The Metacaspase (Mca1p) has a Dual Role in Farnesol-induced Apoptosis in Candida albicans

Manipulating the apoptotic response of Candida albicans may help in the control of this opportunistic pathogen. The metacaspase Mca1p has been described as a key protease for apoptosis in C. albicans but little is known about its cleavage specificity and substrates. We therefore initiated a series of studies to describe its function. We used a strain disrupted for the MCA1 gene (mca1Δ/Δ) and compared its proteome to that of a wild-type isogenic strain, in the presence and absence of a known inducer of apoptosis, the quorum-sensing molecule farnesol. Label-free and TMT labeling quantitative proteomic analyses showed that both mca1 disruption and farnesol treatment significantly affected the proteome of the cells. The combination of both conditions led to an unexpected biological response: the strong overexpression of proteins implicated in the general stress. We studied sites cleaved by Mca1p using native peptidomic techniques, and a bottom-up approach involving GluC endoprotease: there appeared to be a “K/R” substrate specificity in P1 and a “D/E” specificity in P2. We also found 77 potential substrates of Mca1p, 13 of which validated using the most stringent filters, implicated in protein folding, protein aggregate resolubilization, glycolysis, and a number of mitochondrial functions. An immunoblot assay confirmed the cleavage of Ssb1p, a member of the HSP70 family of heat-shock proteins, in conditions where the metacaspase is activated. These various results indicate that Mca1p is involved in a limited and specific proteolysis program triggered by apoptosis. One of the main functions of Mca1p appears to be the degradation of several major heat-shock proteins, thereby contributing to weakening cellular defenses and amplifying the cell death process. Finally, Mca1p appears to contribute significantly to the control of mitochondria biogenesis and degradation. Consequently, Mca1p may be a link between the extrinsic and the intrinsic programmed cell death pathways in C. albicans.

Apoptosis, or programmed cell death (PCD) 1 is an important physiological process involved in key steps of the development of multicellular organisms. It also regulates the fate of cell populations in adverse conditions. Several pathways have been described involving internal triggering factors, often associated with critical alterations of mitochondrial structures and functions, or external signals received by specific cell death receptors. Apoptosis involves a number of cellular processes including active mechanisms of chromatin condensation, DNA fragmentation, and phosphatidylserine externalization to the outer leaflet of the plasma membrane. Mitochondria appear to be major actors in the intrinsic apoptosis pathway: the alteration of mitochondrial membrane permeability, a Bax-Bcl2-dependent process in mammalian cells, induces overproduction of reactive oxygen species (ROS) and the release of cytochrome c, which in turn contribute to the induction of endoproteolytic cascades leading to cell death (1,2). Caspases are cysteine-proteases acting as key components of apoptosis machineries. In mammalian cells, proforms of the effector caspases 9 and 3 are cleaved and activated, and in turn activate many other cellular caspases (3)(4)(5). In mammals, these proteases cleave preferentially after aspartic acid residues (6,7). Unicellular eukaryotes, including yeasts, also display apoptosis or PCD (8,9). The yeast Candida albicans, which is characterized by a yeast-hyphae dimorphism, is an opportunistic human pathogen responsible for benign candidiasis to lifethreatening systemic infections. PCD was shown to be involved in the regulation of collective cellular interactions during biofilm formation by C. albicans where cell density is high, and competition for nutrients may occur (10). Under such adverse conditions, C. albicans secretes metabolites including farnesol or tyrosol that act as quorum sensing compounds (11,12). Farnesol has been shown to inhibit biofilm formation (13) and induce ROS production and apoptosis (10,14) through the activation of a number of signaling pathways, including the Ras c-AMP pathway (15), and Czf1p and Efg1p coordinated activities (16). However, the effects of farnesol may vary according to the growth conditions and physiology of the cells (16), and this has led to some apparently contradictory reports in the literature. For example, in C. albicans yeast cells, proteomic studies based on 2D-gel electrophoresis indicate that farnesol increases the expression of heat shock proteins including Cta1p, Hsp90p, and Hsp70p (10). However, in C. albicans biofilm, transcriptomic analysis identified down-regulation of the corresponding genes (11). Hsp90p is also involved in C. albicans apoptosis by regulating the calcineurin-caspase apoptotic pathway (17). Recent work shows that the effects of farnesol may be mediated by the ABC multidrug transporter Cdr1p-dependent glutathione efflux from cells (18), promoting the conditions for a severe oxidative stress, a triggering signal for apoptosis.
A key player in apoptosis in C. albicans is the cysteineprotease Mca1p, a member of the family of metacaspases, a group of specific caspases present in plants, fungi, and protozoans where they have also been implicated in apoptosis (19 -21). In C. albicans, the MCA1 gene is the only gene encoding a metacaspase and its activation during apoptosis events has been demonstrated (10). Farnesol induces expression of the MCA1 gene and leads to the activation of the metacaspase activity at the protein level (10). C. albicans lacking Mca1p displays resistance to oxidative stress-induced death and changed energy metabolism (22). Mca1p share 55% sequence identity and 69% overall homology with the metacaspase Yca1p from Saccharomyces cerevisiae. Besides its role in the H 2 O 2 -induced apoptosis, Yca1p has been implicated in other functions: the regulation of the composition of the insoluble proteome with or without H 2 O 2 stress (23); clearance of insoluble protein aggregates (24); and cell cycle progression and particularly in G2/M transition (25). Only one substrate of the S. cerevisiae Yca1p has been clearly demonstrated, the glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (26). However, little is known about Mca1p in C. albicans, and in particular about its specificity or direct targets. This is an important issue, because although compounds such as Plagiochin E (27) induce C. albicans apoptosis through a caspase-dependent process, others induce apoptosis through metacaspase-independent pathways (28). Indeed, the precise role of Mca1p in the induction of apoptosis and in mitochondrial functions remains unclear. Therefore, we studied the function of Mca1p to decipher in particular the relative contribution of Mca1p and farnesol in the observed variations of the C. albicans proteome. We used a strain deleted for the MCA1 gene (mca1⌬/⌬) and compared its proteome to that of a wild-type strain, in the absence and in the presence of the quorum sensing molecule farnesol. We combined classical bottom-up and label-free proteomic techniques using trypsin to identify differences in protein levels. We also developed top-down and bottom-up approaches based on GluC digestion to identify and quantify potential metacaspase targets. This analytical pipeline allowed us to provide new insights into Mca1p function and its involvement in apoptosis in C. albicans.

