We identified a gene of the fungal pathogen Candida albicans, designated EFG1, whose high‐level expression stimulates pseudohyphal morphogenesis in the yeast Saccharomyces cerevisiae. In a central region the deduced Efg1 protein is highly homologous to the StuA and Phd1/Sok2 proteins that regulate morphogenesis of Aspergillus nidulans and S.cerevisiae, respectively. The core of the conserved region is homologous to the basic helix–loop–helix (bHLH) motif of eukaryotic transcription factors, specifically to the human Myc and Max proteins. Fungal‐specific residues in the bHLH domain include the substitution of an invariant glutamate, responsible for target (E‐box) specificity, by a threonine residue. During hyphal induction EFG1 transcript levels decline to low levels; downregulation is effected at the level of transcriptional initiation as shown by a EFG1 promoter–LAC4 fusion. A strain carrying one disrupted EFG1 allele and one EFG1 allele under the control of the glucose‐repressible PCK1 promoter forms rod‐like, pseudohyphal cells, but is unable to form true hyphae on glucose‐containing media. Overexpression of EFG1 in C.albicans leads to enhanced filamentous growth in the form of extended pseudohyphae in liquid and on solid media. The results suggest that Efg1p has a dual role as a transcriptional activator and repressor, whose balanced activity is essential for yeast, pseudohyphal and hyphal morphogenesis of C.albicans. Functional analogies between Efg1p and Myc are discussed.
Many fungal pathogens alternatively grow in a budding yeast form or in a hyphal form, an ability contributing to different aspects of fungal virulence. In the most common human fungal pathogen, Candida albicans, the hyphal growth form appears most suited for adherence and penetration of epithelial or endothelial cell layers (reviewed by Calderone and Braun, 1991; Cutler, 1991); also, hyphal growth may allow exit of ingested C.albicans from macrophages (Arai et al., 1977). In contrast, the unicellular yeast growth form (blastoconidium) is especially suited to increase the number of infectious units and may readily achieve systemic distribution in the infected patients. In Histoplasma capsulatum, the yeast form is required for propagation in human tissues, while hyphal growth occurs outside infected organisms (Maresca and Kobayashi, 1989). The yeast form of Ustilago maydis is not pathogenic, while the dikaryotic hyphal form is able to infect plants (Schulz et al., 1990).
Environmental conditions determine the growth mode of C.albicans. While yeast growth is favored at low temperatures and high glucose concentrations, a pseudohyphal form characterized by elongated polar buds that are not separated following cell division is observed at low temperatures on depauperated media; unconstricted ‘true’ hyphae, which in their initial phase of formation are termed germ tubes and grow by continuous apical extension, are induced by a variety of chemical stimuli at high temperature (reviewed by Odds, 1988). Pseudohyphal growth has also been observed in the nonpathogenic yeast Saccharomyces cerevisiae (Gimeno et al., 1992). In this case pseudohyphal growth is induced by starvation for nitrogen on solid media, where it may serve to forage for nitrogen. In contrast, formation of true hyphae in C.albicans occurs in liquid and on solid media and requires positive induction signals, such as serum.
In S.cerevisiae several genes required for the yeast–pseudohyphal dimorphism are known. Analyses of mutants revealed that members of the mitogen‐activated protein kinase (MAP kinase) pathway leading to the activation of the transcription factor Ste12p, as well as the Tec1 protein, are required for pseudohyphal development; overproduction of proteins of the MAP kinase pathway, of Ste12p or Tec1p leads to enhanced pseudohyphal growth (Liu et al., 1993; Gavrias et al., 1996; Mösch et al., 1996). Although the genetic basis of the yeast–hyphal dimorphism in C.albicans has not been established, a Ste12p homolog (Cph1p) recently has been shown to be involved in hyphal growth of C.albicans during some conditions (Liu et al., 1994). Besides involvement of a MAP kinase pathway the role of phosphorylation in morphogenesis was shown by enhanced pseudohyphal formation in elm1 and cdc55 mutants, in which a protein kinase or a subunit of protein phosphatase 2A, respectively, are altered (Blacketer et al., 1993). Saccharomyces cerevisiae pseudohyphal morphogenesis can also be triggered by activation of the Ras2p signal transduction pathway (Gimeno et al., 1992; Mösch et al., 1996). Overexpression of the PHD1 gene also stimulates pseudohyphal growth (Gimeno et al., 1994). The deduced Phd1p sequence contains a segment that is highly homologous to the StuA protein of Aspergillus nidulans, which is required for conidiophore morphogenesis (Miller et al., 1992). This region has also limited homology to the DNA binding motifs of Swi4p and Mbp1p. Recently, the SOK2 gene encoding a second S.cerevisiae protein containing a Phd1p/StuA homology domain was identified (Ward et al., 1996). In contrast to Phd1p, Sok2p appears to inhibit the switch from unicellular to filamentous growth by acting as a repressor of transcription.
