Oma1, a Novel Membrane-bound Metallopeptidase in Mitochondria with Activities Overlapping with the m -AAA Protease*

The integrity of the inner membrane of mitochondria is maintained by a membrane-embedded quality control system that ensures the removal of misfolded membrane proteins. Two ATP-dependent AAA proteases with catalytic sites at opposite membrane surfaces are key components of this proteolytic system. Here we describe the identification of a novel conserved metallopeptidase that exerts activities overlapping with the m -AAA protease and was therefore termed Oma1. Both peptidases are integral parts of the inner membrane and mediate the proteolytic breakdown of a misfolded derivative of the polytopic inner membrane protein Oxa1. The m -AAA protease cleaves off the matrix-exposed C-terminal domain of Oxa1 and processively degrades its transmembrane domain. In the absence of the m -AAA protease, proteolysis of Oxa1 is mediated in an ATP-independent manner by Oma1 and a yet unknown peptidase resulting in the accumulation of N- and C-terminal proteolytic fragments. Oma1 exposes its proteolytic center to the matrix side; however, mapping of Oma1 cleavage sites reveals clipping of Oxa1 in loop regions at both membrane surfaces. These results identify Oma1 as a novel component of the quality control system in the inner membrane of mitochondria. Proteins homologous to Oma1 are present in higher eukaryotic cells, eubacteria and archaebacteria, suggesting that Oma1 is the found-ing member (open reading frame YKR087c) were identified by FastA (36) in the TrEMBL (cid:5) Swiss Protein Database. These were compiled by PileUp (GCG Wisconsin Package, version 10.2 (Accelrys Inc.)) in a multisequence alignment. Those sequences that are homologous to the region surrounding the HE XX H region of Oma1 were included in a second multisequence alignment by ClustalW (37). This resulted in the iden- tification of the conserved proteolytic domain. Prediction of transmembrane regions was done by TopPred (38) and “DAS” (39). Antibody Production— Degradation of endogenous Oxa1 ts in mitochondria was monitored using antisera directed against a C-terminal peptide of Oxa1 (40) and an N-terminal peptide of Oxa1. For generation of the latter, the peptide CSIDELTSSAPSLSASTSD-amide (corre- sponding to amino acid residues 61–78 of the preprotein) was coupled with maleimide-activated carrier protein (Imject, Pierce) to keyhole limpet hemocyanin and used for generation of antibodies in rabbits. a clarifying spin, mitochondrial extracts were fractionated Superose 6 size exclusion chromatography. Eluate trichloroacetic acid-precipitated and analyzed by SDS-PAGE and immunoblotting using a Myc-specific antiserum.

The majority of mitochondrial proteins is nuclear encoded and synthesized at cytosolic ribosomes. Import into mitochondria is mediated by various protein translocases in the outer and inner membrane that allow the passage of preproteins only in a largely unfolded conformation (1,2). Folding and assembly of mitochondrial proteins must therefore occur inside mitochondria. Little is known about the efficiency of this process, but it is clear that mitochondria, as other organelles, harbor a quality control system that ensures the recognition and removal of non-native polypeptides, preventing their potentially harmful accumulation within the organelle (3). Notably, a functional impairment of components of this system leads to neurodegeneration in various forms of hereditary spastic paraplegia, illustrating the importance of protein quality control for mitochondrial function (4,5).
Molecular chaperone proteins and ATP-dependent proteases present in different subcompartments of mitochondria maintain protein quality control within the organelle (3). In line with the endosymbiotic origin of mitochondria, many of these components exhibit significant homology to bacterial proteins. ATP-dependent proteases homologous to Lon proteases (6 -8) and, at least in some organisms, Clp proteases (9,10) have been identified in the mitochondrial matrix space, whereas the inner membrane harbors two AAA proteases homologous to bacterial FtsH proteases (11). These membrane-embedded peptidases were termed m-and i-AAA proteases to indicate their different topology in the inner membrane; the m-AAA protease is active on the matrix and the i-AAA protease on the intermembrane side of the membrane. Both proteases consist of homologous subunits conserved in all eukaryotic cells. The m-AAA protease is a hetero-oligomeric complex of Yta10 and Yta12 subunits in yeast (12) and of AFG3L2 and paraplegin subunits in human (13), whereas Yme1 subunits most likely form homo-oligomeric i-AAA proteases in mitochondria of both organisms (14 -16). Protein degradation by ATP-dependent proteases results in the formation of peptides composed of ϳ5-20 amino acid residues (17)(18)(19)(20). However, a complete proteolytic breakdown of mitochondrial proteins to amino acid residues has already been observed in early studies pointing to the presence of oligopeptidases in mitochondria (21). Until now only two ATP-independent mitochondrial oligopeptidases have been identified, a thimet oligopeptidase localized in the intermembrane space of yeast and human mitochondria (22,23) and the presequencedegrading metallopeptidase PreP in plant mitochondria (24). Other known peptidases act as processing enzymes cleaving off mitochondrial targeting sequences from newly imported preproteins (25). These include the mitochondrial processing peptidase and the intermediate peptidase in the matrix and, in the inner membrane, the inner membrane peptidase and the rhomboid peptidase Pcp1 (also termed Rbd1 or Mdm37), which has been linked to the processing of cytochrome c peroxidase (26,27). Notably, Pcp1 also mediates the processing of the dynamin-like mitochondrial GTPase Mgm1 and thereby controls mitochondrial morphogenesis (27,28).
Here we have analyzed the degradation of a misfolded variant of the polytopic membrane protein Oxa1, and we describe the identification of Oma1, a novel metallopeptidase in the inner membrane of mitochondria. Similar to the m-AAA protease, Oma1 is capable of degrading non-native membrane proteins and therefore represents a novel component of the quality control system that is conserved in higher eukaryotes, eubacteria and archaebacteria.

