Chs1 of Candida albicans is an essential chitin synthase required for synthesis of the septum and for cell integrity

Revision of the CaCHS1 sequence

In the course of performing genomic Southern analyses, we found that the restriction map generated from our data did not match the 5′ region of the restriction map generated from the published CaCHS1 sequence (Au-Young and Robbins, 1990; GenBank accession no. X52420). To address this discrepancy, we recloned the CaCHS1 locus from CAI4 by chromosome walking (see Experimental procedures, pKW035). Sequencing of the newly isolated clone yielded an open reading frame (ORF) of 3078 bp encoding a predicted protein of 1026 amino acids, some 753 bp longer than the original published CaCHS1 sequence. Comparison of the two sequences reveals two major differences. First, X52420 is missing a cytosine residue at nucleotide 189; correction of this frameshift extends the CaCHS1 ORF upstream of the previously published start. Secondly, the first 47 nucleotides of X52420 do not match our newly derived sequence. Identity between the two sequences begins at nucleotide 48. Because this corresponds to a Sau3A site and because the plasmid pJA-IV, the source of X52420, was isolated from a Sau3A library, we believe that pJA-IV contains two non-adjacent chromosomal fragments. The new 3078 bp ORF was also confirmed in the sequence of a 42.5 kb C. albicans cosmid-derived contig Ca35A5 (strain 1161), containing the CaCHS1 gene (accession no. AL033396; Tait et al., 1997).

Disruption of CaCHS1

The first copy of the CaCHS1 gene was disrupted by transforming CAI4 with XhoI–AspI-cut plasmid pKW015. This plasmid contained CaCHS1, with 1806 bp of the 3078 bp ORF replaced with the ura-blaster cassette. Attempts to disrupt the second CHS1 allele by the same procedure failed. To test whether this failure resulted from natural heterozygosity at this locus, we devised a strategy specifically to target the remaining wild-type CaCHS1 allele. Polymerase chain reaction (PCR) primers corresponding to DNA deleted from the Δchs1::hisG allele were used to amplify the remaining wild-type copy of CaCHS1 from genomic DNA of the heterozygote KWC167b. The resulting 600 bp product was cloned into a CaURA3-containing plasmid to create pKW025. This plasmid therefore contained a fragment of CaCHS1 that was only homologous to the remaining non-disrupted allele. Plasmid pKW025 was digested with ClaI to target integration into that site in the remaining wild-type CaCHS1 allele and transformed into KWC167b. Again, no Δchs1/Δchs1 null mutants were isolated. Two disrupted chs1 alleles were observed, but these strains had become triploid at the CaCHS1 locus, and a wild-type CaCHS1 allele still remained (not shown).

Because the viability of S. cerevisiaeΔchs2 mutants was strain dependent, we attempted to generate Δchs1/Δchs1 null mutants in independent genetic backgrounds. CaCHS1 disruption was attempted in the isogenic strains SGY243 (Ura⁻) and NGY10 (Ura⁻, Δchs2::hisG/Δchs2::hisG/Δchs2::hisG;Gow et al., 1994). Southern analysis suggested that these strains were triploid at the CaCHS1 locus, as has been described previously for CaCHS2 in the SGY243 strain (Gow et al., 1994). Attempts to knock out the final wild-type copy of CaCHS1 in these strains were also unsuccessful. A variety of conditions was tested in an attempt to recover potentially slow-growing, debilitated homozygous Δchs1/Δchs1 null mutants. No condition was found in which Δchs1/Δchs1 null mutants were recoverable. These data strongly suggest that CaCHS1 is an essential gene.

