Yeast cell walls are critical for maintaining cell integrity, particularly in the face of challenges such as growth in mammalian hosts. The pathogenic fungus Cryptococcus neoformans additionally anchors its polysaccharide capsule to the cell surface via α(1-3) glucan in the wall. Cryptococcal cells disrupted in their alpha glucan synthase gene were sensitive to stresses, including temperature, and showed difficulty dividing. These cells lacked surface capsule, although they continued to shed capsule material into the environment. Electron microscopy showed that the alpha glucan that is usually localized to the outer portion of the cell wall was absent, the outer region of the wall was highly disorganized, and the inner region was hypertrophic. Analysis of cell wall composition demonstrated complete loss of alpha glucan accompanied by a compensatory increase in chitin/chitosan and a redistribution of beta glucan between cell wall fractions. The mutants were unable to grow in a mouse model of infection, but caused death in nematodes. These studies integrate morphological and biochemical investigations of the role of alpha glucan in the cryptococcal cell wall.
Almost 40 years ago, researchers in Scotland detected α(1-3) glucan in the cell walls of several non-pathogenic species of Cryptococcus (Bacon et al., 1968). Although this polymer had been tentatively identified in other fungi in 1952 (Duff, 1952), and was isolated from Aspergillus niger in 1965 (Johnson, 1965), several decades elapsed before a protein was implicated in its synthesis. Studies of the non-pathogenic fission yeast Schizosaccharomyces pombe then identified a protein, termed Ags1p (for alpha glucan synthase; Hochstenbach et al., 1998) or Mok1p (Katayama et al., 1999), whose reduced expression correlated with loss of α(1-3) glucan and aberrant cell morphology. Recent work further suggests that this 272 kDa protein may function to both synthesize alpha glucan and to join α(1-3) glucan segments via linkers composed of α(1–4) glucan (Grün et al., 2005). S. pombe encodes three additional proteins related to alpha glucan synthase, which are required for spore wall maturation. Two of these (Mok12p and Mok13p) are involved in the synthesis of glucans containing mainly α(1-3) linkages, while a third (Mok14p) acts in formation of glucan polymers which include α(1–4) linkages (Garcia et al., 2006).
Among the pathogenic fungi, alpha glucan synthases have now been studied in Aspergillus fumigatus, Histoplasma capsulatum and Cryptococcus neoformans. A. fumigatus has three such enzymes (Beauvais et al., 2005; Maubon et al., 2006). Ags1p is localized to the cell wall and is involved in formation of ∼50% of the cell wall α(1-3) glucan, while Ags2p is intracellular and its loss has no detectable effect on glucan levels. Deletion of the genes encoding either Ags1p or Ags2p yields cells with altered hyphal morphology and reduced conidiation, but with normal virulence in a mouse model (Beauvais et al., 2005). Deletion of the third gene, encoding Ags3p, results in cells that overexpress Ags1p, presumably to compensate for the lost enzyme activity and maintain normal wall composition. These cells demonstrate enhanced virulence in a mouse model, although that is probably an indirect effect (Maubon et al., 2006). In contrast to the protein families of Aspergillus and Schizosaccharomyces, the H. capsulatum genome encodes only one alpha glucan synthase. Deletion of the gene results in cells that have reduced ability to kill macrophages in culture and to colonize murine lungs, establishing α(1-3) glucan as a major virulence factor of that pathogen (Rappleye et al., 2004).
Our interest in alpha glucan stems from its role in the biology of the important opportunistic pathogen C. neoformans. This fungus causes fatal meningoencephalitis in immunocompromised patients, with the majority of infections caused by serotypes A and D. Central to the virulence of C. neoformans is an extensive polysaccharide capsule, which surrounds and is linked to the cryptococcal cell wall. The wall is composed of alpha and beta glucans, chitin, chitosan and mannoproteins. We previously used RNA interference (RNAi) to target the single cryptococcal alpha glucan synthase gene (AGS1), and showed that this eliminated proper association of capsule material with the cell surface. These cells also grew slowly and were temperature-sensitive (Reese and Doering, 2003).
The temperature sensitivity we observed in cells subjected to RNAi targeting C. neoformans AGS1 interested us for two reasons. First, we wondered whether this indicated that the gene was essential, but that the cells survived because of the inherently incomplete nature of RNAi. Second, even if strains deleted for AGS1 were viable, the observed temperature sensitivity suggested that such cells would not survive in animals infected with C. neoformans. If this were true, Ags1p would be a candidate target for antifungal agents, especially as the mammals affected by cryptococcosis do not produce compounds similar to alpha glucans. In support of this idea, a relatively new class of antifungal drugs targets beta glucan synthase, an enzyme required for cell wall synthesis (Kartsonis et al., 2003).
To explore the questions outlined above, and to understand the role of α(1-3) glucan in C. neoformans biology, we generated alpha glucan synthase disruption and appropriate complemented strains in serotypes A and D. Below we present our findings on these strains and the relationship between morphology, cell wall composition, virulence and drug sensitivity.
