervous system and causes meningoencephalitis(9, 24). C. neoformans occurs in three varieties: C. neoformansvar. grubii (serotype A), C. neoformans var. gattii (serotypesB and C), and C. neoformans var. neoformans (serotype D). Althoughthese varieties can be interfertile (29, 31, 46, 47), intervarietalmating results in hybrid strains (serotype AD) that producelargely inviable spores (34). C. neoformans occurs in the environmentassociated with pigeon guano and certain tree species (9). Humansare thought to be exposed by inhalation of basidiospores, whichare small enough to lodge in the alveoli of the lung (52). Sporesare produced via sexual reproduction, haploid fruiting, or diploidfilamentation (27, 28, 48, 59). Importantly, virulence has beenlinked to mating type. Most clinical isolates are of the ${alpha}$ matingtype, and ${alpha}$ strains have been reported to be more virulent thancongenic a strains in serotype D (9, 30).

A sexual cycle has been reported for C. neoformans var. gattiiand C. neoformans var. neoformans and involves fusion of a and ${alpha}$ cells to produce heterokaryotic filaments (27, 28). These filamentsultimately produce basidia, where nuclear fusion and meiosisoccur, and long chains of infectious spores are produced. However,mating has only recently been characterized for C. neoformansvar. grubii serotype A strains, which are the predominant clinicalisolates. Population studies of this variety indicate that itis predominantly clonal, and only strains of the ${alpha}$ mating typehad been identified (15, 32). Recently, however, two a matingtype strains (125.91 and IUM 96-2828) were identified from aworldwide screen of $~$ 1,500 serotype A strains (36, 41, 55). Whilestrain 125.91 was unable to mate with the serotype A referencestrain H99, strain IUM96-2828 could mate with H99 to produceviable recombinant progeny (26). Thus, a sexual cycle existsfor the most common clinical isolates of C. neoformans. However,the role that this sexual cycle plays in the population structureand virulence of C. neoformans var. grubii remains unclear.

Here we present additional evidence that strain 125.91 is abona fide haploid a mating type serotype A strain and show thatit is capable of undergoing a sexual cycle with a subset ofC. neoformans var. grubii ${alpha}$ mating type strains. We constructedcongenic serotype A strains and used these strains to test whethermating type is linked to virulence in this variety. These congenicserotype A strains will allow characterization of the role thatthe sexual cycle plays in pathogenicity and will enable geneticand molecular analyses of the most prevalent variety of thisubiquitous human pathogen.

MATERIALS AND METHODS

Top

Materials and Methods

Strains. Reference strains used in this study were the serotype A strainH99 (mating type ${alpha}$ , serotype A [ ${alpha}$ A]) (54) and the congenic serotypeD strains JEC20 (aD) and JEC21 ( ${alpha}$ D) (20, 30). The isolates 125.91,8-1, S25J, 4476B, ZG287, ZG290, and KW5 were from the Duke Universitystrain collection (34, 36). Strains CBS132 and ATCC 48184 werefrom the American Type Culture Collection, and strains CDC94-383,CDC92-74, CDC228, and CDC304 were from the Centers for DiseaseControl and Prevention (CDC) (34). The H99 crg1 mutant was derivedfrom strain H99 by gene disruption. A 1.3-kb CRG1 coding domainfragment was generated by PCR with primers 5'-GCAACCAATCACTATGAAand 5'-TGTGATCTGGTTCATGTA. PCR was performed for 30 cycles of94°C for 40 s, 40°C for 1 min, and 72°C for 1 min.PCR products were gel excised, cloned, and verified by sequencing.A two-step PCR amplification procedure (13) was used to generatethe crg1::URA5 disruption allele used to transform strain F99(H99 ura5). Biolistic transformation and mutant isolation wereas described previously (54, 57).

Media. Strains were grown on yeast extract-peptone-dextrose (YPD) medium.Mating assays were on V8 medium (29) (5% [vol/vol] V8 juice,3 mM KH₂PO₄, 4% [wt/vol] agar) at pH 5 or 7. Mating assays withstrain 125.91 used V8 medium at pH 7, V8 medium at pH 5, carrotmedium (5% [vol/vol] carrot juice, 3 mM KH₂PO₄, 4% agar), orSLAD medium (1). Confrontation assays were on SLAD medium.

FACS analysis. Fluorescence-activated cell sorter (FACS) analysis was performedas described previously (34). Approximately 50 µl of vegetativecells was washed with normal saline (0.9%), fixed in 70% ethanol,washed with 10 mM Tris-HCl (pH 7.2)-0.25 M sucrose-1 mM EDTA-1mM MgCl₂-0.1 mM CaCl₂-0.1 mM ZnCl₂, and incubated with 10 µgof propidium iodide per ml and 1 mg of RNase A per ml. Cellswere diluted to $~$ 10⁸ cells/ml, sonicated for 10 s, and subjectedto FACS analysis. More than 10,000 cells of each strain wereanalyzed for DNA content, determined by relative fluorescenceof stained genomic DNA.

NCP1a gene analysis. The NCP1a gene was amplified by touchdown PCR (44) with primers5'-TTGAGCGTCATATTGGTCATGAC and 5'-GAAGACCGTCATCACACACAAG. PCRmixtures contained 50 to 80 ng of genomic DNA and 50 pmol ofeach primer. PCR was performed for 20 cycles of 94°C for30 s, 60°C with a temperature increment of -1°C at everycycle for 1 min, and 72°C for 45 s. After 20 cycles, anadditional 20 cycles were performed at an annealing temperatureof 40°C. The ACT1 gene was amplified with primers 5'-ATGGAAGAAGAAGGTACGand 5'-TTAGAAACACTTTCGGTG. PCR mixtures contained 50 ng of genomicDNA and 50 pmol of each primer, and PCR was performed for 30cycles of 94°C for 15 s, 55°C for 30 s, and 72°Cfor 45 s.

