Cas1p is a membrane protein necessary for the O-acetylation of the Cryptococcus neoformans capsular polysaccharide

Authors


Abstract

The capsule is certainly the most obvious virulence factor for Cryptococcus neoformans. The main capsule constituents are glucuronoxylomannans (GXM). Several studies have focused on the structure and chemistry of the GXM component of the capsule, yet little is known about the genetic basis of the capsule construction. Using a monoclonal antibody specific to a sugar epitope, we isolated a capsule-structure mutant strain and cloned by complementation a gene named CAS1 that codes for a putative membrane protein. Although no sequence homology was found with any known protein in the different databases, protein analysis using the Propsearch software classified Cas1p as a putative glycosyltransferase. Cas1p is a well-conserved evolutionary protein, as we identified one orthologue in the human genome, one in the drosophila genome and four in the plant Arabidopsis thaliana genome. Analysis of the capsule structure after CAS1 deletion showed that it is required for GXM O-acetylation.

Introduction

Cryptococcus neoformans is an opportunistic fungal pathogen that mainly infects immunocompromized patients, causing life-threatening infections of the central nervous system. This basidiomycete yeast exists in three varieties: C. neoformans var. grubii (serotype A) (Franzot et al., 1999) with a worldwide distribution, C. neoformans var. neoformans (serotype D), more frequently found in Europe, and C. neoformans var. gattii (serotypes B and C), which is limited to tropical and subtropical regions (Kwon-Chung and Bennett, 1984). C. neoformans cells have a thick extracellular polysaccharide capsule, which is one of the most important virulence factors along with the production of melanin (Casadevall and Perfect, 1998).

The capsule polysaccharides are composed of glucuronoxylomannans (GXM, 88% of its mass), galactoxylomannans (GalXM) and mannoproteins (MP) (Cherniak and Sundstrom, 1994). The GXM have been well characterized chemically and are the antigenic basis for serotype specificity. They consist of a (1→3)-α-d-mannopyranan backbone bearing β-d-xylopyranosyl, β-d-glucopyranosyluronic acid and 6-O-acetyl substituents. Variations of glucuronic acid and xylose substitution are used to characterize different chemotypes (Cherniak et al., 1998). The capsule protects C. neoformans cells against phagocytosis, alters antigen presentation, affects the production of cytokines, can be responsible for complement depletion in the host and inhibits leucocyte migration into infected sites (for review see Buchanan and Murphy, 1998). Intravenous injection of capsule polysaccharide prior to infection with C. neoformans shortens the survival of treated compared with control mice (Bennett and Hasenclever, 1965).

Recently, Chang and Kwon-Chung (1994; 1998; 1999) and Chang et al. (1996) cloned four capsule-associated genes, CAP10, CAP59, CAP60 and CAP64, and showed that the specific deletion of any one was associated with loss of virulence and an acapsular phenotype. However, the authors did not determine the specific biochemical functions of these genes even though they are necessary for capsule synthesis. In fact, no gene coding for components of the biosynthetic pathways leading to the complete capsule structure has been identified (Doering, 2000). A number of mutants producing a thicker or thinner capsule than the wild type have been described in the literature, but no evidence of a structural modification of their capsules have been reported (Jacobson et al., 1982; Jacobson and Tingler, 1994). UDP glucuronate decarboxylase, UDP xylosyltransferase, UDP glucuronyltransferase, UDP glucose dehydrogenase and, more recently, α-(1→3) mannosyltransferase activities has been identified in C. neoformans extracts, but their involvement in the biosynthesis of the capsule remains to be clearly demonstrated (Jacobson and Payne, 1982; Jacobson, 1987; White et al., 1990; Doering, 1999).

Little is known about the impact of the capsule-polysaccharide structure on the pathophysiology of cryptococcosis, even though several lines of evidence suggest its effect on the virulence of C. neoformans. In vivo in humans, serotypes A versus D, as well as serotype A versus B, differ in terms of the type of infection produced (Speed and Dunt, 1995; Dromer et al., 1996). In vitro, various GXM produce various effects in terms of antiphagocytic activity (Small and Mitchell, 1989) or induction of cytokine synthesis by different host cells (Chaka et al., 1997; Lipovsky et al., 1998). Finally, capsule structure seems to affect the in vitro binding of C3 to C. neoformans cells (Washburn et al., 1991; Young and Kozel, 1993). However, how the capsule influences yeast virulence is unknown.

Our investigation focused on the identification of genes involved in capsule biosynthesis in order to study the relationship between the C. neoformans capsule structure and virulence. Herein, we describe the cloning of the first capsule synthesis gene, CAS1, coding for an enzyme, whose activity is necessary for the synthesis of O-acetyl residues on the GXM structure.

Results

Cloning of CAS1

The isolation and description of acapsular, hypocapsular and hypercapsular mutants were previously reported, but no C. neoformans mutant strain synthesizing a structure modified capsule has been described (Jacobson et al., 1982; Jacobson and Tingler, 1994; Chang et al., 1996). Using an enzyme-linked immunosorbent assay (ELISA) and the GXM-specific monoclonal antibody CRND-8, we isolated a mutant strain (NE28) (Table 1) that synthesized a capsule with a modified structure (Fig. 1).

