OUP user menu

Do major species concepts support one, two or more species within Cryptococcus neoformans?

Kyung J. Kwon-Chung, Ashok Varma
DOI: http://dx.doi.org/10.1111/j.1567-1364.2006.00088.x 574-587 First published online: 1 June 2006

Abstract

Cryptococcus neoformans, the agent of cryptococcosis, had been considered a homogeneous species until 1949 when the existence of four serotypes was revealed based on the antigenic properties of its polysaccharide capsule. Such heterogeneity of the species, however, remained obscure until the two morphologically distinct teleomorphs of C. neoformans were discovered during the mid 1970s. The teleomorph Filobasidiella neoformans was found to be produced by strains of serotype A and D, and Filobasidiella bacillispora was found to be produced by strains of serotype B and C. Ensuing studies revealed numerous differences between the anamorphs of the two Filobasidiella species with regard to their ecology, epidemiology, pathobiology, biochemistry and genetics. At present, the etiologic agent of cryptococcosis is classified into two species, C. neoformans (serotypes A and D) and Cryptococcus gattii (serotypes B and C). Intraspecific genetic diversity has also been revealed as more genotyping methods have been applied for each serotype. As a result, the number of scientifically valid species within C. neoformans has become a controversial issue because of the differing opinions among taxonomists as to the appropriate definition of a species. There are three major species concepts that govern classification of organisms: phenetic (morphologic, phenotypic), biologic (interbreeding) and cladistic (evolutionary, phylogenetic). Classification of the two C. neoformans species has been based on the phenetic as well as the biologic species concept, which is also supported by the cladistic species concept. In this paper, we review and attest to the validity of the current two-species system in light of the three major species concepts.

Keywords
  • Cryptococcus neoformans
  • Cryptococcus gattii
  • species concept

Introduction

The discovery of heterothallism in the pathogenic yeast Cryptococcus neoformans has led to the recognition of profound genetic diversity within the species. In 1975, the first teleomorph of C. neoformans described as Filobasidiella neoformans was found to be produced by matings between two compatible strains of serotype D as well as between compatible strains of serotypes A and D (Kwon-Chung, 1975, 1976). The second species of Filobasidiella, Filobasidiella bacillispora, was found to be produced by matings between compatible strains of serotypes B and C (Kwon-Chung, 1976). The correlation between serotypes and the two different teleomorphs indicated that the differences between these two sero-groups were not limited to their antigenic properties. Subsequent studies have shown that the two species are distinct in their ecological niche (Kwon-Chung & Bennett, 1978; Ellis & Pfeiffer, 1990), epidemiology (Kwon-Chung & Bennett, 1984), biochemistry (Bennett, 1978; Polacheck & Kwon-Chung, 1980, 1986; Min & Kwon-Chung, 1986; Kwon-Chung, 1987) and genomic arrangement (Wickes, 1994; http://www.bcgsc.ca/gc/cryptococcus/). Based on biochemical differences, the anamorph of F. bacillispora was described as the distinct species Cryptococcus bacillisporus (Kwon-Chung, 1978). The species epithet bacillisporus was chosen in accordance with the species epithet of the teleomorph. Subsequently, the canavanine–glycine–bromthymol blue (CGB) agar was formulated as a single-step diagnostic tool to distinguish between C. neoformans and C. bacillisporus (Kwon-Chung, 1982). d-Proline agar medium was also found to be a reliable diagnostic tool for differentiating between the two species (Dufait, 1987). Kwon-Chung (2002) proposed the use of the species epithet gattii over the epithets with priority, C. bacillisporus or C. hondurianus (syn. Torulopsis var. hominis var. hondurianus and Cryptococcus hominis var. hondurianus) (Castellani, 1933). Although the name gattii had widely been used over the past 20 years, the epithet bacillispora had seldom been used and the name hondurianus had never been used since it was first proposed.

The present paper focuses on the validity of the two-species system for the agent of cryptococcosis based on the major species concepts used in the classification of the organism.

Species concepts

Defining a biologic species is an important but difficult subject because of the differing opinions as to the definition of a species. Although Darwin's work on the origin of species has been most influential in modern biology, he focused on the mechanisms of how new species evolved from existing species without offering a criterion that defined the species boundary (Darwin, 1875). How can one species be distinguished from another within the same genus? Kottelat suggested a pragmatic definition of a species: When two or more groups show different sets of shared characters (genetic, morphologic and physiologic) and the shared characters for each group allow all the members of that group to be easily and consistently distinguished from members of another group, then they are considered a different species (Kottelat, 1955).

