Manganese Activation of Superoxide Dismutase 2 in the Mitochondria of Saccharomyces cerevisiae*

Manganese-dependent superoxide dismutase 2 (SOD2) in the mitochondria plays a key role in protection against oxidative stress. Here we probed the pathway by which SOD2 acquires its manganese catalytic cofactor. We found that a mitochondrial localization is essential. A cytosolic version of Saccharomyces cerevisiae Sod2p is largely apo for manganese and is only efficiently activated when cells accumulate toxic levels of manganese. Furthermore, Candida albicans naturally produces a cytosolic manganese SOD (Ca SOD3), yet when expressed in the cytosol of S. cerevisiae, a large fraction of Ca SOD3 also remained manganese-deficient. The cytosol of S. cerevisae cannot readily support activation of Mn-SOD molecules. By monitoring the kinetics for metalation of S. cerevisiae Sod2p in vivo, we found that prefolded Sod2p in the mitochondria cannot be activated by manganese. Manganese insertion is only possible with a newly synthesized polypeptide. Furthermore, Sod2p synthesis appears closely coupled to Sod2p import. By reversibly blocking mitochondrial import in vivo, we noted that newly synthesized Sod2p can enter mitochondria but not a Sod2p polypeptide that was allowed to accumulate in the cytosol. We propose a model in which the insertion of manganese into eukaryotic SOD2 molecules is driven by the protein unfolding process associated with mitochondrial import.

Manganese-dependent superoxide dismutase 2 (SOD2) in the mitochondria plays a key role in protection against oxidative stress. Here we probed the pathway by which SOD2 acquires its manganese catalytic cofactor. We found that a mitochondrial localization is essential. A cytosolic version of Saccharomyces cerevisiae Sod2p is largely apo for manganese and is only efficiently activated when cells accumulate toxic levels of manganese. Furthermore, Candida albicans naturally produces a cytosolic manganese SOD (Ca SOD3), yet when expressed in the cytosol of S. cerevisiae, a large fraction of Ca SOD3 also remained manganese-deficient. The cytosol of S. cerevisae cannot readily support activation of Mn-SOD molecules. By monitoring the kinetics for metalation of S. cerevisiae Sod2p in vivo, we found that prefolded Sod2p in the mitochondria cannot be activated by manganese. Manganese insertion is only possible with a newly synthesized polypeptide. Furthermore, Sod2p synthesis appears closely coupled to Sod2p import. By reversibly blocking mitochondrial import in vivo, we noted that newly synthesized Sod2p can enter mitochondria but not a Sod2p polypeptide that was allowed to accumulate in the cytosol. We propose a model in which the insertion of manganese into eukaryotic SOD2 molecules is driven by the protein unfolding process associated with mitochondrial import.
Superoxide dismutase (SOD) 1 enzymes represent a family of metalloproteins that have evolved to catalytically remove toxic superoxide anions. Most eukaryotes express two distinct forms, a copper-and zinc-containing enzyme (SOD1) that largely resides in the cytosol (1) but is also found in the intermembrane space of mitochondria (2)(3)(4) and a second SOD that contains manganese (SOD2) and is typically localized in the mitochondrial matrix (3,5). In both cases, enzymatic activity is reliant on the redox cycling of the bound copper or manganese ion cofactor. Hence, the post-translational insertion of the metal represents a key step in controlling enzymatic activity in vivo.
Much is known about the mechanism by which SOD1 acquires copper in vivo. Copper is transported and trafficked to the site of SOD1 by the concerted action of cell surface and intracellular copper transporters and a copper chaperone known as CCS (6 -12). CCS can insert copper into a pre-existing apopool of SOD1 with no need for new protein synthesis (13,14). The copper chaperone can also act on newly synthesized molecules of SOD1 (14). In either case, oxygen is required for CCS activity, providing a means for regulating SOD1 activity in response to oxygen status (15).
The delivery of manganese to SOD2 should also involve a carefully controlled trafficking system. Using yeast genetics, we have identified two membrane proteins that help deliver manganese to the enzyme. One is the divalent metal transporter Smf2p that localizes in intracellular vesicles (16,17). Saccharomyces cerevisiae cells lacking Smf2p accumulate very low levels of manganese and show defects in manganese requiring enzymes of the Golgi and in Sod2p of mitochondria (18). A second transporter that affects yeast Sod2p is Mtm1p, a member of the mitochondrial carrier family of proteins (19). Although the precise substrate for transport by Mtm1p is not known, Mtm1p is needed for proper insertion of manganese into mitochondrial Sod2p (19).
