Plasma Membrane Proteolipid 3 Protein Modulates Amphotericin B Resistance through Sphingolipid Biosynthetic Pathway

Invasive opportunistic fungal infections of humans are common among those suffering from impaired immunity, and are difficult to treat resulting in high mortality. Amphotericin B (AmB) is one of the few antifungals available to treat such infections. The AmB resistance mechanisms reported so far mainly involve decrease in ergosterol content or alterations in cell wall. In contrast, depletion of sphingolipids sensitizes cells to AmB. Recently, overexpression of PMP3 gene, encoding plasma membrane proteolipid 3 protein, was shown to increase and its deletion to decrease, AmB resistance. Here we have explored the mechanistic basis of PMP3 effect on AmB resistance. It was found that ergosterol content and cell wall integrity are not related to modulation of AmB resistance by PMP3. A few prominent phenotypes of PMP3 delete strain, namely, defective actin polarity, impaired salt tolerance, and reduced rate of endocytosis are also not related to its AmB-sensitivity. However, PMP3 overexpression mediated increase in AmB resistance requires a functional sphingolipid pathway. Moreover, AmB sensitivity of strains deleted in PMP3 can be suppressed by the addition of phytosphingosine, a sphingolipid pathway intermediate, confirming the importance of this pathway in modulation of AmB resistance by PMP3.

strain, namely defects in salt tolerance, actin polarity and endocytosis, are not responsible for AmB-sensitivity of this strain. Instead, we demonstrate that modulation of AmB resistance by PMP3 is mediated through sphingolipid biosynthetic pathway.

Results and Discussion
PMP3 modulates AmB resistance. The S. cerevisiae PMP3 gene was isolated from a multicopy overexpression library (in plasmid pFL44L) as conferring higher resistance to AmB. A PMP3 clone with 165 bp ORF along with 1196 bp upstream and 275 bp downstream regions was used in further studies. To confirm the phenotype, PMP3 deletion and overexpression strains were compared with their parent strain for AmB resistance (Fig. 1a). While the delete strain was 8-fold more sensitive to AmB than the parent strain, the overexpression strain was about 4-fold more tolerant compared to the parent strain. During the course of this study, Huang et al 13 , while establishing a functional variomics tool for discovering drugresistance genes and drug targets, also identified PMP3 as conferring AmB resistance when present at more than one copies. PMP3 (also known as SNA1) has three paralogs in S. cerevisiae, namely SNA2, SNA3 and SNA4, which encode proteins with 40%, 34% and 41% identity, respectively, to that of PMP3 16 . Deletants of these genes were comparable to the parent strain in their susceptibility to AmB (results not shown), implying that these genes do not have any role in this phenotype.
To test if PMP3 has a similar role in pathogenic yeasts, we searched for homologs in C. albicans and C. glabrata. C. albicans has two homologs, which encode proteins that show 51% and 45% identity at amino acid level to that of S. cerevisiae. The first one is referred to as CaPMP3 ortholog (orf19.1655. 3) and the second one as CaPMP3 best hit (orf19.2959.1) in Candida Genome Database 17 . C. glabrata has a single ortholog CgPMP3 (CAGL0M08552g) encoding a protein with 76% identity to ScPmp3p. The open reading frames of these homologs were PCR amplified and used to replace the ORF in ScPMP3 clone, thereby placing these ORFs under the control of ScPMP3 promoter and terminator in pFL44L vector. These were tested for their ability to modulate AmB resistance after being transformed into pmp3D strain of S. cerevisiae. PMP3 ortholog from C. albicans was earlier shown to increase AmB resistance of S. cerevisiae 13 . In addition, we found CaPMP3 best hit and CgPMP3, besides complementing pmp3 mutation, provided resistance higher than that of wild-type strain (Fig. 1a). While the AmB resistance conferred by CgPMP3 and CaPMP3 best hit (CaPMP3-B) was similar to that of ScPMP3, i.e., 4-fold higher than that of wild-type strain, the CaPMP3 ortholog (CaPMP3-O) provided 2-fold higher resistance (Fig. 1a).
