|Home | About | Journals | Submit | Contact Us | Français|
Augmenter of liver regeneration (ALR) is both a growth factor and a sulfhydryl oxidase that binds FAD in an unusual helix-rich domain containing a redox-active CxxC disulfide proximal to the flavin ring. In addition to the cytokine form of ALR (sfALR) that circulates in serum, a longer form, lfALR, is believed to participate in oxidative trapping of reduced proteins entering the mitochondrial intermembrane space (IMS). This longer form has an 80-residue N-terminal extension containing an additional, distal, CxxC motif. This work presents the first enzymological characterization of human lfALR. The N-terminal region conveys no catalytic advantage towards the oxidation of the model substrate dithiothreitol (DTT). In addition, C71A or C74A mutations of the distal disulfide do not increase the turnover number towards DTT. Unlike Erv1p, the yeast homolog of lfALR, static spectrophotometric experiments of the human oxidase provide no evidence for communication between distal and proximal disulfides. An N-terminal his-tagged version of human Mia40, a resident oxidoreductase of the IMS and a putative physiological reductant of lfALR, was subcloned and expressed in Escherichia coli BL21 DE3 cells. Mia40, as isolated, shows a visible spectrum characteristic of an Fe/S center and contains 0.56 ± 0.02 atoms of iron per subunit. Treatment of Mia40 with guanidine hydrochloride and triscarboxyethylphosphine hydrochloride during purification removed this chromophore. The resulting protein, with a reduced CxC motif, was a good substrate of lfALR. However, neither sfALR, nor lfALR mutants lacking the distal disulfide, could oxidize reduced Mia40 efficiently. Thus, catalysis involves a flow of reducing equivalents from the reduced CxC motif of Mia40, to distal- and then proximal CxxC motifs of lfALR, to the flavin ring, and, finally, to cytochrome c or molecular oxygen.
Hepatic regenerative stimulator substance (1), now termed hepatopoietin (2) or augmenter of liver regeneration (ALR1 (3)), was first identified as a protein that stimulated hepatocyte proliferation and liver regeneration (1–4). ALR is believed to activate the MAP kinase pathway (5, 6), inhibits natural killer cells (7) and influence the regeneration of Drosophila imaginal discs (8). The sequence of this growth factor showed homology with two yeast proteins, Erv1p and Erv2p (3, 9, 10). All three proteins were found to be FAD-linked sulfhydryl oxidases (11–14) and the subsequent crystal structures of Erv2p (15) and ALR (16) identified a new flavin-binding fold with the isoalloxazine ring inserted into the mouth of a 4-helix bundle. A depiction of dimeric ALR is shown in Figure 1A. Mammalian ALR is found in two main alternatively spliced forms (2, 6, 12, 17–20). The short form (sfALR, 15 kDa) is an extracellular cytokine and also participates in intracellular redox-dependent signaling pathways (1–4, 6, 18). The long form of the oxidase (lfALR, 23 kDa) exists predominantly in the mitochondrial intermembrane space (IMS) and contains an 80-amino acid N-terminal extension housing a second CxxC motif. The presence of additional redox-active disulfides in N- or C-terminal extensions from the core flavin binding domain is a common feature of the Erv/ALR family (21–23) where they may mediate the transfer of reducing equivalents between thiol substrate and flavin prosthetic group (see later). Fig 1B shows that these distal disulfides are placed within an N-terminal extension of lfALR and its yeast ortholog Erv1p. These regions show little sequence conservation and, in particular, the CxxC motif of lfALR is some 30 residues closer to the Erv/ALR flavin binding domain than the corresponding distal disulfide in Erv1p.
lfALR and yeast Erv1p play key roles in the generation of disulfide bonds in the mitochondrial intermembrane space (IMS) as depicted for Erv1p in Figure 2. Studies using yeast have suggested that oxidized Mia40 exchanges disulfides with a variety of proteins undergoing oxidative folding in the IMS: for example proteins with twin Cx3C motifs (the Tim 8, 9, 10, 12, and 13 proteins), twin Cx9C motifs (Cox17, Cox19, Mic14, Mic17, and Mdm35), the copper chaperone Ccs1, and Erv1p, itself (24–35). Reduced Mia40, so generated, is then the immediate substrate of Erv1p (Figure 2).
