TUP1 encodes a transcriptional repressor that negatively controls filamentous growth in Candida albicans. Using subtractive hybridization, we identified six genes, termed repressed by TUP1 (RBT), whose expression is regulated by TUP1. One of the genes (HWP1) has previously been characterized, and a seventh TUP1-repressed gene (WAP1) was recovered due to its high similarity to RBT5. These genes all encode secreted or cell surface proteins, and four out of the seven (HWP1, RBT1, RBT5, and WAP1) encode putatively GPI-modified cell wall proteins. The remaining three, RBT2, RBT4, and RBT7, encode, respectively, an apparent ferric reductase, a plant pathogenesis-related protein (PR-1), and a putative secreted RNase T2. The expression of RBT1, RBT4, RBT5, HWP1, and WAP1 was induced in wild-type cells during the switch from the yeast form to filamentous growth, indicating the importance of TUP1 in regulating this process and implicating the RBTs in hyphal-specific functions. We produced knockout strains in C. albicans for RBT1, RBT2, RBT4, RBT5, and WAP1 and detected no phenotypes on several laboratory media. However, two animal models for C. albicans infection, a rabbit cornea model and a mouse systemic infection model, revealed that rbt1Δ and rbt4Δ strains had significantly reduced virulence. TUP1 appears, therefore, to regulate many genes in C. albicans, a significant fraction of which are induced during filamentous growth, and some of which participate in pathogenesis.
Candida albicans is a commensal resident of human mucosal surfaces and is a frequent cause of opportunistic infections, some of which can be life threatening (Odds 1988, 1994a; Dupont 1995). Several features of C. albicans have been implicated as virulence characteristics and, with the advent of molecular genetics in this organism, several of these features have been tested directly by the deletion of relevant genes. One such characteristic is the capacity of C. albicans to grow both in a single-celled yeast form (the blastospore) and in a spectrum of filamentous (hyphal) forms (Corner and Magee 1997; Gow 1997; Loet al. 1997; Kobayashi and Cutler 1998; Mitchell 1998). Other characteristics thought to be critical to pathogenesis are adhesiveness to host cells, secretion of degradative enzymes such as proteases and phospholipases, and interactions with the immune system (Odds 1994b; McCulloughet al. 1996; Fukazawa and Kagaya 1997).
Interactions between pathogens and their mammalian host are typically mediated by molecules that are either secreted or displayed on the cell wall. The cell wall of C. albicans is dynamic, displaying several proteins whose abundance changes dramatically under different growth conditions (Chaffinet al. 1998). Several hyphal-specific cell wall proteins have been characterized and, while some have been assigned roles in cell wall integrity or adhesiveness to host cells, others have no known function. In addition to their functional characteristics, the accessibility of cell wall proteins makes them useful targets for sensitive diagnostic tests and some simple therapeutic approaches.
We recently characterized a gene, TUP1, whose deletion causes C. albicans to grow in a constitutively filamentous state (Braun and Johnson 1997). TUP1's orthologue in the model yeast, Saccharomyces cerevisiae, has been studied extensively and encodes part of a transcriptional repression complex that is brought to promoters by regulated DNA-binding proteins (Keleheret al. 1992). This complex participates in many separate pathways of S. cerevisiae that are regulated in response to environmental or other conditions (Carrico and Zitomer 1998; Proft and Serrano 1999). We therefore proposed that, in C. albicans, TUP1 encodes a repressor of genes whose expression generates filamentous growth. Several questions arose from this hypothesis. One is whether, in C. albicans, TUP1 participates in any pathways other than filamentous growth. If not, then TUP1's own expression might be modulated for regulatory effect, in contrast to the situation in S. cerevisiae, where regulation acts on the DNA-binding proteins that bring Tup1p to selected promoters. A second question concerns the nature of the switch from budding to filamentous growth. Does TUP1 repress one key gene that generates filamentous growth, or does it repress a variety of genes, each of which contributes to filamentous growth? Finally, how does the pathway involving TUP1 relate to other pathways known to influence filamentous growth and to the normal process of filamentous growth control in wild-type cells? We address aspects of each of these questions in this article, although the final question is dealt with in more detail in a related article (Braun and Johnson 2000).
Since deletion of TUP1 causes filamentous growth in the absence of any environmental inducing signal, the tup1Δ strain is a useful tool for the study of filamentation in C. albicans. Filamentous tup1Δ cells can easily be grown in culture and studied in parallel, under the same growth conditions, with their wild-type counterparts in the blastospore (single-cell) state. To find genes downstream of the TUP1 repressor that are responsible for generating the tup1Δ phenotype or otherwise participate in tup1Δ-dependent filamentous growth, we used a cDNA subtraction method to isolate genes overexpressed in the tup1Δ strain compared to a wild-type strain. Several genes, all of which encode cell surface or secreted proteins, were identified by this procedure, and nearly all were induced in wild-type C. albicans by conditions that promote filamentous growth. In this article, we describe their isolation and characterization, and the phenotypic analysis of the respective deletion mutants, in both laboratory culture and animal models of C. albicans infection.
