Simultaneous identification of molecular and mating types within the Cryptococcus species complex by PCR-RFLP analysis

The Cryptococcus species complex contains the most common causative agents of fungal meningoencephalitis in immunocompromised patients, particularly in the AIDS population (Mitchell & Perfect, 1995; Casadevall & Perfect, 1998). Within the Cryptococcus species complex, two species, Cryptococcus neoformans and Cryptococcus gattii, have been named. Both species have a bipolar mating system involving α and a mating types (MAT) (Kwon-Chung et al., 2002; Kwon-Chung & Varma, 2006). The Cryptococcus species complex is currently divided into eight major molecular types: VNI/AFLP1 and VNII/AFLP1A/1B (C. neoformans var. grubii, serotype A); VNIII/AFLP3 (AD hybrid, serotype AD); VNIV/AFLP2 (C. neoformans var. neoformans, serotype D); and VGI/AFLP4, VGII/AFLP6, VGIII/AFLP5 and VGIV/AFLP7 (C. gattii, serotypes B and C) (Boekhout et al., 2001; Latouche et al., 2003; Meyer et al., 2003; Bovers et al., 2008a). In addition to the eight established molecular types with opposite mating types, a genotype, VNB, unique to Botswana and a few novel hybrids of AαDα and C. neoformansxC. gattii have been described in recent studies (Litvintseva et al., 2005a, 2006; Bovers et al., 2006; Lin et al., 2007; Bovers et al., 2008b). Differences in biology, epidemiology, pathogenicity, clinical features and drug susceptibility have been reported to be associated with species, varieties, molecular types and mating types within the Cryptococcus species complex (Kwon-Chung et al., 1992; Speed & Dunt, 1995; Casadevall & Perfect, 1998; Meyer et al., 2003; Trilles et al., 2004; Campbell et al., 2005; Fraser et al., 2005).

C. neoformans is a ubiquitous pathogen associated with avian faeces and decaying wood, being responsible for most cases of cryptococcosis. The most common variety, C. neoformans var. grubii, primarily infects immunocompromised individuals and accounts for most cryptococcal infections (>90 %) (Casadevall & Perfect, 1998). Strains of serotype A MATα (Aα) have been shown to be more pathogenic than Aa strains in several reports (Lengeler et al., 2000; Keller et al., 2003; Barchiesi et al., 2005). Despite molecular types VNI and VNII both appearing to be distributed globally, VNII is recovered less frequently (Meyer et al., 2003). In addition, a unique genotype, VNB, belonging to C. neoformans var. grubii has recently been found to be restricted to Botswana. VNB includes a large proportion of fertile isolates of MATa (Litvintseva et al., 2006). The second variety, C. neoformans var. neoformans, causing cases mainly in Europe, has been reported to be less virulent and more susceptible to fluconazole than C. neoformans var. grubii (Tortorano et al., 1997; Casadevall & Perfect, 1998). In murine models, Dα strains are more virulent than congenic Da strains (Kwon-Chung et al., 1992). AD strains are hybrids between the two varieties, and possess two mating type alleles with one from each of the varieties. It is suggested that AD hybrids may be more common in clinical and environmental sources than was previously thought (Litvintseva et al., 2005a; Viviani et al., 2006). Further studies have shown that AD hybrids exhibit hybrid vigour in the laboratory (Lin et al., 2007; Litvintseva et al., 2007). Interestingly, strains of AαDa have been reported to be more virulent than AaDα strains (Barchiesi et al., 2005). Moreover, a hybrid genotype of AαDα has recently been recovered from the USA and Italy, implicating same-sex mating between α mating types within this species (Litvintseva et al., 2005a; Lin et al., 2007).

