ABSTRACT
Phosphatidylserine decarboxylase 1 (Psd1p), an ancient enzyme that converts phosphatidylserine to phosphatidylethanolamine in the inner mitochondrial membrane, must undergo an autocatalytic self-processing event to gain activity. Autocatalysis severs the protein into a large membrane-anchored β subunit that noncovalently associates with the small α subunit on the intermembrane space side of the inner membrane. Here, we determined that a temperature sensitive (ts) PSD1 allele is autocatalytically impaired and that its fidelity is closely monitored throughout its life cycle by multiple mitochondrial quality control proteases. Interestingly, the proteases involved in resolving misfolded Psd1ts vary depending on its autocatalytic status. Specifically, the degradation of a Psd1ts precursor unable to undergo autocatalysis requires the unprecedented cooperative and sequential actions of two inner membrane proteases, Oma1p and Yme1p. In contrast, upon heat exposure postautocatalysis, Psd1ts β subunits accumulate in protein aggregates that are resolved by Yme1p acting alone, while the released α subunit is degraded in parallel by an unidentified protease. Importantly, the stability of endogenous Psd1p is also influenced by Yme1p. We conclude that Psd1p, the key enzyme required for the mitochondrial pathway of phosphatidylethanolamine production, is closely monitored at several levels and by multiple mitochondrial quality control mechanisms present in the intermembrane space.
INTRODUCTION
In an aqueous environment, the amphipathic chemistry of phospholipids drives the formation of lipid bilayers. These bilayers encapsulate the cell and its internal compartments and are indispensable for life. As one of the major membrane phospholipids, phosphatidylethanolamine (PE) is of fundamental importance for cell physiology, with numerous emerging links to human pathology. PE serves as a precursor to the major cellular phospholipid, phosphatidylcholine (1), and is a critical component of glycosylphosphatidylinositol anchors (2), the cleavage furrow (3), and autophagy (4). In bacteria, PE can dictate membrane protein topology and function (5). Defects in PE levels have been linked to Alzheimer's disease (6), Parkinson's disease (7), and liver diseases (8). In spite of its clear physiologic importance, there remain significant gaps in our basic understanding of PE metabolism, including detailed mechanistic information concerning the enzymes responsible for its synthesis.
Of the four cellular pathways of PE production, two are essential for mammalian development and are differentially localized to the endoplasmic reticulum (ER) or the mitochondrion (9, 10). The role of mitochondria in lipid synthesis is rarely mentioned despite the fact that they are second only to the ER in this functional capacity. Phospholipids made in the mitochondrion are phosphatidic acid (PA), CDP-diacylglycerol, phosphatidylglycerol, cardiolipin (CL), and PE (11, 12). Saccharomyces cerevisiae expresses two phosphatidylserine decarboxylases, Psd1p and Psd2p, which produce PE by decarboxylating phosphatidylserine (PS). Psd1p is localized in the mitochondrion (13, 14), while Psd2p resides in the endomembrane system (15, 16). Psd1p is conserved from bacteria to humans, while Psd2p is found only in yeast. Yeasts lacking both Psd1p and Psd2p (psd1Δ psd2Δ) are ethanolamine auxotrophs and require ethanolamine supplementation to produce PE by the ER-localized CDP-ethanolamine pathway, the second major cellular pathway of PE biosynthesis.
Psd1p is encoded within the nucleus and imported into the mitochondrion postsynthesis as a zymogen (14). Its maturation requires three proteolytic processing steps, the last of which is executed by Psd1p itself and is essential to generate an active decarboxylase. This self-processing event, termed autocatalysis, separates the enzyme into two subunits, α and β, and generates a pyruvoyl prosthetic group at the NH2 terminus of the α subunit that is absolutely required for enzymatic activity (17 – 19). Autocatalysis occurs within a conserved LGST motif between the glycine and the serine residues through a process of serinolysis (17, 18, 20 – 23). The α subunit remains noncovalently associated with the β subunit, which is anchored to the mitochondrial inner membrane so that PS decarboxylation occurs on the intermembrane space (IMS) side (21). Aside from the conserved LGST motif, mechanistic insight into this essential self-cleaving event is limited.
We recently showed that Psd1p rerouted to the secretory pathway undergoes autocatalysis normally and is capable of rescuing the ethanolamine auxotrophy of psd1Δ psd2Δ yeast (23). As such, unique mitochondrial factors—protein, lipid, or other—are not required for Psd1p autocatalysis. Thus, as long as it is anchored in a membrane, everything that is needed for Psd1p autocatalysis is encoded within the polypeptide itself. Based on these results, we hypothesized that membrane anchorage of Psd1p promotes a tertiary structure that is critical for autocatalysis. As such, mutations outside the conserved LGST motif that disrupt this postulated structure would be predicted to impair autocatalysis.
With the goal of identifying novel motifs required for Psd1p autocatalysis and function, we investigated the molecular basis of the temperature-sensitive behavior of a PSD1 allele (encoding Psd1ts) that has four missense mutations in the β subunit (24). We hypothesized that the temperature sensitivity of Psd1ts was due to a defect in autocatalysis, which we demonstrate to indeed be the case. Our additional results indicate that the fidelity of Psd1ts is closely monitored by multiple mitochondrial quality control proteases. Interestingly, the efficient degradation of the Psd1ts precursor requires the sequential activities of two distinct proteases, with Oma1p working prior to Yme1p. In contrast, postautocatalysis, Yme1p, which is not required for the rapid turnover of the small α subunit, is able to degrade the mature Psd1ts β subunit alone. We conclude that the key enzyme required for the mitochondrial pathway of phosphatidylethanolamine production, Psd1p, is closely monitored at several levels and by multiple mitochondrial quality control mechanisms present in the IMS. These results further underscore the efficiency and dedication of the quality control machinery for the preservation of membrane homeostasis within the mitochondrion.
RESULTS
Defective autocatalysis of a temperature-sensitive PSD1 allele.Using error-prone PCR, a temperature-sensitive PSD1 allele that confers a growth defect at the nonpermissive temperature of 37°C was previously established (24). Psd1ts harbors four missense mutations—K356R, F397L, E429G, and M448T—in the β subunit (Fig. 1A). However, the molecular basis for the thermolability of Psd1ts has never been established. We hypothesized that the introduced mutations may disrupt Psd1p autocatalysis at the restrictive temperature, which would in turn prevent it from functioning in mitochondrial PE synthesis. Consistent with the original characterization of the ts allele (24), Psd1ts supported growth of the psd1Δ psd2Δ strain in the absence of ethanolamine at 22°C and 30°C but failed to do so at 37°C (Fig. 1B). Wild-type (WT) Psd1p rescued growth of psd1Δ psd2Δ yeast in the absence of ethanolamine at all temperatures. Consistent with our hypothesis, whereas WT Psd1p was properly matured into separate β and α subunits following growth at each temperature, Psd1ts had a mild autocatalytic defect at the permissive temperature (22°C) that was exacerbated at 30°C. Following overnight growth at the restrictive temperature (37°C), even the Psd1ts precursor was not detected by immunoblotting (Fig. 1C). Interestingly, even though Psd1ts α and mature β subunits were not detected by immunoblotting when grown at 30°C (Fig. 1C), there was apparently still enough functional enzyme present to rescue the ethanolamine auxotrophy of the psd1Δ psd2Δ strain at this temperature (Fig. 1B). Indeed, an analysis of mitochondrial phospholipids revealed that Psd1ts significantly increased PE levels relative to the parental psd1Δ psd2Δ strain following overnight growth at 22°C and 30°C but not at 37°C (Fig. 1D and E).