EXPERIMENTAL PROCEDURES
Yeast Strains and Cloning Procedures-Yeast strains used in this study are derivatives of W303. The ⌬yta10 (YGS101 (29)), ⌬yme1 (15), and oxa1 ts (30) mutant strains were described previously. Cells were grown on YP medium containing 2% galactose and 0.5% lactate at 30°C or, in case of temperature-sensitive oxa1 mutant strains, at 24°C. YTA10, YME1, and OMA1 were disrupted by PCR-targeted homolo-gous recombination using the heterologous marker cassettes HIS3MX6 or KanMX4, respectively (31). The complete open reading frame of OMA1 and an internal fragment of YTA10 (bp 150 -2170) were replaced by the disruption cassettes. Homologous recombination was verified by PCR.
Expression of a C-terminally Myc-tagged variant of Oma1 was achieved by modification of the genomic OMA1 gene by PCR-targeted homologous recombination (32). The primers 5Ј-CAA TCT GAT TGT  AGC AGT ATG GGT AAC TAT TAT AAA AGT TTT TTC TCA ATG  CGG ATC CCC GGG TTA ATT AA-3Ј and 5Ј-GGG TTA TTT ATT GGG  TAC AAA AGA AAA GAG CAT AAC TCG TGG AGT GCG AAT TCG  AGC TCG TTT AAA C-3Ј were used for amplification of the disruption cassette using pFA6a-Myc-kanMX4 as template, which was subsequently transformed into wild type cells. Homologous recombination was verified by PCR.
For in vitro synthesis of Oma1, the OMA1 gene was amplified by PCR using the primer pair 5Ј-GGG CTG CAG ATG TTC CGC TAT GAC AAT GGT-3Ј (5Ј-primer) and 5Ј-GGG GGA TCC TTA CAT TGA GAA AAA ACT TTT-3Ј (3Ј-primer). The PCR product was cloned into the pCRII-TOPO-vector (Promega) to allow in vitro transcription by the SP6 RNA polymerase.
Site-directed Mutagenesis of OMA1-Point mutations were introduced into OMA1 using a PCR-based site-directed mutagenesis kit (QuikChange; Stratagene) according to the instructions from the provider. OMA1 cloned into the yeast expression vector YCplac111 under the control of the ADH1 promoter was used as a template in the PCR. Glutamic acid residue 173 was replaced by glutamine by using the primer pair 5Ј-GATGGGATTGCCACTGTTTTAGCACACCAATTTGC-TCATCAGCTAGCAAGAC-3Ј (forward primer) and 5Ј-GTCTTGCTAG-CTGATGAGCAAATTGGTGTGCTAAAACAGTGGCAATCCCATC-3Ј (reverse primer); the histidine residue 176 was replaced by tyrosine by using the primer pair 5Ј-GCCACTGTTTTAGCACACGAATTTGCTTA-TCAGCTAGCAAGACACACAGTGGC-3Ј (forward primer) and 5Ј-CGG-CTGTGTGTGTCTTGCTAGCTGATAACAAATTCGTGTGCTAAAACA-GTGGC-3Ј (reverse primer); and histidine residue 181 was replaced by tyrosine by using the primer pair 5Ј-GCTCATCAGCTAGCAAGATAC-ACAGCCGAAAATTTGTCGAAGGCTCC-3Ј(forward primer) and 5Ј-GGAGCGTTCGACAAATTTTCGGCTGTGTATCTTGCTAGCTGATGA-GC-3Ј (reverse primer). Nucleotide exchanges were verified by DNA sequencing.
Protein Import into Isolated Mitochondria-Mitochondria were isolated as described previously (33,34). After in vitro transcription of the genes using SP6 polymerase, precursor proteins were synthesized in reticulocyte lysate (Promega) in the presence of [ 35 S]methionine and imported into mitochondria for 15 min at 25°C as described (35). Mitochondria harboring the temperature-sensitive Oxa1 mutant protein were incubated for 15 min at 37°C prior to import to induce the phenotype. Non-imported preproteins were digested by the addition of 100 g/ml trypsin (for 20 min at 4°C). To inhibit the protease, a 20-fold molar excess of soybean trypsin inhibitor was added. Samples were analyzed by SDS-PAGE and autoradiography.
To assess the proteolytic breakdown of endogenous Oxa1 ts , mitochondria (30 g) were incubated in import buffer containing ATP and the ATP-regenerating system at 25 or 37°C. Proteolysis was halted at the indicated times by cooling the samples on ice. Mitochondria were reisolated, and mitochondrial proteins were analyzed by SDS-PAGE and immunoblotting.
Subfractionation of Mitochondria-Mitoplasts were generated by hypotonic swelling of mitochondria. Mitochondria (30 g) were suspended at a concentration of 0.1 mg/ml in 20 mM HEPES/KOH, pH 7.2, supplemented with import buffer (10%). Proteinase K (50 g/ml) was added where indicated. Samples were then incubated for 30 min at 4°C. The protease treatment was halted by adding PMSF 1 (2 mM). Control mitochondria (non-swelling conditions) were diluted to the same extent in SHKCl buffer (0.6 M sorbitol, 80 mM KCl, 20 mM HEPES/KOH, pH 7.2). Mitochondria and mitoplasts were re-isolated by centrifugation, washed once with SHKCl buffer supplemented with PMSF (1 mM), and lysed in SDS sample buffer. Samples were analyzed by SDS-PAGE and immunoblotting. The efficiency of swelling was assessed by immunodecoration with antisera against cytochrome b 2 (soluble intermembrane space protein), the ADP/ATP carrier (inner membrane protein), and Mge1 (soluble matrix protein).