In contrast to the yeast Erv1p/Mia40 system, rather little is known concerning possible interaction between human ALR and potential thiol substrates in the IMS. Indeed, only sfALR has been subject to a detailed enzymological investigation (12, 36). This small dimeric flavoprotein (Figure 1A) is a very weak stand-alone sulfhydryl oxidase for reduced lysozyme (12). Dithiothreitol was later used as a convenient model substrate and showed kcat values of 50–75/min with a range of constructs of sfALR (36). In terms of the physiological oxidant of ALR, Farrell and Thorpe found that cytochrome c was a much better oxidant than molecular oxygen in vitro (36). They suggested that channeling reducing equivalents via the respiratory chain would avoid the generation of hydrogen peroxide associated with disulfide bond generation in the IMS (36). Support for this suggestion has subsequently come from in vitro and in vivo studies with yeast Erv1p (35, 37). Recently, Koehler and colleagues have suggested an alternate pathway involving cytochrome c peroxidase-mediated oxidation of cytochrome c driven by the hydrogen peroxide generated by Erv1p (38).
The present work presents the first enzymological comparison of the long and short forms of human ALR. Reductive titrations and mutagenesis are used to explore the mechanistic consequences of the inclusion of an additional distal CxxC motif within the 80-residue N-terminal extension. While the properties of the full-length Erv1p (13, 39) are one benchmark for our continuing studies on ALR, their distal CxxC motifs occupy different locales in divergent N-terminal regions (see earlier, Figure 1B). We further extend our studies to include the presumed physiological substrate of human lfALR, Mia40. This work suggests unanticipated new roles for Mia40 and reinforces how much more we need to know concerning oxidative protein folding and metal trafficking systems in the mitochondrial IMS.
Ampicillin, kanamycin, chloramphenicol, tetracycline, guanidine hydrochloride, riboflavin, imidazole, urea, FAD, horse heart cytochrome c, GSH, GSSG, β-mercaptoethanol, bovine erythrocyte superoxide dismutase and lysozyme were obtained from Sigma-Aldrich. Tryptone, yeast extract, NaCl, monobasic potassium phosphate, and EDTA were from Fisher Scientific. IPTG was obtained from Promega and TCEP from Pierce. DTT and DTNB were obtained from Acros Organics. Primers for mutagenesis were ordered from Integrated DNA Technologies.
The clone for the long form of ALR was obtained from ATCC (IMAGE ID: 3346679) and was then subcloned into pTrcHisA (Invitrogen) using NheI and HindIII restriction sites. PCR products were purified using Qiagen PCR purification kit and were digested with appropriate restriction enzymes. pTrcHisA (Invitrogen) was also digested with the same restriction enzymes. DNA was purified using Qiagen gel extraction kit and ligated using T4 DNA ligase. Ligation products were transformed into E. coli TOP10 cells (Invitrogen) followed by sequence confirmation of the construct. The resulting plasmid was finally used to transform E. coli BL21 DE3 chemically competent cells (Invitrogen). Plasmids recovered from these cells were further sequenced to confirm that no additional changes had been introduced. Experimental conditions for expression and purification of lfALR were as reported earlier (36) with the modifications mentioned here. ALR bound to the nickel-NTA column (ProBond; Invitrogen) was re-equilibrated with 10 mL of 50 mM phosphate buffer, adjusted to pH 7.5 in the presence of 6 M guanidine hydrochloride and 5 mM TCEP. Treatment with reductant and denaturant allowed for the subsequent removal of contaminants that were linked to folded lfALR via disulfide bonds. This step is not necessary with the short-form of ALR (36). The column was then incubated for 2 h at 25 °C prior to washing with 50 mM potassium phosphate buffer, pH 7.5, at 4 °C. The column was re-equilibrated with 100 µM FAD in the same buffer, incubated overnight, and then eluted with imidazole gradients as detailed previously (36).
