Identification of a Mating Type-Like Locus in the Asexual Pathogenic Yeast Candida albicans

See allHide authors and affiliations

Science  20 Aug 1999:
Vol. 285, Issue 5431, pp. 1271-1275
DOI: 10.1126/science.285.5431.1271

This article has a correction. Please see:


Candida albicans, the most prevalent fungal pathogen in humans, is thought to lack a sexual cycle. A set of C. albicans genes has been identified that corresponds to the master sexual cycle regulators a1, α1, and α2 of the Saccharomyces cerevisiae mating-type (MAT) locus. TheC. albicans genes are arranged in a way that suggests that these genes are part of a mating type–like locus that is similar to the mating-type loci of other fungi. In addition to the transcriptional regulators a1, α1, and α2, the C. albicans mating type–like locus contains several genes not seen in other fungal MAT loci, including those encoding proteins similar to poly(A) polymerases, oxysterol binding proteins, and phosphatidylinositol kinases.

The yeast C. albicans is the most common human fungal pathogen causing most cases of oral and vaginal thrush as well as severe mucosal and systemic infections in immunocompromised individuals (1). A principle difficulty in studying C. albicans, compared with other yeasts, is thatC. albicans has no known sexual cycle and is therefore not amenable to conventional genetic analysis. It is a diploid organism for which no haploid state has been observed, nor has any process resembling meiosis or spore formation been detected. Sexual reproduction in fungi is typically controlled by genes that reside in a genetic locus called a mating-type, or MAT, locus. In the budding yeast Saccharomyces cerevisiae, the genes residing at the MAT locus can be either the a type or the α type. These genes, which code for transcriptional regulators, specify the three cell types involved in the S. cerevisiaesexual cycle (2, 3). A cell containing only theMATa locus is an a cell; a cell containing only the MATα locus is an α cell; and a cell that contains both MATa and MATα (typically diploid and therefore heterozygous at the MAT locus) is ana/α cell. MATa codes for the homeodomain protein a1, which has no known function in acells. MATα codes for a homeodomain protein (α2) and an α-domain protein (α1) that cause the repression ofa-specific genes and the activation of α-specific genes, respectively (Fig. 1A). Because thea- and α-specific genes encode proteins required for each of the cell types to mate, these changes in gene expression differentiate the a cell from the α cell. The product of a successful mating is an a/α diploid cell in whicha1 and α2 are both expressed. The a1 and α2 proteins are the key regulators of the a/α cell type, and together they bind to a specific DNA sequence to repress the transcription of many target genes (including α1). Thisa1/α2 regulatory activity shuts off the ability of cells to mate and at the same time permits meiosis and sporulation in the presence of the appropriate external nutritional signals.

Figure 1

Features of the S. cerevisiaemating-type (MAT) locus and the C. albicans mating type–like (MTL) locus. (A) The S. cerevisiae MAT locus contains ORFs for three gene-regulatory proteins, a1, α1, and α2, that are located on homologous chromosomes. The region of heterologous DNA sequence between the two chromosomes is 642 base pairs (bp) for thea chromosome and 747 bp for the α chromosome. (B) The C. albicans MTL locus contains ORFs for nine proteins from four families of proteins: three gene-regulatory proteins, two phosphatidylinositol kinases, two oxysterol binding protein–like proteins, and two poly(A) polymerases. The region of DNA sequence that differs between the MTLa and MTLαsegments is 8742 bp for MTLa and 8861 bp forMTLα.

Shown in Fig. 1B are two genomic fragments from C. albicansthat contain clusters of genes that bear a marked resemblance to theMATa and MATα genes of S. cerevisiae(denoted mating type–like, or MTLa and MTLα, respectively). The MTLa gene cluster was obtained by chromosome walking with a lambda library of C. albicansgenomic fragments. The beginning probe for the walk was based on a sequence trace from the Stanford C. albicans Sequencing Project that resembled a portion of the S. cerevisiae MATa1gene. MTLα was obtained by walking downstream ofMTLa to its flanking DNA sequence and then back into and through MTLα (4, 5).MTLa contains four open reading frames (ORFs) that encode a gene regulatory protein, a poly(A) polymerase, an oxysterol binding protein–like protein, and a phosphatidylinositol kinase.MTLα contains four genes whose products are closely related to those in MTLa plus an additional ORF coding for another gene-regulatory protein. The DNA sequences within MTLaand MTLα are ∼48% identical overall; however, the DNA sequences flanking them are greater than 99% identical. We have defined the borders of MTLa andMTLα as the points at which their DNA sequences become greater than 99% identical.

