Modulation of azole sensitivity and filamentation by GPI15, encoding a subunit of the first GPI biosynthetic enzyme, in Candida albicans

Glycosylphosphatidylinositol (GPI)-anchored proteins are important for virulence of many pathogenic organisms including the human fungal pathogen, Candida albicans. GPI biosynthesis is initiated by a multi-subunit enzyme, GPI-N-acetylglucosaminyltransferase (GPI-GnT). We showed previously that two GPI-GnT subunits, encoded by CaGPI2 and CaGPI19, are mutually repressive. CaGPI19 also co-regulates CaERG11, the target of azoles while CaGPI2 controls Ras signaling and hyphal morphogenesis. Here, we investigated the role of a third subunit. We show that CaGpi15 is functionally homologous to Saccharomyces cerevisiae Gpi15. CaGPI15 is a master activator of CaGPI2 and CaGPI19. Hence, CaGPI15 mutants are azole-sensitive and hypofilamentous. Altering CaGPI19 or CaGPI2 expression in CaGPI15 mutant can elicit alterations in azole sensitivity via CaERG11 expression or hyphal morphogenesis, respectively. Thus, CaGPI2 and CaGPI19 function downstream of CaGPI15. One mode of regulation is via H3 acetylation of the respective GPI-GnT gene promoters by Rtt109. Azole sensitivity of GPI-GnT mutants is also due to decreased H3 acetylation at the CaERG11 promoter by Rtt109. Using double heterozygous mutants, we also show that CaGPI2 and CaGPI19 can independently activate CaGPI15. CaGPI15 mutant is more susceptible to killing by macrophages and epithelial cells and has reduced ability to damage either of these cell lines relative to the wild type strain, suggesting that it is attenuated in virulence.

Chromosomal disruption of CaGPI15 gene. Heterozygous (CaGPI15Hz) and conditional null (Cagpi15 null) mutants of CaGPI15 were generated in the C. albicans BWP17 strain using a PCR based approach 15,16 . CaGPI15Hz had one allele of CaGPI15 disrupted with a HIS1 nutritional marker 17 . Cagpi15 null strain was made in the CaGPI15Hz background with the second CaGPI15 allele placed under the control of the repressible MET3 promoter. Since URA3 is known to alter gene expressions in C. albicans 18 , one copy of URA3 was inserted at the RPS1 locus in BWP17 (BWP17URA3) as well as in CaGPI15Hz (CaGPI15Hz-URA3) and these were used as controls in studies on all mutants that involved use of URA3 as a selection marker. The downregulation of CaGPI15 expression levels were confirmed by transcript level analysis (Supplementary Fig. 2A).
Depletion of CaGpi15 affects growth of C. albicans. The growth of CaGPI15Hz, on solid and liquid medium was comparable to that of the wild type strain ( Fig. 1A(i,ii)). The Cagpi15 null on the other hand, grew slower on solid minimal media containing Met/Cys ( Fig. 1A(iii)). Further, in liquid medium, the doubling time for the Cagpi15 null in the presence of 10 mM Met/Cys was found to be higher than in the absence of Met/Cys ( Fig. 1A(iv); Supplementary Table 2).

Depletion of CaGpi15 results in reduced GPI-GnT activity in C. albicans.
We previously showed that the GPI-GnT activity was reduced in the CaGPI15Hz strain 17 . Not only was the GPI-GnT activity significantly lower in the CaGPI15Hz (~50%) with or without the URA3 marker as compared to the wild type controls it was further reduced in the Cagpi15 null strain (28% activity) under repressive conditions of growth (Fig. 1C). Under permissive growth conditions in the Cagpi15 null there is no significant decrease in the GPI-GnT activity (47%) as compared to CaGPI15Hz-URA3 (Fig. 1C). That a drop in GPI-GnT activity of roughly 50% does not seem to cause a corresponding reduction in growth of the strain would suggest that relatively low levels of GPI biosynthetic activity are sufficient for the growth of C. albicans. This has also been reported in other GPI biosynthetic mutants 9,11,13,19 . However, when GPI biosynthesis drops below a certain threshold, as is seen in the conditional null strain under repressive growth conditions, then it affects the growth of the fungus. GPI-GnT activity was restored in CaGPI15 revertant strain where one allele of CaGPI15 was reintroduced in the CaGPI15Hz using the constitutively active pACT1 promoter (Fig. 1C).
Depletion of CaGpi15 causes cell wall defects. The CaGPI15 knock-down mutants showed several cell wall defects, including increased clumping when grown to near saturation levels and lower chitin and beta glucan levels in the cell wall versus the wild type strain ( Supplementary Fig. 2B,C; Supplementary Table 3). The cell wall defects were reversed in the CaGPI15 revertant strain ( Supplementary Fig. 2B,C(i-iv)), suggesting that the cell wall defects were specifically due to depletion of CaGpi15. viously 17 , hyphal growth of CaGPI15Hz was noticeably lesser than that of the wild type BWP17 strain on solid as well as liquid hyphae-inducing media at 37 °C but was restored in the CaGPI15 revertant cells (Fig. 1B(i-iv); Fig. 1D(i-iv); Supplementary Fig. 3A), suggesting that this effect was specific to CaGPI15.
