Manganese Superoxide Dismutase in Saccharomyces cerevisiae Acquires Its Metal Co-factor through a Pathway Involving the Nramp Metal Transporter, Smf2p*

Eukaryotes express both copper/zinc (SOD1)- and manganese (SOD2)-requiring superoxide dismutase enzymes that guard against oxidative damage. Although SOD1 acquires its copper through a specific copper trafficking pathway, nothing is known regarding the intracellular manganese trafficking pathway for SOD2. We demonstrate here that in Saccharomyces cerevisiae cells delivery of manganese to SOD2 in the mitochondria requires the Nramp metal transporter, Smf2p. SOD2 activity is greatly diminished in smf2 (cid:1) mutants, even though the mature SOD2 polypeptide accumulates to normal levels in mitochondria. Treating smf2 (cid:1) cells with manganese supplements corrected the SOD2 defect, as did elevating intracellular manganese through mutations in PMR1. Hence, manganese appears to be inaccessible to mitochondrial SOD2 in smf2 mutants. Cells lacking SMF2 also exhibited defects in manganese-dependent steps in protein glycosylation and showed an overall decrease in steady-state levels of accumulated manganese. By comparison, mutations in the cell surface Nramp transporter, Smf1p, had very little impact on manganese accumulation and trafficking. Smf2p resides in intracellular vesicles and shows no evidence of plasma membrane localization, even in an end4 mutant blocked for endocytosis. We propose a model in which Smf2p-containing vesicles

Superoxide dismutases (SOD) 1 are critical anti-oxidant defense enzymes that catalyze the disproportionation of superoxide anion to oxygen and hydrogen peroxide. Eukaryotic cells possess two evolutionary distinct forms of SOD: a copper-and zinc-containing SOD (SOD1) found mainly in the cytosol (1) and a manganese-containing SOD (SOD2) that localizes strictly to the mitochondrial matrix (2). Because both forms of SOD require a transition metal co-factor, enzyme activity in cells may be controlled through availability of the cognate metal ion. Indeed this is the case for the copper-requiring SOD1, where the enzyme relies on components of a defined copper trafficking pathway for acquisition of its metal ion (for reviews see Refs. [3][4][5][6]. However, nothing is known regarding the mechanism by which SOD2 receives its manganese cofactor in vivo. Encoded in the nucleus, SOD2 is predicted to acquire its manganese after import into the mitochondrial matrix. The manganese-binding residues of SOD2 occur at distant regions of the polypeptide (7), and the metal binding site is not likely to remain intact during the protein-unfolding process needed for mitochondrial membrane translocation. Therefore, an intricate manganese trafficking pathway must warrant the accurate delivery of the metal from the cell surface to the mitochondria. To date, none of the factors that participate in this pathway have been identified.
The bakers' yeast Saccharomyces cerevisiae represents an ideal eukaryotic system in which to unravel metal trafficking pathways. This organism has been extremely powerful for elucidating the copper acquisition pathway for copper/zinc SOD1 (8 -11), and a number of candidates for manganese trafficking have already been identified. The transport of manganese (and calcium) into the secretory pathway is accomplished by a Golgilocalized P-type ATPase known as Pmr1p (12)(13)(14)(15). Atx2p is another manganese homeostasis factor that localizes to intracellular vesicles (16) whereas the uptake of manganese into yeast vacuoles is accomplished by vacuolar Ccc1p (17,18). Manganese homeostasis is also affected by the product of S. cerevisiae CDC1 (19 -21); however, the site of CDC1 action is not defined.