There are now several competing species concepts, among which the most widely accepted are: phenetic (morphologic, phenotypic), biologic (interbreeding) and cladistic (phylogenetic, evolutionary) (Mayden, 1997). Each species concept has strengths and weaknesses. For example, the phenetic species concept defines a species as the smallest group that is consistently and persistently distinct and can be distinguished morphologically (phenotypically) by ordinary means (Cronquist, 1978). This system allows one to recognize differences between species easily. It works well for sexual as well as asexual organisms and can even be applied to extinct species. The weakness of this concept, however, is that morphology-based concepts are not necessarily reflective of genetics. The biologic concept defines the species as a group of actually or potentially interbreeding populations that are reproductively isolated from other such groups (Mayr & Ashlock, 1991). This concept fits well within population genetics, offers an unambiguous empirical criterion and provides an important conceptual framework for speciation. However, this system is applicable only to the sexual organisms existing at the present time and has no evolutionary dimension. The cladistic concept defines species as the smallest diagnosable cluster of individuals that are interconnected by genetic relationships (Cracraft, 1983). This concept offers a clear evolutionary dimension, is the richest concept in paleontological studies and its branch points make use of DNA sequences, the most reliable characteristic. The present cladistic system, however, is based only on a fraction of lineages that have been uncovered as a result of the details required using this approach. It is highly subjective, depending on the DNA sequence used for analysis of the genetic relationships and is disconnected from population genetics.

Which of the species concepts is therefore the most appropriate for classification of the agent that causes cryptococcosis? This subject is controversial among taxonomists, as evidenced by the number of species suggested, which ranges from two to eight depending on the species concept adopted.

Is the classification of C. neoformans into two species supported by the phenetic species concept?

When strains of C. neoformans were classified as representing two species, C. neoformans and C. gattii (C. bacillisporus), it was based on a combination of the morphological differences between their teleomorphs, F. neoformans and F. bacillispora, respectively, and the biochemical differences between the anamorphs of the two species. The differences in size and shape of the basidiospores produced by F. neoformans and F. bacillispora are unambiguous, as shown in Fig. 1. Although their basidial structures are clearly different, the morphology of their anamorphs does not offer any markedly distinguishable traits. There are, however, some morphological features that are specifically associated with each of the two species. Colonies of C. gattii grown on conventional mycological agar media are almost always more mucoid and sticky in texture than those of C. neoformans, although this does not always reflect the size of their capsules. The brown pigment produced by colonies of C. gattii, grown on birdseed agar, is less intense than that of C. neoformans. Additionally, a green hue is present around colonies of C. gattii grown on birdseed agar whereas it is absent around colonies of C. neoformans (Kwon-Chung, 1978). Yeast cells of C. gattii are more often reported to be elliptical or tear-shaped, whereas such cell shapes are infrequently observed in C. neoformans (Kwon-Chung, 1982). The two species, however, can be more reliably distinguished by their growth phenotype on certain media formulations based on their biochemical differences. Strains of C. neoformans do not react to CGB agar whereas strains of C. gattii react positively; this thus represents a one-step diagnostic medium for the separation of the two species (Fig. 2). CGB agar was formulated based on differences between C. neoformans and C. gattii in their ability to utilize glycine as the sole source of nitrogen and carbon as well as differences in their susceptibility to l-canavanine, an arginine analog (Min & Kwon-Chung, 1986; Polacheck & Kwon-Chung, 1986). The positive reaction produced by strains of C. gattii on CGB and the negative reaction by strains of C. neoformans suggested a clear difference in their nitrogen metabolisms (Polacheck & Kwon-Chung, 1986). The biochemical difference in the assimilation of nitrogen between the two species was supported by their creatinine deaminase activites as well as their ability to assimilate d-proline. In C. gattii, creatinine deiminase degraded creatinine in the presence of ammonium while the same enzyme in C. neoformans was found to be inhibited (feedback inhibition) by ammonium in the medium (Polacheck & Kwon-Chung, 1980; Min & Kwon-Chung, 1986). Cryptococcusgattii can utilize d-proline as the sole source of nitrogen whereas C. neoformans cannot (Dufait, 1987). Other differential phenotypes included the capsular antigenic properties that could be distinguished by the commercially available serotyping kit (Crypto Kit, Japan). At the time of writing, the Iatron Crypto Kit is no longer commercially available, but it is hoped that a similar serotyping kit will become readily available. One may, however, encounter a few untypable strains that could still be classified as either C. neoformans or C. gattii based on their CGB reaction. The phenotypic differences can be ascertained readily and consistently, thereby justifying the separation of C. neoformans from C. gattii

Figure 1

Phenotypic differences between teleomorphs of Cryptococcus neoformans (Filobasidiella neoformans) and Cryptococcus gattii (Filobasidiella bacillispora) (bars = 1 μm).

