Discovery of a Modified Tetrapolar Sexual Cycle in
Cryptococcus amylolentus and the Evolution of MAT in
the Cryptococcus Species Complex
Keisha Findley1,2., Sheng Sun2., James A. Fraser3, Yen-Ping Hsueh4, Anna Floyd Averette2, Wenjun Li2,
Fred S. Dietrich2, Joseph Heitman2*
1Genetics and Molecular Biology Branch, National Human Genome Research Institute, National Institutes of Health, Bethesda, Maryland, United States of America,
2Department of Molecular Genetics and Microbiology, Duke University Medical Center, Durham, North Carolina, United States of America, 3School of Molecular and
Microbial Sciences, University of Queensland, Brisbane, Australia, 4Division of Biology, California Institute of Technology, Pasadena, California, United States of America
Sexual reproduction in fungi is governed by a specialized genomic region called the mating-type locus (MAT). The human
fungal pathogenic and basidiomycetous yeast Cryptococcus neoformans has evolved a bipolar mating system (a, a) in which
the MAT locus is unusually large (.100 kb) and encodes .20 genes including homeodomain (HD) and pheromone/receptor
(P/R) genes. To understand howthis unique bipolar matingsystem evolved, we investigated MAT in the closely related species
Tsuchiyaea wingfieldii and Cryptococcus amylolentus and discovered two physically unlinked loci encoding the HD and P/R
genes. Interestingly, the HD (B) locus sex-specific region is restricted (,2 kb) and encodes two linked and divergently oriented
homeodomain genes in contrastto the solo HD genes (SXI1a, SXI2a) of C. neoformansand Cryptococcus gattii. TheP/R(A) locus
(Cryptococcus heveanensis) and is linked to many of the genes also found in the MAT locus of the pathogenic Cryptococcus
species. Our discovery of a heterothallic sexual cycle for C. amylolentus allowed us to establish the biological roles of the sex-
determining regions. Matings between two strains of opposite mating-types (A1B16A2B2) produced dikaryotic hyphae with
fused clamp connections, basidia, and basidiospores.Genotyping progeny using markers linked and unlinked to MAT revealed
that meiosis and uniparental mitochondrial inheritance occur during the sexual cycle of C. amylolentus. The sexual cycle is
tetrapolar and produces fertile progeny of four mating-types (A1B1, A1B2, A2B1, and A2B2), but a high proportion of progeny
are infertile, and fertility is biased towards one parental mating-type (A1B1). Our studies reveal insights into the plasticity and
transitions in both mechanisms of sex determination (bipolar versus tetrapolar) and sexual reproduction (outcrossing versus
inbreeding) with implications for similar evolutionary transitions and processes in fungi, plants, and animals.
Citation: Findley K, Sun S, Fraser JA, Hsueh Y-P, Averette AF, et al. (2012) Discovery of a Modified Tetrapolar Sexual Cycle in Cryptococcus amylolentus and the
Evolution of MAT in the Cryptococcus Species Complex. PLoS Genet 8(2): e1002528. doi:10.1371/journal.pgen.1002528
Editor: Hiten D. Madhani, University of California San Francisco, United States of America
Received September 2, 2011; Accepted December 21, 2011; Published February 16, 2012
This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for
any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Funding: This work was supported by NIAID R37 grant AI39115-14 and R01 grant AI50113-08 to JH and an NIH Minority Supplement 5R01-AI063443-04 S1 Sub #
1-P30 that supported KF. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: email@example.com
. These authors contributed equally to this work.
Sexual reproduction is ubiquitous throughout nature, generates
population diversity, and has been described extensively in plants,
animals, and microorganisms . Sex is both costly and advanta-
its benefits outweigh its costs . In sexually reproducing populations,
outbreeding is common, but inbreeding forms of sex also occur that
promote clonality. Additionally, unisexual reproduction may be an
adaptive virulence strategy for several microbial pathogens .
Fungi occur in two mating configurations: bipolar and tetrapolar
. In bipolar species, transcription factors that establish mating-
type (MAT) are encoded by a single locus; in some examples genes
encoding pheromones and their receptors are also present . For
mating to occur compatible cells must differ at MAT (a and a),
although there are examples of bipolar fungi that also undergo
same-sex mating (e.g. Candida albicans and Cryptococcus neoformans ).
In tetrapolar species, two physically unlinked genomic regions (i.e.
MAT loci A and B) control and establish cell identity. These loci are
often multiallelic, and alleles must differ at both loci for sexual
reproduction to occur. Bipolar mating systems support more
efficient inbreeding (50%) and also outbreeding (50%), while
tetrapolar systems promote more efficient outbreeding (.99%)
and restrict inbreeding (25%) . Ascomycetous yeasts such as
Saccharomyces cerevisiae and Candida albicans are bipolar while
basidiomycetous yeasts like Tremella mesenterica and Ustilago maydis
are typically tetrapolar . In contrast to most basidiomycetous
species, Ustilago hordei, Coprinellus disseminatus, C. neoformans, and
Cryptococcus gattii have bipolar mating systems [8,9,10,11,12].
C. neoformans is a haploid, dimorphic fungus that has a bipolar
mating system, represented by two alleles, a and a . MAT
spans 100 to 120 kb, and encodes more than 20 genes, many of
which are involved in mating. Comparison of the MAT gene
cluster among the members of the pathogenic Cryptococcus species
complex revealed that extensive rearrangements and gene
conversions have occurred over time even though recombination
PLoS Genetics | www.plosgenetics.org1February 2012 | Volume 8 | Issue 2 | e1002528
in this gene cluster is generally suppressed [13,14,15,16]. The
sexual cycle and the structure of MAT in the pathogenic
Cryptococcus species have been extensively examined and are well
defined [4,15]. In a laboratory setting, Cryptococcus reproduces via
either opposite-sex or unisexual reproduction [5,11,12,14,17,18].
Mating (a-a) initiates with cell-cell fusion, followed by production
of a filamentous dikaryon with fused clamp cell connections, and
culminates in nuclear fusion and meiosis in the basidia [4,19].
Meiosis produces four haploid nuclei that undergo mitotic division
to produce four chains of basidiospores that germinate into fertile
yeasts that can mate with a partner/parent of the opposite mating-
type. The major differences in a-a unisexual reproduction is that a
monokaryon (instead of a dikaryon) forms, mating can involve two
genetically distinct isolates (a1-a2) or two genetically identical
genomes (a1-a1), and the resulting meiotic spore products are all a.
Fraser et al. proposed that the ancestral form of MAT to the
pathogenic Cryptococcus species was tetrapolar, with the homeodo-
main (HD) and pheromone/receptor (P/R) genes present in two
unlinked sex-determining regions . Sequential rounds of gene
acquisition led to the expansion of the ancestral tetrapolar MAT
loci. In this model, a chromosomal translocation event then fused
the unlinked loci into a contiguous region resulting in the
formation of a transient tripolar intermediate in which MAT is
linked in one partner yet unlinked in the other. This unstable
intermediate underwent gene conversion to link the other MAT
locus alleles, one or the other homeodomain gene was lost, and
MAT was subjected to multiple inversions and gene conversions
events to yield the extant bipolar MAT locus of Cryptococcus [16,20].
The pathogenic Cryptococcus species form a monophyletic cluster
composed of at least two but possibly as many as six species: C.
neoformans var. neoformans, C. neoformans var. grubii, and the sibling
species C. gattii (VGI, VGII, VGIII, VGIV) that all have the
potential to infect humans and other animals . A recent multi-
locus sequence typing (MLST) phylogenetic study resolved the
species relationships in this complex . The monophyletic sensu
stricto Filobasidiella clade is comprised of the pathogenic species and
three closely related saprobic species: Tsuchiyaea wingfieldii,
Cryptococcus amylolentus, and Filobasidiella depauperata [22,23]. The
more distantly related sensu lato sister clade Kwoniella encompasses
several saprobic and one aquatic-associated species: Bullera
dendrophila, Cryptococcus heveanensis, Cryptococcus bestiolae, Cryptococcus
dejecticola, and Kwoniella mangroviensis .