A cell surface manganese transporter known as Smf1p has also been identified for yeast. Smf1p is a member of the Nramp family of metal transporters that have been conserved from bacteria to mammals (22). There are three Nramp transporters in yeast, Smf1p, Smf2p, and Smf3p. Smf3p affects iron trafficking (23) whereas Smf1p and Smf2p appear to function in manganese homeostasis. Overexpression of either Smf1p or Smf2p enhances manganese uptake (24), and the two transporters are co-regulated at the post-translational level by manganese ions (23,25). Yet Smf1p and Smf2p are not redundant. Unlike Smf1p, Smf2p shows no evidence of plasma membrane localization but appears to reside in intracellular vesicles (23). smf1⌬ mutants exhibit a greater sensitivity to metal chelators than smf2⌬ mutants (24), and Smf1p is more effective in proton-coupled metal uptake when expressed in Xenopus oocytes (26). These observations would suggest that Smf1p is a limiting factor for cell surface uptake and trafficking of manganese, whereas Smf2p may be secondary.
In this paper, we have surveyed the known manganese homeostasis factors in yeast for their possible role in trafficking manganese to mitochondrial SOD2. Of all the proteins examined, only Smf2p was found to be critical for activating SOD2 with manganese. The role of Smf2p in manganese trafficking was not specific to the mitochondria, because manganese-dependent enzymes in the Golgi also rely on Smf2p for activity. Our observations are consistent with a model in which the intracellular Smf2p-containing vesicles play a pivotal role in regulating manganese availability to mitochondrial SOD2 as well as other cellular targets.
Stocks of strains were maintained in enriched yeast extract, peptonebased medium supplemented with 2% glucose (YPD) (28) at 30°C unless specified otherwise. A synthetic metal-depleted minimal defined medium was prepared through use of Chelex 100 resin (Bio-Rad) (25,29). YPLG (yeast extract and peptone-based medium with 2% lactate and 0.1% glucose) was employed for studies of invertase activity.
Plasmids-To create bacterial expression vectors for recombinant yeast SOD2, we first isolated a genomic clone of yeast SOD2 by amplifying SOD2 nucleotides Ϫ557 to ϩ894 by PCR, digesting with BamHI and SalI at the termini of the PCR fragment, and ligating the product into the yeast shuttle vector pRS414 (30), generating vector pEL101. The SOD2 gene fragment of pEL101 was then subcloned into the Escherichia coli expression vector pET21a(ϩ) (Novagen), generating pEL001 (for expression of full-length precursor SOD2). To obtain the expression vector for mature yeast SOD2 (amino acids 27 to the stop codon), two NdeI sites were introduced sequentially in vector pEL101 at the start codon of SOD2 (AGGATG 3 CATATG) and at nucleotide position ϩ76 (AGAACC 3 CATATG) by site-directed mutagenesis (QuikChange, Stratagene). The plasmid was digested with NdeI and religated such that the sequence encoding the first 26 amino acids was removed, generating vector pEL104. The mature SOD2-containing fragment was then liberated by NdeI/SalI digestion of pEL104 and subcloned into the E. coli expression vector pET21a(ϩ), generating pEL002. The sequence integrity of all plasmids was confirmed by double-stranded DNA sequencing (Core facility, Johns Hopkins Medical Institutions).
Biochemical and Immunofluorescence Techniques-To monitor SOD activity, cells were propagated non-shaking in YPD medium to an A 600 of 1.5, harvested and washed before homogenization by glass-bead agitation in extraction buffer (10 mM NaPO 4 , pH 7.8, 5 mM EDTA, 5 mM EGTA, 50 mM NaCl, 0.1% Triton X-100, 100 g/ml phenylmethylsulfonyl fluoride, 1 g/ml pepstatin A, 1 g/ml leupeptin). Tween 20 was added to a final concentration of 1.0%, and samples were vortexed for 15 s prior to non-denaturing gel electrophoresis, followed by analysis of SOD activity by staining with nitro blue tetrazolium (Sigma Chemical Co.) as previously described (31,32).
Invertase assays were conducted on cells pre-grown in YPD medium to mid log phase. Cells were harvested, washed, and inoculated to an A 600 of 0.3 in YPLG medium with or without the addition of 10 mM CaCl 2 and/or 20 M MnCl 2 . Following 5 h of incubation, the cells were washed, harvested, and lysed by glass-bead agitation in the extraction buffer described above for SOD assays. Seventy micrograms of cell lysate protein was subjected to electrophoresis at 4°C for 7 h at 80 V in a 4% native polyacrylamide gel (12,33). Invertase activity (conversion of sucrose to glucose) was detected by staining with tetrazolium red according to Gabriel and Wong (34).
