Systematic capsule gene disruption reveals the central role of galactose metabolism on Cryptococcus neoformans virulence

Authors


*E-mail janbon@pasteur.fr; Tel. (+33) 1 45688356; Fax (+33) 1 45688420.

Summary

The polysaccharidic capsule is the main virulence factor of Cryptococcus neoformans. It primarily comprised of two polysaccharides: glucuronoxylomannan (GXM, 88% of the capsule mass) and galactoxylomannan (GalXM, 7% of the capsule mass). We constructed a large collection of mutant strains in which genes potentially involved in capsule biosynthesis were deleted. We used a new post-genomic approach to study the virulence of the strains. Primers specific for unique tags associated with the disruption cassette were used in a real-time PCR virulence assay to measure the fungal burden of each strain in different organs of mice in multi-infection experiments. With this very sensitive assay, we identified a putative UDP-glucose epimerase (Uge1p) and a putative UDP-galactose transporter (Ugt1p) essential for C. neoformans virulence. The uge1Δ and ugt1Δ strains are temperature sensitive and do not produce GalXM but synthesize a larger capsule. These mutant strains (GalXM negative, GXM positive) are not able to colonize the brain even at the first day of infection whereas GXM-negative strains (GalXM positive) can still colonize the brain, although less efficiently than the wild-type strain.

Introduction

The interactions between a pathogen and the infected host are the key to the pathogenesis of many infections. These involve different types of surfaces molecules and can be proteins, lipids or polysaccharides. However, most studies focus on proteins because they are easier to eliminate or modify through gene mutation. Polysaccharides are much more complicated to modify and their genetics, at least in eukaryotes, is far from being completely understood. However, it is obvious that, as in bacteria, their structures have a major effect on their function and on the virulence of the microorganisms (Roberts, 1996; Logan et al., 2000; Späth et al., 2003; Mille et al., 2004; Bates et al., 2006).

The cell surface of the pathogenic basidiomycete yeast Cryptococcus neoformans is mainly composed of polysaccharides. The polysaccharide capsule is actually its main virulence factor. This fungus affects mainly patients with cellular immunity deficiency (primarily AIDS patients) and causes meningoencephalitis, which is fatal if untreated (Casadevall and Perfect, 1998). The composition and structure of the capsule has been well studied, at least for its principle constituent, a large polymer (1.2–1.5 × 106 Da) of (1→3)-α-mannosyl residues with xylose, glucuronic acid and O-acetyl branches. This polysaccharide, called glucuronoxylomannan (GXM), makes up 88% of the mass of the capsule. The serotype of a given strain (determined using monoclonal and polyclonal antibodies) is linked to the percentage and position of the different residues, although the specific structures recognized by these antibodies are unknown (Janbon, 2004). The two other capsule constituents, a smaller polysaccharide (1.5 × 105 Da) called galactoxylomannan (GalXM), and mannoproteins have been much less studied. Thus, although it has recently been shown that GalXM molecules are the most abundant, in molar terms, within in the capsule (McFadden et al., 2006), its structure has been only determined for a serotype D strain (Vaishnav et al., 1998). Also there is no specific antibody available for this polysaccharide and its localization within the capsule remains unknown. Finally, its role in cryptococcosis pathophysiology remains largely unknown (Chaka et al., 1997; Pericolini et al., 2006).

Very few of the genes involved in the biosynthesis of the capsule have thus far been identified. Up to the year 2000, only four genes (CAP59, CAP60, CAP10 and CAP64) have been cloned and shown to be necessary for virulence and for the presence of a visible capsule by India Ink negative staining analysis in a serotype D strain (see Bose et al., 2003 for review). Since then, a few other genes have been identified as participating in the GXM biosynthetic pathways and analysis of the corresponding mutant strains has allowed the structure of this polysac charide to be linked with the virulence of C. neoformans (Janbon et al., 2001; Moyrand et al., 2002). However, information on these pathways remains scarce and no GalXM mutant strain has been reported. Consequently, the relationship between the presence of certain sugar structures and different cryptococcosis pathophysiology parameters is still poorly understood.