Construction of a conditional CaCHS1 mutant

We created a conditional Δchs1 null mutant by placing the remaining wild-type allele under the control of the regulatable promoter MRP1. This bidirectional promoter was isolated from a C. albicans promoter-trap library and is located upstream of an ORF with significant homology to maltase genes (US patent 5824545). The MRP1 promoter is induced by maltose and repressed by glucose. Plasmid pKW044 was constructed, which contained CaURA3 adjacent to the MRP1 promoter fused at the ATG to a 3′-truncated CaCHS1 gene. Integration into the remaining wild-type allele resulted in an inactive 3′-truncated CaCHS1 and a full-length CaCHS1 ORF under the control of the MRP1 promoter. Ura⁺ transformants were selected on maltose, and their ability to grow on glucose was then tested. Southern analysis confirmed that pKW044 had integrated correctly to create conditional Δchs1/MRP1p-CHS1 mutants (data not shown). No wild-type allele was present in these strains. One of these transformants, KWC340, was selected for further analysis.

Northern analysis of CaCHS1 expression under the control of the MRP1 promoter

Regulation of CaCHS1 expression by the MRP1 promoter was investigated by Northern analysis. SC5314 (CHS1/CHS1) and KWC340 (Δchs1/MRP1p-CHS1) cells were grown overnight in SMal, washed and resuspended in SD or SMal, incubated at 30°C and samples removed at hourly intervals up to 6 h. Total RNA was extracted from these cells and probed with a CaCHS1-specific probe N. No CaCHS1 mRNA was detectable after 1 h or at later time points in KWC340 cells resuspended in glucose-containing SD medium. Indeed, CaCHS1 mRNA could not be detected after 30 min (data not shown). In contrast, when KWC340 was grown in maltose medium SMal, the CaCHS1 mRNA was detected at all time points at levels similar to those seen in strain SC5314 (Fig. 1).

Viability of the Δchs1::hisG/MRP1p-CHS1 mutant

Growth and cell viability of strain KWC340 was measured spectrophotometrically. Samples were taken at intervals from cells grown on SMal and SD medium, and a diluted aliquot of cells was also plated on SMal plates. Cells in both media continued to grow exponentially for 6–8 h after dilution into SD and SMal. The number of viable cells grown in SMal continued to increase exponentially in proportion to the measured OD₆₀₀. In contrast, the viability of cells grown in SD was always lower than the viability of SMal-grown cells, and the number of viable cells declined after 6–8 h in SD (Fig. 2). An initial increase in the OD₆₀₀ of SD-grown cells was observed, accompanied by a decrease in cell viability, as assessed by colony counts and an increasing proportion of propidium iodide-staining cells (not shown). These results indicate that cells growing under repressing conditions in SD became non-viable.

Function of CaCHS1 during growth in the yeast form

To examine the role of CaCHS1 during yeast cell growth, the Δchs1::hisG/MRP1pCHS1 conditional mutant KWC340 was grown overnight in SMal medium, washed and resuspended into either SD or SMal at 30°C. Yeast cells of KWC340 grown in SMal were indistinguishable from wild-type cells (Fig. 3A). Under glucose-repressing conditions, aberrant morphologies were observed within one or two generations (3–4 h). Although growth continued, cell number did not increase. Instead, chains were formed with obvious constrictions that are not compartmentalized by septa (Fig. 3B). New compartments in the chains were formed at approximately 2 h intervals, comparable with the normal generation time in this medium, suggesting that each swelling may represent a cell equivalent. As growth under repressing conditions continued transiently, lateral buds appeared from within chains, and branched chains and clusters of chains were seen (Fig. 3C and I). Compartments in chains were larger than wild-type yeast cells and were multinucleate, with up to five nuclei being observed in the apical compartment (Fig. 3D). Compartments towards the ends of the chains lysed progressively (Fig. 3E), and the appearance of lysed cells corresponded with loss of viability. Lysis was observed less frequently at the constricted regions between compartments in the chain. These observations indicate a progressive loss of cell wall integrity under repressing conditions in the conditional mutant.

Calcofluor-staining of KWC340 yeast cells grown under repressing conditions showed that they still formed distinct chitin-rich rings (synthesized by Chs3p) but had no primary septa (Fig. 3C–H). The Δchs1/MRP1-CHS1 mutant also formed aseptate pseudohyphae under these conditions with visible constrictions between cell junctions (Fig. 3F). Nuclei were able to pass between adjacent compartments of such constricted chains (Fig. 3G), again demonstrating that septa were incomplete or absent. This phenotype was also observed in the Δchs1/MRP1pCHS1;Δchs3/Δchs3 mutant (strain KWC359).