Figure 2

Canavanine-glycine-bromthymol blue (CGB) agar reaction produced by (a) Cryptococcus neoformans and (b) Cryptococcus gattii within 48 h, at 30°C.

Is the classification of C. neoformans into two species supported by the biologic species concept?

Heterothallism in C. neoformans was first discovered in strains of serotype D. Single basidiospore cultures were obtained by micromanipulation of the spores produced by matings between a clinical strain, NIH12 (MATα), and an environmental strain, NIH433 (MATa). Genetic recombination was confirmed by analysing the progeny of a cross between strains B-3501(MATα) and B-3502 (MATa), the F1 generation of NIH12 × NIH433. Colonies of strain B-3501 grown on malt extract agar were milky (m type) whereas those of strain B-3502 were translucent (t type). Ninety-nine progeny isolates exhibited independent segregation of the two markers: 27 (α, m); 25 (a, m); 21 (α, t); 26 (a, t). This indicated that the progeny were products of meiotic cell division (Kwon-Chung, 1976). Segregation analysis of other phenotypic markers, such as CAP+ (encapsulated) and CAP– (acapsular), as well as Mel+ (melanin formation on birdseed agar) and Mel–, further confirmed the meiotic event (Kwon-Chung & Rhodes, 1986). Strains of serotype A were also found to mate with serotype D strains and produce viable basidiospores, although genetic characterization of the progeny was not reported. As the DNA sequence of genes such as URA5 (Edman & Kwon-Chung, 1990), LAC1 (Williamson, 1994), CAP59 (Chang & Kwon-Chung, 1994) as well as those comprising different subunits of genomic or mitochondrial rRNA genes (Fell, 2000; Xu, 2000) became available, they were used for phylogenetic analysis. Although the extent of divergence in the DNA sequences varied, depending on the gene used, most strains of serotype D clustered apart from the group containing a majority of the serotype A strains analysed (Franzot, 1999; Nakamura, 2000; Xu, 2000). Based on the observed DNA sequence diversity, a new variety, C. neoformans var. grubii, was proposed to accommodate strains of serotype A (Franzot, 1999). Some investigators suggest that the genetic differences between strains of serotypes D and A warrant the elevation of C. neoformans var. grubii from its current varietal status to the rank of a new species. Such proposals can be scrutinized under the biologic species concept. There has been genetic characterization of serotype AD strains that react with anti-A as well as anti-D factor sera. These strains were determined to be diploids or aneuploids, containing serotype A- as well as serotype D-specific sequences (Cogliati, 2001; Lengeler, 2001). Analysis of the mating type-specific genes, such as STE12 or STE20, in the AD serotype diploid strains indicated that a majority of them were MATa/α type with a few strains homozygous for the MAT allele (Cogliati, 2001). The ploidy and genotypes of these strains suggested that a majority of them were products of mating between MATa and MATα strains of serotype A and D and a few of them were fusion products between the two strains of the same mating type of serotype A and D.