Psd1ts has a temperature-sensitive autocatalytic defect. (A) Schematic of in vivo Psd1p constructs. The missense mutations in Psd1ts are indicated. MTS, mitochondrial targeting signal; TM, transmembrane domain. (B) The indicated strains precultured at 22°C were spotted onto synthetic complete dextrose medium (SCD) with or without 2 mM ethanolamine and incubated at 22°C, 30°C, or 37°C for 3 days. (C) The α and β subunits of Psd1p were analyzed by immunoblotting in whole-cell extracts isolated from cultures grown at the indicated temperatures; Pic1p served as a loading control. β-α, Psd1p that has not performed autocatalysis. The migration of molecular mass markers in kilodaltons is indicated at the left. (D) After steady-state labeling with 32Pi in SCD with 2 mM choline at the indicated temperatures, phospholipids were extracted from mitochondria isolated from the indicated strains and separated by thin-layer chromatography (TLC). The migration of phospholipids is indicated (PC, phosphatidylcholine; PI, phosphatidylinositol). (E) The relative abundance of PE was determined for each strain at each temperature and is expressed as a percentage of the total phospholipid for each strain at a given temperature (means ± SEM; n = 6). For the 22°C and 30°C samples, significant differences were determined by one-way ANOVA with Holm-Sidak pairwise comparisons; for the 37°C samples, the indicated significant differences were determined by Kruskal-Wallis one-way ANOVA on ranks, with Tukey test pairwise comparisons. n.s., not significant. (F) In vitro import of a [35S]methionine-labeled WT, ts, or S463A Psd1p autocatalytic mutant into wild-type mitochondria at 22°C, 30°C, and 37°C and in the presence (+ΔΨ) or absence (−ΔΨ) of the proton motive force across the inner membrane; 5% of each precursor (−) and 100% of every time point were analyzed. P, precursor; I, import intermediate; m, mature β subunit.
Next, radiolabeled WT and ts Psd1 precursors were incubated with wild-type mitochondria, and their import and processing were followed as a function of time, temperature (22°C, 30°C, or 37°C), and the presence and absence of a membrane potential (Fig. 1F). Psd1p precursor harboring an S463A mutation in the conserved LGST motif served as an autocatalysis-defective control (23). As expected, WT Psd1p was imported into mitochondria and underwent autocatalysis in a time- and membrane-potential dependent manner at every temperature tested. In contrast, Psd1ts was imported into mitochondria but failed to undergo autocatalysis at any temperature. Thus, in organello import studies of Psd1ts revealed a severe defect in autocatalysis even when incubated at a temperature that is permissive in vivo. We conclude that the four mutations introduced into the β subunit of the ts allele inactivate Psd1p function at the restrictive temperature by disrupting its autocatalysis.
Psd1p is a self-processing serine protease before becoming a decarboxylase.The fact that ER-targeted Psd1p undergoes autocatalysis and is functional in vivo (23) indicates that everything needed for autocatalysis is self-contained within the polypeptide. Currently, the only structural motif that is known to be required for autocatalysis across many species is the conserved LGST motif (17, 18, 20 – 23). A fundamental question regarding the autocatalytic mechanism is what makes Ser463 of the LGST motif a strong nucleophile. We postulated that Psd1p serinolysis involves a catalytic triad consisting of an acid and a base present within the polypeptide. Indeed, 3 bases (His residues [Fig. 2A, green]) and 12 acids (Asp, Asn, Glu, and Gln) are conserved in the β subunit of Psd1p from bacteria to humans. Of the three conserved bases, only mutation of His345 prevented autocatalysis (Fig. 2B) and failed to support growth of the psd1Δ psd2Δ strain in the absence of ethanolamine (Fig. 2C). Interestingly, His345 is flanked by the four mutations encoded by the ts allele (Fig. 2A, red dashes) suggesting a plausible mechanism by which they disturb the autocatalytic process. Similar to the H345A and S463A mutants, a D210A Psd1p mutant was autocatalytically impaired and nonfunctional (Fig. 2D and E). These results suggest that Ser463-His345-Asp210 of yeast Psd1p form a catalytic triad typical of serine proteases. In sum, Psd1p initially functions as a self-processing serine protease whose activity unmasks its final PS decarboxylase activity.
Identification of an (auto)catalytic triad in Psd1p. (A) Clustal Omega sequence alignment of Psd1p from the indicated species (E. coli, Escherichia coli). Conserved potential bases (His residues; highlighted in green) and acids (Asp, Asn, Glu, and Gln; highlighted in purple) are indicated. The LGST motif is shown in yellow. The positions of the four missense mutations present in Psd1ts are marked with red dashes above the alignment. The determined base, His345, and acid, Asp210, are indicated with darker shading and green and purple dashes. (B and D) psd1Δ psd2Δ yeast (−) and psd1Δ psd2Δ yeasts transformed with the indicated Psd1p constructs were grown overnight in SCD supplemented with 2 mM ethanolamine. The α and β subunits of Psd1p were detected in whole-cell extracts by immunoblotting; Pic1p served as a loading control. The migration of molecular mass markers in kilodaltons is indicated at the left. (C and E) psd1Δ psd2Δ yeast (−) and psd1Δ psd2Δ yeasts transformed with the indicated Psd1p constructs were spotted onto SCD plates with or without 2 mM ethanolamine and incubated at 30°C for 4 days.
Psd1ts is an unstable polypeptide with destabilized intramolecular interactions.The mutations in Psd1ts disturbed the autocatalytic process before it occurred. We wondered if the same mutations could disrupt Psd1ts when shifted to the restrictive temperature postautocatalysis. To determine the relative stability of WT and ts Psd1p at different temperatures, in vivo degradation assays were performed (Fig. 3A). Following growth at the permissive temperature to allow autocatalysis of Psd1ts, cycloheximide was added to inhibit new cytosolic protein synthesis, and the cultures were either maintained at 22°C or shifted to 30°C or 37°C prior to ascertaining the abundances of Psd1p α and β subunits at various times by immunoblotting. Wild-type Psd1p α and β subunits were relatively stable at each temperature (Fig. 3B to E). At the permissive temperature, both Psd1ts subunits were stable, similar to WT Psd1p (Fig. 3B and C). In contrast, upon shifting to 30°C or 37°C, the stability of both Psd1ts subunits was significantly reduced compared to that of WT Psd1p (Fig. 3B, D, and E). Interestingly, at 37°C, the α subunit of Psd1ts disappeared faster than its β subunit. To extend these results, mitochondria isolated from psd1Δ psd2Δ yeast grown at permissive temperature and expressing either WT or ts Psd1p were incubated at 30°C or 37°C for up to an hour prior to determining the levels of Psd1p α and β subunits by immunoblotting (Fig. 3F to H). As observed in vivo (Fig. 3B and E), the steady-state abundance of the Psd1ts α subunit decreased more rapidly at 37°C than that of the β subunit, which was itself less stable than its WT counterpart (Fig. 3H). This suggests that at higher temperature the Psd1ts β subunit might be partially unfolded, leading to the release of the α subunit. Further, these results implicate mitochondrial proteases in the temperature-dependent reduction in Psd1ts levels.
Psd1ts is unstable at nonpermissive temperature. (A) In vivo degradation assay. (B) Whole-cell extracts were isolated at the indicated times following growth at the indicated temperatures in the presence of cycloheximide (CHX). Samples were resolved by SDS-PAGE and immunoblotted as indicated. (C to E) The α and β subunits of WT and ts Psd1p remaining at each time point at 22°C (C), 30°C (D), and 37°C (E) were quantified and plotted as the percentage of protein remaining compared to time 0 (means ± SEM; n = 4). * and #, P ≤ 0.05 (Student's t test). (F) Mitochondria isolated from psd1Δ psd2Δ expressing WT or ts Psd1p at 22°C were heat treated as indicated. Following incubation for the indicated times, mitochondria were recovered and 50 μg was resolved by SDS-PAGE and immunoblotted as indicated. (G and H) The percentages of Psd1 β and α subunits remaining at each time point at 30°C (G) and 37°C (H) were quantified (means ± SEM; n = 6). * and #, P ≤ 0.05 (Student's t test). The migration of molecular mass markers in kilodaltons is indicated at the left of panels B and F.