CaChs1p makes chitin in the septum and is essential for the integrity of the lateral cell walls of hyphae

Hyphal development was induced in 20% serum supplemented with either glucose or maltose. The Δchs1/MRP1p-CHS1 mutant formed normal hyphae for the first three or four generations in the presence of both glucose and maltose in liquid (Fig. 3H) and on solid media (Fig. 4A and B). Germ tube emergence rates and hyphal extension rates were not significantly altered during this period when strains KWC340 and KWC359 were grown on glucose or maltose (data not shown). Cells grown on glucose exhibited a progressively more aberrant morphology, with the lateral cell walls becoming expanded and hyphae developing regular septumless constrictions reminiscent of the morphology of the mutant grown under yeast-form conditions. Calcofluor-staining wall evaginations were observed to balloon out spontaneously from the cell surface (Fig. 4C and D). These balloons were observed at different positions on the hyphae (Table 1), at the apices (Fig. 5A), at points where branches emerged (Fig. 5B) and at constricted sites in the lateral cell wall that presumably represent potential sites of septation (Fig. 5C). In some cases, lysis of the balloons occurred (Fig. 5E, Table 1). Hyphal cells grown on glucose were also multinucleate (Fig. 5F and G). The effect of depleting cells of CaChs1p was also examined in strain KWC359, which had both copies of CaCHS3 disrupted. The Δchs1/MRP1pCHS1;Δchs3/Δchs3 mutant also displayed the aberrant morphologies described above, but lysed more frequently at the hyphal apex and subapically (Fig. 5D and J, Table 1). The balloons formed in the double mutant stained with Calcofluor, indicating that chitin was still present within these regions (Fig. 5H and I). No further growth was observed after cells had formed balloons or had lysed. The addition of 1 M sorbitol as an osmotic support (Fig. 6A–F) did not remediate the ballooning or lysis phenotype. The addition of 1 mM allosamidin, an inhibitor of chitinase (Fig. 6G), did not prevent lysis or the formation of balloons, suggesting that the loss of cell integrity resulted from the loss of chitin synthesis alone and was not augmented by chitinase action.

Ultrastructural analysis of the Δchs1/MRP1p-CHS1 mutant

Examination of thin sections of glucose-repressed KWC340 cells by transmission electron microscopy (TEM) confirmed that there were no septa separating chains of yeast cells and the compartments of hyphae (Fig. 7A and D). Yeast cells grown on SMal medium formed normal septa between the mother and bud. These septa could be labelled with wheatgerm agglutinin–colloidal gold (WGA-Au), indicating the presence of chitin (Fig. 7B). Constrictions in chains of KWC340 grown on SD medium were positive for WGA-Au labelling, showing that chitin rings were present, but abnormally thickened (Fig. 7A and C). Hyphae of strain KWC340 grown in the presence of maltose formed septa that showed a normal pattern of WGA-Au labelling (Fig. 7E). KWC340 hyphae grown under repressing conditions were labelled with WGA-Au only in the lateral walls and in areas where the chitin ring was abnormally thickened (Fig. 7F). These observations are consistent with the view that Chs1p synthesizes chitin of the primary septum in both yeast and hyphal cells.