We have previously shown that the Ras/cAMP dependent PKA activity is altered in mutants of the first step of GPI anchor biosynthesis in C albicans 12,13,17 . Hyperactive Ras mutants are heat shock sensitive 13,20 and reduced Ras signaling correlates with heat-shock resistance 13 . CaGPI15Hz was resistant to heat shock as compared to BWP17 (Fig. 1E(i)), suggesting that Ras-dependent cAMP/PKA signaling was decreased in this mutant. C. albicans has two Ras proteins, CaRas1 and CaRas2 of which CaRas1 is known to be the major determinant of hyphal growth 21 . Overexpression of CaRAS1 restores filamentation in CaGPI5Hz while overexpression of CaRAS2 does not ( Fig. 1E(ii,iii); Supplementary Fig. 3B).
CaGPI15 mutant strains are sensitive to azoles due to compromised ergosterol biosynthesis. CaGPI15Hz as well as Cagpi15 null cells were sensitive to azoles as compared to controls ( Fig. 2A(i,ii); Supplementary Fig. 4A(i,ii)). This sensitivity was reversed in the CaGPI15 revertant strain ( Fig. 2A(iii); Supplementary Fig. 4A(iii)). Azoles target CaErg11, the lanosterol demethylase in the ergosterol biosynthetic pathway of C. albicans 22 . Hence, the levels of CaERG11 transcripts in cells of CaGPI15Hz and Cagpi15 null were examined. CaERG11 levels were significantly reduced in both cases ( Supplementary Fig. 4B(i)). CaERG11 levels were restored in the CaGPI15 revertant strain ( Supplementary Fig. 4B(ii)). The reduction in CaERG11 levels (iv) Cagpi15 null mutant shows growth defect in liquid cultures. Cagpi15 null was grown both in absence (p) and presence (r) of 10 mM Met/Cys in liquid medium. For liquid cultures, cell growth for the various strains was monitored by OD 600nm at different time points and doubling times are calculated and mentioned in Supplementary Table 2. The experiment was done three times in duplicates; arithmetic mean with standard deviations is shown. For solid media experiments, a 5 µl suspension of cells corresponding to 1 × 10 7 , 2 × 10 6 , 4 × 10 5 , 8 × 10 4 and 1.6 × 10 4 numbers were spotted from left to right in each row. The experiments were done thrice using independent cultures. (B) CaGPI15 is required for filamentation. The hyphal growth and quantification of hyphal growth in CaGPI15Hz for up to 120 min in (i,ii) liquid spider media and in (iii,iv) liquid RPMI with 10% serum at 37 °C. A minimum of 100 cells were used for the statistical analysis. The arithmetic mean with standard deviation is plotted. (C) CaGPI15 depletion reduces GPI-GnT activity. GPI-GnT activity was tested in the CaGPI15 mutants as described in Materials and Methods. (D) Hyphal growth in Cagpi15 null and CaGPI15 revertant. The hyphal growth and quantification of hyphal growth in Cagpi15 null and CaGPI15 revertant for up to 90 min in (i,ii) liquid spider media and in (iii,iv) liquid RPMI with 10% serum at 37 °C. A minimum of 100 cells were used for the statistical analysis. The arithmetic mean with standard deviations is plotted. (E) CaRas1 is responsible for the filamentation phenotypes of the CaGPI15 mutants. (i) Cells of the indicated strains were spotted on SD-agar plates and incubated at 30 °C for 48 h after a 10 min heat shock at 48 °C; (ii) and (iii) The hyphal growth and quantification of hyphal growth in various strains for up to 120 minutes in liquid RPMI with 10% serum media at 37 °C. CaRas1 restores filamentation in CaGPI15Hz. These experiments were done twice in duplicates.
www.nature.com/scientificreports www.nature.com/scientificreports/ correlated well with the accumulation of lanosterol, the substrate of CaErg11, and a reduction in total ergosterol levels in the CaGPI15Hz and Cagpi15 null strains (Fig. 2B).
Upc2 is a transcription factor that controls CaERG11 levels in the cell 23 . The occupancy of both RNA Pol II (RNAPII) as well as Upc2 were significantly reduced on the promoter of CaERG11 in CaGPI15Hz-URA3 as compared to BWP17URA3 when probed using primer pair 2 ( Fig. 2C(i,ii)), suggesting that the promoter had reduced accessibility for transcription. This also correlated well with the fact that acetylation of histone H3, as assessed using the H3K56 antibody, was significantly lower on the promoter of CaERG11 in CaGPI15Hz-URA3 as compared to that in BWP17URA3 (Fig. 2C(ii) & Supplementary Fig. 4D).
