Erv1 is a flavin‐dependent sulfhydryl oxidase in the mitochondrial intermembrane space (IMS) that functions in the import of cysteine‐rich proteins. Redox titrations of recombinant Erv1 showed that it contains three distinct couples with midpoint potentials of −320, −215, and −150 mV. Like all redox‐active enzymes, Erv1 requires one or more electron acceptors. We have generated strains with erv1 conditional alleles and employed biochemical and genetic strategies to facilitate identifying redox pathways involving Erv1. Here, we report that Erv1 forms a 1:1 complex with cytochrome c and a reduced Erv1 can transfer electrons directly to the ferric form of the cytochrome. Erv1 also utilized molecular oxygen as an electron acceptor to generate hydrogen peroxide, which is subsequently reduced to water by cytochrome c peroxidase (Ccp1). Oxidized Ccp1 was in turn reduced by the Erv1‐reduced cytochrome c. By coupling these pathways, cytochrome c and Ccp1 function efficiently as Erv1‐dependent electron acceptors. Thus, we propose that Erv1 utilizes diverse pathways for electron shuttling in the IMS.
The mitochondrial intermembrane space (IMS) is conducive to disulfide bond formation (Koehler et al, 2006; Wiedemann et al, 2006; Herrmann and Kohl, 2007), suggesting that the redox environment is oxidizing, similar to that of the bacterial periplasm (Kadokura et al, 2003). Thiol‐trapping assays indicated that the small Tim proteins with the twin CX3C motif (Tim8, Tim9, Tim10, and Tim13) form disulfide bonds (Curran et al, 2002a, 2002b; Koehler, 2004). Structural and mutational studies confirmed that the small Tim proteins contain juxtapositional disulfide linkages (Allen et al, 2003; Lu et al, 2004; Webb et al, 2006).
Subsequently, a new import pathway was identified for the small Tim proteins and other substrates with a twin CX9C motif (Chacinska et al, 2004; Mesecke et al, 2005; Rissler et al, 2005; Terziyska et al, 2005; Gabriel et al, 2007). Specifically, Mia40 functions as a tethering component in the IMS by forming a transient disulfide bond with the incoming substrate, trapping the precursor in the IMS. The substrates are released from Mia40 by a mechanism that may involve rearrangement of disulfide bonds. It has been proposed that the sulfhydryl oxidase Erv1 participates in recycling Mia40 by reoxidizing Mia40's cysteine residues (Mesecke et al, 2005). Thus, electrons may be transferred from the imported proteins to Erv1 via Mia40. Erv1 also functions in the maturation of cytosolic FeS centers (Lange et al, 2001), suggesting its diverse functions in the IMS. Both Mia40 and Erv1 are required for viability (Lisowsky, 1992; Chacinska et al, 2004).
Erv1 was first identified by Lisowsky and co‐workers (Lisowsky, 1992) and is a member of the flavin‐dependent sulfhydryl oxidase family (Coppock and Thorpe, 2006). Erv1 shares homology with Erv2, a protein that functions in the endoplasmic reticulum (Sevier et al, 2001), with Quiescin‐sulfhydryl oxidases that function in the extracellular matrix (Hoober et al, 1996), and with the E10R protein of the pox virus that inserts disulfide linkages into the virion's coat proteins (Senkevich et al, 2002). These sulfhydryl oxidases are predicted to use oxygen (O2) as an electron acceptor for the oxidation of cysteine residues to protein disulfides, generating hydrogen peroxide (H2O2) in the process (Coppock and Thorpe, 2006).
Erv1 has three potential redox centers, two pairs of redox‐active cysteines (C30–C33 and C130–C133) and a noncovalently bound FAD (Hofhaus et al, 2003; Farrell and Thorpe, 2005). The C30–C33 pair is important for the formation of an intermolecular disulfide bond between two Erv1 monomers and is required for in vivo function (Lee et al, 2000; Hofhaus et al, 2003). The C30–C33 pair, however, is not required for sulfhydryl oxidase activity in vitro, suggesting that the C30–C33 pair is not required for Erv1 enzymatic activity, but instead serves a structural role. In contrast, the C130–C133 pair, which is positioned near the likely flavin‐binding site, is important for the redox activity of Erv1 (Hofhaus et al, 2003; Farrell and Thorpe, 2005), as mutagenesis of either cysteine to a serine or alanine abolishes sulfhydryl oxidase activity. Farrell and Thorpe (2005) have estimated an Em value of −178 mV for the midpoint potential of the flavin in ALR, but were not able to reduce the disulfide near ALR's flavin‐binding site (i.e., the disulfide equivalent to C130–C133 in Erv1), raising the possibility that this putative redox‐active cysteine pair in ALR may have a considerably more negative midpoint potential.