For alkaline extraction of mitochondrial membranes, mitochondria were suspended at a concentration of 0.1 mg/ml in 0.1 M Na 2 CO 3 , pH 10.5, and 1 mM PMSF. After incubation for 30 min at 4°C, soluble and insoluble proteins were separated by centrifugation for 60 min at 220,000 ϫ g. Soluble proteins were trichloroacetic acid-precipitated, and both fractions were analyzed by SDS-PAGE and immunoblotting. The fractionation was tested by immunodetection of marker proteins as described above.
Gel Filtration Analysis-Mitochondria expressing a Myc-tagged variant of Oma1 (600 g) were solubilized during an incubation for 15 min at 4°C in 1% (w/v) Triton X-100, 150 mM potassium acetate, 4 mM magnesium acetate, 0.5 mM PMSF, 30 mM Tris/HCl, pH 7.4, at a concentration of 5 mg/ml. After a clarifying spin for 15 min at 30,000 ϫ g, mitochondrial extracts were loaded onto a Superose 6 gel filtration column (Amersham Biosciences). Eluate fractions were collected, trichloroacetic acid-precipitated, and analyzed by SDS-PAGE and immunoblotting by using an antiserum against the c-Myc epitope.
Sequence Alignments-Sequences homologous to yeast Oma1 (open reading frame YKR087c) were identified by FastA (36) in the TrEMBL ϩ Swiss Protein Database. These were compiled by PileUp (GCG Wisconsin Package, version 10.2 (Accelrys Inc.)) in a multisequence alignment. Those sequences that are homologous to the region surrounding the HEXXH region of Oma1 were included in a second multisequence alignment by ClustalW (37). This resulted in the identification of the conserved proteolytic domain. Prediction of transmembrane regions was done by TopPred (38) and "DAS" (39).
Antibody Production-Degradation of endogenous Oxa1 ts in mitochondria was monitored using antisera directed against a C-terminal peptide of Oxa1 (40) and an N-terminal peptide of Oxa1. For generation of the latter, the peptide CSIDELTSSAPSLSASTSD-amide (corresponding to amino acid residues 61-78 of the preprotein) was coupled with maleimide-activated carrier protein (Imject, Pierce) to keyhole limpet hemocyanin and used for generation of antibodies in rabbits.

Proteolysis of a Mutant Variant of Oxa1
Causing a Temperature-sensitive Growth Phenotype-Oxa1 is a polytopic mitochondrial membrane protein with an N-terminal domain facing the intermembrane space and a C-terminal domain exposed to the matrix space ( Fig. 1A) (40 -42). A temperature-sensitive mutant allele of OXA1 has been identified (43), which causes an exchange of leucine 240, located in a loop region in the intermembrane space, to serine (44). Western blot analysis of mitochondria isolated from an oxa1 ts mutant strain grown at permissive temperature revealed reduced steady state levels of Oxa1 when compared with wild type mitochondria (Fig. 1B). We therefore examined the stability of mutant Oxa1 and incubated isolated mitochondria at 37°C (Fig. 1B). Endogenous Oxa1 ts was degraded within mitochondria with a half-life of about 10 min, whereas Oxa1 was stable under these conditions (Fig. 1B). In subsequent experiments, both proteins were synthesized in rabbit reticulocyte lysate in the presence of [ 35 S]methionine and imported post-translationally into isolated wild type mitochondria at permissive temperature. We observed proteolysis of Oxa1 ts but not of Oxa1 when mitochondria were further incubated at 37°C after completion of import (Fig. 1C).
Oxa1 ts retains functional activity under permissive conditions (43). It is therefore conceivable that replacement of leucine 240 by serine destabilizes Oxa1 causing its rapid degradation under non-permissive conditions. Alternatively, the mutant protein might be mislocalized within mitochondria due to an impaired insertion of Oxa1 ts into the inner membrane. To test the validity of either possibility, mitochondria isolated from wild type and the oxa1 ts strain were subfractionated, and the intramitochondrial localization of mutant and wild type Oxa1 was determined. Both proteins were accessible to externally added proteinase K after osmotic disruption of the outer membrane (Fig. 1D). In agreement with previous observations (40), C-terminal fragments of Oxa1 with a molecular mass of about 27 kDa were generated under these conditions by proteolytic removal of the N-terminal domain of Oxa1 in the intermembrane space. In contrast, a C-terminal fragment with a molecular mass of about 19 kDa accumulated in mitoplasts of oxa1 ts mitochondria under these conditions ( Fig. 1D and Fig.  6). The L240S mutation apparently resulted in a destabilization of the loop region between transmembrane regions 2 and 3, making it accessible for proteolytic degradation by proteinase K. Oxa1 and Oxa1 ts as well as the C-terminal fragments generated by proteinase K treatment of mitoplasts were exclusively recovered from the membrane fraction upon alkaline extraction (Fig. 1D). Although its exact membrane topology remains to be determined, these results demonstrate the insertion of Oxa1 ts into the inner membrane. We therefore conclude that local misfolding of the mutant protein rather than its impaired insertion into the inner membrane causes the lability of Oxa1 ts .
Degradation of Oxa1 ts by the m-AAA Protease but Not the i-AAA Protease-In order to identify proteases involved in the degradation of Oxa1 ts , the stability of the mutant protein was analyzed after import into ⌬yta10 and ⌬yme1 mitochondria lacking the m-AAA protease or the i-AAA protease, respectively. Deletion of YME1 did not affect the proteolytic breakdown of Oxa1 ts ( Fig. 2A), even though mutating leucine 240 to serine was found to cause misfolding of an intermembrane space loop of Oxa1 (Fig. 1D). Degradation of full-length Oxa1 ts was impaired, however, in mitochondria lacking the m-AAA protease subunit Yta10 ( Fig. 2A). Proteolytic fragments that were not detectable in wild type mitochondria accumulated in these mitochondria ( Fig. 2A). Thus, the m-AAA protease is involved in but is not essential for the degradation of newly imported Oxa1 ts .