His-tag removal of lfALR’ was done by incubating the protein with a His-tagged TEV protease (40) at 30 °C for 6 h (0.1 A280 units of TEV-protease for every 1 A456 units of flavoprotein) followed by passage through a Ni-NTA resin. The digested product was evaluated by SDS-PAGE to confirm removal of the N-terminal his-tag.
A clone for human Mia40 was obtained from ATCC (IMAGE ID 4538841) in the pCMV-SPORT6 vector. The sequence for Mia40 was PCR amplified using the N-terminal forward primer 5’-GATGGATCCATGTCCTATTGCCGGCAG-3’ (BamHI site underlined) and a C-terminal reverse primer 5’-GGCGAATTCTTAACTTGATCCCTCCTCTTC-3’ (EcoRI site underlined, stop codon italicized). The PCR products were purified, using a Qiagen PCR purification kit, and digested with the restriction enzymes for 2 h at 37 °C. The pTrcHisA plasmid (Invitrogen) was also digested with the same restriction enzymes and the linearized plasmid purified using a Qiagen gel extraction kit. Ligation with T4 DNA ligase was done for 3 h at room temperature. The ligated product was transformed into E. coli TOP10 cells (Invitrogen) and the recovered plasmid was sequenced. The plasmid was re-sequenced after transformation into E. coli BL21 DE3 and E. coli Rosetta-gami cells to ensure that no unanticipated mutations had been introduced.
The primers used for site directed mutagenesis (QuikChange, Stratagene) are listed in Supporting Information. Each construct was sequenced to confirm that only the desired changes were effected.
For expression of Mia40 in E. coli BL21 DE3 cells, a glycerol stock was used to inoculate four 5 mL starter cultures (LB medium containing 50 µg ampicillin/mL) and grown overnight at 37 °C. The cultures were used to inoculate four 2 L flasks containing 500 mL of the same media. At an A600 value of approximately 0.6, the cells were induced with 1 mM IPTG and shaken for an additional 5–6 h at 37 °C. Cells were harvested by centrifugation (4000 g, 30 min, 4 °C) and stored at −20 °C. Expression conditions were essentially the same for Rosetta-gami-DE3 cells, except that the media were additionally supplemented with kanamycin (30 µg/mL), chloramphenicol (20 µg/mL), and tetracycline (12 µg/mL). Bacterial pellets were thawed and resuspended in a total of 25 mL of 50 mM potassium phosphate buffer, pH 7.5, containing 300 mM NaCl, 0.1 mg/mL lysozyme, and one EDTA-free protease inhibitor tablet (Roche). The suspension was passed twice through a French Press (at 10,000 psi), followed by brief sonication to shear DNA. The cell lysate was then centrifuged at 6000 g for 30 min. β-Mercaptoethanol (10 mM) was subsequently added to the supernatant followed by 2.5 mL Ni-NTA resin, previously equilibrated with 50 mM potassium phosphate buffer, pH 7.5. The mixture was rocked for 1 h at 4 °C and then transferred to an empty chromatography column to allow elution of unbound material. The bound protein was then treated in two ways as detailed below.
