Ras signaling activates glycosylphosphatidylinositol (GPI) anchor biosynthesis via the GPI–N-acetylglucosaminyltransferase (GPI–GnT) in Candida albicans

The ability of Candida albicans to switch between yeast to hyphal form is a property that is primarily associated with the invasion and virulence of this human pathogenic fungus. Several glycosylphosphatidylinositol (GPI)-anchored proteins are expressed only during hyphal morphogenesis. One of the major pathways that controls hyphal morphogenesis is the Ras-signaling pathway. We examine the cross-talk between GPI anchor biosynthesis and Ras signaling in C. albicans. We show that the first step of GPI biosynthesis is activated by Ras in C. albicans. This is diametrically opposite to what is reported in Saccharomyces cerevisiae. Of the two C. albicans Ras proteins, CaRas1 alone activates GPI–GnT activity; activity is further stimulated by constitutively activated CaRas1. CaRas1 localized to the cytoplasm or endoplasmic reticulum (ER) is sufficient for GPI–GnT activation. Of the six subunits of the GPI–N-acetylglucosaminyltransferase (GPI–GnT) that catalyze the first step of GPI biosynthesis, CaGpi2 is the key player involved in activating Ras signaling and hyphal morphogenesis. Activation of Ras signaling is independent of the catalytic competence of GPI–GnT. This too is unlike what is observed in S. cerevisiae where multiple subunits were identified as inhibiting Ras2. Fluorescence resonance energy transfer (FRET) studies indicate a specific physical interaction between CaRas1 and CaGpi2 in the ER, which would explain the ability of CaRas1 to activate GPI–GnT. CaGpi2, in turn, promotes activation of the Ras-signaling pathway and hyphal morphogenesis. The Cagpi2 mutant is also more susceptible to macrophage-mediated killing, and macrophage cells show better survival when co-cultured with Cagpi2.

Glycosylphosphatidylinositol (GPI) 5 anchors are ubiquitously present in eukaryotes, and they anchor a wide variety of proteins to the extracellular leaflet of the plasma membrane. In yeast, as well as in fungal pathogens such as Candida albicans and Aspergillus fumigatus, GPI-anchored proteins are also incorporated into the cell wall, covalently linked to the ␤-1,6glucan layer, and are important for cell wall biogenesis as well as its maintenance (1). Given that they are present in the extracellular space, a large fraction of these proteins serve as signal receptors and/or host-pathogen recognition molecules (1). Some of these surface-localized GPI-anchored proteins may even be cleaved off by cell-surface phospholipases to provide eukaryotic pathogens with an armory of enzymes or virulence factors required in colonizing/infecting the host under appropriate conditions (1,2). In yeast and fungi, the GPI biosynthetic pathway is essential, and mutants defective in the pathway are inviable (3).
The GPI anchor, attached at the C termini of proteins carrying the appropriate GPI signal sequence, is preassembled via a sequential set of biosynthetic events in the ER (3,4). In the first step of GPI biosynthesis, a phosphatidylinositol (PI) receives an N-acetylglucosamine (GlcNAc) from UDP-GlcNAc. The GlcNAc-PI thus generated is then deacetylated at the sugar, triply mannosylated, and decorated with 2-3 phosphoethanolamine groups at the different mannoses. The phosphoethanolamine on the third mannose is crucial for the final step that involves attachment of the anchor to a newly translocated protein in the ER lumen to generate a complete GPI-anchored protein. In yeast and fungi, a fourth sequential mannose is obligatory, but in mammals, this requirement is not strictly upheld except in certain specific tissues (5,6). The GPI-anchored protein is then transported via the secretory pathway to the extracellular surface of the cell membranes and/or the cell walls where they function (7).
Despite a rather conserved core, GPI biosynthesis shows species-specific differences both in terms of sequences of the reac-tions involved as well as substrate specificities (3,4). Recently, it has also begun to be recognized that significant variations exist, particularly among accessory subunits of the various enzymes involved in this pathway. One possible hypothesis that would justify these variations is that these subunits determine how and in what context the GPI biosynthetic pathway would be regulated in the organism. The first hint of such differences came up when studies on the mammalian GPI-GnT enzyme showed the presence of Dpm2, a subunit of the dolicholphosphate-mannose synthase, as an activator (8,9). No Dpm2 homolog is known to exist in Saccharomyces cerevisiae or C. albicans. Likewise, studies on the GPI-GnT enzyme complex of S. cerevisiae indicated a negative mutual regulation between the first step of GPI biosynthesis catalyzed by GPI-GnT and Ras signaling, whereas studies in mammalian cells did not indicate a similar interaction (10 -12). Our own investigations revealed that in C. albicans, a close relative of S. cerevisiae, both Ras signaling and ergosterol biosynthesis cross-talk with the GPI-GnT complex (13). However, there were significant differences between S. cerevisiae and C. albicans in how they interact with Ras signaling. In the former, down-regulation of many of the GPI-GnT subunits resulted in hyperfilamentous phenotypes, suggesting that hyperactive Ras is a common response to defects in the first enzyme complex (Fig. 1, left panel) (10,11,14). In C. albicans, however, down-regulating a subunit encoded by CaGPI19 caused hyperfilamentation and Ras hyperactivation, whereas down-regulation of a different subunit encoded by CaGPI2 resulted in the exact opposite phenotypes due to reduced Ras signaling (Fig. 1, right panel) (13,15,16). In other words, the outcome was dictated by the identity of the subunit of the GPI-GnT complex that was down-regulated. Thus, a simple correlation between GPI-GnT activity and Ras signaling could not be drawn. More detailed analysis using double deletion mutants permitted us to conclude that CaGPI19 and CaGPI2 were mutually negatively regulated (Fig. 1, right  panel), and disrupting one allele of CaGPI2 in a Cagpi19 heterozygous mutant could control the filamentation pattern in it (13). Thus, when it came to Ras signaling and filamentation, CaGPI2 acted downstream of CaGPI19 (Fig. 1, right  panel).
However, what happens at the level of the proteins? How does CaGpi2 interact with Ras signaling? Does its interaction require the formation of a functional GPI-GnT complex as was seen in S. cerevisiae? How do other subunits of the GPI-GnT interact with Ras signaling? Can they interact with CaRas proteins independently or do they all act via a single subunit? Is Ras an inhibitor of GPI-GnT in C. albicans as is reported in S. cerevisiae? Does the subcellular localization of Ras and its activation status matter? Is there a physical interaction between CaGpi2 and any of the two C. albicans Ras proteins and, if so, in which compartment does it occur? These are some of the questions we explore in this work.
We show that all the GPI-GnT subunits interact with Ras signaling via CaGpi2. Of the two C. albicans Ras proteins, CaRas1 is the only one that participates in this interaction. The activation of Ras signaling by CaGpi2 does not depend on the formation of a functional GPI-GnT complex. Overexpression of CaGpi2, however, alters the status of Hsp90 in the cell. CaRas1 enhances the GPI-GnT activity of the enzyme complex, and the activation is favored by its GTP-bound active state. A direct physical interaction exists between CaGpi2 and CaRas1 and the extent of this interaction is determined by the localization of CaRas1 in the ER.

Results
The GPI-GnT enzyme in C. albicans is made up of six putative subunits, based on sequence homology with S. cerevisiae. Of these, CaGpi3 is the putative catalytic subunit, and CaGpi1, CaGpi2, CaGpi15, CaGpi19, and CaEri1 are expected to be accessory subunits. Mutants of each of these subunits were generated as described under "Experimental procedures" and studied as described below. Ras2 inhibits GPI-GnT activity; at least two GPI-GnT subunits (Gpi2 and Eri1) appear to physically interact with and inhibit Ras2 activity (11). Whether they do so directly or via another subunit is not clear. Two other GPI-GnT mutants in S. cerevisiae (deficient in Gpi1 and Gpi19) are reported to exhibit hyperactive Ras phenotypes (hyperfilamentation) (11,14). Right panel, in C. albicans a hypofilamentous and a hyperfilamentous mutant of the GPI-GnT complex have been reported (13). Of the two subunits studied, CaGpi2 alone appears to be involved in activation of Ras signaling, and hyphal morphogenesis via CaRas1 and CaGPI19 conditional null is hyperfilamentous due to a mutually negative regulation between it and CaGPI2 (13).