The genome sequences of two C. neoformans serotype D strains, one serotype A and one serotype B strain have now been completed (Loftus et al., 2005) (http://www.broad.mit.edu/annotation/genome/cryptococcus_neoformans_b/Home.html), and post-genomic tools such as microarray analysis have been developed and have started to be used (Missall et al., 2006). In this study, we used this genomic data to identify a large number of genes that may be involved in capsule biosynthesis. We have started to systematically delete these genes and have analysed the phenotypes associated with these gene deletions and their consequences on virulence. We have developed a real-time PCR assay for studying the organ dissemination of each mutant strain in multi-infection experiments. This study has revealed that galactose metabolism plays a central role on virulence of C. neoformans.

Results

Identification and deletion of genes potentially involved in C. neoformans var. grubii capsular polysaccharide biosynthesis

We identified more than 70 genes potentially involved in the capsular polysaccharide biosynthesis in the genome of C. neoformans serotype D using three complementary strategies: functional complementation cloning (Moyrand et al., 2002), two-hybrid screens using as baits two genes identified by complementation cloning (CAS1 and CAS3), and bioinformatics. We then identified orthologous serotype A genes and, in an ongoing programme of systematic gene disruption, 65 were deleted as described in the Experimental procedures. The phenotypes associated with C. neoformans virulence (i.e. growth rates at 30°C and 37°C; production of melanin and urease) were investigated for two independent isolates except 17 of the 65 deletions, for which we only obtained one mutant strain. We also tested the capsule phenotypes (size and structure). The capsule structure of each strain was tested using nine anti-GXM monoclonal antibodies. For some genes, we confirmed previously published observations by studying the serotype D equivalent mutant strain. For example, strains in which the CAS35 gene was deleted were hypocapsular, as was the serotype D cas35Δ strain (Moyrand et al., 2004). By contrast, some phenotypes were not conserved in both serotypes. For example, the cap59Δ serotype A strain and the equivalent serotype D mutant strain displayed no visible capsule after negative coloration by India ink; however, urease production in this serotype A mutant strain was only reduced with respect to the wild-type strain and not negative as previously published for the serotype D cap59Δ mutant strain (García-Rivera et al., 2004).

This first examination of the collection of mutant strains led us to select 22 of them. Our principal choice criteria were modification of the capsule structure or composition. We also included a so-called ‘acapsular’ mutant strain as a control and all the strains in which the homologues of the CAP genes were deleted (Janbon, 2004). The list of the genes and results of the phenotypic analysis of the corresponding mutant strains are given in Table 1. The putative function encoded by these genes and the GenBank accession numbers are given in Table S1.

Table 1.  Phenotypes of the strains used in this study.
StrainGene4H3a5E4a2H1aCRND-8aE1a21D2a13F1a12A1aCapsuleb
size (%)
MelaninUreaseDoubling time at 37°C (h)Doubling time at 30°C (h)
  • a. 

    Reactivities of each antibody towards mutant strains have been tested by dot blot assays. (+) signal similar to the one obtained with the wild type strain; (++) stronger signal; (±) lower signal; (–) very weak signal; (0) no signal.

  • b. 

    Capsule sizes of each strain were compared with the wild-type strain.

NE241WT+++0++++100++2.181.68
NE278WTGFP+++0++++100++2.091.70
NE301CAS10000000095+±1.971.77
NE303UXS1000000070++2.311.84
NE307CAS31+00098++2.131.75
NE311CAS91+++0+±++96++2.081.65
NE313CAS33+++0+±++121++2.101.81
NE315CAS34+++0+±++125++2.051.77
NE317UGT1++++++0++++++++184++2.55
NE319CAS32+++0+±++114++2.031.74
NE323CAP5+++0+±++93++2.011.73
NE331CAS92+++0++++114++1.991.80
NE335CAS4++0098++2.061.89
NE337CAS35043++3.002.30
NE339CAS3+++0++++109++1.991.85
NE341CAS9+++0++++112++1.971.88
NE343CAP2+++0++++118++2.031.88
NE347CAP6+++0++++101++2.061.70
NE349CAP3+++0++++97++2.021.76
NE351CAP4+++0+±++96++2.061.75
NE353CMT1++±0+±++119++2.001.85
NE359CAP1+±++00102++2.011.78
NE365UGE1++++++0++++++++210++2.54
NE367CAP590000000+±2.291.81
NE433Sup1++++++0++++++++199++3.782.01
NE434Sup2++++++0++++++++245++6.062.10
NE435Sup3++++++0++++++++173++5.072.44
NE436Sup4++++++0++++++++185++4.502.04