The conditional Δchs1/MRP1p-CHS1 mutant is avirulent

To examine the role of CaCHS1 in infection by C. albicans, strain KWC340 was assessed in a systemic murine virulence model. Three control strains were included (Table 2): SC5314 (CHS1/CHS1), the clinical isolate that is the parent of the strains used in this study; KWC352 (CHS1/MRP1pCHS1); and MRP2 (Δura3/MRP1p-URA3), a strain that has a single copy of the CaURA3 gene under the control of MRP1p. The five animals infected with SC5314 or KWC352 died within 8 days (for neutropenic mice) or 11 days (for immunocompetent mice). The wild-type virulence of KWC352 indicates that the MRP1p–CHS1 fusion was stable in vivo and that looping out of the duplicated CHS1 to create a Ura⁻ strain did not occur at high frequency. This was unlikely because auxotrophic strains are known to be avirulent (Kirsch and Whitney, 1991; Polack, 1992). All mice injected with KWC340 or MRP2 survived for at least 20 days after inoculation (Fig. 8). Therefore, MRP2 is attenuated in virulence like Ura⁻ auxotrophs, indicating that expression from the MRP1 promoter is switched off in vivo. To investigate further the survival of KWC340-infected animals, these mice were necropsied, and the kidneys were removed and patched onto SMal and YPD plates. No colonies were recovered, indicating that KWC340 was efficiently cleared from tissues. Thus, CaCHS1 is essential for growth in mice in systemic infections.

CaChs1p synthesizes chitin in the primary septum

Calcofluor staining and WGA-Au labelling of the cell wall in TEM thin sections suggested that CaChs1p is responsible for synthesis of the chitinous primary septal plate in both yeast and hyphal forms of C. albicans. Therefore, septum formation in yeast and hyphal cells shares at least some of the same biosynthetic machinery. The implication that septal chitin was synthesized by CaChs1p was also evident in analyses of the C. albicansΔchs2/Δchs2;Δchs3/Δchs3 double null mutant, in which Calcofluor still stained septal plates (Mio et al., 1996). The ballooning phenotype of depleted CaChs1p hyphal cells at variable locations including the apex and subapex suggested that CaChs1p also contributes to lateral cell wall synthesis. This could not be predicted from studies of the ΔScchs2 mutant, which does not exhibit these dramatic phenotypes. The sites of ballooning had a different distribution between CaChs1p-depleted Chs3⁺ and Chs3⁻ hyphal cells. In Chs3⁻ cells, a higher percentage of cells had defects at the apex or subapically, whereas in the Chs3⁺ cells, 50% of the cells had defects at the point where a branch emerges. These observations suggest that CaChs1p synthesizes some chitin in the lateral cell wall, and that this chitin is essential for maintaining the integrity of the cell wall. Chitinase enzymes may be active at some of the cellular sites at which these defects were apparent, and chitinases are thought to be responsible for cell wall softening to allow branches to emerge in most filamentous fungi (Gooday et al., 1992). We tested the hypothesis that chitinases may antagonize the Δchs1 phenotype by growing the cells in the presence of the chitinase inhibitor allosamidin. Neither allosamidin nor osmotic support using sorbitol was able to remediate the ballooning phenotype, suggesting that C. albicans has no compensatory mechanism for the loss of CaChs1p. The inability to rescue the mutant with sorbitol suggests that CaChs1p is required for more than just cell wall integrity. The expression of ScCHS2 is cell cycle regulated, and Δchs2 mutants had multiple nuclei (Bulawa and Osmond, 1990; Pammer et al., 1992; Choi et al., 1994). Similarly, depletion of CaChs1p resulted in the loss of co-ordination between the budding and nuclear cycles. Compartments formed by yeast cells of the Δchs1/MRP1pCHS1 mutant under repressing conditions were larger, elongated and nuclear division was uncoupled from cytokinesis. Often, the terminal compartment in a chain had multiple nuclei, greater in number than the number of compartments in the chain. Hyphal cells were also multinucleate. Therefore, nuclear division and segregation were also affected by loss of CaChs1p. It is not known whether CaChs1p has a direct effect on nuclear division or whether these effects are a consequence of the inability to undergo cytokinesis. Mutants in S. cerevisiae with cytokinesis defects, e.g. Δshs1 andΔhof1, also have increased cell size and are multinucleate (Mino et al., 1998; Vallen et al., 2000). The increased cell size may indicate that CaChs1p-deficient cells have a prolonged G₁ phase. The budding pattern was also affected in the conditional Cachs1 mutant, in which the normal bipolar budding pattern was abolished, and growth occurred at the distal pole of the daughter cell.