EXPERIMENTAL PROCEDURES
Strains and Growth Conditions-C. albicans strains BWP17 (29) and BWP17-mca1⌬/⌬ (mca1⌬::FRT/mca1⌬::FRT) (28) were grown in yeast-extract peptone dextrose (YPD) medium (1% [w/v] yeast extract, 2% [w/v] peptone, and 2% [w/v] dextrose). Cell cultures (100 ml) were inoculated at an initial OD 600 of 0.1 with liquid precultures (10 ml medium inoculated with freshly grown colonies on YPD agar plates) and incubated in an orbital shaker at 30°C. Farnesol (250 M final concentration), or solvent, was added to the cultures in mid-exponential phase of growth, and the cells collected by centrifugation 4h later. C. albicans grew as budding-yeast under these conditions. Stock solutions (250 mM) of farnesol (mixed isomers; Sigma Chemical Co, 5 M solution.) were prepared in methanol just before use. Cell-free extracts were prepared by disrupting the cells (60% wet weight/vol in 0.1 M potassium phosphate buffer, pH 7.2) using a One-Shot apparatus (Constant Systems Ltd, Daventry, UK) operating at 2.5 Kbar. The cell lysate was diluted twice in the same buffer, and large cell debris was removed by centrifugation for 10 min at 5000 ϫ g. The supernatant, referred to hereafter as the cell homogenate, was rapidly frozen in 5 mg proteins/ml aliquot fractions and kept at Ϫ80°C.
Validation of the Apoptosis Condition by Flow Cytometry-Cells from the four growth conditions (WT, WT-farnesol, mca1⌬/⌬, and mca1⌬/⌬-farnesol) were collected by centrifugation and washed twice in 1ϫ PBS. Flow cytometry (BD Accury, 10,000 cell counts, medium flow rate) was used in the following ways to evaluate apoptosis: (1) Aliquots were labeled with propidium iodide (PI) at a concentration of 10 g/ml incubated in the dark for 1h, washed by centrifugation, and counted. (2) Two aliquots were processed to prepare protoplasts. The cells were first preincubated for 10 min at 30°C in a 0.1 M Tris-Cl buffer containing 0.5 M 2-mercaptoethanol. The cells were washed in 0.01 M Tris-Cl, pH 7, 0.5 M KCl, collected by centrifugation, and resuspended in the digestion buffer (0.01 M citrate-phosphate buffer, pH 5.8, 1.35 M sorbitol, and 1 mM EGTA). Zymolyase 100T (MP Biomedicals, Illkirch-Graffenstaden, France) was added to a final concentration of 10 mg/ml. The samples were incubated at 30°C and protoplast formation evaluated by measuring the OD 600 of 1/100 dilutions of the cells in water. After nearly complete protoplast formation (Ͼ85%) the cells were collected by centrifugation and washed in a 0.01 M Tris-maleate buffer, pH 6.8, 0.75 M sorbitol, 0.4 M mannitol, and 2 mM EGTA. An aliquot of the protoplasts was incubated with 5 l of the annexin V-FITC reagent (BD Pharmingen, San Diego, CA) for 1h in the dark, washed, and then diluted in the same buffer for flow cytometry analysis. Another aliquot was processed for TUNEL labeling: the protoplasts were first fixed with a freshly prepared 4% formaldehyde solution in PBS, permeabilized using 0.1% Triton X-100 in 0.1% (w/v) sodium citrate solution, and further processed with enzyme and reagents from the Roche Cell-Death detection kit according to the manufacturer's instructions (3). Changes in total reduced glutathione concentration were evaluated by a modification of the method of Lewicki et al. (30), using naphthalene 2,3 dicarboxaldehyde (NDA) as a glutathione-specific fluorogenic reagent. Intact cells were washed in PBS, incubated with NDA (1 mM) for 30 min in the dark, and analyzed by flow cytometry as described above.
SDS-PAGE-Samples of 20 g of protein prepared from cells in the various conditions investigated were run on a NUPAGE 4 -12% acrylamide gel (Invitrogen) and stained in Coomassie blue (Simplyblue Safestain; Invitrogen, Carlsbad, CA).
Western blot Analyses-20 g protein samples of prepared from cells in the various conditions investigated were separated on a NUPAGE 4 -12% acrylamide gel (Invitrogen) and transferred to nitrocellulose membrane (Protran BA85, GE Healthcare Life Sciences). Ssb1p protein expression was assessed by immunoblot with a primary antibody (rabbit polyclonal IgG fraction) raised against the purified protein from S. cerevisiae anti-Ssb1p (kind gift from Dr Vé ronique Albanese, Institut Jacques Monod, Paris). We used an antirabbit Horseradish peroxidase conjugated antibody (Sigma Aldrich) and an enhanced chemiluminescent reagent (West Pico, GE Healthcare) for detection.
Strategy for the Analysis of Proteome Variations as a Function of the Experimental Conditions-The general strategy for the production of the different data sets and their exploration is summarized in Fig. 2. The individual steps of this pipeline are described in the following sections.
