The Effects of Glutaredoxin and Copper Activation Pathways on the Disulfide and Stability of Cu,Zn Superoxide Dismutase*

Mutations in Cu,Zn superoxide dismutase (SOD1) can cause amyotrophic lateral sclerosis (ALS) through mechanisms proposed to involve SOD1 misfolding, but the intracellular factors that modulate folding and stability of SOD1 are largely unknown. By using yeast and mammalian expression systems, we demonstrate here that SOD1 stability is governed by post-translational modification factors that target the SOD1 disulfide. Oxidation of the human SOD1 disulfide in vivo was found to involve both the copper chaperone for SOD1 (CCS) and the CCS-independent pathway for copper activation. When both copper pathways were blocked, wild type SOD1 stably accumulated in yeast cells with a reduced disulfide, whereas ALS SOD1 mutants A4V, G93A, and G37R were degraded. We describe here an unprecedented role for the thiol oxidoreductase glutaredoxin in reducing the SOD1 disulfide and destabilizing ALS mutants. Specifically, the major cytosolic glutaredoxin of yeast was seen to reduce the intramolecular disulfide of ALS SOD1 mutant A4V SOD1 in vivo and in vitro. By comparison, glutaredoxin was less reactive toward the disulfide of wild type SOD1. The apo-form of A4V SOD1 was highly reactive with glutaredoxin but not SOD1 containing both copper and zinc. Glutaredoxin therefore preferentially targets the immature form of ALS mutant SOD1 lacking metal co-factors. Overall, these studies implicate a critical balance between cellular reductants such as glutaredoxin and copper activation pathways in controlling the disulfide and stability of SOD1 in vivo.

The copper-and zinc-containing superoxide dismutase (SOD1) 6 (1) protects eukaryotic cells against oxidative stress by scavenging toxic superoxide anions. Enzyme catalysis is carried out by the copper co-factor, and in cells, this copper ion is delivered to SOD1 through regimented metal trafficking pathways. One pathway involves the CCS copper chaperone (2) that physically docks with SOD1 (3,4) and transfers its copper cargo in an oxygen-dependent fashion (5). Metazoan SOD1 can also acquire copper by a second "CCS-independent pathway" (6,7). Although the precise copper donor in this case is not understood, the CCS-independent pathway shows a dependence on glutathione and is particularly sensitive to certain perturbations in the SOD1 structure (6,7). Specifically, prolines at SOD1 positions 142 and 144 (based on Saccharomyces cerevisiae and human SOD1) will block activation by the CCS-independent pathway but not by CCS. S. cerevisiae SOD1 naturally contains these prolines and shows total dependence on CCS (2). SOD1 molecules from higher organisms generally lack prolines 142 and 144 and can be activated independent of CCS (6,7). How CCS-independent activation is blocked by prolines 142 and 144 is uncertain, but these residues have been proposed to disrupt the monomer-dimer equilibrium of apoSOD1 (8).
Although SOD1 is normally a protective enzyme, dominant mutations throughout the SOD1 polypeptide have been linked to the fatal motor neuron disease, amyotrophic lateral sclerosis (ALS). The underlying mechanism is incompletely clear, but a well accepted model involves misfolding of SOD1 mutants and consequential accumulation of toxic SOD1 aggregates (9 -11). The misfolding and instability of SOD1 mutants are curious in that SOD1 is normally a highly stable enzyme (12). Studies with purified enzyme indicate that the bound metal co-factors as well as an intramolecular disulfide in SOD1 help stabilize the structure of the SOD1 homodimer, and this is true for both wild type (WT) and ALS mutant polypeptides (13)(14)(15)(16)(17)(18). During disease, loss of the intramolecular disulfide correlates with misfolding of mutant SOD1 (19), and formation of improper intermolecular disulfides helps aggregate the protein (20 -22). As such, the cellular factors that promote faithful oxidation of the correct intrasubunit disulfide in SOD1 should promote SOD1 stability.
Very little is known regarding the cellular factors that impact on the disulfide of human SOD1. O'Halloran and co-workers (23) found that in the case of yeast SOD1, Cu-CCS promotes oxidation of the disulfide. The role of the CCS-independent pathway in oxidizing the disulfide of metazoan SOD1 has not been addressed previously. Without copper activation, the disulfide cysteines may be reduced, as was shown for yeast SOD1 in vivo (23); however, the thiol reductants that promote this process are unknown. The low redox potential of the cytosol should favor cysteine reduction. Additionally, the thiol oxidoreductases thioredoxin and glutaredoxin (GRX) are known to target a limited number of cysteines in polypeptides (24,25). GRX can resolve mixed disulfides between GSH and a polypeptide cysteine (an S-thionylated polypeptide) (24,26), and can also directly reduce intramolecular disulfides in proteins without an S-thionylated intermediate. Such reduction of intramolecular disulfides has been described for Escherichia coli Grx1 (24,27,28). But to date, no intramolecular disulfide target has been described for the GRXs of eukaryotes. Could SOD1 represent such a target?