Table 1. C. neoformans strains used in this study.
StrainsGenotypeSerotypeSource or reference
JEC21 MATαD(Moore and Edman, 1993)
JEC20 MAT a D(Moore and Edman, 1993)
JEC156 MAT a ade2 ura5 D(Wickes and Edman, 1995)
JEC43 MATαura5D(Wickes and Edman, 1995)
NE23 MATαcas1–1DThis study
NE28 MATαcas1–1 ura5DThis study
H99 MATαAJ. Heitman
TYCC33 MATαcap59-Δ::ADE2 ura5D(Chang and Kwon-Chung, 1994)
TYCC77 MATαcap64-Δ::ADE2 ura5D(Chang et al., 1996)
TYCC122 MATαcap60-Δ::ADE2 ura5D(Chang and Kwon-Chung, 1998)
TYCC150 MATαcap10-Δ::ADE2 ura5D(Chang and Kwon-Chung, 1999)
NE30 MAT a cas1-Δ::ADE2 ura5 DThis study
NE31 MAT a cas1-Δ::ADE2 ura5 DThis study
JEC33 MATαlys2D(Wickes and Edman, 1995)
JEC50 MAT a ade2 D(Wickes and Edman, 1995)
JEC52 MATα lys2 ura5 D(Wickes and Edman, 1995)
NE147 MAT a CAS1 DThis study
NE148 MAT a cas1-Δ::ADE2 DThis study
NE149 MATα CAS1 DThis study
NE150MATα cas1-Δ::ADE2DThis study
NE167MATα CAS1DThis study
NE168MATα cas1-Δ::ADE2DThis study
NE169 MAT a CAS1 DThis study
NE170 MAT a cas1-Δ::ADE2 DThis study
Figure 1.

A and B. Comparison of the strains cas1–1 (NE28) and CAS1 (JEC43) by light microscopy.

C and D. Immunofluorescence labelling with the monoclonal antibody CRND-8.

E and F. India ink negative-staining of the capsule.

The CAS1 gene was isolated by complementing the strain NE28′s inability to react with monoclonal antibody CRND-8. The seven CRND-8-positive transformants yielded plasmids with inserts 5.5–8 kb long. Restriction maps of the cloned fragments, subcloning and complementation experiments established that the gene of interest was located on a 4.5 kb BamHI/BamHI fragment. Sequencing of the whole fragment revealed two large open reading frames (ORF) of 758 and 1064 bp, which could correspond to two large exons of the gene. Using two primers each specific to one of these open reading frames (ORF), we amplified, cloned and sequenced the complete corresponding cDNA fragment (see Experimental procedures). This sequence contained a 2880 bp ORF which encoded a putative 108 kDa protein. Alignment of this sequence with the genomic DNA sequence revealed the presence of eight short introns. The polyadenylation site was located 71 nucleotides downstream from the stop codon. By sequencing five independent clones after 5′ rapid amplification of cDNA ends (RACE) (see Experimental procedures), it was determined that the 5′ end of the transcript was located between −44 and −60 upstream from the ATG. We cloned the cas1–1 allele from the strain NE28 by hybridization of a partial genomic library using a CAS1-specific probe. The analysis of the nucleotide sequence revealed a frameshift mutation at the beginning of the third exon, 453 nucleotide downstream the ATG, introducing a stop codon at the nucleotide 558 (data not shown).

Comparison of the sequence with the available databases revealed no homology with any known protein. However, we identified one similar sequence in the human genome located on the chromosome 7. The complete corresponding cDNA from a human placenta cDNA library was amplified by polymerase chain reaction (PCR) and the complete sequence of the human orthologue (hsCAS1) was determined. The amino acid sequence was 22% identical and 51% similar to that of the C. neoformans protein sequence (see alignment of the sequences on Fig. 2). The Drosophila melanogaster genome also contained one orthologue gene (locus CG2938) located on chromosome X (19% identical and 46% similar). We also identified four orthologue genes on the Arabidopsis thaliana genome on chromosomes I, II, III and V. The partial sequences of these proteins were very similar to each other (90% similarity) and shared an average 70% similarity with the C. neoformans protein sequence. All the similarities between the different orthologues were mainly localized within the central region of the protein sequences.

Figure 2.

Alignment of the amino acids sequences of the human (hsCas1p) and the C. neoformans (cnCas1p) orthologues indicating identical (*), strongly similar (:) or weakly similar (.) residues.

The analysis of the cnCas1p amino acid sequence revealed the presence of two putative N-glycosylation sites in positions 3–6 and 169–172. Moreover, the sequence AARKVGKT in position 124–131 matched the ATP/GTP-binding site motif A (P-loop) consensus sequence [AG]-x(4)-G-K-[ST] characteristic of some ATP- or GTP-binding proteins (Walker et al., 1982) that can be present in some proteins binding other nucleotides. However, this sequence was not conserved in the different Cas1p orthologues and might not be functional in this protein.

According to the program tmhmm1.0 (Sonnhammer et al., 1998), Cas1p is a membrane protein with 12 putative transmembrane domains. As shown in Fig. 3, the human and fly orthologues also resembled membrane proteins with 15 putative transmembrane domains. The positions of these domains are very well conserved, with an isolated N-terminal domain and 11 central domains. In cnCas1p, however, the C-terminal is predicted to be cytoplasmic whereas the C-terminal of the human and fly orthologues contained three more transmembrane spannings.

Figure 3.

Hydropathy profiles of cnCas1p, hsCas1p and dmCas1p determined by TMHMM1.0 analysis.

We used the Propsearch software to analyse the Cas1p sequence. Among the 12 best hits obtained, eight were glycosyltransferases (Table 2). Thus, this analysis indicated that Cas1p is of the same family as the Pmt mannosyltransferases responsible for the first step of Saccharomyces cerevisiae mannoprotein O-glycosylation (review in Strahl-Bolsinger et al., 1999).