We generated oxa1 ts mutant strains lacking either the m-or the i-AAA protease to monitor the role of these proteases during degradation of Oxa1 ts in vivo. Mitochondria were isolated from these strains, incubated at non-permissive temperatures where indicated, and subsequently analyzed by immunoblotting with polyclonal antisera directed against N-or C-terminal peptides of Oxa1 (Fig. 2B). Whereas no N-terminal fragments were detected, a C-terminal fragment of about 14 kDa (C14) corresponding to the matrix-exposed domain of Oxa1 accumulated in oxa1 ts mitochondria (Fig. 2B). This fragment was generated by the m-AAA protease because it was not detected in oxa1 ts ⌬yta10 mitochondria (Fig. 2B). In agreement with the import studies ( Fig. 2A), several other proteolytic fragments of Oxa1 ts accumulated in these mitochondria. These include the N-terminal fragments N22, N25, and N27 (the numbers indicate the estimated molecular weights) and the C-terminal fragments C15, C16, C17, and C20 ( Fig. 2B and Fig. 6). Deletion of the i-AAA protease did not impair the degradation of Oxa1 ts in vivo. We only observed the accumulation of the proteolytic fragment C14 as in oxa1 ts mitochondria (Fig. 2B).
These findings assign a crucial role to the m-AAA protease for the degradation of Oxa1 ts in vitro as well as in vivo and provide evidence for the involvement of yet another protease in the proteolytic breakdown of this polytopic membrane protein.
The i-AAA protease is known to exert an overlapping substrate specificity to the m-AAA protease (45). Therefore, it cannot be excluded that the i-AAA protease, although not essential for proteolysis, generates proteolytic fragments of Oxa1 ts in the absence of the m-AAA protease. The evaluation of this hypothesis is made difficult by the synthetic lethality of mutations in both AAA proteases (45,46). We used biochemical means to characterize the generation of the proteolytic fragments of and Oxa1 ts in mitochondria. Energized mitochondria isolated from wild type (WT) or oxa1 ts cells were incubated at 37°C. At the indicated times, aliquots were withdrawn and analyzed by SDS-PAGE and immunoblotting using an antiserum directed against a C-terminal peptide of Oxa1. C, stability of newly imported Oxa1 and Oxa1 ts in wild type mitochondria. 35 S-Labeled Oxa1 and Oxa1 ts were imported at 25°C into wild type mitochondria. Shifting the samples to 37°C induced proteolysis. At different times, the amount of imported protein was determined by SDS-PAGE and autoradiography. Virtually identical results were obtained upon import of Oxa1 and Oxa1 ts in oxa1 ts mutant mitochondria (data not shown). D, submitochondrial localization of Oxa1 and Oxa1 ts . Mitochondria isolated from wild type (WT) and oxa1 ts cells grown under permissive conditions were fractionated by osmotic disruption of the outer membrane (swelling) and by alkaline extraction (Na 2 CO 3 ) and treated with proteinase K (PK; 50 g/ml) when indicated. Samples were analyzed by SDS-PAGE and immunoblotting using an antiserum directed against a C-terminal peptide of Oxa1. f27 and f19, C-terminal fragments of Oxa1 with a molecular mass of ϳ27 and 19 kDa. The appearance of the Oxa1 fragment marked with an asterisk varied depending on the mitochondrial preparation used. T, total mitochondria; P, pellet fraction; S, supernatant fraction. newly imported Oxa1 ts in oxa1 ts ⌬yta10 mitochondria (Fig. 3). Although affecting the accumulation of C20, depletion of mitochondria of ATP did not significantly affect the formation of N22 (Fig. 3A). For further characterization, the requirement for divalent metal ions for the proteolytic breakdown of newly imported Oxa1 ts in oxa1 ts ⌬yta10 mitochondria was examined (Fig. 3B). Mitochondria were depleted of divalent metal ions by adding the chelating agents EDTA and o-phenanthroline, the latter being membrane-permeable. Formation of fragments N22 and C20 was strongly reduced under these conditions (Fig.  3B). Notably, N22 and C20 accumulated if only EDTA was added to the reaction suggesting that divalent metal ions are required in the mitochondrial matrix space (Fig. 3B). An involvement of the ATP-dependent i-AAA protease which exposes its metal-binding site to the intermembrane space can therefore be excluded. Thus, these experiments suggest that a novel metallopeptidase can degrade Oxa1 ts which exposes its catalytic site(s) to the matrix space.
Genome-wide Screening for Mitochondrial Metallopeptidases in Silico-In order to identify this proteolytic activity, we screened the genome sequence of Saccharomyces cerevisiae for open reading frames encoding putative metallopeptidases with consensus metal-binding motifs HEXXH or HXXEH. The pattern search led to the identification of 180 and 93 putative proteins containing the HEXXH and the HXXEH motif, respectively. Proteins with known functions in cellular compartments other than mitochondria were not further considered. In the next step, the probability of localization in mitochondria was evaluated using the PSORTII algorithm (47). As mitochondrial inner membrane proteins often lack a cleavable mitochondrial presequence, making the prediction of the cellular localization more difficult, we also examined the remaining proteins for the existence of hydrophobic segments that might serve as transmembrane regions. This procedure led to the identification of several genes coding for putative mitochondrial metallopeptidases, including both soluble as well as membrane-bound proteins.