Construction of GPI-GnT deletion mutants
The generation of CaGPI2 and CaGPI19 heterozygous and conditional null mutants in the BWP17 strain of C. albicans was described previously (13,15,16). The other Cagpi heterozygous and conditional null mutants were similarly generated in the BWP17 strain. Because C. albicans has two alleles of each gene, the heterozygous mutants of the GPI-GnT genes were constructed by disruption of one allele of a specific gene by one of the two selection markers, HIS1 or ARG4. Repeated attempts to make homozygous nulls were unsuccessful in all cases. Hence, conditional null mutants were generated by replacing the endogenous promoter of the surviving allele with a repressible MET3 promoter in the heterozygous strain background.
The successful generation of the heterozygous and conditional null deletion mutants was first confirmed by PCR and then by monitoring the relative transcript levels of the corresponding GPI-GnT subunit gene vis à vis the WT control using real time-PCR (RT-PCR). The transcript levels of CaGPI1, CaGPI2, CaGPI3, CaGPI15, CaGPI19, and CaERI1 were significantly down-regulated in their respective deletion mutants (Fig. S1). This confirmation allowed us to use these mutants for further studies.

GPI-GnT activity is reduced in the GPI-GnT deletion mutants
The percentage of GPI-GnT activity was measured in the GPI-GnT deletion mutants with respect to that of the WT strain, BWP17. Significantly lower GPI-GnT activity was observed in the heterozygous mutants CaGPI2/Cagpi2 (CaGPI2Hz), CaGPI15/Cagpi15 (CaGPI15Hz), CaGPI19/ Cagpi19 (CaGPI19Hz), and CaERI1/Caeri1 (CaERI1Hz) mutant strains as compared with BWP17, suggesting that they were required for GPI-GnT activity ( Fig. 2A). The heterozygous mutants CaGPI1/Cagpi1 (CaGPI1Hz) and CaGPI3/ Cagpi3 (CaGPI3Hz) had ϳ80%, or greater, relative GPI-GnT activity suggesting haplosufficiency of these genes when it came to maintaining GPI-GnT activity at near normal levels. Hence, the conditional null mutants were examined. Both Cagpi1 and Cagpi3 conditional null mutants were grown in repressive growth media, and GPI-GnT activity was estimated in these mutants relative to WT control strain (Fig. 2B). The Cagpi1 conditional null mutant (ϳ20%) and the Cagpi3 conditional null mutant (ϳ40%) had drastically reduced GPI-GnT activity as compared with the control strain under repressive growth conditions. Thus, CaGpi1 and CaGpi3 are also essential for GPI-GnT activity in C. albicans.

Depletion of any one of CaGpi1, CaGpi2, CaGpi3, CaGpi15, CaGpi19, or CaEri1 affects Ras signaling and hyphal morphogenesis
As mentioned earlier, Cagpi2 is hypofilamentous due to decreased Ras signaling, and Cagpi19 is hyperfilamentous due to enhanced Ras signaling as compared with the WT strain (13,15). Hyphal morphogenesis of the four remaining GPI-GnT subunit mutants of C. albicans were examined in Spider medium as well as YEPD at 37°C. Down-regulation of CaGPI1, CaGPI3, and CaGPI15 led to decreased hyphal growth with respect to BWP17 under hypha-inducing conditions, whereas down-regulation of CaERI1 resulted in hyperfilamentation (Fig. 3A). The phenotypes of CaGPI2Hz and CaGPI19Hz are also shown for comparison (Fig. 3A). The revertant strains were restored in their hyphal morphogenesis phenotypes, indicating that the effect was specifically caused by the gene defect in each case (Fig. 3B). Similar results were obtained for hyphal growth of the strains in liquid Spider medium as well (Fig. 3, C and D).

Hyphal growth in the GPI-GnT mutants correlate with CaGpi2 levels
As seen previously in the CaGPI2Hz and CaGPI19Hz mutant strains (13), are the hyphal morphologies of the other GPI-GnT mutants also correlated with the levels of CaGPI2 in them? To test this hypothesis, we first examined mRNA transcript as well as protein expression levels of CaGPI2 in the various GPI-GnT deletion mutants (Fig. 4A, panels i and ii, and Fig. S2). The transcript and protein expression levels of CaGPI2 were down-regulated in the hypofilamentous strains, CaGPI1Hz, CaGPI2Hz, CaGPI3Hz, and CaGPI15Hz as compared with the control. In contrast, CaGPI2 was up-regulated in the hyperfilamentous CaGPI19Hz and CaERI1Hz strains. Thus, hyphal growth in the GPI-GnT mutants correlates with CaGPI2 expression levels.
Next, we overexpressed CaGPI2 in the hypofilamentous CaGPI1Hz, CaGPI3Hz, and CaGPI15Hz strains. In each case, hyphal growth was restored back to WT levels as was the Rasdependent cAMP/PKA signaling (Fig. 4B, panels i and ii, and Fig. S3, panel i). Disrupting one allele of CaGPI2 in the CaGPI19Hz or CaERI1Hz strain, however, reversed their hyperfilamentation phenotype and reduced cAMP/PKA signaling (Fig. 4C, panels i and ii, and Fig. S3, panel ii). The converse was not true. Disruption of one allele of CaERI1 in the CaGPI2Hz strain did not induce greater filamentation in it (Fig.  4D, panel i). We have previously shown that disruption of one allele of CaGPI19 in the CaGPI2Hz strain also does not induce hyperfilamentation (13). Furthermore, overexpression of CaGPI1, CaGPI3, and CaGPI15 in the CaGPI2Hz strain also did not restore hyphal morphogenesis (Fig. 4D, panel ii). Taken GPI-GnT activity was measured in GPI-GnT deletion mutants relative to WT strain as explained under "Experimental procedures." A, GPI-GnT activity in the heterozygous GPI-GnT mutants. GPI-GnT activity was measured in CaGPI1Hz, CaGPI3Hz, CaGPI2Hz, CaGPI15Hz, CaGPI19Hz, and CaERI1Hz mutants relative to BWP17. B, conditional null Cagpi1 and Cagpi3 mutants show reduced GPI-GnT activity. BWP17URA3 as well as conditional null Cagpi1 and Cagpi3 mutants were grown in both permissive (Ura Ϫ Met Ϫ Cys Ϫ ) and repressive (Ura Ϫ Met ϩ Cys ϩ ) media during secondary culture. GPI-GnT activity was measured in all strains relative to BWP17URA3. All experiments were done three times in duplicate, and averages with standard deviation are plotted.

GPI biosynthesis and Ras signaling in C. albicans
together, it may therefore be inferred that CaGpi2 is downstream of all GPI-GnT subunits in regulation of Ras signaling.

Activation of Ras signaling by CaGpi2 is independent of GPI-GnT activity
GPI-GnT activity is restored in the revertant strain expressing CaGPI2 as compared with the CaGPI2Hz strain (Fig. 5A). However, in all the other hypofilamentous GPI-GnT mutant strains, GPI-GnT activity decreases upon overexpressing CaGPI2 (Fig. 5A). Likewise, disruption of CaGPI2 in the hyperfilamentous GPI-GnT mutant strains does not result in restoration of GPI-GnT activity (Fig. 5B), even though it reverses the hyperfilamentation phenotype. A transcript level analysis in the WT strain overexpressing CaGPI2 suggests that the levels of the different GPI-GnT subunits are significantly altered (Fig.  5C, panel i). Although the levels of CaGPI1, CaGPI3, and CaGPI15 are up-regulated upon overexpression of CaGPI2, the levels of CaGPI19 and CaERI1 are down-regulated, thereby hampering formation of a functional enzyme complex. Likewise, the expression levels of all the GPI-GnT genes are up-regulated in the CaGPI2Hz strain, whereas that of CaGPI2 is reduced (Fig. 5C, panel ii). Thus, it appears that a certain optimal stoichiometry of the subunits is required for constitution of an active GPI-GnT enzyme complex.