Although the clusters of genes in the MTL locus are much larger than those of the S. cerevisiae MAT locus (9 versus 0.7 kb), three features of the MTL locus in C. albicans are markedly similar to those of the MAT locus in other fungi, especially S. cerevisiae (Fig. 1). First, three of the proteins coded for by the C. albicans locus have predicted amino acid sequences very similar to those of the transcriptional regulators a1, α1, and α2 encoded by theS. cerevisiae MAT locus. The C. albicans MTLasegment codes for a homeodomain protein similar in sequence to theS. cerevisiae a1 protein (30% identity and 56% similarity over the entire protein and 43% identity and 59% similarity in the homeodomain region). The C. albicans MTLα segment codes for an α-domain protein similar to theS. cerevisiae α1 protein and for a homeodomain protein similar to the S. cerevisiae α2 protein. The predictedC. albicans α1 protein is 26% identical (49% similar) to the S. cerevisiae α1, and the C. albicans α2 is 28% identical (58% similar) to the S. cerevisiae α2 protein, with particularly strong similarity seen in the homeodomain region (44% identity and 69% similarity) (Fig. 2).

Figure 2

Sequence comparisons between the C. albicans and S. cerevisiae a1, α1, and α2 proteins. Alignment was performed with the GCG8 program pileup (Genetics Computer Group, Madison, WI) and displayed with the program SeqVu 1.1 GES algorithm (Garvan Institute of Medical Research, Darlinghurst, Sydney, Australia). Boxes indicate sequence identity and shading indicates similar and identical resides. (A) Protein alignment based on predicted amino acid sequence for C. albicans a1. Dots indicate possible start codons for the proteins, and inverted triangles show the positions of introns in the C. albicans gene. The first intron is 74 bp long (S. cerevisiae first intron is 51 bp), and the second intron is 256 bp (S. cerevisiae second intron is 53 bp). The diamond shows the position of intron 2 in the S. cerevisiae MATa1 gene. Arrows at C. albicans amino acid positions 140 and 197 delineate the homeodomain. (B) Protein alignment based on predicted amino acid sequence for C. albicans α2. Inverted triangle indicates position of the intron in the C. albicans(59 bp) gene, and arrows at C. albicans amino acid positions 111 and 172 delineate the homeodomain. (C) Protein alignment based on predicted amino acid sequence for the C. albicans α1 protein. Several α1-like proteins have been identified in ascomycete fungi, and a conserved region designated the α domain (29) has emerged. Arrows at C. albicans amino acid positions 90 and 146 delineate the α domain. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

A second similarity between the C. albicans MTL and S. cerevisiae MAT loci concerns the overall organization of these three genes. As is true for the S. cerevisiae genes, theC. albicans MTLα1 and MTLα2 genes are transcribed divergently from one chromosome, and the MTLa1gene is found on the other chromosome (Fig. 1). This feature is particularly notable because gene order and organization are not generally conserved between C. albicans and S. cerevisiae (6, 7).

A third similarity between C. albicans and S. cerevisiae is the conserved positions of the introns in theMATa1 and MTLa1 genes. In S. cerevisiae, the MATa1 gene is one of only a few genes interrupted by two introns (8). The C. albicans MTLa1 gene also appears to be interrupted by two introns, one of which is in the same position (in the “recognition” helix of the homeodomain) as that in the S. cerevisiae MATa1 (Fig. 2). The locations of the introns in the C. albicans MTLa1 gene were initially predicted from the DNA sequence and were verified by observing the sizes of reverse transcription–polymerase chain reaction (RT-PCR) products (9). In S. cerevisiae, theMATα2 gene is free of introns, but the C. albicans MTLα2 gene contains a single intron located in the same position within the recognition helix of the homeodomain as the COOH-terminal introns in the C. albicans MTLa1 and S. cerevisiae MATa1 genes.

Candida albicans is diploid, and several lines of evidence support the idea that MTLa and MTLα reside at the same position on homologous chromosomes. Two types of C. albicans MTLa1 deletion mutants were constructed (a complete deletion of the MTLa1 ORF and a deletion of only the homeodomain) by homologous replacement by disrupted genes (10). After a single round of transformation (11), the resultant strains were tested by PCR (12) and Southern (DNA) analysis (9). The mutant forms of the MTLa1 gene were readily detected (Fig. 3), but the wild-type gene was absent. This result indicates that the MTLa1 gene is present in only a single (haploid) copy in the C. albicans genome and contrasts with the case for many other genes in C. albicans, where two rounds of disruption have been necessary (one for each copy) to destroy a gene. The same approach was used to make a complete disruption of the C. albicans MTLα2 gene. After a single round of transformation, the disrupted allele, but not the naturally occurring MTLα2 gene, was detected by PCR (9). From these results, we conclude that both MTLa1 andMTLα2 exist in only a single copy in the C. albicansgenome. These results also indicate the absence of silent mating-type “cassettes” in the C. albicans genome because PCR primers to the MTLa1 (Fig. 3) andMTLα2 (9) ORFs do not detect these genes in the disrupted strains. These data provide strong support for the idea that the MTL locus of C. albicans is heterozygous, whereas most of the C. albicans genome is homozygous. Because the two MTL loci are each embedded in nearly identical DNA sequences, the simplest interpretation is that theMTL loci reside on homologous chromosomes as they do inS. cerevisiae.