In order to determine whether the genetic diversity between serotype A and D strains hinders genetic recombination during meiosis, we crossed H99 with JEC20 (B-4476) on V-8 juice agar and isolated 20 single basidiospore cultures (F1 progeny) and analysed their genotype. The genes used for the analysis of their genetic profiles were: STE12, STE20 and MF on chromosome 4, URA5 and LAC1 on chromosome 7 (Edman & Kwon-Chung, 1990; Williamson, 1994), and CAP59 on chromosome 1 (Chang & Kwon-Chung, 1994) in JEC20. Sequence-specific primers of genes STE12, STE20, MF, CAP59 and LAC1 from strains H99 and JEC20 (Chang, 2000, 2001; Lengeler, 2001; Petter, 2001) were used to amplify PCR products from the progeny, which were then compared with those from the parents. The URA5 genotype of this progeny was determined by Southern blot analysis of StuI restriction patterns of their genomic DNAs. Consistent with the restriction site Stu1 being present in the middle of the URA5 gene in serotype D strains, the Southern blot showed a pattern of two barely resolvable DNA fragments. By contrast, the serotype A strains showed two widely separated bands, indicative of the StuI restriction site polymorphism within the URA5 gene (Kwon-Chung, 1992) (Fig. 3). Serotype of the progeny was determined using the Iatron Crypto Kit. The progeny could be categorized into four groups: MATa parental type, aneuploid, diploid (hybrids) and haploid recombinants (Table 1). The recovery of serotype AD diploid strains by crossing strains of serotype A and serotype D in the laboratory had also been reported previously (Tanaka, 1999). Figure 3 illustrates the genotypes of representative strains of each group. The ploidy based on DNA content of the progeny was confirmed by FACS analysis (Tanaka, 1996) (Fig. 3). The number of diploids and aneuploids in the progeny was significantly higher than that of the haploids, comprising 70% of the progeny. The ploidy of several of the progeny strains as determined by FACS analysis did not correspond to that based on genetic analysis. For example, the genotype of sb-16 was the same as that of JEC20 but the FACS analysis indicated it to be an aneuploid. This is not surprising as the chromosomal markers used were only on three of the 14 chromosomes (Loftus, 2005). Haploid recombinants carrying genes from both parents comprised 15% (sb-4, sb-18 and sb-19) of the total progeny. These types of recombinant populations may exist in nature, although currently available typing tools may not be adequate for the detection of such strains. In addition, such approaches may necessitate larger numbers of strains than thus far studied. A high frequency of aneuploids was expected in light of differences in the genomic organization between H99 and JEC20 (Schein, 2002). At the chromosomal level, for example, the URA5 and LAC1 genes are both on chromosome 7 in JEC20 and JEC21 (http://www.tigr.org/tdb/e2k1/cna1/index.shtml) whereas they are on chromosome 8 in H99 (J. Kronstad, personal communication). Meiotic nondisjunction appeared to have occurred with considerable frequency. The occurrence of stable diploids is puzzling as laboratory diploids constructed by fusing two strains readily segregate (Whelan & Kwon-Chung, 1986). There have been several reports that support these conclusions. First, genetic analysis of numerous strains of serotype AD MATa/MATα from the environment as well as from clinical specimens indicated that they were the products of mating between strains of serotypes A and D (Tanaka, 1999; Cogliati, 2001; Lengeler, 2001). PCR fingerprinting revealed that the genotypes of 17 of the 68 (25%) serotype D clinical isolates from Italy were typed as either VIN3 or VIN4. These were presumed to be hybrids of VIN1 (serotype D) and VIN6 (serotype A) and all were diploids (Cogliati, 2001). Similarly, four of 43 (9.3%) serotype A strains were hybrid diploids (Cogliati, 2001). In a more comprehensive survey by the European Confederation of Medical Mycology in 2002, the percentage of hybrid clinical strains from Italy alone reached 31% (Fimua Cryptococcosis Network, 2002). By analysis of strains using the amplified fragment length polymorphism (AFLP) method, Boekhout et al. also found that 9.4% of serotype A clinical isolates were hybrids while six of 14 (60%) serotype D clinical isolates expressed the hybrid genotype (Boekhout, 2001). This may be due to instability of serotype in diploid or aneuploid strains. Strains sb2 and sb3 (Table 1) are clear cases of diploid progeny based on genetic as well as FACS analysis but serotyped initially as AD and then typed as D in subsequent testing a year later. Conversely, sb14 is an aneuploid strain that was serotyped first as D but serotyped as A in subsequent testing. Furthermore, some diploid and aneuploid strains showed a significantly stronger reaction with factor 8 (serotype D-specific) than with factor 7 (serotype A-specific) as if the expression of serotype D epitope was dominant (e.g. sb7, sb20). These inconsistencies in serotyping could also result from differences in the degree of specificity and potency among the different batches of factor sera. Regardless, our results of genetic analysis and serotyping of the progeny from the cross between H99 and JEC20 suggest that, although genetically divergent, strains of serotype D and A interbreed and the serotype of each strain does not always corroborate with its genotype. Therefore, serotyping alone is not a reliable method to distinguish var. grubii from var. neoformans.

Figure 3

Ploidy (a) and genetic analysis (b, c) of some representative progeny from crosses between H99 and JEC20.