Here we describe cellular factors that control the status of the SOD1 disulfide in vivo. By using a yeast expression system, we find that in addition to CCS, the CCS-independent pathway for copper activation helps oxidize the disulfide of human SOD1. When both pathways are blocked, the SOD1 disulfide cysteines are reduced, and in the case of ALS SOD1 mutants A4V, G93A, and G37R, the protein is highly unstable and is subject to degradation. We also demonstrate for the first time that cytosolic glutaredoxins (GRX) can reduce the SOD1 disulfide cysteines and thereby affect stability of certain ALS SOD1 mutants.
For most biochemical analyses, 50 ml of yeast cells were propagated overnight at 30°C in either minimal synthetic dextrose (SD) selecting media (starting A 600 ϭ 0.15) or in enriched YPD (yeast extract, peptone, dextrose) medium (starting A 600 ϭ 0.05). In studies with methionine repression, overnight cultures were diluted in a volume of 200 ml to an ϷA 600 ϭ 0.6 and allowed to grow for an additional 2 h to early log phase. 1 mM methionine was then added, and aliquots of 50 ml were harvested at various time points for cell lysis.
For expression of Grx2p in yeast, the S. cerevisiae GRX2 gene was amplified with primers that introduced BamHI and SalI sites at Ϫ962 and ϩ987 and inserted at these same sites in either pRS413 (HIS3 CEN), creating pCO147, or pRS414 (TRP1 CEN), creating plasmid pVCO147. For expression in E. coli, GRX2 was amplified without the mitochondrial targeting signal (residues 1-34) (32) from pCO147 using primers that introduced an NdeI site at Met-35 and a BamHI site 15 bp after the stop codon. The fragment was inserted at the NdeI and BamHI sites of pET21a (Novagen), creating pCO150.
Purification of Recombinant Yeast Grx2p and Human SOD1-For production of recombinant Grx2p, E. coli strain BL21(DE3) (Novagen) transformed with plasmid pCO150 was grown in 5 liters of LB media and induced for 2.5 h with 1 mM isopropyl ␤-D-thiogalactopyranoside when the A 600 reached 0.7. Harvested cells were stored at Ϫ80°C. A frozen cell paste equivalent to ϳ800 ml of cell culture was lysed by freezingthawing and resuspended in 50 ml of 50 mM Tris-HCl, pH 8.0, 5 mM DTT. Cell debris was removed by centrifugation, and proteins were precipitated with 65-90% (NH 4 ) 2 SO 4 followed by resuspension in 20 mM Tris-HCl, pH 8.0, 25 mM NaCl, 5 mM DTT to a final volume of ϳ3 ml. The protein solution was desalted using a Hi-Trap TM desalting column (GE Healthcare) and then loaded onto a Hi-Trap TM Q-Sepharose XL column (GE Healthcare) equilibrated with 20 mM Tris-HCl, pH 8.0, 5 mM DTT. Recombinant Grx2p was eluted with 0 -50 mM NaCl and concentrated to 37 mg/ml (3.1 mM). Protein purity was demonstrated by SDS-PAGE and Coomassie staining (supplemental Fig. S3A). Protein was stored at Ϫ80°C with 5% glycerol added to the buffer. This recombinant Grx2p was found to efficiently catalyze the reduction of 2-hydroxyethyl disulfide (HED) by GSH with activity (ϳ300 units/mg protein) comparable with E. coli Grx1 (33).
Recombinant human WT and ALS mutant SOD1 proteins were obtained from an S. cerevisiae expression system as described previously (18). Purity of these proteins was demonstrated by a single band on SDS-PAGE and are of the correct mass as demonstrated by electrospray ionization mass spectrometry using a Sciex API III triple quadrupole mass spectrometer (PerkinElmer Life Sciences). Highly purified SOD1 proteins purified in this manner have been used in single crystal x-ray diffraction analyses (15,17), differential scanning calorimetry analyses (34), and in vitro proteasomal digestion assays (35). The latter study further verified the purity of wild type and pathogenic SOD1 proteins isolated in this fashion by using reverse phase high pressure liquid chromatography immediately prior to mass spectrometric analysis. By inductively coupled plasma mass spectrometry, WT and A4V SOD1 contained equal amounts of copper and zinc in the copper-binding site and only zinc in the zinc-binding site (36). Metal-free SOD1 proteins were generated by dialysis of this material at low pH in the presence of EDTA as described (18,37), resulting in apoSOD1 containing Ͻ0.05 eq of both copper and zinc per dimer (18,38), and an oxidized disulfide (as determined by AMS analysis; see Fig. 6) (35).