A real-time PCR method for following the fungal burden of each strain in multiple-strain infections

Each constructed mutant strain carries a specific tag (see Experimental procedures). We designed a specific primer for each of these tags and a primer within the nat1 marker for a real-time PCR assay specific to each mutant strain. (See an example of the results obtained in the Fig. 1). Using this strategy, we were able to specifically amplify the DNA of each mutant. Moreover, our method was very sensitive as we were able to detect DNA from less than 10 cells within a mixed DNA preparation (Fig. 1). Finally, standard curve analysis revealed a reproducible Tag-dependant linear correlation between the crossing point value and the natural logarithm of the number of cells used to prepare the DNA sample (data not shown).

Figure 1.

A real-time PCR strategy for estimating the proportion of each strain in a mixed population.
A. Positions of primers (black arrows) and probes (red) on the disruption cassette. The grey box shows the position of the tag.
B. Example of analysis of primer specificity by real-time PCR. A positive signal was obtained with only the corresponding mutant strain for which this primer couple had been designed.
C. Sensitivity of the technique. The DNA from 107 to 10 cells of each strain were prepared, mixed and used as a substrate for real-time PCR. A typical example of the results obtained with the wild type-specific primers and probes is given.

Study of virulence through multiple-strain infection experiments

We co-injected the selected 22 mutant strains in a mouse together with an isogenic wild-type strain. At days 1, 3 and 7, the mice were killed and total fungal DNA was extracted from each target organ (spleen, lungs and brain). The ratios of the fungal burdens of each strain to those of the control wild-type strain were calculated for each organ and at each time point (see Fig. S1). Three mice were studied at each time point. The results obtained were highly reproducible with very little discrepancy from one mouse to another (Fig. S1). Most of the mutant strain appeared to be affected in organ dissemination. For example, the strain NE367 was much less abundant than the wild type in the three organs from day 1. This strain has a mutated CAP59 gene thus confirming its role in the virulence of C. neoformans that was previously demonstrated for a serotype D strain (Chang and Kwon-Chung, 1994). As expected, this mutant strain became less prevalent than the wild-type strain over time, but always remained present in the three organs (Chang et al., 2004). On the other hand, some mutant appeared to be differentially attenuated according to the organ studied. These results were particularly interesting for us because they suggested that some capsule structures could influence differentially the targeting of C. neoformans cells to one or the other organ. Thus, this mutant strain completely disappeared from the brain and lungs and became less prevalent in the spleen than the wild type. NE365 has a mutated UGE1 gene, which encodes a putative UDP-glucose epimerase, an enzyme that interconverts UDP-glucose and UDP-galactose in many organisms (Holden et al., 2003). The NE317 strain carrying a mutation in the UGT1 gene, which encodes a putative Golgi UDP-galactose transporter, had a very similar dissemination pattern. It colonized the brain and the lungs poorly, whereas in the spleen, the ugt1Δ/wild type ratio was close to 1 (1.08) on day 1. Then it was completely eliminated from the brain and lungs and its level diminished strongly in the spleen over time (Fig. 2). Single infection experiments with the uge1Δ and cap59Δ strains confirmed the results obtained in multiple-strain co-infections, with the patterns being even more sharply defined (Fig. 3). Thus, at day 1, the uge1Δ strain did not enter the brain but disseminated in the spleen at a similar rate to the wild type. Dissemination in the lungs was weaker than for the wild type but was not totally absent. The re-introduction of the UGE1 restored the wild-type dissemination pattern of the strain (Fig. 3). As expected, the mutant strain was completely non-virulent, with none of the infected mice dying over a period of 42 days (Fig. 4), whereas the wild-type strain and the complemented strains killed the mice in less than 10 days. We sacrificed the mice infected with the mutant after 42 days, homogenized the target organs and measured the colony-forming unit (cfu) in each of them. Most of the brains were sterile (four out of seven) whereas for the infected brains the levels of colonization were minor (∼102 cfu). In the other hand, all the spleen and lungs were still infected demonstrating that this mutant was not eliminated by the host but clearly restricted to some organs. Similar virulence defect was obtained with the ugt1Δ strain (Fig. 4). We also confirmed the results of organ dissemination of the cap59Δ strain through single infection experiments (Fig. 3). Thus, cap59Δ strains were able to colonize all three organs albeit less efficiently as compared with the wild-type strain. The results obtained through single infection are clearly similar to the ones obtained through multi-infection and real-time PCR analysis a least for the third and seventh day of infection.