Southern blot analysis. Genomic DNA was prepared by using the Camgen (Cambridge, UnitedKingdom) yeast genomic DNA extraction kit. DNA (10 µg)was digested to completion with PstI (New England Biolabs) andfractionated by electrophoresis in a 0.8% agarose gel. The gelwas ethidium bromide stained to verify equal DNA loading anddigestion, transferred to a nylon membrane (Hybond-N+; Amersham),and fixed by UV cross-linking. The gel was also ethidium bromidestained after transfer to verify complete DNA transfer to themembrane. The membrane was hybridized overnight with a gel-purifiedDNA probe labeled with ³²P by random priming (Amersham) andthen washed, dried, and exposed to film for 24 hr. The transposonT1 and T2 probes were generated by PCR with genomic DNA ( ${alpha}$ D strainJEC21) and primers 5'-CACTGTGGGAAAGGGTCGCC and 5'-CACTGTGGGAATTCCTCGCCfor T1 and 5'-GGGGTACAAAATGCA for T2. The T1 and T2 primerswere also used to PCR amplify these transposons from serotypeA and D strains, with PCR performed for 30 cycles of 94°Cfor 15 s, 55°C for 30 s, and 72°C for 45 s.

Serotyping. PCR-based genotyping for the identification of serotype wasas described previously (34, 36). PCR was performed for 30 cyclesof 94°C for 15 s, 60°C (for PAK1, GPA1, and CNA1) or66°C (for STE20) for 30 s, and 72°C for 30 s. PCR productswere cloned and sequenced. Classic serotyping was done withthe Crypto Check serotyping kit from Iatron Laboratories (Tokyo,Japan) as described previously (34).

Mating type analysis. Mating to reference strains was used to determine mating type.Strains were pregrown on YPD agar for 2 days, and a small amountof cells was removed and patched onto V8 solid medium (pH 5.0)either alone or mixed with a reference strain. Plates were incubatedat room temperature in the dark and assessed by light microscopyfor filament and basidiospore formation. PCR-based mating typedetermination was as described previously (34, 36), using primersto the a or ${alpha}$ allele of the STE20 gene for serotypes A and D.

Cell wall and nuclear staining. Strains were grown on V8 medium (pH 5) as described above formating reactions. Areas with filamentation were excised, washedwith deionized water, and stained with 0.1% calcofluor white(Difco) or 1 µg of DAPI (4',6'-diamidino-2-phenylindole-2HCl)(Sigma) per ml. Calcofluor white-stained filaments were analyzedfor fused clamps. DAPI-stained cells were analyzed to establishwhether cells were uni- or binucleate.

Confrontation assays. Two strains were streaked onto SLAD medium in parallel lines5 mm apart without touching. Conjugation tube formation by ${alpha}$ cells and swelling of a cells were assessed by light microscopyafter 24, 48, and 72 h at room temperature.

Single-basidiospore isolation. To isolate basidiospores, strains were grown on V8 medium (pH5) as described above for mating reactions. Areas showing basidiosporeformation upon microscopic examination were excised, and sporeswere transferred to a fresh YPD plate and dissected by micromanipulation.

RFLP analysis. PAK1 and GPA1 gene sequences from strain KNA14 were PCR amplified,cloned, and sequenced as described above. The 125.91, 8-1, KNA14,and H99 sequences were compared to identify restriction fragmentlength polymorphisms (RFLPs). The restriction enzyme PvuI (CGATCG)(New England Biolabs) was used to characterize the GPA1 gene,and the restriction enzyme BstXI (CCANNNNNNTGG) (New EnglandBiolabs) was used to characterize the PAK1 gene. Briefly, genefragments were PCR amplified as described above, and the DNAwas incubated with 1 to 1.25 U of the appropriate restrictionenzyme and buffer for 4 h. The PvuI digests were at 37°C,and the BstXI digests were at 55°C.

PFGE. Cells were grown in yeast nitrogen base minimal medium supplementedwith 1 M NaCl at 30°C with shaking at 225 rpm to an opticaldensity at 600 nm of 0.5 to 0.6. Cells were harvested and plugswere prepared as described previously (35). Samples were runin a 1% pulsed-field gel electrophoresis (PFGE)-grade agarose(Bio-Rad) gel with 0.5x Tris-borate-EDTA running buffer in aBio-Rad contour-clamped homogeneous electric field DRII apparatuswith the following settings: initial A time, 75 s; final A time,150 s; start ratio, 1.0; run time, 30 h; mode, 10; initial Btime, 200 s; final B time, 400 s; start ratio, 1.0; run time,60 h; mode, 11. The voltage was set to 125 V and the buffertemperature was set to 12°C. The running buffer was changedevery 24 h.

Congenic strain construction. Strain 125.91 (aA) was crossed with strain 8-1 ( ${alpha}$ A), and singlebasidiospores were isolated. One of the a progeny from the mating,KNA14, was crossed with strain H99 crg1 ( ${alpha}$ A), and single basidiosporesof a and ${alpha}$ strains were isolated. An a offspring from this mating,KNB-5, was crossed to H99 to yield first-generation progeny.The process of isolating a single-basidiospore cultures andbackcrossing to H99 was repeated a total of 10 times to produceKN99a, an a strain that is congenic with the ${alpha}$ strain H99. StrainKN99-5a is the product of the fifth backcross.