Despite the identification of these components, the mechanistic details of the post-translational events associated with activation of eukaryotic SOD2 are still poorly understood. The protein is encoded in the nucleus, transported across two mitochondrial membranes and once inside the mitochondrial matrix, and the polypeptide folds into a tetrameric enzyme. The stage at which manganese is introduced is not known. For example, can manganese be inserted into a pre-existing pool of apoSOD2, as is the case with copper containing SOD1? In addition, it is not clear whether SOD2 requires a mitochondrial location to acquire its metal cofactor. In certain organisms, Mn-SODs can be activated outside the mitochondria. The pathogenic fungi Candida albicans (20) and the blue crab Callinectes sapidus (21) both express Mn-SODs in the cytosol. When the Mn-SODs from either C. albicans or from the bacterium Bacillus stearothermophilus were targeted to the cytosol of S. cerevisiae, they exhibited certain activity (20,22,23). These results alone would suggest that a mitochondrial location may not be essential for manganese activation of SOD2.
In this study, we explored the pathway for inserting manga-nese into mitochondrial SOD2 using S. cerevisiae Sod2p as a model. We found that efficient metalation of Sod2p requires a mitochondrial localization of the protein; a cytosolic version of Sod2p is poorly activated with the metal. Furthermore, only newly synthesized molecules of Sod2p that are freshly imported into mitochondria can acquire the metal in vivo. Manganese cannot be readily inserted into a pool of Sod2p that is apo for manganese. We provide a model in which manganese insertion into Sod2p is driven by the protein unfolding process associated with mitochondrial import.
Plasmids-The pEL111 vector was constructed by subcloning the BamHI-SalI fragment of pEL101 (18) containing Ϫ557 to ϩ894 of the S. cerevisiae SOD2 gene into the pRS415 vector (25) digested with the same enzymes. To construct the vector pEL1G1 bearing a GAL1-SOD2 fusion, a SacI site (GAGCTC) was first introduced in the SOD2expressing vector pEL101 (18) replacing the sequence TAAAAA 15 bp upstream of the SOD2 start codon by site-directed mutagenesis. This plasmid was subsequently digested with SacI and XhoI, and the resulting 910-bp fragment containing the SOD2 open reading frame and its transcriptional terminator was subcloned downstream of a GAL1 promoter in the pYES2/CT vector (Invitrogen) using the same restriction sites. Sequence integrity was confirmed by DNA sequencing analysis (Core Facility, Johns Hopkins Medical Institutions). The multicopy expression vector for cytosolic Sod2p (amino acids 27 to the stop codon) pEL124 was created by subcloning the BamHI-SalI fragment of pEL104 (18) into pRS425 (25) at the same restriction enzyme sites. The resulting construct for cytosolic Sod2p contains the SOD2 gene promoter (Ϫ558 to Ϫ1) and terminator (ϩ703 to ϩ889) but lacks the mitochondrial presequence. The C. albicans SOD3 expression vector pVTSOD3 was described previously (20).
Biochemical Assays-For preparation of cell lysates, S. cerevisiae strains were inoculated in 50 ml of YPD at a starting A 600 of ϳ0.05 and allowed to grow without shaking at 30°C for ϳ15 h. In general, whole cell lysates were prepared by glass-bead agitation as described previously (18). Where mitochondrial fractionation was required, yeast cells were converted to spheroplasts, lysed with a Dounce homogenizer, and fractionated into mitochondrial and post-mitochondrial cell fractions by differential centrifugation as previously described (26). Whole cell lysates or cellular fractions were analyzed for SOD activity by native gel electrophoresis and staining with nitroblue tetrazolium (27). S. cerevisiae Sod2p and C. albicans SOD3 polypeptide levels were monitored by subjecting whole cell lysates or cell fractions to denaturing gel electrophoresis and immunoblotting with an antibody directed against S. cerevisiae Sod2p (18) that cross-reacts with C. albicans SOD3. Where needed, antibodies directed against Mas2p (gift from Dr. Rob Jensen, Johns Hopkins University) and cytosolic Pgk1p (Molecular Probes, Eugene, OR) were used as described (18).