C. gattii, previously known as C. neoformans var. gattii, has been raised to species level (Kwon-Chung et al., 2002). This sibling species was primarily restricted to tropical and subtropical climates until an outbreak on Vancouver Island, Canada, in 2000. It inhabits decaying wood of eucalypts or other tree species and primarily infects immunocompetent individuals (Casadevall & Perfect, 1998; Kidd et al., 2004). Compared with C. neoformans, the clinical characteristics of C. gattii have been reported to be different. It is the cause of cryptococcomas and has a lower susceptibility to several antifungal agents, a need for prolonged treatment and a higher mortality rate (Speed & Dunt, 1995; Trilles et al., 2004). Like C. neoformans, isolates of C. gattii with the MATα mating type are also predominant in clinical and environmental samples, especially in the former (Campbell et al., 2005; Fraser et al., 2005; Litvintseva et al., 2005b). Among the four previously defined molecular types of C. gattii, molecular types VGI and VGII are widely distributed, whereas VGIII and VGIV are usually isolated from tropical climates (Meyer et al., 2003; Litvintseva et al., 2005b). Strains of VGII MATα have been revealed to be related to a high incidence of C. gattii infections, such as the cryptococcosis outbreak on Vancouver Island (Kidd et al., 2004; Campbell et al., 2005; Fraser et al., 2005). Strains of VGIV MATα have been recovered mainly from patients with AIDS in sub-Saharan Africa (Litvintseva et al., 2005b). Furthermore, several novel C. gattiixC. neoformans hybrids in which the C. gattii alleles originated from the molecular type VGI have been described in clinical samples (Bovers et al., 2006, 2008b).

The population structure of the eight molecular types has been confirmed by sequence and phylogenetic analyses of multiple genes (Fraser et al., 2005; Bovers et al., 2008a). RFLP analysis based on a single gene was developed for the differentiation of the molecular types and confirmed to be a reliable and easy-to-use method (Latouche et al., 2003; Meyer et al., 2003). In a recent study, sequence analysis of the CAP1 gene, which is located at the MAT locus, indicated that sequence variation due to the mating type was greater than that due to the molecular type divergence. The dendrogram showed that strains clustered according to mating type, and in each of the two clades, strains clustered according to molecular type (Kidd et al., 2005). In a previous study by us, partial sequences of the GEF1 gene (a gene also located at the MAT locus) for strains representative of the distinct molecular and mating types were determined. The results showed that this gene exhibited genetic traits similar to those of the CAP1 gene (X. Feng and others, unpublished results). These data demonstrated that the molecular and mating types of strains belonging to the Cryptococcus species complex could be determined simultaneously by sequencing either of the two genes.

In the present study, two alternative approaches using PCR-RFLP analysis, based on the CAP1 and GEF1 genes, requiring only common materials in the laboratory were developed to determine simultaneously the molecular and mating types of isolates of the Cryptococcus species complex.

Strains and DNA extraction. One hundred and forty-four isolates of the Cryptococcus species complex comprising 96 C. neoformans isolates and 48 C. gattii isolates were used in this study. Among the tested strains, 135 represented the 14 distinct groups of the molecular and mating types, namely VNI MATα, VNI MATa, VNII MATα, VNIII MATα/a, VNIII MATa/α, VNIV MATα, VNIV MATa, VGI MATα, VGI MATa, VGII MATα, VGII MATa, VGIII MATα, VGIII MATa and VGIV MATα. The other nine isolates represented the rare subtypes of the genotype VNB (including MATα and MATa) and the AD hybrid of AαDα. The data for the strains tested are shown in Table 1. The molecular and mating types of 69 isolates were determined previously in other studies, and those of the other 75 isolates (Table 1) were determined by PCR fingerprinting (Meyer et al., 2003) and PCR analysis with primers specific for the STE20 and STE12 genes situated within the MAT locus as described previously by Lengeler et al. (2001) and Fraser et al. (2003), respectively. The strains recovered from China were identified as Cryptococcus species by morphology and biochemical and physiological tests, such as microscopic observation of India ink preparation, black or brown colony appearance on L-3,4-dihydroxyphenylalanine agar, a positive urease reaction and the ability to grow at 37 °C (X. Feng and others, unpublished results). Strains were maintained on yeast extract-peptone-glucose medium at 30 °C for 2 days before testing. Genomic DNA was prepared from each strain as described previously (Meyer et al., 2003).

Table 1. Strains of the Cryptococcus species complex tested in this study and RFLP results obtained for the CAP1 and GEF1 genes