CaRas1 stimulates GPI-GnT activity
C. albicans has two Ras proteins. Because both RAS1 and RAS2 are nonessential genes in C. albicans, a complete null mutant, made by deleting both copies of CaRAS1 and CaRAS2, was used for this study (17). Caras1/Caras2 had lower GPI-GnT activity (ϳ50%) as compared with WT BWP17 (Fig. 6A), suggesting that deletion of CaRAS genes causes a reduction in GPI-GnT activity. This is in complete contrast to that reported in S. cerevisiae where deletion of RAS2 (homologous to CaRAS1) leads a to significant increase in GPI-GnT activity (11).
CaRAS1 and CaRAS2 were now separately reintroduced as single copies in the Caras1/Caras2 mutant to generate strains that express only one Ras protein at a time. A strong ACT1 promoter was used to ensure sufficient CaRas1 and CaRas2 levels, given that both alleles of CaRAS1 and CaRAS2 were deleted in the Caras1/Caras2 mutant. The expression of the CaRas proteins was confirmed by transcript level analysis (Fig.  S4A), and their filamentation phenotypes (Fig. S4B) were in complete agreement with what has been reported previously (17).
GPI-GnT activity was assayed in Caras1/Caras2/CaRAS1 and Caras1/Caras2/CaRAS2, compared with Caras1/Caras2/ URA3 (Fig. 6A). Restoration in GPI-GnT activity was observed . Defects in GPI anchor biosynthesis affects hyphal morphogenesis in C. albicans. A, depletion of any one of the GPI-GnT subunits affects hyphal morphogenesis on solid media. Filamentation of the GPI-GnT heterozygous strain relative to BWP17 is shown. A representative image on the 7th day of growth is shown for all strains. B, hyphal growth phenotype was restored in each GPI-GnT revertant on solid media. Filamentation of all the GPI-GnT strains is shown relative to BWP17URA3 on Spider medium. A representative image on 10th day of growth is shown for all strains except CaERI1Hz/pACT1-CaERI1 for which growth on the 7th day is shown. The experiments were done three times again in duplicate using independently grown cultures for confirmation of the result in A and B. C, depletion of any GPI-GnT subunit affects hyphal morphogenesis in liquid media. GPI-GnT heterozygous strains as well as BWP17 control were grown in liquid Spider medium for 2 h. The number of yeast and pseudohyphal and hyphal cells were counted, and the data were plotted as a fraction of total. D, hyphal growth phenotype was restored in each GPI-GnT revertant in liquid media. The GPI-GnT revertant strains displayed normal hyphal growth relative to BWP17URA3 in liquid Spider media after 1 h of growth. Averages of three sets of independent experiments done in duplicate along with standard deviations are shown in C and D.

GPI biosynthesis and Ras signaling in C. albicans
in the former but not in the latter strain. Thus, CaRAS1 alone exerts control on the first step of the GPI anchor biosynthetic pathway. It has been previously shown that in the CaGPI2Hz strain overexpression of CaRas1, but not CaRas2, restores hyphal morphogenesis (13). Thus, CaRas1 exerts control over GPI-GnT as well as hyphal morphogenesis in C. albicans.

CaRas1 G13V is a better activator of the GPI-GnT complex than CaRas1
CaRas1 cycles between an inactive GDP-bound form and an active GTP-bound form. To ascertain whether stimulation of GPI-GnT activity requires the GTP-bound active form of CaRas1, we introduced a G13V mutation in CaRas1. CaRas1 G13V is constitutively activated and cannot interact with the GTPaseactivating protein (GAP), Ira2, to return back to its inactive GDP-bound state (18,19). Expression of CaRas1 G13V was extremely low as compared with that of WT CaRas1 (Fig. 6B, panel i). This is in keeping with what has been previously reported for this mutant protein in C. albicans (20). Interestingly, despite its low expression, CaRas1 G13V was able to enhance GPI-GnT activity by roughly 1.5-fold as compared with WT CaRas1 as well as BWP17URA3 (Fig. 6B, panel ii), suggesting that GTP-bound CaRas1 is a better activator of the GPI-GnT enzyme complex than WT CaRas1.

CaRas1 can stimulate GPI-GnT at the ER
How does post-translational modification and localization of CaRas1 affect GPI-GnT activity in C. albicans? Nascent Ras proteins synthesized in the cytosol are prenylated, allowing them to transiently associate with the ER membrane where they are further palmitoylated before being finally transported to the inner leaflet of the plasma membrane (PM) (21,22). Hyphal morphogenesis in C. albicans involves the PM-localized CaRas1 preferentially (20).
CaRas1 has two cysteine residues at positions 287 and 288 that serve as sites for palmitoylation and prenylation, respectively. CaRas1 C288S cannot be prenylated and is known to be localized in cytosol, whereas CaRas1 C287S cannot be palmitoylated and remains associated with the ER (20). Neither CaRas1 variant efficiently supports hyphal growth, but CaRas1 C288S is the more severely affected of the two. To test which of these variants best interact with the GPI anchor biosynthesis pathway, we studied GPI-GnT activity in the Caras1/Caras2 null strain, wherein these protein variants were exclusively For monitoring protein expression levels, immunofluorescence was done using monoclonal anti-mRFP primary antibody and TRITC-labeled secondary antibody and quantified (40 cells) as described under "Experimental procedures." Three experimental sets were used to obtain averages with standard deviation. For RT-PCR, two independent experiments were done in duplicate and averages of the four measurements along with standard deviation are plotted. B, overexpression of CaGPI2 in the hypofilamentous GPI-GnT mutant strains restores filamentation and Ras-dependent cAMP/PKA signaling. CaGPI2 was overexpressed using pACT1-CaGPI2 in CaGPI1Hz, CaGPI3Hz, and CaGPI15Hz. The filamentation of these strains in liquid Spider medium (panel i) as well as the cAMP/PKA activity in their cell lysates (panel ii) was monitored. C, down-regulation of CaGPI2 in the hyperfilamentous GPI-GnT mutant strains restores filamentation and Ras-dependent cAMP/PKA signaling. One allele of CaGPI2 was disrupted in CaERI1Hz and CaGPI19Hz mutants, and their hyphal growth in liquid Spider medium (panel i) as well as cAMP/PKA activity in cell lysates (panel ii) was monitored relative to controls. All filamentation experiments were done three times in duplicate using independently grown strains for confirmation of the results. The cAMP/PKA activity was done three times in duplicate. Averages with standard deviation are plotted. D, none of the GPI-GnT subunits restore filamentation in CaGPI2Hz. Disruption of one allele of CaERI1 (panel i) or overexpression of CaGPI1, CaGPI3, or CaGPI15 was unable to reverse the hypofilamentation observed in the CaGPI2Hz strain in liquid Spider medium even at 37°C.

GPI biosynthesis and Ras signaling in C. albicans
expressed using the CaRAS1 endogenous promoter. Because these are not overexpression strains, they are expected to behave as heterozygous rather than as overexpression strains.
The expression levels and the specific localizations of the CaRas1 variants in the strains Caras1/Caras2/pRAS1-CaRas1, Caras1/Caras2/pRAS1-CaRas1 C287S , and Caras1/ Caras2/pRAS1-CaRas1 C288S were confirmed by immunofluorescence using confocal microscopy (Fig. S5). The expressed CaRas1 variants were also quantified using Western blots (Fig.  6C, panel i). Expression of CaRas1 C287S and CaRas1 C288S is higher than that of WT CaRas1 and yields single bands of ϳ46 kDa (although CaRas1 C288S had slightly faster migration) as they do not undergo proteolytic cleavage that CaRas1 experiences when localized to the plasma membrane (23). Both CaRas1 C287S and CaRas1 C288S in Caras1/Caras2 strain stimulated higher GPI-GnT activity than CaRas1 (Fig. 6C, panel ii). Thus, the mutant variants of CaRas1 that remain associated with the ER or are present in the cytoplasmic pool are able to stimulate GPI-GnT activity.