To begin to assess the folding status of Psd1ts, a blue-native PAGE (BN-PAGE) analysis was performed. BN-PAGE is a gentle electrophoretic technique that can resolve intact protein complexes, thus providing information on the quaternary structure of a protein. Postautocatalysis, the α and β subunits of Psd1p remain noncovalently associated. Consistent with this, both the α and β subunits of WT Psd1p comigrated in a discrete complex of >669 kDa (Fig. 4A, arrowheads) and as diffuse complexes centered around 140 kDa (Fig. 4A, red lines). In contrast, Psd1ts α subunit-containing complexes were absent, which caused the Psd1ts β subunit-containing complexes to be slightly downshifted compared to the WT (Fig. 4A, gray lines). Given that its steady-state abundance as determined following SDS-PAGE was similar to that of WT Psd1p (Fig. 4A, bottom), the failure to detect Psd1ts α subunit-containing complexes indicates that the Psd1ts β and α subunit interaction is destabilized in BN-PAGE. Since the α subunit is only ∼10 kDa, it would not likely be resolved in this gel system. To directly determine if the noncovalent interaction between the α and β subunits of Psd1ts postautocatalysis is impaired, 3× FLAG-tagged Psd1p α was immunoprecipitated from mitochondria isolated from yeast grown at permissive temperature, and the presence of copurified β subunit was determined (Fig. 4B). Indeed, in contrast to WT Psd1p, the β subunit of the ts allele was only weakly copurified with the α subunit, and a significant fraction of Psd1ts β was instead detected in the unbound flowthrough (Fig. 4C). These results indicate that the interaction of the α and β subunits of Psd1ts is compromised even at the permissive temperature.
Association of α and β subunits is destabilized in Psd1ts. (A) Mitochondria isolated at the permissive temperature from psd1Δ psd2Δ yeast expressing FLAG-tagged WT or ts Psd1p, or untagged WT Psd1p, were solubilized with 1.5% (wt/vol) digitonin, separated by 6 to 16% BN-PAGE, and immunoblotted for the α or β subunit. (Bottom) Immunoblots following SDS-PAGE. The arrowheads and red lines mark Psd1p complexes that were detected by antibodies against both the α and β subunits; the gray lines indicate ts Psd1p complexes that migrated faster and were detected only with the β subunit antisera; the asterisks mark background bands identified in psd1Δ mitochondrial extracts. (B) Following solubilization with 1.5% (wt/vol) digitonin, anti-FLAG resin was used to immunoprecipitate the FLAG-tagged α subunit, and the presence of copurified β was determined by immunoblotting; Tom70p and Pic1p served as controls. SM, starting material; B, bound material; FT, nonbinding flowthrough. The migration of molecular mass markers in kilodaltons is indicated at the left in panels A and B. (C) The percentage of each subunit detected in the nonbinding flowthrough was determined as follows: FT/SM × 100, where FT is the volume of Psd1p α or β subunit detected in the flowthrough and SM is the volume associated with the corresponding starting material. Means ± SEM; n = 3. Student's t test was used to compare the α and β subunits. *, P < 0.001 versus FLAG-tagged WT Psd1p (one-way ANOVA with Holm-Sidak pairwise comparisons).
The α and β subunits are codependent.At elevated temperatures, the α subunit of Psd1ts exhibited a higher rate of decay than the β subunit (Fig. 3E and H), suggesting that postautocatalysis, the α subunit requires the β subunit for its stability. Whether the WT β subunit requires the α subunit for its stability has not been established. Therefore, we generated in vivo constructs that allowed the α and the β subunits to be expressed separately (Fig. 5A). To target the α subunit to the mitochondrial IMS, the mitochondrial targeting sequence from cytochrome c 1 was used (25). The constructs were transformed either individually or in combination into psd1Δ psd2Δ yeast. Whereas the α subunit was readily detected when expressed individually, the β subunit was virtually undetectable (Fig. 5B). The failure of the β subunit to accumulate was not secondary to a defect in its mitochondrial import (Fig. 5D). Interestingly, when both subunits were coexpressed, the steady-state abundance of the β subunit was somewhat stabilized. To more rigorously test the ability of the α subunit to promote the stability of Psd1β in trans, psd1Δ psd2Δ yeasts expressing Psd1β were retransformed with either a plasmid encoding the IMS-directed α subunit subject to doxycycline (Dox) repression or the empty vector (Fig. 5E). In this model, Psd1β levels were directly proportional to the amount of coexpressed IMS-α (Fig. 5F). Since the catalytically essential pyruvoyl prosthetic group is generated by the autocatalytic process, artificial separation of the two subunits, even if they were brought back together physically, failed to rescue growth of the psd1Δ psd2Δ strain in the absence of ethanolamine, as expected (Fig. 5C and G). Finally, we determined whether the capacity of IMS-α to promote the stability of Psd1β reflected its ability to physically interact in trans. Consistent with the idea that IMS-α promotes Psd1β stability through noncovalent association, Psd1β was copurified with 3× FLAG-tagged IMS-α (Fig. 5H). Interestingly, the ability of digitonin to extract Psd1β was improved when coexpressed with IMS-α (Fig. 5I). The failure of digitonin to solubilize Psd1β when expressed in isolation could reflect its folding status and/or accumulation in insoluble protein aggregates. Taken together, these results indicate that preautocatalysis, the β subunit requires the α subunit for its stability, while postautocatalysis, they are mutually codependent.
Preautocatalysis, Psd1 β requires α for stability. (A) Schematic of in vivo constructs to individually express the β and α subunits. The stop-transfer signal of cytochrome c 1 is indicated in pink. (B) Whole-cell extracts from psd1Δ psd2Δ yeasts transformed as indicated were resolved by SDS-PAGE and immunoblotted for the α and β subunits; Tom70p served as a loading control. The asterisks mark IMS-targeted α subunit that had not been fully processed. (C) In parallel, the same yeasts were spotted onto SC medium without Leu and Ura with or without 2 mM ethanolamine and incubated at 30°C for 3 days. (D) In vitro import of [35S]methionine-labeled WT or Psd1 β into wild-type mitochondria at 30°C in the presence (+ΔΨ) or absence (−ΔΨ) of the proton motive force; 5% of each precursor (−) and 100% of every time point were analyzed. P, precursor; I, import intermediate; m, mature β subunit. (E) psd1Δ psd2Δ yeasts expressing Psd1β were transformed with either the empty vector (EV) or IMS-a subject to doxycycline repression (Doxoff). (F) Immunoblots for the α and β subunits in the indicated whole-cell extracts; Tom70p served as a loading control. The asterisks mark IMS-targeted α subunit that had not been fully processed. (G) In parallel, the same yeasts were spotted onto SC medium without Leu and Ura with or without 2 mM ethanolamine and incubated at 30°C for 3 days. (H) Following solubilization of the indicated mitochondria with 1.5% (wt/vol) digitonin, anti-FLAG resin was used to immunoprecipitate the FLAG-tagged α subunit, and the presence of copurified β was determined by immunoblotting; Tom70p and Pic1p served as controls. SM, starting material; B, bound material; FT, nonbinding flowthrough. Background bands detected in each bound lane detected in the FLAG immunoblot are marked with blue dots. (I) To determine digitonin solubilization efficiency, 25 μg of starting material (SM) and nonextracted (NE) pellet post-digitonin solubilization was resolved by SDS-PAGE and immunoblotted as indicated. Decreased signal intensity in the NE lanes (arrowheads) relative to the corresponding SM lanes reflects the fraction of each protein that was extracted by digitonin. The percentages of Psd1 β not extracted by digitonin are shown below the blot and were determined as follows: NE/SM × 100, where NE is the volume of Psd1 β subunit detected in NE and SM is the volume associated with the corresponding starting material. Means ± SEM; n = 3. Significant differences were measured by one-way ANOVA with Holm-Sidak pairwise comparisons. The migration of molecular mass markers in kilodaltons is indicated at the left of every immunoblot.
The i-AAA protease degrades the Psd1ts β subunit following its heat-induced aggregation.Yme1p forms the i-AAA protease, which has a principal role in enforcing quality control within the mitochondrial IMS (26, 27). It was previously reported that deletion of YME1 augments Psd1p activity, with a concomitant increase in mitochondrial PE levels (28). From these results, it was postulated that Yme1p contributes to the proteolytic turnover of Psd1p, although direct evidence of this was never provided. To ascertain if Yme1p is a determinant of Psd1p stability, YME1 was deleted in the context of psd1Δ psd2Δ yeast expressing a given amount of WT or ts Psd1p (Fig. 6A). Next, the expression levels of WT and ts Psd1p were compared following growth at 22°C, 30°C, and 37°C in the presence or absence of Yme1p (Fig. 6B). Notably, in the absence of Yme1p, both WT and ts Psd1p precursor accumulated at every temperature (Fig. 6B, β-α band detected by both antibodies), and the amount of WT α and β subunits detected at 37°C was increased. The absence of Yme1p modestly rescued the autocatalytic defect of Psd1ts at 30°C but not 37°C; at 37°C, three smaller COOH terminus-containing Psd1ts fragments accumulated (f1, f2, and f3). Deletion of Yme1p did not change the thermosensitivity of Psd1ts (Fig. 6C) or its assembly (Fig. 6D to F). These results suggest a role for Yme1p in Psd1p biogenesis; indicate that at elevated temperatures a fraction of WT Psd1p is degraded by Yme1p; and demonstrate that several Psd1ts fragments are produced independently of Yme1p, which is responsible for their normal removal.