In the present study we describe a C.albicans gene, EFG1, which encodes another protein containing a conserved domain as in the StuA, Phd1 and Sok2 proteins. We discovered that the region of homology in all of these proteins encompasses a sequence with features of a basic helix–loop–helix (bHLH) DNA binding/dimerization domain of certain transcription factors in yeast and mammalian cells (Ferré‐D‘Amare et al., 1993; Ellenberger, 1994; Ma et al., 1994). Results of gene disruption and overexpression experiments and of transcript analyses suggest that the function of EFG1 in C.albicans is the regulation of yeast and hyphal morphogenesis. Thus, StuA, Phd1, Sok2 and Efg1 (designated ‘SPSE proteins’) appear to represent a novel class of transcriptional regulators that are involved in cellular differentiation processes in fungi.
Isolation of the C.albicans EFG1 gene
Certain S.cerevisiae diploid strains are able to respond to nitrogen limitation by growth in a filamentous pseudohyphal form (Gimeno et al., 1992). We attempted to clone genes of C.albicans involved in hyphal or pseudohyphal morphogenesis by transforming a S.cerevisiae diploid with genomic C.albicans libraries in multicopy S.cerevisiae vectors and screening for transformants with accelerated pseudohyphal morphogenesis. Among transformants of each of two genomic libraries one such transformant was identified. By isolation and retransformation of the plasmids carried by the respective transformants (p606/1 and p607/2), we confirmed that the pseudohyphal phenotype was plasmid‐borne (Figure 1B and H).
Restriction mapping of the inserts in plasmids p606/1 and p607/2 revealed that they represented overlapping segments of the same C.albicans genomic region. Subclones of this region were constructed in multicopy S.cerevisiae vectors, transformed in S.cerevisiae and tested for enhanced formation of pseudohyphae. A 1.6 kb fragment was shown to be minimally required for this phenotype. To test dependence on high plasmid copy numbers we cloned the 5.2 kb BamHI insert into the centromeric vector YCplac22. Transformants carrying the resulting vector did not show enhanced pseudohyphal morphogenesis indicating that high expression of the putative C.albicans gene is required to stimulate pseudohyphal growth in S.cerevisiae.
At low nitrogen concentrations the transformant phenotype was similar to the phenotype of a YCpR2V transformant expressing the Ras2Val19 protein (Gimeno et al., 1992); however, at 600 μM ammonium sulfate 60% of colonies of the YCpR2V‐transformant showed pseudohyphae compared with only 20% of the p606/1 transformants. To test synergy between the C.albicans gene product and the Ras2Val19 protein S.cerevisiae transformants cotransformed with p606/1 and YCpR2V were analyzed. Such transformants did not show enhanced morphogenesis compared with single transformants indicating a lack of synergy. On the other hand a synergistic effect on morphogenesis was obtained in double‐transformants carrying p606/1 and pCG38, which encodes a high‐copy inducer of pseudohyphal growth, Phd1p (Gimeno and Fink, 1994) (Figure 1, compare B and D). To test further whether the pseudohyphal formation elicited by the identified C.albicans gene required the activity of the MAP kinase module (Mösch et al., 1996) we transformed p607/2 into the ste20 mutant HLY492 (Liu et al., 1993). Pseudohyphal formation was completely blocked in this mutant, while a transformant of a congenic wild‐type strain (CGX31) showed enhanced formation of pseudohyphae (Figure 1G and H). Likewise, pseudohyphal formation elicited by PHD1 overexpression (pCG38), or the activity of the RAS2Val19 allele (YCpR2V), was completely blocked in the ste20 mutant. Thus, the dependence on low‐nitrogen conditions, the lack of synergy with Ras2p and its dependence on STE20 suggested that the heterologous C.albicans gene product functions in the known morphogenetic pathways of S.cerevisiae (Gimeno et al., 1992; Mösch et al., 1996). Synergy between the C.albicans gene product and the homologous Phd1 protein (see below) can be explained by their identical, dosage‐dependent functions in stimulating morphogenesis.
Sequence of the EFG1 gene
The nucleotide sequence of the 2.517 kb EcoRI–SphI fragment of p606/1 was determined (DDBJ/EMBL/GenBank accession number Z32687; data not shown). Sequence analyses revealed a contiguous open reading frame between nucleotide positions 211–1869. The deduced protein contains 552 amino acids (molecular mass 59.9 kDa) and has an unusual composition consisting of 19% glutamine, 10% alanine and 9% each of threonine, serine, proline and glycine residues; the single CUG codon 452 was translated as serine, instead of the standard leucine (Leuker and Ernst, 1994; Santos and Tuite, 1995). Glutamine residues are especially clustered in the amino and carboxyl terminal regions of the deduced protein. Several sequence motifs may have functional importance: (i) a possible RR..RGRR nuclear localization signal at residues 229–230/249–252; (ii) a protein kinase A (PKA) phosphorylation site at residues 204–206 (RVT); (iii) a possible casein kinase II phosphorylation site at residues 208–211 (TMWE) and numerous target sites for casein kinase I of the structure S/T‐XX‐S/T.