Of these species that are phylogenetically closely related to the
pathogenic Cryptococcus species complex, sex has recently been
described for C. heveanensis and K. mangroviensis [24,25]. Specifically, a
heterothallic sexual cycle was observed in these two members of the
Kwoniella clade and basidiospores associated with cruciate-septated
basidia are produced during mating. Additionally in F. depauperata
and T.mesenterica, the nature of sex has alsobeen revealed inprevious
studies and exemplifies homothallic and heterothallic sexual cycles,
respectively [12,26,27,28]. The mating structures of F. depauperata
resemble the basidia and basidiospores of C. neoformans and C. gattii
while T.mesenterica mating products aresimilar to C. heveanensis and K.
mangroviensis [12,26,27,28,29]. However, no sexual reproduction had
been observed in either C. amylolentus or T. wingfieldii.
A recent study of C. heveanensis revealed it has a tetrapolar mating
system, i.e. its sexual reproduction is governed by MAT comprised
of two physically unlinked gene clusters: a multiallelic HD locus (B
locus) and a P/R locus (A locus) that is at least biallelic .
However, it still remains unclear when the bipolar mating system in
Cryptococcus pathogenic species first appeared, that is, did it emerge
earlier in the common ancestor of the sensu stricto group when it split
from the sensu lato group, or did it evolve later and only in the
Cryptococcus pathogenic species? Given the close relationship of T.
wingfieldii and C. amylolentus to the pathogenic Cryptococcus species
complex, understanding their life cycles, as well as their MAT loci
configurations can provide key insights into the evolution of MAT
and sexual reproduction in C. neoformans and C. gattii.
In this study, we provide a detailed description of the
heterothallic sexual cycle of C. amylolentus that we observed under
laboratory conditions. Additionally, we characterized the MAT
loci of T. wingfieldii and C. amylolentus, and discovered in both
species two physically unlinked gene clusters, one encoding the
HD locus and the other encoding the P/R locus. Genes within
these clusters include many homologs of Cryptococcus MAT-
associated genes. Furthermore, our mating assay and genetic
analyses of C. amylolentus meiotic progeny showed that many
meiotic progeny are sterile and one parental type is overrepre-
sented in the meiotic products, suggesting its tetrapolar mating
system deviates from the classic model. We discuss the implications
of our findings in the context of the evolution of the mating type
locus as well as of bipolar sexuality in the Cryptococcus species
complex. Our findings also provide insights into similar evolu-
tionary processes that drive the formation and function of sex
chromosomes in algae, fish, insects, and mammals .
Characterizing MAT in T. wingfieldii and C. amylolentus
To determine the structure of MAT in T. wingfieldii, fosmid
libraries were constructed from the type strain CBS7118 and
probed with several genes within (MYO2, LPD1, and SXI1) or
flanking (FAO1 and NOG2) the C. neoformans MAT locus. Positive
clones (3F11-3A15-5J15 (P/R locus), 2B23-2K10 (HD locus), and
4E07 (FAO1), see Figure S1) were pooled and sequenced, resulting
in the identification of two candidate MAT loci. The FAO1 gene
lies on a distinct fosmid and appears to be unlinked or distant from
MAT. The region obtained containing the P/R locus spans
Fungal gene clusters mediate sex determination, natural
product synthesis, and metabolic functions. Eukaryotic
organisms share features of gene cluster formation
including translocations, inversions, gene conversion, and
suppressed recombination. The C. neoformans/C. gattii
mating-type (MAT) locus spans a single .100 kb gene
cluster encoding .20 genes, many involved in sex. We
examined MAT gene cluster evolution in model and
pathogenic Cryptococcus species. MAT was characterized
from two closely related species, T. wingfieldii and C.
amylolentus, and is organized into two unlinked gene
clusters on different chromosomes. MAT organization in
these species provides insight into evolutionary transitions
from tetrapolar to bipolar mating systems involving fusion
of physically unlinked sex-determinants into one contigu-
ous region. These sex determination transitions occurred
concomitantly with the origin of the pathogenic species
complex from the last common ancestor shared with
tetrapolar non-pathogenic species. We discovered a
tetrapolar sexual cycle in C. amylolentus that generates
recombinant meiotic progeny, many of which are infertile.
Fertile progeny are biased towards one parental mating-
type (A1B1) and may be an evolutionary precursor to
unisexual mating of the closely related pathogenic species.
This study reveals factors orchestrating gene cluster
formation and sex chromosome evolution in fungi,
including features shared with animals and plants.
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org2February 2012 | Volume 8 | Issue 2 | e1002528
,70 kb and the region obtained containing the HD locus spans
,40 kb (Figure 1A).
We also cloned and sequenced MAT in the saprobic yeast-like,
sibling species C. amylolentus (type strain CBS6039) employing the
same approach. Fosmid libraries were generated and probed with
MYO2, LPD1, RPL39, and SXI1. Primers specific for MAT genes in
T. wingfieldii were used to generate probes for C. amylolentus and the
identity of each probe was confirmed via cloning and sequencing.
Positive clones (4E01 (SXI1), 4E22 (MYO2), and 3H19 (LPD1), see
Figure S2) were individually sequenced and assembled into two
MAT loci. An additional fosmid (3N14 (RPL39), see Figure S2) was
later identified and sequenced via primer walking. The regions
that were sequenced span ,20 kb and ,60 kb respectively, and
each contain two small sequence gaps (Figure 2A and Figures S3
and S4). The linear order of the fragments in the P/R locus was
determined based on Southern blotting. Specifically, genomic
DNA from CBS6039 and CBS6273 was digested with five
restriction enzymes (BamHI, BglI, ClaI, EcoRI, and NcoI) and
Southern blot analysis was performed with probes hybridizing to
the ends of each contig in the P/R assembly of C. amylolentus. The
gene content in MAT appears to be largely conserved between T.
wingfieldii and C. amylolentus. However, our Southern blot analysis
indicated at least two major inversions exist between the P/R
regions of these two species (Figure 3).
Analysis of the MAT sequences obtained from T. wingfieldii and C.
amylolentus revealed that the gene content of these regions are similar
to the C. neoformans and C. gattii MAT alleles [10,16]. In both sibling
species, orthologs of both SXI1 and SXI2 are present in the HD
locus implicating this as the ancestral configuration (Figure 1 and
Figure2).Theorientationofthe homeodomain transcriptionfactors
mirrors the organization of the paired, divergently transcribed
genes, bE and bW, in the tetrapolar basidiomycete U. maydis [29,30].
In contrast, inC. neoformansand C. gattii, onlyoneHD gene is present
and SXI1a is specific to the a allele while SXI2a is specific to the a
allele. The region corresponding to the P/R locus contains the
mating pheromone genes, the pheromone receptor gene STE3, and
the five genes that were hypothesized to be those most recently
acquired by the Cryptococcus MAT locus (LPD1, RPO41, BSP2, CID1,
and GEF1). In T. wingfieldii, three pheromone genes (MFa1 and
MFa3 are identical while MFa2 differs in only one amino acid) are
present and share greater identity with the MFa genes of C. gattii
with an identity of 80% compared to 70–75% shared with the MFa
pheromone gene (Figure 4A). The P/R region in C. amylolentus
differs from C. heveanensis in that the pheromone genes are located
.30 kb away from STE3 whereas in C. heveanensis these genes are
closely linked . Moreover LPD1, STE11, ZNF1, and IKS1 are
not within the P/R locus of C. heveanensis , while the P/R region
is more extensive in C. amylolentus and spans .60 kb (Figure 2 and
Figure 1. T. wingfieldii MAT loci and chromosomal locations. (A) Six fosmids were analyzed to generate the assembly for T. wingfieldii. The MAT
gene probes used to probe the T. wingfieldii library are indicated in blue. The HD (B) and P/R (A) loci are embedded within assemblies that span 40
and 70 kb respectively. Grey arrows indicate genes that either flank MAT or are hypothetical genes, black arrows are Cryptococcus MAT-specific genes,
and yellow indicates the genes most recently acquired into the Cryptococcus MAT locus. Scale bar=10 kb. (B) Chromosomes from T. wingfieldii were
separated using PFGE, followed by Southern hybridization using three MAT-specific probes, two from the HD locus and one from the P/R locus.