Bacteria utilize a wide array of terminal electron acceptors for the oxidative formation of protein disulfide bonds (Bader et al, 1999). Erv1 may also display a level of flexibility with regard to electron acceptors, because yeast cells lacking the mitochondrial genome and respiratory pathway still import mitochondrial proteins. In this study, we have used genetic and biochemical approaches to characterize Erv1 and investigate potential pathways in which Erv1 might be oxidized by different terminal electron acceptors. We show that Erv1 forms a complex with cytochrome c (cyt c) in organello and in vitro. With in vitro assays, we demonstrate that Erv1 shuttles electrons to both O2 and then cytochrome c peroxidase (Ccp1) and oxidized cyt c. These studies suggest that, like the prokaryotic disulfide catalytic system, Erv1 utilizes several terminal electron acceptor pathways.
To begin characterization of the electron acceptor pathways utilized by the sulfhydryl oxidase Erv1, we generated temperature‐sensitive (ts) mutants using error‐prone PCR (Koehler et al, 1998). Two strains harboring alleles erv1‐101 and erv1‐12 on centromeric plasmids as well as the tim9‐3 control (Leuenberger et al, 2003) grew at 25°C but arrested growth at 37°C on rich glucose (YPD) and ethanol‐glycerol (YPEG) media (Figure 1A). We subsequently analyzed the steady‐state levels of mitochondrial proteins using α‐ketoglutarate dehydrogenase (Kdh) as a loading control (Figure 1B). The abundance of outer membrane markers porin and Tom40 was not affected (data not shown); however, the steady‐state levels of the substrates of the MIA40 pathway (Erv1, Tim9, Tim10, Tim12, and Tim13) and Mia40 were markedly reduced in mitochondria that were shifted to 37°C. In addition, Tim22, Tim23, and Tim54 levels were reduced. The reduction in the TIM22 import components and Tim23 likely results from a decreased import of the small Tim proteins via the MIA import pathway. Interestingly, Ccp1 migrated aberrantly in mitochondria shifted to 37°C, and cyt c abundance was not obviously affected at 37°C. Taken together, Erv1 dysfunction affects proteins primarily imported by the MIA40 import pathway and the TIM22 import pathway.
Cyt c partners with Erv1 in organello
To identify Erv1 partner proteins, we placed 10X‐His tags on the C‐termini of both Erv1 and Ccp1 and substituted the His‐tagged versions on multicopy plasmids for the endogenous genes (Figure 2A and B). The Erv1‐His and Ccp1‐His strains grew like wild type (WT, data not shown), indicating that the His‐tagged proteins were functional. Initial attempts to express Erv1‐His from a centromeric plasmid yielded a marked decrease in the Erv1‐His protein levels. Therefore, Erv1‐His expression from the multicopy plasmid resulted in increased expression of Erv1‐His relative to WT mitochondria, whereas cyt c and Ccp1 abundance were comparable with that of WT mitochondria (Supplementary Figure 1). Mitochondria were solubilized in 1.0% digitonin (T) and incubated with Ni2+ agarose. After washing to remove non‐bound proteins (S), the bound proteins (B) were eluted in sample buffer and separated by SDS–PAGE. Immunoblotting analysis showed that cyt c and a fraction of the Mia40 co‐purified with Erv1‐His, whereas Ccp1 did not co‐purify. Note that in independent purifications, the fraction of cyt c that co‐purified with Erv1 varied from approximately 70 to 100% (data not shown), most likely caused by the increased abundance of Erv1‐His. With Ccp1‐His, a fraction of the cyt c but not Erv1 or Mia40 co‐eluted. In a control reaction, WT mitochondria were treated identically and the tested proteins showed no affinity for the Ni2+ agarose (Figure 2C). Therefore, cyt c is a partner protein with both Erv1 and Ccp1, but Ccp1 seemingly does not form a complex with Erv1.
Terminal electron acceptor pathways are required in both aerobic and anaerobic growth conditions, and different pathways are utilized in prokaryotes depending on the oxidative state (Bader et al, 1999). First, we tested whether the erv1 mutants grew under anaerobic conditions and in the presence of ethidium bromide, which induces loss of mitochondrial DNA and respiratory ability (Figure 2D). The erv1 mutants grew, albeit more slowly, when the mitochondrial genome was absent (+EtBr) under aerobic conditions (+O2). The erv1‐101 also showed a slower growth rate under anaerobic conditions (−O2), whether mitochondrial DNA was present or absent (Figure 2D). In contrast, the negative control Δtim54, which is petite negative, failed to grow under the conditions tested. Because yeast express two different cyt c genes (CYC1 and CYC7) under different O2 concentrations (Burke et al, 1997), we used a Δcyc3 strain to generate a cyt c null strain, as Cyc3 is the heme lyase for assembly of holocytochrome c (Dumont et al, 1988). The Δcyc3 and Δccp1 strains grew under anaerobic and aerobic conditions in the presence and absence of the mitochondrial genome (Figure 2D).