The fact that Chs1p-deficient cells undergo cell lysis suggests that this mutation results in significant weakening of the integrity of the cell wall. In the yeast form, lysis occurred at the constricted neck region between two compartments. In the hyphal form, sites of constrictions were also most prone to lysis, suggesting that the absence of the primary septal plate increased the fragility of the cell wall at these junctions. Chitin in the lateral cell wall was often thickened at these constricted regions, suggesting that an alternative chitin synthase, possibly Chs3p, may have been activated locally. A range of S. cerevisiae mutants has been shown to elevate chitin synthesis in response to cell wall weakening to restore the structural integrity of the wall (Popolo et al., 1997; Kapteyn et al., 1999a, b). This may be mediated via the cell wall integrity MAP kinase cascade (Smits et al., 1999). Homologues of this pathway have been identified in C. albicans (Navarro-Garcia et al., 1998). Therefore, chitin ring formation may be regulated by the presence or absence of the primary chitinous septum, perhaps through local induction of the cell wall integrity response. In the Δchs1/MRP1pCHS1;Δchs3/Δchs3 strain, the swollen balloons still stained with Calcofluor, suggesting that another chitin synthase, e.g. CaChs2p, may be responsible for chitin synthesis in these structures. Mutant analyses show that this enzyme makes around 10% of the chitin in the hyphal form (Gow et al., 1994; Munro et al., 1998).

Hyphae of C. albicans normally branch behind positions of septa. It was evident from analysis of the Δchs3/Δchs3 mutant that septum formation can proceed without the presence of the chitin ring synthesized by CaChs3p and, in the Δchs1/MRP1pCHS1;Δchs3/Δchs3 strain, branch formation occurs without the septal plate synthesized by CaChs1p and the CaChs3p-derived chitin ring. Loss of CaChs1p results in weakening of the cell wall at these locations and, although branches are formed, many of the cells lyse or balloon at the point where branches emerged. Therefore, CaChs1p must play a vital role in maintaining cell integrity at the hyphal–branch junction and at the tips of growing cells.

Shadow-cast TEM analysis revealed that chitin in the primary septum of C. albicans is more fibrillar than the short rodlets of chitin in the lateral cell wall (Gow et al., 1980; Gooday and Gow, 1983). This suggests that the chitin synthesized by CaChs1p differs in length and crystallinity from the chitin synthesized by CaChs3p. Regulation of the hydrogen bonding and crystallization of nascent chitin microfibrils in different locations in the cell wall is not understood, but may lead to chitin microfibrils being formed with significantly different structural and tensile properties.

Chitin synthase isoenzymes are differentially sensitive to chitin synthase inhibitors such as the nikkomycins and polyoxins (Cabib, 1991; Gaughran et al., 1994). In S. cerevisiae, septum-synthesizing ScChs2p is the least sensitive chitin synthase isoenzyme to both these classes of antifungals. Our studies of CaChs1p demonstrate that this isoenzyme is the most important enzyme for viability of C. albicans in vitro and in infected animals. Many antifungal screens have been based on the inhibition of total cellular chitin synthase activity, the majority of which results from ScChs1p/CaChs2p, and CaChs3p, neither of which are required for viability (Gow et al., 1994; Bulawa et al., 1995; Mio et al., 1996; Munro et al., 1998). This suggests that CaChs1p is the most important target for screens for novel Chs inhibitors directed against Candida species. One such inhibitor was recently reported (Masubuchi et al., 2000) that selectively inhibits CaChs1p and shows in vitro antifungicidal activity against a number of Candida spp. and mild in vivo efficacy in a murine systemic candidiasis model. Treatment of C. albicans cells with this inhibitor resulted in growth arrest, with the yeast cells forming chains indicative of loss of septa (Sudoh et al., 2000). This compound was only weakly active against Aspergillus fumigatus. CaCHS1 homologues have not yet been identified in filamentous fungi, predicting that such compounds may be most active against ascomycetous yeast-like pathogens. The finding that individual members of CHS gene families may be essential for viability highlights the need to understand the specific roles of Chs isoenzymes in fungal growth and their relative importance as potential drug targets.