Arrows depict hybridization of HD genes to an ,1 Mb chromosome distinct from hybridization of the P/R genes to an ,1.1 Mb chromosome.
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org3February 2012 | Volume 8 | Issue 2 | e1002528
Figure S2). In C. amylolentus, two pheromone genes (MFa1 and
MFa2 differin only two amino acids) have been identified and share
73% identity with the MFa protein product of C. neoformans and 65–
70% identity with the MFa pheromone gene. In summary, both
SXI1 and SXI2 were present in the ancestral HD locus of the sensu
stricto Filobasidiella clade. Thus, loss of one or the other HD gene
occurred during the evolution of MAT in the pathogenic Cryptococcus
species. Additionally, the five genes most recently acquired by the
Cryptococcus MAT locus are linked to the ancestral P/R locus and
thus appear to have been acquired into the expanding MAT A locus
rather than entrapped by the MAT fusion event, in contrast to an
earlier evolutionary model, suggesting a revision to the model
(Figure 5) .
In T. wingfieldii and C. amylolentus, the FCY1 and UAP1 genes flank
the 59 end of the MAT HD locus, similar to C. neoformans/C. gattii, but
FAO1 is unlinked and present elsewhere in the genome. We observed
that STE11 is not present in the P/R locus but, based on PCR
C. amylolentus (data not shown). In the MAT locus of the pathogenic
Cryptococcus species, STE11 is present. In C.heveanensis, STE11 islinked
configuration with retention in C. neoformans and C. gattii and
translocation out of MAT in C. amylolentus and T. wingfieldii . In T.
wingfieldii, the flanking gene at the 39 end of MAT, NOG2, was used as
a probe. It was present in a single contig within the larger fosmid
assembly of T. wingfieldii, but has not been linked to either the HD or
islinked tothe P/R locus,althoughthisgapremainsto besequenced.
Interestingly, NCP1 and NCP2 are duplicated genes in T. wingfieldii
and C. amylolentus but not in the pathogenic Cryptococcus species. The
configuration might be ancestral.
We also identified several hypothetical genes (CND06020,
CND06030, CND06040, CND01650, CNBE0480, CNE02690,
and CNE02670) with C. neoformans genes as the most closely related
homolog in other sequenced fungal genomes. Four of these genes
Figure 2. C. amylolentus MAT loci and chromosomal locations. (A) Four fosmids were analyzed to generate the assembly for C. amylolentus. The
MAT gene probes used to probe the C. amylolentus library are indicated in blue. The HD (B) and P/R (A) loci are embedded in regions that span 20 and
60 kb respectively. Grey arrows indicate genes that either flank MAT or are hypothetical genes, black arrows are Cryptococcus MAT-specific genes, and
yellow indicates the genes more recently acquired into the Cryptococcus MAT locus. Several gaps remain in the MAT loci of C. amylolentus. Scale
bar=10 kb. Green bars under the assembly denote gaps in sequence. (B) Chromosomes from C. amylolentus were separated using PFGE, and
analyzed by Southern hybridization using three MAT-specific probes, one from the HD locus and two from the P/R locus. The RPL22 gene was also
used as a probe. Arrows depict hybridization of the HD and P/R locus probes to distinct chromosomes (,1.1 and ,1.15 Mb).
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org4February 2012 | Volume 8 | Issue 2 | e1002528
reside on chromosome 4 and two on chromosome 5 of C.
neoformans, indicating that translocation (intra- and inter-chromo-
somal) events may have occurred between these two chromosomes
during the evolution of MAT in the pathogenic Cryptococcus species
[16,28]. In C. heveanensis, F. depauperata, and T. mesenterica there is
additional evidence for similar exchanges between chromosomes
A considerable level of synteny exists across both MAT loci in T.
wingfieldii and C. amylolentus, but we also observed at least two
major inversion events that have occurred between the two
genomes (highlighted in blue in the P/R locus, Figure 3).
Comparison of each sibling species to the C. neoformans serotype
D strain JEC21 revealed extensive gene rearrangements and
inversions present throughout MAT (Figure S5), similar to the
comparisons of MAT within the C. neoformans/C. gattii species
complex. The arrangement of the MAT loci in T. wingfieldii and C.
amylolentus corresponds to an evolutionary intermediate in MAT
evolution in which the loci (or their linked gene repertoire) have
expanded but not yet fused.
The HD and P/R loci are physically unlinked in
T. wingfieldii and C. amylolentus
Analysis using pulsed-field gel electrophoresis and Southern
hybridization demonstrated that the HD and P/R loci are
physically unlinked in T. wingfieldii, as well as in both strains of
C. amylolentus (CBS6039 and CBS6273). Each genome has
approximately 10–12 chromosomes ranging in size from 800 kb
to 2.2 Mb. Three genes were used to probe the T. wingfieldii
chromosomes, two from the HD locus, SXI1 and RPL22, and one
from the P/R locus, MYO2 (Figure 1B). For C. amylolentus, a total of
three genes were used as probes: one from the HD locus, SXI1,
and two from the P/R locus, MYO2 and ETF1 (Figure 2B). From
the chromoblot analysis, the two loci are located on separate
chromosomes (,1.1 and 1.15 Mb) in both of the sibling species.
That the HD and P/R loci are located on different chromosomes
suggests a tetrapolar mating configuration for both sensu stricto
species T. wingfieldii and C. amylolentus. Moreover, given the finding
that other more distant outgroup species (C. heveanensis, T.
mesenterica) are also tetrapolar , the most parsimonious
interpretation is that the tetrapolar configuration represents the
ancestral form of MAT and the bipolar state observed for the
pathogenic Cryptococcus species therefore arose even more recently
than revealed by previous studies of the more distantly related sensu
lato species C. heveanensis . Thus, the organization of MAT in
the sibling species resembles key aspects of the proposed
intermediates in the evolution of bipolar MAT in the pathogenic
Cryptococcus species from a tetrapolar ancestor.
Identification of key genes that define MAT
MAT is defined as a gene cluster (containing either HD and/or
P/R genes) whose sequence is divergent between two strains of
opposite mating-types. Based on the characterized structure of
MAT in both species, we sought to determine which genes in each
region govern and control sexual identity. The lack of additional
T. wingfieldii strains has made it difficult to assess experimentally
whether it has a sexual cycle and, if so, which genes are involved.
Fortunately, in C. amylolentus two strains are available and this
enabled our analysis of MAT and sex in this species resulting in the
discovery of an extant sexual cycle (described below).
Regions that define MAT typically display polymorphisms when
comparing sequences from strains of opposite mating-type while
the genes that flank MAT share a much higher level of identity
($99%). The SXI1 and SXI2 dimorphic region defines the
diverged region of the MAT B HD locus in C. amylolentus. We
aligned the nucleotide sequences and performed a matrix
comparison for the dimorphic region (,2 kb) spanning the SXI1
and SXI2 genes in CBS6039 and CBS6273. The diversity lies in
the region between the two genes, and their divergently oriented
Figure 3. Synteny analysis of MAT sequences from T. wingfieldii and C. amylolentus. On the left is the comparison of the HD (MAT B) locus,
while the comparison of the P/R (MAT A) locus is shown on the right. Red lines connecting T. wingfieldii and C. amylolentus sequences denote
conserved gene order; while blue lines indicate inverted orientations of the sequences from the two species. Green bars under the assembly denote
sequence gaps in the assembled contigs.