Biochemical Assays-4-Acetamido-4Ј-maleimidylstilbene-2,2Ј-disulfonic acid (AMS) was used to monitor the disulfide status of SOD1. Yeast cells from a 50-ml culture were washed twice in deionized water, and a 100-l cell pellet was resuspended in 200 l of a GdnHCl buffer (6 M guanidine-HCl, 3 mM EDTA, 0.5% Triton X-100, 50 mM Tris-HCl, pH 8.3) that contained, as needed, 15 mM AMS (Molecular Probes). 100 l of glass beads (425-600 m; Sigma) were added and cells lysed by three cycles of vortexing at room temperature for 2 min, interspersed by 1-min incubations on ice. Extracts were clarified by centrifugation at 10,000 ϫ g for 5 min, and the supernatant was incubated at 37°C for 1 h in the dark. A 50-l aliquot was then applied to a MicroSpin G-25 gel filtration column (Amersham Biosciences), and 32 l of flow-through was prepared for SDS-PAGE by incubating with SDS-DTT loading buffer at room temperature for 7 min, followed by quick clarification by centrifugation.
For AMS modification of SOD1 from fibroblasts, cells were plated at a density of 6.67 ϫ 10 5 cells per 100-mm tissue culture dish and cultured for 24 h at 37°C. Media were aspirated, and cells were washed once in phosphate-buffered saline, followed by cell lysis through addition of 200 l of GdnHCl buffer (see above) containing, as needed, 15 mM AMS. The AMS reaction and subsequent gel filtration proceeded as above. 20 l of column flow-through was boiled in SDS-DTT gel loading buffer and clarified by centrifugation prior to analysis by SDS-PAGE on 14% precast gels (Invitrogen).
Immunoblotting with fibroblasts used an antibody (1:1000 dilution) that only recognizes human SOD1 (6), although detection of SOD1 from yeast lysates generally employed a peptide-derived antibody (39) that recognizes both human and mouse SOD1. Cross-reactivity with nonspecific yeast products of Ϸ39 and Ϸ69 kDa was occasionally observed with early preparations of the antibody (e.g. see supplemental Fig S1A), but not with later preparations of higher titer (e.g. see Fig. 5A). Standard immunoblots (no AMS) and native gels for SOD1 activity used 30 -50 g of yeast cell lysate protein. In nonreducing gels (as in Fig. 5A), the gel was pre-soaked in tris(2-carboxyethyl) phos-phine (TCEP) according to published methods (21) prior to electroblotting. SOD1 activity was monitored by native gel electrophoresis on 12% precast gels and by nitro blue tetrazolium staining as described (40,41).
GRX activity of yeast cell lysates was monitored by the HED assay (33). Yeast cells grown to confluency in selecting media were subjected to glass bead lysis. Cell lysates were heated at 85°C to inactivate glutathione reductase and thioredoxin reductase (42). 15 g of cell lysate protein was added to a 1-ml quartz cuvette containing a 300-l HED reaction mixture (100 mM Tris-HCl, pH 8.0, 2.0 mM EDTA, 1 mM GSH, 0.4 mM NADPH, 6 g/ml glutathione reductase, and 0.7 mM HED). GRX reduction of the HED substrate was measured by continuous monitoring of NADPH consumption at 340 nm over a range of 2 min.
An in vitro assay for reduction of the human SOD1 disulfide by recombinant Grx2p was carried out in a 300-l reaction containing the aforementioned GRX assay constituents with 0.2 mM NADPH rather than 0.4 mM NADPH and no HED. 2-3 M purified human SOD1 with an oxidized disulfide was added as substrate, and following the addition of recombinant Grx2p to a concentration of 0.05 M, the reaction was incubated at 30°C. At specific time points, 10-l aliquots were mixed with 60 l of GdnHCl buffer (see above) containing 15 mM AMS. Following incubation for 1 h at 37°C, samples (Ϸ100 ng of SOD1) were subject to gel filtration and analysis of SOD1 by SDS-PAGE and immunoblot. It is noteworthy that in certain preparations of recombinant SOD1, thiol modification at Cys-111 was incomplete. For example, human WT SOD1 (purchased from Sigma) showed no AMS reactivity at Cys-111 unless pretreated with reducing agents. Apparent oxidation at Cys-111 was also observed upon storage (Ͼ2 weeks at 4°C or Ͼ6 months Ϫ70°C) of apo but not metallated A4V and G93A human SOD1. Similar problems with Cys-111 have been reported elsewhere (35,43). As such, it is important to carry out analysis of the disulfide with freshly prepared samples of purified apoSOD1.