We used protein sequence homology to S. pombe to identify the single cryptococcal AGS1 gene from serotype D strain JEC21 (see Experimental procedures). The gene, which is 8480 nucleotides, includes 13 introns ranging from 46 to 80 nucleotides, and encodes a protein with high homology to the corresponding enzymes of other fungi studied. Well-conserved regions occur throughout the protein (Fig. 1), covering each putative functional domain (Hochstenbach et al., 1998). This is particularly notable in the putative glycogen synthase domain (Fig. 1), which has been proposed to be involved in binding UDP-glucose or ADP-glucose and in polymer synthesis, based on comparisons with glycogen synthases from Escherichia coli (Furukawa et al., 1993). A mutation in the S. pombe Ags1p that conferred temperature sensitivity (asterisk in Fig. 1) was located in the putative hydrolase domain (Hochstenbach et al., 1998), and we targeted this region in our gene disruption in C. neoformans.
We performed biolistic transformation of serotype D strain JEC21 to replace part of the putative glycosyl hydrolase domain of AGS1 (Fig. 1) with a C. neoformans URA5 gene. If the inserted sequence were read through from the intact upstream portion of the hydrolase domain, it would encode 20 unrelated amino acids followed by a stop codon. We also generated a version of the resulting ags1Δ strain that was complemented by exogenous integration of the intact wild-type AGS1 gene (ags1ΔAGS1). We confirmed the sequence replacement and subsequent complementation by PCR and DNA blotting (see Experimental procedures).
To assess the loss of alpha glucan, we performed immunoelectron microscopy using a specific antibody developed against S. pombeα(1-3) glucan (Konomi et al., 2003; Sugawara et al., 2003). As shown in Fig. 2, we detected no antibody-reactive material in the cell walls of ags1Δ cells, indicating a complete lack of α(1-3) glucan. In contrast, the parental wild-type strain showed abundant labelling in the cell wall, primarily in the outer portion of that structure. The results from complemented mutant cells were like those from wild type, indicating that the loss of alpha glucan labelling was indeed due to the engineered AGS1 disruption. As expected from our previous studies using RNAi, the mutant also lacked capsule fibres extending from the cell wall, in sharp contrast to the other two strains.
To examine larger cell populations for the potential presence of capsule we stained cells with antibody to the major polysaccharide of the capsule (MacGill et al., 2000). Wild-type cells and the complemented mutant strain (ags1ΔAGS1) bound anti-capsular antibody to their surfaces (Fig. 3, top panel). In contrast, ags1Δ cells did not react at all with this antibody, similar to a known acapsular strain, cap59 (Fig. 3, top panel), and to the RNAi strain (Reese and Doering, 2003).
Wild-type cryptococci constitutively shed capsule material into their surroundings. When mutant strains without capsule are incubated in conditioned medium (CM) from wild-type cells, which contains shed capsule material, they recover the ability to bind anti-capsular antibody, even after extensive washing (Reese and Doering, 2003). This is shown for the acapsular strain cap59 in the lower panel of Fig. 3 (cap59/wt). CM from the ags1Δ mutant similarly transferred capsule material to acapsular acceptor cells (Fig. 3, cap59/ags1Δ), indicating that this mutant, like wild-type cells, sheds capsule components. However, ags1Δ cells were not able to bind capsule material to their own surfaces, even when it was provided by wild-type CM (Fig. 3, ags1Δ/wt). The mutant cells were present in clumps, a characteristic typical of acapsular strains, although the clumps were larger than those seen for cap59. The ags1Δ cells were also less regular in size and shape than either acapsular or wild-type cells. These characteristics were seen in ags1Δ cells of both serotypes A and D (Fig. 3), and in cells in which RNAi targeted AGS1 (Reese and Doering, 2003).
Typical of cryptococcal strains lacking capsule (Bulmer et al., 1967), ags1Δ colonies of both serotypes had a dry, dull appearance, and were not as bright a white colour as the wild type (Fig. 4). Like the corresponding RNAi strains, the mutants did not grow at 37°C and grew slowly at 30°C: the doubling time for serotype A ags1Δ cells at 30°C was 4.8 h, compared with parental and complemented strain doubling times of 2.5 and 2.6 h respectively. Poor growth of the mutant cells was exacerbated by the presence of 0.05% SDS (Fig. 4), or by the presence of 30 μg ml−1 Congo Red or 1 mg ml−1 Calcofluor White (not shown), all phenotypes consistent with poor cell integrity (de Groot et al., 2001). These growth defects were common to both serotypes. We detected no difference in the efficiency of mating when serotype D ags1Δ cells were used as either the MATa or MATα partner in mating reactions (not shown).
We wished to pursue the apparent defect in cell integrity of cells lacking Ags1p by examining the cell wall ultrastructure. Using fixation conditions for electron microscopy that we optimized to preserve even abnormally thickened cell walls (see Experimental procedures for rationale and details), we observed dramatic alterations in the cell structure (Fig. 5). Compared with wild-type or complemented strains, the mutants were misshapen and exhibited abnormally broad junctions at the bud neck (Fig. 5B) and the division septum (Fig. 5E). Wild-type cell walls are uniform, with a more striated inner region that blends smoothly into a more particulate outer region. In the mutants, the wall is thickened and ragged, and appears to be less dense overall. The two regions of the wall no longer form a smooth continuum, but show a hypertrophic striated region with only patchy material clinging to its surface (see Fig. 5E, expanded image). This phenotype was enhanced in serotype A cells (Fig. 5E), and by growth in minimal medium (not shown).