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org5 February 2012 | Volume 8 | Issue 2 | e1002528
59 regions span roughly 600 bp with a similarity score of 92%
(Figure 6). This region encodes the N-terminal dimerization
regions known to be variable and which also defines alleles in other
species (please see Text S1, and Figures S10, S11, S12, S13 for
further information on analyses of HD dimorphic region in
meiotic progeny). Moreover, the sequence length for CBS6273 is
slightly shorter than for CBS6039 at the 39 end of the region we
sequenced for the SXI2 gene. In summary, the SXI1 and SXI2
genes span ,2 kb and define the B MAT locus in C. amylolentus.
Although it is not yet clear whether there are any other sexually
dimorphic regions beyond SXI1 and SXI2 (which could reflect
expansion of the HD locus), our analysis based on PCR assay
showed that the areas flanking the SXI1 and SXI2 genes are
conserved enough between CBS6039 and CBS6273 that primers
designed based on CBS6039 sequence amplify corresponding
regions from CBS6273 (data not shown).
To determine whether the pheromone receptor gene STE3 lies
within the A P/R mating-type locus, we performed Southern blot
analysis using genomic DNA from the two strains of C. amylolentus.
The STE3 PCR product derived from CBS6039 was used as a
probe, and only hybridized to the lanes containing CBS6039 DNA
with no hybridization to CBS6273 (Figure 4B). This analysis
provides evidence that the STE3 gene differs between the two C.
amylolentus strains and the pheromone receptor gene is also linked to
mating-type. Extensive additional Southern and PCR data
(summarized in Figures S2, S3, S4) document that the sequence
divergent region of the P/R locus spans more than 60 kb
encompassing multiple genes (mating pheromone genes, STE3,
STE12, and STE20 among others). This contrasts with C. heveanensis
in which the P/R locus is more restricted, STE3 and the MF
pheromone genes are closely linked, and the LPD1, STE11, ZNF1,
MYO2, and IKS1 genes are linked to but not within MAT . In
conclusion, in C. amylolentus a tetrapolar mating system with
physically unlinked HD and P/R loci appears to define mating-
type identity, and the P/R locus has expanded considerably
compared to C. heveanensis, revealingan evolutionary intermediate in
the transition from the tetrapolar to bipolar state that is even more
closely related to the pathogenic species complex.
Figure 4. Analysis of the pheromone/receptor genes in C. amylolentus. (A) Sequence alignments of the pheromone gene in C. neoformans
var. neoformans JEC21 MFa, C. neoformans var. grubii H99 MFa, C. gattii WM276 MFa, C. neoformans var. neoformans JEC20 MFa, C. neoformans var.
grubii 125.91 MFa, C. gattii E566 MFa, C. heveanensis CBS569 MFa, C. amylolentus CBS6039 MFa1, T. wingfieldii CBS7118 MFa1, and T. mesenterica
ATCC24925 Tremerogen a-13. The black arrow denotes the predicted cleavage site. The pheromone receptor gene, STE3, is MAT specific. (B) Genomic
DNA from the two C. amylolentus strains was digested with BamHI, BglI, ClaI, EcoRI, or NcoI and Southern blot analysis was performed using the STE3
PCR product from CBS6039 as a probe. For each enzyme digestion, CBS6039 was in the left lane and CBS6273 was in the right lane.
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org6 February 2012 | Volume 8 | Issue 2 | e1002528
Phylogenetic analysis of MAT related genes in C.
amylolentus and T. wingfieldii
We conducted phylogenetic analysis of several genes that are
located within the MAT locus of C. neoformans (CID1, ETF1, GEF1,
LPD1, STE3, STE12, STE20, SXI1, and SXI2). This analysis
included C. neoformans var. neoformans, C. neoformans var. grubii, and
C. gattii representatives from the pathogenic species cluster 
and the closely related sibling species C. amylolentus and T.
wingfieldii, as well as the outgroup species C. heveanensis and T.
mesenterica . Based on the phylogeny of the species within the C.
neoformans pathogenic species cluster, these genes can be classified
into three different groups: species specific (CID1, GEF1, LPD1),
mating-type specific (ETF1, STE3, STE12, STE20), and mating-
type unique genes (SXI1, SXI2) (Figure 7 and Figures S6 and S7).
Figure5.Modelfortheevolution ofthemating-typelocus inthepathogenic Cryptococcus species.The physically unlinkedancestraltetrapolar
HD and P/R loci contained both homeodomain genes and the pheromone/receptor genes respectively. Additional genes were acquired into both loci,
expanding the MAT-specific region. A translocation event occurred, likely between chromosomes 4 and 5 of Cryptococcus, resulting in the formation of a
telomeric ends of chromosome 4. The unstable tripolar intermediate later collapsed to a bipolar state. The fused loci were subjected to further gene
rearrangement and gene conversion events,whichled to theformationof the bipolar alleles of the pathogenic Cryptococcus species. White arrowsindicate
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org7 February 2012 | Volume 8 | Issue 2 | e1002528
The species- specific phylogeny of CID1, GEF1, and LPD1 is
consistent with the hypothesis that this region has been recruited into
the MAT locus of C. neoformans during the transition from a tetrapolar
to a bipolar mating system. For the sex-unique genes in the C.
neoformans species complex, SXI1 and SXI2, SXI2 showed a
considerably higher level of polymorphism between the two alleles
from the CBS6039 and CBS6273 C. amylolentus isolates (Figure S7).
ETF1 might have gained its mating-type specific divergence in the C.
neoformans species complex after the common ancestor of the species
complex split from the other sibling species while STE3, STE12, and
STE20 all have mating type specific phylogenetic patterns within the
C. neoformans species complex. In C.amylolentus, PCR primers designed
based on CBS6039 sequences were only able to amplify these genes
from CBS6039, but not from CBS6273, indicating the existence of
CBS6273 for each of these three genes. This is consistent with the
mating type specific pattern observed within the C. neoformans species
complex. Additionally, for STE3 and STE12, the clusters of C.
amylolentus and T. wingfieldii are more closely related to the MATa
alleles of C. neoformans species complex, suggesting a possible common
origin of thesealleles, aswell as an early involvement of the STE3 and
STE12 genes in the evolution of mating type determination.
Discovery of the sexual cycle in C. amylolentus
Following definition of the mating-type locus for both sibling
species, we sought to identify a sexual cycle for C. amylolentus and T.
wingfieldii to determine whether the A, B, or both A and B MAT loci
control sexual reproduction. It was previously thought that both of
these sibling species were asexual ; however, we discovered an
extant heterothallic sexual cycle for C. amylolentus. We conducted
mating assays and found the following optimal conditions: V8
pH=5 solid medium with incubation for one week or longer at
room temperature in the dark. The cross between C. amylolentus
strains CBS6039 and CBS6273 produced hyphae with fused
clamp connections and aseptated basidia terminating in four long
individual spore chains (please see further discussion on strains
CBS6039 and CBS6273 in Text S1, and formal description of
mating in Materials and Methods section), similar to matings in C.
neoformans and C. gattii. Sterigmata were not observed (Figure 8A–
8F). A marked, obvious feature is the shape of the spores which are
ellipsoid in the pathogenic species  whereas C. amylolentus
spores are round and similar in size to yeast cells. Crosses of either
C. amylolentus strain with T. wingfieldii were infertile. Because there
is only one strain of T. wingfieldii available, T. wingfieldii might be
fertile in the presence of a suitable partner, similar to the two
interfertile C. amylolentus strains, or it could be a sterile isolate.