We tested CYC3 and ERV1 for synthetic lethality at 25°C (Figure 2E). Tetrad analysis showed that the strain deleted for cyc3 in the erv1‐101 mutant background was viable in aerobic conditions at 25°C. However, the Δcyc3 erv1‐101 mutant failed to grow in anaerobic conditions. Thus, CYC3 and ERV1 are synthetic lethal in anaerobic conditions, indicating that cyt c and Erv1 interact genetically as well as physically.
Erv1 has three different midpoint potentials corresponding to the two redox‐active cysteine pairs (C30–C33 and C130–C133) and the FAD moiety
Previous studies have suggested that C30–C33 plays a structural role in the formation of Erv1 dimers, whereas the C130–C133 pair has sulfhydryl oxidase activity and is situated near the FAD (Lee et al, 2000; Hofhaus et al, 2003). We measured the disulfide/dithiol redox properties of Erv1 at pH 7.0 using GSH/oxidized glutathione (GSSG) redox buffers to poise samples at defined redox potentials (Eh) values ranging from −100 to −240 mV and reduced dithiothreitol (DTT)/oxidized (DTTox) redox buffers to poise samples at Eh values ranging from −220 to −380 mV (Figure 3; Krimm et al, 1998; Hirasawa et al, 1999). Recombinant Erv1 was incubated in the indicated redox potential buffer and then treated with mBBr to form fluorescent covalent adducts of Erv1 thiols (mBBr does not react with disulfides). The titration results were independent of the redox equilibration time, over the range from 1.5 to 3.5 h, and were also independent of the concentration of the redox buffer, over the range from 1 to 5 mM, indicating the likelihood of good redox equilibration between Erv1 and the ambient potential imposed by the redox buffers. Additional support for the establishment of good redox equilibration during these titrations comes from the fact that all of the titrations gave excellent fits to the Nernst equation for a two‐electron redox couple. Figure 3A shows the results of a typical titration and reveals the presence of two two‐electron components. The average values (based on three titrations) for the redox midpoint potentials (Em) of the two components are −330 and −150 mV, respectively (the use of the average deviation as an approximate measure of the experimental uncertainty suggests that the experimental uncertainty does not exceed±10 mV). The Em value and the n=2 character of the titration curve, as well as the known specificity of mBBr for chemical modification of thiols (Krimm et al, 1998; Hirasawa et al, 1999), indicate that the redox titration results are most readily interpreted in terms of the presence of two separate disulfide/dithiol redox couples in Erv1.
Added support for the notion of two separate thiol‐based redox couples comes from DTNB assays of the thiol content of Erv1 samples poised at defined redox potentials (See Table I). In particular, the difference in thiol content of Erv1 samples poised at a potential where the Em=−150 mV transition would be either completely reduced or completely oxidized is 2.0±0.3 mol of thiol per mol of Erv1, exactly what would be predicted for the reduction of one intramolecular disulfide per Erv1 monomer. Similarly, the difference in thiol contents of Erv1 samples poised at potential where the Em=−320 mV transition would be either completely reduced or completely oxidized is 0.9±0.3 mol of thiol per mol of Erv1, exactly what would be predicted for the reduction of one intermolecular disulfide between two Erv1 monomers.
An attempt was made to confirm the assignment, based on DTNB thiol assays of samples poised at defined Eh values, of the Em=−330 mV to an intermolecular disulfide formed between two Erv1 monomers, using non‐reducing SDS–PAGE gels of Erv1 samples that had been poised at defined Eh values centered around −330 mV before SDS–PAGE analysis. The amount of dimer increases as the Eh is increased over the range from −380 to −220 mV (Supplementary Figure 2), consistent with an Em value of approximately −330 mV for an intermolecular disulfide. Figure 3B shows the results of a redox titration, achieved under conditions identical to those of Figure 3A (except that mBBr was not added), but with tryptophan fluorescence used to monitor redox‐dependent changes in Erv1. The data are most readily interpreted in terms of a two‐electron redox couple, with Em=−315±10 mV at pH 7.0, in which the microenvironment of at least one Erv1 tryptophan residue differs in the oxidized and reduced forms of the protein. The fact that the Em value of the two‐electron couple detected in tryptophan fluorescence is indistinguishable, within the combined experimental uncertainties of the measurements, from the Em value of the more negative couple seen in the mBBr titrations, leads us to conclude that it is the monomer/dimer transition that alters the microenvironment of at least one Erv1 tryptophan.
