Evidence for Mating of the "Asexual" Yeast Candida albicans in a Mammalian Host

See allHide authors and affiliations

Science  14 Jul 2000:
Vol. 289, Issue 5477, pp. 307-310
DOI: 10.1126/science.289.5477.307


Since its classification nearly 80 years ago, the human pathogenCandida albicans has been designated as an asexual yeast. In this report, we describe the construction of C. albicansstrains that were subtly altered at the mating-type–like (MTL) locus, a cluster of genes that resembles the mating-type loci of other fungi. These derivatives were capable of mating after inoculation into a mammalian host. C. albicansis a diploid organism, but most of the mating products isolated from a mouse host were tetrasomic for the two chromosomes that could be rigorously monitored and, overall, exhibited substantially higher than 2n DNA content. These observations demonstrated that C. albicans can recombine sexually.

The yeast Candida albicans is found as a commensal organism in the digestive tract of mammals. It is also the most common human fungal pathogen, causing both mucosal and systemic infections, particularly in immunocompromised people (1). C. albicans, a diploid yeast, has been classified as asexual because no direct observation of mating or meiosis has been reported. Population studies of C. albicansindicate that some low level genetic exchange may occur (2), but there is no conclusive evidence for recent sexual recombination (3–5). The apparent absence of sexual reproduction inC. albicans is especially intriguing because its relatives in the budding yeast family (e.g., Saccharomyces cerevisiaeand Kluyveromyces lactis) have retained sexual cycles. In fact, sexual reproduction is common throughout the fungal kingdom, ranging from yeasts to mushrooms.

Although mating and meiosis have not been observed for C. albicans, homologs of genes that function in both mating [e.g.,GPA1 (6), STE20(7), and STE6 (8)] and meiosis [e.g., DMC1 (9)] in S. cerevisiae have been identified in C. albicans. These genes are intact and code for well-conserved proteins, and several have been shown to complement S. cerevisiae mutants. The full range of the functions of these gene products in C. albicanshas not been determined, but their presence suggests that C. albicans has a cryptic sexual cycle or that C. albicanslacks a sexual cycle and these conserved gene products have been coopted for different purposes.

Sexual reproduction in fungi is typically controlled by genes that reside in a specified genetic locus called a mating-type, orMAT, locus. Two features distinguish this locus from other portions of the genome. First, in diploid cells, the MATlocus is usually heterozygous; it codes for different (but usually related) genes on each of the two homologous chromosomes. Second, the genes encoded by the MAT locus of different fungi generally fall into three specific categories: DNA binding proteins that regulate the expression of sexual cycle genes, structural genes that code for mating pheromones, and structural genes that code for mating pheromone receptors (10–13).

The genes in the MAT locus of the sexually reproducing yeastS. cerevisiae have been well characterized (11,14). They encode three transcriptional regulatory proteins that, together with other proteins encoded elsewhere in the genome, control the transcription of many target genes. The patterns of expression of these target genes give rise to three distinct types of cells (a, α, and a/α) that are responsible for the S. cerevisiae sexual cycle. The two types of mating cells (a and α) are typically haploid and carry different genetic information at the MAT locus: a cells carry MATa, and α cells carry MATα. MATacodes for a single regulatory protein (the homeodomain protein a1), andMATα codes for two proteins (the homeodomain protein α2 and the alpha domain protein α1). Mating between an a cell and an α cell forms the a/α cell, which is usually diploid, carries both MATa and MATα, and is capable of undergoing meiosis and spore formation.

We recently described a mating-type–like (MTL) locus inC. albicans (Fig. 1) that resembles the mating-type (MAT) locus of the sexually reproducing yeast S. cerevisiae in two important respects: First, it encodes proteins similar to the three transcriptional regulators encoded by MAT (a1, α1, and α2), and second, it is heterozygous in the diploid laboratory strain of C. albicans that we analyzed (SC5314), with an a1-like gene carried on one chromosome and α1- and α2-like genes carried on the other. Other similarities in the arrangement of the genes, the positions of introns, and the function of the a1 protein (it is a transcriptional repressor in both organisms) further support a close connection between the MTL locus of C. albicansand the MAT locus of S. cerevisiae(15).