In C. amylolentus, we observed that the periphery of some mating
patchescontains a mixture of both monokaryotichyphae and sectors
in which mating occurs to produce dikaryotic hyphae indicative of
sexual reproduction. The dikaryotic sectoring phenotype is present
in most mating patches and also serves as a visual assay for mating.
visualized in greater detail by microscopy. The four spore chains are
each very long consisting of .15 (quantified by counting 10
individual basidia) spores per chain and clamp cell connections are
visible by light microscopy and SEM (Figure 8A–8F). Based on
fluorescence microscopy with Hoechst 33258 or Sytox green,
dikaryotic hyphae and both uni- and occasional bi-nucleate spores
were observed (Figure S8A–S8D). In the Filobasidiella lineage, C.
neoformans and C. gattii produce both dikaryotic (heterothallic) and
monokaryotic (homothallic) hyphae while F. depauperata produces
only monokaryotic hyphae. The presence of dikaryotic hyphae in C.
amylolentus provides evidence that opposite-sex mating occurs during
the sexual cycle [11,12]. Additionally, the presence of two nuclei in
in the spore or packaging of two nuclei into some spores (as occurs in
pseudo-homothallic species) .
Interestingly, the cap of the spore chain represents a quartet of
basidiospores. These spores are the oldest in the spore chain and
remain tightly attached to each other. Younger spores in the four
Figure 6. The homeodomain genes, SXI1 and SXI2, define MAT. A percent identity plot of both C. amylolentus strains, CBS6039 and CBS6273,
comparing the SXI1 and SXI2 dimorphic region in the HD locus. The red ellipsoid represents an EcoRV site, which only cleaves SXI1 in CBS6039 while
the blue ellipsoid represents an RsaI site, which only cleaves SXI2 in CBS6039.
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org8 February 2012 | Volume 8 | Issue 2 | e1002528
spore chains remain attached to the preceding and following
spores in the chain but often not to their meiotic siblings in the
other three spore chains. Thus, the quartet spore cap appears to
tether the ends of the spore chains together. This feature has not
been described in the pathogenic Cryptococcus species. In summary,
microscopic examination of mating structures in C. amylolentus has
revealed both shared hallmarks with sexual reproduction in the
pathogenic Cryptococcus species and novel features.
Genotypic analysis of meiotic progeny
To determine if recombination occurs, and to further assess
whether the mating system of C. amylolentus is tetrapolar or bipolar,
we performed microdissection of random progeny (F1 set 1) and
individual spore chains (F1 set 2) followed by molecular
genotyping analysis for both MAT markers and a genome-wide
set of RAPD markers. We designate the CBS6039 parent as A1B1
and the CBS6273 parent as A2B2, according to the designation
used for a tetrapolar mating system and our findings, assigning A
as the P/R locus and B as the HD locus (as in T. mesenterica, C.
heveanensis, and U. maydis [24,26]).
For F1 set 1 (F1S1), a total of 40 spores were dissected and 28
(70%) germinated (Tables S1 and S2). The progeny were all
haploid based on FACS analysis with C. neoformans as reference
(data not shown). Genotyping using MAT markers and RAPD
markers revealed that most of the progeny inherited all of the
parental alleles from CBS6039 (A1B1) (Tables S1 and S2) and did
not appear to be meiotic recombinants. Of the 28 progeny, three
(11%) did show recombination within the P/R locus (#17, 27, and
28), whereas only one additional progeny (3.5%) exhibited
reassortment between the P/R and HD loci (#18). We
hypothesize that this is likely due to the dissection of a mixture
of yeast cells, blastospores (mitotic pre-meiotic cells produced by
budding from the hyphae or clamp cells), and basidiospores
(meiotic sexual spores) [5,34], which are all morphologically
similar for this species. Similar to C. neoformans, in C. amylolentus
blastospores can be generated from the clamp cell, and the
following repeated mitotic events tend to produce a cluster of cells
at the hyphal septa. This may explain why we did not observe an
equal distribution of markers from the two parental strains among
the blastospores, as they could have been mitotic products from
one common parental blastospore. That many isolates in F1S1
could be blastospores is also supported by analysis of the
mitochondrial genome segregation (as shown below) that revealed
a majority of this progeny set possess nuclear and mitochondrial
Figure 7. Phylogenetic patterns of four C. amylolentus MAT genes. The phylogenetic relationships of C. amylolentus to the pathogenic
Cryptococcus species and neighboring taxa based on four genes, GEF1, CID1, SXI1, and SXI2, are shown. GEF1 and CID1 display a species-specific
phylogeny and the SXI1 and SXI2 alleles are very diverged from the pathogenic Cryptococcus species. The trees were constructed using the Neighbor-
Joining method implemented in the software MEGA4. Bootstrap values on tree branches were calculated from 500 replicates. (a) indicates strains
with the MATa locus, and (a) indicates strains with the MATa locus.
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org9 February 2012 | Volume 8 | Issue 2 | e1002528
genomes inherited from different parents. Remarkably, 22 (78%)
of the progeny are sterile and unable to undergo sexual
reproduction with either parent or their F1 siblings. It is interesting
that progeny that appear to be derived from blastospores are, for
unknown reasons, frequently sterile.
To analyze meiotic basidiospores specifically, we dissected F1
set 2 (F1S2) from four well-resolved individual spore chains (one
chain each from four different basidia). The germination
frequency was 91% (31/34), and 58% (18/31) of the progeny
were sterile with both parents (Table 1). All of this progeny set
were also haploid based on FACS analysis (data not shown).
Molecular analysis of this set using the same six MAT A or B genes
revealed that 64.5% (20/31) of the progeny resembled one or the
other parent (A1B1 or A2B2) while the other ,35% exhibit
evidence of recombination within the P/R locus and/or between
the HD and P/R loci (i.e. A1B2 or A2B1 progeny) (Table 1). In
contrast to the first F1 progeny set (F1S1), genotyping of the spore
chain derived progeny set (F1S2) using RAPD markers revealed
extensive recombination (Table 2). Linkage analyses clustered
markers analyzed in this study into several linkage groups,
indicating independent inheritance of markers (data not shown).
In addition, analysis of the markers implemented in this study
revealed that for each marker, the two parental alleles were
equally inherited across the entire progeny set (Table 2).
Specifically, for each marker, the percentages of the CBS6039
allele ranged between 35% and 71%, which did not show any
significant bias toward one parental allele (chi-square test,
P.0.05). Similarly, the percentage of the CBS6039 allele that
each progeny inherited ranged from 25% to 80%, and again these
values reflect equivalent inheritance of alleles from either parent
(chi-square test, P.0.05). Moreover, we observed that meiotic
recombination in C. amylolentus resulted in the generation of new
combinations of alleles in the progeny given the multiple
genotypes present in the different spore chains analyzed. The
observed high level of recombination and equivalent inheritance of
the two parental alleles support the conclusion that meiosis occurs
in C. amylolentus.
Mating ability of meiotic progeny
We discovered that some of the progeny that are sterile with
either parent are in fact interfertile with other progeny.
Specifically, of the 59 F1 progeny (mixture of blastospores and
Figure 8. Sexual reproduction of C. amylolentus. Microscopic examination of mating structures produced during sex between the two C.
amylolentus strains, CBS6039 and CBS6273, on V8 (pH=5) medium incubated in the dark at room temperature for 2 weeks. (A) SEM of basidiospores
attached to basidia. Scale bar represents 10 mm. (B) SEM of fused and unfused clamp connections. (C and D) Light microscopy at a magnification of
20X of hyphal filaments, basidia, and basidiospores, scale bar=10 mm. (E) Basidium with youngest spores attached and associated detached spore
chains, scale bar=1 mm. (F) A cluster of basidiospores and basidia. Scale bar=10 mm.