Computer analyses revealed that the central region of the deduced protein (amino acids 210–350) contains sequences able to form α‐helices, while random coils are predicted for the termini. Within its central region, encompassing amino acids 202–303, the deduced protein has high homology to the StuA protein of A.nidulans (Miller et al., 1992), as well as the Phd1 and Sok2 proteins of S.cerevisiae (Gimeno and Fink, 1994; Ward et al., 1996) (80% conserved amino acids in all four proteins) (Figure 2). Because of its induction of filamentous growth in C.albicans (see below) we designated the encoding gene EFG1 (enhanced filamentous growth) and we refer to the conserved group of fungal proteins (StuAp, Phd1p, Sok2p, Efg1p) as ‘SPSE’ proteins. EFG1 is represented by a single genomic locus on chromosome R of C.albicans (B.B.Magee, personal communication; Internet:: http://alces.med.umn.edu/bin/geneinfo).
Northern analyses revealed that the EFG1 transcript has a length of ∼2.8 kb and nuclease S1‐mapping of the 5′‐end of the EFG1 transcript revealed a major transcript start point at position −35 relative to the ATG of the coding region (data not shown). The transcript 5′‐end (GTTCT) differs from the TCA/GA and RRYRR sequences typical for S.cerevisae transcripts (Hahn et al., 1985). The unexpected size of the EFG1 transcript relative to the 1.65 kb of the EFG1 coding region may be due to additional transcript start sites far upstream of the −35 start site or to extensive 3′‐untranslated or poly(A) sequences (Jackson and Standar, 1990). A putative TATA box is present at positions −82 to −85 upstream of the −35 transcript start site, which is followed by a T/C rich sequence. The sequence TAG…TAGTAT….TTC in the 3′‐untranslated region, between residues 1987–2031, agrees with the tripartite consensus sequence for transcription termination in S.cerevisiae (Zaret and Sherman, 1982). Further downstream in the 3′‐untranslated region sequences homologous to ARS1 consensus sequences of S.cerevisae (Newlon and Theis, 1993) are detected. Vectors carrying the consensus sequences on a 735 bp NsiI fragment in an integrating vector (p1367/1; Losberger and Ernst, 1989) indeed replicate episomally in C.albicans and S.cerevisiae (our unpublished results).
To confirm the deduced coding region we deleted sequences upstream of the predicted ATG initiation codon and the coding region was placed under the control of the GAL1 promoter in plasmid pBIST; transformants carrying this plasmid showed enhanced formation of pseudohyphae only on galactose‐, but not on glucose‐containing media (Figure 1E and F). The activity of subclones revealed that the predicted coding region can be partially deleted without loss of its pseudohyphal stimulating activity. Deletions shortening the deduced protein at its carboxyl terminal end by 50 amino acids were still active, while more extensive deletions were inactive (data not shown).
Efg1p as a bHLH protein
The central portion of the conserved region in SPSE proteins has significant sequence and structural homology to the bHLH motif in eukaryotic transcription factors including Max, MyoD and Pho4p (Ferré‐D‘Amare et al., 1993; Ellenberger, 1994; Ma et al., 1994) (Figure 2, bottom). Residues known to contact DNA in Max (Ferré‐D'Amare et al., 1993) and MyoD (Ma et al., 1994) are conserved in SPSE proteins. The presence of arginine and an apolar residue (Met) in position 252 and 247 (Efg1p numbering) is similar to Myc and Max, but unlike MyoD, suggesting that the two central positions in the target sequence are CG and not GC. The conserved glutamic acid residue in bHLH proteins (e.g. position 118 in MyoDp) is replaced by threonine in SPSE proteins (Efg1p position 248). The role of glutamic acid is to donate oxygen to establish hydrogen bonds to the CA residues of the target sequence, which is a function that may also be assumed by a threonine residue. Replacement of this glutamic acid residue by glutamine in the yeast Pho4 protein did not abolish DNA binding (Fisher and Goding, 1992). The conserved histidine residue in Myc/Max‐type proteins is replaced by valine in SPSE proteins (Efg1p position 244) and thus resembles MyoD (which contains alanine). The SPSE proteins (as MyoD) do not contain leucine zipper sequences; therefore, dimerization of Efg1p most likely occurs by the bHLH region. Like other eukaryotic transcription factors Efg1p contains stretches of glutamines (Tjian and Maniatis, 1994); in yeast such stretches occur in repressors (e.g. Ssn6p) as well as activators (e.g. Snf5p) (Winston and Carlson, 1992; Treitel and Carlson, 1995).
