Metacaspase Yca1 is required for clearance of insoluble protein aggregates

Abstract

In complex organisms, caspase proteases mediate a variety of cell behaviors, including proliferation, differentiation, and programmed cell death/apoptosis. Structural homologs to the caspase family (termed metacaspases) engage apoptosis in single-cell eukaryotes, yet the molecular mechanisms that contribute to nondeath roles are currently undefined. Here, we report an unexpected role for the Saccharomyces cerevisiae metacaspase Yca1 in protein quality control. Quantitative proteomic analysis of Δyca1 cells identified significant alterations to vacuolar catabolism and stress-response proteins in the absence of induced stress. Yca1 protein complexes are enriched for aggregate-remodeling chaperones that colocalize with Yca1-GFP fusions. Finally, deletion and inactivation mutants of Yca1 accrue protein aggregates and autophagic bodies during log-phase growth. Together, our results show that Yca1 contributes to the fitness and adaptability of growing yeast through an aggregate remodeling activity.

Caspases are clan CD cysteine-dependent aspartate-specific proteases that contain conserved cysteine and histidine residues, which together form a catalytic dyad essential for protease activity. The death effector domain (DED) and caspase recruitment domain (CARD) oligomerization motifs within the N-terminal prodomain of initiator caspases facilitate regulatory interactions, which direct the zymogen to apoptotic signaling platforms such as the death-inducing signaling complex (DISC) or apoptosome. Induced proximity of initiator caspases at these platforms promotes autocatalytic cleavage and activation (1–3). Caspase activation through scaffold-dependent proximity is unique to initiator caspases, whereas caspase members that lack the N-terminal prodomain require proteolytic activation through additional proteases. The fully mature caspase conformation is a heteromeric compound of catalytic fragments resulting from the removal of the prodomain and cleavage events, which separate the protein into small (∼P10) and large (∼P20) catalytic components. There are no direct caspase homologs in single-celled organisms; however, bioinformatic investigation of distant ancestors to caspases uncovered a family of structural homologs in plants, fungi, and protozoa termed metacaspases (4). Metacaspases are divided into type I and type II variants based on the presence or absence of a prodomain, respectively, and are distributed irregularly across various species.

The budding yeast Saccharomyces cerevisiae expresses a single type I metacaspase named Yca1 (also termed Mca1). Similar to metazoan caspases, Yca1 contains the conserved catalytic cysteine-histidine dyad and undergoes caspase-like autocatalytic liberation of 10-kDa (P10) and 20-kDa (P20) fragments from a 50-kDa proenzyme (5). The Yca1 prodomain does not contain DED or CARD motifs, and the activation mechanism for Yca1 is currently unknown (6). Instead, the N-terminal prodomain of Yca1 is rich in poly-Q/N repeats, a motif predicted to have prion-like function (7, 8). The prion domain of Yca1 is an unexpected feature; however, its position in the N-terminal regulatory domain of Yca1 suggests that the poly-Q/N repeats could facilitate protein interactions and direct the metacaspase to regulatory structures or substrates.

Caspase family proteases regulate multiple cell behaviors that contribute to organism fitness and pathology. Caspases are established initiators and executioners of programmed cell death (PCD), a central process in the patterning of developing organisms and homeostatic maintenance of tissues in adults. However, the influence of caspase activity is not limited to death-centric pathways. Caspases also promote numerous nonapoptotic processes contributing to immune-cell activation, motility, and differentiation of a variety of cell types (9–11). In response to H₂O₂ and other apoptogenic insults, S. cerevisiae are reported to similarly undergo PCD with numerous markers of apoptosis (5, 12). PCD in response to H₂O₂ is substantially diminished in the YCA1-C297A catalytic inactivation mutant, the deletion mutant (Δyca1), and wild-type cells pretreated with caspase inhibitors. From these observations, Yca1 action has been established as a de facto mediator of yeast PCD in response to oxidative stress and environmental insults. However, there has been limited effort to characterize molecular function of Yca1 in nondeath functions analogous to the larger caspase protein family. Interestingly, wild-type yeast cells were observed to overgrow Yca1 disruptants in competitive growth assays (13). In addition, Yca1 deletion and inactivation alters the timing of the cell cycle, resulting in substantial elongation of the G1 phase and a defective G2/M checkpoint (14). Together, these observations suggest that Yca1 may promote the longevity and greater fitness of yeast, a feature that would curtail the selective pressure against death-centric pathways in single-celled eukaryotes.