Virulence studies. Virulence studies were performed with established rabbit andmurine cryptococcal meningitis models (1, 39, 56). Groups of4- to 6-week-old female A/Jcr mice (10 mice per strain) wereanesthetized by phenobarbital injection and infected with 1x 10⁵, 5 x 10⁴, or 1 x 10³ fungal cells in 50 µl of normalsaline pipetted into the nostrils. Mice were monitored twicedaily, and those in pain or sick were sacrificed by CO₂ inhalation.The percentage of surviving mice was plotted against time, andP values were calculated by the Mann-Whitney test. Male NewZealand White rabbits weighing 2.5 to 3 kg (three per group)were administered cortisone acetate at 5 mg/kg intramuscularly1 day prior to inoculation and then daily for the entire testperiod. One day after the initial steroid treatment, the rabbitswere anesthetized with xylazine and ketamine intramuscularlyand inoculated intracisternally with 10⁸ cells in 0.3 ml ofnormal saline. Rabbits were anesthetized on days 4 and 7 followinginfection, and 0.5 ml of cerebrospinal fluid (CSF) was withdrawn,serially diluted, and plated onto YPD medium. The colonies werecounted, and the mean number of CFU was plotted, with the standarderror of the mean indicated. A standard Student t test was performedto establish P values for the differences between groups.

Nucleotide sequence accession numbers. The sequences from strains 125.91 and 8-1 have been depositedin GenBank under accession numbers AY219893 to -96.

RESULTS

Top

Results

Strain 125.91 is a serotype A MATa haploid strain. We recently identified the first a mating type isolate of thepathogenic serotype A C. neoformans var. grubii (strain 125.91)(36). Identification of this strain as serotype A and a matingtype was based on serotyping, PCR-based genotyping for serotype,and the sequence of the MATa-encoded STE20a gene. Because strain125.91 would not mate with the serotype A reference strain H99,we established by additional stringent criteria that strain125.91 is a bona fide haploid a mating type serotype A (aA)strain and not an unusual aneuploid or hybrid isolate. First,by FACS analysis strain 125.91 is haploid (1n/2n) and not aneuploidor diploid (2n/4n) (Fig. 1A). Second, by Southern blotting weconfirmed that strain 125.91 lacks two transposable elements(T1 and T2) that are common in serotype D strains and absentor rare in serotype A strains (Fig. 1B) (12, 36). PCR with primersspecific to the T1 and T2 transposons also showed that onlythe serotype D strains produced PCR products of the appropriatesize; this is consistent with an analysis of the 2x genomicsequence coverage of the serotype A reference strain (H99),which shows no T1 or T2 transposons (not shown). Third, by phylogeneticanalysis based on amplified fragment length polymorphism analysis,strain 125.91 is evolutionarily more related to serotype A strainsthan to serotype D or AD hybrid strains (T. Boekhout, personalcommunication). Fourth, mitochondrial DNA sequences of strain125.91 were characteristic of serotype A and not serotype Dstrains (F. Dietrich, personal communication). Finally, in contrastto serotype AD strains, which typically contain both serotypeA- and D-specific gene alleles at multiple loci (34), strain125.91 was previously found to contain only serotype A allelesby PCR-based genotyping (36). To verify this PCR analysis, thePAK1, GPA1, and CNA1 genes from strain 125.91 were sequencedand found to be more similar to serotype A gene sequences (96to 99%) than to serotype D gene sequences (91 to 94%). In summary,strain 125.91 is a haploid serotype A strain.

View larger version (66K):
[in this window]
[in a new window]
FIG. 1. Identification of a and ${alpha}$ haploid serotype A C. neoformans strains. (A) FACS analysis of the aA strains 125.91, KNA13, and KNA14; the ${alpha}$ A strain 8-1; and the control aD haploid strain JEC20 and aDDa diploid strain CHY600. (B) Southern blots of 125.91 (aA), 8-1 ( ${alpha}$ A), H99 ( ${alpha}$ A), JEC20 (aD), and JEC21 ( ${alpha}$ D) PstI-digested genomic DNAs probed with transposon T1 or transposon T2. (C) Structures of the MAT locus alleles from the aD strain JEC20, the aA strain 125.91, and the ${alpha}$ A strain H99. The mating type-specific regions are shown as thick lines, and the flanking regions are shown as thin lines. Genes are represented as arrows in the direction of transcription. Genes encoding pheromones are shown as black arrows, locus-specific genes are shown as white arrows, and all other genes are in grey. The IKS1 genes are circled with a black line, the RPO41 genes are circled with a dotted line, and the NCP1a gene is boxed. (D) PCR assays were conducted with primers specific to NCP1a. Genomic DNAs from serotype D, serotype A, and AD hybrid strains were assayed. aAD ${alpha}$ hybrid strains were ZG290, ATCC 48184, CDC92-74 (AD ${alpha}$ ), CDC228, and CDC304. ${alpha}$ ADa hybrid strains were ZG287, MMRL774, KW5, CBS132, and CDC94-383. PCR with primers specific to the actin gene served as a positive control for the presence of genomic DNA.