The apparent overgrowth of the inner layer of the cell wall, and overall thickening of the cell wall structure, suggested that the ags1Δ strains were undergoing compensatory changes to maintain cell integrity in the absence of α(1-3) glucan. To address this we wished to examine cell wall composition directly. Such studies in C. neoformans have previously been performed in acapsular mutants, to avoid contamination with the abundant capsule polysaccharides (James et al., 1990). We therefore generated an acapsular version of our serotype D ags1Δ mutant by crossing it to the acapsular strain B4131, which has a point mutation in the CAP59 gene (Chang and Kwon-Chung, 1994). As described in the Experimental procedures, we isolated cell wall material from the resulting double mutant and prepared alkali-insoluble (AI) and alkali-soluble (AS) fractions. The AI fractions in wild-type cells are composed predominantly of beta glucans, along with chitin, chitosan and small amounts of mannan (James et al., 1990; Banks et al., 2005). AS fractions typically contain alpha and beta glucans. Cells bearing both the ags1Δ and cap59 mutations showed profound alterations in the composition and distribution of cell wall components compared with those bearing the cap59 mutation alone (Fig. 6). As expected, α(1-3) glucan itself was completely absent, indicating that no other pathway in C. neoformans significantly contributes alpha glucan to the cell wall under these conditions. Interestingly, beta glucan was redistributed from the AS layer of the wall to the AI fraction, although it remained roughly 80% of the total cell wall mass. Concomitant with the loss of α-glucan from the mutant strain, the content of chitin and chitosan was increased to about twice its normal fraction of the cell wall (Fig. 6).
In recent years, the echinocandins, which inhibit β(1-3) glucan synthase, have been developed as a new class of antifungal agents (Kartsonis et al., 2003). Interestingly, C. neoformans is not susceptible to these compounds (Abruzzo et al., 1997), although the cryptococcal beta(1-3) glucan synthase is essential (Thompson et al., 1999) and is sensitive to them in vitro (Maligie and Selitrennikoff, 2005). The altered distribution of beta glucan in the ags1Δ cell wall made us wonder whether the mutant cryptococci might have altered susceptibility to antifungal compounds that either target beta glucan specifically, or might benefit by a disorganized and weakened cell wall. To investigate this we examined the effect of established antifungal agents on the growth of wild type, ags1Δ and complemented strains. These experiments were performed in serotype A because it is more virulent than serotype D, and is most commonly isolated from immunocompromised patients with cryptococcal infections (Mitchell and Perfect, 1995). We found that the susceptibility of the ags1Δ strain to the echinocandin tested was not significantly different from that of parental cells, although the mutant was slightly more susceptible to several azole antifungal drugs (Table 1).
Table 1. Minimal inhibitory concentrations of antifungal agents.
In one of the three experiments this result was twice the value listed.
Results of three independent experiments are shown, with values in μg ml−1.
The central role of capsule in cryptococcal pathogenesis is undisputed. Less clear, however, are the relative contributions of capsule material that is present on the cell surface and capsule material that has been shed from cells. The ags1Δ strains are unique among cryptococci without capsule, in that they still shed capsule components although they are unable to bind these molecules to their surfaces (Fig. 3). This makes virulence studies with the ags1Δ strain important. We used an intranasal inoculation model (Lim et al., 1980) to examine the growth of these cells in mice. Although recovery of serotype A mutant cells from the lungs of infected mice was similar to wild type at 1 h after infection, by 1 week later we were unable to recover any cells from the lungs of infected animals while recovery of wild type had increased over 100-fold (data not shown). This was not surprising, given the temperature sensitivity and compromised integrity of this mutant.
Because of the temperature sensitivity of the mutant strain, we turned to a less conventional disease model that can be tested at lower temperatures. The soil nematode, Caenorhabditis elegans, has been used to evaluate cryptococcal virulence (Mylonakis et al., 2002; 2003; 2004; Nielson et al., 2005; Tang et al., 2005; London et al., 2006). These investigations have demonstrated that several mutations in cryptococcal cells, which are known to reduce virulence in mammals, also abrogate the shortened lifespan and reduced progeny formation that are caused in worms by C. neoformans with wild-type virulence (Tang et al., 2005). We observed several differences when we compared worms that were exposed to C. neoformans serotype A ags1Δ with those exposed to the parental strain (Fig. 7). First, unlike the parental wild-type cells, no mutant cryptococci were visible as intact spheres in the gut beyond the grinder (Fig. 7), suggesting that most of them were crushed during passage through this structure. Mutant cells also did not induce the digestive organ distension seen when worms were exposed to the parental strain (Fig. 7, white arrows), and were not recovered to the same extent (3.80 ± 2.0 colony-forming units per worm for ags1Δ versus 33.20 ± 7.79 for the parent; P-value < 0.05). Despite these differences, exposure to the mutant strain still resulted in death of the worms, killing them with a half-time of 5 days compared with 3 days for the wild type (P = 0.04). For comparison, worms grown on a non-pathogenic bacterial lawn survive well over 10 days, while those grown with no nutrient source survive no more than 2 days. Both the mutant strain and its parent dramatically inhibited the formation of progeny, with similar counts of 4.2 ± 0.77 progeny for the wild type and 3.3 ± 0.62 for the mutant (P-value = 0.51), compared with over 100 progeny for worms grown on bacterial lawns.