In this study, we show that the yeast metacaspase Yca1 elicits a beneficial cell autonomous action. The Yca1 deletion mutant is enriched for stress-response chaperones and proteins involved with vacuolar catabolism. This finding is validated by increased vacuolation in the Δyca1 strain during log-phase growth, which implicates an enhanced autophagic response in compensating cells. Yca1 fusions colocalize with aggregate remodeling chaperones on insoluble aggregates and coprecipitate during logarithmic growth. Moreover, the ability of Yca1 to interact with insoluble material is severely compromised in metacaspase mutants lacking the prion domain. Finally, metacaspase deletion and inactivation strains accumulate insoluble aggregates during log-phase growth. These data support a model for metacaspase activity, whereby Yca1 promotes the removal of insoluble protein aggregates to maintain the fitness of growing yeast.

Results and Discussion

Δyca1 Cells Are Enriched for Stress-Response Proteins, Vacuolar Peptidases, and Autophagic Bodies.

To screen for evidence of nonapoptotic function for Yca1, we measured relative proteomic changes induced by Yca1 ablation in physiologic growth conditions. Trypsin-digested peptides from wild-type and Δyca1 lysates were labeled with iTRAQ mass tag labels and combined and analyzed by liquid chromatography tandem MS (LC/MS-MS). Proteins with iTRAQ ratios significantly different from general protein variation (1.20 < Δyca1/wild type < 0.80) were selected along with those previously identified in 2D gel experiments (14). The 67 up-regulated proteins (Fig. S1 and Table S1) were evaluated using the Funspec cluster interpreter to identify statistical enrichment of gene attributes (15). Of the Funspec categories available, we chose to focus on GO biological process and MIPS subcellular localization (Fig. S2). We analyzed the 13 down-regulated proteins similarly, but they did not produce significant clusters. Among the up-regulated proteins, we observed a substantial overrepresentation of proteins involved with vacuolar catabolism, stress response, and growth/metabolism based on the GO biological-process category (Fig. 1A). Remarkably, five of six known peptidases responsible for catabolic processes within the vacuolar lumen were increased by over 40% (on average). Δyca1 lysates were also enriched for stress-response proteins, including Hsp70 family chaperones (Ssa1, Ssa2, and Ssa4) and Hsp104, a chaperone involved with active resolubilization of protein aggregates (16) (Fig. 1 B and C). In particular, the 3-fold induction of the general stress marker Hsp12 (17) suggests that the Δyca strain is inherently stressed during log growth.

Fig. 1.

Proteomic analysis of Δyca1 reveals up-regulation of stress-response proteins and vacuolar peptidases. (A) GO biological-process clusters (Funspec: P_{hypergeometric} < 5 × 10⁻⁵, Bonferroni correction-enabled) of proteins up-regulated in Δyca1 lysates as determined by iTRAQ. (B and C) iTRAQ ratios for proteins involved with (B) vacuolar catabolism and (C) stress response. (D) FM4-64 vacuole stain depicting normal and multivacuolated morphology. (E) Quantification of multivacuolated wild type, Yca1 catalytic inactivation mutant C297A, and Δyca1 cells during log growth (n = 3; ± SEM).