CaRas1 physically interacts with CaGpi2 at the ER
Because it was observed that CaRas1 best stimulates the GPI-GnT complex when it is present in the cytoplasm or at the ER, and because CaGpi2 was identified as the only subunit capable of activating hyphal growth via the Ras-signaling pathway, we wondered whether a physical interaction exists between CaRas1 and CaGpi2. To this end, we first examined the localization of the two proteins. For immunofluorescence staining, CaRas1 was detected using an anti-Ras1 antibody and a TRITC-labeled secondary antibody; CaGpi2 was tagged with a His 6 tag and detected using an anti-His antibody followed by a TRITC-labeled secondary antibody (Fig. 7A). Notably, CaRas1 appears to be mainly present either in the ER or the PM. The Pearson's correlation coefficient (PCC) between CaRas1 and ER tracker was 0.41 Ϯ 0.13. CaGpi2 localizes to the ER to a greater extent. The PCC between CaGpi2 and ER tracker was 0.65 Ϯ 0.01.
Fluorescence resonance energy transfer (FRET) was employed to assess the actual likelihood of interaction. It must be pointed out that FRET depends inversely on the sixth power of the distance between two fluorophores and hence drops rapidly with small increases in distance. Thus, a positive FRET is seen as an indication of a physical interaction between the donor and the acceptor. CaGpi2-His 6 was detected using an anti-His antibody and a TRITC-labeled secondary antibody. CaRas1 was detected using an anti-Ras antibody and a secondary antibody that was FITC-labeled to provide us with a fluorescence donoracceptor pair. Once again, we saw that CaGpi2 was predominantly localized to the ER (acceptor-red channel), whereas CaRas1-FITC was localized to PM as well as ER (donor-green channel) in the prebleach image (Fig. 7B). After photobleaching Figure 5. Activation of Ras signaling by CaGpi2 is independent of GPI-GnT activity. A, GPI-GnT activity is not restored on overexpression of CaGpi2 in hypofilamentous GPI-GnT mutants. CaGPI2 was overexpressed in all the heterozygous GPI-GnT mutant strains, and the GPI-GnT activity in microsomal preparations from these strains was measured relative to BWP17URA3. B, disruption of CaGPI2 in the hyperfilamentous strains does not restore of GPI-GnT activity. GPI-GnT activity was also monitored in microsomes prepared from CaGPI19Hz/CaGPI2Hz and CaERI1Hz/CaGPI2Hz as compared with BWP17. These experiments were done three times in duplicate using independent microsomal preparations, and averages with standard deviation are plotted in A and B. C, relative expression of GPI-GnT genes in mutant strains of CaGPI2. Panel i, in BWP17/pACT1-CaGPI2 the transcript levels of CaGPI1, CaGPI2, CaGPI3, and CaGPI15 were up-regulated, whereas the transcripts of CaGPI19 and CaERI1 were down-regulated relative to BWP17URA3. Panel ii, in CaGPI2Hz the transcript levels of all GPI-GnT genes were up-regulated except for CaGPI2 vis à vis BWP17. These experiments were done twice in duplicate, and the averages of the four measurements along with standard deviations are plotted.

GPI biosynthesis and Ras signaling in C. albicans
of the acceptor fluorophore (CaGpi2-His 6 -TRITC), an increased fluorescence of donor molecules (CaRas1-FITC) was detected, indicating that the two proteins were in close proximity (Fig. 7B). More than 10 cells were analyzed and consistently produced a positive FRET between the two proteins (62 Ϯ 2%) in regions where co-localization is distinctly visible (Fig. 7B), suggesting that the two proteins are in close proximity (6.3 Ϯ 0.2 nm) and likely to physically interact. No FRET was observed when regions of interest (ROIs) exclusively at the PM were selected and where only CaRas1 localization was clearly visible.
To ensure that the FRET observed is specific to the two proteins being studied, we tagged CaErg11 (another ER-localized enzyme, lanosterol 14␣-demethylase, involved in ergosterol biosynthesis) with His 6 and repeated the FRET experiments. CaRas1 was probed with anti-Ras antibody and a secondary antibody tagged with FITC, whereas CaErg11-His 6 was detected with an anti-His antibody and a TRITC-labeled secondary antibody in BWP17-CaErg11-His 6 cells. No FRET was observed between the donor acceptor pair, suggesting that CaRas1 and CaErg11 are not in close proximity and do not interact (Fig. 7C).

CaGpi2 overexpression phenocopies a hsp90 mutant
We had previously reported that the strain overexpressing CaGpi2 is hyperfilamentous (13). We discovered that it was also sensitive to heat shock (Fig. 8A), a phenotype typical of hyperactive Ras mutants (18). A transcript level analysis suggested that CaHSP90 was down-regulated in this strain (Fig. 8B). For further confirmation of the role of Hsp90, we disrupted one allele of CaHSP90 in the CaGPI2Hz background (CaGPI2Hz/ CaHSP90Hz) and examined its filamentation phenotype at 30°C. The CaGPI2Hz/CaHSP90Hz was hyperfilamentous as compared with the parental strain even at 30°C, clearly indicat- activates GPI-GnT. GPI-GnT activity was monitored in Caras1/Caras2 mutant and Caras1/Caras2/URA3 with respect to BWP17 and BWP17URA3, respectively. CaRAS1 or CaRAS2 was overexpressed using the strong ACT1 promoter (pACT1) in Caras1/Caras2 mutant, and GPI-GnT activity was measured vis à vis BWP17URA3. B, constitutively active mutant of CaRas1 is a better activator of the GPI-GnT. Panel i, Western blot analysis of CaRas1 G13V versus CaRas1. Western blot analysis of cell lysates from Caras1/Caras2 strain expressing either CaRas1 G13V or CaRas1 along with lysates from control strains. The ϳ46-kDa band is of full-length CaRas1, and the ϳ28-kDa one is a result of regulated proteolysis of CaRas1 localized to the PM. G6PDH was used as the loading control. These blots were repeated at least three times, and a representative image is shown. Panel ii, GPI-GnT activity is stimulated by the CaRas1 G13V . GPI-GnT activity in microsomal preparations of Caras1/Caras2 strain expressing either CaRas1 G13V or CaRas1 is plotted relative to BWP17URA3. The activity of Caras1/Caras2 strain carrying the empty vector is also shown. Averages from three experiments done in duplicate along with their standard deviations are shown. C, cytosolic and ER-associated CaRas1 are able to stimulate GPI-GnT activity. Panel i, Western blots of CaRas1 localization-defective mutants. CaRas1, CaRas1 C287S , and CaRas1 C288S were expressed using pRAS1, and their protein levels were probed by Western blots. G6PDH was used as the loading control. The blots were repeated three times, and a representative blot is shown. Panel ii, GPI-GnT activity in strains expressing localization-defective CaRas1 mutants. GPI-GnT activity in microsomes from Caras1/Caras2/URA3, Caras1/Caras2/pRAS1-CaRas1, Caras1/Caras2/ pRAS1-CaRas1 C287S , and Caras1/Caras2/pRAS1-CaRas1 C288S strains, relative to BWP17URA3, is shown. The data presented here are averages of two sets of experiments done in duplicate along with standard deviations.

GPI biosynthesis and Ras signaling in C. albicans
ing that Hsp90 depletion overcomes the filamentation block in the CaGPI2Hz strain (Fig. 8C).
Hsp90 down-regulation affects the transcription of several proteins, including Hog1 MAPK, as well as their phosphorylation (24). Hence, we monitored the levels of phosphorylated Hog1 as a measure of Hsp90 activity and found that the strain overexpressing CaGpi2 had reduced Hsp90 activity (Fig. 8D). Thus, CaRas1 may be activated upon overexpression of CaGpi2 due to a down-regulation of Hsp90 in the cell. Activating Hsp90 in these cells with the help of tamoxifen (25) results in enhanced Hsp90 activity and reversal of the hyperfilamentation phenotype (Fig. 8, D and E), indicating that Hsp90 is indeed affected upon CaGpi2 overexpression.