When removal of an endogenous Fe/S chromophore (see later) was desired, the bed volume of the resin was exchanged with 50 mM potassium phosphate buffer, pH 7.5, containing 6 M guanidine hydrochloride and 5 mM TCEP, and the beads were incubated for 2 h at room temperature. The column was then transferred to the cold room and washed with 40 mL of 50 mM potassium phosphate buffer, pH 7.5, containing 10 mM β-mercaptoethanol followed by 5 mL aliquots of this solution supplemented with 50, 200, and 500 mM imidazole. Fractions were pooled after evaluation of purity by SDS-PAGE using 12% gels subsequently stained with Coomassie Brilliant Blue. Purified Mia40 was concentrated using Amicon Ultra-15 (Millipore; 10 kDa cut off), and then desalted by applying about 0.5 mL aliquots to PD-10 columns (GE Healthcare Life Sciences) equilibrated with 50 mM potassium phosphate buffer, pH 7.5, containing 0.3 mM EDTA. Complete separation between protein thiols and β-mercaptoethanol was verified by reacting aliquots from each fraction with DTNB. Reduced protein usually emerges in fractions 5–7 and β-mercaptoethanol elutes after fraction 11. Reduced Mia40 fractions were pooled and stored under nitrogen at 4 °C. Thiol titer was rechecked using DTNB before use.
When the iron-containing form of Mia40 was being purified, the Ni-NTA bound protein was washed with 12 mL of 50 mM phosphate buffer, pH 7.5, containing 30 mM imidazole, 300mM NaCl and 10mM β-mercaptoethanol. A second wash omitted NaCl and a third wash, additionally, omitted β-mercaptoethanol. The column was then developed with 5 mL aliquots of 50 mM phosphate buffer, pH 7.5, containing 50, 200, and 500 mM imidazole. Mia40-containing fractions were evident by their brown color and were pooled and desalted with a PD-10 column preequilibrated with 50 mM potassium phosphate buffer, pH 7.5.
Anaerobic methods and cytochrome c assays were performed as described earlier (36). The extinction coefficient for the flavin cofactor bound to sfALR was 11.6 mM−1 cm−1 at 456 nm (36). Extinction coefficients measured in this work were obtained by recording spectra before and 1 min after the addition of 0.1% sodium dodecyl sulfate (from a 10% stock solution in water) to a solution of enzymes in 50 mM Tris buffer, pH 7.5, containing 0.3 mM EDTA. The extinction coefficient of free FAD in 0.1% detergent is 11.3 mM−1 cm−1 at 448 nm (23). The extinction coefficients used in this work were: lfALR’ (11.7 mM−1 cm−1); C71A and C74A distal mutants (11.7 mM−1 cm−1); C142A and C145A mutants of the proximal disulfide (10.6 and 12.2 mM−1 cm−1, respectively). Metal-free Mia40 concentrations were estimated using ProtParam (41) using an extinction coefficient of 19.9 mM−1 cm−1 at 280 nm. By matching the concentrations of metal-free and iron-containing Mia40 using the Bradford protein assay, an extinction coefficient of 27.5 mM−1 cm−1 was determined for the metallated protein.
Metal analysis of LB medium (16 g tryptone, 16 g yeast extract, 5 g NaCl and 2.5 g of monobasic potassium phosphate in 1 L of water adjusted to pH 7.5), and purified recombinant Mia40, was performed using a Dual View Iris Intrepid II XSP ICP spectrometer maintained at the Soil Testing Facility, Department of Plant and Soil Sciences, University of Delaware. Samples (3 mL) were mixed with 10 mL of concentrated nitric acid, sealed in XP 1500 Teflon-coated flasks and digested for 1 min at 175 °C followed by a subsequent digestion at 180 °C for 5 min in a MARS 5 microwave oven. After cooling, the digests were brought to 50 mL with distilled deionized water and the samples were analyzed as above.