View this table:
Table 1

Genetic analysis of 20 F1 progeny from a cross between a serotype A strain, H99 and a serotype D strain, JEC20

StrainSerotypemating(V8)PloidyMFURA5CAP59CNLACSTE12STE20
H-99AαnαAAAαα
B-4476DanaDDDaa
sb-1AD/adaα2nαaADADADαaαa
sb-2AD/Daα2nαaADADADαaαa
sb-3AD/Daα2nαaADADADαaαa
sb-4D/DanaDDAaa
sb-5AD/adαn+1αDDADαα
sb-6AD/ADaα2n−1αaADAADαaαa
sb-7aD/aDn+1αDADAαα
sb-8AD/ad2n−1αaDDAαaαa
sb-9D/DanaDDDaa
sb-10Ad/aDaα2n−1αaADDADαaαa
sb-11D/Daα2nαaADADADαaαa
sb-12A/Aaα2n−1αaADDAαaαa
sb-13AD/Aaα2n−1αaDDADαaαa
sb-14aD/Aaα2n−1αaADAADαaαa
sb-15AD/aDanaDDDaa
sb-16D/Dan+1aDDDaa
sb-17AD/ADan+1aDDAaa
sb-18Ad/adαnαDDAαα
sb-19D/DanaDDAaa
sb-20aD/aDαn+1αADAAαα
  • * Font size is reflective of the reaction intensity with anti-A or anti-D factor sera.

  • FACS patterns: n=haploid; 2n=diploid; n+1=haploid by FACS, aneuploid by genetic analysis; 2n−1=diploid by FACS, aneuploid by genetic analysis.

  • Haploid by genetic analysis but aneuploid by FACS pattern.

Soon after the discovery of F. neoformans, the sexual state of serotype B and C strains was observed by matings between strains NIH 191 (serotype C, MATa) and NIH 444 (serotype B, MATα) (Kwon-Chung, 1976). The teleomorph of C. gattii (C. bacillisporus) was described as a new species based on the distinct morphology of its basidiospore as well as by its failure to undergo meiosis when the type strains of the two Filobasidiella species were crossed (Kwon-Chung, 1976). Prior to the description of C. bacillisporus, Vanbreuseghem and Takashio had isolated an atypical strain of C. neoformans from a leukemic patient in Africa and described it as C. neoformans var. gattii (Vanbreuseghem & Takashio, 1970). The description of this new variety was strictly based on observations of elliptical and oval yeast cells in the histopathological sections of brain tissue. When we examined the ex-type strain of C. neoformans var. gattii (CBS 6289), it showed the same serotypic, biochemical and morphological characteristics as the serotype B strains of C. bacillisporus (Kwon-Chung, 1982). Cryptococcusneoformans and C. bacillisporus were subsequently reduced to varietal status as C. neoformans var. neoformans and C. neoformans var. gattii. The latter was based on the observed DNA–DNA relatedness between the two anamorphs as well as the result of crosses between F. neoformans and F. bacillispora. The DNA–DNA relatedness between C. neoformans and C. bacillisporus was 55–63% (Aulakh, 1981), which is higher than that commonly observed between different species of yeast. Furthermore, viable basidiospores were produced in crosses between certain strains of F. neoformans and F. bacillispora.

The agar medium initially used to discover heterothallism in the two species was malt extract agar (Raper & Fennell, 1965) and although it was useful in screening for mating type of each strain, the efficiency of mating on it was significantly lower than on V-8 juice with 4% agar, which had been formulated for mating of Filobasidiella species (Kwon-Chung, 1982). V-8 juice agar allowed certain pairs of C. neoformans and C. gattii to mate and produce viable basidiospores. Matings between B-3502 (ATCC 34874, MATa) of C. neoformans and NIH444 (MATα) or CBS 6289 (MATα), the ex-type strain of C. neoformans var. gattii, produced a low percentage of viable progeny (20–30%). Phenotypic analysis of the progeny, which included the CGB reaction, maximum growth temperature and agglutination patterns using polyclonal antibodies for serotyping, suggested that some of the progeny were hybrids (Kwon-Chung, 1982). It was not clear whether the hybrids were the result of genetic recombination or reassortment during meiosis, as the only genetic markers that had been used were the MAT alleles determined by mating reactions. It was not known whether these phenotypes were determined by nuclear genes. Moreover, ploidy determination of the progeny had not been carried out. The hybrid phenotype can also result from diploidy or aneuploidy rather than from genetic recombination or reassortment. Kwon-Chung (2002) proposed raising the two varieties to species status, C. neoformans (serotype A, D, AD) and C. gattii (serotype B, C), based on the mounting evidence of their phenotypic and genetic distinctiveness, despite the mating results.