MATERIALS AND METHODS
Strains and media: The strains used are listed in Table 1. BCa7-9 carries the long allele of RBT1, while BCa7-10 carries the shorter allele. Standard media YPD and SD (Guthrie and Fink 1991) were used for most strain growth and maintenance, and the filamentous growth media cornmeal agar (Difco, Detroit; Vidottoet al. 1986) and Spider (Liuet al. 1994) have been described.
Representational difference analysis: RNA was prepared by a hot phenol protocol (Ausubelet al. 1992) from C. albicans strains SC5314 and BCa2-10, grown in YPD at 30° to OD600 of 0.7 from a low inoculum. To achieve rough homogeneity for OD600 readings, the filamentous tup1Δ cultures were vigorously mixed prior to sampling and the cells were then mixed in the cuvette by pipette aspiration immediately before reading. A total of 200 ml of cells were harvested, 4.35 and 3.05 mg of RNA were obtained, respectively, and 2 mg of total RNA was used for poly(A) selection using an Oligotex kit from QIAGEN (Chatsworth, CA). Approximately 12.5 and 13.9 μg of poly(A) RNA were isolated, respectively.
To isolate differentially expressed gene fragments, the CLONTECH (Palo Alto, CA) PCR-Select cDNA subtraction kit, which is based on representational difference analysis (Hubank and Schatz 1994) was used according to its directions. Briefly, double-stranded cDNA was synthesized from the target (tup1Δ) and driver (SC5314) RNA samples and cut into fragments. Target samples were split in two, ligated to different adapters, and then put through two hybridizations. A quick self-hybridization with excess driver cDNA removed the most abundant molecules, after which the two target pools were combined and a more exhaustive hybridization was carried out to provide substrates for target-specific PCR. After 42 cycles of nested PCR, a heterogeneous smear of DNA was seen on an agarose gel that represented molecules that were overrepresented in the original target pool.
T/A ends were generated on pBS SK+ vector DNA with Taq polymerase and ligated to size-selected PCR products. Random clones were picked and their inserts were amplified by PCR; the products were made into probes and hybridized either to Northern blots of electrophoretically separated original total SC5314 and BCa2-10 RNA or to slot blots (Ausubelet al. 1992) containing a filamentous growth series of total RNAs prepared from a MAL2→TUP1 expression strain (BCa5) or wild-type SC5314 treated with serum, as described in Figure 1. Church hybridization conditions were used throughout (Church and Gilbert 1984). These blots were evaluated for differential expression, and promising clones were partially sequenced.
Cloning and sequencing of novel genes: Positive clones contained only small amounts of C. albicans sequence information, so a λ library was used to isolate full-length genes. Typically, several overlapping clones were isolated, and these were restriction mapped to find their regions of interest. One clone was selected for sequencing by primer walking, and both strands of the open reading frame (ORF) were sequenced. Only one clone, WAP1, was not fully contained on one library clone; therefore its sequence is a composite from two overlapping clones. The selection of single clones is significant because, as a constitutive diploid, genes in C. albicans have two alleles that are often different, constituting natural heterozygosity (Scherer and Magee 1990). Sequences are deposited in GenBank under the accession numbers listed in Figure 3.
Sequence analysis was performed with DNA Strider (Kyte-Doolittle hydrophobicity analysis, translations for presentation, and miscellaneous analysis) and the GCG package (Pileup for multiple sequence alignments, Gap for pairwise alignments and identity values, and SignalScan for signal sequence identification). Multiple sequence alignments are presented through SeqVu (Garvan Institute).
Deletion of novel genes: The techniques of Fonzi and Irwin (1993) were used with slight variations. PCR primers were designed to amplify a pair of 0.4- to 1.0-kb regions flanking each open reading frame. These flanks were then cloned into the vector pBB510 (derived from pMB-7; Fonzi and Irwin 1993), placing one on either side of the hisG-URA3-hisG cassette. The PCR primers were typically designed to contain the following restriction sites: upstream flank, SphI on upstream and BglII on downstream side; downstream flank, NsiI on upstream and Asp 718 on downstream side. After the PCR products were cut with the appropriate enzymes, these ends enabled a four-way ligation into quadruply cut and phosphatase-treated pBB510. The selection of either set of enzymes for the vector restriction cuts (SphI, BamHI, NsiI, Asp 718, or SphI, PstI, BglII, Asp 718) allowed the relative direction of the hisG-URA3-hisG cassette and the flanks to be reversed during the ligation, yielding two disruption plasmids for each target gene.