The Disulfide in Human SOD1 Is Oxidized by Both the CCSdependent and -independent Pathways for Copper Loading-
We sought to understand how copper trafficking pathways affect the disulfide of human SOD1. Disulfide status was probed by AMS, which forms stable thioether linkages with free polypeptide cysteines, but is nonreactive toward cysteines bridged in a disulfide. Unlike S. cerevisiae SOD1 and the Cu,Zn-SOD of Caenorhabditis elegans that contain only two cysteines (7), human SOD1 contains four cysteines as follows: the disulfide Cys-57 and Cys-146 cysteines, and Cys-6 and Cys-111. To discern between these, we introduced single and combined mutations at these cysteines. The corresponding SOD1 variants were expressed in yeast, and lysates prepared in 6 N guanidinium were treated with AMS and analyzed by immunoblot. The goal of such a "redox Western" is not to determine total protein levels but to compare the level of oxidized versus reduced cysteines within a particular sample. Typically, the 15.9-kDa human SOD1 migrates on SDS gels at the position of Ϸ21.5 kDa (Fig. 1A, lane 1) (44). When WT SOD1 from CCS1ϩ yeast is reacted with AMS, a major band corresponding to a shift in mobility of Ϸ5.0 kDa is observed (Fig. 1, thin arrow, lane 2). This reflects AMS modification at the nondisulfide Cys-6 and Cys-111 cysteines, because no mobility shift is observed with C6S/C111S SOD1 expressed in CCS1ϩ yeast (Fig. 1, lane  5). The lack of AMS reactivity at Cys-57 and Cys-146 demonstrates that the disulfide is oxidized in CCS1ϩ yeast cells, as would be expected for active SOD1. It is noteworthy that AMS modification at Cys-6 and Cys-111 produces a larger shift in mobility than the 1.07 kDa expected for two AMS moieties. This is due to AMS modification at Cys-111, because C6S SOD1, but not C111S, also produces an aberrantly large shift in mobility (Fig. 1B, also see supplemental Fig. S1B and Tables S1 and S2). Certain small modifications in SOD1 composition can effect anomalous mobility on SDS gels, such as the C57S mutation (Fig. 1A, lane 11) and several other substitutions reported for SOD1 (31,39,41,44). AMS modification at Cys-111 likewise results in anomalous mobility on SDS-PAGE. A more detailed description of the effects of AMS modification at each of the four cysteines is presented in supplemental Tables S1 and S2.
When human SOD1 is expressed in ccs1⌬ null yeast cells lacking the CCS1 gene, a second AMS product bearing increased mobility shift appears (Fig. 1A, heavy arrow, lane 3). This corresponds to additional AMS reactivity at the disulfide Cys-57 and/or Cys-146 because the same shift is seen with C6S/C111S SOD1 expressed in ccs1⌬ cells (lane 6), but not with C146S and C57S SOD1 affecting the disulfide (Fig.  1A, lanes 9 and 13). AMS reactivity at Cys-57 and/or Cys-146 indicates that the disulfide cysteines have been reduced in cells lacking CCS.
We also monitored the disulfide of human SOD1 expressed in mammalian cells. These studies employed skin fibroblasts derived from CCS ϩ/ϩ and CCS Ϫ/Ϫ null homozygous mice that are transgenic for human WT SOD1 or the ALS SOD1 mutant G37R (6,45). When expressed in CCS ϩ/ϩ fibroblasts, human WT SOD1 exhibits the identical AMS reactivity pattern seen with disulfide-oxidized SOD1 from CCS1ϩ yeast (Fig. 1C, compare  lanes 1 and 3). In CCS Ϫ/Ϫ fibroblasts, two products are observed, representative of both oxidized and reduced states of the disulfide cysteines (Fig. 1C, lane 6). With the G37R ALS mutant SOD1, there is evidence for disulfide reduction even in CCS ϩ/ϩ fibroblasts (Fig. 1C,  lane 8), consistent with the notion that ALS mutants are more susceptible to disulfide reduction (46). As has been shown for the endogenous SOD1 of yeast (23), CCS helps oxidize the disulfide of human SOD1 expressed in yeast and mammalian cells.