The role of histone acetylation in the regulation of CaERG11 expression in C. albicans was further investigated. H3K56 and H3K9 acetylation are mediated by Rtt109, a histone acetyltransferase 24,25 . As can be seen from ( Fig. 2D(i,ii); Supplementary Fig. 5D), the expression of RTT109 at the mRNA as well as protein levels was not The H3K56Ac levels in whole cell lysate were estimated in the mutants relative to BWP17URA3. H3 levels were taken as a loading control. A cropped representative image is shown, and the full-length image is displayed in Supplementary Fig. 5D (iii) The CaERG11 promoter occupancy of RNA pol II, H3K56Ac and H3K9Ac in the mutants relative to BWP17URA3. (E) Azole sensitivity was reversed in CaGPI15Hz/pACT1-RTT109 strain. Growth curve analysis of CaGPI15Hz/pACT1-RTT109 mutant in 0.2 μg/ml ketoconazole. The doubling time of CaGPI15Hz/pACT1-RTT109 in the presence of ketoconazole is almost similar with the wildtype strain grown in the same condition (Supplementary Table 2). (F) CaGPI15Hz/pACT1-RTT109 strain accumulates ergosterol. Relative quantification of ergosterol was done using GC-MS. P-values for CaGPI15Hz were calculated relative to the wild type while for CaGPI15Hz/pACT1-RTT109 it was relative to CaGPI15Hz-URA3.
www.nature.com/scientificreports www.nature.com/scientificreports/ significantly altered in the CaGPI15Hz strain. RTT109 was then expressed under the control of pACT1 promoter in the CaGPI15Hz strain and its overexpression was confirmed ( Fig. 2D(i)). CaERG11 transcript levels were upregulated in these cells ( Fig. 2D(i)). ChIP analysis showed that overexpression of RTT109 in CaGPI15Hz cells increased acetylation of H3, as assessed by anti-H3K56 or anti-H3K9 antibodies, and increased the occupancy of RNAPII on the CaERG11 promoter ( Fig. 2D(iii)) Overexpression of RTT109 in CaGPI15Hz cells also reversed the azole sensitivity ( Fig. 2E; Supplementary Fig. 4C) and restored ergosterol levels in these cells (Fig. 2F), confirming that acetylation of H3 by Rtt109 regulates the expression of CaERG11 in this strain.
Separately, in order to ensure that the reduced H3 acetylation was not due to a global reduction in acetylation or due to defects in nucleosome assembly, we checked the H3 acetylation levels at two different intergenic regions (Chromosome 5 and Chromosome R) in BWP17URA3, CaGPI15Hz-URA3 and CaGPI15Hz-pACT1-RTT109 by ChIP ( Supplementary Fig. 4E(i,ii). No reduction in H3 acetylation was observed in the two strains at these positions relative to the control, BWP17URA3, suggesting that the ChIP signal observed is not a nucleosome assembly dependent effect and was instead due to a reduction in acetylation levels at the promoter of CaERG11.
The acetylation mediated by Rtt109 is catalyzed with two histone chaperones, Vps75 and Asf1 26 . While the latter has been shown to be important for H3K56Ac alone, the former has been shown to be important for acetylation of H3 at K9 as well K56 25 . A VPS75Hz mutant should, therefore, mimic a strain defective in acetylation by Rtt109. To confirm that acetylation of H3 at the CaERG11 promoter was important for its regulation in C. albicans, ChIP analysis was done in a VPS75Hz strain generated in the lab. The acetylation of H3 dropped to a significant extent at the promoters of CaERG11, CaGPI2, CaGPI15 and CaGPI19 in this strain when probed using anti-H3K56 antibody ( Supplementary Fig. 4F(i-iv)).
CaGPI2 and CaGPI19 levels in the CaGPI15 heterozygous strain can also be restored by RTT109 overexpression. Transcript levels of both CaGPI2 and CaGPI19 were found to be reduced in CaGPI15Hz as compared to BWP17 (Fig. 3A(i)) and were restored in the CaGPI15 revertant strain (Fig. 3A(ii)), suggesting that this effect was specifically linked to CaGPI15 levels. It should be noted that the expression of two housekeeping genes (CaUBC13, CaACT1), two ergosterol biosynthesis genes downstream to CaERG11 (CaERG3, CaERG4) and www.nature.com/scientificreports www.nature.com/scientificreports/ three other downstream GPI biosynthetic genes (CaGPI12, CaGPI14, CaGPI8) were all found to not be significantly altered in the CaGPI15Hz or the CaGPI15 revertant strains ( Supplementary Fig. 5A).
Further, the acetylation of histone H3 on the promoter of CaGPI15, CaGPI2 and CaGPI19 in the CaGPI15Hz strain was found to be significantly reduced when probed using anti-H3K56 antibody ( Fig Similarly, to confirm that CaGPI19 functions below CaGPI15 in controlling CaERG11 levels and sensitivity to azole drugs, CaGPI15Hz/CaGPI19Hz and CaGPI15Hz/ pACT1-CaGPI19 strains were generated. In the former, one allele of CaGPI19 was disrupted and in the latter CaGPI19 was overexpressed in the CaGPI15Hz background. CaERG11 transcripts were reduced in CaGPI15Hz/ CaGPI19Hz and upregulated in CaGPI15Hz/pACT1-CaGPI19 as compared to the parent strain (Fig. 4B). The sensitivities of these strains to azoles also correlated with their CaERG11 levels. CaGPI15Hz/CaGPI19Hz was more sensitive to azoles while CaGPI15Hz/pACT1-CaGPI19 was resistant to azoles as compared to CaGPI15Hz   Table 2). The experiments were repeated twice in duplicates. (iv) Relative quantification of ergosterol was done in different mutants of CaGPI19 using GC-MS. P-values for all heterozygous mutants were calculated with respect to the wild type controls while that for overexpression mutants are with respect to CaGPI19Hz-URA3.