S. cerevisiae Sod2p (containing the N-terminal mitochondrial targeting sequence) and C. albicans SOD3 were purified as recombinant proteins as described previously (19,20,28). Molar concentrations of these SOD molecules were determined by amino acid hydrolysis analysis (Protein Chemistry Laboratory, Texas A & M University).
To monitor the rate of manganese activation of Sod2p, smf2⌬ mutants deficient in manganese were grown in YPD medium for ϳ15 h to an A 600 of ϳ3.0. 10 M MnSO 4 was then added and after various time intervals, 25 ml of culture samples were removed for preparation of whole cell lysates by glass bead homogenization and for monitoring Sod2p activity and polypeptide levels as above. Where needed, 100 g/ml cycloheximide was added to cultures just prior to manganese supplementation. In a duplicate set of cultures, lysates were prepared from spheroplasts, and crude mitochondria were isolated (26). The mitochondria fractions were analyzed for manganese content as described in Ref. 19 by atomic absorption spectroscopy using a PerkinElmer AAnalyst 600 graphite furnace atomic absorption spectrometer.
To follow in vivo mitochondrial import of Sod2p, sod2⌬ yeast mutant cells were transformed with the GAL1-SOD2 vector pEL1G1 and cul-tured for 17 h in YPR medium to an A 600 of 1.1-1.5. 2% galactose was then added to induce SOD2 expression. Where needed, 20 M of the proton uncoupler carbonyl cyanide m-chlorophenylhydrazone (CCCP) (Sigma) was added to block mitochondrial import (29). Addition of 0.05% (v/v) ␤-mercaptoethenol (␤-ME) served to neutralize CCCP as described previously (29). Cell lysates were prepared by spheroplast homogenization, and mitochondrial and post-mitochondrial supernatant fractions were prepared as described above.

Efficient Metalation of Sod2p Requires Mitochondrial
Localization-We tested whether a mitochondrial localization of Sod2p was needed for manganese insertion into the enzyme. A cytosolic version of S. cerevisiae Sod2p was created by removing the N-terminal mitochondrial presequence ( Fig. 1A) (30). The resulting Sod2p molecule (CytSod2p) was expressed in a sod2⌬ mutant of S. ceresiviae lacking the endogenous mitochondrial Sod2p. As seen in Fig. 1B, CytSod2p co-localizes with the cytosolic marker Pgk1p and is largely excluded from the mitochondria marked by the mitochondrial matrix protein Mas2p. By comparison, expression of native Sod2p harboring the Nterminal presequence (MitoSod2p) resulted in a mitochondrial localization of the enzyme as expected (Fig. 1B).
To test for enzymatic activity, lysates from cells expressing CytSod2p or MitoSod2p were applied to a native gel and analyzed for SOD activity by nitroblue tetrazolium staining. As seen in Fig. 1C, the cytosolic Sod2p was largely inactive compared with the mitochondrial enzyme (compare lanes 1 and 6). The activity of CytSod2p was restored by manganese supplementation in vivo, indicating that the lack of CytSod2p activity under physiological conditions results from a manganese deficiency in the enzyme. It is noteworthy that the amount of manganese required to activate CytSod2p is quite high (Ն100 M). This is a concentration that is somewhat toxic to the yeast,  (30), and this region was removed from S. cerevisiae Sod2p to create CytSod2p. B, cell lysates were prepared from strain BY4741 expressing native mitochondrial Sod2p (MitoSod2p) and from the isogenic sod2⌬ mutant transformed with pEL124 expressing cytosolic Sod2p (CytSod2p). 60 g of total cell lysates (T) were separated by differential centrifugation into a post-mitochondrial supernatant fraction that is largely cytosolic (C) and a crude mitochondria fraction (M). The entire sample of each fraction, along with 60 g of total cell lysate, (to achieve identical cell equivalents) were subjected to denaturing gel electrophoresis and immunoblot analysis with antibodies directed against S. cerevisiae Sod2p, the cytosolic phosphoglycerate kinase (Pgk1p), and the mitochondrial processing protease (Mas2p). C, the sod2⌬ strain expressing either native mitochondrial Sod2p (Mito-SOD2p) on plasmid pEL111 or cytosolic Sod2p on plasmid pEL124 were grown overnight in YPD medium supplemented with the indicated concentrations of MnSO 4 . Whole cell lysates were subjected to either native gel electrophoresis and staining with nitroblue tetrazolium for Sod2p activity (top) or denaturing gel electrophoresis and immunoblotting with anti-Sod2p (bottom). It is noteworthy that, prior to denaturing gel electrophoresis, Sod2p-containing samples were heated in SDS at Ϸ40°C, rather than the standard 100°C, to prevent precipitation of Sod2p.