Flavin analysis of Erv1 indicated the presence of 1.0±0.1 mole of FAD per mole of recombinant Erv1 (no evidence was found for any FMN in the protein). Figure 3C shows the results of a typical titration of the FAD group of Erv1 at pH 7.0, using the absorbance in the visible region to monitor the oxidation state of FAD. The data showed an excellent fit to the Nernst equation for a single n=2 component with an Em value of −215 mV. Representative spectra generated during the potentiometric titration of the flavin group are included (Supplementary Figure 3). The absence of well‐defined isosbestic points in these spectra arises from a slow drift in the instrument baseline during the approximately 2 h required for the full titration. Under the conditions employed, no red anionic or neutral blue flavin semiquinone could be detected, in contrast to the results reported for ALR (Farrell and Thorpe, 2005). It should also be pointed out that while the flavin in ALR has been suggested to have a more positive redox potential than the disulfide/dithiol centers present in ALR (Farrell and Thorpe, 2005), this is not the case for Erv1 where the FAD has an Em value that is significantly more negative than that of the Em=−150 mV intramolecular disulfide.
Erv1 and cyt c interact in a 1:1 complex and Erv1 can directly reduce cyt c
Farrell and Thorpe (2005) have previously suggested that cyt c is the physiological oxidant of Erv1. In Figure 2A, we demonstrated that cyt c partners with Erv1 in vivo. To further probe the stoichiometry of the Erv1:cyt c complex, we examined the interactions of our characterized recombinant Erv1 and yeast cyt c, using a spectral perturbation technique to monitor protein–protein interactions. This technique has been used to characterize a large number of protein complexes, including the one formed between cyt c and Ccp1 (Erman and Vitello, 2002). The solid line in Figure 4A shows the difference spectrum arising from the interaction of Erv1 with cyt c in low ionic strength buffer (i.e., it represents the spectrum of a 1:1 mixture of the two proteins from which has been subtracted the sum of the spectra of the two separate proteins). This difference spectrum was not observed if either BSA or Ccp1 replaced cyt c, suggesting that the difference spectrum of Figure 4A does not arise from some nonspecific artifact. Figure 4A also shows (see the dashed line) that no spectral perturbations are observed at high ionic strength (solid NaCl was added to the sample used for the low ionic strength spectrum to yield an approximate concentration of 250 mM and the spectrum was re‐recorded). The simplest explanation for this observation is that electrostatic forces, which weaken at high ionic strength, make an important contribution to stabilizing the Erv1:cyt c complex. Figure 4B shows the results of an experiment, carried out at low ionic strength, in which the magnitude of spectral changes arising from complex formation were plotted against varying concentrations of cyt c at a fixed concentration of Erv1. The data give a good fit to the theoretical curve for the presence of a 1:1 Erv1:cyt c complex of very high affinity (computer simulations suggest that the Kd for complex formation under these conditions is probably ⩽100 nM). As protein–protein complex formation can alter the Em value of protein prosthetic groups (Ondrias et al, 1985), the FAD group of Erv1 and the heme of cyt c were titrated in a 1:1 mixture of the proteins at the low ionic strength that favors complex formation between the proteins. The Em value of the Erv1 FAD was unaffected by the presence of cyt c and the Em value of cyt c was found to be +250 mV, regardless of whether Erv1 was present or absent (data not shown). The value obtained in the absence of Erv1 is in good agreement with the published literature (Wallace, 1984).
That Erv1 and cyt c form a complex suggests that Erv1 might shuttle electrons directly to cyt c. This possibility was tested directly, under anaerobic conditions, by observing changes in absorbance at 550 nm that arise from reduction of the cyt c heme when DTT‐reduced Erv1 is added to ferric cyt c (Figure 4C). Erv1, at the concentration used in these experiments, does not contribute significantly to the absorbance at this wavelength. As shown in Figure 4C, the addition of reduced Erv1 indeed cause rapid reduction of cyt c, denoted by the marked increase in absorbance after Erv1 addition. In control experiments, in which DTT alone was added to cyt c in the absence of Erv1, a much slower rate of heme reduction was observed, indicating that DTT is a kinetically poor reductant for cyt c and confirming that Erv1 can reduce the heme of cyt c directly in vitro.