OMA1  35 S-Labeled Oxa1 ts was imported in wild type (WT), ⌬yta10, or ⌬yme1 mitochondria. After removal of non-imported preproteins by trypsin treatment, proteolysis at 37°C was monitored by SDS-PAGE and autoradiography as in Fig. 1. B, proteolysis of endogenous Oxa1 ts . Energized mitochondria isolated from oxa1 ts , oxa1 ts ⌬yta10, and oxa1 ts ⌬yme1 cells were incubated for 30 min at 37°C. Samples were analyzed by SDS-PAGE and immunoblotting using antisera directed against a C-terminal (␣C) or an N-terminal (␣N) peptide of Oxa1. Predominant N-and C-terminal proteolytic fragments of Oxa1 were marked with N or C, respectively, and a number indicating the apparent molecular mass in kDa. N25, C16, and C17 represent additional minor proteolytic products not labeled in the figure. Neither Oxa1-specific antisera showed unspecific cross-reactivity in ⌬oxa1 mitochondria (data not shown). The appearance of the Oxa1 fragment marked with an asterisk varied depending on the mitochondrial preparation used.

FIG. 3.
Characterization of the proteolytic activity mediating degradation of Oxa1 ts in the absence of the m-AAA protease. A, ATP dependence. Proteolysis of radiolabeled, newly imported Oxa1 ts was analyzed in oxa1 ts ⌬yta10 mitochondria in the presence (ϩATP) or absence (ϪATP) of ATP and an ATP-regenerating system. To reduce internal ATP levels, mitochondria were incubated for 5 min at 25°C prior to import with apyrase (1 unit/l)(ϪATP). The formation of the proteolytic fragments N22 and C20 after incubation for 30 min at 37°C was quantified by PhosphorImaging analysis. The amount of N22 or C20 accumulating in the presence of ATP was set to 1. An average of several independent experiments (n) and the S.D. is shown. B, metal dependence. After completion of import of radiolabeled Oxa1 ts in oxa1 ts ⌬yta10 mitochondria, samples were depleted of divalent metal ions by adding the chelating agents EDTA (5 mM) or EDTA and ophenanthroline (o-phe) (each 2.5 mM). The proteolytic fragments N22 and C20 accumulating after incubation for 30 min at 37°C were quantified as in C. The statistical analysis was performed as in A.
bound mitochondrial protein of 35.8 kDa that harbors a conserved domain of the metallopeptidase family M48 (48). Homologous proteins are present in higher eukaryotes, including plants, and in eubacteria as well as archaebacteria (Fig. 4,  A and B) but have not been analyzed. Notably, homologues in higher eukaryotic cells have an N-terminal extension that is missing in eubacterial and archaebacterial proteins (Fig. 4, A and B). For reasons outlined below, the yeast protein encoded by YKR087c was termed Oma1 (for overlapping activity with m-AAA protease).
To determine the subcellular localization of Oma1, we generated a variant carrying a Myc tag at the C terminus. Yeast cells expressing Oma1 myc were subfractionated, and samples were then analyzed by immunoblotting. Oma1 myc was exclu- Putative transmembrane segments are shown by gray bars and the consensus metal-binding site by black bars. C, subcellular localization of Oma1. Cells expressing Myc-tagged Oma1 were grown on YP medium containing galactose (2%) and glucose (0.1%). Cells (corresponding to 0.5 optical density units) were lysed by alkaline extraction (Cell extract). Mitochondria (10 g) and the postmitochondrial supernatant fraction (corresponding to 0.5 optical density units) were analyzed by SDS-PAGE and immunoblotting using antisera directed against the c-Myc epitope, Bmh1 or Mge1. Bmh1 and Mge1 are markers of the cytosolic and mitochondrial fraction, respectively. D, import of Oma1 into isolated mitochondria. 35 S-Labeled Oma1 was imported at 25°C into wild type mitochondria. To dissipate the membrane potential across the inner membrane, valinomycin (1 M) and carbonyl cyanide p-trifluoromethoxyphenylhydrazone (25 M) were added prior to import when indicated (Ϫ⌬). All samples were treated with trypsin (100 g/ml) to remove non-imported Oma1. Aliquots were withdrawn at various times and resuspended in SHKCl buffer (ϩ⌬) or in 20 mM HEPES/KOH, pH 7.4, for osmotic disruption of the outer membrane (Swelling). Then samples were treated with proteinase K (PK) (100 g/ml) and analyzed by SDS-PAGE and autoradiography. E, submitochondrial localization of Oma1. Mitochondria harboring Myc-tagged Oma1 were fractionated by alkaline extraction (Na 2 CO 3 ) and osmotic disruption of the outer membrane (Swelling) and treated with proteinase K (PK; 50 g/ml). Fractions were analyzed by SDS-PAGE, and immunoblotting with antisera was directed against the Myc epitope, the intermembrane space protein cytochrome b 2 (Cyb2), matrix-localized Mge1, and the ADP/ATP carrier (AAC) which is an integral part of the inner membrane. F, Oma1 is part of a high molecular weight complex in the inner membrane. Mitochondria harboring Myc-tagged Oma1 were solubilized with Triton X-100. After a clarifying spin, mitochondrial extracts were fractionated by Superose 6 size exclusion chromatography. Eluate fractions were trichloroacetic acid-precipitated and analyzed by SDS-PAGE and immunoblotting using a Myc-specific antiserum. The following proteins were used as standards for calibration: hsp60 (840 kDa, 12.5 ml); thyroglobulin (669 kDa, 13 ml); apoferritin (443 kDa, 14.9 ml); cytochrome b 2 (250 kDa, 16.75 ml). sively recovered from the mitochondrial fraction and was not detectable in the postmitochondrial supernatant (Fig. 4C). These results were further substantiated by mitochondrial protein import studies. Oma1 was synthesized in rabbit reticulocyte lysate in the presence of [ 35 S]methionine and incubated with energized isolated mitochondria. Oma1 accumulated in a protease-resistant location in a time-dependent manner (Fig.  4D). Despite its low efficiency, this reflects the specific import of Oma1 into mitochondria as it was dependent on the presence of a membrane potential across the inner membrane (Fig. 4D). Moreover, Oma1 became accessible to externally added proteinase K upon osmotic disruption of the outer membrane (Fig.  4D). Oma1 does not carry a cleavable N-terminal presequence as we could neither observe proteolytic processing upon import into mitochondria nor upon incubation of radiolabeled Oma1 with purified mitochondrial processing peptidase (data not shown).