Table 2. Protein identified belonging to the same family as Cas1p by Propsearch analysis.
ProteinOrganismDistanceFunctionReference
MdoH E. coli 8.23Glycosyltransferase(Loubens et al., 1993)
YhjO E. coli 8.25Glycosyltransferase(Sofia et al., 1994)
Pmt2 S. cerevisiae 8.71Glycosyltransferase(Lussier et al., 1995)
EnvBIV9.43Coat polyprotein(Garvey et al., 1990)
HrpM P. syringae 9.52Glycosyltransferase(Mukhopadhyay et al., 1988)
EnvBIV9.68Coat polyprotein(Garvey et al., 1990)
Tscc H. sapiens 10.04Na-Cl cotransporter(Simon et al., 1996)
Isp4 S. pombe 10.15Peptide transporter(Sato et al., 1994)
Pmt3 S. cerevisiae 10.17Glycosyltransferase(Immervoll et al., 1995)
Pmt1 S. cerevisiae 10.35Glycosyltransferase(Strahl-Bolsinger et al., 1993)
BcsA A. xylinum 10.38Glycosyltransferase(Standal et al., 1994)
AcsA A. xylinum 10.43Glycosyltransferase(Wong et al., 1990)

CAS1 of the different serotypes

Southern blotting experiments using CAS1 as the probe revealed that a homologue of this gene was also present in the genomes of serotypes A, B and C (data not shown). We cloned the serotype A CAS1 homologue by colony hybridization of a PstI partial genomic library of the strain H99. The sequence of this gene was very similar to its serotype D counterpart (94% identity and 99% similarity at the amino acid level). Moreover, the serotype A CAS1 gene was put under the control of the serotype D promoter and the construct was used to transform strain NE28. The resulting colonies were all CRND-8-positive, showing that the serotype A CAS1 gene could complement the phenotype of the serotype D mutant. The same probe was used to identify the location of CAS1 in C. neoformans serotype D and A genomes after PFGE separation of the chromosomes of strains JEC21, JEC20 and H99. The probe hybridized to only one of the two smallest chromosomes of the strains JEC21 and JEC20, and the smallest chromosome of strain H99 (data not shown). Thus, probably only one copy of this gene was present in the genome. No close homologue could be revealed by Southern blot analysis with either high or low stringency conditions. Moreover, a blast search against the partial genome sequence from the ‘Cryptococcus neoformans Genome Project’ at the Stanford Genome Technology Center (http://sequence-http://www.stanford.edu/group/C.neoformans/index.html) did not reveal any CAS1 homologous sequence other than CAS1 itself.

Regulation of CAS1 transcription

Previous studies have shown that at least four genes are necessary for GXM biosynthesis. Thus, when CAP10, CAP59, CAP60 or CAP64 are deleted, the resulting strains have an acapsular phenotype (Chang and Kwon-Chung, 1994; 1998; 1999; Chang et al., 1996). Using reverse transcription (RT)–PCR, we showed that the deletion of any one of these genes had no influence on the level of CAS1 transcription compared with the original strain (data not shown). Moreover, the level of CAS1 transcription was not changed when the cells were grown in a minimum medium or capsule-induction medium (data not shown).

Disruption of CAS1

The plasmid pNE28 containing the disruption cassette (see Fig. 4A;Table 3) was used to transform the strain JEC156. The transformants were selected on 5-FOA medium screening for the Ade+, Ura phenotype. After 3 weeks of incubation at room temperature, two kinds of transformants were visible. The majority of them were slightly pink, possibly indicating episomal replication of the plasmid, and about 10% of them were white, possibly indicating integration of the plasmid. A total of 56 white transformants were tested for their reactivity with the CRND-8 antibody; eight negative clones were found. PCR analysis showed that the cassette was correctly integrated into the genome and Southern blotting showed that there was only one copy in each. We finally obtained seven independent disrupted strains (see Fig. 4B). As with the cas1–1 strain (NE28), various cas1-Δ strains did not react with any of the five anticapsular monoclonal antibodies tested (see Fig. 4C). This phenotype co-segregated with the presence of the disruption cassette in genetic crosses to the wild-type strain of opposite mating (data not shown). When CAS1 was reintroduced into the cells via an episomal plasmid, the transformant cells reacted with monoclonal antibody E1 while the original strain was completely negative. We tested three different independent cas1-Δ strains and obtained the same results. When the original strain JEC43 was transformed with the plasmid bearing CAS1, the phenotype of the transformant was unchanged even with E1. The same result was obtained with the strains NE30 and JEC43 and the serotype A CAS1 gene (data not shown).

Figure 4.

Disruption of CAS1.

A. Physical map of the CAS1 region and strategy used to construct the disruption cassette. The position of the primers (Cas1 h and Ade2a) used to check the correct integration of the disruption cassette are indicated.

B. Southern-blot analysis of the DNA digested by HindIII and hybridized with the CAS1-specific probe. Lane A: JEC43; Lane B to H: independent cas1-Δ strains.

C. Immunoblot analysis of the various mutants and constructs. For each strain, a cell suspension was directly spotted on a nitrocellulose membrane and probed using the different GXM-specific antibodies indicated (see Table 3) The plasmid pNE10 differs from pCnTEL-1 in that it lacks the NotI fragment containing the telomeric sequences.

Table 3. GXM-specific monoclonal antibodies used in this study.
AntibodySerotype specificityClassReference
CRND-8DIgM(Ikeda et al., 1996),
E1AIgG(Dromer et al., 1987)
4H3DIgG(Casadevall and Scharff, 1991)
5E4A, B, C, DIgM(Casadevall and Scharff, 1991)
2H1A, B, C, DIgG(Casadevall and Scharff, 1991)

GXM structure analysis

The structures of the purified GXM were determined by nuclear magnetic resonance (NMR) spectroscopy. Similar GXM compositions was found for the original strains JEC21 and JEC43, the mutants cas1-Δ (NE30 and NE31), and the reconstructed isolates NE30 + pNE10-CAS1 and NE31 + pNE10-CAS1. NMR data (see Table 4, Figs 5 and 6) were compared with those of previously published de-O-acetylated GXMs from serotype D strains (Skelton et al., 1991; Bacon et al., 1996). They were consistent with a Chem1 chemotype (SRG M1: 70–80%; SRG M6: 20–30%) (Cherniak et al., 1998).