To study the role of CaPMP3 ortholog and CaPMP3 best hit in C. albicans, we deleted both alleles of these genes in strain SN95 and confirmed by diagnostic PCR (Fig. S1). The C. glabrata ortholog CgPMP3 (CAGL0M08552g) was also deleted and confirmed by diagnostic PCR (Fig. S2). The AmB susceptibility of these delete strains with respect to their parent strains was compared (Fig. 1b). While deletion of PMP3 orthologs in C. glabrata and C. albicans sensitized the cells to AmB by about 4-fold, deletion of CaPMP3 best hit did not have any effect. The AmB sensitivity of ortholog deletants in both these species provides strong evidence that PMP3 gene is important for modulation of AmB resistance in pathogenic fungi as well.
AmB resistance mediated by Pmp3p is not dependent on ergosterol or Hsp90 or cell wall integrity. As far as the mechanistic basis of PMP3 effect on AmB resistance is concerned, Huang et al 13 showed that it is not related to its role in ion homeostasis. Absence or severe reduction in the amount of ergosterol in the fungal membranes and its replacement with certain other sterols results in AmB resistance in fungi 7,10,11 . To address this possibility total cellular content of ergosterol was estimated, as described 18 . The ergosterol content, as % wet weight of cells, of parent, delete and overexpression strains, was 0.021 6 0.001, 0.023 6 0.002 and 0.023 6 0.001, respectively. Though these values are comparable, it is possible that the intracellular  distribution of ergosterol might be affected. To check this, cells were stained with filipin, which is specific for sterols 19 , and observed (Fig.  S3a). While wild-type and PMP3 overexpression strains showed intense fluorescent spots within cells, pmp3D strain lacked such spots. Thus, it is possible that more ergosterol is distributed in the plasma membrane of the delete strain, rendering it more accessible for AmB binding and killing. If this is true, then the delete strain should be more sensitive to other polyenes which also act by binding to ergosterol. However, the sensitivity pmp3D strain to the polyenes nystatin, natamycin and filipin was found to be comparable to that of wild-type and PMP3 overexpression strains (Fig. S3b), ruling out ergosterol distribution or content having any role in modulation of AmB resistance by PMP3. Huang et al 13 have also ruled out the involvement of ergosterol in modulation of AmB resistance by PMP3, since this gene did not affect the resistance against other polyenes.
A recent report suggested that AmB resistance of ergosterol biosynthetic pathway mutants is highly dependent on Hsp90 chaperone and these mutants are hypersensitive to Hsp90 inhibitors radicicol and geldanamycin as well as oxidative stress 20 . To check the Hsp90 dependence of AmB resistance conferred by PMP3, the sensitivity of this strain to radicicol and oxidative stress was checked along with erg6D strain as positive control ( Table 1). The AmB resistance of erg6D strain and PMP3 overexpression strain was comparable. However, while erg6D strain was 8-fold and 4-fold, respectively, more sensitive to radicicol and oxidative stress, the sensitivity of PMP3 overexpression strain was comparable to wild-type, implying that Pmp3p is not dependant on Hsp90 for conferring AmB resistance. Cell wall alterations also can affect AmB resistance 7 . Compared to parent strain, PMP3 delete strain showed normal chitin deposition (Fig. S4a), as well as similar resistance to cell wall disrupting agents calcofluor white, sodium dodecyl sulphate and congo red (Fig. S4b), implying that AmB sensitivity of delete strain is not related to cell wall integrity.