Several hydrophobic stretches, which could form membranespanning segments, are present in the sequence of Oma1 (Fig.  4B), suggesting that Oma1 is an integral membrane protein. In order to determine its submitochondrial localization, mitochondria harboring Oma1 myc were fractionated and analyzed with a polyclonal antiserum directed against the Myc epitope. In a first step, mitochondrial membranes were extracted at alkaline pH. Only integral membrane proteins like the ADP/ATP carrier remain in the pellet fraction under these conditions, whereas soluble proteins as well as peripheral membrane proteins accumulate in the supernatant (Fig. 4E). Oma1 myc was almost exclusively recovered from the pellet fraction under these conditions indicating a tight membrane association (Fig.  4E). Oma1 myc was resistant to added proteinase K but became accessible to proteolytic degradation upon osmotic disruption of the outer membrane (Fig. 4E). Cytochrome b 2 , a soluble protein in the intermembrane space, was released upon swelling, whereas matrix-localized Mge1 remained protease-resistant under these conditions (Fig. 4E). We conclude from these results that Oma1 is an integral part of the inner membrane and exposes its C terminus carrying the Myc tag to the intermembrane space.
Oma1 Forms a High Molecular Mass Complex in the Inner Membrane-The native molecular mass of Oma1 was subsequently determined in gel filtration experiments. Mitochondrial membranes harboring Oma1 myc were solubilized with Triton X-100, and extracts were subjected to sizing chromatography. Eluate fractions were analyzed using a Myc-specific polyclonal antiserum. Oma1 myc was recovered in a single peak from the column corresponding to a molecular mass of ϳ300 kDa (Fig. 4F). The formation of this complex was not dependent on ATP (data not shown). Thus, Oma1 myc is part of a high molecular mass assembly in the mitochondrial inner membrane. Considering the subunit size of Myc-tagged Oma1, the observed native mass of the Oma1-containing complex should be consistent with a homo-hexameric structure. However, further experiments remain to be done to establish whether Oma1 forms a homo-oligomeric complex or associates with other yet unknown proteins.
Degradation of Oxa1 ts by Oma1-The results described identify Oma1 as a novel metallopeptidase in the mitochondrial inner membrane. To examine a potential role of Oma1 in the degradation of misfolded polytopic membrane proteins, OMA1 was deleted in wild type, oxa1 ts cells, and in oxa1 ts ⌬yta10 cells by homologous recombination. Cells lacking Oma1 did not exhibit apparent growth defects on fermentable or non-fermentable carbon sources at 15, 30, or 37°C (data not shown).
The stability of Oxa1 ts in mitochondria lacking Oma1 was analyzed in subsequent experiments. 35 S-Labeled Oxa1 ts was imported into mitochondria isolated from ⌬oma1, ⌬yta10, and ⌬yta10⌬oma1 cells which were then further incubated at 37°C to allow proteolysis (Fig. 5, A and B). When compared with wild type mitochondria, degradation of Oxa1 ts was only slightly affected by deleting either Oma1 or the m-AAA protease subunit Yta10 (Fig. 5A). However, Oxa1 ts was stabilized in mitochondria lacking both proteases (Fig. 5A). These results demonstrate that both Oma1 and the m-AAA protease are capable of degrading this mutant polytopic membrane protein and thus exert an overlapping substrate specificity.
Both proteases can apparently act independently of each other as the proteolytic breakdown of Oxa1 ts proceeded in the absence of either protease. In wild type as well as in ⌬oma1 mitochondria proteolysis of Oxa1 ts is mediated by the m-AAA protease; the C-terminal domain is cleaved off, whereas the transmembrane domain of Oxa1 ts is completely degraded (Fig.  5B). In the absence of the m-AAA protease, on the other hand, proteolytic fragments of Oxa1 ts accumulated in mitochondria (Fig. 5B). These fragments were apparently generated by Oma1 as they were not detectable in oxa1 ts ⌬yta10⌬oma1 mitochondria lacking Oma1 (Fig. 5B).
We also determined the steady state level of Oxa1 ts in mitochondria lacking one or both proteases by immunoblotting (Fig.  5C). Oxa1 ts accumulated at slightly increased levels in the absence of the m-AAA protease (Fig. 5C). Moreover, N-and C-terminal proteolytic fragments of Oxa1 ts were detected in these mitochondria as described before (see Fig. 2B). Deletion of OMA1 had different effects on the stability of Oxa1 ts depending on the presence or absence of the m-AAA protease, whereas Oxa1 ts accumulated at similar levels in oxa1 ts and oxa1 ts ⌬oma1 mitochondria, deletion of OMA1 in oxa1 ts ⌬yta10 cells resulted in a significant stabilization of Oxa1 ts (Fig. 5C). Oma1 is required for the formation of the proteolytic fragments N22 and C20 (as well as the minor proteolytic products C16), which were not detectable in oxa1 ts mitochondria lacking both proteases (Fig. 5C). Notably, Oxa1 ts is not completely stabilized in the absence of the m-AAA protease and Oma1. The proteolytic fragments N27 and C15 accumulated in oxa1 ts ⌬yta10⌬oma1 mitochondria pointing to the existence of yet other peptidase(s) (Fig. 5C).