LC-MS/MS Acquisition-All digests or protein extracts were analyzed using a LTQ Velos Orbitrap equipped with an EASY-Spray nanoelectrospray ion source and coupled to an Easy nano-LC Proxeon 1000 system (all devices are from Thermo Fisher Scientific, San Jose, CA). Chromatographic separation of peptides was performed with the following parameters: Acclaim PepMap100 C18 precolumn (2 cm, 75 m i.d., 3 m, 100 Å), Pepmap-RSLC Proxeon C18 column (50 cm, 75 m i.d., 2 m, 100 Å), 300 nl/min flow, gradient rising from 95% solvent A (water, 0.1% formic acid) to 35% solvent B (100% acetonitrile, 0.1% formic acid) in 217 min followed by column regeneration for 23 min giving a total run time of 4 h. Peptides were analyzed in the orbitrap in full ion scan mode at a resolution of 30,000 (at m/z 400) and with a mass range of m/z 400 -1800. Fragments were obtained with a collision-induced dissociation (CID) activation with a collisional energy of 40%, an isolation width of 2 Da, and an activation Q of 0.250 for 10 ms. MS/MS data were acquired in the linear ion trap in a data-dependent mode in which the 20 most intense precursor ions were fragmented, with a dynamic exclusion of 20 s, an exclusion list size of 500 and a repeat duration of 30 s. The maximum ion accumulation times were set to 100 ms for MS acquisition and 50 ms for MS/MS acquisition. All MS/MS data were processed with an in-house Mascot search server (Matrix Science, Boston, MA; version 2.4.1). The mass tolerance was set to 7 ppm for precursor ions and 0.5 Da for fragments. The following modifications were used in variable modifications: oxidation (M), phosphorylation (STY), acetylation (K, N-term), and deamidation (N, Q). Glutathionylation (C) was added for additional analyses of trypsin digests. The maximum number of missed cleavages by trypsin was limited to two for all proteases used. MS/MS data were searched against a C. albicans ORF database (# 6209) retrieved from the Candida Genome Database website (31) (http://www.candidagenome.org/).
Quantitative Analysis in Label-free Experiments-Proteins in the various homogenates were digested in solution, in duplicate, overnight at 37°C by sequencing grade trypsin (12.5 g/ml; Promega, Madison, WI) in 20 l of 25 mmol/L NH 4 HCO 3 . LC-MS/MS acquisition was performed with a 4-hour gradient. MS/MS data were processed with an in-house Mascot search server (Matrix Science, Boston, MA; version 2.4.1). Label-free quantification in between subject analysis was performed on raw data with Progenesis-LC software 4.1 (Nonlinear Dynamics Ltd, Newcastle, U.K.) using the following procedure: (1) chromatograms alignment, (2) peptide abundances normalization, (3) statistical analyses of features, and (4) peptides identification using the Mascot server. A decoy search was performed and the significance threshold was fixed to 0.05. The resulting files were imported into Progenesis-LC software. Peptides with an ion score less than 15 were rejected. Conflicts for the identification of some peptides were resolved manually. Proteins with similar normalized abundance variations (ANOVA p-value lower than 0.05) were classified by the Auto-Class Bayesian clustering system (webserver Autoclass@IJM, http:// ytat2.ijm.univ-paris-diderot.fr/) (32). Farnesol and mutation effects were investigated with a two-factor analysis. p-values associated were calculated with a dedicated R script. Gene ontology analyses were performed on all clusters obtained with the GO Slim mapper from the Candida Genome Database website. Functional enrichments for each GO slim term were calculated, relative to the whole identified proteome, using the hypergeometric distribution.
Quantitative Analysis by TMT Labeling Experiments-Aliquots of 100 g of protein from each cell homogenate preparation were reduced with 10 mM Tris(2-carboxyethyl)phosphine hydrochloride (TCEP), alkylated with 18 mM iodoacetamide according to the TMT sixplex kit protocol (Thermo Fisher scientific, San Jose, CA) then digested in solution overnight at 37°C by sequencing grade trypsin (12.5 g/ml; Promega) in 20 l of 25 mmol/L triethyl ammonium bicarbonate (TEAB). Samples from the WT, WT-farnesol, mca1⌬/⌬, and mca1⌬/⌬-farnesol conditions were labeled with 126 (monoisotopic m/z ϭ 126.12773), 127 (m/z ϭ 127.13108 Da), 128 (128.13444 Da), and 129 (m/z ϭ 129.13779 Da) labels, respectively. The samples were mixed, desalted, and concentrated with a 50 mg C18 SPE column (Biotage, Uppsala, Sweden) and the peptides precipitated with acetone. The resulting peptide mixtures were analyzed in triplicate by LC-MS/MS with a 4-hour gradient. MS/MS were acquired with HCD fragmentation and the orbitrap analyzer set at a resolution of 7500 at m/z 400 in a top20 protocol. The maximum ion accumulation time was set to 100 ms for MS acquisition and 50 ms for MS/MS acquisition. Data were processed with Proteome Discoverer 1.4 software (Thermo Fisher scientific, San Jose, CA) and the Mascot search engine with the trypsin digest parameters described above and the additional modifications carbamidomethylation (C) and TMTsixplex (K and N-term). The precursor mass tolerance threshold was set at 0.7 ppm and 0.02 Da for fragment tolerance. The percolator node was used and a 5% FDR was applied. Quantification results in triplicate were treated as replicates and were visualized with Proteome Discoverer 1.4 to allow comparisons with the results obtained by label-free methods.