Strains, media and growth conditions

The C. albicans strains used in this study are listed in Table 2. C. albicans Ura⁻ strains were grown in YPD [1% (w/v) yeast extract, 2% (w/v) peptone, 2% (w/v) glucose] at 30°C (Sherman, 1991). Ura⁺ strains were grown in either SD [0.67% (w/v) yeast nitrogen base without amino acids, 2% (w/v) glucose] or SMal [0.67% (w/v) yeast nitrogen base, 2% (w/v) maltose] at 30°C (Sherman, 1991). Solid medium contained 2% (w/v) purified agar (Oxoid). Counterselection of Ura⁻ segregants was on SD medium containing 1 mg ml⁻¹ 5-fluoroorotic acid (5-FOA) and 100 µg ml⁻¹ uridine (Fonzi and Irwin, 1993).

Transformation of C. albicans

Ura⁻C. albicans cells were transformed after treatment with lithium acetate (Bulawa et al., 1995) or using a spheroplast method (Kurtz et al., 1986). Ura⁺ transformants were streaked on SD plates, single colonies were picked and grown in 5 ml of SD medium, and genomic DNA was extracted for Southern analysis.

Construction of plasmids and C. albicans strains

A 9.2 kb SalI–NheI fragment containing CaCHS1 and 200 bp of YEp13 sequence was subcloned from plasmid pJA-IV (Au-Young and Robbins, 1990) into the SalI–XbaI sites of pBluescript SK– producing PKW013. Plasmid pKW014 was generated from plasmid pKW013 by cutting with XhoI and ligating the digested DNA in a large volume to promote intramolecular ligations. pKW014 contained 3.5 kb of downstream CaCHS1 sequence with a 5.5 kb upstream XhoI fragment deleted.

The 4.5 kb PstI–AflIII fragment of pMB-7 containing the ura-blaster cassette (Fonzi and Irwin, 1993) was inserted into the 5′NsiI and 3′NcoI sites of the CaCHS1 ORF in plasmid pKW014 to create plasmid pKW015, which was used for the disruption of the CaCHS1 gene. This resulted in the deletion of nucleotides 921–2726 of CaCHS1 (GenBank accession no. X52420). Plasmid pKW015 was cut with XhoI and AspI, leaving 540 bp of upstream and 613 bp of downstream CaCHS1 sequence for targeted integration, and the linearized cassette was introduced into C. albicans Ura⁻ strain CAI4 and a Ura⁻ derivative of CACB8B-5 (Δchs3/Δchs3) by electroporation.

For the construction of the conditional CaCHS1 mutant, a 1.7 kb PCR product (nucleotides 271–2030; accession no. X52420) was first amplified from the CaCHS1 locus of strain CAI4 using primers CaCHS1-1 5′-GATGATAACAATCACACC (nucleotides 271–288) and CaCHS1-2 5′-GACACCAATGAACCTGTC (nucleotides 2013–2030). The amplified product was cut with Asp718 and EcoRI, and the resulting 0.6 kb fragment was ligated into the Asp718–EcoRI sites of pBluescript SK– to generate plasmid pKW024.

The selectable marker, CaURA3, was amplified by PCR from plasmid pKW015 using primers 5′-ATATCCGCGGTACTAATAGGAATTGATTTG (nucleotides 3–22 of GenBank accession no. X14198; KspI site underlined) and (5′-CTAGAGCTCCCACCTTTGATTGTAA (nucleotides 1355–1340; SacI site underlined). The resulting 1.35 kb SacI–KspI fragment was subcloned into the SacI–KspI sites of pKW024 to generate plasmid pKW025. The same SacI–KspI fragment was subcloned into pBluescript SK to form plasmid pKW016.