To substantiate a proteolytic activity of Oma1, point mutations were introduced into the consensus metal-binding site of Oma1 representing the predicted proteolytic center. The potential metal ligands histidine 176 and 181 and the catalytic glutamic acid residue 173 were exchanged by site-directed mutagenesis. Wild type Oma1 and the mutant proteins were expressed in oxa1 ts ⌬yta10⌬oma1 cells, and the steady state level of mutant Oxa1 was determined (Fig. 5D). Whereas the proteolytic fragments N22 and C20 accumulated at significantly increased levels upon expression of wild type Oma1, they were not detectable in cells harboring the mutant variants of Oma1. Thus, the integrity of the consensus metal-binding site is essential for proteolytic activity identifying Oma1 as a metallopeptidase.
We conclude from these experiments that Oma1 can degrade the misfolded polytopic membrane protein Oxa1 ts in vivo and in vitro and thereby partially substitutes for a loss of the m-AAA protease activity. In contrast to the m-AAA protease, however, Oma1 does not mediate the complete turnover of Oxa1 ts but rather generates N-and C-terminal proteolytic fragments. The respiratory deficiency of yeast cells lacking the m-AAA protease was not suppressed upon overexpression of Oma1 providing further evidence for differences in the substrate specificity of both peptidases (data not shown).
Proteolytic Cleavage of Oxa1 ts at Both Membrane Surfaces by Oma1-The submitochondrial localization of Oma1 together with our studies using metallopeptidase inhibitors indicate that the catalytic center of Oma1 is exposed at the inner surface of the mitochondrial inner membrane. To examine whether Oma1 cleaves misfolded Oxa1 ts exclusively at this membrane surface, we determined the cleavage sites of Oma1 in the mutant protein. In a first step, truncated OXA1 alleles were generated that encode C-terminal fragments of Oxa1 ts starting at amino acid residues Leu-201, Met-221, Tyr-246, Leu-263, Phe-280, or Ser-309, i.e. amino acid residues flanking transmembrane segments 2-5. In addition, Oxa1 variants were constructed encoding N-terminal fragments of mature Oxa1 ts terminating at amino acid residues Phe-236, Phe-257, and Thr-291. The mutant proteins were synthesized in rabbit reticulocyte lysate in the presence of [ 35 S]methionine and used as size standards during SDS-gel electrophoresis (Fig. 6, A and B).
The length of the proteolytic fragments, which accumulated upon proteolysis of newly imported Oxa1 ts , was determined by an electrophoretic analysis using the derived calibration curve. In oxa1 ts mitochondria, only the C-terminal fragment C14 was generated by the m-AAA protease, whereas the N-terminal part of Oxa1 ts appears to be degraded in a processive manner (see Fig. 2). Amino acid residues 286/287 were estimated as the N-terminal amino acids of C14 by the electrophoretic analysis (Fig. 6) confirming previous findings that the m-AAA protease can cleave substrates within membrane-spanning segments (15). In oxa1 ts ⌬yta10 mitochondria, the C-terminal fragment C16 was generated by Oma1-mediated cleavage of Oxa1 ts at amino acid residues 264/265 that directly follow the third transmembrane segment (amino acids 245-262) and that are located in the matrix space (Fig. 6). Unexpectedly, the electro- FIG. 5. Proteolysis of Oxa1 ts by Oma1. A and B, stability of newly imported Oxa1 ts . 35 S-Labeled Oxa1 ts was imported into mitochondria isolated from wild type (WT), ⌬yta10, ⌬oma1, and ⌬yta10⌬oma1 cells, and proteolysis at 37°C was monitored by SDS-PAGE and autoradiography. The stability of full-length Oxa1 ts was quantified by PhosphorImaging analysis (A). B, the autoradiography of the gel is shown. C, steady state level of Oxa1 ts in mitochondria lacking the m-AAA protease and Oma1. Mitochondria isolated from wild type (WT), oxa1 ts , oxa1 ts ⌬yta10, oxa1 ts ⌬oma1, and oxa1 ts ⌬yta10⌬oma1 cells were analyzed by SDS-PAGE and immunoblotting using antisera directed against a N-(␣Oxa1-N) or a C-terminal peptide of Oxa1 (␣Oxa1-C). As a gel loading control, steady state levels of the outer membrane protein Tom70 and the ADP/ATP carrier in the inner membrane was determined by immunoblotting (data not shown). D, steady state level of Oxa1 ts in oxa1 ts ⌬yta10⌬oma1 mitochondria expressing Oma1 or the mutant variants of Oma1 E173Q , Oma1 H176Y , or Oma1 H181Y . Immunoblotting was performed as described in C. phoretic analysis of the Oxa1 ts fragment N22 identified at its C terminus the amino acid residues 245/246, whereas amino acid residues 240/241 were established as the N-terminal amino acids of C20 (Fig. 6). Although the exact determination of Oma1 cleavage sites requires sequencing of the proteolytic fragments, these experiments strongly suggest that Oma1 is capable of cleaving Oxa1 ts not only within loop regions in the matrix space but also in a loop region that harbors the mutated site and is exposed to the intermembrane space. DISCUSSION We have identified a conserved metallopeptidase, Oma1, as a novel component of the quality control system in the inner membrane of mitochondria. Oma1 has functions overlapping with the m-AAA protease and cleaves a misfolded polytopic membrane protein at multiple sites. Proteins homologous to Oma1 comprise a large protein family with members present in higher eukaryotes, including plants, as well as in eubacteria and archaebacteria. Although different in their domain structure, all of them are predicted to be integral membrane proteins and contain a metallopeptidase domain characteristic of the M48 family, suggesting that Oma1 represents a novel enzyme class capable of degrading membrane proteins.