We tested whether the CCS-independent pathway affects the human SOD1 disulfide (6). This auxiliary pathway for activating SOD1 is blocked by introducing prolines in SOD1 at positions corresponding to amino acids 142 and 144 (6,7). Human SOD1 contains Ser and Leu at these positions, and an S142P/ L144P mutant of human SOD1 is only activated by CCS (6). When S142P/L144P human SOD1 is expressed in CCS1ϩ yeast cells, the enzyme is active (6), and the disulfide is oxidized (Fig.  2A, lane 3). However, when expressed in ccs1⌬ yeast cells, S142P/L144P human SOD1 cannot obtain copper by either pathway; the enzyme is inactive (6), and the disulfide cysteines are completely reduced (Fig. 2A, lane 4). Hence, both CCS-dependent and -independent pathways for activating SOD1 contribute to disulfide oxidation with human SOD1.
The Disulfide and Stability of ALS Mutant SOD1-We explored how changes in copper loading in vivo affect ALS mutants of SOD1. Human SOD1 mutants G37R, G93A, and A4V were expressed in yeast under conditions where copper activation by CCS and/or the CCS-independent pathways were blocked. CCS was inhibited by expression in a ccs1⌬ null yeast strain, and CCS-independent activation was blocked by introducing prolines 142 and 144 into SOD1. As seen in Fig. 2B, lane  10 and 11) as has been reported for other SOD1 mutants (31,39,41,44). B and C, lane 1, human SOD1 expressed in the CCS1ϩ strain KS107. C, lanes 2-9, analysis of immortalized fibroblasts from CCS ϩ/ϩ and CCS Ϫ/Ϫ mice that were transgenic for either WT or G37R human SOD1 as described (6,45). Y indicates CCS1ϩ yeast expressing WT human SOD1; M indicates fibroblasts from CCS ϩ/ϩ mice expressing WT human SOD1. It is noteworthy that with long exposures, disulfide-oxidized SOD1 often runs as a doublet, with a minor band Ϸ1.5 kDa smaller than the major disulfide-oxidized product (e.g. A, lane 2). Because this doublet is observed with WT, C146S, C57S, and C6S, but not with C111S SOD1 (also see supplemental Fig. S1B), AMS modification at Cys-111 might effect both Ϸ4.5 kDa (major) and Ϸ3.0 kDa (minor) shifts in apparent molecular mass. However, we cannot exclude other SOD1 modifications, such as oxidative products and retention of some metal binding during electrophoresis as has been reported previously to cause multiple SOD1 isoforms during electrophoresis (23,56). Although AMS reactivity generally goes to completion, there occasionally is some low level of incomplete modification as is seen in C, lanes 8 and 9. 4, loss of the CCS-independent pathway alone through an S142P/L144P substitution resulted in some lowering of the steady state level of the ALS mutants, particularly A4V (also see supplemental Fig. S2). Moreover, when the S142P/L144P variants of G37R, G93A, and A4V were expressed in ccs1⌬ cells, no polypeptide could be recovered (Fig. 2B, lane 3). By comparison, the S142P/L144P variant of WT SOD1 stably accumulated in ccs1⌬ cells (Fig. 2B, lane 3, top panel, also see supplemental Fig. S2). The ALS mutants, but not WT SOD1, appeared highly unstable and degraded when both the CCS-dependent and -independent pathways for copper loading and disulfide oxidation were eliminated.
Loss of CCS alone can also affect the stability of certain ALS mutants expressed in yeast. For example, G41D accumulates to very low steady state levels in ccs1⌬ yeast (Fig. 2C). A4V SOD1 also appears somewhat unstable in ccs1⌬ cells, but to a lesser degree ( Fig. 2C and also see Figs. 4C and 5B). It is noteworthy that this same pattern of instability has been observed in mammalian expressions systems; G41D and A4V are more prone to degradation and aggregation than other ALS mutants (39,47). It is important to note that unlike mammalian expression systems where SOD1 misfolding can lead to both degradation and aggregation of the polypeptide, protein degradation is the primary end point of SOD1 misfolding in yeast. There is no evidence of SOD1 aggregation in yeast expression systems (see "Discussion").
To monitor turnover of disulfide-reduced versus disulfideoxidized polypeptides, SOD1 synthesis in yeast cells was controlled by the methionine-repressible MET25 promoter. In this manner, A4V SOD1 is actively synthesized in yeast cells not treated with methionine but is repressed upon addition of methionine to the growth medium, allowing us to monitor loss of the SOD1 polypeptides over time. In the absence of methio-nine, a good fraction of A4V SOD1 expressed in ccs1⌬ cells is seen in the disulfide-reduced state, and the ratio of reduced to oxidized disulfide remains relatively constant over 3 h (Fig. 3A, lanes 1 and 2; also see supplemental Fig. S2). But when A4V expression was repressed by methionine supplements, the disulfide-reduced form of A4V SOD1 was lost, whereas the disulfide-oxidized form was more stable (Fig. 3A, lanes 4 -6; also see supplemental Fig. S2). This result is consistent with in vitro studies with recombinant SOD1 showing that disulfide-reduced SOD1 is more prone to degradation by the proteosome than disulfide-oxidized SOD1 (35).