To recover the 5′ end of CHS1, plasmid pKW025 was cut with ClaI and transformed into CAI-4. Transformants were analysed by Southern analysis of genomic DNA, and strain CAI-4A, which contained pKW025 integrated at the ClaI site of CHS1, was identified. CAI-4A DNA was cut with HindIII, diluted, ligated and transformed into Escherichia coli. Plasmids were screened for CHS1 sequence by PCR using primer CaCHS1-1b, 5′-GGCGCAGCACACCAAGAGCAC-3′ (base numbers 1840–1861) and the 3′ primer CaCHS1-3 5′-GGCGTGAGCACAAATGAGCG-3′ (the complement of base numbers 2598–2617). Clones producing the expected 780 bp product were confirmed as having CHS1 sequence. Plasmid pKW030 was one such clone identified.

The entire CHS1 locus was reconstructed by ligating the 3.6 kb HindIII–PstI fragment of pKW030 and the 3.5 kb BstEII–NotI fragment from plasmid pKW013 into pBluescript SK. The resulting plasmid, pKW035, contains a 5.3 kb fragment of C. albicans DNA including the entire CHS1 ORF.

A 1 kb portion of the maltose-regulatable promoter MRP1p was amplified by PCR with primers Malp-1 (BamHI site underlined), 5′-ATAGGATCCTCACCAATTCCATCAC, and Malp-2 (with an RcaI site at the ATG), 5′-CTTAGTGTTGACTGTCATGACGATC.

The MRP1pCHS1 cassette was finally constructed using a three-way ligation that included: (i) a 2 kb CaCHS1 RcaI–Asp718 fragment (nucleotides −1 to 2084) from plasmid pKW035; (ii) a 4.4 kb BamHI–Asp718 fragment from pKW016 carrying CaURA3 and pBluescript; and (iii) a 1 kb BamHI–RcaI-cleaved MRP1p PCR product. This generated plasmid pKW044, containing the MRP1 promoter fused at the CaCHS1 start codon to a 3′ truncated CaCHS1 lacking 1610 bp of the ORF, in pBluescript SK– with URA3 as the selectable marker. Plasmid pKW044 was cut with XhoI before transformation into the CHSI/Δchs1::hisG Ura⁻ strain to direct integration into the corresponding XhoI site of the chromosomal copy of CaCHS1, thus generating the conditional mutant.

To examine the effect of strain background on the viability of CaCHS1 disruptants, a further series of heterozygous CHS1/Δchs1::hisG mutants was created. This alternative CaCHS1 disruption cassette was constructed from a 4 kb XbaI–PstI fragment containing the CaCHS1 gene derived from plasmid pJA16 (Au-Young and Robbins, 1990). This was ligated into XbaI- and PstI-cut pUC18 to form plasmid pAB1. The hisGURA3hisG cassette was excised from plasmid p5921 (Gow et al., 1994) with BamHI and BglII and blunt ended using mung bean nuclease (Boehringer Mannheim, Roche Diagnostics). Plasmid pAB1 was cleaved with ClaI and the ends rendered blunt by treatment as before. The hisGURA3hisG fragment was ligated into the ClaI-cut and blunted pAB1 to form plasmid pAB9 (co-ordinates 1158–2555 deleted). The CaCHS1 disruption cassette was linearized with PvuII before transformation into strains CAI4, SGY243 and NGY10 (see Table 2 for genotypes) by the spheroplast method (Kurtz et al., 1986).

Northern analyses

Total RNA was extracted from C. albicans, and Northern blots were prepared according to the protocol of Hube et al. (1994). A 533 bp RcaI–XhoI fragment corresponding to positions −1 to 532 of the CaCHS1 ORF from plasmid pKW035 was radiolabelled with [α-³²P]-dCTP using the random-primed labelling kit (Boehringer) to give the CHS1-specific probe N. A 0.7 kb fragment representing co-ordinates 126–829 bp of the C. albicans TEF3 coding region cloned in pUC19 (Colthurst et al., 1992) was used as a positive control for Northern analysis. The hybridization conditions described by McCreath et al. (1995) were used.