EFG1 transcript regulation
Candida albicans ATCC10261 was induced to form hyphae using three established protocols, during which the expression of housekeeping genes including ACT1 is not significantly affected (Delbrück and Ernst, 1993). Logarithmically growing cells were diluted in YP‐medium (P1), 5% horse serum/YPD‐medium (P2) or 5% horse serum/YP‐medium (P3) and incubated at 37°C. YPD‐medium, in which no hyphae are induced (Delbrück and Ernst, 1993), was used as a control. Hyphal formation proceeded similarly in conditions P1–P3: 5–10% (20 min), 35–40% (40 min) and 80–90% (P1 and P3) or 40% (P2) (60 min) of cells produced hyphae (germ tubes). Total RNA was isolated at different time points during incubation of the cultures and assayed for the presence of EFG1 mRNA by Northern blotting, using ACT1 mRNA as control (Delbrück and Ernst, 1993) (Figure 3).
A single EFG1 mRNA species of ∼2.8 kb RNA was detected, whose level declined significantly preceding hyphal morphogenesis induced by protocols P1–P3, while the level of the ACT1 mRNA remained constant (Figure 3). In control cells (C), the EFG1 transcript level was unchanged during incubation. Transcript downregulation was most pronounced in hyphal‐forming cells induced by protocol P1, which contained almost undetectable mRNA levels after 10 min of hyphal induction. In all induction protocols the decline of the EFG1 transcript level started prior to the emergence of visible hyphae. Thus, these studies reveal a clear correlation between the decline of the EFG1 transcript levels and subsequent hyphal formation.
To demonstrate that the downregulation of EFG1 transcript levels occurs on the level of transcriptional initiation, a fusion of the EFG1 promoter to the LAC4 reporter gene (Leuker et al., 1992) was constructed and expressed in C.albicans. Transformants were induced to form hyphae by the same conditions (P1–P3) that were used for EFG1 mRNA analysis (Figure 3) and β‐galactosidase (βgal) activity in cell extracts were determined at different time points. In this system, a rapid decline in βgal activity was observed for all cultures forming hyphae, while the activity in control cells remained constant (Figure 4). The extent of the initial decline varied from a factor of 500 (conditions P1 and P3) to a factor of about 5000 (P2). These results demonstrate that the EFG1 promoter activity is significantly downregulated prior to hyphal formation.
We were not able to obtain mutants in which both EFG1 alleles are disrupted by standard gene disruption protocols (Sadhu et al., 1992; Fonzi and Irwin, 1993), suggesting that EFG1 is essential for the viability of C.albicans. Therefore, we adopted a strategy to disrupt the coding region of only one allele and to replace the promoter region of the second allele with a regulatable promoter. To this end we first identified the C.albicans PCK1 gene encoding phosphoenolpyruvate carboxykinase; the PCK1 promoter was found to be active during growth on carbon sources such as succinate, ethanol, or sucrose, but to become effectively repressed by glucose (C.Leuker, A.Sonneborn and J.F.Ernst, manuscript in preparation). The first EFG1 allele was disrupted by transforming strain CAI8 (Ura3− Ade2−) with an EFG1 fragment, in which the EFG1 promoter was replaced by the PCK1 promoter and the URA3 gene as selectable marker (Figure 5, top). Uridine prototrophs, in which the desired disruption event was confirmed (see below), were again transformed with an EFG1 fragment, in which the ADE2 gene interrupts the EFG1 coding region and Ura+ Ade+ prototrophs were selected. In this manner two identical doubly disrupted strains designated SS4 and SS14 were obtained, in which one EFG1 allele is disrupted by ADE2 and the second allele is under control of the regulatable PCK1 promoter. SS4 and SS14 showed the identical phenotypes discussed below.
Disruption of both EFG1 alleles as shown in Figure 5 (top) was demonstrated by Southern blotting (Figure 5, bottom). Genomic DNA of transformants was digested using EcoRI or HindIII, which both cut outside the region present on the disruption fragments. In the wild‐type strain each EFG1 allele is present on EcoRI fragments of different lengths (6.0 kb and 7.0 kb). Insertion of the PCK1p/URA3 cassette increased the size of the 6 kb fragment to 7.4 kb, but left the 7 kb fragment unchanged (lane 2). Because of an EcoRI site within ADE2, its insertion into the second EFG1 allele removed the 7.4 kb fragment, but created a new 2 kb EcoRI fragment (lane 3). In HindIII digests a single 25 kb fragment was detected in the wild‐type (lane 6), while insertion of the PCK1p/URA3 cassette in one allele generated an additional fragment of 26.4 kb (lane 5). The second disruption step, because of a HindIII site in ADE2, led to disappearance of the remaining wild‐type copy and generation of a 2.9 kb fragment, as expected (lane 4).