Three COOH terminus-containing Psd1ts fragments accumulate in the absence of Yme1p. (A) Schematic for generating strains expressing a set amount of WT or ts Psd1p in the presence or absence of Yme1p. (B) Immunoblots for Yme1p and the α and β subunits of WT and ts Psd1p in the indicated whole-cell extracts from yeasts grown at the temperatures shown; Pic1p served as a loading control. β-α, Psd1p that had not performed autocatalysis; f1, f2, and f3, COOH terminus-containing Psd1ts fragments. (C) The indicated strains precultured at 22°C were spotted onto SCD with or without 2 mM ethanolamine and incubated at 22°C, 30°C, or 37°C for 3 days. (D) Mitochondria isolated at the permissive temperature from psd1Δ psd2Δ yeasts expressing FLAG-tagged WT or ts Psd1p, or untagged WT Psd1p, in the presence or absence of Yme1p were solubilized with 1.5% (wt/vol) digitonin, separated by 6 to 16% BN-PAGE, and immunoblotted for the α subunit or β subunit. The arrowheads and red lines mark Psd1p complexes that were detected by antibodies against both the α and β subunits; the gray lines mark ts Psd1p complexes that migrated faster and were detected only with the β subunit antisera; the asterisks mark background bands identified in Δpsd1 mitochondrial extracts. (E) Following solubilization with 1.5% (wt/vol) digitonin, anti-FLAG resin was used to immunoprecipitate the FLAG-tagged α subunit, and the presence of copurified β was determined by immunoblotting; Tom70p, Atp2p, and Pic1p served as controls. SM, starting material; B, bound material; FT, nonbinding flowthrough. (F) The percentage of each subunit detected in the nonbinding flowthrough was determined as follows: FT/SM × 100, where FT is the volume of Psd1p α or β subunit detected in the flowthrough and SM is the volume associated with the corresponding starting material (means ± SEM; n = 3). Student's t test was used to compare the α and β subunits. *, P < 0.001 versus FLAG-tagged WT Psd1p (one-way ANOVA with Holm-Sidak pairwise comparisons). The migration of molecular mass markers in kilodaltons is indicated at the left of every immunoblot.
When the temperature-dependent stability of WT and ts Psd1p was assessed postautocatalysis in vivo, a striking Yme1p-dependent difference in the thermostability of the β and α subunits of Psd1ts was identified (Fig. 7A to C). In the presence of Yme1p, the β and α subunits of Psd1ts were degraded at roughly similar rates at elevated temperatures (α slightly faster than β, as previously noted). However, in its absence, while the α subunit still rapidly disappeared at higher temperatures, the Psd1ts β subunit was significantly stabilized (Fig. 7B and C). Importantly, the discordant behavior of the α and β subunits of Psd1ts without Yme1p was also observed upon in vitro incubation of isolated mitochondria at 37°C (Fig. 8A to C). Yme1p has a documented role in resolving mitochondrial protein aggregates (27). Therefore, we speculated that at elevated temperatures, the folding of Psd1ts β may be affected so that its interaction with α is compromised, causing it to adopt an aggregation-prone state. To test this idea, the α/β association and aggregation states of WT and ts Psd1p were determined in isolated mitochondria kept on ice or incubated, as indicated, at 37°C. The α/β association of WT and ts Psd1p was not noticeably affected at elevated temperatures, although less total Psd1ts α and copurified β was noted after 20 min at 37°C (Fig. 8D). In the presence or absence of Yme1p, WT Psd1p (both subunits) was almost completely extracted by digitonin even following a 1-h heat treatment (Fig. 8E to G). In contrast, whereas both ts subunits were readily solubilized by digitonin in the presence or absence of Yme1p when kept on ice, after only 1 h at 37°C, a significant fraction of Psd1ts β, but not α, accumulated in TX-100-resistant protein aggregates (Fig. 8E, arrowheads; quantified in F and G). Surprisingly, the heat-induced aggregation of Psd1ts β occurred even in the context of Yme1p. Given that the levels of the ts β subunit were decreased by ∼77% after only 2 h of growth at 37°C (Fig. 3E), this suggests that aggregation of Psd1ts β precedes, and likely provokes, its ultimate removal by Yme1p. Taken together, our results support the following model (Fig. 8H). Upon exposure to heat, Psd1ts β subunits are partially denatured so that they rapidly accumulate in protein aggregates that are eventually resolved by Yme1p. The Psd1ts β aggregates do not include the α subunit, which is instead rapidly degraded by an unidentified mitochondrial protease upon its release from β.
Yme1p is responsible for the rapid turnover of ts β but not ts α. (A) In vivo degradation assay. Whole-cell extracts were isolated at the indicated times following growth at the indicated temperatures in the presence of CHX. Samples were resolved by SDS-PAGE and immunoblotted as indicated. The migration of molecular mass markers in kilodaltons is indicated at the left. (B) The percentages of α and β subunits remaining after 24 h at 37°C for WT and ts Psd1p in the presence or absence of Yme1p were quantified. Means ± SEM; n = 4. *, P ≤ 0.05 (Student's t test). (C) The percentages of α and β subunits remaining for WT and ts Psd1p at each time point with or without Yme1p were quantified. Means ± SEM; n = 4. #, P ≤ 0.05 (Student's t test).
ts β subunit aggregates require Yme1p for their removal. (A) Schematic for assays performed in isolated mitochondria heat treated or not. (B) Mitochondria were isolated from the indicated yeast strains grown at 22°C. Following incubation for the indicated times at 37°C, mitochondria were recovered, and 50 μg was resolved by SDS-PAGE and immunoblotted as indicated. (C) The percentages of α and β subunits remaining for WT and ts Psd1p at each time point with or without Yme1p were quantified. Means ± SEM; n = 3. (D) Mitochondria incubated for the indicated times at 37°C were solubilized with 1.5% (wt/vol) digitonin, the FLAG-tagged α subunit was captured with anti-FLAG resin, and the presence of copurified β was determined by immunoblotting; Tom70p and Atp2p served as controls. SM, starting material; B, bound material; FT, nonbinding flowthrough. (E) Mitochondria (mitos) were kept on ice or incubated at 37°C for 1 h prior to performing a sequential detergent solubilization assay to detect protein aggregation. Fractions were resolved by SDS-PAGE and immunoblotted as indicated. (F and G) The percentages of WT and ts β subunit (F) and α subunit (G) in TX-100-insoluble aggregates were determined. Means ± SEM; n = 3. (H) When shifted to high temperatures, the β subunit of Psd1ts forms aggregates that are resolved by Yme1p. The α subunit is not included in the ts β subunit aggregates and is rapidly degraded by an unidentified protease. The migration of molecular mass markers in kilodaltons is indicated at the left of every immunoblot.
Evidence of a role for Yme1p in the life cycle of endogenous Psd1p.We sought to obtain evidence for the involvement of Yme1p in the stability and/or folding of endogenous Psd1p. Since the levels of WT α and β subunits detected at 37°C were increased in the absence of Yme1p (Fig. 6B), we first determined the expression of endogenous Psd1p in mitochondria isolated from WT and yme1Δ yeasts grown at 30°C or 37°C (Fig. 9A). Psd1p levels were significantly decreased in mitochondria isolated from WT yeast grown at 37°C versus 30°C (Fig. 9A and B). The same trend was observed in mitochondria isolated from yme1Δ yeast grown at 37°C versus 30°C. However, the amount of Psd1p in temperature-matched mitochondria was consistently higher in the absence of Yme1p than in its presence. Thus, when Yme1p is not expressed, Psd1p levels are aberrantly increased.