To obtain mitochondrial and post-mitochondrial cell fractions, cells grown in YPD medium were converted to spheroplasts prior to gentle lysis by Dounce homogenization as described previously (35,36). The resulting cell lysates containing intact mitochondria were subsequently fractionated into mitochondrial and post-mitochondrial fractions by differential centrifugation (37).
For immunoblot analysis of SOD2, total cell lysates or mitochondrial and post-mitochondrial supernatant fractions prepared as described above were subjected to electrophoresis on a 14% SDS-PAGE gel and analyzed by Western blot using an anti-yeast SOD2 antibody diluted 1:1000 and a secondary anti-rabbit IgG diluted 1:12,500 (Amersham Biosciences, Inc.). As needed, antibodies against mitochondrial matrix Mas2p (diluted 1:10,000) and cytosolic Pgk1p (Molecular Probes diluted 1:1000) were used as controls. Detection employed the ECL kit (Amersham Biosciences, Inc.) according to the manufacturer's specifications. For production of recombinant SOD2 proteins, E. coli transformants harboring the aforementioned pET expression plasmids were pre-grown to log phase in LB medium supplemented with 100 g/ml ampicillin; isopropyl-1-thio-␤-D-galactopyranoside was added to a final concentration of 0.4 mM and following 3 h of induction, the bacteria are harvested and lysed by vigorous vortexing in the extraction buffer described above for SOD assays. The lysate proteins were cleared by centrifugation prior to SDS-PAGE and immunostaining.
To monitor steady-state manganese levels in whole cells, strains were grown in YPD to an A 600 of 1.5, harvested, and washed in water prior to atomic absorption analysis. Twenty microliters of samples containing 10 5 -10 6 cells was applied to a PerkinElmer Life Sciences model 4000 graphite furnace atomic-absorption spectrophotometer, and manganese content measured according to the manufacturer's specifications. Concentrations of total cellular manganese were calculated based on the approximated yeast cell volume of 7 ϫ 10 Ϫ14 liter (38). Measurements of manganese content in mitochondrial fractions or whole cell lysates was similarly determined except that cells were grown to stationary phase (A 600 ϭ 4.0) to improve yield of mitochondria, and manganese concentrations were determined relative to extract protein.
For immunofluorescence studies, the end4 ts and isogenic wild type parental strain expressing Smf2-HA or Ste6-HA were grown aerobically in minimal defined medium without supplementation of manganese and iron. Growth in this medium minimizes vacuolar degradation of Smf2p (23,25) and enhances immunodetection of the protein. Because of the temperature-sensitive nature of end4 ts mutants, cells were all first pre-grown at 25°C to an A 600 of 0.5, and were then either shifted to the non-permissive temperature at 37°C for 40 min or remained at 25°C prior to preparation for immunofluorescence microscopy. Immunofluorescence microscopy was conducted as described previously (39) except fixed cells were digested with 100 g/ml Zymolyase 20T (ICN) for 30 min at 30°C. Ste6-HA and Smf2-HA were probed with a mouse anti-HA primary antibody (12CA5) as before (39) except 1:8,000 antibody dilution was used for Smf2-HA. An anti-mouse antibody conjugated to Cy3 (Jackson ImmunoResearch Laboratory Inc.) was used as a secondary antibody. Fluorescence and Normarski Differential Interference Contrast microscopy was carried out on a Zeiss Axiovert 135TV microscope (Microscopy Facility, Johns Hopkins Medical Institutions) at a magnification of ϫ 1000.