www.nature.com/scientificreports www.nature.com/scientificreports/ The cross-talk between CaGPI2, CaGPI15 and CaGPI19. The data presented in Fig. 3A, suggested that CaGPI15 is an activator of both CaGPI2 and CaGPI19. To examine the interaction between CaGPI2 and CaGPI19 in the CaGPI15Hz background, we studied two double heterozygous strains, CaGPI15Hz/CaGPI2Hz and CaGPI15Hz/CaGPI19Hz. The levels of CaGPI19 were increased in CaGPI15Hz/CaGPI2Hz while that of CaGPI2 were increased in CaGPI15Hz/CaGPI19Hz. Overexpressing either CaGPI2 or CaGPI19 in CaGPI15Hz resulted in downregulation of the other (Fig. 5A). Thus, the mutually negative regulation between CaGPI2 and CaGPI19 continues to function in the CaGPI15Hz strain.
GPI-GnT activity assays also corroborate such a model. Overexpression of CaGPI15 in CaGPI2Hz or CaGPI19Hz pushes up transcript levels of both CaGPI2 and CaGPI19 and causes enhanced GPI-GnT activity CaGPI2 and CaGPI19 are mutually negatively regulated and function downstream of CaGPI15 to control Ras signaling and CaERG11 levels, respectively, in the organism. Azole response via regulation of CaERG11 continues to correlate with CaGPI19 levels while Ras1-dependent hyphal morphogenesis continues to correlate with CaGPI2 levels. (E) CaGPI15 can mediate active GPI-GnT complex formation in CaGPI2Hz and CaGPI19Hz strains. Relative GPI-GnT activity in CaGPI15Hz strain overexpressing either CaGPI2 or CaGPI19 as well as in CaGPI2Hz and CaGPI19Hz strains overexpressing CaGPI15. P-values for heterozygous mutants were calculated with respect to the wild type controls while that for double deletion or overexpression mutants was calculated relative to the heterozygous parent control.
www.nature.com/scientificreports www.nature.com/scientificreports/ in comparison to the parent strains (Fig. 5E). However, due to the negative regulation between CaGPI2 and CaGPI19, overexpressing either CaGPI2 or CaGPI19 in the CaGPI15Hz strain cannot enhance its GPI-GnT activity (Fig. 5E).
CaGPI15 heterozygous strain is more susceptible to killing by MH-S macrophage cells and is less able to kill the macrophage cell line. We tested the effect of CaGPI15 knock-down on C. albicans virulence. For this, a murine alveolar macrophage cell line (MH-S) was co-cultured with BWP17 and CaGPI15Hz strain for 3 h (Fig. 6A,B) and 18 h (Fig. 6C,D). Both strains of Candida formed hyphae when co-cultured with MH-S cells for 18 h. BWP17 co-cultured with MH-S had 55% longer hyphae than CaGPI15Hz co-cultured with MH-S for 18 h (Fig. 6E). The internalization of C. albicans cells by MH-S was roughly similar for the CaGPI15Hz and BWP17 strains (Fig. 6F). This was also the case when phagocytosis was inhibited with the help of cytochalasin D (Cyt D) (Fig. 6F). Thus, mutation of CaGPI15 does not appear to alter the phagocytosis of C. albicans cells by either cytoskeleton-dependent or independent pathways. The colony forming units (CFU) recovered after incubation with the macrophage cells was significantly lower in CaGPI15Hz as compared to BWP17 (Fig. 6G). Hence, more numbers of CaGPI15Hz cells were killed by MH-S in comparison to BWP17. In addition, more live MH-S cells were recovered after co-culture with CaGPI15Hz as compared to BWP17, suggesting that CaGPI15Hz was less virulent (Fig. 6H). There was also no difference in cell death by pyroptosis after co-culturing MH-S with BWP17 and CaGPI15Hz at 1:5 multiplicity of infection (MOI) for 3 h and 18 h (Fig. 6I). But there is a significant difference in pyroptosis between 3 h and 18 h for MH-S co-cultured with either strain. Similarly, there was no difference in cell death by apoptosis (Annexin V staining) when MH-S cells were co-cultured with either with MH-S (0.3 million) were seeded in 24 well cell culture plate and allowed to adhere overnight in CO 2 incubator at 37 °C. BWP17 and CaGPI15Hz at MOI 1:5 were added for 3 h and 18 h and processed as discussed in Methods. Cells were harvested and 1 µl Annexin V APC was added. Each point represents mean ± SEM of values obtained from three independent assays. *p ≤ 0.05, **p ≤ 0.005 and ***p ≤ 0.0005 represent statistically significant difference between control and treated cells, ns is no statistically significant difference. The significance of any difference was calculated by using one-tailed distribution in a two-sample equal variance student's t test. (2019) 9:8508 | https://doi.org/10.1038/s41598-019-44919-4 www.nature.com/scientificreports www.nature.com/scientificreports/ BWP17 or CaGPI15Hz strain at 1:5 MOI for 18 h. Nevertheless, the level of apoptosis was significant in MH-S on infection with either strain of C. albicans for 18 h. So, C. albicans seems to cause MH-S macrophage cell death by apoptosis as well as pyroptosis.