Figure 3

Products of PCR reactions showing deletion of the C. albicans MTLa1 gene and the absence of genes fora1 at other loci. (A) Schematic of the genomic locus. Labeled arrows indicate locations of primers used in the PCR reactions in (B). (B) Agarose gels showing fragments produced in PCR reactions with indicated primer sets on genomic DNA isolated from different strains. Lanes: WT, genomic DNA from wild-type cells as the template; a1 KO, genomic DNA from the complete MTLa1 deletion strain as template; a1 HD KO, genomic DNA from the MTLa1 homeodomain deletion strain as template. Predicted sizes of the PCR products are indicated on the left (in kilobases). All primer sets yielded the appropriately sized products or no product as predicted. Lane 1, DNA marker fragments; lanes 2 to 4, primers A and B on indicated genomic DNA; lanes 5 to 7, primers A and hisG2 on indicated genomic DNA; lanes 8 to 10, primers B and hisG1 on indicated genomic DNA; lane 11, DNA marker fragments; lanes 12 to 14, primers C and D on indicated genomic DNA; lanes 15 to 17, primers E and F on indicated genomic DNA.

One hypothesis for the absence of a sexual cycle in C. albicans is that C. albicans was originally ana/α cell (to use the S. cerevisiaenomenclature) and through recombination lost one allele of the mating locus, becoming an a/a or α/α cell unable to undergo meiosis and return to a haploid state (13). The discovery of two MTL gene segments inC. albicans, one that resembles MATa and one that resembles MATα, appears to rule out this simple idea. The identification of MTLa and MTLα also suggests that the a/α configuration of the C. albicans MAT-like locus is a stable one. Because homologous recombination by way of the sequences flanking the locus could in principle result in the loss of either MTLa orMTLα, it seems likely that some sort of recombinational suppression exists. For example, one or more of the genes in each locus could be essential for cell viability, or the region could be under some type of mechanistic recombinational suppression.

In S. cerevisiae, the a1 and α2 proteins form a heterodimer that binds to specific DNA sequences (the haploid-specific gene, or hsg, operators) and represses transcription of the haploid-specific genes (2, 14,15). To see whether C. albicans has ana1/α2 repression activity and whether it is dependent on the genes of the MTL locus, we used two different hsg operators: One is the consensus hsg operator from S. cerevisiae, and the other is an hsg-like sequence found upstream of the C. albicans CAG1 gene, which encodes the α subunit of a trimeric GTP-binding protein. Although the function of this hsg-like sequence in C. albicans is not known, the sequence is recognized by S. cerevisiae a1/α2 (16). Five GFP (green fluorescent protein) reporter constructs were analyzed in C. albicanscells (17). In each construct, theGFP gene was placed under transcriptional control of theC. albicans ADH1 promoter, and in addition contained (i) no insert, (ii) three S. cerevisiae hsg operator consensus sequences, (iii) three S. cerevisiae hsg operators mutated in two nucleotide positions to prevent recognition bya1/α2 (18), (iv) three copies of the hsg operator-like sequences found upstream of the C. albicans CAG1 gene, or (v) three copies of the C. albicans CAG1sequence mutated in a way predicted to destroy recognition bya1/α2. The operators were inserted in the ADH1upstream region 260 base pairs upstream of the GFP gene.

Candida albicans cells containing the construct that lacks the operator expressed GFP, as evidenced by their bright green fluorescence (Fig. 4). The introduction of the S. cerevisiae hsg operators into the promoter markedly decreased fluorescence, whereas the mutant S. cerevisiae hsg operators showed no significant difference from the control construct that lacks the operators. The presence of the hsg-like sequence from CAG1 also significantly repressedGFP production, and the point-mutations introduced into it relieved this repression.

Figure 4

Identification of ana1/α2 transcriptional repression activity in C. albicans mediated by the hsg operator and the MTLa1gene. Transcriptional repression activity was determined as a measure of GFP fluorescence in a heterologous reporter system in the presence of test DNA binding sites as inserts. In each case, the reporter consists of the C. albicans ADH1 promoter controlling GFP expression. Reporter (1) no insert, (2) three S. cerevisiae hsg operator consensus sequences, (3) three S. cerevisiae hsg operators mutated at two nucleotide positions to prevent recognition by a1/α2 (18), (4) three copies of the hsg operator-like sequences located upstream of the C. albicans CAG1 gene, and (5) three copies of theC. albicans CAG1 sequence mutated in a way predicted to destroy recognition by a1/α2. Repression of the reporter is observed only when the reporter contains an intact hsg operator and when the MTLa1 gene is present.