RESULTS
The Role of S. cerevisiae SMF2 in the Trafficking of Manganese to Mitochondrial SOD2-A number of S. cerevisiae genes have previously been shown to participate in cellular manganese homeostasis, for example, ATX2 (16), CCC1 (17,18), CDC1 (19 -21), and PMR1 (12,13). To test whether these genes are required for delivery of manganese to mitochondrial SOD2, we monitored SOD2 activity in atx2⌬, ccc1⌬ and pmr1⌬ null mutants, in the cdc1-1-temperature-sensitive mutant and in strains overexpressing CCC1 and PMR1. In all cases, we found that modulating activity of these genes had no detectable effect on SOD2 activity (data not shown). Therefore, the ATX2-, CCC1-, CDC1-, and PMR1-encoded proteins do not seem critical for delivering manganese to mitochondrial SOD2.
S. cerevisiae expresses three members of the Nramp family of metal transporters (Smf1p, Smf2p, and Smf3p) and we addressed whether these were needed for SOD2 activity. In the experiment of Fig. 1A, lysates from cells harboring single or multiple mutations in the three SMF genes were applied to a native polyacrylamide gel and the corresponding SOD2 activities were analyzed by staining with nitro blue tetrazolium (31,32). Surprisingly, a mutation in SMF1 encoding the cell surface transporter for manganese had no effect on SOD2 activity and a similar result was obtained with smf3⌬ mutants (Fig. 1A). However, inactivation of SMF2 led to a striking decrease in SOD2 activity. The double smf1⌬ smf2⌬ and triple smf1⌬ smf2⌬ smf3⌬ mutants likewise exhibited SOD2 inactivation (Fig. 1A). This effect was specific to the manganese containing SOD2, as the activity of copper/zinc-containing SOD1 was not affected in any of the mutants (Fig. 1A). The decrease in SOD2 activity observed with smf2⌬ mutants was not due to loss of the SOD2 polypeptide. When the cell lysates of Fig. 1A were applied to a denaturing gel and analyzed by immunoblotting with an anti-SOD2 antibody, the levels of the SOD2 polypeptide remained constant in all the smf mutants (Fig. 1B).
Mature SOD2 is targeted to the mitochondrial matrix by a Ϸ26-amino acid N-terminal pre-sequence that is proteolytically removed following mitochondrial import (40). Because protein mislocalization may lead to SOD2 inactivation, we addressed whether smf2 mutations affected the mitochondrial import and/or processing of SOD2. Whole cell lysates were fractionated into intact mitochondria and a post-mitochondrial supernatant (largely cytosolic) fraction by differential centrifugation. As markers for fractionation, cytosolic Pgk1p, and mitochondrial matrix Mas2p were employed. As seen in the immunoblot of Fig. 2A, the SOD2 polypeptide was restricted to the mitochondrial fraction of both the wild type strain and the smf2⌬ mutant. To test whether the SOD2 mitochondrial pre-sequence was properly cleaved in smf2⌬ cells, the electrophoretic mobility of the SOD2 polypeptide was compared with that of a recombinant SOD2 precursor (all 233 amino acids of SOD2) and mature SOD2 protein (amino acid 27 to the stop codon), both prepared from E. coli. As seen in Fig. 2B, the SOD2 polypeptide expressed in yeast smf2⌬ cells migrated to the same position in a denaturing gel as recombinant mature SOD2, indicating that the smf2⌬ mutation does not interfere with processing of the SOD2 polypeptide pre-sequence.
Because the SOD2 polypeptide is accurately expressed and targeted to the mitochondria in smf2⌬ mutants, the lack of enzymatic activity might be due to low manganese availability. If so, supplementing the growth medium with manganese might bypass the smf2⌬ defect. As seen in Fig. 3, SOD2 activity in wild type strains was not affected by manganese supplementation; however, addition of manganese corrected the smf2⌬ defect and restored SOD2 activity to normal levels. This would suggest that manganese availability to SOD2 is very low in smf2⌬ mutants. To address this further, we attempted to bypass the requirement for Smf2p by genetically manipulating intracellular manganese. PMR1 encodes a P-type manganese-and calcium-transporting ATPase (12)(13)(14)(15), and mutants of PMR1 have been shown to accumulate high levels of manganese, apparently in the cytosol (41,42). As seen in Fig. 4A, a mutation in PMR1 effectively elevated total manganese levels in a smf2⌬ deletion strain. Moreover, this pmr1⌬ mutation suppressed the smf2⌬ defect in SOD2, because the pmr1⌬ smf2⌬ double mutant exhibited wild type levels of SOD2 activity (Fig. 4B). Together with the manganese supplementation studies shown in Fig. 3, these experiments indicate that SOD2 is inactivated in smf2⌬ strains due to low manganese availability.