CaGPI15 heterozygous strain is more susceptible to killing by LA-4 epithelial cells and is less able to kill the epithelial cell line. We further investigated the interaction of CaGPI15Hz cells with an epithelial cell line. Murine epithelial cell line LA-4 cells were co-cultured with BWP17 and CaGPI15Hz strain for 3 h (Fig. 7A,B) and 18 h (Fig. 7C,D). Both strains formed hyphae on co-culture with LA-4 cells for 18 h (Fig. 7C,D) but BWP17 exhibited 51% longer hyphae than CaGPI15Hz (Fig. 7E). CaGPI15Hz cells were more susceptible to killing by LA-4 cells as significantly lower CFU were obtained for CaGPI15Hz co-cultured with LA-4 in comparison to that for BWP17 (Fig. 7F). Phagocytosis of BWP17 cells by LA-4 cells was significantly higher than that of CaGPI15Hz cells at MOI 1:5 (Fig. 7G), suggesting a role for fungal GPI anchored proteins in their uptake 27 . However, the live cell recovery of LA-4 cells was significantly higher for those co-cultured with CaGPI15Hz versus those with BWP17 at MOI 1:5 (Fig. 7H). No significant difference in phagocytosis or live cell recovery was observed in experiments using either of these strains at MOI 1:1. Significantly lower pyroptosis (~30% of wildtype) was seen in LA-4 cells when co-cultured for 18 h with CaGPI15Hz in comparison to those co-cultured with BWP17. No pyroptosis was seen when LA-4 cells were infected with either strain for 3 h. Further, we found a small but significantly higher degree of apoptosis in LA-4 cells infected with BWP17 cells versus CaGPI15Hz after were seeded in 24 well cell culture plate and allowed to adhere overnight in CO 2 incubator at 37 °C. BWP17 and CaGPI15Hz at MOI 1:5 were added for 3 h and 18 h and processed as discussed in Methods. Cells were harvested and 1 µl Annexin V APC was added. Each point represents mean ± SEM of values obtained from three independent assays. *p ≤ 0.05, **p ≤ 0.005 and ***p ≤ 0.0005 represent statistically significant difference between control and treated cells, ns is no statistically significant difference. The significance of any difference was calculated by using one-tailed distribution in a two-sample equal variance student's t test. (2019) 9:8508 | https://doi.org/10.1038/s41598-019-44919-4 www.nature.com/scientificreports www.nature.com/scientificreports/ incubation for 18 h at 1:5 MOI (Fig. 7J). Hence, the killing induced by C. albicans infection in epithelial cells also involved apoptosis and pyroptosis. The data from epithelial cell infection studies also supports the notion that the CaGPI15Hz strain is attenuated in virulence.

Discussion
The GPI anchor glycolipid is produced in the endoplasmic reticulum in 10-12 sequential biochemical steps. In lower eukaryotes this pathway is essential to the growth and viability of the organism while in higher eukaryotes it is critical only at certain stages of organismal development, such as in embryogenesis. Given their essentiality for eukaryotic pathogens, several steps of the pathway have been the focus of study as probable drug targets 28,29 . However, isolating and studying the individual enzymes of the GPI biosynthetic pathway is challenging because it involves mostly multi-subunit membrane-bound enzymes of relatively low abundance. No high-resolution X-ray crystallographic data are available for any of the enzymes till date and there are no commercially available substrates for most steps of the pathway. The study of the GPI biosynthetic pathway in C. albicans is made more challenging due to its codon biasness. C. albicans proteins heterologously expressed in another host may at times need to be codon optimised for function. Further, until recently no cell free assay system was available for the GPI biosynthetic pathway in this organism since protocols used to generate microsomes from the closely related yeast, S. cerevisiae, did not yield active microsomes from C. albicans 13,19 .
Despite the many challenges, studying the GPI anchor biosynthesis of different organisms without depending solely on model organisms can be very rewarding. For example, the first and committed step of the GPI biosynthesis pathway in eukaryotes is an important site for regulation. The presence of poorly conserved accessory subunits to assist the highly conserved catalytic subunit suggests that the regulation of the GPI biosynthetic pathway could be mediated via these proteins. Some evidence in support of such a hypothesis also exists. In S. cerevisiae, Ras2 was shown to inhibit the GPI-GnT complex and vice versa while no such regulation was observed in mammals 4,6,30 . Similarly, Dpm2 is known to regulate the mammalian GPI-GnT, but an equivalent regulation is not observed in other organisms 31 . In a series of papers establishing such a link in C. albicans, we showed that the GPI-GnT complex interacts with and regulates two other important pathways, ergosterol biosynthesis and hyphal morphogenesis in C. albicans [11][12][13] . Specifically, CaGpi19 controls CaERG11 levels and modulates azole drug response while CaGpi2 regulates hyphal morphogenesis by controlling Ras signaling. Moreover, CaRas1, the C. albicans homolog of S. cerevisiae Ras2, activates rather than inhibits the GPI-GnT activity 17 . In turn, the CaRas1-dependent PKA signaling pathway is activated by CaGpi2, but this is independent of the GPI-GnT activity itself 17 .