Figure 1

Features of the C. albicansmating-type–like (MTL) locus (15). The C. albicans MTL locus contains open reading frames for nine proteins from four families of proteins: three gene regulatory proteins (a1, α1, and α2), two phosphatidylinositol kinases (PIK), two oxysterol-binding proteinlike proteins (OBP), and two poly(A) polymerases (PAP). The region of DNA sequence that differs between the MTLa and MTLα segments is 8742 base pairs (bp) for MTLa and 8861 bp forMTLα.

Like its counterpart in S. cerevisiae, theMTL locus could regulate a sexual cycle in C. albicans. If the analogy between the MTL locus inC. albicans and the MAT locus of S. cerevisiae holds, then the SC5314 laboratory strain of C. albicans would be an a/α strain and would not be expected to mate. To test the hypothesis that C. albicanshas the inherent capacity to mate, we genetically altered an SC5314 derivative (CAI4) to create two types of “a” strains and two types of “α” strains. For the astrains, either the entire MTLα locus was deleted from CAI4 (to give an MTLa/mtlαΔ strain) or only theα1 and α2 genes were deleted (to give anMTLa/mtlα1mtlα2 strain). Likewise, α strains were constructed either by deleting the entire MTLa locus (mtlaΔ/MTLα) or by deleting only thea1 gene (mtla1/MTLα) (15,16). In addition, ade2/ade2(AdeUra+) and ura3/ura3(Ade+Ura) derivatives of these strains were constructed to allow successful mating events to be detected by selecting for cells with the ability to grow on media lacking adenine and uracil (Ade+Ura+ prototrophs) (16).

To test for mating, we mixed various strains as outlined in Fig. 2A. In addition to mixtures that contained putative mating pairs (an a strain and an α strain), we also included pairs that would not be expected to mate, for example, an MTLa/MTLα ura3/ura3 and an MTLa/MTLα ade2/ade2 (mix 1, Fig. 2A). Because mammals are natural hosts for C. albicans, we introduced the combinations of strains into mice through tail vein injection (17). Pilot experiments had revealed that, after injection, theade2/ade2 strains persisted in mice at substantially lower levels than did the ura3/ura3 strains (18). To compensate for this difference, we injected three times as manyade2/ade2 cells as ura3/ura3 cells for each test mating. Twenty-four hours after injection, the mice were euthanized, the kidneys were removed and homogenized, and the mixture was plated on various media. For all the test matings, about 103 C. albicans colonies per kidney formed on rich medium, conditions under which all the starting strains readily grow. On medium lacking adenine and uracil, we recovered 44 colonies from the a + α mix (MTLa/mtlα1mtlα2 + mtla1/MTLα) (mix 2 of Fig. 2A), one colony from the mixture containing the aand α locus deletion strains (mtlaΔ +mtlαΔ) (mix 3 of Fig. 2A), and no colonies from the control mix (mix 1 of Fig. 2A). Additional test matings revealed similar results (19). Other experiments showed that prototrophs did not appear when comparable titers of the starting strains were simply mixed together and incubated on various types of laboratory media and then plated to selective medium (20).