Actin polarity and endocytosis, though impaired in pmp3D strain, are not responsible for its AmB sensitivity. To gain further insight into PMP3 mechanism of action, we tried to predict its possible functions on the basis of biological roles of genes that interact with PMP3. The list of interacting genes was analyzed using DAVID Bioinformatics Resources 21 for enrichment of gene ontology terms for biological processes. The top-two annotation clusters corresponded to endocytosis and actin cytoskeleton ( Table 2). To MIC for AmB and radicicol was determined in SC-ura broth at 30uC. Sensitivity to oxidative stress was determined by dilution spotting on SC-ura agar medium with tert-butyl hydroperoxide (TBH) at 37uC. MIC is the concentration at which no growth was observed. www.nature.com/scientificreports check if impaired endocytosis would result in AmB sensitivity, we screened mutants of several genes having role in endocytosis for their AmB sensitivity. Deletants of RVS161 and RVS167 were about 4-fold more sensitive to AmB compared to the parent strain (Fig. S5). These strains, besides defects in endocytosis have several other phenotypes including salt sensitivity and altered actin cytoskeleton [22][23][24][25] . SUR7, encoding an eisosome protein involved in endocytosis, partially suppresses several of these phenotypes upon multicopy overexpression [26][27][28] . Thus, we exploited overexpression of SUR7 to understand if AmB sensitivity of pmp3D strain is a consequence of defects in actin cytoskeleton or endocytosis, or it is an independent phenotype. A large scale survey using GFP-Snc1-Suc2 reporter has indicated that endocytosis is decreased in pmp3D strain 29 . We monitored rate of endocytosis with a different reporter, namely methionine permease (Mup1) tagged with ecliptic pHluorin, which is a pH-sensitive green fluorescent protein variant that does not fluoresce after internalization to an acidic compartment like vacuole 30,31 . Mup1-pHluorin is internalized rapidly upon exposure to methionine. Wild-type cells showed substantial decrease in Mup1-pHluorin intensity within 20 min after adding methionine (Fig. 2a). However, in pmp3D strain 40 min was needed for a similar decrease, confirming that the rate of endocytosis is slowed down in this strain. SUR7 expressed from a multicopy plasmid restored the rate of endocytosis of pmp3D strain to normal level (Fig. 2a). Mup1-pHluorin fluorescence was also monitored by flow cytometry (Fig. 2b). Though background fluorescence was high for all the strains, the rate of decrease in fluorescence is indicative of rate of endocytosis. While it was slow in the pmp3D strain, it was restored to wild-type level upon SUR7 overexpression.
Actin cytoskeleton plays a central role in endocytosis 25 and rvs161D and rvs167D strains impaired in endocytosis also have actin polarization defects 23 . Moreover, as PMP3 interacts with genes having role in actin cytoskeleton (Table 2), we visualized actin in PMP3 strains. The pmp3D strain showed pronounced defect in actin polarity, which is suppressed by overexpression of SUR7 (Fig. 3 and Fig.  S6). SUR7 also suppressed the sensitivity of pmp3D, rvs161D and rvs167D strains to NaCl (Fig. 4a). However, it could not reverse the sensitivity of these strains to AmB (Fig. 4b), demonstrating that AmB sensitivity of these mutants is not mediated by defects in actin polarity, endocytosis or NaCl tolerance.
Sphingolipid biosynthetic pathway is essential for PMP3 mediated increase in AmB resistance. We had recently shown that sphingolipid biosynthetic pathway genes FEN1 (ELO2) and SUR4 (ELO3) modulate AmB resistance 12 . While inhibition of sphingolipid biosynthesis with myriocin sensitized cells to AmB, addition of phytosphingosine, a sphingolipid pathway intermediate, reversed this phenotype 12 . To check the importance of this pathway for PMP3 mediated increase in AmB resistance, multicopy ScPMP3 was transformed into a few sphingolipid pathway mutants and the resistance was checked (Fig. 5a). In the wild-type parent strain (BY4741) ScPMP3 could increase AmB resistance at least by 4-fold. However, it increased AmB resistance by 2-fold or less in mutants of sphingolipid biosynthetic genes FEN1 and SUR4, and regulatory genes YPK1 32,33 and SAC1 34 . If PMP3 overexpression effect is independent of sphingolipid pathway, then fold-increase in AmB resistance by PMP3 in these mutants should have been comparable to that of the parent strain. Only 2-fold or less increase in resistance shows that PMP3 is dependent on this pathway for enhancing AmB resistance. Even this increase appears to be due to genetic redundancy. FEN1 and SUR4 are involved in fatty acid elongation and can partly compensate for each other's loss, since double deletion is lethal 35 . YPK1 and YPK2 are synthetic lethal 36 and arose from the whole genome duplication 37 . Sac1p is a phosphatidylinositol phosphate phosphatase, and its catalytic domain (Sac1-like domain) is seen among several phosphatases with partially overlapping function 38 . Sac1p is known to modulate sphingolipid metabolism 34,39 . Physical interaction of Figure 3 | Actin polarization defect of pmp3D strain is suppressed by multicopy SUR7 overexpression. Cells were grown to log phase and actin was visualized by rhodamine phalloidin staining. About 200 cells with small buds were scored according to their polarization state. Cells with actin patches concentrated in the small bud, with fewer than four patches in the mother cell, were classified as polarized cells. Other cells with more actin patches in the mother cell than in the small bud were classified as depolarized cells. Representative images are shown in Figure S6. Mean values of two independent experiments are given. The error bars indicate the range. Pmp3p and Sac1p has also been reported in a large-scale study 40 . Thus it appears likely that Pmp3p modulates sphingolipid biosynthesis and AmB resistance by interacting with Sac1p. Dependence of Pmp3p on Sac1p provides possible link between Pmp3p and sphingolipid pathway.