In a related study by Lengeler et al. (35) the MAT loci of strain125.91 and reference serotype A and D strains were sequenced.Comparison of these MAT loci supports the conclusion that strain125.91 is a serotype A, a mating type strain (Fig. 1C). TheMAT locus of strain 125.91 contains only mating factor a genesand no mating factor ${alpha}$ pheromone genes. Sequence homology washighest between strains 125.91 and JEC20 (aD) for most genesin the MAT locus, with the notable exception of the serotypeA RPO41a gene, which was more similar to the serotype A RPO41 ${alpha}$ allele of strain H99. The IKS1 gene flanks the MAT locus inserotype D but is part of the MAT locus in both serotype A strains(H99 and 125.91). Importantly, the NCP1a gene was present onlyin the MAT locus of the serotype D mating type a strain. Herethe NCP1a gene was found to be present in all aD strains testedand to be absent in all ${alpha}$ strains of either serotype tested (Fig.1D and data not shown). In AD hybrid strains heterozygous forthe MAT locus, the NCP1a gene was present only in hybrids thatinherited the MATa allele from the serotype D parent ( ${alpha}$ ADa) andnot in those that inherited the serotype A MATa allele (aAD ${alpha}$ )(Fig. 1D). Based on these findings, we conclude that strain125.91 contains a serotype A MATa allele at the MAT locus.

Strain 125.91 mates with a subset of serotype A, ${alpha}$ mating type strains. By molecular analysis strain 125.91 is an aA strain, yet itfails to mate with the ${alpha}$ reference strains H99 and JEC21 (36).Comparison of the entire MAT locus sequences of the sterileaA strain 125.91 and the fertile aD strain JEC20 reveals thatall of the genes in the fertile MATa allele are present in the125.91 MAT locus except the NCP1a gene, which shares homologywith a Neurospora crassa gene of unknown function (35). ncp1amutants are fertile (22), and aAD ${alpha}$ hybrids that result from intervarietalmating lack NCP1a (Fig. 1C); thus, the absence of this genecannot account for the apparent sterility of strain 125.91.Alterations in temperature, medium, or supplementation withcyclic AMP did not stimulate mating of strain 125.91.

Because strain 125.91 (aA) cannot mate directly with the ${alpha}$ A referencestrain H99, we tested the ability of strain 125.91 to mate withother ${alpha}$ A strains. A total of 150 strains were tested, and matingwas examined on four different media that support mating ofserotype D strains (V8 [pH 7], carrot, or SLAD medium) or serotypeB strains (V8 [pH 5]). Three strains mated with strain 125.91.Strain S25J is an ${alpha}$ A clinical isolate from Asia that mated withstrain 125.91 only on V8 (pH 5). Strain 4476B, a clinical isolatefrom Uganda, mated with strain 125.91 only on carrot mediumand was found to be an ${alpha}$ AD hybrid strain. Strain 8-1 is an ${alpha}$ Aclinical isolate from a U.S. organ transplant center, and itmated with strain 125.91 on all media tested and to the greatestextent on V8 (pH 5) (Fig. 2). We focused on strain 8-1.

View larger version (135K):
[in this window]
[in a new window]
FIG. 2. Morphological features of C. neoformans var. grubii mating. The a mating type, serotype A strain 125.91 was mixed with the ${alpha}$ mating type, serotype A strain 8-1. (A) Example of filaments observed after 10 days of growth on V8 (pH 5.0) at 25°C in the dark. Bar, 1 mm. (B) Example of filaments, basidia, and spores produced after 10 days of growth on V8 (pH 5.0) at 25°C in the dark. Bar, 25 µm. (C) DAPI staining of nuclei in filaments produced in a 125.91 (aA)-to-8-1 ( ${alpha}$ A) cross. Arrows designate paired nuclei in the dikaryons. Bar, 25 µm. (D) Differential contrast image of the filaments seen in panel C. (E) Calcofluor White staining of the septa of filaments produced in a 125.91 (aA)-to-8-1 ( ${alpha}$ A) cross. The arrowhead designates a fused clamp connection. Bar, 25 µm. (F) DIC image of the filament seen in panel E.

We first verified that strain 8-1 is an authentic ${alpha}$ A strain.Strain 8-1 typed as serotype A in an antibody agglutinationtest (reacted with antigens 1 and 7) and was haploid by FACSanalysis (Fig. 1A). By PCR, strain 8-1 contains only the serotypeA alleles of the STE20 ${alpha}$ , PAK1, GPA1, and CNA1 genes, and thesequences of these genes are more similar to serotype A thanto serotype D reference sequences. Finally, by transposon fingerprinting,strain 8-1 lacks the T1 and T2 elements characteristic of serotypeD strains (Fig. 1B). Thus, strain 8-1 is a haploid serotypeA strain.

Characterization of strain 125.91 mating. Mating between the aA strain 125.91 and the ${alpha}$ A strain 8-1 producedfilaments characteristic of mating. Filament cells containedtwo nuclei (dikaryons), were linked by fused clamp cells whosesepta stained with calcofluor white, and produced basidium fruitingbodies decorated with four long chains of basidiospores (Fig.2). These morphological features are all hallmarks of matingof serotype D C. neoformans var. neoformans strains (28) andare distinct from those of haploid fruiting filaments (monokaryotic,unfused clamps, and short spore chains) (59).

Basidiospores isolated from the 125.91 (aA)-to-8-1 ( ${alpha}$ A) matinggerminated ( $~$ 25%) to produce viable meiotic progeny. Of 15 progenytested, all were serotype A by antibody agglutination testsand PCR analysis of the STE20, PAK1, GPA1, and CNA1 genes (notshown). Seven meiotic segregants were ${alpha}$ , six were a, and twowere indeterminate, based on mating to the parental strainsand PCR analysis with STE20 primers specific for the ${alpha}$ and amating types (not shown). These results are consistent withan approximate 1:1 ratio of ${alpha}$ to a progeny.