A mutant variant of the inner membrane protein Oxa1 carrying a point mutation in the loop region between the second and third transmembrane segment was employed as a model substrate to analyze the quality control of polytopic membrane proteins. Misfolding of mutant Oxa1 can be easily monitored in isolated mitochondria by protease protection experiments; although resistant in the wild type protein, the loop region between the second and third transmembrane segment is protease-sensitive in the mutant variant. Our analysis revealed that misfolded Oxa1 is degraded in an ATP-dependent manner by the m-AAA protease or, in an ATP-independent manner, by Oma1 and another yet unknown peptidase resulting in the accumulation of proteolytic fragments. Notably, Oma1-mediated substrate cleavage is not a prerequisite for proteolysis by the m-AAA protease but appears to occur under conditions of limited m-AAA protease activity.
Subfractionation of mitochondria and inhibitor studies indicate that the proteolytic center of Oma1 is present on the matrix side of the inner membrane. Its activity was only inhibited in the presence of a membrane-permeable chelating agent, whereas depletion of divalent metal ions of the intermembrane space did not impair Oma1 activity. The consensus metal-binding motif is in close proximity to a transmembrane segment that represents the only membrane-spanning segment C-terminal to the proteolytic site. Consistently, the C-terminal end of Oma1 was demonstrated to be accessible in mitoplasts to externally added protease. The length estimation of the proteolytic fragments generated by Oma1, on the other hand, point to cleavage events in loop regions of the substrate protein exposed to both the intermembrane and matrix space. An impaired membrane insertion of the misfolded protein seems unlikely as the loop region harboring both the mutated site as well as an Oma1 cleavage site is protease-accessible in isolated mitoplasts, i.e. exposed to the outer surface of the inner membrane. We therefore conclude that Oma1 can cleave a polytopic membrane protein at opposite membrane surfaces. This is reminiscent of mitochondrial Oma1 (resulting in the formation of N22, C20, and C16). The fragments C15 and N27 were generated by proteolytic cleavage after amino acid residues 275/276 and 273/274, respectively. It is conceivable that both fragments were formed by a single cleavage by an unknown peptidase. f27 and f19 represent fragments generated upon addition of externally added protease to mitoplasts (see Fig. 1D).
AAA proteases that degrade membrane proteins after their dislocation from the membrane bilayer (45). Oma1 was found to assemble in a larger complex in the inner membrane that could provide a hydrophilic environment within the lipid bilayer facilitating a vertical movement of membraneembedded segments. It remains to be examined, however, whether Oma1-mediated proteolysis indeed involves membrane dislocation of substrate proteins or occurs by another mechanism.
Although Oma1 is able to clip membrane proteins, its properties differ from the recently defined group of intramembrane peptidases like the S2P protease, rhomboid-1, the signal peptide peptidase, and presenilins (49,50). Mutational analysis has identified the consensus metal-binding motif HEXXH of Oma1 as the proteolytic center. Whereas intramembrane peptidases harbor their catalytic site residues within the membrane bilayer, the conserved metal-binding site of Oma1 as well as that of other family members is in close proximity to a transmembrane segment but most likely not embedded within the membrane. Moreover, intramembrane peptidases appear to recognize only substrates containing one transmembrane segment, whereas Oma1 was found to cleave a polytopic membrane protein. On the other hand, Oma1 can be distinguished from peptidases mediating the complete turnover of membrane proteins, like the 26 S proteasome or AAA proteases (51,52), as proteolysis by Oma1 seemingly does not depend on ATP and results in the accumulation of proteolytic fragments of a polytopic substrate protein. Oma1, together with its homologues, therefore comprises a novel class of peptidases that recognize membrane-embedded polypeptides as substrates.
The analysis of the proteolytic breakdown of the polytopic Oxa1 protein also revealed new insights into the mechanism of membrane protein degradation by AAA proteases. As other ATP-dependent proteases, AAA proteases are thought to completely degrade substrates into small peptides (53). Proteolysis of mutant Oxa1 by the m-AAA protease, however, results in the accumulation of the C-terminal domain in the matrix space whose tight folding may preclude its degradation. It seems unlikely that the m-AAA protease degrades the substrate from the N terminus which is exposed to the intermembrane space of mitochondria. We therefore favor the possibility that the m-AAA protease can initiate proteolysis of a polytopic membrane protein at internal sites which results in the generation of the C-terminal fragment of Oxa1. This is reminiscent of the proteolytic processing of membrane-bound transcription factors by 26 S proteasomes, which is believed to involve a hairpin loop formation of the substrate within the cavity of the proteasome (54). In a similar manner, the m-AAA protease may mediate the processing of the newly imported precursor form of cytochrome c peroxidase (26,27).
Previous studies employing model substrate proteins with single transmembrane segments revealed that their stability, as well as the involvement of one or the other AAA protease, is determined by the folding state of solvent-exposed domains and the number of amino acid residues present at one membrane surface (45,55). The m-AAA protease recognizes misfolded Oxa1, even though an amino acid residue in an intermembrane space loop of this protein was mutagenized. Whereas matrixexposed segments of the mutant Oxa1 protein may attain a non-native conformation which is recognized by the m-AAA protease, the stability of misfolded Oxa1 toward degradation by the i-AAA protease is unexpected. These findings therefore emphasize the critical importance of analyzing the turnover of a variety of model substrate proteins with different membrane topologies to fully understand features guiding the turnover of membrane proteins by AAA proteases.