Five micrograms of one of the resulting plasmids was cut to release the entire cassette (SphI to Asp 718), phenol-extracted, ethanol-precipitated, and transformed into C. albicans strain CAI4 to integrate into, and therefore replace, the first allele of the selected gene. PCR primers residing outside of the cloned gene flanks were used in conjunction with primers directed against hisG in analytical PCR to verify the junctions of each new integrant, and correct integrants were plated on 5-fluoroorotic acid + uridine to select for loss of URA3 by intralocus recombination between the tandem hisG repeats. A second cycle of transformation was then performed with the opposite-orientation disruption plasmid, which yielded a set of integration junctions distinguishable by PCR from those of the original integration event. Absence of the wild-type gene was also verified by PCR at this point (Figure 1). This strain, with one URA3 gene integrated in the genome, was used for all further phenotypic characterization. The following intervals were deleted: RBT1, from 17 bp into the predicted ORF to 41 bp before its end; RBT2, from 60 bp into the predicted ORF to 55 bp beyond its end; RBT4, from 44 bp inside the predicted ORF to 64 bp before its end; RBT5, from 24 bp before the beginning of the predicted ORF to 29 bp inside its end; WAP1, from 1 bp before the start of the predicted ORF to 424 bp before its end. Details of the oligonucleotides used and the plasmids constructed are available on request.
Pathogenesis studies in rabbit corneas: Strains of C. albicans were maintained at −70° in YPD broth to which glycerol (5 ml glycerol in 25 ml YPD) had been added. Budding yeast, for use as inoculum in the rabbits, was grown by inoculating 25 ml of YPD with 100 μl of the stored culture and incubating on a shaker at 200 rpm overnight at 30°. A thick paste of budding yeast was prepared by centrifuging the YPD culture at 2500 × g for 5 min and pouring off the supernatant. The tube was left inverted until it was used so that it could drain thoroughly.
Three rabbits were inoculated with each of the five strains studied. Under general and local anesthesia, the nictitating membrane was removed and the central 7 mm of corneal epithelium was removed from the right eye of each rabbit. Fifty microliters of “bud paste” was placed on the debrided cornea and a contact lens was placed over the inoculum. A temporary tarsorrhaphy was done to prevent extrusion of the contact lens. The tarsorrhaphy sutures, contact lens, and any residual inoculum were removed after 1 day. After a total of 6 days, the rabbits were killed, and their corneas were excised and divided in half for quantitative isolate recovery and histology studies.
Postmortem tissue dissection: Rabbits were killed by rapid intravenous injection of euthanasia solution containing sodium pentobarbital (Sleepaway; Fort Dodge Laboratories, Fort Dodge, IA). The infected corneas were removed at the limbus and bisected longitudinally. One-half of the cornea was used for quantitative isolate recovery and the other half was fixed in formalin for histology.
Quantitative isolate recovery: One corneal half from each rabbit was cut into small pieces, placed into 3 ml of sterile normal saline in a sterile test tube, and ground for three 10-sec intervals with a Tissumizer (Tekmar, Cincinnati). Serial 10-fold dilutions of the corneal suspension were done in sterile normal saline and three 100-μl samples from each dilution were spread with a glass rod over the surface of Sabouraud's dextrose agar plates. The plates were incubated until the colonies were easily countable. The set of three plates with between 10 and 100 colonies was counted and used to estimate the total number of colony-forming units present in each infected cornea.
Histology: For each cornea, a single 5-μm, paraffin-embedded section was cut from the midcornea. The section was stained with periodic acid-Schiff's reagent and counterstained with fast green stain. Micrographs were taken with an Olympus BH-2 microscope with a ×20 objective and ×2.5 photo eye-piece.
Hyphal penetration: Pictures were taken of one entire corneal section from each infected cornea with an Olympus PM-20 exposure control unit attached to an Olympus BX-40 microscope. The final magnification was ×150, which allowed us to mark the hyphal ends and measure the corneal thickness and hyphal length with a ruler for each observation of a hypha and calculate the depth of penetration as a percentage of corneal thickness. All hyphae were assumed to originate at the outside corneal surface at the site of inoculation. The fungal biomass was calculated by multiplying the mean depth of penetration by the number of hyphae observed in each section and averaging over all three rabbits.
Statistics: The data were analyzed using SAS version 6.12 using the Windows NT operating system. First, the depth of hyphal penetration data for each rabbit was analyzed to determine the mean number of observations, mean depth of penetration, quartile depth of penetration, and biomass. These data were then analyzed along with the isolate recovery data and the clinical evaluation data. Due to the nature of the clinical scoring system and the small number of rabbits in each group, the data were analyzed by the method of least squares to fit general linear models.
Pathogenesis studies in mice: Strains of C. albicans were first verified for their general health by testing their growth rates (in YPD), which were equivalent for all strains, except for tup1Δ. The tup1Δ strain was grown in SD plus glycerol to keep its filament length to a minimum to allow injection. Wild-type cells grew very poorly in this medium and therefore could not be used as a control. For injection into mice, cultures were grown in YPD to an OD600 of 1 and then harvested and washed once or twice in sterile saline solution (0.9 m NaCl) at room temperature. Cells were counted with a hemacytometer and diluted to 2E6 cells per milliliter, of which 0.5 ml was injected into the tail vein of anesthetized 4- to 6-wk-old BALB/C mice. The maximum time between harvesting and injection was 4 hr. Cells for earlier experiments shown in Figure 8A, were rinsed only once, leading to residual cell growth and injection of between 1 and 2 million cells. Five or six mice were injected per strain of C. albicans. Mice were assessed daily or more often for weight and general condition and were killed if moribund, in accordance with UCSF humane treatment guidelines. The number of mice surviving through the day was recorded for each group.