Figure 2

(A) MTLconfigurations of strains used in test mating experiments. Strains containing different deletions at the MTLlocus were mixed together and tested for mating in a mouse tail vein model. The MTLa and MTLα configurations for the different mutants are represented schematically for each of three different injection mixes labeled 1 though 3. EachMTL configuration has been labeled “a” or “α” to indicate the analogous behavior of a similarly modified diploid in S. cerevisiae. Mix 1, intact MTLAde + intact MTL Ura; mix 2,mtla1 deletion Ade +mtlα1mtlα2 deletion Ura; mix 3,mtlaΔ Ade, mtlαΔ Ade + mtlαΔ Ura,mtlaΔ Ade (19). (B) FACS analysis of prototrophs. Strains were analyzed by fluorescence-activated cell scan for DNA fluorescence. Thex axis of each graph (Sytox) represents a logarithmic scale of fluorescence, and the y axis (counts) represents a linear scale of cell number. In each case, the control strain (CAI4) is in black, and the test strain is in gray. The first graph [a/α (Ade)] shows an overlay of an Ade MTLa/MTLα strain over the Ura MTLa/MTLα strain (CAI4). The mtla1 andmtlα1α2 starting strains gave similar profiles (18). The graphs labeled 3-4 and 3-6 show profiles for two of eighteen prototrophs tested and represent the majority of the FACS profiles, showing an increase in fluorescence that is consistent with an increase in DNA content (21).

After the Ade+Ura+ phenotype of the prototrophs was confirmed by purifying for single colonies on selective medium, they were subjected to several tests. In a representative sample, treatment with serum elicited rapid germ tube formation, indicating that the prototrophs were indeed C. albicans(18). Fluorescence-activated cell sorting (FACS) analysis suggested that cells from most of the colonies tested (12 of 15) had substantially increased DNA content compared with that of the starting strains (Fig. 2B) (21). DAPI (4′,6′-diamidino-2-phenylindole) staining showed the presence of a single nucleus in the cells of all of the prototrophs (21).

We next determined whether the prototrophs carried genetic markers derived from both parents. For theMTLa/mtlα1mtlα2 × mtla1/MTLαcross (mix 2 of Fig. 2A), all four alleles of MTL were distinguishable, two from each starting strain (22). Southern analysis (23) of 16 of the prototrophs recovered from this experiment revealed that 12 strains carried all four alleles of MTL [MTLa/mtla1/MTLα/mtlα1mtlα2(Fig. 3, examples shown in lanes 10 through 16)] and four strains carried three alleles ofMTL (Fig. 3, example shown in lane 9). These four strains represent three of the four possible MTL locus combinations (MTLa/mtla/MTLα, MTLa/mtla/mtlα, andMTLa/MTLα/mtlα). The starting strains each carried only one version of MTLa and one version ofMTLα (Fig. 3, lanes 7 and 8); hence, all the prototrophs contained genetic information from both of the parent strains, and we will hereafter refer to them as conjugants. “Wild” strains ofCandida could not have confounded this analysis because two of the MTL alleles in this experiment (one from each starting strain) were created for use in this experiment and do not exist outside the laboratory.

Figure 3

Southern blot confirming the presence of three or more alleles of MTL in the conjugant strains.C. albicans genomic DNA was isolated from conjugants and subjected to restriction enzyme digestion before electrophoresis. Lane 1, marker lane; lane 2,MTLa/MTLα ura3/ura3 auxotroph; lane 3,MTLa/MTLα ade2/ade2 auxotroph; lane 4,mtlaΔ/MTLα ade2/ade2auxotroph; lane 5, MTLa/mtlαΔura3/ura3 2; lane 6, prototrophic conjugant from mating test 3 ofFig. 2A; lane 7, mtla1/MTLα ade2/ade2 auxotroph; lane 8,MTLa/mtlα1mtlα2 ura3/ura3 auxotroph; lane 9, prototrophic conjugant from mating test mix 2 of Fig. 2A showing the presence of three configurations of MTL; lanes 10 to 16, prototrophic conjugants from mating test 2 of Fig. 2A showing the presence of four configurations of MTL. See (25) for details.

The strain recovered from the MTL locus deletion experiment (mix 3 of Fig. 2A) was also analyzed and found to carryMTLa, MTLα, and at least one mtldeletion. (DNA restriction fragments for the mtladeletion and the mtlα deletion are identical in size and could not be distinguished in this experiment.) Again, because none of the starting strains in this test mix carried both MTLa andMTLα (Fig. 3, lanes 4 and 5). We can conclude that the conjugant carried genetic information from each parent. Moreover, although there was a mixture of several different strains injected in this experiment (mix 3 of Fig. 2A), the conjugant must have formed from one a parent and one α parent.