The Golgi Also Appears Starved for Manganese in the smf2⌬ Mutant-We addressed whether the effect of smf2⌬ mutations was specific for mitochondrial SOD2. For example, manganese needs to be delivered to the Golgi apparatus where the metal activates a variety of sugar transferases involved in protein processing (43)(44)(45)(46). In yeast, a useful marker for Golgi manganese is invertase, an enzyme that is modified by N-glycosylation via manganese-dependent mannosyltransferases (45)(46)(47)(48). Defects in invertase glycosylation can be scored by electrophoretic shifts on a non-denaturing gel. In the experiment of Fig. 5, the mobility of active invertase was monitored by staining with tetrazolium red. The pmr1⌬ mutants lacking the Golgi manganese pump are defective for invertase glycosylation (12), and the enzyme exhibits a shift in electrophoretic mobility (Fig.  5A, lane 2). Notably, invertase was also underglycosylated in smf2⌬ mutants (Fig. 5A, lane 4), although the shift in mobility was not as severe as with pmr1⌬ mutants. In comparison, a   FIG. 1. Gene disruption of SMF2 but not SMF1 or SMF3 leads to SOD2 inactivation. Total cell lysates were prepared from AA255 (wild type) and the isogenic EL201 (sod2⌬), XL112 (smf1⌬), XL117 (smf2⌬), MP112 (smf3⌬), XL131 (smf1⌬, smf2⌬), and MP113 (smf1⌬, smf2⌬, smf3⌬) strains grown in YPD medium to log phase. A, 40 g of lysate protein was analyzed by native gel electrophoresis and SOD activity staining using nitro blue tetrazolium (31,32). Positions of SOD1 and SOD2 are indicated. B, 70 g of lysate protein was analyzed by denaturing gel electrophoresis and immunostaining with an anti-SOD2 antibody. Inactivation of SOD2 associated with smf2⌬ mutations is highly reproducible and was also observed in two unrelated yeast strain backgrounds.
FIG. 2. SOD2 is targeted to the mitochondria and processed normally in smf2⌬ mutants. A, total cell lysates (T) from the indicated yeast strains were fractionated by differential centrifugation into post-mitochondrial supernatants (P) and mitochondria (M) as previously described (36). 60 g of total (T) cell lysate protein, and the same cell equivalents of post-mitochondrial supernatants and mitochondria, were subjected to Western blot analysis using antibodies directed against SOD2, the mitochondrial processing protease (Mas2p), or cytosolic phosphoglycerate kinase (Pgk1p). B, yeast SOD2 produced either as recombinant forms from E. coli or endogenously from yeast cells was examined by electrophoresis on a 14% SDS-PAGE gel followed by immunoblotting using an anti-SOD2 antibody. Recombinant yeast SOD2, 10 or 100 ng of lysate proteins was analyzed from E. coli transformed with pEL001-or pEL002-expressing yeast precursor SOD2 (containing mitochondrial pre-sequence) and mature SOD2 (lacking pre-sequence), respectively. Endogenous yeast SOD2, 30 g of lysate protein was examined from the indicated yeast strains. Strains utilized are as described in Fig. 1.   FIG. 3. Supplementing the smf2⌬ mutant with manganese in the growth medium restored SOD2 activity to wild type levels. Yeast strains AA255 (WT) and XL117 (smf2⌬) were grown in YPD supplemented with 0 -40 M MnCl 2 , and SOD activity was assayed as in Fig. 1A. single null mutation in either SMF1 or SMF3 had no effect on invertase glycosylation (lanes 3 and 5). Mutations in SMF1 did enhance the defect of strains already lacking SMF2 (lane 6), indicating that Smf1p may contribute to invertase glycosylation when Smf2p is absent. In any case, Smf2p is clearly critical for efficient processing of invertase.