as indicated by slowed growth ( (19) and (not shown)). The cytosolic form of Sod2p is only active when cells hyperaccumulate manganese. Under physiological conditions, Sod2p needs to be inside the mitochondria to be efficiently activated.
Cytosolic SOD3 from C. albicans Is Largely Active When Expressed in S. cerevisiae-The pathogenic fungi C. albicans expresses a manganese-containing SOD in the cytosol (Ca SOD3) that is reported to be active when expressed in the cytosol of S. cerevisiae (20). We therefore addressed whether Ca SOD3 has a unique ability to acquire manganese in the cytosol.
Consistent with earlier studies (19,20), Ca SOD3 expressed in S. cerevisiae exhibits some activity under physiological conditions ( Fig. 2A, lanes 2 and 8). Expression was observed in both a sod2⌬ strain (lane 8) and a strain expressing the endogenous mitochondrial Sod2p of S. cerevisiae (lane 2). Activity of Ca SOD3 expressed in S. cerevisiae is limited by manganese bioavailability. Decreasing intracellular manganese through a deletion of the Smf2p manganese transporter (18) abolished Ca SOD3 activity (Fig. 2A, lane 5), and activity was rescued by growing cells in the presence of 100 M manganese (lane 6). In fact, such manganese supplementation also had a dramatic effect on Ca SOD3 activity in SMF2 wild-type cells (lanes 3 and  9). Hence, there appears to be a large inactive pool of Ca SOD3 that is manganese-deficient and can be activated at high intracellular manganese, reminiscent of the scenario seen with S. cerevisiae cytSod2p (Fig. 1C).
The expression of Ca SOD3 in S. cerevisiae is driven by a high copy vector and the strong constitutive ADH1 promoter (20). To estimate how much Ca SOD3 is being produced relative to endogenous Sc Sod2p, purified Sc Sod2p and Ca SOD3 proteins of known concentrations were used as standards in a semiquantitative immunoblot against lysates from cells expressing Sc Sod2p and Ca SOD3. As seen in Fig. 2B, Ca SOD3 is expressed in S. cerevisiae on a per mole basis at levels that are roughly 10-fold higher than the endogenous Sc Sod2p. This overexpression of Ca SOD3 protein may explain why activity can be detected in S. cerevisiae, despite the fact that a large fraction of the protein lacks manganese. Overall, the findings obtained with Ca SOD3 and with cytosolic Sod2p demonstrated that, in S. cerevisiae, efficient activation of Sod2p requires a mitochondrial localization. There is clearly a component absent from the cytosol that is required for efficient activation of Sod2p.
Insertion of Manganese into Sod2p Requires New Protein Synthesis-How is mitochondrial Sod2p activated with manganese? We know that, in the case of copper-containing Sod1p, a pre-existing apopool of the enzyme is rapidly activated with copper in the absence of new protein synthesis (13,14). We tested whether the same was true for manganese-containing Sod2p of the mitochondria.
To monitor activation of a pool of Sod2p that is largely apo for manganese, we utilized the manganese-deficient smf2⌬ mutant. In these cells, the Sod2p polypeptide still accumulates in the mitochondria, but is largely inactive because of low mitochondrial manganese (18). Sod2p activity is fully restored in this mutant by culturing cells in the presence of 10 M manganese (Fig. 3, A and C, lanes 3). We monitored the time required to activate Sod2p following the addition of manganese to the growth medium. As shown in Fig. 3, A and C, Sod2p was activated very slowly by manganese and required at least 2-3 h of treatment with the metal. By comparison, activation of cytosolic Sod1p by copper in S. cerevisiae cells is complete in Ͻ5 min (13). The slow activation of mitochondrial Sod2p is not a result of slow trafficking of the metal to the mitochondria, as mitochondrial manganese was restored to near wild-type levels after 15 min of treatment with manganese (Fig. 3B). Such a delay in metalation of the enzyme suggests that new protein synthesis may be required.