Alpha glucan is critical to the normal function of yeast cell walls and plays an important role in the virulence of multiple fungal pathogens. Beyond Aspergillus, Histoplasma and Cryptococcus, which have been mentioned above, the cell wall α(1-3) glucan of Blastomyces dermatitidis has been suggested to play a role in yeast–phagocyte interactions (Hogan and Klein, 1994). In Paracoccidioides brasiliensis a shift in glucan polymer linkage from beta to alpha occurs as cells convert from the environmental mycelium form to pathogenic yeast (reviewed in Borges-Walmsley et al., 2002).
In this study, we investigated C. neoformans cells disrupted in their only copy of the alpha glucan synthase gene, AGS1. These cells are viable, although they grow slowly and are temperature sensitive. One notable result of their lack of cell wall α(1-3) glucan is that they no longer display capsule polysaccharide on their surfaces, although they do shed this material into their environment, as we observed earlier for a strain in which AGS1 was targeted by RNAi (Reese and Doering, 2003). Examination of the ags1Δ strain showed that the mutant cells have poorly formed cell walls and have trouble dividing. Direct analysis of cell walls showed that a loss of α(1-3) glucan is accompanied by an increase in chitin/chitosan content and the redistribution of beta glucan from the AS fraction to the AI fraction. The latter changes may be linked, as there is precedent for beta glucan bound to chitin being alkali insoluble (Hartland et al., 1994), although we have not tested this directly. However, even the compensatory changes we observe are insufficient to restore the integrity of mutant cells, so they remain sensitive to cell wall stress. Our electron microscopy studies highlight the corresponding morphological changes in ags1Δ cells, demonstrating that a reduction of the outer region of the cell wall (Fig. 5) accompanies loss of α(1-3) glucan. The small amount of this region that remains is tremendously disorganized, and is adjacent to a hypertrophic variant of the usual striated inner zone (Fig. 5E). The perturbed wall, lacking α(1-3) glucan, can no longer serve as an attachment site for the capsule fibres that are essential for cryptococcal cells to cause disease.
Studies in other fungal pathogens have demonstrated a range of alpha glucan expression within individual species, depending on the strain examined (Hogan and Klein, 1994; Rappleye et al., 2004). To test whether our results could be generalized within C. neoformans, we generated ags1Δ mutant and appropriate complemented control strains in both serotypes A and D. For all phenotypes tested, these strains were the same, with the exception of more extensive changes in the appearance of serotype A ags1Δ cell walls (Fig. 5). We have noted that the cell walls of serotype A are generally thicker and denser in appearance, and are more resistant to both enzymatic and physical disruption, than those of serotype D (A. Yoneda and T.L. Doering, unpubl. results). The more dramatic mutant phenotype probably reflects this difference, rather than any fundamental difference in the role of alpha glucan.
Cell wall synthesis is an attractive target for antifungal agents, analogous to the many successful antibiotics that target the corresponding process in bacteria. Echinocandin compounds inhibit fungal beta(1-3) glucan synthesis, and are approved for clinical use for certain manifestations of candidiasis and aspergillosis, as well as for prophylactic or empiric treatment in selected contexts (reviewed in Morrison, 2006). Interestingly, these compounds are not effective against C. neoformans infections, although in vitro studies indicate they do inhibit cryptococcal beta(1-3) glucan synthase (Maligie and Selitrennikoff, 2005), which is essential for cell viability (Thompson et al., 1999). Although all of the cell wall glucans in our mutant cells are present in beta linkage, this did not significantly alter susceptibility to the echinocandin we tested. This supports the conclusion that the resistance to echinocandins displayed by cryptococci is not due to any lack of importance of beta glucan to cell wall structure in this organism, but rather to one or more other mechanisms. These might include degradation or elimination of the drug, modulation of pathways involved in maintaining cell integrity (Kraus et al., 2003), or inherent features of the target enzyme. We did note a modest increase in susceptibility of the ags1Δ cells to several other antifungal compounds currently in use (Table 1), but this probably reflects the general reduction in cell wall integrity and health of this mutant.
A unique feature of the ags1Δ mutant is that the surface display and secretion of capsule polysaccharides have been unlinked. This may shed light on the individual roles in pathogenesis of these two populations of capsule material. Because of the temperature sensitivity of the ags1Δ mutant, these studies cannot be carried out in mammals, but we did examine the behaviour of the mutant and parental strains in a C. elegans model of infection. The mutant strain was predominantly lysed upon ingestion, but nonetheless caused accelerated death of the worms. This was similar to the phenotype we previously described for a strain that is mutated in the CAP59 gene (Mylonakis et al., 2002). Strains deleted for CAP59 lack capsule as assessed by India Ink staining (Chang and Kwon-Chung, 1994), and their surfaces do not react with antibody to glucuronoxylomannan (GXM), the dominant capsule component (Fig. 3). However, such cells do contain limited GXM that is detectable by ELISA assay upon cell lysis (Garcia-Rivera et al., 2004). Point mutants in CAP59 also produce the second major capsule component, galactoxylomannan (Vaishnav et al., 1998), although deletion strains have not been specifically tested for this compound. Further studies, perhaps using other low temperature experimental models of infection (London et al., 2006; Mylonakis and Fuchs, 2006), are clearly needed to define the key determinants of toxicity in the C. elegans model. While these determinants may include the presence of one or both capsule polysaccharides, it is clear that neither surface presentation of GXM nor an intact yeast cell is required. In the future it will be important to identify additional C. neoformans mutants that shed capsule components without displaying them on their surfaces, yet survive at mammalian body temperatures. Such mutants will help to extend our understanding of how the mode of capsule presentation affects cryptococcal virulence.