Evidence that Yme1p is important for endogenous Psd1p fidelity. (A) Mitochondria (50 μg) from WT and yme1Δ yeasts grown at the indicated temperatures were immunoblotted as shown. (B) Densitometry analyses of protein steady-state levels from mitochondrial extracts from panel A. Shown is expression relative to the WT at 30°C ± SEM; n = 6 for Psd1p and Aac2p and n = 3 for Pic1p. Significant differences were determined by Student's t test. If normality or equal variance tests failed, significant differences were instead determined by a Mann-Whitney rank sum test. (C) Mitochondria were isolated from WT and yme1Δ yeasts grown at the indicated temperatures in the absence (left) or presence (right) of CHX for 24 h, and protein aggregation was determined using the sequential detergent solubilization assay. Fractions were resolved by SDS-PAGE and immunoblotted as indicated. The migration of molecular mass markers in kilodaltons is indicated at the left in panels A and C. (D) The percentages of Psd1p β subunit, Pic1p, and Aac2p in TX-100-insoluble aggregates following growth of WT and yme1Δ yeasts at 30°C or 37°C in the absence of CHX were determined. Means ± SEM; n = 5. (E) The percentages of Psd1p β subunit, Pic1p, and Aac2p in TX-100-insoluble aggregates following growth of WT and yme1Δ yeasts for 24 h at 30°C or 37°C in the presence of CHX were determined. Means ± SEM; n = 5. (D and E) Significant differences were determined by one-way ANOVA with Holm-Sidak pairwise comparisons. If normality or equal variance tests failed, significant differences were instead determined by Kruskal-Wallis one-way ANOVA on ranks, with Tukey test pairwise comparisons. n.s., not significant.
Next, we determined the aggregation status of endogenous Psd1p in WT and yme1Δ mitochondria isolated following growth at 30°C or 37°C in the presence or absence of new protein synthesis (Fig. 9C). Very little Psd1p was detected in aggregates in mitochondria isolated from WT yeast grown at 30°C (Fig. 9C and D). In contrast, more Psd1p was aggregated when yme1Δ yeast was grown at this temperature, although the observed difference just failed to reach significance. Interestingly, when WT yeast was grown at 37°C, significantly more Psd1p, as well as additional integral inner membrane proteins, was present in aggregates. In contrast, the proportion of Psd1p detected in aggregates in yme1Δ mitochondria did not change following growth at this temperature. Combined, these results suggest that Yme1p has a routine role in monitoring the status of Psd1p (and other inner membrane proteins), that its capacity to perform this function is impaired and/or overwhelmed at elevated temperature, and that its absence shifts the mitochondrial proteostatic equilibrium under normal growth conditions to a stressed state.
In the absence of new protein synthesis, the amount of Psd1p detected in aggregates did not change in mitochondria purified from WT and yme1Δ yeasts, regardless of the incubation temperature (Fig. 9C and E). Since the proportion of Psd1p in aggregates was increased following growth of WT yeast at 37°C with continued protein synthesis (Fig. 9D), this suggests that the stability of mature Psd1p is not overtly sensitive to elevated temperature, a conclusion that is reinforced by the yme1Δ-based results. Overall, our findings support the conclusion that Yme1p is an important determinant of WT Psd1p stability. Further, they underscore the difficulty of addressing this issue in yme1Δ yeast that expresses endogenous Psd1p and highlight the value of using the ts Psd1p allele as a tool to isolate the importance of Yme1p, and perhaps additional quality control proteases, in monitoring Psd1p fidelity.
The removal of the Psd1ts precursor requires the sequential actions of two mitochondrial proteases.The three COOH terminus-containing Psd1ts fragments (f1, f2, and f3) that accumulated at 37°C in the absence of Yme1p required continued protein synthesis (Fig. 10A). This, combined with the simple fact that they by definition represent NH2-terminal truncations, indicates that they are products of the Psd1ts precursor. We initially hypothesized that as a serine protease, perhaps these fragments were generated by inappropriate Psd1ts self-proteolysis. To test this postulate, using homology-integrated clustered regularly interspaced short palindromic repeats (CRISPR)-Cas (HI-CRISPR), related isogenic yeast strains were established that express a set amount of ts, S463A (which cannot perform autocatalysis [Fig. 2B and D]), or ts-S463A Psd1p in the presence or absence of Yme1p (Fig. 10B). However, a ts mutant unable to perform autocatalysis (ts-S463A) still generated these fragments (Fig. 10C and D), indicating that a mitochondrial protease(s) is instead likely responsible for their production.
The three COOH terminus-containing Psd1ts fragments are not generated by aberrant self-proteolysis. (A) Immunoblots for Yme1p and the α and β subunits of WT and ts Psd1p in the indicated whole-cell extracts from yeasts grown at 37°C for 24 h with or without CHX; Pic1p served as a loading control. (B) Schematic for generating strains expressing a set amount of ts, S463A, or ts-S463A Psd1p in the presence or absence of Yme1p. (C) The indicated strains precultured at 22°C were spotted onto SCD with or without 2 mM ethanolamine and incubated at 22°C or 37°C for 5 days. (D) Immunoblots for Yme1p and the α and β subunits of ts, S463A, and ts-S463A Psd1p in the indicated whole-cell extracts from yeasts grown at the temperatures shown; Pic1p served as a loading control. β-α, Psd1p that had not performed autocatalysis; f1, f2, and f3, COOH terminus-containing Psd1p fragments. The migration of molecular mass markers in kilodaltons is indicated at the left of every immunoblot.
With the goal of identifying the mitochondrial protease responsible for generating the Psd1ts precursor-derived fragments, we focused on Oma1p, a stress-activated mitochondrial protease that resides in the inner membrane (29 – 32). Using HI-CRISPR, a series of related isogenic yeast strains were established that express a set amount of WT or ts Psd1p in the absence of Yme1p, Oma1p, or both (Fig. 11A). With respect to its temperature-dependent expression, maturation, and stability, WT Psd1p was insensitive to the loss of Oma1p (Fig. 11B and 12A, C, and D). Additionally, the absence of Oma1p, singly or in combination with Yme1p, did not change the thermosensitivity of Psd1ts (Fig. 11D) or its autocatalytic competency at the permissive temperature (Fig. 11C). However, when Oma1p was missing, either by itself or together with Yme1p, the ts Psd1p precursor accumulated at 37°C while two of the three COOH terminus-containing Psd1ts fragments (f1 and f3) did not (Fig. 11C, arrowheads). Interestingly, postautocatalysis, the temperature-dependent stability of both Psd1ts subunits was unaffected by deletion of Oma1p (Fig. 12B to D). In sum, these results indicate that the degradation of a Psd1ts precursor unable to undergo autocatalysis depends on the cooperative and sequential actions of two inner membrane proteases, Oma1p and Yme1p (Fig. 11E).
Oma1p and Yme1p work sequentially to degrade the Psdts precursor. (A) Schematic for generating strains expressing a set amount of WT or ts Psd1p in the presence or absence of Oma1p and/or Yme1p. (B and C) Immunoblots for Yme1p and the α and β subunits of WT Psd1p (B) or ts Psd1p (C) in whole-cell extracts from the indicated yeasts grown at the temperatures shown; Pic1p served as a loading control. The migration of molecular mass markers in kilodaltons is indicated at the left. (D) The indicated strains precultured at 22°C were spotted onto SCD with or without 2 mM ethanolamine and incubated at 22°C, 30°C, or 37°C. (E) When Psd1ts is unable to adopt an autocatalytically competent tertiary structure at restrictive temperature, Oma1p generates two COOH terminus-containing fragments that are then completely degraded by Yme1p.
Oma1p is not required for the degradation of either ts subunit postautocatalysis. (A and B) In vivo degradation assay. Whole-cell extracts from yeasts expressing WT Psd1p (A) or ts Psd1p (B) were isolated at the indicated times following growth at the indicated temperatures in the presence of CHX. Samples were resolved by SDS-PAGE and immunoblotted as indicated. The migration of molecular mass markers in kilodaltons is indicated at the left. (C) The percentages of α and β subunits remaining for WT and ts Psd1p at each time point in the absence of Oma1p or the combined absence of Oma1p and Yme1p were quantified. Means ± SEM; n = 4. #, P ≤ 0.05 (Student's t test). (D) The percentages of α and β subunits remaining after 24 h at 37°C for WT and ts Psd1p in the presence or absence of Oma1p and/or Yme1p were quantified. Means ± SEM; n = 4. *, P ≤ 0.05 (Student's t test).