Myriocin inhibits the first committed step of sphingolipid biosynthesis catalyzed by serine palmitoyltransferase 33 . Sphingolipid pathway regulatory genes YPK1 32,33 and SAC1 34 modulate myriocin resistance. To test if PMP3 also regulates sphingolipid pathway, we checked myriocin resistance of deletion and overexpression strains. While deletion of PMP3 decreased myriocin resistance by 2-fold, its overexpression increased myriocin resistance by 4-fold, both with respect to parent strain (Fig. 5b), indicating that PMP3 is possibly involved in regulation of this pathway in S. cerevisiae. We also checked the myriocin sensitivity of C. glabrata strain deleted in PMP3 ortholog, and C. albicans strains deleted in PMP3 ortholog or best hit. However, the sensitivity of these strains was found to be comparable to that of their respective parent strains (Fig. S7). Another approach used to establish the role or dependence of genes on sphingolipid pathway is by supplementing with phytosphingosine (PHS), a sphingolipid pathway intermediate 33,41 . Addition of PHS increased the AmB resistance of pmp3D strain of S. cerevisiae to wild type level. It also decreased the AmB resistance of PMP3 overexpression strain to nearly wild type level (Fig. 6a), perhaps by its known antifungal activity at high concentration 42 . PHS also suppressed AmB sensitivity of C. glabrata and C. albicans strains deleted in PMP3 orthologs (Figs. 6b and 6c). These results further establish that PMP3 modulates AmB resistance through sphingolipid pathway in S. cerevisiae as well as in pathogenic Candida species.
Sphingolipid bases and complex sphingolipids have multiple roles in cells, both as structural components and as signalling molecules 43,44 .
Mutants of sphingolipid pathway show pleiotropic phenotypes 44 , of which those affected in actin cytoskeleton 45 , endocytosis 46 and AmB resistance 12 are pertinent here. Since actin is critical for endocytosis 25 , defective endocytosis could be a consequence of impaired actin polarity. Thus, impaired actin cytoskeleton and slow rate of endocytosis of pmp3D strain are consistent with the regulatory role played by PMP3 in sphingolipid pathway.
In conclusion, we have shown that a few striking phenotypes of PMP3 mutant, such as impaired actin polarity, endocytosis and salt tolerance are not related to its AmB-sensitivity. Rather, we show that modulation of AmB resistance by PMP3 is dependent on sphingolipid biosynthetic pathway, since AmB sensitivity of PMP3 deletants is suppressed by phytosphingosine, a sphingolipid pathway intermediate. Moreover, enhanced AmB resistance conferred by overexpression of PMP3 is dependent on functional sphingolipid biosynthetic and regulatory genes. Efforts are underway to elucidate the precise mechanism underlying PMP3 effect or dependence on sphingolipid pathway for modulating AmB resistance.

Methods
Fine chemicals and yeast synthetic drop-out medium supplements without uracil were procured from Sigma. All other media components were obtained from BD (Difco). Oligonucleotides were custom synthesised from Sigma-Genosys, India. Restriction enzymes, DNA polymerases and other DNA modifying enzymes were obtained from New England Biolabs, and DNA purification kits were obtained from Qiagen.
Strains, media and growth conditions. S. cerevisiae and Candida strains and plasmids used in this study are listed in Table S1 and S2. The Escherichia coli strain DH5a was used as a cloning host. YPD and Synthetic complete (SC) media were prepared and used as described 12 . Uracil supplement is omitted in SC medium to provide SC-ura medium. Yeast transformations were carried out using the modified Growth assays by dilution spotting. For dilution spotting assays, the strains/ transformants were grown overnight in SC or SC-ura medium, reinoculated in fresh medium to an A 600 of 0.1 and grown for 6 h. The exponential phase cells were harvested, washed and resuspended in sterile water to an A 600 of 1.0 (,2 3 10 7 cells/ ml). Ten-fold serial dilutions were made in water and 5 ml of each dilution was spotted on SC or SC-ura plates with desired concentration of compounds, as mentioned in Figures. DMSO alone was included in control plates, corresponding to its concentration in experimental plates, where appropriate. Plates were incubated for 2 days at 30uC before taking photographs. These experiments were repeated at least three times with comparable results.