We used a differential cDNA subtraction technique patterned after representational difference analysis (Hubank and Schatz 1994) to isolate genes under TUP1 regulation. For this analysis, tup1Δ and wild-type cells were grown under identical noninducing conditions (YPD) and formed, respectively, filaments and single budding blastospores (Braun and Johnson 1997). Since TUP1 encodes a transcriptional repressor, excess wild-type cDNA was subtracted from experimental tup1Δ cDNA to yield cDNAs that were overrepresented (derepressed) in the mutant cells. Representatives of the resulting products were tested for differential expression by hybridization to blots of C. albicans total RNA. Initial Northern blots contained the input RNAs, to screen the clones for overexpression in tup1Δ cells vs. wild-type cells. Of a total of 150 clones tested, 75 displayed greater expression by this assay. Subsequent slot blots posed a further question: is the represented gene also induced during filamentous growth in wild-type cells treated with serum and, if so, with what kinetics (Figure 2, h–m)? Of 66 clones that were tested against slot blots, 47 appeared to be induced in wild-type cells undergoing filamentous growth. These results indicated that the screen was successful in isolating TUP1-regulated genes and also suggested that most of the genes downstream of TUP1 repression are induced during filamentous growth.
The screen yielded one previously characterized gene, HWP1 (Staab et al. 1996, 1999; Staab and Sundstrom 1998), and a set of novel genes. The novel genes were named RBT (repressed by TUP1) and, while the five (RBT1, RBT2, RBT4, RBT5, and RBT7) described here were differentially expressed in tup1Δ cells vs. wild-type cells, only three (RBT1, RBT4, and RBT5) were also induced by serum treatment of wild-type cells. An additional genes, WAP1 (wall protein one), was discovered in the course of isolating RBT5 and also was induced during filamentous growth in wild-type cells (not shown). A related article describes the regulation of most of these genes in more detail (Braun and Johnson 2000). We do not yet know how direct the regulatory relationship is between TUP1 and any of the RBT genes.
Sequence analysis of the RBT genes: A wild-type C. albicans genomic DNA phage library was screened with inserts from each of the differential cDNA clones, and plasmids were excised from the resulting phage. Areas matching the original cDNA fragments and containing open reading frames were sequenced on both strands and conceptually translated (see Figure 3 for the protein sequences and GenBank accession numbers). Each putative protein had a hydrophobic signal sequence at the NH2 terminus, indicating that it enters the secretory pathway. Several (HWP1, RBT1, RBT5, and WAP1) also had hydrophobic segments at the very COOH terminus, signifying addition of a glycosyl phosphatidylinositol (GPI) moiety (Figure 4B). GPI is added to the COOH terminus of such proteins after the hydrophobic tail is clipped off. In fungi, the GPI moiety can either serve to anchor the protein on the extracellular face of the plasma membrane or be reprocessed to enable covalent attachment of the protein to the cell wall (Udenfriend and Kodukula 1995; Caroet al. 1997). The HWP1, RBT1, RBT5, and WAP1 gene products all lacked a dibasic tag near the GPI addition site, indicating that these proteins are destined for cell wall integration.
RBT1 characteristics: The protein encoded by RBT1 was 43% identical to that encoded by the previously characterized HWP1 gene, which was also recovered in our screen (Figure 4A). HWP1 was isolated by Staab et al. (1996) as a gene encoding a prominent, antigenic, and hyphal-specific C. albicans cell wall protein and has since been found to play a unique role in adhesion by covalently linking to host tissues (Staabet al. 1999). In addition to high similarity in the NH2- and COOH-terminal hydrophobic segments (Figure 4B), the proteins share high similarity in the COOH-terminal two-thirds of their length, which are 53% serine, threonine, and proline residues. Such regions are common in fungal cell wall proteins, constituting preferred sites for O-linked glycosylation (Jentoft 1990; Lipke and Ovalle 1998). This glycosylation imparts a rod-like structure to the protein domain, enabling it to span the periplasmic space/cell wall to display the mature NH2-terminal portion of the protein to the outside (Jentoft 1990). RBT1's product also contains one predicted N-glycosylation site (NX[T/S]), compared to three for HWP1 (Staab and Sundstrom 1998). The mature NH2 third of the RBT1 and HWP1 gene products, which are presumably displayed on the cell exterior, was the least similar portion, with 16% identity (Figure 4A). This area of RBT1 has no significant similarity to any other known proteins, nor have its composition and other characteristics given us any other clues about its function. Thus although RBT1 and HWP1 may be processed and displayed on the cell surface in similar ways, they are predicted to display significantly different determinants to the cell surroundings.
In addition to sharing sequence characteristics, RBT1 and HWP1 also shared some regulatory properties. In wild-type C. albicans, RBT1 and HWP1 were not expressed in blastospores but were rapidly induced in response to serum treatment, which also induces filamentous growth. HWP1 was more strongly induced, but RBT1 was induced with equal speed, indicating that its expression is part of a very early transcriptional response to serum. Figure 2 (a–g) also shows that RBT1 and HWP1 were both derepressed as TUP1 was depleted. The activation of HWP1 in serum was much greater than its derepression upon loss of TUP1 expression, indicating that while the serum response may partly depend on rapid lifting of TUP1 repression, it must also depend on other regulated forms of activation or derepression (Braun and Johnson 2000).