Peptide Analysis by Native Peptidomic Techniques-Aliquots of 500 g of protein from cell homogenates from each condition were resuspended (in duplicate) in 1 ml of 50 mM NH 4 HCO 3 buffer. Sizeexclusion chromatography was performed with a 10 kDa cut-off Amicon membrane (Millipore). Peptides and proteins under 10 kDa were isolated and concentrated with a Speedvac and resuspended in 0.1% formic acid. Peptides were isolated and concentrated with five zip-tip C18 tips for each condition. Peptides were analyzed by LC-MS/MS with a 4-hour gradient. MS/MS data were processed with Proteome Discoverer 1.4 software (Thermo Fisher Scientific) and the Mascot search engine with a "no enzyme" search parameter. The percolator node was used for FDR evaluation with a 5% filter in relaxed FDR and 1% in strict FDR. Peptides were flagged as identified when present in at least one condition with a Q-value under 0.05. The Venny web resource (http://bioinfogp.cnb.csic.es/tools/venny/) was used to produce graphical outputs and visualize specific peptides from different conditions. overnight at 37°C by the GluC protease (12.5 g/ml; Sigma Aldrich) in solution in 20 l of 20 mM sodium phosphate buffer, pH7.8. The resulting peptides were analyzed by LC-MS/MS with a 4-hour gradient. MS/MS data were processed with Proteome Discoverer 1.4 software (Thermo Fisher scientific, San Jose, CA) and the Mascot search engine with semi-GluC search parameter. The percolator node was used for false discovery rate (FDR) evaluation with a 5% filter in relaxed FDR and 1% in strict FDR. The Venny web resource was used to visualize peptides specific to the different conditions.
Evaluation of Mca1p Cleavage Specificity-Peptides identified by these analyses (native peptidomics and bottom-up digestions), at all FDR, were considered and only those exclusively present in the WT-farnesol condition with a Q-value under 0.05 were selected for further analysis. The "Extract Cleavage Site Context" algorithm in the Galaxy-P web resource (https://usegalaxyp.org/) was used to generate a list of sequences surrounding the cleavage sites of the peptides. The parameters used for this algorithm were: five amino acids at the N-terminal and C-terminal for the cleavage site context and a protease cleavage C-terminal to the cleavage site. All peptides with cleavage of initiation methionine only were eliminated. Cleavage specificities were represented by the Weblogo (http://weblogo.berkeley.edu/) (33) and Two Sample Logo algorithm with binomial test (http://www. twosamplelogo.org) (34). Extended cleavage specificities were investigated with IceLogo (https://code.google.com/p/icelogo/) (35) using the natural abundance of amino acids in the C. albicans protein sequence database (# 6209). For substrate validation, label-free approaches with Progenesis-LC software 4.1 (Nonlinear Dynamics Ltd, Newcastle, U.K.) of GluC and native peptidomics data sets were performed in a between-subject way. Potential substrates of Mca1p found on an identification basis were validated by considering those over-represented in WT-farnesol condition relative to the other conditions. Anova p-values under 0.05 calculated after label-free analysis of native peptidomics and GluC data sets were considered to select best potential substrates. A two factor analysis was performed to discriminate farnesol and/or mutation dependent or independent effects. GO analyzes were performed on the obtained list of potential substrates with the GO Slim Mapper tool of the Candida Genome Database website. P-values for functional enrichments were calculated with R 3.0.1 software for each GO term.
Data Availability-The complete sets of data are available as .raw files, Proteome Discoverer 1.4 .msf files and the Candida albicans protein sequence database provided as a multiple fasta in the Peptide Atlas repository under the identification number: PASS00561. The msf files can be visualized using a freely available java application Thermo-msf-parser 2.0.5 available at https://code.google.com/p/ thermo-msf-parser/.

Mca1p is Required for Farnesol-induced Apoptosis in C.
albicans-Farnesol, at the concentration and time of action chosen, had only a limited effect on the growth of the mca1⌬/⌬ strain, although it slowed the growth of the wild-type strain. Annexin V-FITC and TUNEL labeling experiments showed that under these conditions, farnesol induced apoptosis in WT cells, but not in mca1⌬/⌬ cells. Little or no specific fluorescence was detected in farnesol-treated mca1⌬/⌬ cells, whereas more than 70% of the farnesoltreated WT cells were labeled (Fig. 1). No PI labeling was found in any of the four conditions indicating that the cells did not become necrotic. Consistent with this observation, the electrophoretic profile in SDS-PAGE experiments of total protein samples from the four conditions were very similar (supplemental Fig. S1). By comparing results between samples following these four experimental conditions, we investigated the role of the metacaspase Mca1p in the early phase of the apoptotic response in C. albicans.
Production of the Various Data Sets- Table I summarizes the results of the various analyses for protein (or peptide) content: trypsin digests, TMT-labeled samples, and GluC digests, all acquired in triplicates for each condition and the native peptidomic analyses run in duplicate for top-down evaluation of endogenous proteolysis in controls and farnesol-treated samples. The complete data sets are available in supplemental Data Files S1 to S5.