PCR amplification. The sequences of the CAP1 and GEF1 genes from the reference strains representing distinct molecular and mating types, including H99 (VNI MATα; GenBank accession no. AF542529), 125.91 (VNI MATa; AF542528), JEC21 (VNIV MATα; AF542531), JEC20 (VNIV MATa; AF542530), WM276 (VGI MATα; AY710430), E566 (VGI MATa; AY710429) and R265 (VGII MATα; DQ096809), were obtained from GenBank. The sequences for the reference strains of WM626 (VNII MATα), CBS1930 (VGII MATa), WM161 (VGIII MATα), B4546 (VGIII MATa) and WM779 (VGIV MATα) were generated in our study using TA cloning and a nested sequencing approach. Primers were designed from the consensus sequence based on alignment of the targeted gene sequences from the above reference strains. For the CAP1 gene, two primers (CAP1F, 5'-TCAGATCAACCGCGACATCG-3'; and CAP1R, 5'-TGAACCCAAGGGGTAAGCCA-3') comprising nt 664–683 and nt 1879–1898, respectively, based on the CAP1 sequence of strain JEC21 were designed to amplify a portion of the CAP1 gene from all strains investigated (Fig. 1a). For the GEF1 gene, two primers (GEF1F, 5'-GGACCCATGCCTGAAATGTG-3'; and GEF1R, 5'-TACGCTTGCCCCGATCTG-3') comprising nt 1703–1722 and nt 3024–3041, respectively, based on the GEF1 sequence of the strain JEC21 were designed to amplify a portion of the GEF1 gene from all strains tested (Fig. 1b).

(11K): Fig. 1. Schematic representation of the amplified regions of the CAP1 gene (a) and GEF1 gene (b). Primer binding sites are shown by short arrows. Open boxes represent exons, with the order indicated by numbers, whilst lines between exons represent introns.

Amplification was performed in 1x PCR buffer containing 1.5 mM MgCl₂, 0.2 mM each dNTP, 10 µM each primer and 1.5 U Taq polymerase. The PCR conditions for both amplifications were: initial denaturation at 94 °C for 5 min, followed by 31 cycles of 94 °C for 30 s, 57 °C for 30 s and 72 °C for 1.2 min, with a final extension step at 72 °C for 5 min. PCR products were analysed on a 1 % (w/v) agarose gel.

RFLP analysis. Using DNAMAN 5.2.2 software, analyses of sequence comparisons, restriction endonuclease selection, restriction maps and unique cutting sites were conducted based on the amplified DNA sequences of the CAP1 and GEF1 genes from type strains H99, 125.91, WM626, JEC21, JEC20, WM276, E566, R265, CBS1930, WM161, B4546 and WM779, representing 12 distinct groups of molecular and mating types. Additionally, AD hybrids were analysed using a combination of serotype A (VNI) and serotype D (VNIV) alleles. The restriction endonucleases AvaI, HgiCI, EcoRII and PstI were chosen and predicted to yield specific RFLP patterns for the differentiation of the distinct groups of all molecular and mating types following digestion of the PCR products from the CAP1 gene. The PCR products were subjected to separate restriction digestions with AvaI and HgiCI enzymes in a first-line identification; PCR products with patterns identical to those of molecular type VNI or VNII (both MATα) were further digested with EcoRII, and PCR products with patterns identical to those of molecular type VGIII or VGIV (both MATα) were further digested with PstI. The restriction endonucleases EcoT14I and HapII were selected and predicted to yield specific RFLP patterns for the differentiation of the distinct groups of all molecular and mating types following digestion of the PCR products from the GEF1 gene. The PCR products were subjected to separate restriction digestions with EcoT14I and HapII enzymes for identification.

All restriction digestions were performed according to the manufacturer's instructions (TaKaRa). The reaction mixture was incubated at 37 °C for 3 h before separation on a 2.0 % (w/v) agarose gel at 100 V for 1 h and then visualization under UV light.

RFLP identification of the molecular and mating types
The 144 C. neoformans and C. gattii isolates tested gave positive amplification with both primer pairs: CAP1F/CAP1R and GEF1F/GEF1R. For all strains assayed, a fragment of approximately 1235 bp was amplified from the genomic DNA with primers CAP1F and CAP1R, and one of approximately 1339 bp with primers GEF1F and GEF1R.

By separate digestions with AvaI and HgiCI, RFLP analysis of the fragments amplified from the CAP1 gene generated 12 different patterns, compatible with the distinct groups of the eight molecular types (VNI, VNII, VNIII, VNIV, VGI, VGII, VGIII and VGIV) and the two mating types (α and a) within the 135 isolates of the Cryptococcus species complex. The same restriction pattern was shown by the molecular types VNI and VNII (both MATα) and a different pattern was shown by both VGIII and VGIV (both MATα), as predicted by in silico analysis. In concordance with the in silico analysis, molecular types VNI and VNII (both MATα) could be differentiated further by EcoRII digestion, and VGIII and VGIV (both MATα) could be differentiated further by PstI digestion (Tables 1 and 2; Fig. 2). RFLP analysis of the fragments amplified from the GEF1 gene via separate digestions with EcoT14I and HapII generated 14 different patterns corresponding exactly to the distinct groups of the eight molecular types and the two mating types for the 135 isolates as expected from the in silico analysis (Tables 1 and 2; Fig. 3). Thus the 135 cryptococcal isolates were successfully grouped corresponding to their unique molecular and mating types using either of the two PCR-RFLP approaches. No profile variation was observed when the patterns were analysed, confirming the reliability of the two typing approaches. In addition, the unusual AD hybrid of AαDα was also distinguished by the established RFLP patterns by either of the two typing approaches (Tables 1 and 2; Figs 4 and 5).