EFG1 transcript levels in SS and control strains were analyzed (Figure 6). Cells were grown in conditions leading to PCK1 promotor repression (SD) or derepression/induction (B‐medium) into the exponential growth phase, total RNA was isolated and 20 μg of RNA was analyzed by Northern blotting; the ACT1 transcript was used as an internal control (Delbrück and Ernst, 1993). Control cells grown on SD contained moderate levels of an EFG1 transcript of ∼2.8 kb (Figure 6, lane 5), which increased during growth on B‐medium (lane 6) indicating that EFG1 expression is glucose‐regulated. The disruptant strain SS14 was lacking an EFG1 transcript as in control cells; instead, it contained an EFG1 transcript of ∼2.3 kb (Figure 6, lanes 7 and 8). The reduced transcript size in strain SS14 as compared with the control strain could be due to replacement of the authentic EFG1 promoter by the PCK1 promoter, which conceivably could reduce the sizes of the 5′ or 3′‐untranslated regions of the transcript. The transcript level in strain SS14 grown in SD, relative to the ACT1 control transcript, was clearly reduced as compared with control cells (Figure 6, compare lanes 5 and 7). As expected, growth on B‐medium led to significant overexpression of the PCK1p::EFG1 gene in strains CAI4[pRC2312P‐H] and SS14 (lanes 4 and 8).
The consequence of a reduced EFG1 transcript level on cell morphology was analyzed microscopically after growing strain SS14 in liquid SD‐medium, in which the PCK1 promoter is repressed (Figure 7A). Cells had a more elongated, rod‐like appearance as compared with control cells (Figure 7D) which was reminiscent of the opaque phenotype of the ‘switching’ strain WO‐1 (Soll et al., 1993); furthermore, such cells often remained attached leading to pseudohyphal filaments. The tendency to filamentation was also apparent on solid SD‐medium, where colonies formed lateral extensions (Figure 1J), similar to cph1 mutants in some conditions (Liu et al., 1994). Although glucose‐grown SS14 cells had a pseudohyphal phenotype, the addition of 5% horse serum did not induce the appearance of true unconstricted hyphae, as in control cells. Thus, a reduced EFG1 expression level has at least two effects, (i) by inducing cell elongation and pseudohyphal filaments and (ii) by prohibiting formation of true hyphae.
EFG1 overexpression could be achieved by the growth of disruptant strains SS4/14 on B‐medium (Figure 6, lane 6). An even greater overexpression could be achieved in transformants carrying multiple copies of plasmid pRC2312P‐H, which contains the PCK1p::EFG1 expression module (Figure 6, lane 4). Interestingly, overexpression of PCK1p::EFG1 was accompanied by disappearance of the authentic EFG1 transcript, which is predominant in glucose‐grown cells (Figure 6, lane 3). This result is an indication of autoregulation of EFG1, a finding that we recently confirmed using promoter fusions (unpublished results).
The majority of SS4 cells and pRC2312P‐H transformants grown in B‐medium had a pronounced tendency to form long filaments, consisting of extremely elongated cells (Figure 7B and C). Because of constrictions between cells and their growth starting with an apical bud we consider these filaments as pseudohyphae, but not true hyphae, which are formed by continuous apical extension (Odds, 1988). The pseudohyphal phenotype could also be observed on solid B‐medium, in which SS4 colonies formed extensive lateral pseudohyphae (Figure 1K). As expected from higher EFG1 expression in pRC2312P–H‐transformants compared with SS4 cells, filamentous growth was enhanced (Figure 7, compare B and C). pRC2312P‐H transformants grew as normal yeast cells under conditions of PCK1p repression, consistent with a wild‐type level of the EFG1 transcript (Figure 6, lane 3). To test whether an enhanced EFG1 expression level interferes with hyphal formation, we added 5% serum to a culture of SS4 grown in B‐medium. Such cells rapidly (within 30 min) began to form typical germ‐tubes that often were seen to emerge from pseudohyphal or yeast cells (Figure 7B, insert) and that continued to extend by apical growth as true hyphae. A similar result was obtained with strain CAI4[pRC2312‐P‐H]. Thus, EFG1 overexpression enhances pseudohyphal growth and allows the formation of true hyphae.
The C.albicans EFG1 gene encodes a new member of a conserved class of fungal proteins that are involved in morphogenetic processes. This class includes StuAp of A.nidulans, as well as the Phd1 and Sok2 proteins of S.cerevisiae. StuAp is involved in the formation of conidiophores in A.nidulans (Miller et al., 1992). Phd1p enhances and Sok2p suppresses pseudohyphal formation in S.cerevisiae (Gimeno and Fink, 1994; Ward et al., 1996), while Efg1p regulates the morphogenesis of C.albicans. The homology among the three proteins (referred to as ‘SPSE proteins’) extends over a region of ∼100 amino acids with 80% similarity. Efg1p and Phd1p, which both stimulate pseudohyphal growth in S.cerevisiae, do not contain homologous sequences other than the region conserved in SPSE proteins. However, this region cannot be responsible for pseudohyphal induction, since it is also present in Sok2p. Although a limited homology of SPSE proteins to the DNA‐binding region of transcription factors including Swi4p has been discussed (Gimeno and Fink, 1994; Ward et al., 1996), these authors did not recognize that the core of the conserved region can be aligned to the bHLH domain of eukaryotic transcription factors, such as the mammalian Max, Myc, MyoD and E47 proteins or the yeast Cbf1 and Pho4 proteins (Ferré‐D'Amare et al., 1993; Ellenberger, 1994; Ma et al., 1994) (see Results; Figure 2).