To address the requirement for protein synthesis, the time course for Sod2p activation was monitored under conditions in which in vivo protein translation was blocked by cycloheximide. Fig. 3C shows that cycloheximide treatment (lanes 5, 7, and 9) FIG. 2. SOD3 from C. albicans is only partially active when expressed in the cytosol of S. cerevisiae (A). Wild-type strain BY4741 and the isogenic smf2⌬ and sod2⌬ mutants, transformed where indicated (SOD3, ϩ) with pVTSOD3 (20)-expressing C. albicans (Ca) SOD3, were grown in YPD medium that was supplemented where indicated (Mn 2ϩ , ϩ) with 100 M MnSO 4 . Total yeast cell lysates were analyzed for SOD activity by the native gel assay as described in the legend to Fig. 1C.
The positions of C. albicans SOD3 and the endogenous Sod2p and Sod1p enzymes from S. cerevisiae are indicated. B, the specified amounts of whole cell lysate protein from either the sod2⌬ cell-expressing C. albicans SOD3 (left) or from wild-type BY4741-expressing endogenous Sod2p (right) were subjected to immunoblot analysis and compared with known amounts of the corresponding recombinant Mn-SOD molecule, which was purified to homogeneity as described previously (19,20). The purified recombinant Ca SOD3 and S. cerevisiae (Sc) Sod2p contain an N-terminal His 6 tag (20) and mitochondrial targeting sequence (18), respectively, that account for the slightly higher molecular weights on the immunoblot. completely abolished the increase in Sod2p activity with manganese treatment. Trafficking of manganese into the mitochondria, however, was unaffected by cycloheximide, as indicated by atomic absorption spectroscopy (Fig. 3B). Thus, manganese insertion requires new protein synthesis. The metal is not readily inserted into a pre-existing pool of apoSod2p, and only a freshly synthesized Sod2p molecule appears competent for metalation.
Mitochondrial Import and Synthesis of Sod2p Appear Closely Coupled-Synthesis of the Sod2p polypeptide occurs outside the mitochondria, whereas metalation of Sod2p takes place within mitochondria. Based on these distinct cellular locations, why would polypeptide synthesis be required for manganese insertion? As a likely explanation, the synthesis of Sod2p may be closely coupled with mitochondrial import, and it is the import process that facilitates manganese insertion. In fact, it has been suggested that certain mitochondrial proteins are imported co-translationally (31)(32)(33)(34), because folding of the polypeptide in the cytosol would prohibit mitochondrial uptake. We tested whether this was the case for Sod2p.
An experiment was designed in which Sod2p synthesis was controlled via the S. cerevisiae GAL1 promoter. A sod2⌬ deletion strain was transformed with an inducible Sod2p expression vector (Gal-Sod2p), and following 15-30 min of treatment with galactose, newly synthesized Sod2p became apparent (Fig. 4A). All of the newly synthesized Sod2p migrated as a single species (Fig. 4A), representing the mitochondrial processed ("P") form. Unprocessed, precursor Sod2p ("U") is completely absent in the induced samples, suggesting that the mitochondrial import of Sod2p occurs immediately following, or concomitant with, Sod2p synthesis.
To address this further, we uncoupled mitochondrial import and protein synthesis. This was achieved through use of the proton ionophore CCCP, which blocks import by disrupting the mitochondrial membrane potential (29) . Fig. 4B, lanes 4 -6, shows that CCCP completely blocked import of Sod2p into the mitochondria. As a result, the unprocessed Sod2p precursor (U) accumulated in the cytosolic fraction (Fig. 4B, lane 5). The effect of CCCP can be neutralized by ␤-ME (29), and when ␤-ME is added shortly following CCCP, there is no inhibition of import, and newly synthesized Sod2p was taken into mitochondria and processed (P) (Fig. 4B, lanes 7-9).