Reliability of the ploidy analysis together with the availability of gene sequences and serotype-specific factor sera enabled us to re-examine the evidence of genetic recombination between C. neoformans and C. gattii. The F1 progeny of CBS 6289 × B-3502 maintained as glycerol stocks were revived and subjected to FACS analysis for ploidy determination (Tanaka, 1996). The genetic markers used for PCR were LAC1, URA5, CAP59 (Petter, 2001) and STE12 (AY168185). Serotyping was performed using the Crypto Check Kit (Iatron). Table 2 summarizes the antigenic, physiologic and genetic profiles of 16 progeny that were subjected to analysis. Fifty per cent of the progeny were haploid, serotype D, and contained alleles of genes specific to serotype D. The other 50% were diploid or aneuploid and contained alleles of genes from both parents. Interestingly, five of the eight nonhaploid progeny were serotyped as D and the remaining three switched from their initial BD serotype to D serotype in subsequent tests. Six of the eight diploid and aneuploid strains were weakly positive for the CGB test by 72 h while the two remaining diploid strains remained negative. These results indicate that CGB reaction or serotyping alone cannot be used for the identification of hybrids or recombinants. Unlike in the cross between serotype A strains and serotype D strains, true haploid recombinants were not obtained. Although the number of progeny analysed was small due to low viability of basidiospores produced by the cross, it appears that C. gattii and C neoformans can produce stable diploid hybrids, but not true recombinants of haploid progeny. This suggests that strains of C. gattii are genetically close enough to fuse and form basidiospores of diploids and aneuploids but may not be able to undergo meiosis and produce true recombinants. It was interesting to note that serotyping with serotype D-specific factor was more stable and dominant in diploid strains formed by the cross of CBS 6289 and B-3502. An explanation of this observation awaits elucidation of genetics on serotype-specific biochemical or antigenic features.

View this table:
Table 2

Genetic and biochemical analysis of 16 F1 progeny from a cross between a serotype B strain, CBS6289, and a serotype D strain, B3502

StrainSerotypeMating(V8)PloidyCGB CAP59 CNLAC STE12 URA5
24 h72 h
CBS6289Bαn++BBαB
B-3502DanDDaD
sb-17DanDDaD
sb-18DanDDaD
sb-26DanDDaD
sb-27Daα2n±BDBDaαD
sb-28B/D;Daα2n±BDBDaαBD
sb-30DanDDaD
sb-31B/D;Daα2n±BDBDaαBD
sb-32DanDDaD
sb-37Daα2n±BDBDaαBD
sb-38Daα2n±BDBDaαBD
sb-40Daα2n±BDBDaαBD
sb-41B/D;Dα2nBDBDaBD
sb-52DanDDaD
sb-63DanDDaD
sb-65Dan+1BDBDaD
sb-69DanBDDaD
  • * FACS patterns: n=haploid; 2n=diploid; n+1=haploid by FACS and aneuploid by genetic analysis.

  • While diploid by FACS analysis, mated as α but PCR detected only the STE12a allele.

Is the two-species system of C. neoformans supported by the cladistic (phylogenetic) species concept?

As the sequence of ribosomal DNA subunits and their intervening regions, various housekeeping genes, inteins and virulence-associated genes of C. neoformans became available, phylogenetic trees or dendrograms were constructed to show the genetic relatedness of all the serotypes. Regardless of the sequence used for comparisons of genetic relatedness, at least two distinct large clusters, one consisting of serotype A, D and AD and the other consisting of serotypes B and C became clear (Fan, 1995; Fell, 2000; Xu, 2000; Butler & Poulter, 2005; Diaz, 2005). These studies revealed that C. neoformans is a complex species and that the strains belong either to C. neoformans or C. gattii. Figures 4 and 5 show examples of trees with two major clusters: one comprised mostly of serotype A, D and AD strains and the other of serotype B and C strains. Monophyletic clusters of serotype A and serotype D strains are generally well separated within C. neoformans. This suggests that strains of serotype A and strains of serotype D can be readily classified into two varieties: serotype A strains in the var. grubii and serotype D strains in the var. neoformans. However, such a clear correlation between serotype and varietal status no longer persists when phylogenetic trees are constructed using AFLP genotypes or IGS1+5S rRNA+IGSII gene sequences. Figures 6 and 7 illustrate this point (Litvintseva, 2003; Diaz, 2005). Phylogenetic trees constructed based on IGS1+5S rRNA+IGSII sequences appear to support the three-species concept of C. neoformans. The first species contains the serotype A, D and AD strains of genotype 1a, 1b and 1c, while the second species comprises serotype A, D and AD strains, but their genotypes are 2a, 2b and 2C. The third species comprises serotype B and C strains of genotypes 3, 4 and 5. Thus, the difference between var. grubii and var. neoformans can no longer be based on the serotype of the strains but can only be based on the genotype. A phylogenetic tree constructed on the genotype identified by AFLP also did not support the classification of serotype D and A strains into two different species because of the considerable number of AD hybrids and the genetic diversity among the strains of serotype A. The controversy regarding classification of the var. grubii and var. neoformans into two species can be resolved by the application of the biologic species concept to the taxonomy of C. neoformans. However, C. gattii is composed of serotype B and C strains, which are more divergent and can readily be differentiated from C. neoformans regardless of the sequence used to analyse the genetic relatedness.