Table 4. 1H and 13C chemical shifts of glycolytic residues in the GXM isolated from the original (JEC21), reconstructed (cas1-Δ + CAS1) and mutant strains (cas1-Δ + pNE10).
Strain positionGlcAXylManManXylManGlcA
  1. a . Two resonances for Man and Man GlcA H-5/C-5 were detected owing to incomplete 6-O-acetylation. The first and the last rows for each strain refer to the 6-de-O-acetylated and to the 6-O-acetylated mannosyl residues respectively. The majority of Man residues were 6-O-acetylated and the majority of the ManGlcA residues were 6-de-O-acetylated.

JEC21
H-1/C-14.47/104.34.36/105.65.17/104.35.21/103.74.23/102.7
H-2/C-23.39/75.03.32/75.44.22/72.64.23/80.74.27/80.3
H-3/C-33.47/78.03.42/78.24.03/80.44.09/78.94.07/79.5
H-4/C-43.60/73.63.63/72.03.85/69.03.69/69.93.79/69.1
H-5/C-53.65/79.64.00(), 3.26 (ax)/67.54.05/73.43.85/76.44.13/72.7
H-6/C-63.80–3.88/63.5 (de-O-acetylated)
4.30–4.35/66.5 (O-acetylated)
  
6-O-ac2.18, 2.20/22.3  
cas1-Δ + CAS1   5.19/103.75.22/102.7
H-1/C-14.46/104.44.37/105.85.15/104.14.20/81.04.25/80.5
H-2/C-23.36/75.23.29/75.24.22/72.74.08/79.04.06/79.8
H-3/C-33.45/78.13.39/78.14.01/80.53.68/69.73.82/68.9
H-4/C-43.57/73.03.62/71.83.84/68.93.83/76.24.02/75.7 a
H-5/C-53.64/79.93.99(), 3.26(ax)/67.53.96/75.9a
4.06/73.2a
 4.11/73.2 a
H-6/C-63.80–3.89/63.3 (de-O-acetylated)
4.32–4.38/66.4 (O-acetylated)
  
6-O-ac2.20/22.2  
cas1-D + pNE10   5.17/103.85.23/102.6
H-1/C-14.48/104.34.38/105.75.19/104.54.23/80.84.27/79.9
H-2/C-23.39/75.13.32/75.34.24/72.34.08/78.74.10/79.9
H-3/C-33.46/78.13.44/78.23.97/81.13.68/69.83.82/68.9
H-4/C-43.59/74.23.65/71.83.83/68.83.83/76.14.02/75.9
H-5/C-53.65/79.34.00(), 3.29(ax)/67.83.96/76.0  
H-6/C-63.78–3.88/63.2, 63.5  
Figure 5.

1H NMR spectra of GXM isolated from JEC21 (A), a cas1-Δ strain (NE30) (B) and a cas1-Δ + CAS1 strain (C). The spectra were recorded at 600.13 MHz.

Figure 6.

2D [1H, 13C] HSQC spectra of GXM isolated from JEC21 (A), a cas1-Δ strain (NE30) (B) and a cas1-Δ + CAS1 strain (C) recorded at 600.13 MHz. The bottom spectrum of each of the three HSQCs represents the anomeric region of the mannosyl residues. The assignments in the spectra refer to residues as denoted by letters, as follows: G = glucuronic acid, X = xylose, M = mannose, MX = mannosyl residue with 2-xylose attached, MG = mannosyl residue with 2-glucuronic acid attached, M? = unassigned mannosyl residue (owing to overlapping cross-peaks). The subscripts with these assignments refer to the carbon/hydrogen atoms in the respective sugar rings. HSQC cross-peaks for the O-acetyl groups are not shown. The complete signal assignment is only labelled in spectrum (B). Only cross-peaks are labeled in (A) and (C), that are relevant for the O-acetylation. Acquisition parameters are as listed in Experimental procedures. For more details see Skelton et al. (1991) or Cherniak et al. (1998). *Signals from a second unassigned GXM of approximately 15% of the main GXM.

In addition to resonances of the carbohydrate residues, intensive O-acetyl group signals were found at approximately 2.2 p.p.m. for the original and reconstructed strains. Molar ratios of O-acetylation were determined from [1H] NMR spectra by comparison of the acetyl resonances with those of the anomeric protons of mannose residues. These ratios varied between 0.63 (JEC43) and 0.58 (JEC21) acetyl groups per mannose residue in the original strains and 0.25 (NE30 + pNE10-CAS1) and 0.24 (NE31 + pNE10-CAS1) for the reconstructed strains. No acetylation was detected for the cas1-Δ strains (NE30 and NE31).

O-Acetylated residues were assigned using two-dimensional NMR methods and comparison with known chemical shift effects for carbohydrate acetylation (Cherniak et al., 1988). Comparison of the chemical shifts of strains JEC21 and JEC43 with those of the cas1-Δ strains (NE30 and NE31) indicated that no or only minor changes occurred for xylose and glucuronic acid residues, thereby indicating that they were not O-acetylated. Table 4 shows that 1H and 13C chemical shifts for some H-6 and C-6 of mannosyl residues in the original strain shifted by approximately +0.5 and +3 p.p.m., respectively, when compared with the mutant strain. Other substantial changes were observed for H-5/C-5 (+0.1/−3 p.p.m.). These changes are consistent with 6-O-acetylation (Cherniak et al., 1988). Most of these changes were observed for the mannosyl residues without glycosylation in position 2 (Man) and for the mannosyl residue with a (1′2)-β-d-glucopyranosyluronic residue (ManGlcA). Although it is difficult to specifically quantify 6-O-acetylation of the mannosyl residues because of partial signal overlapping, there are only minor changes in relative signal intensity (cross-peaks in HSQC spectra, Fig. 6) for the mannosyl residues with (1′2)-β-d-xylopryranosyl residue (ManXyl), indicating less or no 6-O-acetylation. These results are in agreement with two intense, resolved O-acetyl resonances in the [1H] NMR spectra of the original strains at 2.18 and 2.20 p.p.m. (Fig. 5).