Given the enriched expression of vacuolar peptidases in our iTRAQ screen, we examined live cells for alterations to the size, number, and morphology in the Yca1 mutant. To examine vacuolar structure, we incubated cells with FM4-64, a vital dye that is rapidly internalized and localizes to the vacuolar membrane. Although most wild-type cells had one to three well-defined vacuoles, a subpopulation of cells was multivacuolated with more than three vacuolar rings per cell. We quantified the proportion of cells with normal or multivacuolated appearance in wild-type, Δyca1, and C297A (catalytically inactive mutant of Yca1 where an essential cysteine in its active site is mutated to an alanine) strains (5, 14). In the Δyca1 strain, we observed a 2.5-fold increase in the number of multivacuolated cells and a 2-fold accumulation of these structures in the C297A strain compared with wild type (Fig. 1 D and E). The delocalized stain and multiple intravacuolar structures within the larger vacuolar space of Δyca1 cells were consistent with the morphology of autophagic bodies (18). The proteomic alterations and accumulation of autophagic bodies in the Δyca1 strain suggested a possible autophagic fate; however, Δyca1 is a viable strain that grows at rates comparable with those of wild type, suggesting that the substantial vacuolation and enrichment of vacuolar peptidases are markers of a limited autophagic process in viable cells. Furthermore, the various stress-response chaperones induced by Yca1 deletion showed that log-phase yeasts are challenged without the action of Yca1. The increased expression of molecular chaperones and activation of stress pathways in Δyca1 cells suggest that particular care should be taken with the interpretation of apoptotic assays, because Δyca1 cells may be preconditioned to sudden insults. Indeed, it is not unreasonable to suggest that the compensatory changes in protein expression in the Δyca1 cells may offer enhanced protection against apoptotic induction. This is in contrast to the prevailing interpretation that loss of YCA1 enhances fitness by simply reducing apoptotic engagement.

Yca1 Copurifies with Aggregate Remodeling Chaperones and Colocalizes Together on Protein Aggregates.

To further define the relationship between Yca1 and the altered stress response, we used a single-step magnetic-bead tandem affinity protein tag (TAP) purification system to identify Yca1-associated complexes. Proteins that consistently copurified with Yca1-TAP and C297A-TAP but were absent from control purifications (Fig. 2A) were identified by LC-MS and evaluated using the Funspec cluster interpreter (Fig. 2B and Fig. S3). We observed populations of stress-response chaperones and translational machinery in Yca1 precipitates. Based on the association of Yca1-TAP with protein translation machinery and the DEAD box helicase Ded1, we hypothesized that Yca1 may directly interact with or impinge on the formation of a class of mRNA processing bodies termed P-Bodies (19). To address this possibility, we expressed a centromeric plasmid containing the enhancer of mRNA decapping protein fused to mCherry Edc3-mCh (20) in wild-type, Yca1-GFP, and Δyca backgrounds as an in situ marker of cytosolic P-Bodies (Fig. S4). We did not observe any difference in the size or number of P-Bodies. Similarly, colocalization of Yca1-GFP with Edc3-mCh positive foci only occurred rarely. These data suggested that the association of Yca1-TAP with these proteins occurred through interactions with non–P-Body complexes, and consequently, the Δyca1 phenotype cannot be explained by a P-Body–dependent effect.

Fig. 2.

Yca1-TAP interactors are enriched for aggregate-remodeling chaperones. (A) Yca1/C297A-TAP–associated proteins are enriched for aggregate-remodeling proteins and chaperones. (B) GO biological-process clusters of Yca1-TAP–associated proteins as determined by Funspec (P_{hypergeometric} < 3 × 10⁻⁴, no Bonferroni correction).

Yca1 predominantly copurified with Ssa1/Ssa2 (Hsp70), Hsp42, and Ydj1 (Hsp40) chaperones and the AAA ATPase Cdc48 (Fig. 2A), a cell-cycle control protein associated with the formation of aggresomes and polyglutamine aggregates (21). These interactions complimented our iTRAQ data, showing that the Hsp70 and Hsp104 aggregate-remodeling chaperones are overexpressed as a compensatory response to Yca1 deletion. Together, these data suggested that YCA1 may contribute to protein quality control by functionally interacting with protein aggregates or the Hsp40/Hsp70/Hsp104 protein aggregate-remodeling complex (22).