Figure 2.

Multi-infection experiments. Results obtained with the strains uge1Δ, ugt1Δ and cap59Δ. Each mouse was infected with 22 mutant strains and the wild-type strain. At each time point (day 1 □, day 3 bsl00005 and day 7 ▪) the organs [B (brain), L (lungs) and S (spleen)] were recovered, C. neoformans genomic DNA was extracted and real-time PCR carried out. The ratios of each mutant strain-specific signal to the wild-type-specific signal were calculated. The experiments were carried out in three different mice at each time point and the means of the results are given. *Below detectability threshold.

Figure 3.

Confirmation of the results obtained with the multi-infection experiments by single-strain infection experiments. At each time point (day 1 □, day 3 bsl00005 and day 7 ▪), the different organs [B (brain), L (lungs) and S (spleen)] were homogenized and plated and the numbers of cfu per organ were recorded after 3–5 days of incubation at 30°C. The experiments were carried out in three different mice at each time point and the means of the results are given.

Figure 4.

Top panel. Left. Survival of mice after in infection by the strains KN99α (WT), NE365 (uge1Δ) and NE427 (uge1Δ + UGE1). Right. Survival of mice after in infection by the strains KN99α (WT), NE317 (ugt1Δ) and NE429 (ugt1Δ + UGT1).
Bottom panel. The seven surviving mice (M1 to M7) after 42 days of infection with the uge1Δ strain were sacrificed and cfu in the different organs [B (brain), L (lungs) and S (spleen)] were measured.

UGE1 and UGT1 are both necessary for GalXM biosynthesis

The putative functions of the proteins encoded by UGE1 and UGT1 prompted us to investigate the galactose content of the total polysaccharide secreted by the strains with mutations in these genes (Vaishnav et al., 1998). Both mutant strains were completely devoid of galactose (ratio Gal/Man = 0) in the total capsular secreted polysaccharide, whereas the re-introduction of the genes restored the normal Gal/Man ratio (Gal/Man = 0.28, 0.31 and 0.33 in the reconstructed UGE1, the reconstructed UGT1 and the wild-type strain respectively). So as to check that the absence of galactose in the secreted polysaccharide was not due to an inadequate release of some kind of polysaccharide, we analysed the mutant strains and wild type whole cell galactose content. No galactose was detected for both mutant strains although analysis of the size of the capsule after negative staining by India ink showed that the uge1Δ and ugt1Δ strains synthesize a larger capsule than the wild-type strain (see Table 1 and Fig. 5). The uge1Δ and ugt1Δ strains are thus GalXM negative but GXM positive. By contrast, the analysis of the total secreted polysaccharide from the cap59Δ strain gave a Gal/Man ratio (Gal/Man = 1.6) consistent with the synthesis of GalXM only as previously obtained with the GXM negative mutant strain Cap67 (Vaishnav et al., 1998). Thus the cap59Δ strains are GXM negative but GalXM positive, consistent with previously published results on a so-called acapsular serotype D strain. Urease and melanin production in the uge1Δ and ugt1Δ strains were no different to that seen in the wild-type strain (Table 1). The analysis of the antigenicity of the capsule through dot blot assay using anti-GXM antibodies did not reveal any modification of the GXM structure in the GalXM negative strain (Table 1). However, both mutant strains had a strong growth defect at 37°C that was completely suppressed by adding 1 M sorbitol or galactose 1% to the medium, suggesting that the strain lacking these genes had a cell wall defect (Fig. 5). This was further supported by the sensitivity of the uge1Δ and ugt1Δ strains towards 0.1% SDS or 1.5 M NaCl (data not shown). The re-introduction of these genes restored the wild-type in vitro phenotypes for both mutant strains.

Figure 5.