GXM from the reconstructed strains (NE30 + CAS1 and NE31 + CAS1) were partially 6-O-acetylated but to a lesser extent than the original strains. Chemical shifts for all residues were identical to those of the original strain. However, relative cross-peak intensities in the HSQC spectra suggest less of 6-O-acetylation of the ManGlcA residue. The NMR spectra indicate that 6-O-acetylation was mainly found in the Man residue, as is also supported by the presence of only one O-acetyl resonance in the [1H] NMR spectra (2.20 p.p.m.).

Capsule thickness measurement

Capsule thickness of the 28 strains isolated by crossing after back-crossing the strain cas1-Δ (NE30) was determined as described in Experimental procedures. Genotype cas1-Δ strains (1.80 ± 0.39) had a slightly thinner capsule than the genotype CAS1 strains (2.24 ± 0.41 µm).

Virulence study

The virulence of 16 strains obtained after different crosses was tested in a murine model of disseminated infection. In two independent experiments, using an inoculum of 107 yeast cells, there were only a slight difference although statistically significant (log rank test P < 0.0001) between cas1-Δ strains and the original type ones (see Fig. 7A). However, in two other independent experiments, using an inoculum of 106 yeast cells, the cas1-Δ strains appeared clearly more virulent than the original strains. As shown in Fig. 7B, all mice infected with the cas1-Δ strains died between days 10 and 16 post inoculation, whereas all the mice infected with the original strains were still alive on day 30 (see Fig. 7B).

Figure 7.

Virulence assay.

A. Survival of mice infected with 107 viable yeast cells from representative strains (NE147 to NE150).

B. Survival of mice infected with 106 viable yeast cells from representative strains (NE167 to NE170) (▪) MATα CAS1; (▴) MATaCAS1; (□) MATα cas1-Δ; (▵)MATacas1-Δ.

Discussion

In a recent review, Doering (2000) estimated that at least 12 transferases are required to synthesize the GXM and the GalXM of C. neoformans. Capsule-structure analysis clearly showed that the genotype cas1-Δ strains synthesized a GXM devoid of O-acetyl residues but that the glycosyl linkage between carbohydrate residues on the polysaccharide structure was not modified by CAS1 deletion. Moreover, the Propsearch analysis indicated that Cas1p belongs to the same protein family as different glycosyltransferases. This finding suggests that Cas1p might be an O-acetyltransferase. However, the amino acid sequence of Cas1p does not share any homology with other putative O-acetyltransferases involved in bacterial polysaccharide synthesis (Reuber and Walker, 1993; Franklin and Ohman, 1996; Bhasin et al., 1998) or protein N-terminal acetylation (Polevoda et al., 1999). In fact, prediction of a function of a protein by analysis of its amino acids sequence may sometimes be misleading and, even though we have clearly demonstrated that Cas1p is an element of the biosynthetic pathway leading to the O-acetylation of the GXM, we have no in vitro assay demonstrating the real function of Cas1p.

No paralogue gene could be identified in the C. neoformans var. neoformans genome by Southern blotting using low stringency conditions or in the partial genome sequence available at Stanford University. This finding suggests that Cas1p might be also necessary for the GalXM acetylation. It also suggests that the strain from which CAS1 has been deleted is producing a completely de-O-acetylated capsule polysaccharide. Alternatively, we showed that at least one CAS1 homologue was present in the genomes of the four serotypes, a finding that is in good agreement with the presence of O-acetyl residues in all serotype GXM structures, even though their percentages vary according to the serotype (Turner and Cherniak, 1991).

Cas1p is an evolutionary well-conserved protein from sequence and structure points of view. In mammals, O-acetylation is one of most common modifications that occur on sialic acids. The presence of O-acetylated sialic acid has selective and widespread distribution and is developmentally regulated (Varki, 1992; Klein et al., 1994). Even though the biological function of these residues is still unknown, it has been demonstrated that numerous cancers are associated with a deregulation of the sialic acids O-acetylation (Brockenhausen and Kuhns, 1997; Klein and Roussel, 1998; Pal et al., 2000) and that O-acetylation of plant cell wall polysaccharides exists (Pauly and Scheller, 2000). Interestingly, we were unable to identify any homologue in the complete S. cerevisiae genome or in other partially available ascomycete yeast genomes. Similar results were obtained with two other genes cloned from C. neoformans and involved in GXM biosynthesis that have clear homologues in the genomes of various higher eukaryotes but none in the ascomycete genomes (G. Janbon, unpublished data).

Although GXM structures from the four serotypes have been thoroughly analysed from a chemical point of view, the position of the O-acetyl group was largely unknown. We showed here that O-acetylation occurs on C-6 of either Man or ManGlcA residues. CAS1 deletion led to complete de-O-acetylation of the GXM showing that Cas1p is necessary for O-acetylation of the mannosyl backbone.

Our results also provide some clues about the cascade of events leading to the complete GXM structure. First, no O-acetylation of ManXyl residues was observed. Second, no or poor O-acetylation of ManGlcA was observed in the reconstructed strains. Third, xylosylation and glucuronosylation of the mannosyl residues was not affected by the mutation. The most likely explanation is that the GlcA and Xyl residues are transferred to the mannose backbone before the O-acetyl residues. In analogy with the biosynthesis of bacterial alginate, the GXM acetylation would be the last step in their biosynthesis and could be linked with polysaccharide secretion (Gacesa, 1998).

Reintroduction of the CAS1 gene on a multicopy plasmid did not fully restore the amount of O-acetylation. The reconstructed strains harboured fewer O-acetyl residues, and they were mostly present on unglycosylated Man residues. The simplest explanation is that the prolonged growth of the strains in induction medium before GXM purification led to a partial loss of the plasmid and, subsequently to less Cas1p per cell. Our results also suggest that the Man residues are preferentially O-acetylated compared with ManGlcA residues and that variation of the Cas1p cell content can lead to changes in GXM structure. Our immunoblot results also showed some differences among the E1 reactivities with the original strain, the reconstructed one and the original strain transformed with a plasmid containing CAS1. These results are in good agreement with the hypothesis of a lower Cas1p cell concentration when CAS1 is on a plasmid than when it is at its natural chromosomal location.