Cagpi2 heterozygous strain is more susceptible to killing by MH-S macrophage cells and is less able to kill the macrophage cell line
Given the effect of CaGpi2 on GPI anchor biosynthesis, we expected it to have reduced levels of GPI-anchored proteins that are involved in adhesion/virulence in C. albicans. To confirm this, Western blottings were performed using a polyclonal anti-Als5 antibody. As expected, levels of Als5, a heavily glycosylated adhesin belonging to the agglutinin-like sequence family of proteins, were reduced to ϳ65 Ϯ 12% in the CaGPI2Hz strain (Fig. 9A). In addition, we have already shown above that Ras signaling as well as hyphal growth are also affected in this strain. Hence, we expected it to be attenuated in virulence. To examine whether this was indeed the case, we co-cultured cells from a murine alveolar macrophage cell line (MH-S) with BWP17 and CaGPI2Hz strain for 18 h. As can be seen from Fig.  9, B and C, the internalization of C. albicans cells was found to be higher for the CaGPI2Hz strain than for BWP17 strain; 25.83 Ϯ 2.8% MH-S was able to phagocytose BWP17, whereas 53.01 Ϯ 1.9% MH-S was able to phagocytose CaGPI2Hz strain. This was found to be the case even when phagocytosis was inhibited with the help of cytochalasin D (Cyt D); 14.4 Ϯ 2.5% MH-S was able to phagocytose BWP17, and 36.7 Ϯ 2.1% MH-S was able to phagocytose CaGPI2Hz cells after pretreatment with Cyt D (Fig. 9D). Thus, down-regulation of CaGPI2 resulted in enhanced phagocytosis of C. albicans cells by cytoskeleton-dependent and -independent pathways. For studying the localization of CaGpi2 inside the cell, BWP17-Gpi2-His 6 strain was grown as explained under "Experimental procedures." A, co-localization of CaGpi2 and CaRas1 with ER tracker. CaRas1 was probed with anti-Ras1 antibody followed by a TRITC-labeled secondary antibody (red), and CaGpi2-His 6 was probed with anti-His antibody followed by a TRITC-labeled secondary antibody (red). The co-localization with green ER tracker was observed using an Olympus Fluoview FV1000 confocal microscope. Independent experiments were done twice in duplicate for confirmation, and representative images are shown. B, FRET between CaGpi2 and CaRas1. CaRas1 and CaGpi2-His 6 were labeled first with their respective primary antibodies (anti-Ras and anti-His) and then with secondary antibodies conjugated to FITC (green channel) and TRITC (red channel) as the donor and acceptor pair, respectively, in BWP17-CaGpi2-His 6 . CaRas1-FITC fluorescence significantly increased after acceptor photobleaching. FRET was observed between CaGpi2 and CaRas1 with efficiency of 62 Ϯ 2% in ROIs within the ER. ROIs in PM where only CaRas1 is localized showed no FRET (FRET efficiency 0%). C, negative control for FRET. CaRas1 was labeled with anti-Ras antibody and a FITC-labeled secondary antibody, and ER-localized CaErg11-His 6 was labeled with anti-His antibody and a TRITC-labeled secondary antibody in BWP17-CaErg11-His 6 as explained under "Experimental procedures" and analyzed for FRET. FRET efficiency was 0% in BWP17-CaErg11-His 6 . FRET experiments were repeated twice for confirmation.

GPI biosynthesis and Ras signaling in C. albicans
To study the survivability of the C. albicans cells, the macrophage cells were then lysed, and the fungal cells obtained were plated on YEPD. Colony-forming units (CFU) were counted in each case. CFU of BWP17 alone after incubation at 30°C for 24 h were 0.5 Ϯ 0.02 ϫ 10 6 . This was reduced by roughly 20% after it was co-cultured with MH-S cells (Fig. 9E). CFU of CaGPI2Hz strain, however, were reduced by 54% after co-culturing with MH-S as compared with the same strain in the absence of MH-S (0.5 Ϯ 0.03 ϫ 10 6 ) as shown in Fig. 9E. The significantly greater reduction in CFU of the CaGPI2Hz strain compared with that of BWP17 co-cultured with MH-S suggests that CaGPI2Hz strain is more susceptible to killing by macrophages.
We also examined the live cell recovery of MH-S after co-culturing these cells with C. albicans strains (Fig. 9F). Live-cell recovery of MH-S was ϳ67% when co-cultured with BWP17 at a 1:1 multiplicity of infection (m.o.i.) and ϳ53% when co-cultured with BWP17 at 1:5 m.o.i. Cell recovery of MH-S was ϳ80 and ϳ86% when co-cultured with CaGPI2Hz strain at m.o.i. 1:1 and 1:5, respectively. Thus, the survival rates for MH-S cells co-cultured with CaGPI2Hz cells are higher, suggesting that the CaGPI2Hz strain is likely to be less virulent than the WT strain.

Discussion
Unlike S. cerevisiae, C. albicans is a true filamentous fungus and a pathogen. The ability to switch between yeast and hyphal forms is a major virulence trait of the organism and is regulated, among others, by the Ras-signaling pathway (26). The organism is also adept at forming drug-resistant biofilms on catheters or medical implants, another trait that depends quite considerably on the Ras-signaling pathway (27,28). These attributes of the organism are also closely regulated with a transcriptional program that induces expression of virulence factors like Als3, Ecm33, and Hwp1, a large majority of which are GPI-anchored proteins (1,29). Mutants defective in GPI-anchored proteins at their cell surface are attenuated in virulence as well (30,31). Hence, the idea that the first step of GPI anchor biosynthesis cross-talks with Ras signaling in C. albicans is interesting specifically from the point of view of the virulence of the organism.
In S. cerevisiae, an organism often chosen as the model organism for studying different fungal pathways, Ras2 is shown to Figure 8. CaGpi2 activates CaRas1 signaling through Hsp90. A, strain overexpressing CaGPI2 is heat-shock-sensitive. BWP17/pACT1-CaGPI2 and BWP17URA3 were subjected to heat shock at 48°C, then spotted on YEPD plates, and incubated at 30°C. A representative image at 48 h of growth is shown. The experiment was repeated at least three times for confirmation. B, HSP90 is down-regulated in CaGPI2 overexpression strain. Transcript levels of HSP90 in BWP17/pACT1-CaGPI2 was plotted relative to that in BWP17URA3. Two independent measurements were done, each in duplicate, and averages of the four measurements with standard deviations are plotted. C, CaHSP90 disruption in CaGPI2Hz promotes filamentation. One allele of HSP90 was disrupted in the CaGPI2Hz background (CaGPI2Hz/CaHsp90Hz), and its filamentation phenotype was observed on solid YEPD medium at 30°C relative to CaGPI2Hz and CaHSP90Hz. The experiment has been repeated three times for confirmation. A representative image taken on the 11th day is shown. D, activity of Hsp90 in BWP17/pACT1-CaGPI2 is lower than that in BWP17URA3. Panel i, to monitor the activity of Hsp90, phosphorylation of Hog1 (pHog1), a client of Hsp90, Western blotting was performed using cell lysates from BWP17URA3, BWP17/pACT1-CaGPI2, and BWP17/pACT1-CaGPI2 treated with tamoxifen (TfX) for 4 h and overnight (O/N). The blots were repeated three times, and a representative image is shown. Panel ii, bar graph shows the quantification of the Western blots. E, hyphal morphology in the presence of tamoxifen. BWP17/pACT1-CaGPI2 showed enhanced filamentation as compared with BWP17URA3. The activation of Hsp90 by tamoxifen led to suppressed filamentation in both BWP17/pACT1-CaGPI2 and BWP17URA3. A representative image on the 10th day of growth at 30°C is shown for all strains. Two independent experiments were done, each in duplicate.