Evaluation of the Mca1p Cleavage Specificity-We used the Proteome Discoverer 1.4 software to determine changes to the peptide profile and abundance after GluC digestion, native peptidomic analyses, and trypsin digestion ( Fig. 3; see supplemental Files S1, S2, and S4 for the complete data sets). The peptides found are represented in a Venn diagram according to the conditions (Fig. 3A). More peptides could be identified after Tryptic and GluC digests than native peptidomic analysis. Comparisons of protein profiles by SDS-PAGE indicated that farnesol and/or mca1 disruption did not lead to any massive degradation of proteins (supplemental Fig. S1). We compared the cleavage site context for peptides found by native peptidomic analyses only in the WT-farnesol condition with those found only in the mca1⌬/⌬-farnesol condition. We examined the C-terminal residues of all the peptides identified in the top-down analyses, and we searched for partial cleavage specificity (semi-trypsin and semi-gluC) in the tryptic digests, where only K or R are expected except for the C-terminal fragment of the proteins, and in the GluC digests where an E, and to a lesser extent, a D are expected, except for the C-terminal fragment of the proteins. We expected to find enrichment in peptides with a D at the C-terminal end in the showing the overlaps in differential peptide abundance in tryptic and GluC digests and native peptidomic analyses. B, Logo diagrams computed using the N-ter and C-ter extremities of peptides found only in WT-Farnesol and mca1⌬/⌬ -farnesol conditions from the whole data sets produced in this study. Cleavage is considered at position five for all logos. Weblogo diagrams were used to visualize the general sequence context surrounding the cleavage site for all peptides (insets diagrams for GluC and trypsin experiments and diagrams for native peptidomic) and for peptides without D/E cleavage site (diagrams for GluC experiment) and without K/R (diagrams for trypsin experiment). The Two Sample Logo java applet was used to evaluate the cleavage specificity. Peptides found only in WT-farnesol are considered as the positive data set, whereas those found in mca1⌬/⌬ -farnesol condition as the negative data set. A binomial test was used. IceLogo tool was used to determine extended cleavage specificities taking into account the natural abundance of amino-acids present in the C. albicans database used for all data sets. tryptic digest if Mca1p has the specificity of a caspase, or a K or R in GluC digests, if it does not. In wild-type cells where Mca1p is activated, we found a substantial enrichment in K and R cleavage sites in the semi-GluC database searches (Table IIIB). The comparison between peptides found in both mca1⌬/⌬ and WT conditions with those observed in both farnesol conditions did not reveal any cleavage specificity for the trypsin digest other than the expected K/R residue in P1; this was consistent with the results from native peptidomic and GluC digest analyses. Calculated K/R enrichment of peptide ends for the different conditions confirmed the graphical visualization (Table II). Consideration of only those peptides with K/R ends is informative about the specificity of the protease Mca1p (Fig. 3B). Native peptidomic and GluC digest analyses showed that an acidic amino acid residue in P2 may be important for the cleavage specificity of Mca1p.
Native Peptidomic and GluC Digest Experiments to Identify Potential Mca1p Targets-GluC-directed bottom-up experiments identified 68 peptides (from 54 proteins) among a total of 8549 identified peptides (from 1261 proteins) as potential substrates of Mca1p. Native peptidomic analyses identified nine peptides (from eight proteins) among 416 (from 137 proteins) as potential substrates (supplemental Table S1). Substrate candidates were validated by a label-free approach on GluC digests and native peptidomics experiments (entire data sets in supplemental Files S6 and S7 respectively) with Progenesis-LC software. Peptides candidates (77 candidates) quantified (28 candidates) with higher abundances in WT- farnesol (23 candidates) than in the three other conditions and with an Anova p-value under 0.05 (13 candidates) were considered as best potential substrates. A p-value based on a two-way ANOVA was calculated to classify them and 12 reported an effect dependent of both strain and farnesol factors (supplemental Table S1). Table III provides examples of potential Mca1p substrates as found by native peptidomic or GluC digestion approaches, only in the WT-farnesol condition and with K or R ends. Quantitative Assessment of Proteome Differences Between WT and mca1⌬/⌬ in Response to Farnesol-We performed a label-free (Progenesis-LC) analysis of tryptic digests (four conditions, two replicates each) to investigate protein compositions: 2727 proteins were quantified with the parameter "trypsin for enzyme" and 503 proteins were thereby identified as differing between the conditions (differences of normalized abundance were significant with an ANOVA p-value lower than 0.05). The complete .htm files are supplied as supplemental File S5. We classified these differentially expressed proteins using the AutoClass@IJM Bayesian clustering system (32) (Fig. 4A). All clusters and accessions are presented in supplemental data (supplemental Table S2). The largest differences of protein abundance were between farnesol-treated and untreated cells, whether or not they carried the MCA1 deletion. Most of the proteins that were more abundant in the WT-farnesol condition were also abundant in mca1⌬/⌬-farnesol extracts. We also ran independent analyses (in triplicate) of TMT-labeled digests. Quantifications of the reporter ions, and calculation of the abundance ratios confirmed the differences in protein abundance found in by label-free experiments. The TMT experiments allowed the identification of 1601 proteins with a 5% FDR threshold (see supplemental File FIG. 4. Classification of the proteins differentially expressed in the four experimental conditions. A, Proteins for which the normalized abundance differed (ANOVA p value lower than 0.05) were sorted by the AutoClass Bayesian clustering system (webserver AutoClass@IJM, http://ytat2.ijm.univ-paris-diderot.fr/). All clusters and accessions are described in supplemental Fig. S2. B, Proteins candidate to be substrates of Mca1p found in native peptidomic and GluC digest experiments are distributed according to (B-1) their expression levels (black sector: no significant expression variation; green sector: significant variation in abundance) in label-free analyses of tryptic digests and to (B-2) their belonging to specific abundance profile clusters. C, Normalized abundance level of protein candidates to be substrates of Mca1p are represented in boxplot generated with BoxPlotR tool (http://boxplot.tyerslab.com/). S3 for the identified proteins). There was a very good correlation between the two quantification methods (supplemental Fig. S2). Scatter graphs of the ratios of protein abundances (WT-F/WT and mca1⌬/⌬-F/WT) indicated that proteins with the largest differences in abundance were common to the lists identified by both TMT labeling (supplemental Fig. S2A) and label-free (supplemental Fig. S2B) quantification analyses. Also, trend curves of the scatter graphs revealed that overall protein level in the farnesol condition was higher in mca1disrupted than WT cells. Mca1p-specific peptides corresponding to the part of the gene deleted from the mca1⌬/⌬ strain were found only in the wild-type extracts, and the abundance of Mca1p in the WT increased under farnesol treatment. However, Mca1p was not analyzed further because it was above the threshold in ANOVA (p-value of 0.06) and was not identified in TMT-labeling experiments. Labelfree analyzes of tryptic digests indicated that 33% of proteins found to substrates of Mca1p were differentially expressed (80% being overexpressed) between mca1⌬/⌬-farnesol condition and the WT-farnesol condition ( Fig 4B); this is significantly higher than the 18% of all proteins identified as being differentially expressed in label-free of tryptic digests. This is confirmed by a boxplot representation of potential substrates normalized abundances found in label-free of tryptic digests (Fig. 4C). Thus, not only were potential substrates degraded by Mca1p, but their production appears to be directly or indirectly inhibited by the protease.