To test if Yca1 associated with cytosolic protein aggregates, we monitored the localization of a Yca1-GFP fusion protein that was expressed from the endogenous YCA1 locus. We observed that Yca1-GFP is diffuse throughout the cell and generally excluded from vacuoles in baseline growth conditions (Fig. 3A) (T = 0). However, when subjected to 42 °C heat stress to promote the formation of microscopically discernable cytosolic protein aggregates, Yca1-GFP converged into discrete foci, which colocalized with visible features of the cells detectable by Nomarski optics (Fig. 3A). Yca1-GFP foci were apparent as early as 7 min after exposure to heat, and the Yca1-GFP signal returned to a diffuse cytosolic distribution within 60–90 min after return to 30 °C (Fig. 3A and Fig. S5). To mark protein aggregation, we expressed Hsp104-mRFP under its own promoter in a centromeric plasmid (23) and observed robust colocalization of Yca1-GFP with Hsp104-mRFP-positive aggregates during heat stress (Fig. 3B). Furthermore, we observed formation and colocalization of Yca1-GFP and Hsp104-mRFP foci in aged stationary-phase cultures (Fig. S6), a second physiologic stress capable of promoting protein aggregation (24). We examined whether protease activity or the catalytic cysteine of Yca1 is important for its localization to protein aggregates. To test this supposition directly, the C297A catalytic inactivation mutant was fused to mRFP and expressed in wild-type, YCA1-GFP, and Δyca1 backgrounds from a plasmid under control of the YCA1 gene-promoter sequence (Fig. 3 C and D); C297A also converged into discrete foci during heat stress in all backgrounds, and C297A-mRFP colocalized strongly with Yca1-GFP, indicating that the catalytic core of Yca1 is not required for its aggregate-targeting function.

Fig. 3.

Yca1 targets insoluble protein aggregates irrespective of catalytic activity. (A) Yca1-GFP reversibly condenses into discrete foci during heat stress. Arrows depict colocalization between Yca1-GFP and visible features of the differential interference contrast (DIC) channel. (B) Yca1-GFP colocalizes with Hsp104-mRFP during heat stress. (C and D) Catalytic inactivation of Yca1 (C297A-mRFP expressed in Δyca1) does not affect its aggregate-targeting function. (E) Western blots of soluble and insoluble fractions confirm association of full-length Yca1 with the insoluble phase during heat stress quantified by densitometry in F.

To confirm biochemical association of Yca1 with protein aggregates, lysates from logarithmic-phase Yca1-TAP cells were separated into soluble and insoluble fractions by differential centrifugation over a time course of heat stress and recovery (Fig. 3 E and F). Interestingly, Yca1-TAP was distributed between the soluble and insoluble fractions during logarithmic growth. After 1 h of heat stress, full-length Yca1-TAP and catalytic fragments were depleted from the soluble fraction and increased in the insoluble fraction. Within 90 min of recovery, the pattern of Yca1-TAP fractionation was similar to baseline, albeit with slightly higher expression levels and a greater amount of catalytically active Yca1 in the soluble fraction. The distribution of full-length C297A-TAP in soluble and insoluble fractions behaved similarly to Yca1-TAP over the time course, with the exception of the catalytic fragment, which did not form in the mutant. Together, these data show that Yca1 is actively expressed in response to heat stress and shuttles to insoluble protein aggregates in a manner that does not require catalytic activity or prodomain cleavage.

Yca1 Inactivation Mutants Accumulate Protein Aggregates.

Does Yca1 target itself to aggregates to prevent self-activation, or does Yca1 play a role in aggregate clearance? In light of the aggregate-specific localization of Yca1 and the potential for proteolytic activity, we reasoned that Yca1 in the insoluble fraction may actively remodel or disassemble protein aggregates during exponential growth. To examine this question, we measured the nonidet P40-insoluble protein-aggregate fraction from log-phase wild-type, Δyca1, and C297A protein lysates by filter-trap analysis (Fig. 4 A and B). We observed a 2-fold increase of insoluble aggregates in Δyca1 and a 1.4-fold increase in the endogenous C297A catalytic-inactivation mutant. The partial correction of the aggregate phenotype in the C297A strain suggests that Yca1 prevents protein-aggregate accumulation through a noncatalytic mechanism as well as metacaspase-dependent catalytic activity.

Fig. 4.