To determine whether the C. albicans MTL gene cluster was required for the a1/α2-like repression activity, theGFP reporters were transformed into MTLa1deletion strains and evaluated for fluorescence. In contrast to the wild-type C. albicans strains, the MTLa1 mutant strains showed the same levels of fluorescence for all of the reporter constructs, indicating that the MTLa1 gene is required for the transcriptional repression activity (Fig. 4). Similar behavior was seen for both the complete deletion of the MTLa1 gene and for the MTLa1 homeodomain deletion, consistent with the DNA-binding domain of a1 being required for the repression activity (9). Northern (RNA) analysis also showed that transcription from the reporter constructs containing the functional hsg operators was derepressed in the MTLa1deletion mutants compared with the wild-type strain; however, in the absence of a1, the functional hsg operators still showed a slight amount of repression when compared with the mutated hsg operators (9). This repression could be due to the effect of the α2 alone or to a different C. albicans activity that has some overlapping function with a1/α2. Taken together, these results show that C. albicans has ana1/α2-like transcriptional repression activity and that this activity is dependent on the C. albicans MTLa1 gene. We think it likely that MTLa and MTLα together encode the a1/α2 activity. In S. cerevisiae,a1 and α2 mediate transcriptional repression by bringing the corepressor Tup1 to DNA, and the a1/α2 repressor activity observed in C. albicans is at least partially dependent on the C. albicans Tup1 protein (9).

In S. cerevisiae, genes regulated by the products of the MAT locus encode proteins involved in many aspects of the sexual cycle. C. albicans contains close relatives of many of these genes, including those involved in S. cerevisiae mating [for example, GPA1(16), STE20 (19),STE6 (20)] and meiosis [for example, DMC1(13)]. Although some of the C. albicansrelatives of the S. cerevisiae sexual cycle genes control filamentous growth, the complete functions of most of these genes have not been determined. These observations raise the question of whether any of these C. albicans genes are regulated by theMTL locus. We showed above that a DNA sequence found upstream of the C. albicans CAG1 gene (GPA1homolog) can bring about MTL-dependent transcriptional repression when placed into a test promoter. In addition, we know that expression of the endogenous CAG1 gene is indeed regulated by the MTL locus (9). Thus, like the analogous situation in S. cerevisiae, C. albicans CAG1 is a natural target of the products of the MTL locus. These results have prompted us to reexamine the possibility of a sexual cycle in C. albicans. In preliminary experiments, mating inC. albicans has not been observed; however, a more extensive analysis is now underway to construct strains with the appropriate configurations of the MTL locus and to screen them under numerous environmental conditions, including those conducive to mating in a variety of fungi.

In addition to the three transcriptional regulatorsa1, α1, and α2, six other ORFs were identified in theMTL locus: three in MTLa and three inMTLα (Fig. 1). These additional ORFs are arranged in pairs, one member of which is in MTLa and the other inMTLα. One pair of the ORFs is similar to the S. cerevisiae PIK1 gene, a phosphatidylinositol kinase (PIK); the second pair of ORFs is similar to theS. cerevisiae YKR003W, a member of a class of genes similar to the human oxysterol binding protein (OSBP) gene, referred to here asOBP; and the remaining pair of ORFs is similar to S. cerevisiae PAP1, a poly(A) polymerase (PAP) (21). The proteins encoded by the related ORF pairs are ∼60% identical to one another. This level of divergence between the members of each gene pair suggests that, although clearly related, they may have subtly different functions from each another. MATloci have been characterized in several fungi, and to date, the proteins encoded by them fall into the categories of sequence-specific DNA binding proteins, pheromones, and pheromone receptors (22–26). A few other MATgenes (Schizosaccharomyces pombe Mm, and Ustilago maydis LGA2 and RGA2) do not closely resemble any known genes, and their functions are as yet unknown. In contrast, six of the nine genes in the C. albicans MTL locus code for types of proteins not found in any of the previously described MATloci.

The ORFs for all nine genes of the C. albicans MTLlocus appear intact, suggesting that they code for functional proteins. We know this is the case for C. albicans a1 because it can mediate transcriptional repression from a test promoter bearing an hsg operator. The clustering of all nine genes into the C. albicans MTL locus suggests, by analogy with other fungi, that all of the genes may be involved in a single biological process.These genes could function to regulate a sexual cycle in C. albicans that has remained hidden from investigators, or they could be sexual cycle components derived from an evolutionary ancestor but now used to regulate another cellular process.

  • * To whom correspondence should be addressed. E-mail: ajohnson{at}


Stay Connected to Science

Navigate This Article