The sequence released by the systematic C. albicans genome sequencing effort (assembly no. 4, contig no. 2768; Stanford DNA Sequencing and Technology Center) differed from our sequence by lacking DNA-encoding amino acids (aa) 612–640, while being virtually identical elsewhere. PCR assays across this area of genomic DNA from C. albicans wild-type strain SC5314 yielded products of two different sizes that correctly matched each of the predicted DNA lengths, confirming the presence of each sequence and showing that RBT1 has two alleles that can be easily distinguished (not shown).
RBT2 characteristics: RBT2 encodes a protein similar to ferric reductases located in the plasma membrane, which perform the first step in the assimilation of iron and related metals—the reduction of external chelated Fe III (ferric) to the more soluble Fe II (ferrous). Human pathogens commonly go to great lengths to acquire iron, since they face a highly developed host defense termed iron withholding (Weinberg 1993; Jurado 1997). Both C. albicans and S. cerevisiae have numerous plasma membrane ferric reductases and, in S. cerevisiae, they are known to differ both in function (i.e., which metals they prefer) and in the conditions under which they are expressed (Martinset al. 1998).
The protein product of RBT2 had greatest identity (37%) to the product of CFL1, the only complete ferric reductase so far reported in C. albicans (Yamada-Okabeet al. 1996). Its similarity to the FRE gene products of S. cerevisiae ranged from 21% (FRE7) to 29% (FRE2) over their entire lengths. All of these proteins are predicted to contain seven transmembrane domains, in addition to their hydrophobic signal sequence (Figure 5A). While RBT2 expression was strongly induced in the tup1Δ strains, it was not significantly induced during serum-induced filamentous growth (Figure 2). We do not know what signals normally regulate RBT2 expression.
RBT4 characteristics: A set of genes that encode proteins termed pathogenesis related (PR) have been isolated from plants. These genes are induced upon bacterial and fungal infection, and their secreted products fall into groups termed PR-1 to PR-5. The PR-1 family has an unknown mode of action, but does have antifungal activity and has an extraordinarily stable three-dimensional structure (Stintziet al. 1993; Nidermanet al. 1995). In its 131-amino-acid core region of similarity, RBT4's product is 46% identical to the PR1a, b, and c products of tobacco (Nicotiana tabacum); 59% identical to the S. cerevisiae PRY1 gene product; and 39% identical to the P14A protein from tomato (Lycopersicon esculentum) whose structure has been solved by NMR (Fernandezet al. 1997; Figure 5B).
RBT4 expression was induced by serum treatment of filamentous cells as well as depletion of the TUP1 gene product (Figure 2) and, therefore, is closely linked to filamentous growth. Its intriguing sequence characteristics suggest that its gene product, upon secretion, may damage either host cells or competing microbes.
Characteristics of RBT5 and WAP1: RBT5 and WAP1 encode proteins that share a highly similar 115-aminoacid sequence, which is present once in RBT5 and four times in WAP1 (Figure 6A). The identity of these repeats ranges from 60% between the third repeat of the WAP1 and RBT5-encoded proteins (wap1-c vs. rbt5-a) to 99% between the first and second repeats within the WAP1 protein (wap1-a vs. wap1-b). WAP1 was recovered from the hybridizations intended to isolate full-length RBT5 due to their high sequence similarity (80% identical over 305 bp) and due to the high representation of WAP1 repeats in the library.
RBT5 and WAP1 were related to only one sequence in the general database, the gene encoding proline-rich antigen (PRA) on the cell surface of the pathogenic fungus Coccidioides immitis (Duggeret al. 1996). Its NH2-terminal half comprises 1 unit with 24–28% identity to the 115-aa repeats found in RBT5 and WAP1, which we term the CRoW motif (C. immitis PRA, RBT5, or WAP1). More striking than its overall sequence similarity is that the CRoW motif from C. immitis PRA shares a set of eight cysteines (Figure 6A) and also four glycines and two prolines, suggesting that the structure of this motif may be conserved and may rely on disulfide bond formation in the extracellular environment. All three proteins have sequence characteristics indicating they are secreted and modified for covalent attachment to the cell wall, as described above. Based on these considerations, we predict that the CRoW motifs of RBT5 and WAP1 are stably exposed on the exterior of C. albicans cells. A large class of fungal surface proteins termed hydrophobins also possess eight cysteines that are thought to form disulfide bonds (Kershaw and Talbot 1998). These cysteines are, however, arranged quite differently, and these proteins have no other similarity with RBT5 or WAP1.
The expression of RBT5 was strongly induced by TUP1 depletion and also by treatment of wild-type cells with serum and Spider media (Figure 2; and see Braun and Johnson 2000, a related article). WAP1 also showed complex induction effects upon TUP1 deletion and serum/Spider treatment (data not shown), and it is probable that these two genes have diverged from a common ancestor relatively recently and share similarity in some of their regulatory properties as well as in their sequence.