Media components were from Difco, nourseothricin (ClonNat) was from Werner BioAgents (Germany), 5-fluoroorotic acid was from Midwest Scientific, and oligonucleotides were from Integrated DNA Technologies, and Invitrogen Life Technologies. Reagents for DNA isolation and biolistic transformation were from Bio-Rad, DH5α cells and some restriction enzymes were from Invitrogen, and other restriction enzymes were from New England Biolabs. Specialty reagents for electron microscopy were from Polysciences or Ted Pella, and colloidal gold-conjugated antibody was from Jackson Immunoresearch. Mice for virulence studies were from Jackson Laboratories and 100 mm tissue culture plates for worm studies were from Falcon. Other reagents were from Sigma Chemical.
Strains and culture conditions
Encapsulated C. neoformans strains used in this work included JEC21 [serotype D MATα (Kwon-Chung et al., 1992)], JEC43 [serotype D MATα ura5 (Wickes et al., 1997)] and JEC34 [serotype D MATaura5 (Wickes et al., 1997)] from Joseph Heitman (Duke University Medical Center), as well as H99 [serotype A MATα (Perfect et al., 1993)] and H99R [serotype A MATα ura5 (Gorlach et al., 2002)] from Gary Cox (Duke University Medical Center). Acapsular strains were provided by June Kwon-Chung (National Institutes of Health). These were cap59[serotype D MATα cap59 (Chang and Kwon-Chung, 1994)], in which the CAP59 gene is deleted, and B4131 [also called Cap67; serotype D MATα cap59 (Jacobson et al., 1982)], which has a point mutation in the same gene. Cells were grown at 30°C with shaking in YPD medium (1% w/v yeast extract, 2% w/v peptone, 2% dextrose) or in minimal medium as indicated. Strains in which the AGS1 gene was partially replaced by the URA5 gene (see below) are termed ags1Δ, and these strains complemented by exogenous integration of the wild-type AGS1 gene are designated ags1ΔAGS1.
Sequences and databases
To search for a cryptococcal homologue of the S. pombeα(1-3) glucan synthase, we performed a TBLASTN-type Basic Local Alignment Search Tool [blast (Altschul et al., 1997)] search against the 9× sequence coverage of serotype D strain JEC21 that was available in February 2003 from The Institute of Genomic Research (TIGR). The resulting genomic DNA sequence (gDNA from contig 502641, chromosome 7) was PCR-amplified, sequenced and aligned with a gene model predicted by TWINSCAN (Tenney et al., 2004) to obtain predicted coding sequence (AY436751). Full sequence information is available at TIGR (http://www.tigr.org/tigr-scripts/euk_manatee/shared/seq_display.cgi?db=cna1&orf=177.m03284) (Loftus et al., 2005). Final annotation predicts that the protein contains 10 amino acids at the N-terminus that are not in our initial sequence complementing construct, but our results indicate that the shorter protein is functional. The full protein sequence is available with GenBank accession number XP_572121.
Preparation of nucleic acids
Genomic C. neoformans DNA was prepared from JEC43 as described by Nelson et al. (2001). Total RNA was isolated from JEC43 C. neoformans according to published methods (Missall et al., 2005), with the modifications that 15 ml conical tubes capable of high speeds (Falcon #352059) were used for centrifugation and DEPC-treated water was used for the final suspension. First strand cDNA synthesis was accomplished with the SuperScriptTM First-Strand System (Invitrogen).
Polymerase chain reaction conditions
For amplification of most nucleic acid sequences of 2 kb or less, the following conditions were used with a RoboCycler® Gradient 96 Temperature Cycler with Hot Top Assembly (Strategene): initial denaturation of 2 min at 94°C; 30 cycles of 1 min denaturation at 94°C, 1 min annealing at 55°C, and 2 min elongation at 68°C; and one final 7 min extension at 68°C. For larger segments, the conditions were altered as indicated in the text.