Using this system, we tested whether pre-existing cytosolic Sod2p can be chased into mitochondria. In the experiment of Fig. 4C, Sod2p synthesis was induced for 3 h in the presence of CCCP to allow accumulation of unprocessed cytosolic Sod2p (U) (lane 1). Where indicated, ␤-ME was then added for an additional 45 min. ␤-ME clearly reversed the effects of CCCP during this time frame, because the shorter form of Sod2p representing processed mitochondrial Sod2p (P) became apparent (Fig. 4C, lane 3). However, the unprocessed (U) Sod2p that accumulated prior to ␤-ME treatment was unchanged when mitochondrial import function was restored with ␤-ME (compare U in Fig. 4C, lanes 1 and 3). Furthermore, the appearance of processed mitochondrial Sod2p required new protein synthesis, as cycloheximide specifically prevented formation of processed (P) Sod2p in ␤-ME treated cells (lane 4). Together these findings are consistent with the notion that mitochondrial import requires freshly translated Sod2p. Overall, the synthesis, mitochondrial import, and manganese insertion steps for Sod2p are closely coordinated in time.

DISCUSSION
The mitochondrial SOD2 enzyme is well known for its role in eukaryote survival and fitness (35)(36)(37)(38)(39)(40)(41)(42). Yet despite this widespread importance, virtually nothing is known about the maturation of the SOD2 polypeptide in vivo. How is the inactive protein encoded by the nucleus converted into an active manganese-containing enzyme in the mitochondrial matrix? We have shown here that activation of S. cerevisiae Sod2p through insertion of the manganese cofactor must occur within the mitochondria. When expressed in the cytosol of S. cerevisiae, Mn-SOD molecules are poorly activated. Efficient manganese activation also requires new protein synthesis and mitochondrial import. Our data are consistent with a model in which the translation, mitochondrial import, and manganese activation of Sod2p are closely coupled in time.
Although Sod2p is largely inactive when expressed in the cytosol of S. cerevsiae, activity could be restored by exposing cells to high toxic concentrations of manganese. Under normal physiological conditions, the bioavailability of manganese in the cytoplasm appears too low to activate newly synthesized Sod2p. This may be a universal phenomenon, because most eukaryotes do not express a cytosolic Mn-SOD. However, there are rare exceptions, as in the case of the cytosolic Mn-SOD of C. albicans (20) and of decapod crustaceans (21). Our studies here show that C. albicans SOD3 does not possess an inherent ability to acquire cytosolic manganese, as a large fraction of the protein remained inactive when expressed in the cytoplasm of S. cerevisiae. Instead, C. albicans, as well as the crustaceans, may have evolved novel methods for delivering manganese to Mn-SOD in the cytosol, e.g. mechanisms that involve a manganese chaperone or elevated bioavailability of the metal.
Our studies show that import of S. cerevisiae Sod2p into mitochondria requires a freshly synthesized Sod2p polypeptide. If allowed to accumulate and fold in the cytosol, Sod2p is refractory to mitochondrial uptake. When folded, Sod2p is a notoriously stable molecule. Human SOD2 is stable at 60°C (43), and the S. cerevisiae enzyme can be purified following treatment at 70°C with little loss in activity (5). Studies with homologous manganese SODs from bacteria have shown that even the apoform of the enzyme forms a tight stable structure resistant to thermal denaturation (44,45). As such, it is not surprising that import of SOD2 into mitochondria must occur before the protein has a chance to fold in the cytosol. In this regard, it is noteworthy that the mRNA for SOD2 in mammalian cells and the mRNA/ribosomes for Sod2p in S. cerevisiae are both found associated with the outer membrane of mitochondria (46,47). With other mitochondrial proteins, the 3Ј- FIG. 4. Sod2p accumulated in the cytosol cannot be imported into mitochondria. sod2⌬ mutants harboring the pEL1G1 vector for galactose-inducible expression of SOD2 were grown in YPR medium to mid log before galactose was added to induce Sod2p synthesis. The Sod2p polypeptide in whole cell lysates (panel A and T) or in mitochondrial (M) and post-mitochondrial supernatant/cytosolic (C) fractions was analyzed by immunoblot as described in the legend to Fig. 1. The U-position of unprocessed Sod2p containing the mitochondrial leader sequence and P-position of processed or mature mitochondrial Sod2p are shown. A, galactose induction proceeded for the indicated time points in min prior to preparation of cell lysates. Lane 1 contains unprocessed Sod2p (U) used as a molecular weight control. B, galactose induction proceeded for 3 h. Where indicated, 20 M CCCP and 0.05% (v/v) ␤-ME were added at t ϭ 15 and t ϭ 30, respectively. C, galactose induction in the presence of 20 M CCCP proceeded for 3 h. Cells were either harvested (t ϭ O) or treated for 45 min with 0.05% (v/v) ␤-ME in the presence or absence of 100 g/ml cycloheximide (to block protein synthesis) as indicated. Cycloheximide was added Ϸ15 min prior to the addition of ␤-ME.