Figure 4

The most-parsimonious tree for 34 isolates of Cryptococcus neoformans derived from the sequence of 462 bases of the LAC1 gene (courtesy of Dr. J. P. Xu).

Figure 5

Phylogenetic tree based on an alignment of the DNA sequences of the Cryptococcus neoformans PRP8 inteins (Butler & Poulter, 2005).

Figure 6

Genetic diversity among Cryptococcus neoformans on the basis of 96 AFLP markers, presented in a neighbor-joining dendrogram with Nei-Li genetic distances (Livintseva,2005). With Permission from The University of Chicago Press.

Figure 7

Phylogenetic tree of Cryptococcus neoformans derived from IGSI+5S rRNAt+IGSII sequence data (Diaz, 2005).

Discussion and concluding remarks

The nomenclature of pathogenic species such as C. neoformans affects infectious disease specialists and clinical microbiologists more so than cryptococcal taxonomists. It is imperative, therefore, to use the most appropriate species concept for the classification of C. neoformans in order to achieve nomenclatural stability. This will minimize confusion in scientific communications between investigators in different fields of cryptococcal research and physicians as well as diagnostic mycologists. The need for recognition of the two species within C. neoformans is clear not only due to their biologic and ecological differences, but also due to the differences between C. neoformans and C. gattii in their association with host immune status. Cryptococcus gattii infections have been documented more commonly in immunocompetent individuals, whereas a majority of the patients infected with C. neoformans have been immunocompromised, especially those with HIV infection (Speed & Dunt, 1995; Chen & Sorrell, 2000). Furthermore, some reports have suggested that prognosis is worse with infections due to C. gattii (Mitchell, 1995). Differences in the immunological status of the host and infections caused by the two serotypes A and D within C. neoformans, however, has not been documented.

The valid number of species thus far suggested for C. neoformans has been two, three or eight as discussed at the 6th International Conference on Cryptococcus and Cryptococcosis (Boekhout, 2005; Kwon-Chung, 2005; Meyer, 2005). Different numbers of species were proposed because of the differing opinions among taxonomists as to the definition of a species. For example, those who accept the cladistic (phylogenetic) species concept disregard the biology of the organism and focus only on DNA sequences chosen for the analysis of genetic relatedness, and proposed up to eight species. Fingerprinting based on PCR amplifications using M13 (minisatellite) primers (Kidd, 2003) or AFLP analysis recognized eight major molecular types. The genotypic variation between these types may be comparable with that seen between fungal species established by classical methods (Meyer, 2005). However, the genotypic variation between these molecular types does not always reflect their biology as is the case with species established by classical methods. Comparative analysis of intergenic spacer (IGS) regions of rRNA genes revealed up to 12 different genetic lineages, although no investigator had suggested recognition of the 12 lineages as different species (Diaz, 2005). Because the IGS regions are the most rapidly evolving regions of the rRNA gene cluster, several more molecular types will surface as more strains become available for genotyping from different geographic regions where C. neoformans isolates have not yet been studied. Should the number of C. neoformans species increase as more genotypes are found without having any clear biologic difference? When populations of C. neoformans become isolated, either through geographic distribution or through differences in their reproductive biology, they will ultimately result in speciation by genetic divergence. During this process of genetic divergence, it is expected that there will be distinct populations on the way to species formation. If these incipient species are described as species or subspecies, it will create mass confusion as it is extremely difficult and subjective to decide when a population has sufficiently diverged from other populations to merit a subspecies or a species rank. Does the genotypic variation between different groups represent a different biologic species? This problem can be overcome in C. neoformans because it has an interbreeding system. The two-species concept employs a combination of biologic (interbreeding), morphologic and genetic species definition. If two different genotypic groups interbreed and produce genetic recombinants, they are classified as the same species. Among strains of serotype A and D, about 15% did not belong to one particular genotypic group (Fell, 2000; Boekhout, 2001). Investigators who accept the classification of C. neoformans in three varieties, var. neoformans, var. grubii and var. gattii, simply designated all serotype A strains as var. grubii as if serotype A is exclusive to that variety. Such a simple approach in classification has created confusion in the C. neoformans nomenclature in several ways:

  1. The serotype is based on the antigenic property of the strain that is not always stable. Some diploid strains have been serotyped as AD, A or D. A good example is the type strain of C. neoformans, CBS 132T, which was originally serotyped as D (Boekhout, 2001), but also reported as serotype A (Lengeler, 2001) or AD (K. J. Kwon-Chung, unpublished data) depending on the laboratory where it was serotyped. A similar phenomenon was observed among the diploid and aneuploid F1 strains obtained by crossing H99 and JEC20, as shown in Table 1. Although changes in serotype from AD to A or D may be interpreted as instability of diploid strains during prolonged maintenance (Brandt, 1995), changes in the serotype of the F1 diploid strains was seen without prolonged propagation. Molecular analysis of these strains has shown them to contain alleles of genes specific for both A and D.

  2. There are hybrids that contain both serotype D and A alleles, but are serotyped as either A or D but not AD (Boekhout, 2001; Cogliati, 2000). Such populations have been reported mostly from European countries where strains of serotype D and A coexist in the same environment. Our experimental results of crosses between H99 (the type strain of var. grubii) and JEC20 (=B-4476, the MATa, serotype D strain isogenic to JEC21) clearly showed that they produce diploids, aneuploids, parental types and recombinants. The aneuploid or recombinant progeny was unstable, in that their serotype switched between D, A or AD in 2–5 separate serotyping tests performed a few days to a year apart using the commercial serotyping kit. Should the varietal status of these progeny change in accordance with switching of their serotypes?

  3. There is confusion in the nomenclature of diploids or aneuploids where the serotype is consistently AD. A substantial number of serotype AD strains have been isolated from various parts of the world and they cannot simply be ignored as rare strains. Immunofluorescence patterns using the monoclonal antibody 13F1 suggest that strains of the AD serotype represent a fifth C. neoformans subgroup (Cleare, 1999). The nomenclature of this fifth serotype will be a problem as it is neither var. neoformans nor var. grubii. Under the current classification of the three varieties, some investigators have avoided classifying the AD strains in any particular variety whereas others have included them in C. neoformans var. grubii (Franzot, 1999; Diaz, 2005) causing further complications.

  4. Multiple subgroups may exist within each serotype. For example, at least two distinct subgroups of serotype A strains have been reported from the sub-Saharan region alone (Litvintseva, 2005). As further genotyping of strains from different geographic areas becomes available, it is bound to reveal hitherto unreported subgroups among the serotype A strains. Genetic divergence between serotype A strains collected from AIDS patients in Botswana and the serotype A strains from North Carolina was just as great as between the serotype A strains and the serotype D strains from North Carolina (Litvintseva, 2003) (Fig. 6). If we use the names var. grubii and var. neoformans to recognize the genetic divergence between serotype A strains and serotype D strains, which subgroup of A serotype should represent var. grubii?

  5. Although relatively few in number, there are serotype untypable strains that do not react with any factor sera (Brandt, 1995). These strains have to be genotyped to identify their varietal status. In laboratories where the means either to genotype or to serotype are unavailable, these strains can still be classified under the two-species system, as the one-step diagnostic methods using simple media such as d-proline agar (Dufait, 1987) or CGB agar are readily available (Kwon-Chung, 1982).

In order to be certain about the varietal status of each C. neoformans strain, one cannot rely on its serotype alone. Unfortunately, the literature is full of inaccurate statements such as ‘serotype A strains are classified in var. grubii and strains of serotype D are classified in var. neoformans’. The IGS data clearly show it not to be fully concordant with the classification based on serotype (Diaz, 2005). The simplest way to avoid confusion in the identity of each strain would be to designate the strains by their serotype under the two-species system: C. neoformans serotype A, D and AD; C. gattii serotype B and C. In order to determine to which subgroup the strain of interest belongs, one can further identify the strain by PCR fingerprinting, AFLP genotyping or other molecular typing methods. It is desirable, though, to employ a standard molecular typing method that offers the highest discriminatory power in order to identify the genotype of strains. The genotype determined by a combination of the DNA sequence of IGSI+5S rRNA+IGSII appears to be an appropriate method to subtype each strain within the two species. The rRNA gene sequence has proven to be a stable and reliable marker for genetic relatedness between strains as well as species. rRNA gene sequences have thus been the most widely used molecular marker for the phylogenetic study across the five kingdoms (Wainright, 1993).

Acknowledgements

This work was supported by funds from the intramural program of the National Institute of Allergy and Infectious Diseases. Figure 5 was reprinted from Butler & Poulter (2005), with permission from Elsevier.

Footnotes

  • Editor: Stuart Levitz

References

View Abstract