We used fusion PCR (Davidson et al., 2002) to generate an AGS1 disruption construct with the cryptococcal URA5 gene (Edman and Kwon-Chung, 1990) as a selectable marker, flanked by portions of the AGS1 gene. The construct was designed so the marker replaced 203 bp of the AGS1 sequence from the proposed hydrolase domain (Hochstenbach et al., 1998). First, a 981 bp upstream AGS1 gene segment (5′AGS1, corresponding to genomic nucleotides 1399–2380) was amplified by primers AJR-O11 and AJR-O14 (Table 2); a 1021 bp downstream AGS1 gene segment (5′AGS1, corresponding to genomic sequence 2583–3572) was amplified by primers AJR-O15 and AJR-O12; and a 1931 bp segment containing the URA5 promoter and gene from plasmid pRCD69 [courtesy of R. Davidson (Davidson et al., 2002)] was amplified with primers AJR-O13 and AJR-O16. Each of these three segments was then gel purified and combined as template with primers AJR-O11 and AJR-O16 to produce a 3933 bp product (PCR conditions were modified to annealing temperature of 58°C and extension time of 4 min). This final product was gel purified and introduced into JEC43 (a serotype D ura– strain) using a Bio-Rad biolistic PDS-1000/He gene gun (Toffaletti et al., 1993). We screened transformants that grew on medium lacking uracil for the expected disruption, using PCR and DNA blotting as previously described (Griffith et al., 2004). For PCR, primers were designed to regions of AGS1 within and flanking the planned deletion, and to the URA5 coding sequence. PCR reactions confirmed the absence of the expected portion of AGS1 and the presence of URA5 in its place. Gel analysis of PCR products from primers flanking the replacement also demonstrated the size increase expected because a 203 bp sequence was replaced with one of 1931 bp (not shown). For DNA blots, total genomic DNA was digested with NheI and SacI, transferred to nylon membranes, and blotted with a radiolabelled PCR product corresponding to either the deleted segment or a portion of URA5. The blots confirmed the gene replacement and showed that no additional insertions of the marker had occurred at other genomic sites (not shown). Details of these experiments are available from the authors. Following confirmation of the serotype D ags1Δ strain, primers AJR-O11 and AJR-O16 were used to amplify the disruption cassette from this strain. The resulting DNA fragment was gel purified and used in similar biolistic transformation of H99R (a serotype A ura– strain), to derive a serotype A ags1Δ strain.
The disruption strains from both serotypes were complemented with the entire cDNA corresponding to AGS1. Full-length AGS1 cDNA was amplified from JEC43 total RNA using primers AJR-O143 and AJR-O144 (Table 2) to incorporate flanking 5′ XbaI and 3′ SacII restriction sites. The cryptococcal actin promoter [ACT1p (Gorlach et al., 2002)] was similarly amplified from plasmid pGMC200 from Gary Cox (Duke University Medical School) using primers AJR-O53 and AJR-O54 to incorporate flanking SpeI and XbaI restriction enzyme sites. This ACT1p was then cloned into the appropriate sites of pGMC200 downstream of the nourseothricin N-acetyl-transferase (NAT1) resistance marker (McDade and Cox, 2001) in that plasmid. The resulting plasmid pGMC200-ACT1p was transformed into DH5α cells (Gibco), the DNA extracted by Quantum Prep Plasmid Miniprep (Bio-Rad) and the plasmid construction confirmed by PCR and restriction digests (not shown). The amplified AGS1 cDNA was then cloned in-frame downstream of the ACT1 promoter sequence, between the XbaI and SacII sites, and confirmed as above. The final 12 733 bp plasmid (pAJR-B16) was transformed into DH5α cells. Purified DNA was linearized with XmnI and PmeI and the resulting 11.8 kb fragment gel purified and introduced by biolistics into both ags1Δ strains. Transformants were plated on YPD medium, incubated for 24 h at 30°C, and then replated onto YPD plates containing 50 μg ml−1 nourseothricin with 1 M sorbitol for osmotic support. Nourseothricin-resistant transformants were further screened by PCR and DNA blotting. These studies demonstrated the presence of wild-type AGS1 and NAT1 sequences in the complemented mutant strain, each present in only one copy.
Strain growth studies
Strain growth rates in liquid culture were assessed by serial haemacytometer counts of cells from freshly grown overnight cultures that were subcultured to 1–5 × 105 cells per ml into room temperature YPD medium. Counts were performed in quadruplicate from two samples from each time point, and experiments were repeated. Growth on solid media was tested by spotting 5 μl of fivefold serial dilutions of cells onto the appropriate plates, starting from 1 × 106 cells per ml. For antibiotic sensitivity, cells were grown in YPD at 30°C for 48 h, then inoculated into Sensititre YeastOne Susceptibility plates (TREK Diagnostic Systems) as directed by the manufacturer and incubated for 72 h at the same temperature. The minimal inhibitory concentration was recorded as the lowest concentration of an antifungal agent that prevented the development of a red or purple colour in a well that was initially blue.
Cells were stained as previously described (Pierini and Doering, 2001) with anti-capsular monoclonal antibody 1255 (from Tom Kozel, University of Nevada at Reno) and an Alexafluor 546-tagged anti-mouse antibody (Molecular Probes). Samples were visualized on a Zeiss Axioskop2 MOT Plus microscope (Carl Zeiss, Thornwood, NJ) under phase-contrast and fluorescence-filtered conditions. To evaluate capsule transfer, CM from wild-type cells was prepared as described previously (Reese and Doering, 2003). CM from other strains was prepared similarly, but without a dialysis step. For capsule transfer assays, 2.5 × 106 cells (of either acapsular strain cap59 or the ags1Δ mutant) were washed twice in phosphate-buffered saline (PBS), resuspended in 0.5 ml of PBS with 1 μl of CM, rotated for 1 h at 23°C, and then washed twice with PBS before imaging.