Phenotypes associated with the deletion of UGE1 and UGT1.
A. The mutant strains uge1Δ and ugt1Δ do not grow at 37°C on YPD plates but this defect is reversed by adding 1 M sorbitol. The C. neoformans cells were grown in liquid YPD overnight, washed with sterile water and serial dilutions (106, 105 and 104 cells) of each strain (1) KNH99, (2) sup1, (3) uge1Δ + UGE1, (4) uge1Δ, (5) ugtΔ + UGT1, (6) ugt1Δ were spotted onto different media and observed after incubation for 3 days. sup1 is the mutation responsible for growth at 37°C in the suppressor mutant strain NE433.
B. Deletions of UGE1 or UGT1 are associated with a larger capsule phenotype. The cells were grown on capsule-inducing medium (Janbon et al., 2001) and the capsule was revealed by India ink negative staining.

Suppressor mutation

We investigated the phenotypes responsible for the absence of brain colonization by the uge1Δ strains by isolating four independent stable suppressor mutant strains able to grow at 37°C from the strain NE365 (uge1Δ) after light UV mutagenesis (Fig. 5). It has to be noted that these mutants were still growing less well than the wild type at 37°C. We failed to obtain stable better growing suppressor mutant strains suggesting that strong growth defect associated with the absence of GalXM could not be compensated by any single mutation in C. neoformans. All the suppressor mutants we obtained were similar to the mutant strain for phenotypes associated with the deletion of UGE1 (larger capsule, sensitivity towards 1.5 M NaCl and 0.1% SDS, and no melanization defect). Analysis of the polysaccharides produced by these strains confirmed that they were still unable to produce GalXM. However, all had higher glucose/mannose ratios in the culture supernatant when grown at 37°C than the wild-type strain (from 0.10 for the wild-type to 1.43–2.89 for the suppressor mutant strains). This suggested that the compensatory mutations affected the cell wall composition. When injected into mice, three of the mutants colonized the spleen at the same rate as the uge1Δ strain the fourth one being attenuated. However, none was able to colonize the brain after 1 day of infection (Fig. 6). We checked whether the mutations responsible for the suppressor phenotype were not involved in the strain-specific dissemination by re-introducing the UGE1 gene in one of these suppressor mutant strains. As expected, the newly constructed strain produced GalXM (ratio Gal/Man = 0.20) and colonized the brain as efficiently as the wild-type strain (Fig. 6). This suggested that the brain colonization defect of the uge1Δ strain is not linked to the temperature sensitivity phenotype. Moreover, different studies have shown that a larger capsule does not impair brain dissemination of C. neoformans (D'Souza et al., 2001). Altogether, these results suggest that galatose metabolism plays a key role in the brain dissemination of C. neoformans.

Figure 6.

Dissemination of the suppressor mutant strains in the brain and in the spleen after 1 day of infection. The different organs [spleen (□), and brain (▪)] were homogenized and plated and the numbers of cfu were recorded after 3–5 days of incubation at 30°C. The experiments were carried out in three different mice at each time point and the means of the results are given.

Discussion

The genomes of more than 40 fungi have thus far been sequenced and a numbers of tools have been derived from these data. One of the most powerful for studying microorganisms are collections of deletion mutant strains. A complete set of mutant strains for all the annotated genes in S. cerevisiae is available (Giaever et al., 2002). A partial set of mutant strains has been produced for pathogenic microorganisms such as Candida albicans and has been used for different phenotypic screens (Bruno and Mitchell, 2004; Nobile and Mitchell, 2005). However, for economic and ethical reasons it is not possible to test the effect of each deletion on the virulence of these strains in an animal model. Multiple infection experiments have been widely used for analysing random signature-tag-mutant collections and have proven very useful for identifying virulence genes before the development of genomics (Cormack et al., 1999). Our strategy here is a post-genomic tool that may be considered as a new way to understand the influence of each gene on the virulence of each strain of a collection of deletion mutant strains. Our real-time PCR virulence assay is very easy to set up and can be adapted to many different organisms and models. It also eliminates the problems associated with some mutant strains having a clumpy phenotype as it measures the DNA content and not the number of cfu. It has to be noted that a real-time PCR strategy to measure organ dissemination and virulence has been recently used to analyse a random signature-tag mutant collection of the pathogenic bacteria Burkholderia cenocepacia but never on any deletion-mutants one (Hunt et al., 2004). Moreover, the use of fluorescent probes increased greatly the specificity and sensitivity of the technique. We used only 22 tags for this study, although this can certainly be increased to improve the efficiency of the technique. Finally, the results we obtained by single-strain infections for the two mutant strains confirmed those obtained by multi-strain infections at least at day 3 and day 7, suggesting a poor synergy or an antagonistic effect possibly due to the mix of the different kind of mutant strains in vivo. At day 1, both mutant strains displayed a reduction in organ colonization that was more marked than that observed during the mixed infection, suggesting that the presence of other strains may influence the dissemination of these mutants during the early hours of infection. One can also hypothesize that, as this PCR method is very sensitive, part of the signal from day 1 could be due to remnant of dying cells which will not show up as positive in the cfu method. Finally, we compared the results obtained by real-time PCR and by cfu counting for 13 mutant strains (including uge1Δ and cap59Δ) at day 3 in the brain and obtained of a very good consistency between the two methods of analysis (see Fig. S2).