CAS1 isolated from C. neoformans var. grubii was able to complement the mutation of the C. neoformans var. neoformans strain, thereby suggesting that the small divergence between the two Cas1p sequences was not responsible for the antigenic specificity of the two serotypes. However, the E1 reactivities with cas1-Δ strains reconstructed with CAS1 either from serotype A or serotype D differed from that of the original strain. This finding raises two hypotheses: (1) serotype A Cas1p is able to O-acetylate Man and ManGlcA residues in a Chem1-chemotype context, but less protein is synthesized in reconstructed strains than in the original one; and (2) serotype A Cas1p acetylates only one position on the GXM.

The role of external sugar O-acetylation has been previously studied for other microorganisms. The O-acetylation of the capsule has proved essential for the virulence of Staphylococcus aureus (Bhasin et al., 1998) but, for Salmonella typhimurium, de-O-acetylation of its lipopolysacharide has no effect on the virulence of the strains (Michetti et al., 1992; Slauch et al., 1995). Comparison of the virulence of cas1-Δ and original strains showed that the former with a non-O-acetylated capsule were, at least in two experiments, more virulent than the latter in a murine model. This result represents the first demonstration that the capsule structure can influence the pathophysiology of cryptococcosis. However, it must be kept in mind that, even though this murine model of cryptococcosis which mimics disseminated infection in humans has been previously validated (Lortholary et al., 1999a; 1999b), it does not mean that it is relevant for other clinical situations. For example, this model does not take into account the most common route of infestation, i.e. inhalation, or the dormant phase of the infection (Garcia-Hermoso et al., 1999). The higher virulence of cas1-Δ strains is difficult to explain. As postulated above, the changed structural composition might influence many in vitro parameters that could interfere with the course of the infection and therefore could be responsible for the apparent modification of in vivo virulence. Cleare et al. (1999) recently reported the attenuated virulence of a spontaneous variant strain that synthesized a poorly O-acetylated GXM and suggested a causal relationship between the two phenotypes. This finding contrasts strongly with our results probably because isogenic strains were not compared in their virulence assay. Recent studies on insertional signature tagged mutants and on the pkr1 mutants have identified other hypervirulent strains (D′Souza et al., 2001; Nelson et al., 2001). However, it is not known whether the hypervirulent phenotype of these strains was also caused by any modification of the capsule structure. Similarly, the cAMP-GPA1 signalling pathway was been shown to regulate the virulence and the capsule size in C. neoformans, but no information of any capsule structure alteration of the corresponding mutant strains has been published (Alspaugh et al., 1997; D′Souza and Heitman, 2001). The de-O-acetylation of GXM was associated with a thinner capsule in our strains, which is in good agreement with the previously reported reduction of capsule thickness after its chemical de-O-acetylation (Young and Kozel, 1993). Our results show that a thinner chemically altered is sufficient for virulence of C. neoformans.

Further studies are needed to explain the influence of each GXM structural element on the pathophysiology of cryptococcosis. In this context, it will be necessary to identify other genes involved in the capsule-structure biosynthesis of C. neoformans, which represents a very exciting area of research.

Experimental procedures

Strains and culture media

The C. neoformans strains used in this study are listed in Table 1. The strains were routinely cultured on YPD medium at 30°C (Sherman, 1992). Minimum medium (YNB) contained 6.7 g of yeast nitrogen base without amino acids (Difco) and 20 g of glucose l−1 of water. Induction medium contained 1.7 g of yeast nitrogen base without amino acids and without ammonium sulphate (Difco), 1.5 g of asparagine and 20 g of glucose l−1 of buffer (12 mM in NaHCO3, 35 mM in MOPS, pH 7.1) (Granger et al., 1985). 5-fluoroorotic acid (5-FOA) medium contained 6.7 g of nitrogen base (Difco), 1 g of 5-FOA, 50 mg of uracil and 20 g of glucose l−1. Bacterial strains Escherichia coli DH5α (Hanahan, 1983) and XL1-blue (Stratagene) were used for the propagation of all plasmids.

Monoclonal antibodies

The anticapsular monoclonal antibodies E1 (Dromer et al., 1987), CRND-8 kindly provided by T. Shinoda (Ikeda et al., 1996), 4H3, 2H1 and 5E4 kindly provided by A. Casadevall (Casadevall and Scharff, 1991) (Albert Einstein College of Medicine) were used in immunoblotting and immunofluorescence experiments.

DNA and RNA handling

Genomic DNA purification was carried out as previously described (Garcia-Hermoso et al., 1999). The kit systems RNeasy® Mini and Oligotex™ kits, respectively (Qiagen S.A.) were used to purify total RNA and mRNA from C. neoformans cells. RT–PCR experiments were performed with the Access RT-PCR System (Promega) using total RNA as the substrate and the primers Cas1b (5′-TATCTCTTCCTCGCCGACAG) and Cas1e (5′-TAGCCATTCAGTGATTTCGC). The primers Act1L (5′-CCTTGGTCATTGACAATGGC) and Act1R (5′-GATCGATACGGAGGATAGCG) were used to amplify a part of the actin gene cDNA as an internal control for each experiment. Southern blotting and colony hybridization were carried out using standard protocols (Sambrook et al., 1989). Probe labelling, hybridization, washing and detection of hybridized bands or colonies were performed using the digoxygenin (DIG)-non-radioactive nucleic acid labelling and detection system (Roche Diagnostic) according to the manufacturer's instruction. DNA was sequenced by Cybergene (Evry) using synthetic primers. Programs of the University of Wisconsin Genetics Computer Group were used for analysis of nucleic acid sequences (Devereux et al., 1984). Protein transmembrane domains were predicted using the program tmhmm1.0 (Sonnhammer et al., 1998). Propsearch software (http://www.embl-heidelberg.de/prs.html) (Hobohm and Sander, 1995) was used to identify protein families. Restriction endonuclease digestion and ligation were carried out using standard methods, as recommended by the suppliers.