Immunoelectron microscopy was performed at the Molecular Microbiology Imaging Facility at Washington University School of Medicine. Cells were grown in 50 ml YPD medium at 30°C with shaking until they reached log phase and subjected to centrifugation [5 min, room temperature (RT), 1500 g] in 50 ml conical tubes. the supernatant fractions were discarded, and the cells were fixed in 2% glutaraldehyde in 100 mM phosphate buffer, pH 7.2 for 2 h at 4°C. Cells were washed in phosphate buffer and post-fixed in 1% osmium tetroxide for 1 h at 4°C. Samples were then rinsed in buffer followed by dehydration and embedding as above. 70–90 nm sections were blocked with 5% fetal calf serum, 5% normal goat serum for 30 min, and then incubated with a mouse monoclonal antibody to α(1-3) glucan for 1 h (Sugawara et al., 2003). After washing with blocking buffer, samples were probed with 18 nm colloidal gold-conjugated anti-mouse secondary antibody for 1 h. Sections were washed in phosphate buffer and rinsed in water before examination as above. Parallel controls omitting the primary antibody were consistently negative at the concentration of secondary reagent used in these studies.
The fixation conditions described above (for immunoelectron microscopy) are not effective on cells with thick walls. This effectively selects against imaging of such cells because they are not well preserved. Further, the cell wall is a low contrast structure in images from cells prepared in this way, making it difficult to resolve details of its structure and to detect its outer boundary. For these reasons, we developed an alternate procedure to specifically examine cell walls. Cells were grown and sedimented as above, supernatant fractions were discarded, and the cell pellets were rapidly resuspended in 1 ml of pre-fixation mix (0.1 M sorbitol, 1 mM MgCl2, 1 mM CaCl2, 2% glutaraldehyde in 0.1 M PIPES, pH 6.8). The cell suspension was transferred to microfuge tubes for centrifugation (0.5 min, RT, top speed), and the cell pellets were resuspended in 1 ml of fresh pre-fixation mix and fixed overnight at 4°C. Cells were next pelleted (as above), washed three times in water (each wash involved resuspension of the pellet followed by a 10 min incubation at RT), transferred to borosilicate tubes, pelleted (5 min, RT, 1500 g), and resuspended in 2% KMnO4 in water (Wright, 2000). The final spin and resuspension was repeated, and the cells were then post-fixed for 45 min at RT before being washed repeatedly in water until no purple colour was visible (Wright, 2000). The samples were next transferred to microcentrifuge tubes and dehydrated for 10 min intervals in increasing concentrations of ethanol in water (once each with 30%, 50%, 70%, 90% and 95%; then three times with 100%). The ethanol-substituted samples were next substituted in propylene oxide (PO; twice for 30 min each), and then infiltrated in PO and finally embedded in Eponate 12 resin. Samples were sectioned with a Leica Ultracut UCT ultramicrotome. 60–80 nm sections were stained with uranyl acetate and lead citrate and viewed with a JEOL 1200EX transmission electron microscope.
For complete cell wall analysis, we required a control strain without capsule, to avoid potential contamination with capsule polysaccharides. To allow comparison of strains that were isogenic except at the AGS1 locus, we crossed ags1Δ with an acapsular strain. We preceded this cross with intermediate steps to incorporate markers into the strains of interest and simplify progeny selection and screening. All strains used were of serotype D, and all crosses were performed on freshly poured V8 medium plates as in Kwon-Chung (1994). Briefly, we first crossed B4131 (MATα cap59) with JEC34 (MATaura5), selecting for dull (acapsular) colonies that survived on 5-fluoroorotic acid medium. The resulting cap59 ura5 strains were tested for mating type by crossing to known α or a strains and observing filamentation, and the presence of the cap59 point mutation from strain B4131 (Chang and Kwon-Chung, 1994) was confirmed by sequencing genomic DNA. In parallel, we crossed our MATα ags1Δ strain to a MATaade2 ura5 strain, and plated the resulting spores on minimal plates lacking uracil to select against the MATa parent (Edman and Kwon-Chung, 1990). Colonies that were both pink (ade2) and dull (probably ags1) were selected, tested by PCR and temperature sensitivity for the presence of the ags1 mutation, and checked for mating type. Finally, a cap59 ura5 strain product of the first cross was mated to an ags1 ade2 strain from the second cross, and dull white colonies were selected from spores plated on minimal medium lacking both uracil and adenine. These candidates were tested by PCR and growth at 37°C to confirm the presence of the ags1 mutation, by sequencing to confirm the presence of the cap59 mutation, and by crosses to assign mating type. Confirmed MATαags1 cap59 and MATαcap59 strains were then analysed for cell wall composition.