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org 10February 2012 | Volume 8 | Issue 2 | e1002528
basidiospores), there are 14 A1B1 F1 progeny that are fertile with
the A2B2 parent CBS6273, one A2B2 progeny that is fertile with
the A1B1 parent CBS6039, and one that is fertile with both
parents. Among the spore chain derived (F1S2), progeny #13
(A2B2) was found to be able to mate with progeny #3 and #24
(A1B1). Successful mating was also observed between some MAT
recombinant progeny. Specifically, successful mating was observed
when A1B2 progeny (F1S2 #10 and #16) and A2B1 progeny
(F1S1 #18, F2 #1, #2, and #5) were co-cultured together (with
the exception of mating between F1S2 #16 and F1S1 #18). None
of these MAT recombinant progeny mated with either parent,
further confirming that C. amylolentus possesses a tetrapolar mating
system (Figure 9).
Because only 32% (19/59) of the progeny are fertile in both
progeny sets, we assessed whether fertility increases with an
additional sexual cross or mitotic passage. Even after several
passages on YPD, the sterile phenotype remained stable (data not
shown). We crossed F1S2 progeny #3 and CBS6273 to generate a
backcross progeny set (F2). Interestingly, most of the basidia in the
cross were barren and if spore chains were present, the number of
spores per chain was significantly reduced when compared to
matings between the parental strains. We were successful in
dissecting spores from two individual spore chains. The germina-
tion rate was 54% (6/11) and all of the progeny were fertile (50%
with the CBS6039 parent and the remaining A2B1 progeny are
interfertile with the F1S2 progeny #10 and #16 (Table 1)). All of
the progeny examined are haploid with the exception of F2 #4,
which is diploid by FACS yet remains self-sterile (data not shown).
In summary, taken together our genotyping data indicates that
meiotic recombinants are present among the sexually produced
progeny and our evidence is that the sexual cycle of C. amylolentus
conforms to a modified tetrapolar mating system in that 1) sterile
progeny are also frequently produced, and 2) the ratio of the four
mating types is unbalanced.
Uniparental mitochondrial DNA inheritance
To assess the mitochondrial inheritance pattern during sexual
reproduction of C. amylolentus, SNPs were first identified in two
mitochondrial genes, NAD4 and NAD5, between the two parental
strains, CBS6039 and CBS6273, by PCR amplification and
sequencing. Of the 65 progeny screened, no intra- or inter-genic
recombination between the two genes was observed, and all of the
progeny (with the exception of two from F1S1) typed as the
CBS6273 (A2B2) parent (Table S3). The two progeny (F1S1 #13
and #16) that contain the A1B1 mitochondrial genome are likely
dissectedparentalyeast cells,because they alsoboth possessed A1B1
alleles at all of the other markers that were typed. For the other
nuclear non-recombinantprogeny that type as theA1B1 parent, the
fact that they have the A1B1 nuclear genome and the A2B2
mitochondrial genome suggests that they descend from blastospores
produced after cell-cell fusion and a result from cytoduction of the
CBS6039 nuclear genome and CBS6273 mitochondrial genome.
All other progeny that are derived from meiotic basidiospores
contained a recombinant nuclear genome paired with the
(CBS6273). These results demonstrate that mitochondria are
uniparentally inherited from the A2B2 parent during C. amylolentus
sexual reproduction, similar to C. neoformans in which mtDNA is
inherited uniparentally from the a parent [35,36,37,38].
fromthe A2B2 parent
The current study extends the previous analyses of the MAT
locus in the pathogenic Cryptococcus species to the closest known
Table 1. Summary of mating abilities and genotypes at the
MAT genes of F1 set 2 and F2 progeny.
SXI2 SXI1 RPL39 GEF1 ETF1STE3
Basidium A F1S2-1 sterilebbbbbb
F1S2-4 A1B1+ +
Basidium B F1S2-8A1B1aaaaaa
Basidium D F1S2-25sterileaabbbb
Basidium FF2-3 A2B2bbbbbb
Bold: fertile with parents;
Bold and Italics: fertile with siblings from F1 set 1 and F2 progeny;
Underlined genotypes indicate intra-MAT (A or B locus) recombinant progeny;
1: ‘‘a’’ represents allele from A1B1 parent CBS6039; ‘‘b’’ represents allele from
A2B2 parent CBS6273.
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org11February 2012 | Volume 8 | Issue 2 | e1002528
Table 2. Summary of genotypes of F1 set 2 and F2 progeny using RAPD markers.
Pi_Random_5Pi_Random_8Pi_Random_9 Pi_Random_15Pi_Random_20Pi_Random_21 Pi_Random_24_No.1 Pi_Random_24_No.2
JOHE22492JOHE22621 JOHE22631 JOHE22643JOHE22655_No.1 JOHE22655_No.2JOHE22655_No.3JOHE22656_No.1 JOHE22656_No.2JOHE22660
% of allels from CBS60393
Number of genotypes5
% of alleles from
61 58 3838 5255 58 35 356858 48 45 4558 61527139 61
1‘‘a’’ represents CBS6039 allele;
2‘‘b’’ represents CBS6273 allele;
3Percentage of CBS6039 alleles within each progeny at the markers analyzed;
4Genotypic category based on the markers analyzed in this study for each progeny from the same basidium;
5Number of unique genotypes among the progeny from the same basidium;
6Percentage of CBS6039 alleles among the F1S2 progeny at each marker.
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org 12February 2012 | Volume 8 | Issue 2 | e1002528
species, T. wingfieldii and C. amylolentus. To determine the structure
of MAT in both species, we cloned and sequenced the HD and P/
R loci. Due to their close phylogenetic relatedness ,
characterization of MAT has provided key insights into the
evolution of MAT and revealed important aspects of the transition
from an ancestral tetrapolar to a bipolar mating system in C.
neoformans and C. gattii [13,14,39,40].
A previous phylogenetic analysis using a six-gene multi-locus
sequencing (MLS) approach identified the most closely related
species to the pathogenic Cryptococcus species complex . This
analysis identified the sensu stricto (closely related) and sensu lato
(more distantly related) species that provide unique vantage points
to address questions such as: when and how did the bipolar mating
system evolve? And what, if any, affects does the emergence of
bipolar mating systems have on the pathogenesis of C. neoformans
and C. gattii? Previous studies on a more distantly related sensu lato
species, Cryptococcus heveanensis, revealed it to be tetrapolar .
The key advances presented here provide additional critical
insights. First, as sensu stricto strains, C. amylolentus and T. wingfieldii
are much more closely related to the pathogenic species C.
neoformans/C. gattii than is C. heveanensis; hence the transition to
bipolarity in the pathogens was even more recent than could be
concluded based on the studies of C. heveanensis alone. Second, by
providing additional tetrapolar outgroup species, we can conclude
that the transition was from tetrapolar to bipolar, not vice versa.
Third, the P/R locus is much more expanded in C. amylolentus
compared to C. heveanensis, providing further insights on the
evolution of the MAT and this key step in the process.