Ccp1 and cyt c compete with O2 in Erv1‐mediated oxidation of DTT
Ccp1 serves as a general catalyst to remove H2O2 formed in mitochondria during aerobic metabolism (Boveris, 1976). However, despite extensive in vitro characterization of Ccp1 (Erman and Vitello, 2002), few studies have clarified its specific physiological role in yeast. We tested CCP1 and ERV1 for synthetic lethality at 25°C (Figure 5A). Tetrad analysis showed that strains deleted for ccp1 in the erv1 mutant background were viable at 25°C when the mitochondrial genome was absent. In addition, the erv1‐101 mutant grew, albeit slowly, in anaerobic conditions, similar to that in Figure 2D, regardless if CCP1 was present. Thus, CCP1 and ERV1 are not synthetic lethal, indicating that CCP1 is not strictly required for growth under the conditions investigated.
Reactions catalyzed by flavin‐dependent sulfhydryl oxidases generate disulfide bonds with the reduction of O2 to H2O2. We, therefore, measured H2O2 levels in the erv1 mutants (Figure 5B) in vivo to determine if H2O2 levels were altered. The parental strain and erv1‐101 and erv1‐12 mutants were grown to mid‐log phase and maintained at 25°C or shifted to 37°C for 3 h. H2O2 levels were measured using the fluorescent indicator (DCFH‐DA) and fluorescence was monitored at 530 nm. H2O2 levels were significantly elevated in the erv1 mutants at 37°C, supporting that Ccp1 function is impaired.
Using recombinant Ccp1 and Erv1 and oxidized cyt c, we measured O2 consumption of Erv1 in the presence of excess DTT, a non‐physiological electron donor, using an O2 electrode. In the absence of Erv1, cyt c and Ccp1 did not consume O2 whether alone or in combination (data not shown). When Erv1 was added alone, or in the presence of either oxidized cyt c or Ccp1, the rate of O2 consumption was similar (Figure 6A; Table II). Analysis of complete oxygen electrode traces from air saturation to zero oxygen concentration allowed rates to be determined by drawing tangents to the curve for a range of oxygen concentrations. The addition of cyt c, however, caused a slight delay in O2 consumption of 16±0.2 s, suggesting that oxidized cyt c could suppress the oxygen consumption by Erv1 in presence of DTT in vitro. Interestingly, previous in vitro studies have shown that cyt c is about a 100‐fold better substrate than oxygen for ALR (Farrell and Thorpe, 2005). Furthermore, when Ccp1 was added in combination with oxidized cyt c, the rate of O2 consumption by Erv1 was slowed by 50–60%, suggesting a synergistic interaction between cyt c and Ccp1 (Figure 6B; Table II; Supplementary Figure 4). As an independent measure of peroxide formation, aliquots from the above Erv1 oxidation reactions were mixed with a colorimetric reagent for the quantitation of H2O2. The levels of H2O2 decreased significantly when both cyt c and Ccp1 were pre‐incubated with Erv1 (Figure 6C). Non‐related proteins BSA and GST were used as controls to confirm the specificity of the reaction. Thus, these in vitro assays support the notion that Erv1 can potentially donate electrons to both O2 and cyt c; however, cyt c is the preferred oxidant in vitro. The observation that the addition of both cyt c and Ccp1 is synergistic can be explained by the known ability of Ccp1 to oxidize cyt c, which would make re‐oxidized cyt c continuously available as an electron acceptor for Erv1. Thus, Ccp1 addition ‘perpetuates’ the reaction by re‐oxidizing the cyt c that is reduced by Erv1.
Ccp1 and cyt c fail to bind to heme in the erv1 mutants
Lill and co‐workers have previously shown that Erv1 plays an unknown role in the maturation of cytosolic FeS clusters (Lange et al, 2001), thereby implying that Erv1 might have several functions. During biochemical analyses, we observed that Ccp1 migrated aberrantly in the erv1 mutants (Figure 1B) and postulated that Erv1 might function in heme loading for Ccp1 and/or cyt c. We therefore tested the ability of cyt c and Ccp1 to bind to heme in the erv1 mutants by investigating protease resistance and assaying for heme binding in colorless native gels (Figure 7). Typically, cyt c and Ccp1 are resistant to exogenous proteases, because they fold tightly around heme (Kaput et al, 1989; Claypool et al, 2006). We incubated mitochondrial lysates from the erv1‐12 (Figure 7A) and erv1‐101 (data not shown) mutants in the presence of increasing concentrations of proteinase K (Figure 7B and C). As expected, protease‐sensitive Kdh was degraded in the presence of proteinase K. In the erv1‐12 mutants, both cyt c and Ccp1 were surprisingly protease sensitive. The erv1 mutants were then grown in media supplemented with hemin, which acts as an exogenous heme source for heme‐binding proteins in the absence of heme synthesis. Upon heme supplementation, protease resistance to Ccp1 and cyt c was nearly restored, indicating that the protein's folding state was corrected by heme addition. These results indicate that heme binding may be impaired or rate‐limiting for both cyt c and Ccp1 assembly in the erv1 mutants and that Erv1 may also function in heme maturation.