UTR was found to mediate association with the mitochondrial outer membrane (47), and the same may be true for Sod2p. In any case, translation of S. cerevisiae Sod2p appears to occur at the site of the mitochondria to facilitate co-translational import of the protein into mitochondria (31,32,34,47).
Once imported into mitochondria, Sod2p needs to rapidly acquire its metal, because a pre-existing inactive pool of mitochondrial Sod2p failed to acquire manganese in vivo. This is consistent with in vitro studies preformed on the homologous bacterial Mn-SOD enzymes (44,45,48). When folded, these enzymes cannot acquire the metal. Activation with manganese is only possible when the proteins were thermally denatured and the metal was present during, but not after, refolding of the polypeptide (44,45,48). With Mn-SOD molecules, metal access to the active site is guarded by a large transition state barrier that only becomes accessible when the polypeptide is unfolded (44,45,48). We propose that, in the case of eukaryotic manganese SOD2, the requisite protein unfolding step is achieved by mitochondrial import.
Passage of polypeptides through the inner mitochondrial membrane requires extensive protein unfolding, followed by refolding once in the matrix (49,50). But in the case of eukaryotic SOD2, manganese insertion must take place prior to refolding. SOD2 has four amino acid ligands for manganese at positions 52, 107, 198, and 194 in the S. cerevisiae enzyme. Perhaps the metal begins to insert at the N-terminal ligands as the polypeptide emerges from the inner membrane (see schematic, Fig. 5). An accessory protein may also be involved to prevent protein folding prior to metal insertion. Currently, the only mitochondrial protein known to facilitate manganese activation of SOD2 is S. cerevisiae Mtm1p (19). Located in the inner membrane of mitochondria, Mtm1p is in a perfect position to assist in Sod2p metalation, as the polypeptide enters the mitochondrial matrix. The precise activity of Mtm1p is not known but may involve direct insertion of the manganese cofactor or maintaining Sod2p in a conformation that is competent for metal activation. These possibilities are under current investigation.
Overall, our studies have provided a more detailed mechanistic picture for the post-translational activation of SOD2 with manganese. As shown in our model of Fig. 5, the ribosomes for S. cerevisiae Sod2p synthesis are juxtaposed to the outer mitochondrial membrane (46,47). This allows for the coupling of Sod2p synthesis and mitochondrial import. As the polypeptide emerges from the inner membrane, manganese ions are inserted through a process that is facilitated by Mtm1p in the inner membrane and perhaps other accessory proteins as well. Last of all, the manganese-containing protein is folded in the mitochondrial matrix in a stable quaternary tetramer. With the clear importance of SOD2 in eukaryotic survival and fitness (35)(36)(37)(38)(39)(40)(41)(42), these ordered steps must be carefully controlled.  (47), which would facilitate co-translational import of Sod2p into mitochondria. As Sod2p emerges into the mitochondrial matrix, the protein immediately acquires its manganese cofactor prior to folding of the polypeptide. Manganese binding may begin prior to complete translocation of the Sod2p polypeptide. Manganese insertion requires the action of Mtm1p, a member of the mitochondrial carrier family of transporters in the inner membrane (19). The precise substrate for transport by Mtm1p is not known (circled X), but possibilities include manganese itself or a solute that facilitates manganese insertion. Following insertion of manganese, the Sod2p polypeptide folds and monomers associate to form the active tetrameric enzyme. H and D, manganese binding ligands H52, H107, D194, and H198.