Changes in EFG1 expression have a greater effect on the morphology of C.albicans than the absence of elements of the MAP‐kinase pathway in cph1 and hst7 mutants (Liu et al., 1994; E.Leberer, personal communication). In such strains a defect of hyphal growth is detected only on certain solid media, but no other media or in liquid. Our experiments suggest that Efg1p is essential for morphogenesis of the yeast, pseudohyphal and hyphal growth forms in C.albicans. Although a strain could not be constructed in which both EFG1 alleles are disrupted (possibly, because EFG1 is essential for growth), the consequences of lowered EFG1 expression could be analyzed in a strain which only contained one functional EFG1 allele under the control of the regulatable PCK1 promoter (C.Leuker, A.Sonneborn and J.F.Ernst, manuscript in preparation). This strain, under conditions of EFG1 repression (glucose media), did not produce regular yeast cells, but formed rod‐like, elongated cells that resemble opaque cells of strain WO‐1 (Soll et al., 1993). These cells often remained attached leading to the formation of pseudohyphal filaments. This phenotype was obtained on solid and in liquid media. These results indicate that Efg1p, directly or indirectly, acts as a repressor of an elongated cell form and pseudohyphal growth. On the other hand lowered EFG1 expression also impaired cells to form true unconstricted hyphae in the presence of serum. Thus, Efg1p could function as a repressor, but also as an activator that is needed for hyphal growth. A dual function of a transcription factor as a repressor and an activator is not uncommon, as is discussed below for Myc, and may be explained by dimer formation and interaction with regulatory proteins.
Overexpression of EFG1 mediated by the PCK1 promoter also led to stimulation of filamentous growth in liquid and on solid media. Such filaments consisted of extremely extended cells, whose appearance was unlike the rod‐like cells formed during lowered EFG1 expression and unlike the pseudohyphae of S.cerevisiae. Although these filaments resembled true hyphae they contained constrictions at the site of cell connections and arose by budding; thus, the filaments induced by EFG1 overexpression must be considered as pseudohyphae. It is a surprising finding that reduced EFG1 expression and its overexpression both stimulate pseudohyphal growth of a different type. Conceivably, the same pathway or different pathways lead to both pseudohyphal growth forms. If the same intracellular pathway were involved, Efg1p could act, e.g. by inactivation of a repressor complex containing Efg1p by a ‘squelching’ mechanism (Ptashne and Gann, 1990). Alternatively, excess Efg1p could activate a morphogenetic pathway different from the pathway activated during Efg1p limitation. The finding that overexpression of EFG1 did not suppress the formation of true hyphae by serum is consistent with the above hypothesis that Efg1p is an activator of hyphal growth. Since the EFG1 transcript level is downregulated rapidly prior to hyphal formation it is possible that the Efg1 protein is required only for early events during hyphal formation, for example by creating a state of ‘competence’ to undergo hyphal morphogenesis.
The presumed role of Efg1p as a transcriptional activator and repressor and the homology to Myc (see Results) suggest a model for Efg1p function in accordance with the known functions of Myc. A heterodimer consisting of Myc, which contains a transactivation domain rich in serine and proline residues and the Max protein is required for the activation of target genes during growth of mammalian cells (Ferré‐D'Amare et al., 1993). High levels of Myc activate gene expression and favor cell growth, but block terminal cellular differentiation, presumably because Myc also acts as a transcriptional repressor (Li et al., 1994); in contrast, elevated production of Max represses transcriptional activation by formation of Max–Max homodimers or Max–Mad heterodimers binding to the same targets as Myc–Max (Ayer and Eisenman, 1993; Bernards, 1995). Myc transcript levels decline during the onset of cellular differentiation (Ayer and Eisenman, 1993). As with Myc, Efg1p appears to function both as a transcriptional activator and repressor, because on the one hand it is required for hyphal development and possibly, growth in general, since an EFG1 null strain could not be constructed; its termini, like the transactivation domain of Myc, are rich in serine and proline residues. On the other hand, Efg1p appears to repress the formation of elongated, pseudohyphal‐like cells (the phenotype observed at low Efg1p production levels). Similar to Myc transcripts and consistent with a repressor function, Efg1p transcript levels decline drastically during hyphal induction. Also, we obtained evidence that EFG1 is autoregulated, thus encoding a repressor of its own promoter activity (unpublished results). The repressing activity of Efg1p, as in the case of Myc, may be mediated by association with accessory transcriptional repressors, such as Sin3p (Ayer et al., 1995). A C.albicans protein forming a heterodimer by its bHLH domain with Efg1p is not known. Possibly, the Phd1p and Sok2p bHLH proteins in the yeast S.cerevisiae may form a heterodimer of an activator and a repressor protein as Myc–Max. Sok2p appears to represent a transcriptional factor that represses genes activated by cAMP‐dependent PKA (Ward et al., 1996). Since Efg1p overexpression stimulates pseudohyphal formation in S.cerevisiae and C.albicans it may represent the C.albicans analogue of the S.cerevisiae Phd1 protein. However, while EFG1 is essential for the morphogenesis of C.albicans, a dependence on Phd1p function becomes apparent only if Sok2p function is absent (Ward et al., 1996). As EFG1 in C.albicans, stuA is an essential gene for morphogenesis in A.nidulans (Miller et al., 1992).