Furthermore, the tetrapolar mating system in C. amylolentus showed
indications of deviation from the classic tetrapolar model in that
many MAT loci recombinant progeny are sterile and progeny that
resemble one parent at the MAT loci dominate the progeny
population. Moreover, the organization of MAT in these sibling
species mirrors key aspects (gene acquisitions, chromosomal
Figure 9. C. amylolentus has a tetrapolar mating system. (A) In a bipolar mating system, haploid a and a cells fuse to form a diploid a/a cell. Sex
culminates in meiosis, which gives rise to four meiotic progeny, 2 a and 2 a. The a progeny can mate with the a parent (50%) while the a progeny can
mate with the a parent (50%). In a tetrapolar mating system, haploid A1B1 and A2B2 cells fuse to form a dikaryon/diploid A1B1/A2B2. Meiosis then
results in the production of four haploid meiotic progeny: A1B1 can mate with the A2B2 parent and progeny (25%), A2B2 can mate with the A1B1
parent and progeny (25%), and A1B2 and A2B1 are recombinants (50%) that are sterile with either parent but interfertile with one another. (B) An
example of a RAPD and genotyping marker analysis on four progeny and the two parental strains that represent the different gentoypes in a
tetrapolar mating system (1=F1S2 #3 (A1B1), 2=F1S2 #13 (A2B2), 3=F2 #1 (A2B1), 4=F1S2 #10 (A1B2), 5=CBS6039 (A1B1), and 6=CBS6273
(A2B2)). (C) Results of mating assays of all possible combinations among the four mating types. Mating was performed by mixing strains on V8 plate
(pH=5). (‘‘2’’ indicates lack of sexual reproduction and ‘‘+’’ indicates sexual reproduction occurs). (D) Microscopic images of hyphae and spore chains
generated during C. amylolentus mating assays described in Figure 9C (the mating-type of each strain is indicated in parenthesis). Dikaryotic hyphae
and spore chains were produced in matings between CBS6039 (A1B1) and CBS6273 (A2B2) and between F1 set2 #10 (A1B2) and F2 #1 (A2B1).
Monokaryotic hyphae were produced in all of the other mating combinations, including individual strains grown in the absence of a mating partner.
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org13 February 2012 | Volume 8 | Issue 2 | e1002528
rearrangements, etc.), which shaped the evolution of the mating-
type locus in the pathogenic Cryptococcus species complex.
Previous analysis resolved the phylogeny surrounding the
pathogenic Cryptococcus species cluster and revealed that T. wingfieldii
and C. amylolentus are sibling species, the closest relatives of the
pathogenic species, and members of the Filobasidiella clade . The
MAT loci of T. wingfieldii and C. amylolentus share overall synteny,
with two major inversion events present between the P/R loci of the
two species (Figure 3). For this analysis, the type strain, the only
isolate of T. wingfieldii available, was employed. Two strains of C.
amylolentus are available and we characterized MAT for the type
strain CBS6039 and representative sequences for CBS6273. The
two MAT loci of T. wingfieldii and C. amylolentus are physically
unlinked and present on different chromosomes (Figure 1B and
Figure 2B). The MAT assembly for C. amylolentus is similar to T.
wingfieldii in that both homeodomain transcription factors are
present and opposite in their orientations, similar to the paired,
divergentlyorientedbE and bWgenesinU.maydis. Severalotherkey
genes (SPO14, RPL22, and CAP1) are present and these lie within
MAT in C. neoformans but appear to lie outside of MAT in C.
amylolentus. The configuration of the HD genes in the sibling species
provides evidence that the ancestralform of the HD locus contained
both SXI1 and SXI2, similar to tetrapolar mating systems in other
basidiomycetes, and that loss of one or the other of the HD genes
punctuated the formation of a bipolar mating system.
Of the .20 genes identified in the HD (B) and P/R (A) loci of the
sibling species, we determined which genes define MAT. Because
only one strain of T. wingfieldii is available, we were unable to
By comparing sequences from the two strains of C. amylolentus, we
determined that the MAT-specific region in the HD locus is likely
restricted to the ,3 kb SXI1 and SXI2 dimorphic region. The
divergence is present in the 59 regions of SXI1 and SXI2, similar to
recent findings on the B MAT locus alleles of C. heveanensis . This
is also consistent with findings in other fungi where the N-terminal
regions of the homeodomain proteins are typically variable and
heterodimerization only occurs when compatible (or different allelic
versions) of the proteins are brought together promoting activation
of genes required for sexual development [30,41].
We also sought to define the extent of the sex-specific region in
the MAT A locus. Our extensive Southern and PCR analysis
document that the P/R locus has been expanded to encompass
.60 kb in C. amylolentus, including the STE3 and MF pheromone
genes that lie .30 kb apart in contrast to their close linkage in the
P/R MAT A locus of C. heveanensis (Figures S2, S3, S4). In addition,
several genes encompassed within this expanded C. amylolentus P/R
locus are linked to but outside the defined P/R locus of C.
heveanensis . Thus, one of the two MAT loci has expanded in C.
amylolentus but the two remain unfused.
We also report the discovery of sexual reproduction in C.
amylolentus. Fortunately, the only two strains of C. amylolentus
available in the world are of opposite mating-type and fertile,
enabling us to define the sexual cycle for C. amylolentus. Mating
structures in C. amylolentus resemble those observed in C. neoformans,
and differed from C. heveanensis, consistent with its closer
phylogenetic relationship with C. neoformans than with C. heveanensis.
Mating in C. amylolentus produces many sterile progeny,
suggesting that sexual reproduction may pose a risk in which not
all of the progeny produced are fertile. Although the underlying
mechanism(s) causing sterility in the C. amylolentus progeny is not
clear, there are several possible explanations. First, it is possible
that aneuploids (1N+1) are generated during meiosis that could be
sterile. FACS analysis of the examined progeny suggested that all
of the progeny are haploid with the exception of a single diploid
(F2 progeny #4), but FACS is not sensitive enough to detect 1N+1
(CGH) of the C. amylolentus parental strains with the sterile
progeny will be necessary to address the issue of possible
aneuploidy generated during mating. Second, meiosis is mutagenic
and sexual reproduction may also increase transposition in the
genome. The resulted mutations and/or the insertion of
transposons in MAT or elsewhere might result in sterility. Third,
the increased sterility among progeny could be due to sex induced
silencing of repetitive elements within MAT and linked fertility
genes  or damage to MAT caused by gene conversion events.
Sex induced silencing requires the RNAi machinery. However it is
not known yet if C. amylolentus possesses these genes. The C.
amylolentus genome sequence will allow this question to be
answered. Additionally, we cannot exclude the possibility that
the sterility observed among the progeny is due to divergence/
incompatibility between the mating machineries of the two C.
amylolentus strains, or to nuclear-mitochondrial incompatibility that
has been observed in other yeasts .
From the genotyping analysis, it is evident that extensive
recombination occurred among the progeny produced by sexual
reproduction. Additionally, we observed a 1:1 segregation pattern
of the two parental alleles in the progeny population. This
segregation data and the high level of genetic exchange in the
progeny (especially the F1S2 spore chain derived progeny set)
provide strong evidence that meiosis occurs within the basidium
during sexual reproduction. Additionally, RAPD analysis revealed
that in some spore chains from the F1S2 and the F2 progeny sets,
more than four genotypes are present in a single chain. There are
several possible explanations. In C. neoformans, meiosis typically
gives rise to four meiotic products and it was recently shown that a
single meiotic event occurs in each basidium . In C. amylolentus,
more than one meiotic event could occur in the basidium.
However, this would have to involve post meiotic nuclear fusion
and a second round of meiosis. In this case, up to eight genotypes
could be produced from one basidium. Also, high gene conversion
events favoring some alleles over others could result in a non-
Mendelian inheritance pattern and skew the resulting genotypes in
each individual spore chain.
Another possible explanation for the observed .4 genotypes/
basidium that we favor is the presence of aneuploids in the progeny
population. The RAPD markers employed in this analysis
differentiate the two parental strains by the presence or absence
of a PCR product. If progeny are aneuploid for one or more
chromosome, they could appear unique and differ from the two
parental strains. In this aneuploidy model we expect the
basidiospores from one basidium to share four common genotypes
with the exception of a few rarer genotypes (potential aneuploids).