Cell wall analysis
For analysis of total cell wall composition, cells were grown in YPD for 8 h, washed with water, and resuspended at 1–2.5 OD600 units per 500 ml of minimal medium with 2% glucose. Cells were further grown at 30°C with shaking for 2 days, at which point volumes corresponding to 1200 OD600 units of each strain were sedimented, resuspended in 10 ml of 50 mM Tris-HCl pH 7.5, and broken at 1100 psi in a French pressure cell. The lysed cells were washed six times in distilled water, frozen on dry ice, and then lyophilized to yield ∼100 mg of cell wall material. Dried cell walls were boiled twice in 50 mM Tris-HCl, pH 7.4 containing 50 mM EDTA, 2% SDS and 40 mM β-mercaptoethanol, and extensively washed with water. The AS and AI fractions were extracted as described (Beauvais et al., 2005). Determination of total hexose was performed by the phenol sulphuric acid procedure, with glucose as a standard (DuBois et al., 1956). Total hexosamine was measured after hydrolysis with 4 N HCl for 4 h at 100°C (Johnson, 1971). Monosaccharides were determined by gas chromatography after hydrolysis, reduction and peracetylation of the AI and AS fractions (Mouyna et al., 1998). Aliquots of each fraction (15–30 μg) were treated with recombinant glucanases for 24 h at 45°C, and the reducing sugars that were released by digestion were measured by the p-aminohydroxybenzoic acid hydrazide method as previously described (Beauvais et al., 2005). For α-1,3-glucanase digestion, incubation was with 0.2 μg of Mutanase (from Claus Crone Fuglsang, NovozymesA/S, Bagsvaerd, Denmark) in 100 μl of 50 mM sodium acetate buffer (pH 5.6). For β-1,3-glucanase digestion, incubation was with 2 ng of recombinant Thermoyoga neapolita LamA, purified from E. coli bearing a plasmid provided by Vladimir Zverlov (Institute of Molecular Genetics, Moscow) (Zverlov et al., 1997), in 100 μl of 50 mM sodium phosphate citrate buffer (pH 6.2).
C. elegans killing assays. C. elegans worms were monitored for survival on C. neoformans lawns, essentially as described previously (Mylonakis et al., 2002). C. neoformans serotype A strains H99R and ags1ΔAGS1 were inoculated in 2 ml of YPD with kanamycin (45 μg ml−1), ampicillin (100 μg ml−1) and streptomycin (100 μg ml−1), and grown with shaking at 30°C for 24 h; the slower-growing C. neoformans serotype A ags1Δ strain was grown similarly for 48 h. Lawns were prepared by spreading 10 μl of each culture on 35 mm tissue-culture plates containing brain heart infusion (BHI) agar with the same antibiotics. The plates were incubated at 30°C for 24 h and then at 25°C for 24 h. Approximately 70 nematodes at the L4 stage were transferred from Nematode Growth Media (NGM) seeded with E. coli OP50 to each of three BHI plates per C. neoformans strain. The plates were incubated at 25°C and the worms were examined for survival at 24 h intervals with a Nikon SMZ645 dissecting microscope. At each interval, dead worms were counted and removed.
C. elegans progeny quantification. C. neoformans lawns for progeny quantification were prepared as described above, and experiments were performed as in Tang et al. (2005). One nematode at the L4 stage was moved from NGM plates seeded with E. coli OP50 to each of 16 lawns per Cryptococcus strain. At 24 h intervals living worms were transferred to new lawns for each yeast strain, and at 72 h, the living worms were removed and each BHI agar pad was inverted onto a 100 mm tissue culture plate containing NGM agar with streptomycin (100 μg ml−1) and seeded with E. coli. The plates were incubated at 25°C for 48 h and viable progeny, defined as larvae at or beyond first stage, were counted. Only worms that survived all 3 days were included in calculating the total progeny laid over the course of 72 h.
Growth of C. neoformans in C. elegans. The number of C. neoformans colony-forming units in C. elegans were quantified essentially as described (Garsin et al., 2001), but with 10 worms at the L4 stage in each group and only one 48 h time point of exposure to Cryptococcus lawns prepared on BHI plates as above. After exposure, worms were washed twice in 8 μl drops of M9 medium on a BHI agar plate containing antibiotics as above, in order to remove surface cryptococci. Each group of 10 worms was then placed in an 1.5 ml microcentrifuge tube containing 40 μl of M9 medium with 1% Triton X-100, and ground with a pestle. The volume was adjusted to 300 μl with M9 medium containing 1% Triton X-100, and the final suspension was serially diluted and plated on YPD agar containing the same antibiotics. The plates were incubated for 48 h at 30°C, and colonies were counted.
Virulence in mice. A total of 2–5 × 104 cryptococcal cells in 50 μl of deionized water were delivered intranasally into 4–6-week-old C57Bl/6 mice. Groups of three and five mice were humanely sacrificed at 1 h and 7 days after inoculation respectively. The lungs were harvested, homogenized in water, serially diluted, and plated for colony-forming units as described previously (Sommer et al., 2003).
We thank Morgann Reilly, Stacey Klutts and Lindsay Horvath for comments on the manuscript. We also thank them, other members of the Doering and Reese laboratories, Cameron Douglas and Jennifer Nielsen Kahn for helpful discussions. We are grateful to Jeramia Ory for comments throughout the project and assistance with database analysis, to Wandy Beatty for advice and for performing immunoelectron microscopy studies, and to Tomoko Sugawara and Naohito Ohno for their roles in the preparation of the anti-glucan antibody. We also thank Tom Kozel for anti-capsular antibody, Joe Heitman and Gary Cox for plasmids, Claus Fuglsang and Vladimir Zerlov for glucanase reagents, and June Kwon-Chung, Joe Heitman and Gary Cox for cryptococcal strains. We appreciate the support of the cryptococcal genome projects that has been provided by the Burroughs Wellcome Fund and the National Institutes of Health to the Institute for Genome Research, Duke University, the Broad Institute and Stanford University. This work was supported by departmental startup funds to A.J.R.; NIH K08 award AI63084 and a New Scholar Award in Global Infectious Diseases of the Ellison Medical Foundation to E.M.; and NIH R01 award GM71007 to T.L.D.