The long form of ALR was expressed with an N-terminal hexa-histidine tag for ease of purification (Supplemental Figure S1). As observed with sfALR (36), the presence of two non-conserved, and catalytically non-essential, cysteine residues (C154 and C165) in the long form led to the slow formation of a yellow-colored protein precipitate on storage. Accordingly, a double C154A/C165A mutant was generated: this construct (termed lfALR') showed comparable activity to the wild type lfALR using DTT (entries 3 and 4; Table 1) and remained soluble upon prolonged storage at 4 °C. Removal of the histidine tag on lfALR' by TEV protease led to only minor changes in activity towards DTT (entry 5; Table 1; see Experimental Procedures). Hence the bulk of the experiments in this paper have used His-tagged constructs for ease of purification. There are small, but significant, differences in kcat/Km values for the oxidation of DTT between short and long forms of ALR: the long form is about 3-fold less active (a value dominated by a 4-fold increase in Km value; entries 2 and 4; Table1). While we do not yet understand this effect, we later describe a more marked difference in the reactivity of short and long forms of ALR.
lfALR' showed no detectable free thiols when assessed with DTNB, consistent with the expected combined presence of 8 disulfides in this covalent dimer ((36); see Experimental Procedures). Hence the additional distal CxxC motif (C71/C74) was disulfide bridged in the oxidase as isolated. Figure 3A shows selected spectra recorded during a dithionite titration of lfALR' under anaerobic conditions. Each subunit of lfALR' contains three potential redox centers: the flavin cofactor, the proximal disulfide shown in Figure 1A, and the distal CxxC disulfide housed in the flexible N-terminal region (Figure 1B and Figure S1, Supporting Information). However, the inset to Figure 3A shows that the flavin moiety is essentially completely reduced, as evident by the bleaching in absorbance at 456 nm, before the significant accumulation of reducing equivalents on either the proximal or distal CxxC motifs. Thus the flavin in lfALR' has a significantly more positive redox potential than either of these additional redox centers. As was observed for sfALR' (36), reduction of lfALR' is accompanied by the formation of significant levels of blue flavosemiquinone radical as evident by the appearance of the structured long-wavelength absorbance beyond 530 nm at the midpoint of the titration (42). In addition, inclusion of the 80 amino acid N-terminal extension does not change one of the most striking aspects of the reductive behavior of ALR: the generation of almost quantitative blue flavin semiquinone as oxygen is depleted during turnover with DTT ((36, 43), Figure 3B).
The C142A mutation of lfALR' would leave C145 able to interact with the isoalloxazine ring (the sulfur atom of C145 is within 3.2 Å of the C4a position of the isoalloxazine ring; Figure 1A). This mutation does indeed generate the expected (44) charge-transfer complex characterized by a wedge-shaped absorption feature extending from 520 nm to beyond 700 nm (Figure 4). Subsequently the thiolate would be expected to form a C4a flavin adduct as an intermediate in the reduction of the enzyme flavin (44, 45). In contrast, the C145A mutant shows a normal yellow color devoid of significant charge-transfer absorbance (Figure 4). There is, however, a significant blue shift in the absorption maximum (from 456 to 453 nm) consistent with other proteins in which a disulfide bond proximal to a bound flavin is broken by reduction or mutagenesis (46–48). As expected for the role of a proximal disulfide in sulfhydryl oxidase catalysis, both C142 and C145 mutants of lfALR' showed very low activity towards DTT (Table 1 entries 9 and 10 respectively).
In contrast, two mutants of the distal disulfide (C71A and C74A) showed almost the same kcat values with DTT and 6- to 8-fold lower Km values towards DTT (Table 1 entries 6 and 7 respectively). Thus the absence of the distal disulfide (either in the context of mutation of the long form of ALR, or when the N-terminal extension is totally truncated as in sfALR) leads to a several-fold decrease in Km. While we do not understand the basis of these subtle effects, it is clear that DTT can access the proximal disulfide directly: the distal disulfide is not required when this non-physiological substrate is employed.