RFLP analysis of the GPA1 and PAK1 genes combined with matingtype analysis established that recombinant progeny were producedby the 125.91 (aA)-to-8-1 ( ${alpha}$ A) cross. Strain 125.91 containsa PvuI restriction site near the 3' end of the GPA1 gene thatis not present in strain 8-1. Digestion of a PCR product spanningthe GPA1 gene from strain 125.91 resulted in cleavage of thegene fragment to produce 709- and 82-bp products, compared tothe 791-bp product from strain 8-1 (Fig. 3A). Similarly, thePAK1 gene in strain 125.91 contains a BstXI restriction siteat the 3' end of the gene that is absent in strain 8-1. Digestionof the PCR-amplified PAK1 gene from strain 125.91 resulted incleavage of the gene fragment to produce 271- and 47-bp products,compared to the 318-bp product from strain 8-1 (Fig. 3A). Ofthe 15 progeny tested for recombination, 9 progeny exhibitedthe RFLP and mating type of the parental strains 125.91 (aA)and 8-1 ( ${alpha}$ A), whereas $~$ 4 segregants with either parental genotypewould be expected based on random segregation of three markersand 16 progeny analyzed. Other progeny were clearly productsof meiotic recombination. Four progeny (two a, one ${alpha}$ , and oneindeterminate mating type [Fig. 3A]) inherited the strain 8-1GPA1 allele and the strain 125.91 PAK1 allele, and one progeny(KNA13, a mating type [Fig. 3A]) contained both the 8-1 and125.91 GPA1 alleles and was aneuploid by FACS analysis (Fig.1A). One of the meiotic progeny, designated strain KNA14, inheritedthe GPA1 and PAK1 strain 8-1 alleles and the a mating type fromstrain 125.91 (Fig. 3A).

View larger version (44K):
[in this window]
[in a new window]
FIG. 3. RFLP analysis of progeny from C. neoformans var. grubii crosses shows recombination. The presence or absence of the PvuI restriction site in the GPA1 gene and the BstXI restriction site in the PAK1 gene was examined by digestion of the PCR-amplified gene with the appropriate enzyme. An asterisk indicates the strain used to generate the serotype A congenic strains. (A) RFLP analysis of the parental strains 125.91 (aA) and 8-1 ( ${alpha}$ A) and progeny from a cross of the parental strains. Strain KNA14 was used to generate the serotype A congenic strains. (B) RFLP analysis of the parental strains KNA14 (aA) and H99 crg1 ( ${alpha}$ A) and progeny from a cross of the parental strains. KNB-5 was used to generate the serotype A congenic strains. (C) RFLP analysis of the parental strains KNA14 (aA) and H99 ( ${alpha}$ A), the 5th-backcross strain KN99-5a, and the 10th-backcross strains KN99a and KN99 ${alpha}$ .

Progeny from the 125.91 (aA)-to-8-1 ( ${alpha}$ A) cross exhibited differencesin ability to mate with the parental strains and the referenceserotype A and D strains (Table 1). Strain KNA14 mated withthe ${alpha}$ A parental strain 8-1 and the ${alpha}$ A and ${alpha}$ D reference strainsH99 and JEC21 to produce viable spores (Fig. 4A and B). PFGEwas used to compare the karyotypes of strains 125.91, 8-1, KNA14,and H99 to determine whether differences in chromosome numberor size might prevent successful meiosis and account for matingcompatibility differences (Fig. 5). The chromosomal bandingpatterns of strains 125.91 (aA) and 8-1 ( ${alpha}$ A) were similar, andboth strains lacked a high-molecular-weight band present inH99 ( ${alpha}$ A). The karyotype of strain KNA14 (aA) was similar to thoseof the 8-1 ( ${alpha}$ A) and 125.91 (aA) parental strains but containedtwo lower-molecular-weight bands, one from each parent. We notethat strain KNA14 (aA) mates normally (Table 1) and is haploidby FACS analysis (Fig. 1A).

View this table:
[in this window]
[in a new window]
TABLE 1. Mating abilities of serotype A strains of C. neoformans var. grubii

View larger version (129K):
[in this window]
[in a new window]
FIG. 4. The crg1 mutation enhances filamentation during mating of serotype A strains. The aA strain KNA14 was mated with the ${alpha}$ A strain H99 or an isogenic crg1 ${alpha}$ A mutant strain on V8 (pH 5.0) medium at 25°C in the dark for 10 days. (A) Example of aA (KNA14)-to- ${alpha}$ A (H99) mating. Bar, 500 µm. (B) Filaments, basidia, and spores produced during an aA (KNA14)-to- ${alpha}$ A (H99) mating. Bar, 50 µm. (C) Example of an aA (KNA14)-to- ${alpha}$ A crg1 (H99 crg1) mutant mating. Bar, 100 µm. (D) Filaments, basidia, and spores produced during an aA (KNA14)-to- ${alpha}$ A crg1 (H99 crg1) mutant mating. Bar, 50 µm.

View larger version (73K):
[in this window]
[in a new window]
FIG. 5. Karyotype analysis of C. neoformans var. neoformans strains. Chromosomes from the aA strains 125.91, KNA14, KN99-5a, and KN99a and the ${alpha}$ A strains 8-1, H99, and KN99 ${alpha}$ were separated by PFGE, and the gel was stained with ethidium bromide. Size markers are based on S. cerevisiae chromosomes.

Production of C. neoformans var. grubii congenic strains. Congenic strains of serotype D have been invaluable in classicalgenetic studies of the physiology and virulence of C. neoformansvar. neoformans (20, 30). However, the majority of clinicalisolates of C. neoformans are serotype A (C. neoformans var.grubii). Therefore, it would be of great experimental valueto have congenic strains in the more common serotype A geneticbackground. The mating observed between the serotype A strains125.91 (aA) and 8-1 ( ${alpha}$ A) suggested that congenic strains couldbe generated in serotype A. However, strain H99 ( ${alpha}$ A) is the sequencereference strain with 11x genome-wide sequence coverage anda BAC fingerprint map (45). Thus, it would be preferable togenerate congenic strains in this strain background.