We have previously shown that the GXM structure plays a key role in the pathophysiology of the infection. In a serotype D strains, de-O-acetylation of the GXM increases the virulence of the strains, and the xylose residues are necessary for C. neoformans to maintain a disseminated infection (Janbon et al., 2001; Moyrand et al., 2002). We also demonstrated that the GXM structure influences more precise parameters of the pathophysiology of the infection, such as GXM clearance or the anti-chemostatic effect and complement binding (Kozel et al., 2003; Ellerbroek et al., 2004). Among the genes we have identified as being potentially involved in capsule biosynthesis, 22 deleted genes were chosen because either the corresponding mutant strains had a clear capsule phenotype or the corresponding proteins belonged to families involved in capsule formation. Most of these capsule mutants displayed a dissemination phenotype, warranting further study. However, those with the most interesting phenotypes were strains with mutations in a gene encoding a putative UDP-glucose epimerase. uge1Δ strains did not colonize the brain at all. They were still synthesizing a capsule, which was even larger than the wild-type strain. The UDP-glucose epimerase is an enzyme of the Leloir pathway that reversibly catalyses the transformation of UDP-glucose to UDP-galactose (Holden et al., 2003). Thus, as well as having a larger capsule and being temperature sensitive, these mutant strains did not synthesize GalXM. Similarly, mutations in the UDP-glucose epimerase gene RHD1 in plants alter root development and the galactosylation of xyloglycan and arabinogalactan (Seifert et al., 2002). The NE317 strain with a mutation in the UGT1 gene encoding a putative UDP-galactose transporter displayed a very similar pattern. The ugt1Δ and uge1Δ mutant strains colonized the spleen as efficiently as the wild-type strain 1 day post infection. After this early time point the mutant to wild type ratio decreased progressively. However, the data presented in Figs 3 and 4 suggest that the number of cfu of the uge1Δ mutant in the spleen and in the lungs increased over time although less quickly as compared with the wild type. These mutant cells were not eliminated from the spleen and the lungs after 42 days of infection although the mice looked completely healthy. These data suggest that the similar spleen colonization by the wild-type and the mutant strains was not due to the role of the spleen in filtering the blood at early time point. Overall, these results suggest that the galactose metabolism plays a central role in the dissemination of C. neoformans in the target organs.

The uge1Δ and ugt1Δ strains have a major growth defect at 37°C and thus cannot grow well in mammalian host. So to determine whether the temperature sensitivity of the uge1Δ strains was responsible for the inability to colonize the brain, we isolated some suppressor mutant strains that grew at 37°C and were GalXM negative. These strains still not colonize the brain. Moreover, the re-introduction of the UGE1 gene in one of these strains restored a normal organ colonization pattern and normal GalXM production. This last result proves a clear genetic linkage between the presence of a functional UGE1 and the ability of the strain to colonize the brain in the suppressor mutant strain. It thus proves that the suppressor mutation allowing this strain to grow at 37°C does not affect the brain colonization process. It has to be noted that the colonization of the brain by C. neoformans has been shown to occur within a few hours post infection (Charlier et al., 2005). If the temperature growth defect was the only important phenotype of the GalXM-negative strains in term of brain colonization, the suppressor mutant strains, although growing slower than the wild type, should colonize the brain a least to some extent. Our data suggest that the temperature defect of the GalXM-negative mutant strains is not totally responsible for the defect in brain colonization.