Much of the recent interest in C.albicans is based on the demand for effective antifungal agents to cope with the increasing incidence of life‐threatening systemic candidiasis. Efg1p may be an effective antifungal target, because it is essential for C.albicans morphogenesis and presumably for its growth, while it contains specific features present only in fungal SPSE proteins, on which selectivity of possible agents may be based. Since Efg1p may dimerize, as other bHLH proteins, such interacting proteins, which remain to be discovered, are also possible drug targets. Finally, target genes of Efg1p must be defined, which may present additional opportunities for the action of antifungal agents.
Materials and methods
Strains and media
Strains are listed in Table I. A haploid S.cerevisiae laboratory strain, B76 (derivative of MB331‐17A), was diploidized by transformation with YCp50‐HO (Herskowitz and Jensen, 1991); a plasmid‐free derivative was isolated and designated B76/4. This strain shows similar pseudohyphal growth on low nitrogen media as a strain described previously (Gimeno et al., 1992). Transformation of S.cerevisiae was by the lithium acetate method and of C.albicans by the spheroplast method (Sherman et al., 1986). Yeast cells were grown in YPD, YP (as YPD without glucose) or on supplemented SD minimal medium (Sherman et al., 1986). The PCK1 promoter was induced in B‐medium [0.67% yeast nitrogen base (Bacto), 2% Na‐succinate, pH 6.5] or in CAA‐medium [0.67% yeast nitrogen base (Bacto), 2% casamino acids]. To induce pseudohyphal growth of S.cerevisiae cells were grown on SLADH plates (Gimeno et al., 1994) containing 50 μM ammonium sulfate and the required amino acids. Pseudohyphal growth was inspected microscopically after 2 days at 30°C, using a Zeiss Axioscop. To induce hyphae C.albicans cells were pregrown at 30°C in YPD to the logarithmic growth phase and then diluted (OD600 nm= 0.1) into YP, 5% horse serum or YPD containing 5% horse serum at 37°C (Delbrück and Ernst, 1993).
A genomic library was constructed by inserting random genomic Sau3AI fragments of C.albicans ATCC10231 into the 2μ vector pLG‐B/1 (Leuker and Ernst, 1994); this library contained 50% inserts with an average size of 6 kb. B76/4 was transformed with the pLG‐B/1‐ and YRp7‐ (Losberger and Ernst, 1989) banks selecting for tryptophan or uracil prototrophy, respectively. Single transformants were grown on SLADH plates and their colony and cell morphology was examined after 2 days. In a limited screening we identified one transformant with enhanced formation of pseudohyphae among 2300 transformants of the YRp7 library (p606/1) and one such transformant among 1500 transformants of the pLG89‐B/1 library (p607/2). Control transformants contained plasmid YCpR2V encoding the dominant Ras2Val19 protein or pCG38 encoding PHD1 (Gimeno et al., 1992; Gimeno and Fink, 1994).
Subcloning and sequencing
Plasmids p606/1 and p607/2 were recovered from transformants with enhanced pseudohyphal formation of pseudohyphae (Sherman et al., 1986) and retransformed into B76/4 to confirm their phenotypic effects. Subclones of the 5.2 kb BamHI insert in plasmid p606/1 were generated by insertion of subfragments into YEplac vectors (Gietz and Sugino, 1988) or pRS424 (Christianson et al., 1992). A subclone carrying the 3.8 kb EcoRI subfragment was cut with MscI and PstI and shortened by exonuclease III, followed by nuclease S1 treatment and religation; deletions from the opposite end of the insert were obtained after ClaI and ApaI digestion. Subclones were sequenced by the dideoxy chain termination method using double‐stranded plasmid DNA as templates.