Because both pmr1⌬ and smf2⌬ mutations inhibit invertase glycosylation, we analyzed the effects of their combined mutations. As seen in Fig. 5A, lane 8, the pmr1⌬ smf2⌬ strain exhibited the same invertase defect as a single pmr1⌬ mutant.
If the pmr1⌬ smf2⌬ defect had been more severe, we would have detected an enhanced shift in invertase mobility, because pmr1⌬ mutations do not completely block glycosylation (12). The effects of pmr1⌬ and smf2⌬ mutations are clearly not additive, suggesting that the encoded proteins operate in the same pathway to ensure proper glycosylation of invertase (see "Discussion").
Previous studies indicate that deficiency of either manganese or calcium in the secretory pathway will lead to underglycosylation of invertase (13,47). In accordance with these results, we found that the supplementation of either manganese or calcium to the growth medium partially alleviated the invertase glycosylation defect of pmr1⌬ mutants (Fig. 5B, lanes 6  and 7). However, in the smf2⌬ mutant, only the addition of manganese, not calcium, corrected the invertase defect (Fig.  5B, lanes 10 and 11). Overall, the data strongly indicate that Smf2p is necessary for proper delivery of manganese to the secretory pathway and that mitochondrial SOD2 is not the only downstream target of this Nramp metal transporter.
Deletion of SMF2 Leads to a Global Decrease in Steady-state Levels of Manganese-The effect of smf2⌬ mutations on manganese enzymes in both the mitochondria and the Golgi suggests that this Nramp transporter may act as a global regulator of manganese accumulation and availability. To investigate this further, we measured the total level of manganese accumulated in the various smf⌬ mutants of yeast using atomic absorption spectroscopy. As shown in Fig. 6A, a smf2⌬ deletion led to a striking decrease in steady-state levels of cellular manganese when compared with an isogenic wild type strain. Under the same conditions, a deletion in the cell surface Smf1p transporter only caused a marginal decrease in total cellular manganese. Consistent with earlier studies (23), the smf3⌬ null mutant exhibited an elevated level of intracellular manganese (Fig. 6A). We have also monitored manganese in isolated mitochondria and find that mutations in SMF2 decrease   FIG. 4. Effects of a pmr1⌬ mutation on manganese accumulation and SOD2 activity in smf2⌬ mutants. The indicated yeast strains were grown in YPD to log phase. A, steady-state manganese content of whole cells was measured by atomic absorption spectroscopy. Results represent averages of triplicate samples and three experimental trials. Error bars indicate standard deviation. B, cell lysates were prepared and SOD2 activity (top) and SOD2 protein (bottom) were assayed as in Fig. 1; only SOD2 activity was shown for simplicity. Strains utilized: AA255 (WT), XL117 (smf2⌬), EL202 (pmr1⌬), and EL203 (smf2⌬, pmr1⌬).
FIG. 5. The effects of smf2 mutations on invertase glycosylation. The indicated yeast strains were grown in YPD medium to early log phase and were then incubated for 5 h in YPLG medium to derepress invertase expression. 70 g of protein from total cell lysates was subjected to native gel electrophoresis, and invertase activity (conversion of sucrose to glucose) was monitored by staining with tetrazolium red (34). A, cells were grown in medium not supplemented with manganese or calcium. B, cells were grown in YPLG medium supplemented with either 10 mM CaCl 2 and/or 20 M MnCl 2 as indicated. Strains utilized are as described in Figs. 1 and 4.   FIG. 6. Deletion of SMF2 leads to a global decrease in steady state levels of manganese. A, the indicated yeast strains were grown to log phase in YPD medium, and the manganese content of whole cells was analyzed by atomic absorption spectroscopy as in Fig. 4A. B, cells were grown in YPD medium to stationary phase to facilitate isolation of mitochondria. Cell lysates were prepared, and mitochondria were isolated by differential centrifugation. Manganese content of isolated mitochondria (M) and unfractionated cell lysates (T) was monitored by atomic absorption spectroscopy. All data represent averages of triplicate samples and three experimental trials. Error bars indicate standard deviation. Strains utilized are as described in Fig. 1. mitochondrial levels of manganese (Fig. 6B). Therefore, it appears that, of the three Nramp metal transporters, Smf2p is the most critical for controlling intracellular levels of manganese.