RBT7 characteristics: A protein similar to secreted RNases was also uncovered in this screen. This gene, RBT7, was induced upon TUP1 depletion (Figure 2). Though it initially appeared to be slightly induced during serum treatment of wild-type cells, later work showed that its expression was unaffected by serum or Spider treatment (not shown). RBT7 is similar to several secreted RNase T2 enzymes, including YPL123c from S. cerevisiae, the RNase T2 precursor from Aspergillus oryzae, and the RNase T2 product from Tricoderma viride (34–36% identical over 234–295 amino acids; Inadaet al. 1991; Ozekiet al. 1991). Both A. oryzae and T. viride are filamentous fungal pathogens of plants. The role of these enzymes is unknown, but may be involved with degradation and utilization of RNA molecules from the environment, particularly the host.
Deletion of the RBT genes: We analyzed the biological role of a selected set of TUP1-regulated genes (RBT1, -2, -4, -5, and WAP1) by deleting both copies of each gene from the genome using the URA3-replacement techniques of Fonzi and Irwin (1993). Under laboratory conditions on a variety of nutritional media, no significant differences in growth rate, morphology, or other visible characteristics were detected between any of the mutant strains and the control strain, CAF2-1. The morphology of growing cells was assayed on YPD, SD, S-maltose, S only, YPD + 10% serum, Spider, cornmeal (both with and without coverslip), and plain agar plates. We therefore conclude that, while these genes are coregulated with filamentous growth, and their products probably reside in or near the cell wall, none are required for filamentous growth per se.
Studies of C. albicans infection: While many interesting aspects of C. albicans biology are accessible on laboratory media, factors influencing virulence are not revealed. We therefore tested a set of mutant strains in a rabbit model of C. albicans infection of the cornea of the eye as well as in the more typical mouse model of intravenous infection. First we describe the rabbit infection studies, and then the corroborating studies in mice. Each C. albicans strain was placed under a contact lens on one cornea of the subject rabbit, and the eye was then sewn shut for 1 day. The eyes were then observed for evidence of fungal infection and invasion for the following 5 days. The rabbits were killed after a total of 6 days, and the corneas were photographed and then processed to recover both live C. albicans and samples for histological study. Control strains included CAF2-1, which has a single URA3 allele as do the rbt deletion strains, and its parent, the clinical isolate SC5314 (Fonzi and Irwin 1993).
By visual assays, both rbt4Δ and rbt1Δ showed dramatic defects in cornea invasion, relative to the control strains and to the rbt2Δ and rbt5Δ strains. Figure 7, A and B shows infections by the control strains. D and F show infections by rbt2Δ and rbt5Δ, respectively, which did not differ in any observable way from those in A and B. As shown in C and E, infections by strains deleted for RBT1 and RBT4 showed profoundly reduced levels of infection, both in the clinical photo of the eye taken after 6 days of infection (top) and in the thin cornea sections (bottom) that show extents of hyphal invasion. The noninvasive behavior of these strains was particularly interesting since the strains showed no growth defects on laboratory media.
Consistent with the visual assays, significantly fewer organisms were recovered from the rabbits infected with rbt4Δ than from the rabbits infected with either SC5314 (P = 0.0126) or CAF2-1 (P = 0.0135; Table 2). Maximal depth of hyphal penetration from the outside of the cornea was also significantly less in rabbits infected with rbt4Δ compared with either SC5314 (P = 0.0092) or CAF2-1 (P = 0.0091). The clinical evaluation of disease by three independent observers was also consistent with these measures (data not shown). Since the sequence characteristics of RBT1 predict that its product is displayed on the exterior of the cell wall, and those of RBT4 predict that its product is secreted, we speculate that their functions may be (1) to interact with and impair the immune system components that normally clear C. albicans infections, (2) to provide adhesive interactions that allow the fungal filaments to grow into the host tissue, or (3) to directly damage host tissues.
To corroborate the results seen in the novel rabbit model system, we also performed studies in mice, injecting 106 C. albicans cells of various strains into the tail vein and assaying resultant mortality (Figure 8). As seen in A, these results mirror those seen in the rabbit model system, showing reduced virulence of the rbt1 and rbt4 deletion mutant strains, compared with the high virulence of the wild-type and rbt5 deletion strains. Autopsies of mice infected by the control strains (performed at time of death) showed large numbers of C. albicans hyphae in histologically stained sections of the kidney (periodic acid-Schiff's reagent stained, as in Sheehan and Hrapchak 1987). Mice infected with the rbt1Δ or rbt4Δ strains were likewise autopsied at time of death or at the end of the study and showed no detectable C. albicans cells in the kidney, consistent with lower virulence of those strains.
We also assayed the RBT1/rbt1 and RBT4/rbt4 heterozygous strains for virulence (Figure 8, B and C). As discussed above, the two alleles of RBT1 can be distinguished, and we constructed heterozygous strains missing each allele. All of these heterozygous strains appeared significantly less virulent than the control strain, suggesting that gene dosage may have significant effects on the roles of these genes in pathogenesis.