Metacaspase mutants accumulate insoluble aggregate material during logarithmic growth. (A) Filter-trap assay of soluble and insoluble fractions from equal amounts of total wild-type (BY4741), Δyca1, and C297A lysates. (B) Densitometry analysis of insoluble fractions from filter-trap assays (n = 3; ±SEM). (C) Anti-HIS Western blot of genomic HIS-tagged Ssa1 in the BY4741 and Δyca1 backgrounds. (D) Silver-stained 1D PAGE gel of insoluble fractions from midlogarithmic phase BY4741 and Δyca1 lysates. (E) Domain structure of Yca1. The prodomain (tan) contains the poly-Q (QXXQ) motif. The catalytic histidine (H) and cysteine (C) are marked. (F) Anti-mRFP Western blots of full-length, C374, and C323 N-terminal truncation mutants are quantified in G. Zymogen (Zym) and catalytic (Cat) fragments are marked.

To elucidate the aggregate phenotype associated with caspase deletion and inactivation, the contents of the insoluble fraction were further analyzed. First, the Hsp70 protein Ssa1 was polyhistidine tagged (HIS) at its endogenous locus in wild-type and Δyca1 cells. Ssa1-HIS was found to accumulate in the insoluble fraction of Δyca1 cells compared with wild type (Fig. 4C). This observation suggested that compensatory expression of Ssa1 in response to metacaspase deletion functionally contributed to countering the aggregate-retention phenotype and confirmed the aggregate-retention trend observed in the filter-trap assay. Second, insoluble aggregates from equal amounts of total protein extract were separated by 1D PAGE and silver stained (Fig. 4D). The pattern of protein bands between the Δyca1 and wild-type samples was not altered, suggesting that Δyca1 cells accumulate protein aggregates of a nonspecific composition rather than target individual protein species. Consistent with the filter-trap assay, the intensity of the bands in the aggregate fraction from Δyca1 lysates was directly comparable with the intensity of the aggregate fraction from two times as much wild-type protein, providing a second confirmation of the trend observed in the filter-trap assay. Together, these findings show a role for the Yca1 protein and protease activity in general aggregate clearance.

Prion Domain of Yca1 Mediates Its Delivery to the Insoluble Fraction During Logarithmic Growth.

The molecular mechanisms that direct Yca1 to protein aggregates are not clear. The N-terminal domains of initiator caspases direct mammalian caspases to aggregates of DED and CARD oligomerization motifs that promote autocatalytic activation. Intriguingly, mammalian caspases have also been observed to directly converge on inclusions of extended polyglutamines, similarly promoting their activation (25). As such, we hypothesized that the prion domain of Yca1 could directly target aggregates during log growth.

To examine this possibility, we generated a series of N-terminal truncation mutants of Yca1 fused to mRFP and expressed them from centromeric plasmids under control of the ADH1 promoter sequence (Fig. 4E). During logarithmic growth, all of the mRFP fusions were diffuse in the cytoplasm, similar to the localization of Yca1-GFP fusions. We separated the lysates by differential centrifugation to determine their distribution between the soluble and insoluble fractions (Fig. 4 F and G). Full-length Yca1-mRFP were distributed similarly to the Yca1-TAP fusion. The C374 mutant, which lacks the poly-Q (QQXX) motif in its prodomain, showed a divergent trend to full-length Yca1 and was predominantly soluble during log growth. Finally, the C323 mutant that lacks the entire prodomain underwent spontaneous maturation and was detected principally as a catalytic fragment in both fractions. Residual association of both truncation mutants with the insoluble fraction was evident, suggesting that the C-terminal portion of Yca1 retains some affinity for insoluble material; however, the substantial shift from an insoluble species of Yca1 to a soluble species strongly supports the contention that the prion domain is largely responsible for delivery of Yca1 to insoluble aggregates during log growth. Furthermore, these data show that the prodomain of Yca1 stabilizes the Yca1 zymogen, thus providing regulatory control over metacaspase activation.

Model for Yca1 in Aggregate Clearance.