The same type of marker analysis was performed on the conjugants for the ADE2 locus, which is located on a different chromosome from that of MTL(24). Of the four ADE2 alleles that entered the host, three can be unambiguously identified: The twoade2 disruptions can be distinguished from each other and from the intact ADE2 alleles, but the two intactADE2 alleles are indistinguishable (22). Southern analysis indicated that all of the conjugant strains carried all three of the distinguishable alleles of ADE2 (examples shown inFig. 4); moreover, the normalized ratio of the signal of the intact ADE2 gene to that of either disruptant was about 2:1 in at least four of the conjugants (3-4, 3-6, 3-1, and 3-3) (18), consistent with the idea that these conjugants carried two copies of the intact ADE2 gene and one copy each of the two disrupted alleles. Because all the strains in this experiment have the same type of disruption at theura3 locus, the analysis of this locus is not informative; the Ura+ parental strains have URA3 integrated at the ade2 locus.

Figure 4

Southern blot confirming the presence of at least three alleles of ADE2 in the conjugant strains.C. albicans genomic DNA was isolated from conjugants and subjected to restriction enzyme digestion before electrophoresis. Lane 1, mtla1/MTLα ade2/ade2 auxotroph; lane 2,mtlaΔ/MTLα ura3/ura3 auxotroph; lanes 3 to 16, prototrophic conjugants from mating test mix 2 of Fig. 2A showing the presence of at least three configurations ofADE2. See (25) for details.

In summary, all the prototrophs we recovered from the “a” × “α” crosses contain genetic markers from both parental strains. For the MTLa/mtlα1mtlα2 ×mtla1/MTLα cross (mix 2, Fig. 2A), 12 of 16 prototrophs tested contain all four alleles of the MTL locus, two derived from each parent. The simplest interpretation of this result is that the two diploid parental strains mated to form a tetraploid cell. The four prototrophs recovered from this cross that lacked oneMTL allele probably arose through chromosome loss or homozygosis of the MTL locus after mating, perhaps during the passaging of strains on laboratory media required for their analysis. Although not as definitive as the analysis of the MTLlocus, the analysis of the ADE2 locus is consistent with the idea that the prototrophs arose from the formation of a tetraploid strain from the two diploid parents. To date, mating has been detected in six independent crosses; in all cases, mating was observed only between an “a” and an “α” strain. Control crosses (“a” × “a,” “α” × “α,” and “a/α” × various strains) did not produce Ade+Ura+ prototrophs in the same experiments.

Although we cannot rigorously conclude that successful conjugation occurs only when an “a” strain is crossed with an “α” strain, this is the only combination we have observed to date, and the results correlate well with this idea. This strong correlation supports the idea that the observed conjugation arose from bona fide mating and not from nonspecific cell and nuclear fusion events, the latter would be expected to occur between any two strains irrespective of the MTL configuration. C. albicans appears inherently able to mate and raises the question of why this appears to happen so rarely in nature. Assuming that theMTLa/MTLα configuration of MTL (as found in SC5314) is the prevalent wild form in C. albicans, it is possible that mating requires “a” and “α” strains to arise by homozygosis of the MTL locus or by chromosome loss. These derivatives may be lost quickly from natural populations, and without laboratory intervention, the appropriate pairs ofa and α cells may arise in the same host only rarely.

These data raise the possibility that C. albicans has a complete sexual cycle, perhaps one in which diploids can mate to form tetraploids that can undergo meiosis to produce diploids. Experiments are under way to test for meiosis in tetraploid strains of C. albicans. It is also possible that after mating, chromosomes could be lost from tetraploids, gradually reducing them to the diploid state. The recovery of conjugants that appear trisomic for theMTL-containing chromosome locus provides some support for this notion. Finally, the fact that mating was observed in mice but not on a variety of laboratory media suggests the possibility that a signal or condition conducive to mating is provided by the mammalian host.

  • * These authors contributed equally to this work.

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


View Abstract

Navigate This Article