Both C71A and C74A mutants of lfALR' exhibit normal flavin spectra with no evidence of long wavelength bands (Supplemental Figure S2A). This behavior is distinct from that of the yeast Erv1p studied by Lisowsky and colleagues (39). They observed that, while C33S of the distal disulfide gave a yellow enzyme, the corresponding C30S form showed an intense long wavelength charge transfer band (39). The feature most reasonably reflects an internal mixed disulfide C33-C130 releasing C133 to interact with the flavin (49). While these data with yeast Erv1p provide strong circumstantial evidence for communication between distal and proximal disulfides, mutation of either C30 or C33 had only minor effects using DTT as a substrate of the yeast oxidase (39). Nevertheless, Lisowsky and colleagues report that the distal disulfide of yeast Erv1p appears essential in vivo (39).
The short form of ALR was previously found to show undetectable activity towards reduced glutathione (13). The presence of the additional distal disulfide in lfALR' did not accelerate turnover significantly (with values of < 1/ min using 5 mM glutathione; data not shown). We were further unable to detect measurable turnover of lfALR' with unfolded reduced RNase (at 250 µM RNase thiols in 50 mM phosphate buffer, pH 7.5; see Experimental Procedures; data not shown). Lisowsky and colleagues report that sfALR has very low activity towards reduced lysozyme in 2 M urea (12, 50). Hence it appears that neither short nor long forms of ALR are efficient general catalysts for the oxidation of unfolded reduced proteins in vitro.
While DTT has proved a very useful model substrate for several flavin-dependent sulfhydryl oxidases (16, 36, 48, 51), its small size and marked chemical reactivity can complicate mechanistic inferences. For example, the catalytic advantage of a distal disulfide may be underestimated if DTT can reduce the proximal disulfide directly: such is the case with Erv2p and QSOX (22, 23). Since Mia40 has been shown to be a substrate of yeast Erv1p (Figure 2; (24, 35, 37, 38, 52–57) we decided to test human Mia40 with short- and long-forms of ALR. It is important to note that there are significant differences in amino acid sequence between the yeast and human Mia40 proteins and wide variation between the N-terminal regions of yeast Erv1p and human ALR (Figure 1B). Hence commonalities of behavior between yeast and human mitochondrial oxidative folding systems cannot be assumed without experimental verification.
Human Mia40 is a member of the twin Cx9C protein family (52, 53, 55–58). The cartoon depiction (Figure 5) is based on its homology with the human Cox17 structure (59) and coincides adequately with a very recent NMR structure of a construct of human Mia40 (56). A single non-conserved cysteine residue (C4) close to the flexible N-terminus of human Mia40 is followed by a redox-active, and functionally important, CxC motif (58, 60). Two structural disulfides (58, 60) link the helix-turn-helix structural core of human Mia40 (Figure 5). In yeast/fungi, Mia40 has a much longer N-terminal extension that contains a transmembrane segment tethering the core domain facing the IMS (52, 57, 58, 60). Human Mia40 lacks this transmembrane feature and could be expressed as a soluble protein in E. coli (56, 58). An N-terminal His-tagged version of Mia40 was purified under reducing conditions and treated with guanidine hydrochloride to remove an endogenous chromophore (see later; see Experimental Procedures). The protein, as isolated, contains 3 thiols by DTNB titer. Since the Cx9C structural disulfide pairs of the yeast (60, 61) and human (56) proteins are very resistant to reduction, we can assume that C4, C53 and C55 of our construct of human Mia40 are reduced as isolated (Figure 5). This reduced protein is a significant substrate of lfALR' (Figure 6). After 30 min incubation with 30 µM reduced Mia40 (90 µM protein thiols), approximately two out of three thiols have been oxidized. This is consistent with the generation of a disulfide bond between C53-C55, while leaving the non-conserved N-terminal cysteine residue reduced Figure 5. In accord with this expectation, a C4A mutant (showing two free thiols) was oxidized by lfALR' at a comparable rate to that shown in Figure 6 (data not shown). Hence this non-conserved residue is not essential for catalysis in vitro. The inset to Figure 6 plots the turnover numbers for lfALR' as a function of the concentration of reduced wild-type Mia40 (yielding kcat and Km values of 13/min and 20 µM respectively). While this turnover number appears modest, the catalytic efficiency (kcat/Km of 11,025 M−1s−1) is some 70-fold higher than for the dithiol reductant, DTT (156 M−1s−1; derived from kcat and Km values of 76/min and 8 mM respectively; Table 1 entry 4). The his-tag of the lfALR' construct was removed using TEV-protease (see Experimental Procedures) to assess its impact on the oxidation of reduced Mia40. Here, the initial rate was approximately two-fold faster than observed above (data not shown).