Oma1p-dependent generation of f1 and f3 is temperature independent.When Yme1p is missing, the abundance of fragments f1 and f3 was increased for ts-S463A regardless of temperature (Fig. 10D). This suggests that the ts mutations negatively impact the stability of the S463A-Psd1p precursor. Moreover, their accumulation implies that this Oma1p-dependent process can occur at both permissive and nonpermissive temperatures, if in fact f1 and f3 from ts-S463A are produced downstream of Oma1p. To directly test this possibility, related isogenic yeast strains were established that express a set amount of ts, S463A, or ts-S463A Psd1p in the presence or absence of Yme1p (Fig. 10B) or Oma1p (Fig. 13A and B). In Yme1p-replete yeast, fragment f1 was not detected for S463A, and even though the amount of precursor was significantly less following growth at 37°C (Fig. 13C, arrows), none of the COOH terminus-containing fragments accumulated (Fig. 10D and 13C). When Yme1p was not expressed, fragments f1 and f3 produced from S463A accumulated at 37°C, providing evidence that they are degraded, but not produced, by Yme1p (Fig. 13C, arrowheads). In the absence of Oma1p, fragment f1 was not detected, and the amount of f3 was reduced for S463A. For ts-S463A, the absence of Yme1p increased the abundance of the ts-S463A precursor and fragments f1 and f3 regardless of temperature. In contrast, in the absence of Oma1p, fragment f1 was not detected, and the amount of f3 was reduced for ts-S463A, again at both incubation temperatures (Fig. 13B and C, arrowheads). Thus, the Oma1p-dependent generation of fragments f1 and f3 does not require thermal stress, since it can occur at both permissive and nonpermissive temperatures (Fig. 13D).
Oma1p-dependent generation of f1 and f3 can occur at 22°C. (A) Schematic for generating strains expressing a set amount of ts, S463A, or ts-S463A Psd1p in the presence or absence of Oma1p. (B) The indicated strains precultured at 22°C were spotted onto SCD with or without 2 mM ethanolamine and incubated at 22°C or 37°C for 5 days. (C) Immunoblots for Yme1p and the α and β subunits of ts, S463A, and ts-S463A Psd1p in the indicated whole-cell extracts from yeasts grown at the temperatures shown; Pic1p served as a loading control. β-α, Psd1p that has not performed autocatalysis; f1, f2, and f3, COOH terminus-containing Psd1p fragments. The migration of molecular mass markers in kilodaltons is indicated at the left. (D) The Oma1p-dependent accumulation of f1 and f3 fragments from ts-S463A Psd1p occurs at both 22°C and 37°C.
The generation of f1 and f3 depends on Oma1p and requires structural perturbation of the Psd1p precursor.Finally, we wanted to determine if Oma1p activation alone is sufficient to drive the accumulation of f1 and f3 fragments from the S463A autocatalytic mutant (Fig. 14A). Three types of stress have been shown to activate Oma1p in yeast (30): oxidative stress (hydrogen peroxide), membrane potential depolarization (carbonyl cyanide m-chlorophenylhydrazone [CCCP]), and membrane potential hyperpolarization (complex V inhibitor; oligomycin). As such, the ability of these agents to provoke the degradation of S463A, as well as WT and ts Psd1p, was determined at 22°C (S463A and ts Psd1p) or 30°C (WT Psd1p) (Fig. 14B to D). None of these Oma1p-activating agents either decreased the levels of the S463A precursor or increased the amounts of f1 and f3 fragments detected (Fig. 14D). Collectively, these results indicate that activation of Oma1p by itself is insufficient to drive the degradation of the Psd1p precursor (WT, ts, or S463A).
Oma1p activation is not sufficient to drive the accumulation of f1 and f3. (A) Can known Oma1p-activating agents result in the accumulation of f1 and f3 fragments from the autocatalytic mutant S463A Psd1p? (B to D) Immunoblots for Yme1p and the α and β subunits of WT (B), ts (C), and S463A (D) Psd1p in whole-cell extracts from yeasts grown at 22°C (ts and S463A) or 30°C (WT Psd1p) in the absence or presence of 4 mM H2O2, 45 μM CCCP, or 4 μM oligomycin for 4 h; Pic1p served as a loading control. β-α, Psd1p that had not performed autocatalysis; f1, f2, and f3, COOH terminus-containing Psd1p fragments. The migration of molecular mass markers in kilodaltons is indicated at the left. (E) Two possible models that explain the generation of f1 and f3 from misfolded Psd1p precursor. According to model 1, some fraction of Oma1p is always basally active and thus available to degrade misfolded Psd1p precursor in the absence of any overt mitochondrial stress. In model 2, the misfolded Psd1p precursor itself is the signal that activates inactive Oma1p, thus initiating its own removal.
DISCUSSION
Until recently, the LGST motif was the only structural motif known to be required for Psd1p autocatalysis. In this study, we tested the hypothesis that mutations outside the conserved LGST motif that disturb the generation of an autocatalytically competent tertiary structure would interfere with this essential self-activating event. To test this postulate, we initially focused on a previously established ts allele of PSD1 that has four missense mutations in the β subunit (24) and whose temperature-sensitive mechanism had not been established. Indeed, here, we demonstrated that the failure of the Psd1ts allele to rescue the ethanolamine auxotrophy of the psd1Δ psd2Δ strain at restrictive temperatures was due to a severe autocatalytic defect. Interestingly, the four missense mutations in Psd1ts cluster near a conserved histidine (His345) that, together with Asp210 and Ser463 of the LGST motif, we demonstrated forms a classic Ser-His-Asp catalytic triad. We thus speculate that the ts-associated mutations compromise the ability of the Psd1ts precursor to adopt a folded structure that properly juxtaposes the catalytic triad and enables autocatalysis. Our results confirm and extend conclusions derived from recent in vitro studies with membrane-anchorless Plasmodium knowlesi Psd1 (33) by demonstrating that the initial serine protease activity of Psd1p is evolutionarily conserved and occurs in its native membrane environment in vivo.
Postautocatalysis, the missense mutations destabilize Psd1ts. Even when not exposed to heat, the four missense mutations in the ts β subunit weaken its noncovalent interaction with α. Further, fully processed Psd1ts that was allowed to accumulate at the permissive temperature was rapidly degraded upon shifting to the restrictive temperature. Both in vivo and in isolated mitochondria, the stability of the α subunit of Psd1ts was more sensitive to heat than that of its β subunit. This is perhaps not surprising given its small size and the fact that it is anchored to the IMS side of the inner membrane through its noncovalent interaction with the β subunit, an association that is impaired for Psd1ts. To probe the relationship of α and β further, we artificially separated the two parts of Psd1p and expressed them in isolation or together. In the absence of only 38 amino acids that form the COOH terminus of the Psd1p precursor and which are destined to become the α subunit, Psd1β was hardly detected. This suggests that the ability of Psd1β to fold properly is dependent on its COOH terminus. Indeed, the importance of the α subunit for Psd1β folding and stability is evidenced by the fact that an IMS-directed α subunit increased the steady-state abundance of Psd1β in trans. Given their codependence, it is tempting to speculate that Psd1p function could be tightly regulated by factors that decrease the stability of the interaction between the α and β subunits. Released α subunit would be quickly removed by an as-yet-unidentified mitochondrial protease, and since it contains the essential pyruvoyl group, Psd1p activity would be abruptly turned off.
Our efforts to identify the mitochondrial protease(s) responsible for the efficient removal of Psd1ts at elevated temperatures revealed a striking difference that varied depending on the autocatalytic status of Psd1ts. Whereas postautocatalysis, Yme1p activity alone is sufficient to clear aggregated ts β subunits, preautocatalysis, Oma1p activity generates two NH2-terminally truncated fragments that are then fully degraded by Yme1p. To our knowledge, this is the first demonstration that two mitochondrial proteases function sequentially and cooperatively to degrade a terminally misfolded protein. The identity of the protease(s) working parallel to Yme1p and responsible for the rapid turnover of ts α at elevated temperatures remains unresolved. An intriguing aspect of our findings is the requirement for Oma1p in resolving the misfolded Psd1ts precursor but not fully matured ts β, given that both contain the same NH2-terminal regions. The basis for this is not immediately clear but may reflect the misfolded state obtained for an enzyme that has never properly folded versus one that has.