We similarly examined stability of WT human SOD1. In the experiment of Fig. 3B, the disulfide of WT SOD1 at steady state is more oxidized than that of A4V SOD1 examined in parallel (compare t ϭ 0 samples for A4V and WT SOD1). Following 3 h of methionine repression, both the reduced and oxidized pools of WT SOD1 were retained, compared with A4V SOD1 that exhibited instability particularly with the disulfide-reduced fraction (Fig. 3B, also see supplemental Fig. S2). The disulfidereduced form of WT SOD1 is not subject to the same dramatic turnover as the ALS mutant.
The Role of Cytosolic Glutaredoxins in Reducing the SOD1 Disulfide and Destabilizing ALS Mutant Polypeptides-In the absence of copper activation, what cellular factors favor reduction of the SOD1 disulfide cysteines? We tested the possible role of GRX. S. cerevisiae expresses two GRXs in the cytosol, namely Grx1p and Grx2p. Double grx1 grx2 null mutations do not alter cellular redox or GSH/GSSG ratios (38). We tested how loss of GRX affects the disulfide of SOD1 expressed in FIGURE 2. The effects of copper activation on the disulfide and steady state levels of human SOD1 variants expressed in yeast. Yeast strains expressing WT human SOD1 or the indicated mutant variants were analyzed as follows: A, for AMS modification of SOD1 cysteines as in Fig. 1; B and C, for total human SOD1 protein by immunoblot. A, strains as described in Fig. 1A expressed either WT or S142P/L144P human SOD1 under control of the S. cerevisiae PGK1 promoter. Heavy and light arrows indicate disulfide-reduced and disulfide-oxidized SOD1. B, the ccs1⌬ sod1⌬ strain LS101 was transformed where indicated (CCSϩ) with the human CCS expressing plasmid pPS015 and also with the indicated variants of human SOD1 under control of the S. cerevisiae SOD1 promoter. In lanes 3 and 4, the S142P/L144P substitution was introduced in WT, G37R, G93A, and A4V human SOD1. Quantification of the levels of SOD1 can be found in supplemental Fig. S2. C, human WT, G41D, and A4V SOD1 under control of the PGK1 promoter were expressed in strains described in Fig. 1A. ccs1⌬ null yeast strains, where the SOD1 is normally a mixed population of disulfide-reduced and -oxidized forms (Fig. 4A,  lanes 2 and 5). Loss of GRX through grx1⌬ grx2⌬ null mutations shifted the disulfide of SOD1 toward the oxidized state, and the effects were particularly pronounced with A4V SOD1 (Fig. 4A,  lane 6).
Loss of yeast GRXs not only affected the SOD1 disulfide but also the stability of ALS mutants. In Fig. 4B, A4V SOD1 stability was monitored through methionine repression. The disulfideoxidized form of SOD1 that accumulates in grx1⌬ grx2⌬ ccs1⌬ strains is quite stable over 3 h of methionine repression (Fig. 4B,  lanes 3 and 4). As such, the steady state levels of total A4V SOD1 increases, as do levels of G41D SOD1 expressed in grx1⌬ grx2⌬ ccs1⌬ yeast (Fig. 4C, lanes 2 and 3).
Of the two cytosolic GRXs in yeast, Grx2p is the predominant form (38). Single grx2⌬ mutations were sufficient to increase steady state levels of G41D SOD1 (Fig. 4C, lane 5), whereas grx1⌬ mutations were not (not shown). To examine the effects of Grx2p further, grx1⌬ grx2⌬ ccs1⌬ cells were transformed with a low copy plasmid expressing GRX2 under its native promoter. Plasmid-borne Grx2p was indeed enzymatically active, as monitored by the standard in vitro assay for GRX activity using HED as substrate (33) (Fig. 4D, top). Expression of Grx2p in grx1⌬ grx2⌬ ccs1⌬ yeast also correlated with reduction of the disulfide in A4V SOD1 (Fig. 4D,  bottom).
Loss of GRX clearly affects oxidation of the intramolecular disulfide in SOD1, but what about non-native disulfides? Recently, ALS SOD1 mutants have been shown to oligomerize and form intermolecular disulfide cross-links that can be visualized by electrophoresis under nonreducing conditions (19 -22, 48). However, when analyzed under nonreducing conditions ("ϪDTT"), A4V SOD1 expressed in grx1⌬ grx2⌬ ccs1⌬ yeast only exists as a monomer, and there were no unique high molecular species consistent with intermolecular disulfides (Fig. 5A, right). Therefore, GRXs appear to only target only the chief intramolecular disulfide of SOD1.