We have attempted to reintegrate the RBT1 and RBT4 genes into the double-mutant strains. Although we obtained transformants with the expected characteristics of correct integration site and expression level, none of these strains have showed significantly increased virulence. We have also made and tested one reintegrant of RBT1 into the RBT1/rbt1 heterozygote, restoring the strain to its wild-type complement of two wild-type alleles, and again saw no noticeable restoration of virulence (not shown). We and other laboratories have observed that for some mutant loci, reintegration of the wild-type gene (leading to restoration of gene function) is not difficult (for an example from this laboratory, see Braun and Johnson 1997). However, for other loci it is problematic. For example, deletion of the ASH1 gene provides a clear morphological phenotype on laboratory media (D. O. Inglis and A. D. Johnson, unpublished data). When a wild-type ASH1 gene is reintegrated at the mutant locus, as determined by PCR screening, only 1 in 48 independent reintegrant strains also shows restoration of the phenotype. The reason for this low success rate is unknown. In the case of genes whose deletion causes no detectable phenotypes outside of pathogenesis, such as RBT1 and RBT4, the prospect of testing many candidate reintegrants for those that restore virulence is impractical and somewhat circular in logic. To date we have constructed and analyzed in mice 6 independent rbt1Δ or RBT1/rbt1 mutant strains and 3 independent rbt4Δ or RBT4/rbt4 mutant strains. In all cases we observed a marked loss of virulence, strongly suggesting that the loss of virulence is due to disruption of the indicated gene and not to a secondary effect.
We have isolated a set of genes based on their overexpression in tup1Δ vs. wild-type C. albicans cells. Most of the isolated genes were also induced during filamentous growth in serum-treated wild-type cells, suggesting that, in C. albicans, the principal role of TUP1 may be the regulation of filamentous growth. On the other hand, genes such as RBT2 and RBT7, which are TUP1 regulated but show no filamentous growth regulation, indicate that TUP1 is also involved in separate regulatory pathways and therefore cannot be dedicated solely to filamentous growth regulation. These results also mean that TUP1 itself is not globally inactivated on the transition to filamentous growth. Each of the newly identified genes showed hallmarks of encoding secreted or cell surface proteins. Indeed, three (RBT1, RBT5, and WAP1) of the six appear to encode cell wall proteins, consistent with the idea that the cell wall of C. albicans is extensively remodeled during the switch from single-cell to filamentous growth (Alloushet al. 1996; Staabet al. 1996; Chaffinet al. 1998). Knockout strains for most of these genes showed no growth defects under a variety of artificial nutritional conditions, but those for RBT1 and RBT4 each exhibited strongly defective pathogenesis in both rabbit cornea and mouse tail-vein injection models of C. albicans infection.
There are two explanations for why cell surface and secreted proteins were exclusively obtained in this screen. First, such genes tend to be highly expressed. While the differential hybridization technique that we used in this study attempts to correct for the variable abundance of cDNA species, it remains most sensitive to abundant and highly differential molecules. The second reason lies in the nature of the TUP1 pathway itself. For all known instances of TUP1 repression in S. cerevisiae (with the exception of the initiation of meiosis), TUP1 acts at the end of a regulatory circuit, directly repressing a set of downstream effector genes as opposed to regulating a second tier of regulators (Keleheret al. 1992; Varanasiet al. 1996; Proft and Serrano 1999). Since remodeling of the cell wall appears to be a major aspect of the transition between blastospore and filamentous growth, we think it likely that TUP1 directly represses many of these cell wall protein-encoding genes. Previous work has indicated that the transition to filamentous growth involves not only a dramatic change in the cell's shape, but equally dramatic changes in the composition of the cell's exterior and secreted products (Alloushet al. 1996; Baileyet al. 1996; Staabet al. 1996; Chaffinet al. 1998; and see Braun and Johnson 2000, a related article). Our work supports this concept by identifying additional proteins of this class. Different environmental conditions may even produce different types of filaments that may look similar under the microscope, but that may display and secrete substantially different protein products. For example, conditions of starvation (such as Spider, milk-Tween, or cornmeal agar) and those that mimic the host environment (serum and 37°) each lead to filamentous growth, but may well produce filaments that differ in their molecular details (Braun and Johnson 2000).
We anticipated that the screen might yield genes that are required for filamentous growth, but recovered only one, HWP1 (Staabet al. 1999). No filamentation defects were observed when any of the other genes were deleted. In addition, overexpression of several of the cloned genes did not generate filamentous growth on noninducing media (data not shown). This result is consistent with the idea that only a fraction of the genes induced during filamentous growth are required for filament formation. In any case, our screen is incomplete, and we predict that many more TUP1-regulated genes exist in C. albicans. As an example, the recently described PLB1, which is a secreted phospholipase that contributes significantly to virulence in wild-type cells, is derepressed in tup1Δ cells (Hooveret al. 1998; Leidichet al. 1998).