Cloning methods. The ORFs of putative homologs of ScPMP3 in C. albicans [CaPMP3-ortholog (orf19.1655.3), CaPMP3-Best hit (orf19.2959.1)] and C. glabrata (CAGL0M08552g) were PCR amplified from the genomic DNA of C. albicans and C. glabrata with specific primers sets (Table S3). The PCR products were then used to replace the ScPMP3 ORF in a ScPMP3 clone in multicopy vector pFL44L, using Circular Polymerase Extension Cloning (CPEC) method 48,49 , thereby retaining the ScPMP3 promoter and terminator regions for all PMP3 orthologs as well. For cloning ScSUR7 gene, the SUR7 ORF of S. cerevisiae along with its promoter and terminator (1568 to 2326 bp) was amplified from strain BY4741 with forward primer ScSUR7-OCS1 and reverse primer ScSUR7-OCA1 (Table S3) and cloned in pFL44L by CPEC method 48,49 .
Construction of C. glabrata strain deleted in CgPMP3. PMP3 ortholog in C. glabrata (CAGL0M08552g) was deleted using a selection cassette conferring nourseothricin resistance containing CaNAT1 gene with codon usage adapted for Candida species 50 . A 508 bp region upstream of, and 472 bp region downstream of CgPMP3 ORF were PCR amplified from wild type genomic DNA using primers for upstream (CgPMP3-US1 and CgPMP3-UA1) and downstream regions (CgPMP3-DS1 and CgPMP3-DA1). The upstream flanking region was fused with the 59 region of CaNAT1 cassette using amplified upstream region and plasmid (pCR2.1-NAT 51 ) with CaNAT1 as templates and primers CgPMP3-US1 and CaNAT1-US-R1 to generate upstream split marker. Similarly, the downstream flanking region was fused to 39 region of CaNAT1 cassette with amplified downstream region and pCR2.1-NAT as templates and primers CaNAT1-DS-F1 and CgPMP3-DA1 to generate downstream split marker. These fusion products, which share 401 bp homology between them in the cassette, were mixed together, transformed 52 into C. glabrata wild type strain CG462, and plated on YPD plate. After incubation at 30uC for 24 h, cells were replica-plated onto YPD plate with 200 mg/ml nourseothricin and further incubated for 24 h. Nourseothricin resistant colonies were purified and checked for gene deletion by diagnostic PCR using cassette specific primers and primers outside the flanking region of homology (Table S3).
Fluorescence microscopy. Mup1-pHluorin internalization assay was performed as reported 31,53 . Mup1-pHluorin localization was visualized using a Nikon A1R confocal microscope using FITC optics and 100X oil immersion objective. Images were analysed using NIS Elements software. Visualization of actin by rhodamine phalloidin staining was carried out as described 54 . Calcofluor staining of cell wall was done as described 55 . The subcellular localization of sterols was monitored by staining with filipin as described 19 with slight modification. Exponentially growing cells (0.5 OD cells/ml) were fixed with 3.7% paraformaldehyde for 10 min at 30uC, washed with phosphate-buffered saline (PBS) and incubated with 5 mg/ml of filipin (Sigma F9765) in the dark at 30uC for 5 min. The stained cells were directly observed under a confocal laser scanning microscope (Nikon A1R) using 405 nm laser and images were analysed using NIS element software.
Flow cytometry. Log-phase cells were grown in SC medium without uracil and methionine for 6 hours, and then methionine was added to 20 mg/ml final concentration. At different time intervals cells were collected by centrifugation, washed and resuspended in PBS. Mup1-pHluorin fluorescence was measured with BD Accuri TM C6 flow cytometer in FL1 channel. Excitation and emission wavelengths were 488 nm and 530 nm, respectively. For each sample 10 4 cells were analysed. Three independent experiments were done with two replicates each time.