Importantly, Figure 6 shows that reduced Mia40 is not a significant substrate of sfALR' (open triangles). In contrast, DTT is a slightly better substrate for sfALR' than for lfALR' (Table 1, entries 2 and 4). Further, a C71A/C74A double mutant of the distal N-terminal disulfide in lfALR' abolished activity towards reduced Mia40 (Figure 6, closed circles). Hence the N-terminal extension of ALR, per se, does not confer reactivity to the core ALR domain towards reduced Mia40: the distal disulfide within this sequence is a critical determinant. Finally, reduced Mia40 is not a substrate of proximal disulfide mutants of lfALR' (either C142A or C145A; not shown). In sum, these data provide the first in vitro support for the role of human Mia40 as a direct substrate of human lfALR, and for the involvement of the distal disulfide in the oxidase as a mediator in the transfer of reducing equivalents from the CxC motif of Mia40 to the proximal disulfide of ALR.
Cytochrome c was initially suggested as a potential electron acceptor for ALR on the basis of experiments with the short form of this enzyme (36). However, lfALR is the principal ALR variant in the IMS (34, 62), and hence we determined whether it was also capable of directly reducing cytochrome c. Figure 7 shows that cytochrome c is able to serve as an effective oxidant in air-saturated solution. As observed with sfALR', this reaction is only slightly inhibited by superoxide dismutase (Figure 7) showing that lfALR reduces the cytochrome largely directly, rather than via the intermediacy of the superoxide anion (36). After correction for non-enzymatic reduction of the cytochrome by reduced Mia40, the resulting initial rates show a concentration dependence that could be fitted to kcat and Km values of 33/min and 14 µM respectively for cytochrome c. The corresponding catalytic efficiency kcat/Km of 3.9 × 104 M−1s−1 towards cytochrome c is some 12-fold lower than that obtained using sfALR' (270/min, 10 µM; 4.5 × 105 M−1s−1) measured with DTT as the reducing substrate (36). Several factors might impact this decreased catalytic efficiency. For example, the 80-residue N-terminal extension in lfALR may partially impede the approach of cytochrome c. In addition, the N-terminal 80-residues (calculated pI of 8.9, see Supplemental Table S1) might confer some measure of electrostatic repulsion to the highly-basic cytochrome (pI of about 10 (63)) at pH 7.5. In contrast, the positive charge of the N-terminal region of lfALR might favor the approach of the negatively-charged human Mia40 (pI 4.2; Supporting Information Table S1).
Previously Cox17, a twin Cx9C protein, has been shown to bind copper via its N-terminal CC motif (59, 64, 65). Further, Terziyska et al. have demonstrated that yeast Mia40 binds both copper and zinc and have suggested that this interaction is functionally important (52). This study used a maltose binding protein fusion construct of yeast Mia40 expressed in E. coli cells that were grown in media supplemented with 100 µM of either cupric sulfate or zinc acetate (52). When we purified human MIA40 in the presence of β-mercaptoethanol, but in the absence of denaturant (see Experimental Procedures), the protein always showed a distinct brown color with a visible spectrum suggestive of an iron-sulfur cluster (Figure 8). The chromophore bleached in air-saturated phosphate buffer, pH 7.5, with a half-time of approximately 2 weeks at 4 °C using 130 µM Mia40 (data not shown). Inductively-coupled plasma spectrometry (see Experimental Procedures) of three independent preparations showed 0.56 ± 0.02 atoms of iron compared to 0.010 ± 0.003 and 0.065 ± 0.012 atoms of copper and zinc respectively (See Supporting Information, Figure S3). These data were obtained with standard LB growth media unsupplemented with additional metals. Analysis of the media used for these experiments showed metal contents (15 µM Fe, 28 µM Zn and 0.5 µM Cu) that were considerably lower than the 100 µM supplements of zinc and copper used earlier (52).