Localization of SMF2-The dramatic effect of smf2 mutations on manganese accumulation was unexpected for a transporter that localizes to intracellular vesicles. As one possibility, Smf2p might actually operate at the cell surface but with a residence time too short to be detected. In yeast, a number of cell surface transporters are rapidly internalized and recycled through endocytosis, a classic example being the ABC transporter for mating factor, Ste6p (49). To test whether Smf2p also transiently resides at the cell surface, we exploited an end4 ts mutant, which is blocked for endocytosis at the non-permissive temperature (37°C). This strain and its isogenic wild type parent were transformed with an epitope-tagged version of Smf2p (Smf2-HA (23)) that is functional for complementing the smf2⌬ defect in SOD2 activity (not shown). An epitope-tagged version of Ste6p (Ste6-HA (39)) was used as control. As seen in Fig. 7 (A and B), both Ste6-HA and Smf2-HA exhibited an intracellular punctate staining in the END4 ϩ wild type strain, and there was no evidence of plasma membrane staining. By comparison, Ste6-HA accumulated at the cell surface in the end4 ts cell at the non-permissive temperature (Fig. 7A), con-sistent with earlier findings (39). Yet, under the same experimental conditions, there was no detectable accumulation of Smf2-HA at the plasma membrane and the protein remained at the intracellular sites in the end4 ts cells (Fig. 7B). We also failed to detect cell surface Smf2-HA when the end4 ts mutant was incubated at the non-permissive temperature for extended periods (i.e. 2 h; data not shown). Based on this result, Smf2p does not appear to reside at the cell surface, but rather acts at an intracellular location to control cellular accumulation and trafficking of manganese ions. DISCUSSION In eukaryotes, the multitude of enzymes that require heavy metals as co-factors are housed in diverse cellular locations. As such, the intracellular dissemination of these metals must be tightly controlled. This is clearly the case for copper where specific membrane transporters and soluble copper chaperones cooperate to distribute the metal to target enzymes at various cellular sites (3)(4)(5)(6). In comparison, very little is understood regarding the trafficking of other metals such as manganese. In fact, prior to this study, no factors have been identified that help deliver manganese to superoxide dismutase in the mitochondria.
We show here that the transport of manganese to mitochondrial SOD2 is absolutely dependent on yeast Smf2p, a member of the Nramp family of metal transporters. The SOD2 polypeptide accumulates normally in the mitochondria of cells lacking SMF2, but the enzyme is largely inactive due to low availability of manganese. The role of Smf2p in manganese trafficking is not restricted to mitochondria, because this transporter was also needed to efficiently activate manganese enzymes in the Golgi. Moreover, a null mutation in SMF2 was seen to cause a cell wide loss in steady-state levels of manganese. Smf2p is therefore a critical player in manganese accumulation and distribution pathways.
Even though Smf2p inactivation had a widespread effect on intracellular manganese, we obtained no evidence that this is a cell surface transporter for manganese. By immunofluorescence microscopy, there was an absence of detectable cell surface Smf2p even in mutants blocked for endocytosis. One might argue that the epitope tag may interfere in localization of Smf2p. However, precisely the same tag at the same position does not interfere with the cell surface localization of Smf1p (Ϸ60% identical in amino acid sequence to Smf2p) (25,50), and Smf2-HA is fully functional. Therefore, physiological Smf2p appears to reside at a vesicular-like compartment. The nature of these vesicles is not presently known and will form the basis for future studies. Vesicular structures have previously been implicated in metal trafficking (e.g. in the storage and release of zinc (51-54)), and the Smf2-containing vesicles may harbor manganese. Because Nramp transporters in general are predicted to transport metals in the direction of the cytosol (22,26), Smf2p may very well transport manganese out of the vesicles into the cytosol for trafficking to various cell compartments (Fig. 8). It would therefore seem curious that inactivation of Smf2p leads to a total drop in cellular manganese levels. However, a similar effect has been observed with the mammalian Nramp protein DMT1 (also known as DCT1 and Nramp2). DMT1 localizes to intracellular endocytic vesicles in non-intestinal cells, and even though DMT1 is not at the cell surface, its inactivation can lead to a decrease in cellular uptake of iron (55)(56)(57).