GPI biosynthesis and Ras signaling in C. albicans
be an inhibitor of GPI-GnT activity, and the GPI-GnT complex, in turn, is shown to be an inhibitor of Ras signaling (10,11). The results presented here show that this model does not hold true in C. albicans, and generalizations based on model organisms alone can sometimes be misleading.
In C. albicans, Ras signaling enhances GPI-GnT activity. Of the two Ras proteins in C. albicans, only CaRas1 (a close homolog of S. cerevisiae Ras2) enhances GPI-GnT activity. The interaction of CaRas1 with the GPI-GnT complex is favored by the GTP-bound active form of CaRas1 and can be achieved by CaRas1 that is either cytoplasmic or ER-localized. Thus, it appears that the pool of CaRas1 present in the endomembranes, during its transit to the PM via the ER, is able to activate GPI-GnT activity (Fig. 10). A positive correlation between CaRas1 and GPI-GnT activities would ensure that not only the virulence factors but also the biosynthesis of GPI anchors necessary for their localization and function are coordinated with hyphal morphogenesis. Thus, C. albicans is programmed for optimizing its chances for successful infections.
CaRas1 activity also is in turn stimulated by the CaGpi2 subunit of the GPI-GnT complex resulting in hyphal morphogenesis. No other GPI-GnT subunit appears to participate in this, and the activation of CaRas1 does not require the formation of a functional GPI-GnT complex. This too is in contrast with what has been reported in S. cerevisiae where two subunits, Gpi2 and Eri1, were both shown to interact with Ras2 in coimmunoprecipitation assays (11). However, it remains unclear whether both subunits independently interact with Ras2 or whether one of the subunits acts as a bridge for the other. It is also unclear whether other subunits of the GPI-GnT may be involved in the interaction, nor is it clear whether this interaction occurs with Ras2 localized at the PM or with Ras2 that associates with the ER during its transit to the PM. In C. albicans, using co-localization studies along with FRET analysis, we show that CaRas1 physically interacts with CaGpi2 in the ER (Fig. 10). This is also in keeping with our observation that CaRas1 localized to endomembranes is able to stimulate GPI-GnT activity.
One issue, however, continued to puzzle us. CaRas1 needs to be localized at the plasma membrane for sensing external cues and inducing hyphal morphogenesis (20,26). How then does CaGpi2, acting at the ER, activate CaRas1 signaling and hyphal morphogenesis? One possible explanation is that the physical interaction of CaRas1 with CaGpi2 stabilizes the association of the former with the ER and promotes its palmitoylation, thereby increasing its chances of transport and localization at the PM. Thus, depletion of CaGpi2 could perhaps reduce the levels of CaRas1 available at the PM for signaling in the cAMP/ PKA pathway for hyphal morphogenesis. Overexpression of CaGpi2, however, could perhaps increase the available pool of

GPI biosynthesis and Ras signaling in C. albicans
CaRas1 at the PM and promote filamentation (Fig. 10). This might be one possible explanation for why the WT BWP17 strain overexpressing CaGpi2 is hyperfilamentous (13).
The strain overexpressing CaGpi2 is also heat-shocksensitive, a phenotype typical of hyperactive Ras mutants (18). A more detailed analysis suggested that CaGpi2 overexpression also results in Hsp90 down-regulation, which is reflected in a decrease in transcript levels as well as in Hsp90 activity. Activating Hsp90 in these cells with the help of tamoxifen (25) results in reversal of the hyperfilamentation phenotype, indicating that Hsp90 is indeed affected upon CaGpi2 overexpression. Furthermore, disrupting one allele of Hsp90 in CaGPI2Hz could reverse its filamentation defect and promote filamentation even at 30°C. It is well known that Hsp90 along with its co-chaperone, Sgt1, modulates the Ras-signaling pathway by inhibiting the interaction of GTP-bound CaRas1 with the adenylyl cyclase, Cyr1, while promoting instead its interaction with the GAP, Ira2 (Fig. 10) (32). Hsp90 compromise has been shown to cause activation of the Ras pathway (32). Thus, a loss in Hsp90 activity upon CaGpi2 overexpression should lift the inhibition on Ras signaling and promote filamentation (Fig. 10). We recently demonstrated this to be the case in the CaGPI19 conditional null strain, where CaGpi2 is also overexpressed (33).
These results join the growing trend of reports in the literature suggesting multiple roles for the same protein. In the GPI biosynthetic pathway alone, two such proteins, Arv1 and PIGN, have been identified. Human and yeast Arv1 play a role not only in the delivery of the early GPI anchor intermediate to the first mannnosyltransferase during GPI biosynthesis but also work independently to control lipid homeostasis (34,35). Similarly, PIGN protein transfers phosphoethanolamine to the first mannose in the GPI biosynthetic pathway and also independently works to prevent protein aggregation in the ER (36,37). We propose that CaGpi2 also performs more than one function: one within the GPI-GnT to form a functional enzyme complex as well as to coordinate the activation of the enzyme by CaRas1 and another at the level of Hsp90 to regulate Ras signaling and hyphal morphogenesis.
Unlike Ras proteins that tend to be highly conserved and therefore difficult to specifically target, CaGpi2 in C. albicans shares barely 30% identity with the human homolog Pig-C. Targeting it specifically could result in not only inhibiting GPI biosynthesis but also controlling hyphal induction via the Ras-signaling pathway and thereby attenuating the virulence of the organism. Indeed, the results presented here demonstrate that the CaGpi2-deficient mutant strain is defective in its ability to kill murine macrophage cells and is more likely to be cleared by macrophages, suggesting considerably attenuated virulence. Interestingly, in the human host no interaction between any of the Ras forms and the GPI-GnT complex has been observed (12), making CaGpi2 an even more attractive candidate for development of new anti-fungals. Depending on their filamentation phenotypes, GPI-GnT mutants of C. albicans can be classified into two sets, hyperfilamentous and hypofilamentous. In each mutant, CaGpi2 expression levels alone dictate the filamentation phenotype. CaRas1 physically interacts with CaGpi2 to stimulate GPI-GnT activity. GTP-bound CaRas1 is a better activator. CaGpi2, in turn, helps activate CaRas1-dependent cAMP/PKA activity and promote filamentation. This does not require formation of a functional GPI-GnT. It is possible that physical interaction of CaRas1 with CaGpi2 assists ER association of CaRas1 and thereby promotes its palmitoylation as well as transport/localization to the PM. This perhaps increases the pool of CaRas1 available for hyphal signaling at the PM. At the PM, CaRas1 cycles between an active GTP-bound and an inactive GDP-bound form by the action of Ira2, a GAP, and Cdc25, a guanine nucleotide exchange factor (20,49). For cAMP-dependent signaling, CaRas1-GTP needs to associate with Cyr1, the adenylyl cyclase, and activate cAMP production. However, this interaction is inhibited by Hsp90 along with its co-chaperone, Sgt1 (32). As a consequence of this modulation, CaRas1-dependent cAMP signaling is maintained at basal levels. When Hsp90 levels are down-regulated, due to CaGpi2 overexpression as shown in this work or by CaGpi19 down-regulation (33), the interaction of CaRas1 with Cyr1 is promoted at the cost of its interaction with Ira2. Thus, the filamentation pathway remains turned on even at 30°C, resulting in a hyperfilamentous phenotype.

Materials
All chemicals were purchased from Sigma, SRL, Merck, or Qualigens. Growth media were from HiMedia. Primers used in this study were synthesized by Xcelris or Sigma. DNA ladders and enzymes were from Bangalore Genei (India), Fermentas, or New England Biolabs. UDP-[6-3 H]GlcNAc was from American Radiochemicals. Rabbit anti-His polyclonal antibody was purchased from Santa Cruz Biotechnology, goat anti-rabbit TRITC IgG, goat anti-mouse TRITC IgG, and goat anti-mouse FITC IgG were from Bangalore Genei; mouse anti-Ras mAb was from Merck Millipore; and anti-glucose-6-phosphate dehydrogenase (G6PDH) antibody was from Sigma, anti-phospho-p38 MAPK (Thr-180/Tyr-182) was from New England Biolabs. The production of polyclonal anti-Als5 antibody in rabbit was outsourced to Merck India Ltd. (38). ER tracker BODIPY FL GL was purchased from ThermoFisher Scientific.