The size of the capsule probably does not play a key role in brain colonization because large capsule mutant strains are known to colonize the brain at least as efficiently as the wild-type strain (D'Souza et al., 2001). We have also confirmed that the GXM-negative strains can still colonize the brain, although less efficiently than the wild-type strain (Chang et al., 2004).

Finally, it has to be noted that, although we did prove that the uge1Δ and ugt1Δ strains are GalXM negative, there is still a possibility that a C. neoformans protein which was ignored here could be regulated by both genes and could have an influence in C. neoformans organ targeting. For example, an unknown molecule important for brain colonization could be galactosylated. It is also possible that a source of galactose absent in the brain could complement the uge1Δ and ugt1Δ strains growth defects in the spleen and the lungs. However, the simplest interpretation of our results is, that the absence of GalXM modified the C. neoformans organs targeting.

Our present results raise some questions concerning previously published results on the action of the capsule on different parameters of the cryptococcosis pathophysiology. All the studies published to date on the effect of capsule loss on the pathophysiology of infection have been carried out using so-called ‘acapsular’cap mutant strains (Chang and Kwon-Chung, 1994; 1998; 1999; Chang et al., 1996). However, these ‘acapsular’ strains are GXM negative and GalXM positive. This was already known for at least one strain (James and Cherniak, 1992), but has been completely ignored in all pathophysiological studies. It is likely that the elimination of GXM readily exposed GalXM at the cell surface and thus these ‘acapsular’ mutant strains should also be considered as ‘GalXM-exposed’ mutants (Van de Moer et al., 1990). Therefore, certain conclusions on the involvement of the capsule in the different processes of cryptococcosis infection may need to be re-assessed. It is clear that for many years the importance of this smaller capsule polysaccharide has been underestimated. Almost all studies on the immunoregulation properties of the capsule components have focused on GXM (Vecchiarelli, 2000), with only a few in vitro experiments suggesting that GalXM may be important in cryptococcosis pathophysiology (Chaka et al., 1997; Pericolini et al., 2006).

We observed no effect on the dissemination of the wild type, the mutant and the suppressor mutant strains upon addition of different concentrations of GalXM (300 ng to 3 mg per mouse) in the cell suspension injected in mice (data not shown). Thus, the GalXM effect may be more complicated than a simple receptor-ligand effect at the blood–brain barrier (BBB). GalXM may need to be linked to the cell to have any effect, or it may need a special conformation within the capsule to be efficient. More studies are needed to determine whether GalXM is necessary for C. neoformans to cross the BBB or to multiply within the brain, or both.

Experimental procedures

Strains and culture conditions

All the strains used in this study are C. neoformans var. grubii and are listed in Table S2. The strains were routinely cultured on YPD medium at 30°C (Sherman, 1992). Synthetic dextrose was prepared as described (Sherman, 1992). The capsule sizes were estimated after 24 h of growth in capsule-inducing medium at 30°C as previously described (Janbon et al., 2001). Melanin and urease production were estimated after spotting 105 cells of each strain on L-Dopa or Christensen agar medium respectively (Roberts et al., 1978; Walton et al., 2005). The plates were read after 48 h of incubation at 30°C. The bacterial strain Escherichia coli XL1-blue (Stratagene, La Jolla, CA) was used for the propagation of all plasmids.

Strains construction

The genes described in this report have been deleted by transforming the KN99α strain using a disruption cassette constructed by overlapping PCR as previously described (Moyrand et al., 2004). The primer sequences used are given in Table S3. The transformants were then screened for homologous integration, first by PCR and then by Southern blotting, as previously described (Moyrand and Janbon, 2004). The tagged plasmids, pNATSTM, used to amplify the selective marker were kindly provided by Dr Jennifer Lodge (Saint Louis University School of Medicine). The uge1Δ strain was reconstituted by inserting a 3.5 kb PCR-amplified fragment (using the primers UGE1-5′5 and UGE1-3′3) into the NotI site of a plasmid containing a hygromycin-resistance cassette (Hua et al., 2000). The resulting plasmid, pNE376, was digested by SpeI and used to transform the NE365 strain (MATα uge1Δ) by biolistic DNA delivery. Transformants were selected on YPD medium containing 200 U ml−1 of hygromycin (Calbiochem). Five hygromycin-resistant strains were obtained; all grew at 37°C. Two were stored at −80°C for further studies. A similar procedure was used to reconstitute the ugt1Δ strain.