We were surprised to find that inactivation of the yeast cell surface Nramp transporter, Smf1p, had no significant consequence on manganese homeostasis. The only effect observed was an enhanced deficiency in protein glycosylation when smf1⌬ mutations were introduced in a strain already lacking Smf2p. Based on previous studies, smf1⌬ mutants appeared more starved for manganese than smf2⌬ mutants, because they exhibited a greater sensitivity to metal chelation by EGTA (26). However, our analysis of manganese-dependent enzymes and total manganese accumulation implies that the converse is true. Therefore, it is possible that the EGTA sensitivity of smf1 mutants reflects chelation and depletion of other cations, and accordingly, Nelson and co-workers (24) have shown that the smf-EGTA sensitivity is ameliorated by supplements of copper as well as manganese. We suggest that Smf1p does have the capacity for cell surface uptake of manganese, but it is certainly not the only means by which yeast obtain extracellular manganese. Other cell surface transporters, such as the Fet4p divalent metal transporter (58), may also contribute to plasma membrane uptake of the metal.
The accurate delivery of manganese to enzymes in the secretory pathway requires both Smf2p and Pmr1p, the Golgi P-type ATPase for manganese (47,59). We observed no additive effect of smf2⌬ and pmr1⌬ mutations on protein glycosylation, indicating that these transporters act in the same pathway for routing manganese to the Golgi. Yet Smf2p cannot function in the same cellular compartment as Pmr1p, because the proteins are predicted to transport the metal in opposite directions and have similar effects on protein glycosylation. Moreover, unlike smf2⌬ mutations, inactivation of PMR1 does not block manganese delivery to mitochondrial SOD2. In fact, pmr1⌬ mutations suppress the smf2⌬ defect in mitochondrial manganese, presumably though a build-up of cytosolic manganese. We therefore invoke a model in which Smf2-containing vesicles lie upstream of Golgi-localized Pmr1p and also upstream of the mitochondria, facilitating delivery of manganese ions to both compartments (Fig. 8). Furthermore, this pathway is likely to be conserved in mammalian cells where functional homologues for both Pmr1p and Smf2p exist (60,61). Yet this picture is far from complete. Smf2p represents the first protein identified to date that participates in the manganese delivery pathway for mitochondrial SOD2 and others are likely to exist, including putative manganese chaperones that may shuttle the metal between organelles and also inside the mitochondrial matrix.
Molecular genetic studies in the bakers' yeast are likely to continue providing new insight into the conserved players of manganese trafficking pathways. FIG. 8. A model for Smf2p trafficking of manganese ions in yeast. Smf2p resides at a vesicular-like compartment and is predicted to transport manganese in the direction of the cytosol. The means by which manganese is delivered to the Smf2 vesicles is not known but may involve cell surface transporters for the metal (indicated by dotted arrows). The Smf2p-containing vesicles lie upstream in the manganese delivery pathways for both SOD2 in the mitochondria (Mito) and mannosyltransferases (MNNs) in the Golgi. The P-type ATPase, Pmr1p, is downstream of Smf2p and translocates the metal directly into the lumen of the Golgi (47,59). A putative mitochondrial membrane transporter for manganese (blue circle in Mito) is also predicted to lie downstream of Smf2p. The role of Smf1p in manganese trafficking may become important under cases of manganese starvation stress, e.g. when SMF2 is deleted. See text for details.