Plasmids and strains
The C. albicans parental strain used in this study is BWP17, and all mutants were made in this background (39). Plasmids and strains used in this study are listed in Tables 1 and 2. The primers used are listed in Table S1. WY-ZY4 strain (BWP17-Caras1/Caras2 mutant) was a kind gift from Prof. Yue Wang (Institute of Molecular and Cell Biology, National University of Singapore) (17). Plasmids expressing different mutant forms of CaRas1 were a kind gift from Prof. Deborah A. Hogan (Dept. of Microbiology and Immunology, Geisel School of Medicine, Dartmouth College, Hanover, NH) (20). The pACT1-GFP vector was a kind gift from Prof. Alistair Brown (Aberdeen Fungal Group, University of Aberdeen, Aberdeen, UK). The pADH1-GFP vector was made in the laboratory.

Growth conditions of strains
All fungal strains were grown at 30°C in yeast extract /peptone/dextrose (YEPD) media or synthetic defined dextrose minimal media (SD media) supplemented with specific amino acids based on the auxotrophy status of the strain. MET3-regulatable conditional mutants were grown in permissive (Met Ϫ Cys Ϫ ) or repressive (5-10 mM Met/Cys) minimal media growth conditions for regulating gene expression from the MET3 promoter. Spider medium or YEPD medium was used for the hyphal induction of the fungal strains at 37°C. The bacterial strains were grown in Luria-Bertani (LB) broth at 37°C.

Generation of deletion mutants
Gene disruption in C. albicans is based on PCR-mediated disruption and homologous recombination between target gene and selection marker disruption cassette (39). Lithium acetate method was used for C. albicans transformations (40). The CaGPI2Hz and CaGPI19Hz mutants were previously reported (13,15). Using the same strategy, CaGPI1Hz, CaGPI3Hz, CaGPI15Hz, and CaERI1Hz mutants were made in BWP17 using the PCR-mediated disruption approach (39). For the generation of CaGPI1Hz mutant, the ARG4 marker was used ( Table 2 and Table S1). CaGPI3Hz, CaGPI15Hz, and CaERI1Hz mutants were made using the HIS1 marker ( Table 2 and Table S1). The colonies were confirmed by PCR using gene-flanking primers (Table S1) in all cases.

Generation of conditional null mutants
Conditional null mutants, Cagpi1 and Cagpi3, were also made using PCR-mediated promoter replacement approach as described previously for other GPI-GnT mutants (13,15). The pMET3-URA3-GFP plasmid was used to replace the endogenous gene promoter with MET3 (41). Gene expression from the MET3 promoter can be repressed by controlling concentrations of Met/Cys in the growth medium (42).

Generation of revertant strains
Revertant strains were generated as reported previously (13). Revertant strains for all GPI-GnT subunits (with the exception of CaGPI3) were created in their respective heterozygous strains. For generating CaRAS1 and CaRAS2 revertant strains, pACT1-GFP vector carrying either CaRAS1 or CaRAS2 was used to transform Caras1/Caras2 mutant. The CaGPI3 revertant was generated by transforming the CaGPI3Hz mutant with the pADH1-CaGPI3 plasmid that had the target gene cloned under the control of the ADH1 promoter. Gene-specific as well as locus-specific primers were used to confirm the integration in all cases by PCR.
URA3 gene produces significant effects on cellular proteome and virulence of C. albicans (43). Thus, to rule out the effect of URA3 on other genes in revertant strains, appropriate control strains were generated by reintegration of one copy of URA3 at RPS1 locus in BWP17 and all mutant strains as necessary. Integration at the RPS1 locus in the positive clones was confirmed by PCR using GFP forward primer, FPGFP-BamHI, and RPS1 reverse internal primer, RPCaRPS1 (Table S1).

Generation of overexpression strains
Overexpression strains were generated as reported previously (13). CaGPI2 was overexpressed in each GPI-GnT heterozygous mutant using the plasmid pACT1-CaGPI2. Integration at the RPS1 locus in the positive clones was confirmed using CaGPI2 forward primer, FPCaGPI2-HindIII, and RPS1 reverse internal primer, RPCaRPS1 (Table S1).
For tagging of CaGpi2 with His 6 at its C terminus for immunofluorescence and FRET studies, primers FP CaGPI2-His and RP CaGPI2-URA3 (Table S1) were used to amplify His 6 -URA3, and the amplicon was used to transform BWP17 to generate BWP17-CaGpi2-His 6 . Similarly, CaErg11 was tagged with His 6 -URA3 at its C terminus using primers FP CaERG11-His and RP CaERG11-URA3 (Table S1) to generate BWP17-CaErg11-His 6 .

Generation of double heterozygous strains
Double heterozygous mutant strains were generated as reported earlier (13). Briefly, appropriate selection markers (ARG4 or HIS1) were amplified by PCR using primers that included sequences of the gene to be disrupted as well as the nutritional marker and a heterozygous strain transformed with

GPI biosynthesis and Ras signaling in C. albicans
the amplicon (39). Selection was done on S.D. His Ϫ Arg Ϫ plates, and colonies were confirmed by PCR using gene-flanking primers (Table S1).

Quantification of transcript levels through RT-PCR
RNA was extracted as described previously (13). The cDNA was prepared from total RNA (3 g) using cDNA preparation kit (Bio-Rad). Transcript levels of different genes were quantified using SYBR Green PCR Master Mix (Applied Biosystems) by comparative C t method using RT primers (Table  S1) (44). GAPDH was taken as an internal control for all the experiments.

Preparation of microsomes from C. albicans
Microsomes from C. albicans were prepared using a protocol previously standardized in our lab (13) with minor modifications. Tris-EDTA (TE) buffer (50 mM Tris-HCl, pH 7.5, 10 mM EDTA) was used in place of TM buffer (Tris-MgCl 2 ) to improve GPI-GnT activity. Briefly, cells were pelleted down from secondary cultures (200 ml) of each strain grown overnight at 30°C, washed with ice-cold TE buffer, resuspended in 4 ml of same buffer, and lysed by vortexing with glass beads. Cell lysates were centrifuged at 1000 ϫ g for 10 min at 4°C, and the supernatant was centrifuged at 12,000 ϫ g for 15 min at 4°C. Ice-cold 8.0 mM CaCl 2 was added dropwise to the supernatant with constant stirring at 4°C. The sample was centrifuged at 8000 ϫ g and 4°C for 10 min to pellet out microsomal membranes. The pellet was resuspended in 300 l of ice-cold Tris-EDTA buffer (50 mM, pH 7.5) containing 10% glycerol, aliquoted in microcentrifuge tubes, and flash-frozen in liquid nitrogen for storage at Ϫ80°C.

GPI-GnT assay
GPI-GnT assay was performed using microsomes (ϳ1500 g of total protein) as described previously (45). Briefly, the microsomes were incubated at 30°C for 1 h with UDP-[6-3 H]GlcNAc (1-2 Ci; 60 Ci/mmol) in presence of tunicamycin (0.25 mg/ml) in a 120-l reaction volume. Lipids were extracted in 10:10:3 chloroform/methanol/water, dried under nitrogen, resuspended in water-saturated n-butanol, and partitioned against water. The butanol phase was dried, and glycolipids were dissolved in 10 -20 l of water-saturated butanol, spotted on HPTLC plates, and resolved in 65:25:4 chloroform/methanol/water. Radiolabeled glycolipids were detected and quantified by Bioscan AR-2000 TLC scanner using Winscan software. Because endogenous Gpi12 present in the microsomes deacetylate the product of the GPI-GnT reaction, both [6-3 H]GlcNAc-PI and [6-3 H]GlcN-PI are detected in our assays. The sum of the areas under the peaks for the two species obtained by integration is used to quantify the activity of the microsomal fraction. The total radioactive counts obtained in the control strain was taken as 100%, and activity of all other strains was calculated relative to the control strain. Heat-killed microsomes were used as the negative control.