Since the KNA14 (aA) F₁ progeny of the 125.91 (aA)-to-8-1 ( ${alpha}$ A)cross mated with H99 ( ${alpha}$ A), we used this strain to generate congenicserotype A strains. However, strain KNA14 (aA) mated with eitherthe ${alpha}$ A strain 8-1 or the ${alpha}$ D strain JEC21 to produce long hyphalfilaments with basidia and spores (Fig. 2A and B), whereas basidiaand spores were produced on very short hyphae closely associatedwith the yeast colony when KNA14 was crossed with the ${alpha}$ A strainH99 (Fig. 4A and B). Thus, isolation of spores from the KNA14(aA)-to-H99 ( ${alpha}$ A) cross was technically challenging due to theatypical short-hypha mating morphology. We sought to increasefilamentation in this mating to spatially separate spores fromvegetative yeast cells and allow microdissection. We used confrontationassays to examine filamentation of these strains. In confrontationassays, ${alpha}$ D strains filamented and aD strains enlarged in responseto mating pheromones prior to any direct cell-cell contact (Fig.6, left panel) (41). However, the ${alpha}$ A strain H99 failed to filamentwhen confronted with aA or aD strains (Fig. 6, middle panel,and data not shown). Likewise, the aA strains 125.91 and KNA14failed to filament or produce enlarged cells when confrontedwith ${alpha}$ A or ${alpha}$ D strains. Based on these results, we deduced thatpheromone insensitivity might impair filamentation in the KNA14(aA)-to-H99 ( ${alpha}$ A) cross.

View larger version (67K):
[in this window]
[in a new window]
FIG. 6. The crg1 mutation enhances pheromone response. The a and ${alpha}$ strains were grown in confronting lines $~$ 1 mm apart on SLAD medium and incubated at 25°C in the dark for 72 h. Strains were examined microscopically for the presence of conjugation tubes. aD, JEC20; ${alpha}$ D, JEC21; ${alpha}$ A, H99; ${alpha}$ A crg1, H99 crg1. Bar, 1 mm.

We addressed this apparent lack of pheromone response by creatinga pheromone-supersensitive strain. In Saccharomyces cerevisiae,mutation of a regulator of G-protein signaling (SST2) confersdramatic hypersensitivity to mating pheromone (10, 14). TheC. neoformans gene CRG1 (for Cryptococcus regulator of G proteins)encodes an SST2 homolog (P. Wang and J. Heitman, unpublisheddata), and an H99 crg1 mutant strain ( ${alpha}$ A) forms robust conjugationtubes in confrontation with the aD strain JEC20 (Fig. 6, rightpanel). Most importantly, the aA strain KNA14 mated with theH99 crg1 mutant ( ${alpha}$ A) to produce increased filamentation, abundantbasidia, and spores (Fig. 4C and D).

When spores from the KNA14 (aA)-to-H99 crg1 ( ${alpha}$ A) cross were microdissected,a and ${alpha}$ meiotic progeny segregated in a ratio of 1:1 based onmating assays. All of the a progeny mated with wild-type H99( ${alpha}$ A), producing long hyphae and abundant viable spores. Comparisonof the H99 and KNA14 GPA1 and PAK1 sequences showed that H99lacks the PvuI and BstXI restriction sites, respectively. RFLPanalysis of progeny from the KNA14-to-H99 crg1 cross showedthat recombinant progeny were produced (Fig. 3B).

Figure 7 depicts the mating scheme used to generate congenicaA and ${alpha}$ A strains. An aA progeny (KNB-5) from the KNA14 (aA)-to-H99crg1 ( ${alpha}$ A) cross was used to initiate nine backcrosses to wild-typeH99 ( ${alpha}$ A). The aA strain generated by the fifth backcross, designatedKN99-5a, and its descendents all contained the wild-type CRG1gene yet still produced long hyphae and abundant viable spores.The aA and ${alpha}$ A strains generated by the 10th backcross were designatedKN99a and KN99 ${alpha}$ , respectively. The 5th and 10th backcross strainswere examined by RFLP analysis and PFGE karyotyping for similarityto H99. KN99-5a (aA), KN99a (aA), and KN99 ${alpha}$ ( ${alpha}$ A) all containedthe H99 GPA1 and PAK1 alleles as determined by RFLP analysis(Fig. 3C). All three strains also had karyotypes identical tothat of H99 (Fig. 5). Additionally, the 5th and 10th backcrossstrains were equivalent to H99 in melanin production, capsuleproduction, and growth at 30 and 37°C (not shown). We havecurrently analyzed a limited number of genetic markers, butif we assume a 50% recombination frequency, the 5th backcrossstrain KN99-5a (aA) and the 10th backcross strains KN99a (aA)and KN99 ${alpha}$ ( ${alpha}$ A) are expected to be 96.9 and 99.9% similar to H99( ${alpha}$ A), respectively.

View larger version (16K):
[in this window]
[in a new window]
FIG. 7. Schematic diagram of the mating scheme used to produce a serotype A congenic strain pair. KN99-5a is the product of the 5th backcross and KN99a and KN99 ${alpha}$ are products of the 10th backcross to H99.