Table 2. Categories of restriction profiles from the CAP1 and GEF1 gene fragments with the corresponding molecular and mating types

(28K): Fig. 2. RFLP patterns of CAP1 gene fragments. Fourteen RFLP patterns (C1–C14) were generated with the enzymes AvaI (A), HgiCI (H), EcoRII (E) and PstI (P). Left panel, RFLP patterns of C. neoformans; right panel, RFLP patterns of C. gattii. M, Molecular marker.

(27K): Fig. 3. RFLP patterns of GEF1 gene fragments. Fourteen RFLP patterns (G1–G14) were generated with the enzymes EcoT14I (E) and HapII (H). Left panel, RFLP patterns of C. neoformans; right panel, RFLP patterns of C. gattii. M, Molecular marker.

(60K): Fig. 4. RFLP patterns of CAP1 gene fragments. Pattern C16 (AαDα) contains the bands from C1 (VNI MATα) rather than C3 (VNII MATα). A, AvaI; H, HgiCI; E, EcoRII; M, molecular marker.

(54K): Fig. 5. RFLP patterns of GEF1 gene fragments. Pattern G4 (AαDa) contains the bands from G7 (VNIV MATa) and G1 (VNI MATα). Pattern G16 (AαDα) contains the bands from G6 (VNIV MATα) and G1 (VNI MATα). E, EcoT14I; H, HapII; M, molecular marker.

Identification of VNB strains
The ability of the two RFLP typing approaches to identify the rare genotype VNB and its mating type was also investigated. The VNB MATα strains exhibited the same restriction profile as the molecular type VNI MATα isolates after separate digestions of the CAP1 gene with AvaI, HgiCI and EcoRII, whereas the VNB MATa strains showed a unique pattern distinct from the 12 patterns established by separate digestions with AvaI and HgiCI (Tables 1 and 2; Fig. 6). The VNB MATα strains displayed a unique pattern in comparison with all patterns established for the GEF1 gene (Tables 1 and 2; Fig. 5), but the restriction pattern obtained from the VNB MATa strains was identical to the pattern of the molecular type VNI MATa strains. These results suggest that VNB MATα and VNB MATa strains could not both be distinguished from VNI strains by either of the two methods. Therefore, a combination of the two typing approaches is required for the identification of the VNB strains.

(43K): Fig. 6. RFLP patterns of CAP1 gene fragments. Pattern C5 (AaDα) contains the bands from C6 (VNIV MATα) and C15 (VNB MATa) rather than C2 (VNI MATa). A, AvaI; H, HgiCI; M, molecular marker.

Analysis of AD hybrid subtypes
Multilocus genotyping methods such as PCR fingerprinting, amplified fragment length polymorphism analysis and multigene studies (Boekhout et al., 2001; Lengeler et al., 2001; Meyer et al., 2003; Litvintseva et al., 2005a; Bovers et al., 2007, 2008b) have been needed until now for the precise identification of cryptococcal hybrids. However, in the current study, all three mating types (α/a, a/α and α/α) of the AD hybrids could be differentiated by either of the two RFLP approaches. They showed distinct restriction patterns from those of the other groups of the molecular and mating types. In spite of the similarity in patterns of AαDα and Aα strains in the RFLP analysis of the CAP1 gene, an additional band was observed in AαDα hybrids with EcoRII digestion following gel electrophoresis.