Analysis of the sugar composition

Cells were grown at 30°C or 37°C for 3 days in capsule induction medium. After elimination of the cells by centrifugation and filtration, the total polysaccharide in the culture supernatant was precipitated with ethanol, resuspended in distilled water, filtered and lyophilized. The polysaccharide compositions were analysed by gas liquid chromatography after acid hydrolysis with trifluoroacetic acid as previously described (Sawardeker et al., 1965).

In vivo experiments

Six-week-old male BALB/c mice (Charles River Laboratories, France) were used for all in vivo experiments. Strains were grown on YPD medium at 30°C overnight. For single-strain infection experiments, 105C. neoformans cells were injected into the tail vein of each mouse. At each time point (day 1, day 3 and day 7), the mice were killed and the brain, spleen and lungs were recovered. Each organ was homogenized and adapted dilutions were plated on Sabouraud-chloramphenicol plates. For the brains of mice infected by the uge1Δ strains, the complete organ homogenate was plated. The experiments were repeated independently three times. For the multiple strain infections experiments, 104C. neoformans cells from each strain were mixed and injected into the tail vein of each mouse. At each time point (day 1, day 3 and day 7), the mice were killed and the brain, spleen and lungs were recovered and homogenized in 5 ml of distilled water. The organ residues were spun-down together with C. neoformans cells. The pellet was then washed twice with 1 ml of lysing buffer (Triton X-100 2%, SDS 1%, NaCl 100 mM, Tris-HCl 10 mM, EDTA 1 mM, pH 8). The C. neoformans DNA was then extracted using a glass bead procedure (Lee and Janbon, 2006) and the fungal burden of each strain was estimated by real-time PCR analysis. Three mice were studied at each time point, giving reproducible results.

The relative virulence of the strains was evaluated by intravenously inoculating groups of seven mice with 105 of each cryptococci strain. Animals were observed daily and any deaths were recorded.

Real-time PCR

The real-time PCR was carried out using a LightCycler 2.0 system (Roche). For each mutant strain, a forward primer specific for the tag associated with the deletion marker was designed (see Table S4). The reverse primer was common to all the mutant strains used in this study. The primers were chosen to have no cross-reactivity between the different tags. The sensitivity and specificity of the PCR amplifications was improved by using the LightCycler FastStart DNA MasterPLUS hybridization probes kit (Roche), choosing a couple of fluorescent probes specific to the amplified fragment common to all the mutant strains (see Table S4). The control strain was a GFP tag strain derived from the KN99α strain, kindly provided by J. Kronstad (Vancouver, Canada). We first checked that the expression of none of the common virulence factors or the organ dissemination in this strain were different the wild type. For each strain, standard curves linking the crossing point values and the cell concentration were drawn. The PCR programme for the mutants was one cycle at 95°C for 600 s, 40 cycles of 95°C for 10 s, 64–66°C for 10 s and 72°C for 5 s, and one cycle at 40°C for 30 s.

Capsule structure analysis

The anti-capsular monoclonal antibodies E1 (kindly provided by F. Dromer, France) (Dromer et al., 1987), CRND-8 (kindly provided by T. Shinoda, Japan) (Ikeda et al., 1996), 4H3, 2H1, 21D2, 13F1, 12A1 and 5E4 (kindly provided by A. Casadevall, NY) (Casadevall and Scharff, 1991) were used for the immunoblotting experiments, as previously described (Janbon et al., 2001).

Acknowledgements

We are grateful to T. Shinoda (Japan) for the generous gift of the Mab CRND-8; A. Casadevall (USA) for the Mab 4H3, 5E4 and 2H1; Françoise Dromer (France) for E1. We thank Jenny Lodge (USA) for signature-tagged markers and J. Kronstad (Canada) for the GFP-strain. We also thank V. Caro (France) for her technical advice in the use of the LightCycler apparatus. We thank Nancy Lee, Christophe d'Enfert and Marta Riera for critical reading of this manuscript. This work was supported by a grant from SIDACTION (AO-13) and the Pasteur Institute (DVPI2003).

Ancillary