Virulence of congenic serotype A strains in animal models. We showed previously that the aA strain 125.91 is less virulentthan the ${alpha}$ A reference strain H99 in a murine model of cryptococcosis(36). The relative pathogenicities of the parental strains 8-1( ${alpha}$ A) and KNA14 (aA) were compared to that of H99 ( ${alpha}$ A) in a murineinhalation model (Fig. 8A). Both parental strains had attenuatedvirulence compared to H99 (Fig. 8A) (P = 0.001). In contrast,no difference in virulence was observed between H99 and thefifth-backcross strain KN99-5a (aA) when mice were inoculatedwith either 1 x 10⁵ (Fig. 8A) (P = 0.57) or 5 x 10⁴ (Fig. 8B)(P = 0.85) fungal cells. In addition, we tested the virulenceof the congenic 10th-backcross strains in both the mouse andrabbit models of cryptococcal meningitis. In the rabbit modelthere was no difference in the viability of the serotype A congenicstrains KN99a and KN99 ${alpha}$ in the CSF as determined by comparingquantitative yeast counts over the first week of infection (P> 0.05), and these strains were more virulent in comparisonto the reduced yeast counts of the serotype D strain JEC21 (P< 0.05) (Fig. 8C). While the rabbit model is not intendedto measure host survival, we observed that both serotype A congenicstrains caused 100% mortality in this model over a 10-day observationperiod, whereas rabbits infected with the serotype D strainwere alive and appeared well. Furthermore, there was no differencein virulence between the congenic strains H99 ( ${alpha}$ A), KN99a (aA),and KN99 ${alpha}$ ( ${alpha}$ A) in the murine model when animals were infectedwith either 10⁵ (P = 0.11 for KN99a and P = 1.0 for KN99 ${alpha}$ ) or10³ (P = 0.68 for KN99a and P = 0.31 for KN99 ${alpha}$ ) fungal cells(Fig. 8D). In summary, the serotype A congenic strains werefully virulent, and no difference in virulence was observedbetween the congenic a and ${alpha}$ mating type strains in either amurine inhalation model or the rabbit immunosuppressed infection.

View larger version (28K):
[in this window]
[in a new window]
FIG. 8. Virulence of serotype A strains. (A) Groups of 10 A/Jcr mice were infected with 10⁵ cells of the ${alpha}$ A strains H99 and 8-1, the aA strain KNA14, and the fifth-backcross progeny KN99-5a (aA), and survival was monitored. (B) Groups of 10 A/Jcr mice were infected with 5 x 10⁴ cells of H99 ( ${alpha}$ A) and KN99-5a (aA), and survival was monitored. (C) Groups of three immunosuppressed rabbits were inoculated with 10⁸ cells of strains KN99a (aA), KN99 ${alpha}$ ( ${alpha}$ A), and JEC21 ( ${alpha}$ D). CSF was withdrawn on days 4 and 7 postinfection, and the number of surviving yeast cells was determined by plating serially diluted CSF on YPD medium. (D) Groups of 10 A/Jcr mice were infected with either 10⁵ or 10³ cells of strains H99 ( ${alpha}$ A), KN99a (aA), and KN99 ${alpha}$ ( ${alpha}$ A), and survival was monitored. Error bars indicate standard errors of the means.

DISCUSSION

Top

Discussion

C. neoformans var. grubii serotype A strains cause $~$ 95% of allC. neoformans infections and >99% of the infections in AIDSpatients (9). The predominance of serotype A in clinical isolatesunderscores the importance of understanding the mechanisms ofvirulence in this serotype. Basidiospores are thought to bethe infectious propagule for C. neoformans, and yet spores aretypically produced by mating, which had not been observed inserotype A. Moreover, serotype A strains were thought to beclonal and only of the ${alpha}$ mating type. The recent identificationof diploid AD hybrid strains with a MATa mating type allelederived from serotype A first suggested that fertile a matingtype, serotype A strains might exist (34).

Subsequently, two aA strains were identified: one was a Tanzanianclinical isolate, strain 125.91, and the other was an Italianenvironmental isolate, strain IUM96-2828 (36, 41, 55). StrainIUM96-2828 (aA) has recently been shown to mate with the ${alpha}$ A referencestrain H99, and karyotypic analysis of progeny revealed evidenceof recombination (26), thus defining a sexual cycle in C. neoformansvar. grubii. While the aA strain IUM96-2828 mates with the referencestrain H99 ( ${alpha}$ A), the other aA strain, 125.91, appeared to havea mating defect (36). Here we examined mating of strain 125.91(aA) and discovered that this strain mates with only a subsetof ${alpha}$ A strains. Mating produced dikaryotic filaments with fusedclamp connections and basidial fruiting structures with longchains of spores. By RFLP analysis and mating type tests, wedocument that mating between the aA strain 125.91 and the ${alpha}$ Aclinical isolate 8-1 produced recombinant progeny with variablemating ability. The meiotic progeny strain KNA14 (aA) exhibitedthe widest mating repertoire and was used here with the serotypeA reference strain H99 ( ${alpha}$ A) to generate a congenic pair of aand ${alpha}$ mating type serotype A strains (KN99a and KN99 ${alpha}$ ). The productionof congenic strains was facilitated by the identification ofa crg1 pheromone-supersensitive mutant that lacks a regulatorof G protein signaling for pheromone response. The observationthat the 125.91 MAT locus could be used to produce congenicserotype A strains with apparently normal mating ability suggeststhat this locus is fully functional.