In contrast to the in silico analysis in which AD hybrids were analysed using a combination of molecular type VNI and VNIV alleles, AaDα hybrids showed a mixed profile from the genotype VNB MATa profile rather than molecular type VNI MATa in the RFLP analysis of the CAP1 gene (Fig. 6). The analysis showed that AαDα hybrids displayed partial bands from the VNI MATα profile rather than the VNII MATα profile (Fig. 4), and AαDa hybrids exhibited a mixed profile from VNI or VNII (both MATα) and VNIV MATa profiles (Fig. 2). In the RFLP analysis of the GEF1 gene, AαDa hybrids exhibited a combination of bands from the patterns of VNI MATα and VNIV MATa, AαDα hybrids exhibited a combination of bands from the patterns of VNI MATα and VNIV MATα (Fig. 5), and AaDα hybrids exhibited a combination of bands from the patterns of VNI or VNB (both MATa) and VNIV MATα (Fig. 3).

A recent study suggested that many globally recovered AaDα strains originated in sub-Saharan Africa because their MATa serotype A alleles were closely related to genotype VNB, which is geographically restricted to sub-Saharan Africa (Litvintseva et al., 2007). Likewise, the RFLP profile generated from the AaDα strains tested here was found to contain bands from genotype VNB but not from molecular type VNI in the RFLP analysis of the CAP1 gene. The AαDa strains exhibited a mixed profile of VNI MATα and VNIV MATa in the RFLP analysis of the GEF1 gene. This hybrid genotype has also been detected by RFLP analysis of the PLB1 gene and confirmed by phylogenetic analysis in previous studies (Latouche et al., 2003; Litvintseva et al., 2007), suggesting that the predominant distribution of the molecular type VNI may result in the detection of this genotype in many AαDa strains. Similarly, strains of AαDα, a progeny originated from same-sex mating, also had a MATα serotype A allele derived from molecular type VNI rather than VNII or VNB in accordance with other research (Litvintseva et al., 2007).

AD hybrids are diploid or aneuploid. Strains with loss of serotype A- or D-specific alleles in some loci, such as CLA4, LAC1, URA5 and IGS1, have been detected in several studies (Lengeler et al., 2001; Cogliati et al., 2006; Bovers et al., 2007). Just like the genes at the MAT locus, the absence of one mating-type allele was found in a few AD strains (less than 7 % in AD hybrids) and a novel AB hybrid (Cogliati et al., 2001; Lengeler et al., 2001; Litvintseva et al., 2005a; Viviani et al., 2006; Bovers et al., 2008b). Because the MAT locus is inherited as a single unit (Lengeler et al., 2002), the absence of the CAP1 or GEF1 allele in the RFLP analysis indicates loss of the whole MAT locus.

In summary, for identification of the eight major molecular types of the Cryptococcus species complex, various molecular typing techniques have been developed such as PCR fingerprinting, amplified fragment length polymorphism analysis, RFLP, sequencing and luminex xMAP technology (Boekhout et al., 2001; Meyer et al., 2003; Diaz et al., 2005; Bovers et al., 2007), and the mating types are usually determined by separate PCR analyses or mating assays (Cogliati et al., 2001; Lengeler et al., 2001; Fraser et al., 2003; Campbell et al., 2005). In the present study, the molecular and mating types of all cryptococcal isolates tested were determined simultaneously by RFLP analysis based on either the CAP1 or the GEF1 gene. However, the RFLP analysis of the GEF1 gene seems to be more convenient than the RFLP analysis of the CAP1 gene and should be a preferred typing method. Overall, the RFLP analysis developed here provides an alternative choice for easy identification of the molecular and mating types of strains belonging to the Cryptococcus species complex and may be a useful tool in clinical microbiological and epidemiological studies of this pathogenic yeast.

We thank Joseph Heitman, John R. Perfect, Anastasia P. Litvintseva, Wiley A. Schell, Sheryl Frank and Anna Floyd (Duke University Medical Center, Durham, USA), Wieland Meyer and Dee Carter (University of Sydney, Sydney, Australia), C. De Vroey (Prince Leopold Institut of Tropical Medicine, Belgium), Maria Anna Viviani (Istituto di Igiene e Medicina Preventiva, Milano, Italy), Kyung J. Kwon-Chung (National Institutes of Health, Bethesda, MD, USA), James W. Kronstad (University of British Columbia, Vancouver, BC, Canada), Teun Boekhout (Centraalbureau voor Schimmelcultures–Fungal Biodiversity Centre, Utrecht, The Netherlands), Jianping Xu (McMaster University, Hamilton, Canada) and Claudete R. Paula (University of São Paulo, São Paulo, Brazil) for strains. We also thank Wieland Meyer for his assistance in correction of the manuscript. This work was supported by a grant (no. 30870108) from the National Natural Science Foundation of China.