Functional Analysis of Protein Variations Induced by Farnesol Treatment and/or mca1 Deletion-Representative examples of proteins the expression of which was affected by farnesol and/or the deletion of MCA1 are presented in Table  IV. The differences in abundance were measured by a labelfree approach with trypsin digests and confirmed by a TMTlabeling approach ( Table V). Some of the potential substrates for Mca1p, including chaperone proteins Hsp70p, Hsp90p, and Ssb1p were found differentially expressed in label-free and TMT-labeling analyses of tryptic digests. Gene ontology analyses of the potential substrates of Mca1p confirmed a significant enrichment in proteins implicated in the stress response and folding (supplemental Table S3). Several targets, such as Hsp90p, Rps3p, and Cdc48p, were predicted to interact with Mca1p in C. albicans (excepted for the last protein: just in S. cerevisiae) by the String 9.0 web resource (Fig. 5). Gene ontology analyses also revealed that many of the potential substrates found by GluC experiments are mitochondrial proteins (e.g. Qcr2p, Rim1p, Pil1p, Tpi1p, Rct1p, and Atp5p) or associated with the plasma membrane or the cell wall, or are extracellular (supplemental Table S3). Some potential substrates are involved in carbohydrate metabolism and in the generation of precursor metabolites and energy, functions associated with mitochondria and glycolysis (Tpi1p, Fba1p, and Gpm1p). Farnesol increased the abundance of heat shock proteins (Hsp70p, Hsp90p, Hsp104p, Ssc1p, and Kar2p for example) in the WT, with the effect being much larger in the mca1deleted strain (Tables IV and V). The level of a putative mitochondrial cell death effector (orf19.2175), an ortholog of Aif1p in S. cerevisiae, was increased by farnesol in the WT and to a lesser extent in the mutant (Table IV). Ras1p, a protein involved in apoptosis through the RAS1/cAMP/PKA pathway in C. albicans (36), did not differ between any of the conditions analyzed. The abundance of Tpk2p, a catalytic subunit of PKA, was higher in WT-farnesol cells than in mca1⌬/⌬ cells in the presence of farnesol (supplemental File S5).
We performed a gene ontology analysis on the various classes of proteins with similar abundance profiles (Autoclass clusters) that allowed us to evaluate the effect of farnesol and Mca1p on the "cellular component," the "biological process," and the "molecular function" terms. A Mca1p-dependent effect of farnesol was clearly demonstrated for a number of proteins involved in carbohydrate metabolism. However, the abundance of numerous proteins was not dependent on the presence of the Mca1p protease. The level of proteins involved in the response to oxidative stress increased with farnesol and was higher in the absence of Mca1p in the farnesol condition. The abundance of proteins implicated in processes such as "translation," "ribosome biogenesis," and "protein folding" in the presence of farnesol was still higher in the mca1⌬/⌬ than WT strain (supplemental Table S4). The same was found by analysis of the "structural activity" and "RNA binding" terms of "molecular function" entries (supplemental Table S4). Differences in the control of translation between farnesol-treated mca1⌬/⌬ and farnesol-treated WT cells were confirmed by the overexpression of proteins in the nucleolus observed in extracts. Expression of some Mca1p partners predicted by the String 9.0 database web resource (http://string-db.org/) was increased by farnesol and was higher in the Mca1p-deleted than WT strains (Fig. 5).

Farnesol Induces Cdr1p Overexpression and its Glutathionylation in a Mca1p
-Independent Way-Label-free analysis of tryptic digests and TMT-labeling experiments showed that farnesol increased the abundance of Cdr1p, an ABC multidrug transporter, in a Mca1p-independent way (Fig. 6A). Cdr1p has been described as exporting glutathione from farnesol-treated cells, so we searched for signatures of this activity in our data. A database search of tryptic digests for the post-translational modification glutathionylation (C) revealed no overall difference (Q-value threshold at 0.05) in the glutathionylated proteomes in the different conditions (Fig.  6C). We found that Cdr1p itself was glutathionylated on cysteine 712 (Fig. 6B) in the WT and the mca1⌬/⌬ extracts prepared after farnesol treatment. No differences in reduced glutathione between the different conditions was found by flow cytometry and naphthalene-2,3-dicarboxaldehyde labeling (Fig. 6D).