The current study suggests that metacaspase function is rooted in both death and vital nondeath roles. We propose a model, whereby Yca1 acts to target and limit the formation of insoluble-protein aggregates during the mitotic cell cycle and chronological aging (Fig. 5). In our model, Yca1 contributes to the clearance of protein aggregates over the natural yeast lifespan. Here, the proteolytic activity of Yca1 may direct aggregate degradation, or Yca1 may also act as a stabilizing scaffold for other aggregate-remodeling enzymes. The accumulation of protein aggregates in the absence of Yca1 leads to compensatory expression of stress-response genes and protective autophagy, a response known to mediate aggregate clearance (26–28). We observed that expression of C297A can partially alleviate, but not correct, the autophagic and aggregate accumulation phenotypes. These observations are indicative of a nonproteolytic function, whereby Yca1 might additionally enhance the activity of another protease or act as a stabilizing scaffold of aggregate-remodeling complexes at the aggregate surface. In light of our current findings, the Yca1 prion-domain–dependent targeting of aggregated material may represent a rudimentary activation scaffold analogous to signaling platforms such as the DISC and apoptosome.

Fig. 5.

Model depicting the role of Yca1 in the balance of protein-aggregate assembly and dissolution. In wild-type cells (A), Yca1 targets insoluble-protein aggregates and cooperates with heat-shock proteins to promote aggregate clearance. The improved aggregate clearance in the C297A inactivation mutant suggests that Yca1 may participate in a recruitment or stabilization process to regulate aggregate-remodeling complexes at the aggregate surface. In caspase-deletion mutants (B), the balance of aggregate formation and disassembly is disrupted, favoring their accumulation. The amassing aggregates are too large to be processed by the 20-s proteasome pathway and instead, require autophagic encapsulation and a concomitant stress response for efficient removal.

The dynamics of protein-aggregate formation and dissipation have relevance to stress, disease, and the replicative/chronological aging process (29). Our model for metacaspase function in aggregate dynamics explains a number of defects observed in metacaspase deletion mutants. Autophagy contributes to G1/S timing and arrest (30), which suggests that aggregate accumulation and consequent autophagic degradation is the cause for the lengthened G1/S phase observed in Δyca1 cells (14). Replicative and chronological aging is typified by accumulation of oxidative damage and aggregated proteins, which reduces clonogenicity and predisposes aging mother cells to death (31, 32). Colocalization of Yca1-GFP with Hsp104-mRFP in stationary cultures (Fig. S6) may represent the clearance of inclusions and a mechanism that explains why aged Δyca1 cultures do not regrow (13). Similarly, aggregates of an ectopic polyglutamine construct were observed to localize in nuclei of wild-type yeast but accumulate in the cytoplasm of Δyca1 cells (33). The Yca1-dependent localization of these aggregates suggests that metacaspase function could promote their disassembly in the cytoplasm or directly alter their structure, thus permitting entry into the nucleus. Finally, oxidative stress induced by H₂O₂ causes Δyca1 cells to accumulate excessive amounts of protein carbonyls (12). Here, Yca1 would counter the general aggregates of oxidized proteins that form in response to nonlethal oxidative stress, contributing to cell viability.

The yeast metacaspase Yca1 has been previously characterized as an ancestral caspase-regulating apoptosis in yeast. Here, we have shown a number of fitness defects associated with deletion and inactivation of Yca1 in ambient conditions. Baseline metacaspase function in our model for Yca1 protects cells against by-products of the aging process and toxic amyloids. The role of Yca1 in the dissipation of protein aggregates suggests that the teleology of metacaspases includes vital nondeath cell function.

Materials and Methods

Strains and Growth Conditions.

All S. cerevisiae strains used in this study were derived from the wild-type haploid BY4741 strain. For fractionation and vacuole morphology experiments, samples were grown in YPDm (1% yeast extract, 2% peptone, 2% glucose pH buffered to 3.5 with HCl), which is the same growth condition that we used previously to examine a Yca1-specific cell-cycle defect (14). To examine the localization of fluorescent fusion proteins, samples were grown in synthetic media (complete or with trophic-marker dropout as required).

Microscopy.

Cells were viewed with an Axioplan 2 compound microscope fitted with a Zeiss Axiocam (63× magnification; PlanApochromat 1.4 numerical aperture objective lens, 22 °C) with appropriate filters.

Filter Trap.

Protein-aggregate fractions were collected, and serial dilutions were subjected to vacuum filtration through a PVDF membrane using a BioDot SF microfiltration apparatus. Filter-trap membranes were stained with coomassie blue, and slot intensity was determined by densitometry.

Detailed procedures are in Tables S2, S3 and SI Materials and Methods.