We adapted the staining method used for detecting heme‐binding proteins in denaturing gels to colorless native gels (Stugard et al, 1989; Grandier‐Vazeille and Guerin, 1996), because Ccp1 does not bind to heme covalently. Mitochondria from the parent and erv1‐12, erv1‐101, Δccp1, and ccp1H242P (Tatsuta et al, 2007) strains were solubilized in 1% digitonin and separated on colorless native gels (Figure 7D). The ccp1H242P strain contains a mutation in the heme‐binding site that renders the enzyme inactive, but is postulated to still bind heme weakly (Kaput et al, 1989). From immunoblot analysis of colorless native gels, Erv1, Ccp1, and cyt c seemed to migrate at a similar molecular weight in mitochondria that have been shifted to 37°C (indicated by the arrows). As purified cyt c does not migrate into the colorless native gel because of its very basic isoelectric point, the data of Figure 7D suggest that cyt c might be complexed with both Erv1 and Ccp1 (to form a ternary complex) at the nonpermissive temperature. In addition, the heme staining assay indicated that Ccp1 in the erv1 mutants binds to heme weakly, because it shares the same aberrant heme‐binding pattern observed in Ccp1H242P. We, therefore, propose that in addition to disulfide bond assembly and FeS cluster maturation, Erv1 may play a role in heme maturation/export in the IMS.
Erv1 does not play a role in Ccp1 processing
Ccp1 from the erv1 mutants at 37°C migrated aberrantly on SDS–PAGE gels (Figure 1B); we, therefore, inspected Ccp1 processing in detail (Supplementary Figure 5). Lysates were purified from the Δafg3 strain, which mediates the first cleavage of the Ccp1 targeting sequence, and Δpcp1, which mediates the second cleavage, as controls (Esser et al, 2002; Tatsuta et al, 2007). Ccp1 migration was investigated on an 11% SDS gel by immunoblot analysis. Ccp1 migrated slowly in the Δafg3 and Δpcp1 lysates because of inhibited processing. However, Ccp1 from the erv1 mutants at both 25°C and 37°C migrated identically to Ccp1 from WT mitochondria. Erv1 most likely does not mediate Ccp1 processing; instead Ccp1 may migrate aberrantly in different gel systems (15% SDS–PAGE in Figure 1B versus 11% in Supplementary Figure 5).
This study presents the first detailed examination of the redox behavior of Erv1, putting it into perspective with other sulfhydryl oxidases. In addition, we investigated potential pathways that might be utilized as terminal electron acceptors by Erv1.
Erv1 has three redox active centers
Using the average of the results of redox titrations carried out with either mBBr (Figure 3A) or intrinsic tryptophan fluorescence (Figure 3B) used to monitor the redox state of this dithiol/disulfide couple, the C30–C33 pair seemingly has a midpoint potential of −320 mV (at pH 7.0). DTNB analysis of the thiol content of Erv1 samples poised at defined ambient potentials, as well as non‐reducing SDS–PAGE analysis of these samples, both confirm that this Em=−320 mV couple involves an intermolecular disulfide and is consistent with the idea that this cysteine pair most likely mediates Erv1 dimerization in vivo (Hofhaus et al, 2003). The very low redox potential ensures that this disulfide bond is maintained in the oxidized state continuously, as it is most likely important in maintaining the catalytically active conformation of Erv1 but not involved in its redox cycle.
The Em for the flavin component was determined to be −215 mV at pH 7.0, a value that is more negative than the Em=−178 mV measured for the FAD prosthetic group of ALR at pH 7.5 (Farrell and Thorpe, 2005). As reduction of an FAD involves the uptake of two protons per FAD at all pH values below the pKa values of the reduced FAD, one would predict pH dependence for the Em of an FAD of −59 mV per pH unit. Thus, extrapolation suggests that the Em value for the FAD group of ALR at pH 7.0 would be approximately −150 mV, making the difference in the Em values for the flavins of these two related proteins strikingly different. Another difference in the properties of the Erv1 FAD, compared to this prosthetic group in related proteins, is that a small amount of semiquinone was detected during reductive titrations of both ALR and Erv2 (Farrell and Thorpe, 2005; Wang et al, 2007), whereas no FAD semiquinone was detected during equilibrium electrochemical titrations of Erv1. However, it is possible that an appreciable amount of Erv1 semiquinone could be formed transiently during enzyme turnover.