Light microscopy

Yeast cells were examined either directly or after fixation in 10% (v/v) neutral-buffered formalin (Sigma). Hyphal cells were grown on poly-l-lysine-coated microscope slides (Crombie et al., 1990) placed in Petri dishes containing 20% (v/v) newborn calf serum (Sigma) plus 2% (w/v) maltose or glucose. Cells were mixed with a drop of 50 µg ml⁻¹ Calcofluor (American Cyanamid) or 2 µg ml⁻¹ DAPI (Sigma) after permeabilization in 70% (v/v) ethanol and examined by fluorescence microscopy. A slide culture technique was used for time-lapse microscopy of hyphal cells. A sterile microscope slide was placed in a Petri dish and overlaid with 9.5 ml of 20% (v/v) serum agar containing 2% (w/v) glucose or maltose. The agar was inoculated with 1 × 10⁵C. albicans cells, a coverslip placed on the slide and excess agar removed. The coverslip was then sealed to the slide with 1:1:1 Vaseline–lanolin–paraffin (Kron et al., 1994). The slide was incubated at 37°C on the heated stage of a Zeiss microscope, and the cells were examined using Nomarski differential interference optics over a 24 h period.

WGA-colloidal gold staining of chitin

Cells were grown as described above, harvested after 6 h and fixed in 2.5% (v/v) glutaraldehyde in 0.1 M sodium phosphate, pH 7.6, and stored at 4°C. The cells were washed three times in 0.1 M sodium phosphate, pH 7.6, and fixed with 1% osmium tetroxide in the same buffer for 1 h. Cells were washed three times in phosphate buffer and embedded in 1% (w/v) molten agarose. Blocks of samples were dehydrated first in graded ethanol solutions and then in polypropylene oxide before embedding in Araldite resin (Agar Scientific). Thin sections (60–80 nm) were prepared, mounted on 300 mesh gilded grids (Agar Scientific) and labelled with 10 nm WGA-gold particles (British Biocell International). The grids were immersed for 1 h in WGA-gold, diluted 1:5 or 1:10 with Tris-buffered saline (TBS). To control for non-specific binding of colloidal gold, grids were incubated with a 1:10 dilution of goat F(ab′)₂ anti-mouse IgG/IgM conjugated to colloidal gold (British Biocell International). Grids were dipped in TBS, rinsed in TBS and then washed in dH₂O. The thin sections were post-stained with 5% (w/v) aqueous uranyl acetate and lead citrate (Reynolds, 1963) and examined in a Jeol 1200ExB Temscan electron microscope operated at 120 kV.

Cryogenic scanning electron microscopy (SEM)

Stationary phase yeast cultures were washed in sterile dH₂O, spread over the surface of sterile polycarbonate Nucleopore membranes (Whatman; diameter 47 mm, pore size 2 µm) that had been overlaid on agar plates and incubated overnight. Membranes were lifted from the plate, submerged in 10% (v/v) neutral-buffered formalin (Sigma), rinsed in phosphate-buffered saline (PBS) and then dH₂O. Samples of membrane with cells attached were prepared using the Oxford CT1500C Cryotrans system according to the manufacturer's recommendations. Each sample was cooled in nitrogen slush and transferred under vacuum to the cold stage of the SEM. Samples were coated with platinum and examined on the cryomicroscope stage at −180°C in a Jeol JSM-35CF SEM operated at an accelerating voltage of 10 kV.

Virulence tests

Candida albicans strains were grown in SMal at 30°C, harvested, washed and resuspended in sterile water to the required density. Five male ICR mice (18–22 g; Harlan–Sprague–Dawley) per test were injected with 10⁶ cells via the lateral tail vein. Neutropenic mice were also tested. These were injected intraperitoneally with 150 mg kg⁻¹ cyclophosphamide 96 and 24 h before challenge with 10⁴C. albicans cells and subsequently at 3 day intervals. Neutropenia was confirmed by comparing the ratio of neutrophils with the total number of leucocytes before and after administration of cyclophosphamide. Checks were made for dead or moribund mice three times daily for 20 days after infection. Moribund mice were euthanized by cervical dislocation and necropsied.