Since we have previously expressed other disulfide-containing proteins in E. coli Rosetta-gami™ DE3 cells (23), a strain with a more oxidizing cytosol than normal E. coli cells, we investigated their utility in the expression of Mia40 (see Experimental Procedures). The resulting protein showed a thiol titer of 0.90 ± 0.18, consistent with a disulfide at the CxC motif with the non-conserved C4 as a free thiol. Since metal binding would likely involve the reduced CxC motif (see above), the oxidized protein would be expected to contain little or no iron. In accord with this expectation, oxidized Mia40 showed little absorbance in the visible region (Figure 8) and an iron content of 0.053 atoms/subunit (Figure S3 in Supporting Information). The recent NMR structure of human Mia40 by Banci et al. also utilized E. coli Origami cells to obtain oxidized protein and it is thus not surprising that their preparations did not contain appreciable iron: their structure shows a CxC disulfide in the N-terminal region of the protein.
It is recognized that considerable caution needs to be exercised when drawing inferences from the metal content of eukaryotic proteins that have been expressed heterologously in bacteria (66). For example, the metal homeostasis mechanisms, including the metal chaperone systems and the intracellular concentrations of metal ions, may differ profoundly between organisms and hence confound conclusions of physiological relevance (66–69). However our results are of interest because Lill and coworkers have reported that yeast expressing a temperature-sensitive mutant of Erv1p were unable to assemble cytosolic Fe/S centers at non-permissive temperatures (34, 67). They further demonstrated that lfALR and Erv1p were functional orthologs in Fe/S center trafficking across the IMS (34). We examined whether an Fe/S center could be incorporated into human lfALR' (exploiting the usual methods of incubation with ferrous ammonium sulfate, sodium sulfide and β-mercaptoethanol under anaerobic conditions (70–72)). These exploratory experiments were unconvincing (data not shown). Correspondingly, Lange et al. found no stable association of an Fe/S cluster with yeast Erv1p even under anaerobic conditions (34). Our present finding that Mia40 binds iron strongly in vitro suggests an alternative hypothesis: that Mia40 may participate in an Fe/S delivery system that is enabled, directly or indirectly, by Erv1p in yeast or by lfALR in human mitochondria.
This contribution shows that the reduced CxC motif in human Mia40 transfers reducing equivalents to the mitochondrial form of ALR, but not to the shorter variant that functions as a circulating growth factor. The reaction is obligatorily dependent on the distal CxxC found in the N-terminal extension of lfALR. While communication between distal and proximal disulfide/dithiol redox centers in lfALR is not evident from static titrations, it proves functionally important in the oxidation of reduced Mia40. In addition to the roles of Mia40 in the oxidative trapping of small proteins within the IMS, our data suggest that Mia40 may also play a direct role in Fe/S traffic in the IMS.
Table S1, Figure S1 – Figure S3 and a listing of primers used for site-directed mutagenesis provide supplementary data, sequences and analysis for human Mia40 and lfALR'. This material is available free of charge via the Internet at http://pubs.acs.org.
We thank Drs. Deborah Fass and David Waugh for a plasmid expressing TEV protease, Cathy Olsen for generous help with metal analyses, and Mr. Vamsi Kodali for insightful comments on the manuscript.
†This work was supported in part by National Institutes of Health Grant (GM26643). The content of this work is solely the responsibility of the authors and does not necessarily reflect the official views of the National Institute of General Medical Sciences or the National Institutes of Health.