Western blot Analyses of a Putative Substrate of Mca1p-Immunoblot with anti-Ssb1p antibody confirmed abundances variation found for this protein in label-free experiment of tryptic digests (Fig. 7). We showed a higher Ssb1p abundance in mca1⌬/⌬-farnesol condition relative to other conditions. Moreover, a fragment of this protein was detected (around 20 kDa) only in WT-farnesol condition indicating that the Ssb1p cleavage is caused by the activated Mca1p protease as shown in GluC digest bottom-up approach (Table III). DISCUSSION When analyzing the C. albicans metacaspase function, we wondered whether it is possible to address its endopeptidase activity from in vivo experiments. Therefore, we conducted a series of proteomic studies to: (1) evaluate the role of Mca1p in the early phase of apoptosis in C. albicans, (2) investigate the proteolytic activity of Mca1p and determine whether it is a general purpose protease, or an enzyme with specific substrates, and (3) decipher the cleavage specificity of Mca1p. We analyzed, in parallel, the proteomes of two C. albicans strains, a wild-type strain and an isogenic strain from which both alleles of the MCA1 gene were deleted, grown under non-filamentous conditions (yeast form) in the presence and absence of a low concentration of the apoptosis inducer farnesol. This allowed analysis of the events during an early phase of apoptosis, before the massive degradation events lea ding to necrosis. The action of farnesol action is dependent on Mca1p expression (10). Under the experimental conditions used (250 M farnesol applied for 4h to yeast cells in the mid log phase of growth), there was a clear difference in biological responses between the WT and the mca1⌬/⌬ strains. In the WT cells but not in the mca1⌬/⌬ cells farnesol induced the externalization of phosphatidyl serine and some fragmentation of nuclear DNA indicating apoptosis. However, none of the cells were significantly labeled with propidium iodide, indicating that they were not necrotic. These observations confirmed that the presence of an active metacaspase is required for C. albicans programmed cell death. Most studied caspases have a relatively strict specificity for an aspartic acid residue (D) in P1 (6, 7) whereas several metacaspases from plants, yeasts or some metazoans tend to be specific for basic residues (K/R) (7,37,38). The clear enrichment in K/R cleavage sites we found for Mca1p indicates a cleavage specificity similar to that described for several fungal metacaspases (as determined using synthetic peptides) including the Allomyces arbuscula metacaspase CDPII (37) and Yca1p in S. cerevisiae (25,38); this is also the specificity displayed by plant metacaspases 4 and 9 in Arabidopsis thaliana (7), or mcII-PA in the Norway spruce tree (39). There were only a small number of peptides fulfilling our selection criteria in the various conditions and no cleavage specificity other than a K/R residue was found by comparison of WT-farnesol and mca1⌬/⌬-farnesol using native peptidomic analyses, or gluC and trypsin digests. This indicates that the apoptosis events analyzed did not result from a massive and nonselective degradation of cellular protein. Mca1p appears to be specific to a limited number of substrates, and does not seem significantly to activate other proteases. Moreover, our results showed an interesting bias with over-representation of an acidic residue (D or E) in the P2 position relative to the cleavage site. An acidic D/E predominance at P1Ј and to a lower extent at P2 has been described for the metacaspase 9 in A. thaliana (40) and a basic residue (K or R) at P2 inhibits totally the activity of the CDPII metacaspase of Allomyces arbuscula (37). Overall, we report 54 proteins (77 peptides) as potential substrates of Mca1p in vivo during the early phase of farnesol-induced apoptosis in the yeast form of C. albicans. Among these 77 peptides at least 13 were considered as the best substrates candidates (supplemental Table S1) with more drastic filters. One of the most striking features of the specificity of Mca1p is that many of these substrates belong to either the family of heat-shock proteins (Ssb1p, Hsp70p, and Hsp90p), and translation machinery (Tef1p, Rps3p, Rpl17Bp, Rpp2Bp, and Hcr1p), or to mitochondria (Qcr2p, Aco1p, Cox5p, and Atp5p) and carbon metabolism (Tpi2p, Tal1p, Tkl1p, Gpm1p, Fba1p, Lsc1p, Lpd1p, and Pmm1p). We confirmed by immunoblot the cleavage of Ssb1p, an HSP family heat-shock protein, in the experimental condition where the metacaspase is activated. This suggests that the onset of apoptosis is associated with a loss of general stress response capabilities and a major metabolic rewiring.
Our quantitative proteomic experiments clearly show that diverse biological responses are elicited by farnesol treatment. Some are Mca1p-dependent, others are Mca1p-independent. Also, the proteome differs according to the presence/absence of Mca1p without farnesol treatment. One purely farnesol effect was the strong induction of Cdr1p, a plasma membrane associated ABC transporter and that has been linked to farnesol toxicity (18,41). Cdr1p mediates the extrusion of glutathione from farnesol-treated cells, leading to a depletion of the intracellular pools of GSH, and therefore enhanced susceptibility to oxidative stress (41). We examined differences between the glutathionylated proteomes in the various bottom-up data sets to study the role of Cdr1p in our experimental model of farnesol-induced apoptosis. There were no massive differences in overall protein glutathionylation, with a core of 108 proteins remaining glutathionylated in all conditions. This does not reflect the expected effects of massive glutathione extrusion triggered by farnesol. This observation was confirmed by an evaluation of the cellular content of reduced glutathione by measuring the formation of a fluorescent adduct with naphthalene-2,3-dicarboxaldehyde (30) and flow cytometry analysis: no significant differences were detected between the GSH pools in the different conditions assayed. There were however some specific targets of protein glutathionylation upon farnesol treatment, and in particular, the presence of farnesol led to Cys712 of Cdr1p being glutathionylated irrespective of the presence or absence of Mca1p. This cysteine residue is located in a transmembrane helix (TM4), and a mutational analysis did not find that it has any critical function in the transport activity of Cdr1p (42). The role of Cdr1p glutathionylation remains unclear. The crystal structures of two ABC transporters in the presence of glutathione, the mitochondrial iron-sulfur cluster exporter Atm1p from S. cerevisiae and the heavy metal Atm1-type ABC exporter from Novosphingobium aromaticivorans (43,44), may provide some clues. Others proteins, such as Hsp104p, were glutathionylated (on Cys544 and Cys640) in farnesol-treated samples but only when Mca1p was absent.