The most positive redox center found in Erv1 is the redox‐active, intramolecular C130–C133 pair, which has a midpoint potential of −150 mV at pH 7.0. As the most positive redox center in the molecule, it would be expected to serve as the electron sink for the protein under conditions where all three of Erv1's redox centers are allowed to come into redox equilibrium. Interestingly, the equivalent cysteine pair (C142–C145) in ALR has been proposed to be more negative than −178 mV, because this pair was not reduced during reductive titrations (Farrell and Thorpe, 2005). Considering the midpoint potentials measured in this study, the thermodynamically most favorable direction for the flow of reducing equivalents would be from the FADH2 to the C130–C133 disulfide of Erv1. However, the 65 mV difference in the midpoint potentials of the two centers indicates that electron flow in the reverse direction does not face an insurmountably large thermodynamic barrier. In fact, published data from our lab and others (Farrell and Thorpe, 2005; Coppock and Thorpe, 2006; Wang et al, 2007) support the notion that reducing equivalents instead flow in the opposite direction (Figure 8). Furthermore, the transfer of reducing equivalents from the C130–C133 dithiol pair to FAD is likely to be kinetically favorable in that Erv1 oxidizes substrates such as Mia40 in the IMS (Terziyska et al, 2005). Although the midpoint potential of Mia40 is not known, we would predict that it would be more reducing than the C130–C133 dithiol pair.
Erv1 shuttles electrons to cyt c
Sulfhydryl oxidases, like Erv1, catalyze the formation of disulfide bonds using molecular O2 and releasing H2O2. Therefore, it seemed likely that O2 should serve as the final electron acceptor for Erv1‐mediated reactions. From our study, cyt c is a likely candidate for a direct electron acceptor, because it partners with Erv1 in organello (Figure 2A) and in vitro as shown by the spectral perturbation assays (Figure 4B). In addition, Erv1 directly reduced cyt c, indicating that electrons can flow from Erv1 to cyt c (Figure 4C). From our genetic studies, the erv1‐101 mutant required CYC3 for viability under anaerobic conditions. However, an obvious mechanism for the reduction of cyt c by the redox‐active C130–C133 dithiol pair is not apparent. While this reaction would be thermodynamically favorable given the Em for cyt c is +250 mV and for C130–C133 pair is −150 mV, cyt c reduction is a one‐electron reaction and dithiol oxidation to a disulfide is a two‐electron reaction. As our spectral perturbation measurements indicate that there is only one high‐affinity binding site for cyt c on Erv1, this pathway for cyt c reduction would require the coupling of a two‐electron oxidation to a one‐electron reduction. Reduction of cyt c by Erv1's flavin, if it were to involve a one‐electron transfer involving a transiently formed FAD semiquinone seems more likely, but clearly additional experiments are required before a detailed mechanism is available.
Studies by Tokatlidis and co‐workers suggest that the protein import process is linked to the electron transport chain via the connection of Erv1 with cyt c (Allen et al, 2005); however, we show that the erv1 mutants are viable on ethidium bromide, which disrupts the electron transport chain. Therefore, additional electron acceptors must be present to reoxidize cyt c. Ccp1 is one logical candidate for this role and our demonstration that Ccp1 acts synergistically with cyt c in competing with O2 for reduced Erv1 (Figure 6B and C) is consistent with this hypothesis. In addition, H2O2 levels accumulate in the erv1 mutants at the restrictive temperature (Figure 5B). Thus, we propose that one of the Erv1 electron acceptor pathways relies on cyt c (Figure 8A). A large proportion of oxidized cyt c is available to accept electrons from Erv1 in the mitochondrion (Morgan and Wikstrom, 1991; Cortese et al, 1995). Reduced cyt c can then be re‐oxidized using a variety of pathways, including the electron transport chain and Ccp1. Because cyt c is oxidized and reduced on the same face of the protein (Bosshard et al, 1986), interactions with Erv1 are most likely dynamic. This potentially dynamic interaction may allow Erv1 to use O2 and additional acceptors. Interestingly, the yeast genome has two genes for cyt c, CYC1 and CYC7, which are expressed under high and low O2 concentrations, respectively (Burke et al, 1997). These different forms of cyt c may accommodate Erv1 function under different O2 concentrations.