Hyphal morphology
Hyphal morphology was monitored in different strains as described previously (13). Equal numbers of cells (0.1 OD 600 nm ) of each strain were spotted on YEPD medium plates and Spider medium plates and incubated at 37°C. For hyphal growth in the presence of tamoxifen (52 M), the plates were incubated at 30°C for 10 -15 days. Hyphal morphology was observed in Nikon SMZ1500 microscope. Images at different time intervals (7-10 days) were captured. Hyphal morphology in liquid media was also monitored in different strains as described previously (13).

Heat-shock sensitivity assay
Sensitivity to heat shock was monitored in different strains as described previously (16). Briefly, equal numbers of cells before and after heat treatment at 48°C for 8 min were spotted on YEPD plates and incubated at 30°C. Growth was monitored for 2-3 days, and images were captured at different time intervals.

Immunofluorescence
Primary cultures (10 ml) for each strain were grown until late log phase. Secondary culture cells were grown to early log phase and fixed at room temperature in 3.5% formaldehyde for 30 min. The cells were washed with 1ϫ PBS, resuspended in lyticase buffer (Tris-Cl, pH 7.2, 0.1 M MgCl 2 , sorbitol), and incubated with 2 l of lyticase (Sigma) for 20 -30 min to digest cell walls. The cells were permeabilized with 1% Triton X-100 in 1ϫ PBS for 15-20 min at room temperature. The cells were incubated in blocking solution (1% BSA in 1ϫ PBS) for 30 min at room temperature. Primary antibody (anti-Ras1; 05-516; Merck-Millipore) was added to cells and incubated overnight at 4°C. The cells were washed twice with 1ϫ PBS and incubated with secondary antibody (goat anti-mouse IgG-TRITC;

GPI biosynthesis and Ras signaling in C. albicans
11503801A; GeNei) for 2 h at room temperature. After washing twice with 1ϫ PBS, the cells were resuspended in 80% glycerol. The cells were transferred onto polylysine-coated coverslips and mounted onto glass slides. Images of immunostained cells were recorded using a confocal microscope (Olympus Fluoview FV1000).
To visualize CaGpi2-mRFP expression levels in the C. albicans strains, because protein expression levels for GPI-GnT subunits are low (15,46), immunofluorescence experiments with rabbit anti-mRFP antibodies (Abcam) and TRITC-labeled goat anti-rabbit IgG antibodies (GeNei) were performed. The fluorescence intensity in the cells was quantified using Olympus Fluoview software.

FRET analysis
For FRET, BWP17-CaGpi2-His 6 was used as experimental strain, and BWP17-CaErg11-His 6 was used as negative control strain. In experimental strain BWP17-CaGpi2-His 6 , CaRas1-FITC was the fluorescence donor, and CaGpi2-His 6 -TRITC was the fluorescence acceptor. In negative control strain BWP17-CaErg11-His 6 , CaRas1-FITC was the donor and CaErg11-His 6 -TRITC was used as the acceptor. The tagging did not affect the functioning of the protein, and the strains behave like the WT in each case. The primary antibody His probe sc-803 (Santa Cruz Biotechnology) was used to detect His 6 -tagged proteins. Samples were prepared as described under "Immunofluorescence." Acceptor photobleaching method was used for estimating FRET (47). This method measures the increase in donor emission upon acceptor photobleaching. Secondary antibodies conjugated to the FITC (goat anti-mouse IgG-FITC, 112038001A, GeNei) and TRITC (goat anti-mouse IgG-TRITC, 11503801A, and/or goat anti-rabbit IgG-TRITC, 115028001A, GeNei) were used to generate the donoracceptor pairs. An Olympus Fluoview FV1000 confocal microscope with a ϫ100 oil immersion plan-apochromat lens with a numerical aperture of 1.4 with ϫ3 optical zoom was used for image acquisition. FITC was excited with a 488-nm HeNe laser at 8% power, and emission was detected after passage through a 500 -530-nm bandpass filter. TRITC was excited with a 543-nm HeNe laser at 10% power, and emission was detected after passage through a 550 -590-nm long-pass filter. Photobleaching of TRITC, the acceptor dye, was performed with a 543-nm HeNe laser at 100% power. A control neighboring cell was also monitored before and after photobleaching to ensure that the photobleaching was specific. Images were acquired in a specific order, i.e. an image of FITC alone (pre-bleach donor) and an image of TRITC alone (pre-bleach acceptor), followed by photobleaching of TRITC, another image of FITC alone (post-bleach donor), and an image of TRITC (post-bleach acceptor) to show complete photobleaching of the acceptor molecule. FRET intensity was expressed as the increase in fluorescence (post-bleach donor minus pre-bleach donor) in arbitrary units. FRET efficiency was obtained by the increase in fluorescence normalized to the fluorescence intensity of the pre-bleach donor. As a negative control, CaRas1 was detected using anti-Ras antibody and a FITC-labeled secondary antibody, whereas CaErg11-His 6 was detected with anti-His antibody and a secondary antibody tagged with TRITC in BWP17-CaErg11-His 6 strain and analyzed for FRET. Background corrections were done using an ROI outside the cell. Bleedthrough corrections were done by monitoring fluorescence in the red channel in control samples that were stained with the FITC-labeled antibody alone.

Statistical significance of data
Unless otherwise stated, statistical significance of the data (p value) was calculated in SigmaPlot 8.0 using Student's t test. The p value Յ0.05 is considered not significant and is depicted by n.s.; p value Յ0.01 is depicted by *; p value Յ0.001 is depicted by **; and p value Յ0.0001 is depicted by ***.

Macrophage-mediated killing and phagocytic assays
Maintenance of cell line-Murine alveolar macrophage (MH-S) cell line was grown in RPMI 1640 medium (Sigma) supplemented with 10% fetal bovine serum (Gibco, Life Technologies, Inc.), 50 g/ml gentamicin sulfate, in a humidified atmosphere containing 5% CO 2 at 37°C. Exponentially growing cells were used in all experiments. Cell line was maintained as adherent cultures and subcultured by trypsinization. Live cell recoveries were counted using trypan blue exclusion method using hemocytometer.
CFSE labeling of C. albicans-C. albicans cells (100 million) were labeled with 10 M CFSE (Sigma) for 30 min at room temperature followed by two washes with phosphate-buffered saline at 2000 ϫ g for 5 min at 4°C. Flow cytometric analysis by a FACSCalibur flow cytometer (BD Biosciences) at FL1 channel using CellQuest software indicated that by using this protocol more than 95% of the C. albicans was labeled with CFSE, and labeling was stable up to 18 h. For all FACS experiments, relative fluorescence intensity of 10,000 cells was recorded as single parameter histograms (log-scale 1024 channels, 4 decades).
Confocal microscopy-For visualization of uptake of C. albicans by MH-S cells, 0.3 million cells were cultured on glass coverslips overnight. Cells were then co-cultured with CFSElabeled BWP17 and CaGPI2Hz cells at m.o.i. (1:5) for 3 h at 37°C. Cells were then washed, fixed with 2% paraformaldehyde (PFA), followed by washing twice with quencher (ammonium chloride), and examined using a confocal laser-scanning microscope (Olympus FluoView FV1000). Five images each were captured having Z-sections (depths of 0.1 m).
Co-culture assay in vitro-MH-S cells (0.3 million) cultured in 24-well plates were incubated with CFSE-labeled BWP17 and CaGPI2Hz for 3 h at different m.o.i. (1:5) for indicated time periods at 37°C in a CO 2 incubator. The cells were then harvested with PBS and fixed in 2% PFA. Uptake of stained BWP17 and CaGPI2Hz by MH-S cells was assessed by flow cytometry. For monitoring phagocytosis-independent uptake of C. albicans, MH-S cells were incubated for 1 h with Cyt D (Sigma) (2.5 g/ml) to inhibit actin polymerization and cytoskeletal rearrangement (48). These cells were then washed and incubated with stained BWP17 and CaGPI2Hz strains for 3 h following similar conditions and procedures as mentioned above.
Macrophage-mediated killing of BWP17 and CaGPI2H-To determine the extent of killing and/or phagocytosis, 0.1 million MH-S cells were seeded in 48-well cell culture plates and kept for adherence followed by addition of C. albicans at m.o.i. 1:5