We defined clusters of proteins with similar expression profiles across the different experimental conditions, as determined by the quantitative label-free and the TMT-labeling analyses. The resulting protein classes (Fig. 5) revealed some characteristic expression profiles. Clusters C3 and C10 contain proteins the expression of which was lowered by farnesol treatment independently of the presence of Mca1p; clusters C2 and C10, and to a lesser extent C4, include proteins showing a much larger increase in protein abundance upon farnesol treatment when Mca1p is absent than in wild-type cells. Statistical analysis of the enrichment of gene ontology terms (cellular component, biological process, and molecular function) in the various clusters indicated that the main processes altered are related to stress response, and ribosome biogenesis, protein translation and folding. Farnesol increases the expression of proteins involved in folding and response to stress such as Hsp90p, Ssa1p, Hsp60p, and Glr1p (10). Here, we show that the mca1 deletion considerably enhances the induction of these proteins, but also the induction of other chaperones including Hsp104p, Ssc1p, Ddr48p, and orf19. 7085p. This result raises important questions about the molecular function of Mca1p. Some phenotypes of mca1-deleted C. albicans strains have been described (22): the absence of Mca1p leads to elevated resistance of the cells to oxidative stress, alteration of mitochondrial energy production and rerouting of glucose metabolism. Resistance to oxidative stress is also increased by the down-regulation of the RAS1/cAMP/ PKA pathway by farnesol in C. albicans. This pathway accelerates PCD by inhibiting the expression of stress proteins or by inducing factors of apoptosis (36). Although we did not observe any effects on Ras1p expression, farnesol increased the expression of Tpk2p, a catalytic subunit of PKA, but this effect was not observed in the Mca1p mutant. This is coherent with the HSP-related phenotypes of the mca1⌬/⌬ cells in the presence of farnesol, but cannot be traced back to a PKAdependent inhibition of the expression of the catalase Cat1p, a major contributor to farnesol-induced oxidative stress (15). The differences in expression of Tpk2p are corroborated by those observed for the apoptosis-inducing factor Aif1p ortholog, in our label-free experiments (Table I). Possibly, the RAS1/cAMP/PKA pathway increases the expression of Aif1p ortholog to induce apoptosis events, and this activation may be dependent on the presence of Mca1p.
The sequences and structures of Mca1p in C. albicans and the S. cerevisiae ortholog of Mca1p, the metacaspase Yca1p are very similar, so they are likely to share various functions. An important function of Yca1p is its contribution to the clearance of aggregated proteins in stressed cells (23,24). Yca1p has been found associated with aggregated proteins, together with Cdc48p, a member of the HSP family, and with the aggregate-remodeling chaperones Ssa1p and Ssa2p (24). It has been suggested that mechanism of action of Yca1p is bind to insoluble aggregated proteins and then recruit unfolding/degradation machinery to eliminate them. Our results suggest that Mca1p functions differently: the protein chaperones, including Hsp104p, a major disaggregase in C. albicans (45) appear to be the substrates of Mca1p. The main function of Mca1p appears therefore to be to reduce the capacity of stressed cells to respond to protein alterations, thereby favoring cell death.
Our analyses suggest a second important function of Mca1p. A number of mitochondrial proteins appear to be direct (Cox1p, Atp5p) or indirect (Aif1p) substrates for Mca1p; proteolytic cascades and degradation of mitochondrion are major events of apoptosis (10,46,47), so this reveals a link between Mca1p-dependent apoptosis and the mitochondrial pathway for PCD. Mutations in Cdc48p in S. cerevisiae result in mitochondrion impairments and apoptosis (48). Here, we show that Cdc48p is a potential substrate of Mca1p, with three cleavage sites found, two of which found with the most stringent filters applied. This suggests thus that the endopeptidase activity of Mca1p may directly degrade or at least impair mitochondria. This is relevant to a previously unexplained observation that deleting YCA1 from a frataxin-deficient yfh1⌬ strain that exhibits severe mitochondrial dysfunctions, allows the restoration of cytochromes assembly (49). YCA1 deletion may lower the rate of mitochondria degradation and thereby allow the synthesis and maintenance of partly functional mitochondria. The deletion of MCA1 in C. albicans results in a lower abundance of ROS and a lower mitochondrial membrane potential (22). We report here that the expression of a putative effector of death (orf19.2175), ortholog of the S. cerevisiae Aif1p described to be a mitochondrial apoptosis effector (50), is increased by farnesol in the WT and by less in the mca1⌬/⌬ mutant. Differences in the expression of the apoptosis effector are corroborated by flow cytometric analysis of the activation of apoptosis. However, the expression of this apoptosis inducer in the presence of farnesol was high even in the mca1⌬/⌬ strain, higher than would be expected in the light of the flow cytometry results. This suggests that the induction of apoptosis by the AIf1p ortholog is mostly dependent of Mca1p, consistent with what has been described in S. cerevisiae (50).
Native peptidomic analyses identified only 9 of 416 peptides as potential substrates of Mca1p, and GluC experiments found only 68 among 8549 peptides. This suggests that the metacaspase has a targeted activity. Here, we show that the endopeptidase activity of Mca1p is tightly targeted and directed toward mitochondria, chaperone proteins, and glycolysis proteins, causing mitochondrial damage and thus apoptosis. It would be interesting to investigate the contribution, if any, of cytoplasmic proteases including Mca1p in proteasome and vacuole degradation during apoptosis. No quantitative differences in the numbers of ubiquitinated species were found in our proteomic analyzes of whole protein extracts according to the presence or absence of farnesol and/or Mca1p disruption (data not shown).
In conclusion, the results presented help elucidate the possible functions of Mca1p in C. albicans. Mca1p appears to act as a specific protease, with a limited number of substrates. Another function of Mca1p is to reduce the capability of the cells to respond to stress, by degrading key components of the protein chaperone machineries. Also, Mca1p contributes significantly to the control of mitochondrial biogenesis and degradation, and thereby probably constituting a link between the extrinsic and the intrinsic programmed cell death pathways in C. albicans. Mca1p function could be further described by studying the kinetics of apoptotic responses using a combination of apoptosis inducers.