This model is consistent with our RAPD data (Table 2 and Table
S2) in which several spore chains contain four distinct major
genotypes and several anomalous minority genotypes that are
closely related to one of the four consensus majority genotypes in a
givensporechain(Figure S9).One limitation isthat wearecurrently
unable to score the heterozygous state of the aneuploids, which can
and this provides fertile ground for future studies.
Micromanipulation of the individual spore chains representing
F1S2 and the F2 progeny generated progeny that are recombinant
at the MAT loci (A1B2 and A2B1), and these MAT recombinant
progeny are inter-fertile, but cannot mate with either of the two
parental strains, proving that C. amylolentus has a tetrapolar mating
system. However, among those isolates that were fertile the A1B1
genotype was overrepresented, whereas the other three genotypes
were underrepresented (Table 1 and Table S1).
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org14 February 2012 | Volume 8 | Issue 2 | e1002528
MAT in the pathogenic Cryptococcus species evolved from an Download full-text
ancestral tetrapolar system with physically unlinked B and A loci
and these loci fused into a large bipolar MAT locus. In C.
amylolentus, the structure of MAT indicates a tetrapolar mating
system with allelic diversity in both the B and A loci. Although
evidence from T. wingfieldii, C. amylolentus, C. heveanensis , and C.
disseminatus  suggests that MAT evolved from a tetrapolar to a
bipolar system, an alternative hypothesis could be just the
opposite: namely that the ancestral form of MAT was bipolar
and instead evolved into a tetrapolar mating system in these
species. In such a scenario, a bipolar locus could have suffered a
chromosomal break resulting in the formation of physically
unlinked HD and P/R loci in a derived rather than ancestral
tetrapolar fungal species. In this model, the tetrapolar state would
then be ancestral in some species and derived in others. We do not
favor this alternative model and the one we propose (Figure 5)
instead illustrates the evolution of the bipolar MAT in the
pathogenic Cryptococcus species from an ancestral tetrapolar system.
The evidence adduced now for three sibling species supports the
more parsimonious model that C. amylolentus, T. wingfieldii, and C.
heveanensis all reflect a common, shared ancestral tetrapolar state
rather than multiple independent derived states.
MAT evolution in fungi has defined a continuum of transitions in
modes of sexual reproduction from outcrossing tetrapolar multi-
allelicsystemsto bipolar biallelicsystemsthat promote inbreeding to
unipolar uniallelic same-sex mating that promotes extreme
inbreeding and clonality [15,45]. Aside from bipolar and tetrapolar
mating systems, some deviations from these classic mating systems
have been reported recently. For example, a pseudo-bipolar mating
system has been recently found in the red yeast Sporidiobolus
salmonicolor [46,47]. The authors found that in this species, mating is
normally bipolar and governed by a large continuous MAT locus
with the A and B regions located at either end. However, meiotic
recombination may occur between the MAT locus alleles,
generating novel mating types, and thus increasing MAT allele
number and evolutionary rates for some MAT genes.
Results from our studies illustrate features of both the transition
from tetrapolarity to bipolarity in the closely aligned saprobic and
pathogenic Cryptococcus species, and also the emergence of sexual
reproduction in which one mating-type has an advantage resulting
in a higher proportion of fertile progeny of one mating-type
(A1B1) that might have ultimately led to the emergence of
unisexual same-sex mating in C. neoformans. In conclusion, C.
amylolentus and C. heveanensis have physically unlinked HD and P/R
loci and this arrangement further supports tetrapolarity as the
ancestral configuration, and that the transition to bipolarity
occurred recently and concomitantly with the emergence of the
pathogenic C. neoformans/C. gattii species cluster.
These studies on the molecular events leading to the fusion of
two unlinked sex determining regions of the genome in the
ancestral tetrapolar state to the derived bipolar mating systems
mirror aspects in the hypothesized origin of sex chromosomes of
more complex multicellular eukaryotes, including plants, insects,
fish, and mammals. Namely, Ohno hypothesized that sex
determinants arise on an autosome, and then gradually capture
this chromosome, which evolves to become a sex chromosome
. These steps include the original emergence of the sex
determinant, the recruitment of other genes that function in sex to
the incipient sex chromosome, and rearrangements and the
acquisition of repetitive elements that lead to two sexually
dimorphic chromosomes. The transition from two unlinked sex
determinants in tetrapolar fungi to two linked sex determinants in
bipolar fungi, and the fact that this transition has occurred
repeatedly and independently, provides further support for the
hypothesis that sex determinants arise at distant genomic locations
and then become linked through gene movement or chromosomal
translocations in both mating type loci and sex chromosomes.
Materials and Methods
Strains and media
ThetwostrainsofC. amylolentus, CBS6039 and CBS6273,and the
one T. wingfieldii isolate, CBS7118, were obtained from the
Centraalbureau voor Schimmelcultures (CBS) Fungal Biodiversity
Centre in the Netherlands. Both CBS6039 and CBS6273 were
originally isolated from insect frass in South Africa, while CBS7118
was originally isolated from rubber sheet in Indonesia. All species
were grown and maintained on yeast extract-peptone-dextrose
(YPD) medium at 24uC. Mating assays were performed on V8
medium pH=5 in the dark and also at 24uC. Random spore
dissection was performed on YPD medium as previously described
. Spore chain dissection was performed by first transferring a
well separated spore chain onto a drop of zymolyase on YPD, and
after incubation at 24uC for 15 minutes, individual spore from the
spore chain was dissected as previously described .
To isolate genomic DNA from T. wingfieldii and C. amylolentus,
cells were cultured in 50 ml of liquid YPD shaking overnight at
24uC. The pellets were then lyophilized overnight and the CTAB
method of fungal DNA isolation was performed as described
before . Plasmid DNA from positive TOPO clones was
extracted using the QIAprep Spin Miniprep Kit (Qiagen,
Valencia, CA), fosmid DNA was isolated using a modified
miniprep protocol, and DNA from the shot-gun sequencing
libraries was extracted using the DirectPrep96 Miniprep Kit
(Qiagen, Valencia, CA). Additionally, progeny DNA was isolated
using a modified miniprep protocol and colony lifts were
performed to isolate DNA from individual colonies in each fosmid
library according to the protocol described in .
We designed degenerate PCR primers using the online computer
program, COnsensus-DEgenerate Hybrid Oligonucleotide Primer
(CODEHOP, http://blocks.fhcrc.org/codehop.html) to identify
MAT specific genes in T. wingfieldii. The primers consist of a
relatively short 39 degenerate core and a longer 59 non-degenerate
consensus clamp designed by multiple sequence alignments .
We aligned sequences for two flanking genes (FAO1 and NOG2) and
two recently acquired MAT genes (RPO41 and LPD1) from C.
neoformans var. neoformans and var. grubii, C. gattii, U. maydis, and C.
cinerea to design the degenerate PCR primers (see Table S4 for
primer information). PCR was performed on genomic DNA
isolated by the CTAB extraction method as template and products
were separated by gel electrophoresis. Products with the strongest
ethidium bromide-staining signal were then gel extracted using the
QIAquick Gel Extraction Kit (Qiagen, Valencia, CA) followed by
transformation into E. coli using the TOPO-TA cloning Kit
(Invitrogen, Carlsbad, CA). Plasmid DNA was purified from
transformants and then sequenced. For C. amylolentus, degenerate
primers were not used. Instead, primers from T. wingfieldii were
directly used to amplify MYO2, LPD1, SXI1, and SXI2 in both C.
amylolentus strains (see Table S4 for primer information).
Fosmid library preparation and fosmid library screening
We employed the CopyControl Fosmid Library Production Kit
(Epicentre, Madison, WI) to generate fosmid libraries for T.
wingfieldii and C. amylolentus strain CBS6039. At least 2.5 mg of
Sexual Cycle and MAT of Cryptococcus amylolentus
PLoS Genetics | www.plosgenetics.org 15 February 2012 | Volume 8 | Issue 2 | e1002528