A BamHI site was placed upstream of the putative EFG1 ATG start codon (5′‐GGATCCTA ATG) by PCR‐amplification of a 0.6 kb BamHI–XhoI fragment; this was verified by sequencing. The reassembled EFG1 coding region, as a 2.3 kb BamHI–SphHI fragment, was inserted downstream of the GAL1‐promoter on a EcoRI–BamHI fragment (Johnston and Davis, 1984) and inserted in YEplac195 (EcoRI, SphI) to generate pBIST. To overexpress EFG1 in C.albicans, pBI‐1, a derivative of pRC2312 (Cannon et al., 1992) containing a 1.5 kb BamHI–BglII promotor fragment of the C.albicans PCK1 gene was used (C.Leuker, A.Sonneborn and J.F.Ernst, manuscript in preparation); in this plasmid the BglII site was created at the position of the PCK1 ATG translational start sequence by PCR mutagenesis. Downstream of the PCK1 promoter the entire EFG1 coding region was reconstructed by consecutively inserting the 1 kb BamHI–BglII and the 0.7 kb BglII–HindIII fragments of pBIST, to create plasmid pRC2312P‐H.
To study transcriptional regulation of EFG1 pCL76 was constructed, in which the EFG1 promoter regulates expression of the LAC4 reporter gene (Leuker et al., 1992). First, a HincII–SalI site was constructed by PCR mutagenesis at the ATG start sequence (CCCATATTAATG > CCCATGTCGAC). The BamHI–HincII promoter fragment was ligated with a 4.2 kb PvuII–XbaI LAC4 fragment of pRS‐LAC4 (unpublished results) and inserted into pRC18, a derivative of pRC2312 (Cannon et al., 1992) containing the pUC‐18 multiple cloning site, to construct pCL76.
To replace the EFG1 promoter in one of the C.albicans alleles by the regulatable PCK1 promoter a plasmid carrying an appropriate disruption fragment was constructed. Plasmid pCA01, containing the 1.5 kb BamHI–BglII PCK1 promoter fragment in pUC‐21 (C.Leuker, A.Sonneborn and J.F.Ernst, manuscript in preparation), was modified by first inserting the 1.55 kb BamHI–BglII fragment of pBIST carrying part of the EFG1 coding region into the BglII site at the 3′ end of the promoter; this was followed by insertion, in the presence of an oligonucleotide adapter, of a 1 kb BamHI–PstI URA3 fragment (Losberger and Ernst, 1989) into the BamHI site upstream of the promoter. This construction retained a single BamHI site flanking URA3, into which a 3.5 kb BglII–BamHI fragment containing EFG1 upstream promoter sequences was inserted. From the final plasmid, pCAHUH, a 7.8 kb NotI–BglII fragment was isolated (NotI is situated within the plasmid 5′ of BglII–BamHI fusion point) and used to transform strain CAI‐8 to uracil prototrophy.
To disrupt the second EFG1 allele the 2.4 kb EcoRV fragment of pMK16 carrying the C.albicans ADE2‐gene (Kurtz et al., 1987) was ligated into the single MscI site within the EFG1 coding region of a pUC19‐subclone containing the 3.8 kb EcoRI–BamHI EFG1 fragment (pUC19EFG1). The 5.2 kb XhoI fragment of the resulting vector pUC19/EFG1ADE2 was transformed into the CAI‐8 transformant carrying the PCK1::EFG1 allele selecting for prototrophy. Genomic DNA of transformants was analyzed on Southern blots using the 250 bp EcoRV–XhoI EFG1 fragment as probe.
Total RNA and poly(A) RNA of C.albicans ATCC10261 were isolated and separated by denaturing gel electrophoresis as described (Delbrück and Ernst, 1993). For transcript mapping a 341 bp EcoRI–NheI fragment labelled at its NheI‐end by filling‐in using Klenow enzyme, dCTP, dTTP and [α‐32P]dATP was isolated and hybridized to 2–4 μg poly(A)‐RNA. Treatment with nuclease S1 was performed essentially as described (Sambrook et al., 1989) and the single product, a 168 nt RNA, was identified on a sequencing gel. Total DNA of C.albicans strains was prepared as described (Sherman et al., 1986). Southern blots on fungal genomic DNAs were performed according to standard procedures (Sambrook et al., 1989). Nucleic acids blots on nylon (Hybond, Amersham) were probed using the 1.5 kb NheI fragment of p606/1 as EFG1 probe; as controls an ACT1 fragment (Delbrück and Ernst, 1993) was used. Probes were labelled to high specific activities with 32P using random oligonucleotides as primers (Sambrook et al., 1989). LAC4‐specified β‐galactosidase activity was measured using the Galacto‐Light Plus kit (Tropix); luminescence was determined in a LB953 luminometer (EG&G Berthold GmbH) and expressed as relative light units (RLU) per μg protein. In this assay 100 pg of Escherichia coli βgal had an activity of 20 000 RLU; the assay was linear from 10 pg to 500 pg βgal. Protein was determined using the Bradford assay (Bio‐Rad).
We thank B.B.Magee for mapping information. We thank B.Strickling for contributing to the construction of pBIST and acknowledge the excellent technical help of M.Gerads. We thank M.Wigler for plasmid YCpR2V, G.R.Fink for pCG38 and H.Liu for strains CGX‐31 and HLY492. This work was supported by the Deutsche Forschungsgemeinschaft.
- Copyright © 1997 European Molecular Biology Organization