We also examined the effects of GRX loss on SOD1 activity. As seen in Fig. 5B, there was no significant change in SOD1 activity with WT, G93A, or A4V variants expressed in grx1⌬ grx2⌬ ccs1⌬ cells compared with ccs1⌬ single mutants. A4V SOD1 shows poor CCS-independent activity in ccs1⌬ cells, and this does not change with additional grx1⌬ grx2⌬ mutations (Fig. 5B). Despite oxidation of the A4V SOD1 disulfide in this strain (shown in Fig. 4, A, B, and D), the SOD1 remains largely inactive. Presumably, the disulfide-oxidized SOD1 is still copper-deficient in grx1⌬ grx2⌬ ccs1⌬ cells (see "Discussion").
To more directly test whether GRX can reduce the SOD1 disulfide, we designed an in vitro assay using purified recombinant yeast Grx2p and purified human WT or A4V SOD1. To regenerate reduced GRX, the in vitro reactions also contained GSH. In the experiment of Fig. 6A, recombinant Grx2p at 50 nM was allowed to react with disulfide-oxidized A4V SOD1 that was apo for metals and present at a concentration of 2.0 M. Within 1 h, the disulfide was reduced (Fig. 6A, lane 3, and also see Fig. 6C, lane 6). Similar results were obtained with apoG93A SOD1 (not shown). GSH alone was not sufficient to reduce the disulfide (Fig. 6A, lane 6), but GSH was required for Grx2p-dependent reduction of the disulfide (supplemental Fig. S3B), as strain MC120 (identical to MC119 except ccs1⌬::ADE2 rather than ccs1⌬::URA3) also expressing A4V SOD1 under PGK1 was transformed where indicated (GRX2), with the pVC0147 plasmid for expressing S. cerevisiae Grx2p or with empty vector pRS313 (V). Top, lysates were assayed for GRX activity using the standard HED assay. Results represent the averages of two independent assays; error bars represent range. Activity is defined in terms of nanomoles of NAPDH consumed per min per mg of lysate protein. Bottom, status of the A4V SOD1 disulfide was monitored as in Fig. 1. Heavy and light arrows disulfide-reduced and disulfide-oxidized SOD1, respectively. would be expected for GRX reactions (24). Compared with A4V SOD1, the disulfide of apoWT human SOD1 exhibited poor reactivity toward Grx2p and GSH (Fig. 6B) even after 2 h of incubation (Fig. 6C, lane 3). At best, Ϸ2.0% conversion to the reduced form was seen in one experimental trial out of eight with apoWT SOD1 (see supplemental Fig. S3B).
We also tested Grx2p reactivity toward metallated SOD1. As seen in Fig. 6D, the disulfide of metallated A4V SOD1 containing both copper and zinc was refractory to reduction by Grx2p in vitro. The apo version is the preferred substrate for disulfide reduction by Grx2p.

DISCUSSION
Heren we describe how post-translational modification factors for SOD1 can impact on the intramolecular disulfide and stability of ALS mutant SOD1. Two classes of intracellular factors are shown to work in opposite ways to control status of the disulfide. First, the CCS-dependent and -independent pathways for copper activation promote oxidation of the human SOD1 disulfide and enhance stability of ALS mutants A4V, G93A, and G37R expressed in yeast. Without copper activation, intracellular reductants such as GRX promote disulfide reduction, contributing to SOD1 instability.
The ALS mutants seemed particularly vulnerable to loss of copper activation. Although WT SOD1 stably accumulated in yeast cells without the copper co-factor or an oxidized disulfide, the three ALS mutants we examined (G93A, G37R, and A4V) were degraded when both CCS-dependent and -independent pathways were blocked. In the yeast expression system, misfolded SOD1 mutants lacking metals and the disulfide are effectively cleared by protein degradation. There is no evidence of SOD1 aggregation in yeast by either formation of high molecular weight species on SDS gels (as in Fig. 5A and supplemental Fig. S4) or by formation of detergent-insoluble precipitates (data not shown). In mammalian cells, the clearance of misfolded SOD1 may be incomplete, allowing for accumulation of misfolded aggregates. Regardless of whether the end point is degradation or aggregation, the initiating misfolding event in SOD1 can be promoted by the absence of copper and the intramolecular disulfide. Copper loading of SOD1 is incomplete in various cells and tissues (5, 49 -51), and in the case of certain ALS mutants, this pool of immature SOD1 may very well seed formation of misfolded aggregates.