3′ and 5′-RACE analysis

The C and N termini of Cas1p were determined using the Smart RACE system kit from Clontech. The first-strand cDNA synthesis and PCR amplification of the cDNA fragments were performed according to the manufacturer's instructions using 1 µg of mRNA from C. neoformans JEC21 grown in YPD liquid medium as the starting material. The primers Cas1m (5′-TGTATTTGGAGCCGATAGCCATAGTCGC) and Cas1l (5′-TATGGGCTATCTCTTCCTCGCCGACA) were used for the 5′ and 3′-end amplifications respectively. The cDNA fragments were then cloned into the pGEM′-T Easy Vector (Promega) and sequenced. The 3′ and 5′ ends of human homologue cDNA was cloned using the Marathon-Ready™ cDNA (Clontech) and huCASR (5′-CCCCTTGGTCACTGTATGGTTCATGG) and huCASR2 (5′-CCATGAACCATACAGTGACCAAGGGG) primers. The PCR program was: 94°C for 1 min, five cycles of 30 s at 94°C, 4 min at 72°C, five cycles of 30 s at 94°C, 4 min at 70°C, 25 cycles of 20 s at 94°C, 4 min at 68°C. The resulting PCR RACE products were ligated into the pGEM®-T Easy Vector and sequenced. These sequence data have been submitted to the DDBJ/EMBL/GenBank databases under accession numbers AF355592, AF355593and AF355594respectively.

Polysaccharide studies

Cells were grown in liquid induction medium for 3 d at 30°C, then pelleted by centrifugation and capsule polysaccharides were precipitated by adding three volumes of 95% ethanol to the culture supernatant. GXM extracts were finally prepared as previously described (Cherniak et al., 1991). Before NMR analysis, GXM (15–25 mg) were sonicated for 10 min, exchanged twice in 99.5% deuterium oxide (D2O) and lyophilized. The samples were finally dissolved in 0.5 ml of 99.96% D2O (Sigma).

NMR spectroscopy: NMR spectra were recorded at 57°C with a Bruker Avance 600 NMR spectrometer using a 5-mm [1H, 13C] inverse detection probe equipped with pulsed field gradients and operated at 600.13 and 150.913 MHz respectively. 1H NMR spectra were acquired with a spectral width of 3000 Hz, time domain 32 K, 128 acquisitions, repetition time of 6.5 s 2D NMR spectra were acquired using standard Bruker pulse sequences (xwinnmr 2.6 program). Double quantum filtered [1H, 1H] COSY, TOCSY, ROESY and NOESY spectra were obtained in the phase-sensitive mode. The spectral width was 3000 Hz, 256 increments of 16 acquisitions were acquired, each containing 2K data points. The acquisition time was 340 ms and the relaxation delay 1.5 s. Squared sine-bell functions shifted by π/2 were applied for data processing in t1 and t2. Zero-filling was used to expand the data matrix to 1K in the t1 dimension. The mixing time for the TOCSY spectra was either 40 or 150 ms and for ROESY spectra 70 ms [1H, 13C] one-bond shift correlation spectra were obtained in the 1H detection mode using a gradient enhanced HSQC pulse sequence. Acquisition parameters were: spectral width in t1 12000 Hz and in t2 3000 Hz, optimization for 1JC,H coupling constants of 140 Hz, acquisition of 512 increments in t1 each consisting of 64 acquisitions of 2K data points, acquisition time 170 ms, relaxation delay 1 s 13C decoupling during acquisition was achieved by GARP-1. A sine-bell function shifted by π/2 was applied in t2 and a Gaussian-Lorentzian function in t1. Zero filling to 1 K was used in t1 prior to Fourier transformation. Cross-peak volumes were determined for relative comparison of cross-peaks in the HSQC spectra. [1H, 13C] multiple bond heteronuclear correlation spectra (HMBC) were acquired similarly to the HSQC spectra. No 13C decoupling was applied. The spectral width in t1 was 15000 Hz. Experiments were optimized for nJC,H long-range coupling constants of approximately 2 Hz. The number of acquisitions per increment was 128.

Mutagenesis and screening for Cas mutant strains

C. neoformans strain JEC21 was removed from −70°C and transferred onto a YPD plate as independent colonies. After 2 d incubation, cells from one colony were resuspended in water. Appropriate dilutions were then plated on YPD and immediately irradiated with a UV lamp (254 nm) to obtain 90% survival. After 3 d incubation in the dark, each colony was picked up and cultured individually in 100 µl of induction medium in 96-well plates. On each plate, four wells were inoculated with one colony of the strain JEC21 as control. After 2 d, 100 µl of a 40% glycerol solution were added to each culture and mixed. The 96-well plates were then stored at −20°C until tested. Each culture was tested by ELISA as follows. Using a multichannel pipette, 20 µl of each culture were transferred to a 96-well MultiScreen filtration system plate (Millipore) (Dromer et al., 1993). The cells were washed twice with phosphate-buffered saline (PBS) (Sambrook et al., 1989) and then incubated with the anticapsule monoclonal antibody, CRND-8 (Ikeda et al., 1996), and a alkaline phosphatase-conjugated goat antimouse immunoglobulin M (IgM) secondary antibody (Sigma) for 30 min. After two washes with PBS, the cells were transferred to a new 96-well plate and incubated with p-nitrophenyl phosphate as recommended by the supplier (Sigma). The optical density was recorded using an ELISA plate reader (Labsystem multiscan plus). The presence of a capsule was checked by Indian ink negative staining. Finally, the putative Cas mutants were checked for CRND-8 reactivity by immunofluorescence assay using strain JEC21 as the positive control. The single CRND-8 negative mutant strain isolated by this method, named NE23, was then crossed on V-8 juice agar with the strain JEC156 using a classic procedure (Kwon-Chung et al., 1982). After selection on 5-FOA medium, a Ura derivative strain, named NE28, was isolated.