In the present study, we examined a third subunit of the GPI-GnT complex to elucidate its importance for C. albicans. CaGpi15, like CaGpi2 and CaGpi19, is a poorly conserved protein sharing very low sequence homology with its yeast counterpart. Yet, at the functional level it complements a conditionally lethal S. cerevisiae gpi15 mutant. Thus, significant functional similarities exist between the Gpi15 homologs in the two organisms. In C. albicans, as in S. cerevisiae, CaGPI15 appears to be important for cell growth. Gene-dosage effects also operate here since the heterozygous mutants were only marginally affected in doubling times while the conditional null showed significantly longer doubling times as compared to the wild type strains. Cell wall defects and clumping observed in CaGPI15 mutants appear to be a result of GPI anchor deficiency since these have also been observed in other GPI biosynthetic mutants 9,11-13 .
CaGPI15 mutants were ergosterol deficient and sensitive to azoles. Thus, in its response to azoles, it appeared that CaGPI15 mutants mirrored the CaGPI19 mutants. Yet, surprisingly, the hyphal morphogenesis phenotype of CaGPI15 mutant cells was quite unlike the hyperfilamentous phenotype of CaGPI19 mutants 11,12 . CaGPI15 mutants were hypofilamentous as compared to the wild type strains due to reduced Ras signaling, a feature also observed in CaGPI2 mutants 13 . Thus, CaGPI15 exhibited phenotypes of a mutant that had both CaGPI2 and CaGPI19 downregulated. A transcript level analysis confirmed this. Significantly reduced histone H3 acetylation was observed on the promoters of CaGPI2, CaGPI15 and CaGPI19 in the CaGPI15 mutant strain. Overexpression of RTT109 could restore H3Ac as well as the transcript levels of the three genes, suggesting that expression levels of these genes are regulated via histone H3 acetylation in the CaGPI15 mutant strain. Overexpressing CaGPI15 could also restore CaGPI2 and CaGPI19 levels. Phenotypic assays as well as GPI-GnT activity assays corroborated these results.
Using double mutants in which one allele of either CaGPI2 or CaGPI19 was disrupted in the heterozygous CaGPI15 background, we discovered that CaGPI2 and CaGPI19 continued to be mutually negatively regulated. Further, the hyphal morphogenesis and azole drug response phenotypes correlated with changes in CaGPI2 and CaGPI19 levels, respectively, in these strains.
Finally, we addressed the question of whether CaGPI2 and CaGPI19 were activators or repressors of CaGPI15. We discovered that simultaneous downregulation of CaGPI2 and CaGPI19 in C. albicans results in downregulation of CaGPI15 and overexpression of either of them activates CaGPI15. In other words, both CaGPI2 and CaGPI19 could independently activate CaGPI15. Thus, we propose a model for how the three subunits of the GPI-GnT complex interact in C. albicans (Fig. 5D): CaGPI15 stimulates activation of CaGPI2 as well as CaGPI19. Downregulating it can simultaneously decrease levels of both CaGPI2 and CaGPI19. In turn, both CaGPI2 and CaGPI19 can independently activate CaGPI15. Disrupting either CaGPI2 or CaGPI19 results in upregulation of the other, which in turn upregulates CaGPI15. Both CaGPI2 and CaGPI19 are mutually negatively regulated and function downstream of CaGPI15 as far as hyphal growth and azole drug response are concerned. Sensitivity to azoles via CaERG11 regulation correlates with CaGPI19 levels while Ras-dependent hyphal morphogenesis correlates with CaGPI2 levels.
Two additional issues needed to be addressed here. The first is the mechanism by which CaERG11 is downregulated in the CaGPI15 mutant. The presence of Upc2, a transcription factor for CaERG11, was found to be reduced on the promoter of the CaERG11 gene in CaGPI15 deficient strain. In exploring the possible reason (2019) 9:8508 | https://doi.org/10.1038/s41598-019-44919-4 www.nature.com/scientificreports www.nature.com/scientificreports/ for this, it was observed that H3Ac on the promoter of CaERG11 is also reduced in the azole sensitive CaGPI15 mutant strain. Since CaGPI19 functions downstream of CaGPI15, this strain was also tested for H3Ac and found to have reduced levels of it on the CaERG11 promoter. This is specifically due to loss of Rtt109 activity in these strains. Overexpressing RTT109 could restore the levels of the GPI-GnT subunits as well as of CaERG11 and reverse the response to azoles for both CaGPI15 and CaGPI19 mutant strains.
The second is the mechanism by which Ras signaling is altered in the CaGPI15 mutant. Overexpressing CaRAS1, but not CaRAS2, could restore filamentation in CaGPI15 mutant. Further, CaGPI2 functions downstream of CaGPI15 in this. In a recent manuscript, we showed that CaGpi2 physically interacts with CaRas1 in the endoplasmic reticulum and this interaction helps CaRas1 activate GPI-GnT activity 17 . CaGpi2 also regulates Ras signaling that occurs at the plasma membrane to trigger hyphal morphogenesis, but it does so by its effect on Hsp90 levels in the cell. It is well known that Hsp90 along with its co-chaperone, Sgt1, interacts with Cyr1, the effector of CaRas1, thereby preventing the interaction of GTP-bound CaRas1 with Cyr1 for initiating cAMP-dependent hyphal morphogenesis 32 . Overexpressing CaGPI2 results in downregulation of Hsp90 and a lifting of the inhibition exerted by Hsp90 on the Ras signaling pathway. Since CaGpi2 seems to control hyphal morphogenesis in the CaGPI15 mutant strain, it is reasonable to expect that this mechanism functions here too.