ALS mutants may also be more vulnerable to disulfide reduction by GRX. Yeast Grx2p was seen to promote reduction of the disulfide cysteines of A4V SOD1 both in vivo and in vitro although WT human SOD1 was less reactive. It is possible that misfolding of certain ALS mutants allows for greater access of the GRX molecule toward the disulfide. This increased reactivity with GRX, together with the high instability of the disulfidereduced state, makes the SOD1 mutant a prime target for protein misfolding and degradation.
To date, very few in vivo substrates have been documented for eukaryotic dithiol GRXs. Mammalian GRX can act as a dethionylase for actin, Hsp70, and Ras (26,52,53), but no substrates have been identified for S. cerevisiae Grx1p and Grx2p. Furthermore, there have been no reports of an intramolecular disulfide target for eukaryotic GRXs, only S-thionylated targets.  . Loss of glutaredoxins in yeast does not affect SOD1 oligomerization or alter SOD1 activity. A, lysates from the indicated yeast strains expressing A4V SOD1 under PGK1 where indicated (ϩ) were subjected to denaturing gel electrophoresis and immunoblot analysis for steady state levels of SOD1. Prior to electrophoresis, samples containing 30 g of total extract protein were heated at 95°C in SDS-buffer that either contained (ϩDTT) or lacked (ϪDTT) 10 mM DTT as a reducing agent. Gels were soaked in TCEP to help reduce polypeptide cysteines according to published methods (21) prior to immunoblotting. Vertical numbers indicate size of molecular weight markers run in parallel. B, lysates from the indicated yeast strains expressing WT, A4V, or G93A SOD1 under PGK1 were subjected to either nondenaturing gel electrophoresis and nitro blue tetrazolium staining for SOD1 activity (40, 41) (top), or to SDS-PAGE and immunoblot for human SOD1 protein (bottom). Strains utilized are as follows: GRX1/GRX2ϩ CCS1ϩ, CY4; GRX1/GRX2ϩ CCS1⌬, the ccs1⌬ strain MC108; GRX1/GRX2⌬ CCS1⌬, the grx1⌬ grx2⌬ ccs1⌬ strain MC120. This particular strain background (CY4) shows lower levels of CCS-independent SOD1 activity than other strains (e.g. BY4741 or EG103), perhaps due to lower abundance of intracellular GSH needed for efficient CCS-independent activation (6). We favor a model in which GRX acts on the intramolecular disulfide of SOD1 rather than an S-thionylated intermediate. If SOD1 were S-thionylated, such an intermediate would be detected by AMS modification and would hyper-accumulate in grx1⌬ grx2⌬ yeast mutants lacking dethionylase activity. To our knowledge, SOD1 represents the first reported intramolecular disulfide substrate for a eukaryotic GRX. It is quite possible that other polypeptide disulfides serve as targets, including those noted in the prion (54) and transthyretin redox-sensitive proteins (55) implicated in disease.
Our studies strongly indicate that GRX preferentially acts on a SOD1 molecule that contains an oxidized disulfide, yet lacks copper. First, the in vivo effects of GRX on the SOD1 disulfide were only observed in yeast strains where copper activation was low (e.g. in ccs1⌬ strains). Loss of GRX correlated with disulfide oxidation, but the SOD1 enzyme remained largely inactive, indicative of no copper co-factor. Moreover, recombinant Grx2p could reduce the disulfide of apo but not metallated A4V SOD1. If copper-deficient, disulfide-oxidized SOD1 is indeed the substrate for GRX, this would imply that SOD1 can obtain an oxidized disulfide in vivo without copper insertion. How the SOD1 disulfide is oxidized without copper is still unclear but is the subject of current investigations. In any case, the GRXand/or GSH-mediated reduction of the disulfide in copper-deficient SOD1 would be beneficial to the cell, as it would provide additional substrate for CCS that is normally inert toward disulfide-oxidized SOD1 (23).
Although these studies on the human SOD1 disulfide were largely conducted in yeast, they are predicted to have important implications for SOD1 folding and stability in mammalian cells as well. First, the factors that control the SOD1 disulfide are well conserved in yeast and mammals, including CCS (2), the CCS-independent pathway (6), and the dithiol GRX molecules of the cytoplasm (38). Moreover, the relative instability observed with ALS mutants A4V and G41D expressed in yeast is remarkably similar to what has been reported in mammalian cells (39,47). Therefore, the effects of copper loading pathways and thiol reductants on the disulfide and the stability of SOD1 are expected to be conserved. In a previous transgenic mouse study, loss of CCS was reported to not affect motor neuron disease associated with expression of ALS mutants G93A, G37R, or G85R (45). Based on our studies in yeast, these mutants are stable without CCS due to compensatory effects of the CCS-independent pathway. It is therefore important to consider the impact of both copper loading pathways, as well as thiol reductants such as GRX on the fate of ALS mutants in motor neuron disease.