Oma1p was originally identified using a ts allele of the gene for Oxa1p, a polytopic inner membrane protein (34). Similar to our study, it was shown that Oma1p lacks the ability to completely degrade Oxa1ts and instead generates proteolytic fragments that are normally resolved by the matrix-facing m-AAA protease. These results were taken as evidence that Oma1p has activity overlapping that of the m-AAA protease; hence its name. However, it is possible that, similar to our results with the Psd1ts precursor, in fact Oma1p functions upstream of the m-AAA protease to promote the degradation of Oxa1ts at the nonpermissive temperature. If correct, then these results together raise the possibility that Oma1p has a general role in facilitating the disposal of proteins that are either unable to fold correctly or become misfolded because of stress.
Given that Oma1p is activated by stress (29 – 32), we initially expected that its involvement in the turnover of the Psd1ts precursor would be temperature dependent, since 37°C is deemed a stressful temperature for yeast. However, the Oma1p-dependent accumulation of f1 and f3 fragments from ts-S463A Psd1p in the absence of Yme1p was as evident at 22°C as at 37°C (Fig. 10D and 13C). Thus, the temperature-dependent activation of Oma1p is not required to explain our results. Moreover, exogenous addition of Oma1p-activating agents did not drive the accumulation of f1 and f3, or the diminution of ts-S463A (or WT and ts) precursor, at the permissive temperature. Combined, these results suggest that Oma1p detects only Psd1p that has not undergone autocatalysis and that has been structurally perturbed, as in the case of unprocessed ts Psd1p at the nonpermissive temperature or the ts-S463A variant regardless of temperature. While the accumulation of f1 and f3 is Oma1p dependent, it is unclear if Oma1p itself needs to be activated. Potentially, a small fraction of Oma1p is always basally active (Fig. 14E, model 1). According to this scenario, Oma1p can respond whenever a suitable substrate is encountered. Alternatively, perhaps the folding status of a substrate can itself trigger Oma1p activation (Fig. 14E, model 2). In this case, any treatment that perturbs the structure of a potential substrate would have the innate capacity to stimulate a protease involved in its subsequent clearance.
In mammals, OMA1 and YME1L work in parallel to regulate mitochondrial dynamics by processing long isoforms of OPA1, a dynamin-related GTPase that mediates inner membrane fusion, into short forms (35 – 40). However, where and when these proteases cleave OPA1 and the functional consequences of their processing are different. Robust OXPHOS activity stimulates inner membrane fusion that requires YME1L-mediated cleavage of OPA1 (41). In contrast, when mitochondrial function is challenged by various types of stress, inner membrane fusion is inhibited by the complete processing of long OPA1 isoforms at a different site by OMA1 (29, 35, 38, 39, 42). Intriguingly, following mitochondrial depolarization, YME1L and OMA1 reciprocally degrade each other, depending on cellular ATP levels (31, 32). Cleavage of a long OPA1 isoform is performed by YME1L or OMA1 but not by both proteases (35, 37 – 40). In contrast, we demonstrate here that Oma1p and Yme1p function sequentially to cooperatively degrade a common substrate that if left unresolved could compromise mitochondrial proteostasis. Thus, it appears that, depending on the substrate and/or cellular context, Yme1p and Oma1p can work alone, together, or antagonistically to ensure mitochondrial health.
Our work extends an emerging story of an intimate relationship between mitochondrial phospholipid metabolism and the quality control machinery therein. Mitochondrial phospholipid synthesis requires transport pathways for lipid precursors and products across the aqueous IMS and leaflets of a bilayer. Ups1p and Ups2p, each in association with Mdm35p (26, 43), contribute to the flux of PA and PS across the IMS by grabbing their lipid ligands in the outer membrane and transporting them to the inner membrane (44 – 46). In so doing, they help provide substrates for CL and PE synthesis. Interestingly, both Ups proteins have, for mitochondrial proteins, unusually short half-lives, which is explained by their constitutive degradation by Yme1p and/or Atp23p (26). Previously, we demonstrated that Yme1p has a critical role in degrading certain Barth syndrome-associated mutants of the monolyso-CL transacylase tafazzin that retain some enzymatic activity but are prone to aggregation (47). In the present study, we have documented a role for Yme1p in the biogenesis of Psd1p (the Psd1p precursor accumulates in its absence), the turnover of an aggregation-prone allele of the β subunit postautocatalysis, and the stability of endogenously expressed WT Psd1p. Further, we demonstrated that two proteases, Oma1p and Yme1p, work sequentially to remove a Psd1ts precursor that is unable to fully mature. A common feature of these lipid-related proteins is that they each work in or at the boundary between the hydrophilic surface and hydrophobic core of membranes, a denaturing environment. As such, we posit that the structurally destabilizing environment in which mitochondrial lipid-metabolizing and transport proteins work has provoked the mitochondrial quality control machinery as a whole to evolve multiple means to monitor their fidelity.
Very recently, a novel tumor suppressor, LACTB, was discovered; when overexpressed in certain cancer cell lines, it reduces cell proliferation and increases cellular differentiation via a mechanism that is at least in part explained by a significant decrease in the levels of PISD, the mammalian Psd1p equivalent (48). Importantly, the decrease in PISD abundance is posttranscriptionally mediated and results in significant reductions in certain PE and lyso-PE species. However, the actual mechanism of PISD turnover was not determined. These findings clearly establish that the posttranscriptional regulation of PISD is clinically relevant. Moving forward, it will be interesting to determine if some of the proteolytic pathways that we have identified in yeast using ts Psd1p are harnessed in certain physiological and/or pathophysiological contexts to regulate mammalian PISD levels and, by extension, the phenotype of cells.
MATERIALS AND METHODS
Yeast strains and growth conditions.All the yeast strains used in this study were derived from GA74-1A (MAT a his3-11,15 leu2 ura3 trp1 ade8 [rho + mit +]). The psd1Δ (MAT a trp1 leu2 ura3 ade8 psd1Δ::HISMX6), psd1Δ psd2Δ (MAT a leu2 ura3 ade8 psd1Δ::TRP1 psd2Δ::HISMX6), and yme1Δ (MAT a trp1 leu2 ura3 ade8 yme1Δ::HISMX6) strains have been described previously (23, 49). The psd1Δ psd2Δ yme1Δ and psd1Δ psd2Δ oma1Δ strains were generated by taking psd1Δ psd2Δ or psd1Δ psd2Δ with WT, ts, S463A, or ts-S463A PSD1 integrated into the Leu2 locus and using a CRISPR-Cas9 gene block targeted against YME1 or OMA1. Specifically, yme1 and oma1 knockout were achieved using the HI-CRISPR system (50). The plasmid pCRCT was a gift from Huimin Zhao (Addgene plasmid 60621). The CRISPR-Cas9 targets for YME1 and OMA1 were selected using the Yeastriction (yeastriction.tnw.tudelft.nl) (51) and Benchling online CRISPR guide design tools. The CRISPR construct was designed to recognize residues 12 to 31 on the forward strand of YME1 and residues 118 to 137 on the reverse strand of OMA1 (position 1 is the adenine of the AUG site). A 100-bp homology repair template was designed to have 50-bp homology arms on both sides flanking the Cas9 cutting site and incorporated an 8-bp deletion to induce a frameshift mutation, as described previously (50). psd1Δ psd2Δ yme1Δ oma1Δ strains were generated from the corresponding psd1Δ psd2Δ yme1Δ strains. The spacer sequences (including the CRISPR-Cas9 target and the homology repair template) were ordered as gBlocks (Integrated DNA Technologies) and assembled into the pCRCT plasmid using the Golden Gate assembly method (52). OMA1 disruption was confirmed by DNA sequencing.