Virulence of C. neoformans var. neoformans strains has beenlinked to the MAT locus. The effect of mating type is unclearin a heterogeneous population, but comparison of the virulenceof a and ${alpha}$ congenic strains of serotype D showed that ${alpha}$ strainswere more virulent than a strains (30, 33, 60). We examinedthe role of mating type in virulence in C. neoformans var. grubiistrains. The aA strain 125.91 is less virulent that the ${alpha}$ A referencestrain H99 in animal models (36) and is less virulent than itsmating partner strain 8-1 ( ${alpha}$ A). Interestingly, the virulenceof congenic aA and ${alpha}$ A strains in the H99 background was equivalentin murine inhalation and rabbit meningitis models of cryptococcosis.Further studies conducted under a broad range of conditionsmay be necessary to reveal differences between these strains,but at present it appears that mating type does not play a predominantrole in virulence of C. neoformans var. grubii strains.

The observation that mating type does not appear to play a dominantrole in virulence of serotype A strains raises an interestingquestion: why are the majority of clinical serotype A isolatesof the ${alpha}$ mating type? Sexual reproduction is expected to generatea 1a:1 ${alpha}$ ratio of spores, and this ratio is seen in our laboratorycrosses. If sexual reproduction is the primary method by whichthe infectious propagule is generated, then a higher proportionof clinical isolates would be expected to be of the a matingtype. Some serotype AD strains can filament and produce sporesthat germinate poorly (<5%) and typically are still AD hybrids,and thus this cannot account for the predominance of matingtype ${alpha}$ , serotype A clinical isolates (34). An alternative possibilityis that spores could be produced via the asexual process ofhaploid fruiting. In serotype D strains, haploid fruiting islinked to the ${alpha}$ mating type (59). If this were also true in serotypeA, it could account for an increased prevalence of ${alpha}$ A clinicalisolates. However, we have not yet observed haploid fruitingof any serotype A isolates, and as-yet-unknown environmentalconditions may be necessary. It is also possible that aA strainsare less virulent than ${alpha}$ A strains in humans yet due to host organismdifferences, this lack of virulence is not detected in the mouseor rabbit models.

The fact that the majority of clinical serotype A isolates are ${alpha}$ mating type could also be due to the relative abundance of ${alpha}$ A strains in the environment. Analysis of $~$ 1,500 serotype A strainsworldwide has identified only two aA strains, one a clinicalisolate from Africa and the other an environmental isolate fromsoil near an aviary in Italy. Because aA strains appear to berare, sexual reproduction to produce the infective propagulemay take place only in geographically restricted areas whereboth mating types are present. Alternatively, aA strains mightnot survive well in the environment. For example, ${alpha}$ strains maybe more resistant than a strains to predation by amoebas andnematodes, resulting not only in increased virulence of some ${alpha}$ strains (42, 50) but also in a decreased a strain populationsize.

With the characterization of a sexual cycle in C. neoformansvar. grubii, sexual reproduction has been defined or reportedfor all C. neoformans varieties. However, the function of thesexual cycle in C. neoformans and its role in pathogenicityof this organism still remain unclear. The observation thatthe aA strain 125.91 mates with only a few ${alpha}$ A strains might suggestthat C. neoformans is evolving to be asexual and that the aand ${alpha}$ mating types are increasingly genetically isolated. Thissituation may be similar to that observed with other fungalpathogens and parasites. The ability of the human pathogen Histoplasmacapsulatum to undergo a sexual cycle is lost with laboratorypassage of isolates (62). Loss of sexual reproduction upon prolongedculture has also been reported for C. neoformans var. neoformansstrains (63). The obligate diploid C. albicans is primarilyclonal (17) and may only rarely utilize its recently discoveredsexual cycle (25, 38, 40). Recent studies of the parasite Toxoplasmagondii reveal that clinical isolates are largely clonal andrestricted to three highly related sibling clades that descendedfrom a single recent sexual cross (<10,000 years ago) thatenabled efficient oral transmission without the sexual phase(51). In addition, completion of the sexual cycle in T. gondiican yield recombinant progeny with dramatically enhanced virulence(18). In the oomycete Phytophera infestans, mating-competentstrains are geographically restricted and produce recombinantprogeny that emigrate to cause clonal outbreaks of fungal diseaseon plants (16). Our studies suggest that the sexual cycle ofC. neoformans may similarly be geographically restricted andmay infrequently lead to the production of recombinants withaltered fates that could be important in the evolution of thishuman pathogen.

ACKNOWLEDGMENTS

We thank Arturo Casadevall, Vishnu Chaturvedi, Mahmoud Ghannoum,Harish Gugnani, Sam Messer, Juneann Murphy, Michael Pfaller,and Wiley Schell for strains. We thank James Fraser, ChristinaHull, Alex Idnurm, and Klaus Lengeler for helpful comments anddiscussions; Cristl Arndt for technical assistance; and othermembers of the Heitman lab for their support.

This work was supported in part by NIAID R01 grant AI50113 toJoseph Heitman, NIAID R01 AI28388 grant to John Perfect, andan NIAID program project grant AI44975 to the Duke UniversityMycology Research Unit. Gary Cox is a Burroughs Wellcome NewInvestigator, and Joseph Heitman is a Burroughs Wellcome Scholarin Molecular Pathogenic Mycology and an Associate Investigatorof the Howard Hughes Medical Institute.

FOOTNOTES

* Corresponding author. Mailing address: Department of Molecular Genetics and Microbiology, 322 CARL Building, Research Dr., Duke University Medical Center, Durham, NC 27710. Phone: (919) 684-2824. Fax: (919) 684-5458. E-mail: heitm001{at}duke.edu.

Editor: T. R. Kozel

Present address: Department of Pediatrics and Microbiology,Children's Hospital, New Orleans, LA 70118.

REFERENCES

Top