Cloning of CAS1 from C. neoformans var. neoformans

The strain NE28 was transformed by electroporation (Edman and Kwon-Chung, 1990) with a genomic library constructed in the plasmid pCnTEL-1 and kindly provided by B. Wickes. The NE28 transformants were grown on induction medium plates for 2 d and then transferred onto nitrocellulose paper (Sartorius). The blots were dried using a hair dryer for 1 min and incubated for 2 h in a 50-mM Tris (pH 7.5), 200 mM NaCl, 0.1% Tween 20 and 5% non-fat dry milk buffer (TBS-Tween-milk buffer). After two washes with TBS-Tween, the blots were incubated with the CRND-8 antibody and horseradish peroxidase-conjugated goat antimouse IgM secondary antibody (Sigma) for 1 h. The membranes were then washed twice with TBS-Tween and once with TBS. CRND-8-positive clones were detected using the cheminoluminescent substrate ECL (Amersham Pharmacia Biotech) as recommended by the manufacturer. DNA from each of the CRND-8-positive NE28 transformants was extracted from cells grown on minimum medium, digested with NotI, ligated and used to transform E. coli (Chang and Kwon-Chung, 1994). The plasmids, purified from the resulting E. coli transformants, were subsequently used to transform C. neoformans NE28. Finally, among a total of 56 000 Ura+ transformants that were screened, seven had the following properties: their cells reacted with antibody CRND-8 as determined by immunoblotting; they were plasmid dependent, in that the cells from their Ura derivatives no longer reacted with CRND-8 antibody; and they produced CRND-8-positive transformants when the corresponding plasmid was introduced into C. neoformans strain NE28. The restriction maps of the seven plasmids indicated that the inserted DNA contained common sequences. A physical map of a portion of the 5.5 kb insert from plasmid pNE1 is shown in Fig. 4.

Gene disruption

The CAS1 gene was disrupted using the positive–negative selection protocol described by Chang and Kwon-Chung (1994). First, the BamHI/BamHI fragment from pNE1 was subcloned into pUC18 resulting in pNE8. The plasmid pNE25 was then constructed by subcloning the EagI/SmaI fragment from pNE8 into pUC18. The EcoRV/StuI fragment was then replaced by the SmaI/SmaI fragment from the plasmid pRCD28, kindly provided by J. Heitman, which carries the ADE2 marker (Sudarshan et al., 1999), to result in pNE27. Finally, pNE28 was constructed by subcloning the pNE27 XbaI/EcoRI fragment containing the disruption cassette at the XbaI/EcoRI sites of pCIP3, kindly provided by K. J. Kwon-Chung which contains the URA5 marker (Chang and Kwon-Chung, 1994). After electroporation of strain JEC156, the strains that had integrated the disruption cassette in the correct position were screened by PCR using the primers Cas1 h (5′-GCAGTTCAAACCAAGGAGG-3′) and Ade2a (5′-GACAAGTACATCGAGAGGCT-3′) and checked by Southern blotting using the CAS1 gene as the probe.

Immunofluorescence and immunoblotting experiments

Immunofluorescence assays were carried out as previously described (Dromer et al., 1993). For immunoblotting experiments, 3 µl of a 107 cells ml−1 suspension of a 2 d minimum medium culture, were spotted on a nitrocellulose membrane and dried using a hair dryer for 1 min. The membranes were then treated as described above.

Pulse-field gel electrophoresis (PFGE)

C. neoformans chromosome plugs for PFGE were prepared as previously described (Wickes et al., 1994) and the chromosomes were separated on a 0.8% Agarose gel (Tris acetate EDTA 1X) using a CHEF DR II apparatus (Bio-Rad) with the following parameters: 14°C, 96 h, pulse from 400 to 800 s, 2 V cm−1.

Capsule thickness measurement

Successive rounds of crossing using the strains JEC50 (MATaade2), JEC52 (MATα lys1 ura5), JEC33 (MATα lys2), JEC156 (MATaade2 ura5) were performed starting with the cas1-Δ strain NE30. Out of the 28 Ura+, Ade+, Lys+ randomly chosen strains finally selected, seven were MATα CAS1, eight were MATα cas1-Δ, seven were MATaCAS1 and six were MATacas1-Δ. To determine the thickness of the capsules, cells were grown at 30°C overnight in induction medium and washed twice in PBS. The mean value of the distance from the cell wall to the outer border of the capsule for each strain was calculated from 20 cells in a suspension of India ink under 100× magnification using a grid with a resolution of 0.12 µm (Rivera et al., 1998).

Virulence study

After each cross, four strains representative for each genotype (i.e. MATα CAS1, MATα cas1-Δ, MATaCAS1 and MATacas1-Δ) were randomly chosen. Six-week-old male BALB/c mice (Charles River) were injected in the lateral tail vein with cells from each of the yeast strains (seven/group) as previously described (Lortholary et al., 1999a). Mortality was monitored daily.

Acknowledgements

We are grateful to T. Shinoda (Tokyo, Japan) for the generous gift of the CRND-8 monoclonal antibody, and B. Wickes (San Antonio, TX, USA) for providing the C. neoformans genomic library. We thank A. Casadevall (Bronx, NY, USA) for providing us the 4H3, 5E4 and 2H1 monoclonal antibodies. We thank K. J. Kwon-Chung (Bethesda, MD, USA) for providing the acapsular strains. We thank J. Heitman (Durham, NC, USA) for providing the plasmid pRCD28. We also thank J. Jacobson for editorial assistance. Financial support for this work was provided by SIDACTION and the Pasteur Institute (Contrat de Recherche Clinique). The NMR work was funded by the National Health and Medical Research Council of Australia (#980116).

Ancillary