Yeast cells were grown in either rich lactate (1% yeast extract, 2% tryptone, 0.05% dextrose, 2% lactic acid, 3.4 mM CaCl2-2H2O, 8.5 mM NaCl, 2.95 mM MgCl2-6H2O, 7.35 mM KH2P04, 18.7 mM NH4Cl, pH 5.5) or YPD (1% yeast extract, 2% peptone, 2% dextrose) medium. To assess the functions of the assorted Psd1p constructs and mutants, overnight cultures grown in YPD medium were spotted on synthetic complete (SC) dextrose plates in the absence or presence of 2 mM ethanolamine and grown at 30°C or the indicated temperature. Steady-state labeling of mitochondrial phospholipids was achieved by growing yeast for 24 h at the indicated temperature in synthetic complete dextrose medium supplemented with 2 mM choline and 2.5 μCi/ml 32Pi. To activate Oma1p, starter cultures grown in YPD medium at the permissive temperature were used to inoculate fresh tubes with 1 optical density at 600 nm (OD600) unit/2 ml YPD, which were spiked with nothing, 4 mM H2O2, 45 μM carbonyl cyanide m-chlorophenyl hydrazine, or 4 μM oligomycin and then incubated for 4 h at 22°C (ts and S463A Psd1p) or 30°C (WT Psd1p) with shaking at 220 rpm prior to yeast protein extraction.
Psd1p with a COOH-terminal 3× FLAG tag subcloned into pRS315 and pRS305 or harboring 6 COOH-terminal methionines and subcloned into pSP64 has been described previously (23). The 6 methionine residues added to the COOH terminus allow detection of the α subunit postautocatalysis in in vitro experiments, while the addition of the 3× FLAG tag enables detection of the α subunit postautocatalysis in vivo. PSD1 point mutations were generated by overlap extension (53) using pRS305Psd3XFLAG as the template. The temperature-sensitive PSD1 allele (K356R, F397L, E429G, and M448T) was described previously (24) and was assembled here in the context of pRS305Psd3XFLAG by overlap extension. Psd1β was generated by introducing a stop codon immediately after Gly462 and cloned into pRS315. To target the α subunit to the IMS, α-3× FLAG (starting at S463) was placed downstream of the first 64 amino acids of cytochrome c 1 but still under the control of the PSD1 promoter by overlap extension and cloned into pRS316. IMS-α was subcloned into pCM189 to allow doxycycline-based repression. For inducible expression of IMS-α, starter cultures grown in SC medium without Leu and Ura but containing ethanolamine (0.17% yeast nitrogen base, 0.5% ammonium sulfate, 0.2% dropout mixture complete, 2 mM ethanolamine, 2% dextrose) were used to inoculate fresh SC medium without Leu and Ura but containing ethanolamine cultures, with or without 10 μg/ml doxycycline, which were grown overnight with shaking at 220 rpm at 30°C.
In organello import.Radiolabeled precursors were produced using an SP6 quick coupled transcription/translation system (Promega) spiked with 35S Easy-Tag (PerkinElmer). Import of radiolabeled precursors utilized mitochondria isolated from D273-10B grown in rich lactate and was performed as described previously (23, 47). Where indicated, valinomycin (1 μM) and carbonyl cyanide m-chlorophenyl hydrazine (5 μM) were added 5 min prior to precursor addition to collapse the mitochondrial proton motive force. Upon addition of radiolabeled precursor, the import reaction mixtures were incubated at the indicated temperatures. At the indicated times, import was stopped with an equal volume of ice-cold BB7.4 (0.6 M sorbitol and 20 mM HEPES-KOH, pH 7.4), and mitochondria were reisolated by spinning at 21,000 × g for 5 min at 4°C. One hundred percent of each time point (i.e., the entire sample) and 5% of imported precursors were resolved on 15% SDS-PAGE gels and analyzed by phosphorimaging.
In vivo degradation experiments. In vivo degradation experiments were performed as previously described (47). In brief, following growth at the permissive temperature, fresh tubes were inoculated with 6 OD600 units/3 ml YPD and incubated for 5 min at the permissive temperature. Cycloheximide was added to a final concentration of 200 μg/ml to inhibit cytosolic protein synthesis, and the cultures were incubated at the indicated temperature with shaking at 220 rpm. At each time point, 1 OD600 equivalent was transferred to a tube containing an equal volume of ice-cold 2× azide mixture (20 mM NaN3 and 0.5 mg/ml bovine serum albumin [BSA]). Cells were collected (845 × g for 10 min at 4°C), the supernatant was removed, and the pellets were stored at −80°C until all the time points were harvested. Proteins were extracted, and SDS-PAGE and immunoblotting were performed as previously described (54, 55).
In organello heat shock assay.Mitochondria isolated from yeast grown at the permissive temperature were incubated at the indicated temperature for the indicated times. Mitochondria were harvested by spinning at 21,000 × g for 5 min at 4°C and either further analyzed or resuspended in reducing sample buffer, boiled at 100°C for 5 min, resolved on custom-made 10 to 16% SDS-PAGE gels, and analyzed by immunoblotting.
Immunoprecipitation.Mitochondria (0.2 mg) were solubilized (2.5 mg/ml) for 30 min on ice with 20 mM HEPES-KOH, pH 7.4, 20 mM imidazole, 10% glycerol, 100 mM NaCl, and 1 mM CaCl2 supplemented with 1.5% (wt/vol) digitonin (Biosynth International, Inc.) and protease inhibitors (1 mM phenylmethylsulfonyl fluoride [PMSF],10 μM leupeptin, and 2 μM pepstatin A). Clarified extracts (21,000 × g for 30 min at 4°C) were diluted to 0.24% digitonin with lysis buffer base and incubated with 15 μl anti-FLAG affinity gel (Genscript) with rotation at 4°C for 2 h. Following a low-speed spin, 0.1 mg of unbound material was transferred to a tube, trichloroacetic acid (TCA) precipitated, and processed for SDS-PAGE. The bound material was subjected to three 10-min washes (1st, 0.1% digitonin wash buffer; 2nd, 0.1% digitonin wash buffer with 250 mM NaCl; 3rd, 0.1% digitonin no salt) and processed for SDS-PAGE.
Antibodies.Most of the antibodies used in this study were generated in our laboratory or in the laboratory of J. Schatz (University of Basel, Basel, Switzerland) or C. Koehler (UCLA) and have been described previously (13, 23, 47, 49, 56). Other antibodies used were mouse anti-FLAG (clone M2; Sigma) and horseradish peroxidase-conjugated (Fig. 2, 4, 5, 6D and E, and 8D) and fluorescence-conjugated (Fig. 1C, 3B and F, 6B, 7A, 8B and E, 9C, 10A and D, 11B and C, 12A and D, 13C, and 14B to D) secondary antibodies (Pierce).
Miscellaneous.Isolation of mitochondria, one-dimensional (1D) BN-PAGE, detergent-based aggregation, phospholipid analyses, and immunoblotting were performed as previously described (13, 23, 47, 49, 55 – 57). Statistical comparisons were performed by t test or one-way analysis of variance (ANOVA) with Holm-Sidak pairwise comparison using SigmaPlot 11 software (Systat Software, San Jose, CA); P values of ≤0.05 were deemed significant. All the graphs show means ± standard errors of the mean (SEM).
ACKNOWLEDGMENTS
We thank Jeff Schatz and Carla Koehler for antibodies, Sin Urban (JHMI) for the suggestion that the serine of the LGST motif may be part of a catalytic triad, and Pingdewinde N. Sam and James O. Owusu (Department of Physiology, Johns Hopkins University School of Medicine) for technical assistance.
This work was supported by a National Institutes of Health grant (R01GM111548) to S.M.C.; a Biochemistry, Cellular, and Molecular Biology Program training grant (T32GM007445) to O.O. and E.C.; and a predoctoral fellowship from the American Heart Association (15PRE24480066) to O.B.O.
We declare no competing financial interests.
O.B.O., O.O., and S.M.C. designed research; O.B.O., O.O., E.C., and S.M.C. performed research and analyzed data; and O.B.O., O.O., and S.M.C. wrote the paper.
FOOTNOTES
- Received 3 February 2017.
- Returned for modification 20 February 2017.
- Accepted 8 June 2017.
- Accepted manuscript posted online 12 June 2017.
- Copyright © 2017 American Society for Microbiology.