Of the new genes we obtained, RBT7 and RBT2 have relatively clear functions, based on their sequence characteristics. RBT7 appears to encode a secreted RNase T2, while RBT2 encodes a plasma membrane ferric reductase. The probable substrates of each enzyme (RNA and Fe III, respectively) are known, and a general role can be surmised, which is nutrient acquisition. The other gene sequences provide much less specific information. RBT4 has clear homologs in a wide variety of plants as well as in S. cerevisiae, but little is known of their function. There have been reports of antifungal activity of the PR-1 class of these proteins (Nidermanet al. 1995), but the sequence similarity of RBT4 to these proteins is sufficiently low to prevent any definitive conclusion about its activities.
RBT1, RBT5, and WAP1 (as well as HWP1) all show attributes of fungal cell wall proteins: a signal sequence, a GPI-addition signal without a plasma-membrane retention signal, and one or more regions of short serine/threonine-rich repeats near the COOH terminus. According to the current “lollipop” model, these cell wall proteins are predicted to be heavily glycosylated on the serine/threonine-rich areas, stiffening these regions to allow the proteins to project through and above the cell wall (Jentoft 1990). According to this model the mature NH2 termini would thereby be exposed to the exterior of the cell. Aside from the limited structural similarity that RBT5 and WAP1 share with the antigenic C. immitis PRA protein, the regions of these proteins predicted to be exposed on the C. albicans cell surface show no significant similarities to other proteins in the database. It is particularly striking that the closely related HWP1 and RBT1 share high identity over their sequences everywhere except in this extracellular domain. The exposed domain of HWP1 has recently been reported to serve as a substrate for transglutaminases that crosslink C. albicans to host epithelial cells (Staabet al. 1999). One can predict that the exposed domain of RBT1 also interacts with host cells in some capacity, but is likely to have a function different from that of HWP1.
The results reported in this article also bear on the relationship between filamentous growth and virulence in C. albicans. A C. albicans strain carrying deletions of both CPH1 and EFG1 genes is defective in undergoing the blastospore to filament transition and is avirulent in the mouse tail-vein injection model (Loet al. 1997). Likewise a strain deleted for the TUP1 gene is locked in a filamentous form and is also avirulent in the same model (Figure 8A). On the surface it would appear as though both morphological forms are required for full virulence; since a mixture of these two strains is also avirulent (B. R. Braun and A. D. Johnson, unpublished data), the possibility exists that the transitions between these forms may also be required for full virulence. However, as pointed out by Brown and Gow (1999), Kobayashi and Cutler (1998), Mitchell (1998), and others, this level of discussion fails to take into account the many changes in gene expression that accompany the switch between morphological forms but are not central to it. In other words, it is possible that the “phase-locked” mutants are avirulent for reasons tangential to the morphological defect. For example, RBT1 and RBT4 are filament-specific genes that are necessary for full virulence but not for the morphological transition per se. Since EFG1 is necessary for the expression of both RBT1 and RBT4 (Braun and Johnson 2000), the avirulent phenotype of the efg1, cph1 strain could result simply from lack of expression of genes such as RBT1 and RBT4, which play no direct role in the morphological transitions. Likewise the tup1 strain could be avirulent, not because it is “locked” in the filamentous form, but because it expresses genes such as RBT1 and RBT4 constitutively. Although a strong correlation exists between the proficiency of the blastospore-filament transitions and virulence (Gow 1997; Mitchell 1998), a true assessment of the role of the morphological change in virulence requires an experiment (if indeed one is possible) that uncouples the morphological transitions from the many changes in gene expression that accompany them.
The RBT1 and RBT4 deletion strains were defective in two animal models of infection but showed no phenotypes under a variety of laboratory conditions including different growth media. These results indicate that these two genes are specifically important for success in the host. The rabbit cornea model of infection has the virtue of testing the invasion of C. albicans filaments into a living tissue and of allowing easy and detailed visualization of infection; the two mutant C. albicans strains showed reduced levels of infection and, more significantly, reduced extents of hyphal invasion. An important advantage of the mouse tail-vein model of infection is that it allows a comparison with virulence studies from other laboratories. At the inocula used in our experiments (see Figure 8), the RBT1 and RBT4 deletion strains were significantly reduced in virulence whereas the control strains resulted in rapid death. The latter result is consistent with published reports from other laboratories. Since the sequence characteristics of RBT1 predict that its product is displayed on the exterior of the cell wall, and those of RBT4 predict that its product is secreted, we speculate that their functions may be to interact with and impair the immune system components that normally clear C. albicans infections, to provide adhesive interactions that allow the fungal filaments to grow into the host tissue, or to directly damage host tissues. Since it seems unlikely that RBT1 and RBT4 are regulatory genes, the defects caused by their deletion may be very specific. In principle, both of these gene products could provide targets for antifungal therapeutics.
We are grateful for the guidance and advice of Denis M. O'Day in whose laboratory the rabbit eye model studies were carried out with support from his National Institutes of Health (NIH) grant EY01621 and a grant from Research to Prevent Blindness. The remainder of the work was supported by NIH grant GM37049 to A.D.J.
Communicating editor: A. P. Mitchell
- Received June 24, 1999.
- Accepted May 8, 2000.
- Copyright © 2000 by the Genetics Society of America