The CaGPI15 heterozygous mutant was also far more susceptible to killing by macrophages and epithelial cells in our assays. It showed reduced internalization by epithelial cells, and also shorter hyphae length in comparison to BWP17. In addition, its ability to damage macrophages and epithelial cells was significantly lesser than that of the wild type strain. The damage caused to macrophages and epithelial cells by infection with BWP17 and CaGPI15 heterozygous mutant involved apoptosis as well as pyroptosis, but significantly the pyroptosis seen in epithelial cells on infection with CaGPI15 heterozygous mutant was only 30% of that seen on infection with BWP17. This could be an important contributing factor in reduced damage to epithelial cells by this mutant. Hence, we infer that the CaGPI15 mutant strain is attenuated in virulence.
In conclusion, what is evident from our previous and current studies is that alterations in levels of different subunits of the same complex, all of which affect GPI biosynthesis, receive varying responses from ergosterol biosynthesis and hyphal filamentation pathways in C. albicans. Given the importance of GPI biosynthesis for the viability and growth of C. albicans, the multiple modes of interaction and regulation probably allow the GPI biosynthetic pathway to rapidly respond to multiple signals.
The clinical implications of these observations are hard to miss. Hyphal filamentation and invasive growth are important for the establishment of infection by C. albicans and several hyphae-specific factors are known to be GPI anchored 7,8 . Strains with reduced levels of GPI anchored proteins are known to show attenuated virulence 8 . Additionally, ergosterol and its biosynthetic pathway continue to be the most important targets of the currently available antifungals. However, one of the major problems in treatment of fungal infections has been the rapid drug resistance that develops upon continued usage of these drugs 33 . CaGpi15 as an alternative target, a master regulator that simultaneously affects hyphal morphogenesis as well as ergosterol biosynthesis could be an interesting candidate. Indeed, strains in which CaGPI15 levels are reduced are less virulent, as shown here. Additionally, GPI biosynthesis itself is an essential pathway in C. albicans and targeting the first step of GPI biosynthesis via a protein that bears little homology with its mammalian counterpart could be an effective strategy.

Methods
Strains and media. Yeast and C. albicans strains used in this study are described in Table 1. Murine alveolar epithelial type I cell line (LA-4) and Murine alveolar macrophage cell line (MH-S) were procured from the American type tissue culture collection (ATCC), Rockville, MD, USA. All primers used in this study are listed in Supplementary Table 1. Strains were grown in yeast extract-peptone-dextrose (YEPD) media or synthetic dextrose minimal media (SD media). Ura − strains were grown in YEPD or SD media supplemented with 60 µg/ ml uridine. Similarly, His − strains or Arg − strains were grown in SD medium supplemented with the appropriate amino acid (85.6 μg/ml His or Arg). Transformations were performed using lithium-acetate method 34  Eisenhaber's web site (http://mendel.imp.ac.at/SEQUENCES/gpi-biosynthesis/pigs-main.html) and also confirmed by BLASTp analysis using the mammalian PIG-H sequence as query. Forward primer FPCaGPI15-pCaDis and reverse primer RPCaGPI15-pCaDis (Supplementary Table 1) were used to amplify CaGPI15 gene from genomic DNA of C. albicans. The amplified product was visualized on 1% agarose gel.
Chromosomal disruption of CaGPI15 to make heterozygous CaGPI15 mutant. CaGPI15 heterozygote in BWP17 strain of C. albicans (CaGPI15Hz) was made by PCR-mediated disruption using HIS1 as a selection marker 15,17 . www.nature.com/scientificreports www.nature.com/scientificreports/ Complementarity between C. albicans and S. cerevisiae genes. ScGPI15 gene in S. cerevisiae YPH500 strain was placed under the glucose responsive GAL1 promoter by a PCR-mediated approach 11,35 . The transformed colonies were selected on SDUra − plates and confirmed with locus specific PCR amplification. YPH-pGAL1-ScGPI15 was unable grow on glucose media but could easily grow on media containing 4% galactose and 1.5% sucrose. For complementation studies, CaGPI15 gene, placed under the constitutive PMA1 promoter, was cloned between BamHI and MluI site into YEpHIS plasmid and used to transform this strain 36 . Preparation of microsomes from yeast and GPI-GnT assay. Primary cultures of yeast strains were grown for 16 h in 4% (w/v) galactose. It was inoculated into 250 ml SD medium containing either 4% galactose or 2% glucose at 30 °C till OD 600nm of 2.0 was reached. The cells were harvested and microsomes prepared from it as described previously 37 .