Ccp1 functions as a terminal electron acceptor for Erv1
Erv1 also can use O2 as an electron acceptor, producing H2O2 in the IMS. We propose that Ccp1 efficiently removes accumulating H2O2, reducing it to water (Figure 8B). Oxidized Ccp1 in turn can be reduced and recycled by reduced cyt c. Our studies with the O2 electrode suggest that Ccp1 and cyt c are synergistic with Erv1 and that both proposed pathways in Figure 8 may occur simultaneously. The results from colorless native gel analysis (Figure 7D) argue that Erv1, Ccp1, and cyt c may form a ternary complex in the erv1 mutants at the restrictive temperature; however, this was not confirmed with other biochemical methods such as co‐purification (Figure 2A and B). Subsequent studies will be required to identify additional bonafide Erv1 partners and substrates.
Based on our genetic studies, we also propose that an anaerobic pathway is present to recycle Erv1, because the erv1 mutants grew in anaerobic conditions and in the absence of an electron transport chain. It may be that the cyt c coded by CYC7 facilitates electron transfer under these conditions, but other unidentified acceptors may also be present. This is not unexpected because similar strategies are employed by bacteria dependent upon the aerobic/anaerobic state (Bader et al, 1999).
Erv1 has also been proposed to play a role in the export of FeS clusters and cellular iron homeostasis (Lange et al, 2001), indicating Erv1 may have more than one function in the IMS. In our study of erv1 mutants, cyt c and Ccp1 showed compromised heme binding as demonstrated by the increased sensitivity to protease. This phenotype was partially restored by the addition of exogenous hemin, suggesting that erv1 mutants are probably heme deficient. Additionally, the erv1 mutants showed the same aberrant heme staining pattern on colorless native gels as observed in the ccp1H242P strain. Ccp1 in the ccp1H242P strain shows reduced heme binding, because the imidazole of the specific histidine residue is involved in a pseudocovalent bond with the heme iron (Bosshard et al, 1986). This may be a direct effect caused by nonfunctional Erv1 or a secondary effect caused by general changes in the redox status of the IMS. At present, the details of heme export into the IMS are not clearly understood.
This study begins to dissect the different possible electron‐accepting pathways for re‐oxidizing reduced Erv1. Cyt c, which forms a high‐affinity 1:1 complex with Erv1, has been shown to serve as a direct electron acceptor from reduced Erv1. Ccp1 serves a dual function, reducing the H2O2 generated by Erv1's reduction of O2 and also re‐oxidizing the cyt c reduced directly by Erv1. Additional studies, designed to elucidate these pathways in greater detail, are currently underway in our laboratories.
Materials and methods
Plasmids and strains
To generate recombinant Erv1 and Ccp1, both containing a C‐terminal His6 tag, the open reading frames were cloned separately into the pET28a vector (Novagen). Erv1‐His6 and Ccp1‐His6 were induced in BL21‐CodonPlus (DE3)‐RIL Escherichia coli (Stratagene) and purified under native conditions using Ni2+‐agarose (Qiagen) as per the manufacturer's instructions.
Standard genetic techniques were used for the yeast strains. Strains listed in this study are described in the strain table (Supplementary Table I) Temperature‐sensitive (ts) erv1 strains were generated using error‐prone PCR and plasmid shuffling as described previously (Koehler et al, 1998). Briefly, the erv1 mutants were cloned by in vivo recombination with a gapped centromeric plasmid in strain yEPL1. After plasmid shuffling to remove the WT ERV1 plasmid, the erv1 mutants were identified based on growth at 25°C and arrest at 37°C; the mutants were designated erv1‐101 and erv1‐12. The Erv1‐His tagged strain was constructed by subcloning ERV1 with a 10XHis tag into pRS425, and transforming this plasmid into the yEPL1 strain for plasmid shuffling. For heme studies, yeast cells were grown in the presence of 20 μM hemin before mitochondrial isolation. The yeast strain ccp1H242P (Tatsuta et al, 2007) and plasmid pERV1:URA32μ were kindly provided by Dr Thomas Langer and Dr Patrice Hamel, respectively.
Detailed Materials and methods for additional techniques are provided in the Supplementary data.
Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).
We thank T Langer (University of Cologne) for the ccp1H242P strain, P Hamel (Ohio State University) for the plasmid pERV1:URA32μ and S Claypool (UCLA) for the Δafg3 strain. This work was supported by USPHS National Service Award GM070415 (EPL), NIH R01GM061721 (CMK), Muscular Dystrophy Assoc Grant 022398 (CMK), American Heart Association Grant 0640076N (CMK), the Robert A. Welch Foundation D–0710 (DBK), and the US Department of Energy DE‐FG03‐99ER20346 (DBK). CMK is an Established Investigator of the American Heart Association.
- Copyright © 2007 European Molecular Biology Organization