Construction of regulatable null mutant of
GPI-GnT activity of the microsomes was assayed as described previously 37 . The assay produces both [6 3 H] GlcNAc-PI (N-acetyl glucosaminylphosphatidylinositol) and [6 3 H]GlcN-PI (glucosaminylphosphatidylinositol, the product of the next step of the pathway) and were detected using Bioscan AR2000 TLC scanner. The sum of the radioactive counts detected for the two products in the case of the control strain was considered to be 100% www.nature.com/scientificreports www.nature.com/scientificreports/ Chromatin Immunoprecipitation (ChIP) assay. ChIP was carried out as described previously 13 . Briefly, cells were harvested after cross-linking DNA to protein with 1% (v/v) formaldehyde. Glycine (25 mM) was added and cells lysed by glass beads in lysis buffer (50 mM HEPES, pH 7.4; 140 mM NaCl; 1 mM EDTA; 1% Triton X-100; 1 mM PMSF). After sonication the supernatant was incubated overnight with 1 µg anti-RNA PolII or H3K56Ac antibody. ProteinA-CL agarose beads (20 µl) were added. The beads were spun down after 2 h, washed with lysis buffer, high-salt buffer (lysis buffer with 500 mM NaCl), wash buffer (10 mM Tris-Cl, pH 8.0; 250 mM LiCl; 1 mM EDTA; 0.5% NP-40) and Tris-EDTA buffer. Eluates collected in elution buffer (150 µl Tris-EDTA buffer containing 1.0% SDS) were treated with 20 µg proteinase K. DNA was separated using phenol:chloroform (1:1) and precipitated. The samples were analyzed by PCR using primers for specific regions of promoters. , 50 µg/ml gentamicin sulfate, 0.05 mM 2-mercaptoethanol, 300 µg/ml L-glutamine and 50 µg/ml and 10% heat inactivate FBS (Gibco, Life technologies, Grand Island, NY, USA) in a humidified atmosphere containing 5% CO 2 at 37 °C. Cell lines were maintained as adherent cultures and sub cultured by trypsinization. The cells were then harvested by using 0.25% w/v trypsin from bovine pancreas in 10 mM EDTA disodium salt to detach the monolayer. The harvested cells were collected and pelleted down by centrifugation at 240 g for 5 min at 4 °C. The pelleted cells were suspended in 1 ml media and counted for viability by haemocytometer using trypan blue dye.
MH-S or LA-4 cells (0.3 million) cultured in 24-well plate were incubated with CFSE labelled BWP17 and CaGPI15Hz at different MOI such as 1:1 and 1:5 for 3 h and 18 h at 37 °C in CO 2 incubator. The cells were trypsinized and then harvested with PBS. The harvested cells were fixed in 2% PFA and the uptake of stained BWP17 and CaGPI15Hz by MH-S or LA-4 cells was assessed by using BD FACS Calibur flowcytometer in FL1 channel using Cell Quest software.
Co-culture assay in vitro. Co-culture assay was done as described previously 17 . MH-S cells or LA-4 cells (0.3 million) were incubated with CFSE labelled fungal cells for 3 h at 1:1 or 1:5 MOI. The phagocytosis-independent uptake of fungal cells by MH-S was monitored using 2.5 µg/ml of Cyt D, an inhibitor of actin polymerization 39 as described previously 17 . Confocal microscopy. For visualization of uptake of C. albicans by MH-S or LA-4 cells, 0.3 million cells were cultured on glass cover slips overnight. Cells were then co-cultured with CFSE-labelled BWP17 and CaGPI15Hz at MOI (1:5) for 3 h and 18 h at 37 °C. Cells were then washed, fixed with 2% paraformaldehyde (PFA) followed by washing thrice with quencher (ammonium chloride) and examined using a confocal laser scanning microscope (Olympus FluoView FV1000). Five images each were captured having Z sections (depths 0.1 µm) 17 . The above co-cultured cells of 18 h time point were used to determine the length of hyphae. Nikon NIS element software was used to measure the length of the hyphae.
Macrophage or epithelial cell mediated killing of BWP17 and CaGPI15Hz. The macrophage or epithelial cell mediated killing of C. albicans was studied as described previously 17 . The number of C. albicans was determined as CFU/ml = number of colonies x dilution factor / volume of culture plate.
Detection of pyroptosis through lactate dehydrogenase (LDH) enzymatic assay. LA-4 and MH-S cells were cultured at a concentration of 50,000 cells/100 µl with complete media in a 96 well cell culture plates. After overnight culture, cells were washed with complete medium to remove debris and dead cells. Cells were co-cultured with BWP17 and CaGPI15Hz for 3 h and 18 h at different MOI (1:5) at 37 °C in CO 2 incubator. The co-cultured cells were then subjected for cell cytotoxicity assay. The cell cytotoxicity assay was performed using Cytotoxicity Detection Kit (Pierce TM , USA). The standard protocol assay reported here were performed according to the manufacturer's instructions. The amount of LDH released either in control (Positive control, spontaneous and maximum release) or in the experimental wells was used to calculate the % specific lysis 40 . The % specific lysis was calculated as: % Specific lysis = (Experimental release − Spontaneous release) × 100/(Maximum release − Spontaneous release).

Assessment of apoptosis. LA-4 and MH-S cells were cultured at a concentration of 0.3 million cells/ml
in a 24 well cell culture plates. After overnight culture, cells were washed with complete medium to remove debris and dead cells. Cells were co-cultured with BWP17 and CaGPI15Hz for 18 h at MOI (1:5) at 37 °C in CO 2 incubator. The cells were then washed with PBS to remove BWP17 and CaGPI15Hz. The cells were trypsinzed and washed followed by staining with annexin V APC conjugate (Biolegend, San Diego, CA, USA) to assess the apoptotic cells using BD FACS Calibur 41 and analyzed through Cell Quest Software.

Data Availability
The datasets and material generated during and/or analysed during the current study are available from the corresponding author on reasonable request.