Mia40 Is Optimized for Function in Mitochondrial ... - ACS Publications

Jul 1, 2014 - More recently, disulfide-bonded proteins and a thiol oxidase, termed Mia40 (mitochondrial import and assembly), were discovered in the ...
0 downloads 0 Views 3MB Size
Articles pubs.acs.org/acschemicalbiology

Mia40 Is Optimized for Function in Mitochondrial Oxidative Protein Folding and Import Johanna R. Koch and Franz X. Schmid* Laboratorium für Biochemie und Bayreuther Zentrum für Molekulare Biologie, Universität Bayreuth, 95440 Bayreuth, Germany S Supporting Information *

ABSTRACT: Mia40 catalyzes oxidative protein folding in mitochondria. It contains a unique catalytic CPC dithiol flanked by a hydrophobic groove, and unlike other oxidoreductases, it forms long-lived mixed disulfides with substrates. We show that this distinctive property originates neither from particular properties of mitochondrial substrates nor from the CPC motif of Mia40. The catalytic cysteines of Mia40 display unusually low chemical reactivity, as expressed in conventional pK values and reduction potentials. The stability of the mixed disulfide intermediate is coupled energetically with hydrophobic interactions between Mia40 and the substrate. Based on these properties, we suggest a mechanism for Mia40, where the hydrophobic binding site is employed to select a substrate thiol for forming the initial mixed disulfide. Its long lifetime is used to retain partially folded proteins in the mitochondria and to direct folding toward forming the native disulfide bonds.

M

folding was compared in the presence of DsbA from the periplasm of E. coli, of PDI from the yeast ER, or of Mia40, from yeast mitochondria. In addition, we assessed the reactivity of Mia40 by determining the pK values and the reduction potentials of its catalytic cysteines, as well as of the mixeddisulfide intermediates between Mia40 and Cox17. The long lifetime of the mixed disulfides allowed us to purify these crucial reaction intermediates and to elucidate how interactions between Mia40 and the covalently bound substrate favor the mixed disulfide state. Energetic coupling stabilizes both the folded conformation of Mia40 and the mixed disulfide with its substrate as long as it is not properly folded.

any extracytoplasmic proteins contain disulfide bonds. These stabilizing covalent cross-links are established during the oxidative folding of proteins in the endoplasmic reticulum (ER) of eukaryotes or in the periplasm of prokaryotes. Their formation and the isomerization of incorrect disulfides are catalyzed by oxidoreductases that are well understood.1,2 In the ER, protein disulfide isomerase (PDI) mediates both formation and isomerization of disulfides. In the bacterial periplasm, DsbA serves as a thiol oxidase and DsbC as an isomerase for correcting non-native disulfide bonds.3,4 All these enzymes contain thioredoxin-like domains with a catalytic disulfide bond in a CXXC motif. It engages in a covalent mixed disulfide with a substrate protein and thus initiates thiol− disulfide exchange reactions. Ultimately, this results in the transfer of the disulfide bond from the thiol oxidase to the folding protein. More recently, disulfide-bonded proteins and a thiol oxidase, termed Mia40 (mitochondrial import and assembly), were discovered in the intermembrane space (IMS) of mitochondria.5−16 Mia40 is a thiol oxidase with unique properties. Its small catalytic domain does not belong to the thioredoxin family, and its active site disulfide is arranged not in a CXXC but a CPC motif. It is adjoined by a hydrophobic binding site,17,18 which is used to predispose substrate Cys residues with neighboring hydrophobic residues for the formation of the initial mixed disulfide.19 As in the case of DsbA and PDI, mixed disulfide formation is extremely fast,19 but in contrast to these thiol oxidases, the mixed disulfides formed by Mia40 are longlived both in vitro and in organello.5,20−22 To elucidate the catalytic mechanism of Mia40 at the molecular level and to understand the differences between Mia40 and the thioredoxin-like thiol oxidases, we used Cox17, a natural substrate of Mia40. The time course of its oxidative © 2014 American Chemical Society



RESULTS Mia40 Differs from DsbA and PDI in Forming LongLived Mixed Disulfides. We analyzed the catalysis of oxidative folding by using a variant of the mitochondrial copper-binding protein Cox17. As several other substrates of Mia40, it belongs to the group of CX9C proteins, and it has been used as a model protein by several groups that focused on protein import and oxidation in vivo and in isolated mitochondria.5,23,24 As the other CX9C proteins, Cox17 is composed of two short antiparallel helices that are linked by two structural disulfide bonds:25 Cys36−Cys47 (the “inner” disulfide bond) and Cys26−Cys57 (the “outer” disulfide bond) (Figure 1a). Cox17 from yeast contains a copper ion binding site formed by three additional cysteines (Cys16, 23, and 24) that lie outside the compact helical domain. As in our previous Received: May 23, 2014 Accepted: June 30, 2014 Published: July 1, 2014 2049

dx.doi.org/10.1021/cb500408n | ACS Chem. Biol. 2014, 9, 2049−2057

ACS Chemical Biology

Articles

Figure 2. Oxidation of Cox17* by various thiol disulfide oxidoreductases. Twenty micromolar reduced Cox17* was incubated with 50 μM (a) PDI, (b) DsbA, and (c) Mia40, respectively, in 100 mM potassium phosphate and 1 mM EDTA, pH 7.4, at 15 °C. At different time points, samples were taken, and the reaction was stopped by adding 60 mM iodoacetate. Samples were analyzed by nonreducing SDS-PAGE. The broken arrows in panels a and b indicate the expected position of the mixed disulfides between Cox17* and DsbA or PDI, which were not present in amounts comparable to the mixed disulfide in panel c.

Figure 1. Interaction of Cox17 with Mia40. (a) Structure of Cox17 from S. cerevisiae. The cysteines that form the structural disulfides (yellow) are labeled, the cysteines that were replaced by serines in the Cox17* variants are shown in red. The residues of the MISS/ITS sequence (F50, I51, Y54) are shown in purple; the residues of the ILFsequence (I37, L38, F39) are shown in blue. The structure was created using Protein Data Bank file 1Z2G25 and the program PyMol.52 (b) Schematic representation of the interaction between Mia40 and singlecysteine variants of Cox17*. They engage with Mia40 in a highly dynamic noncovalent complex, before the covalent mixed disulfide is formed.

unlike in the cases of DsbA and PDI, the mixed disulfides are long-lived, and only in the time range of an hour, it is converted to the native form. Incorrectly disulfide-bonded Cox17* (Figure 2b) could not be detected. Either, Mia40 avoids the formation of incorrect disulfides and/or it can also act as a disulfide isomerase. Together, these experiments confirm that PDI is an excellent and generic folding enzyme. DsbA is equally fast, but produces mainly incorrect disulfides. Mia40 differs from the two by forming long-lived mixed disulfides, which finally are transformed into the native form. The formation of long-lived mixed disulfides with substrates has been detected before, both in vitro and in organello.5,20−22 We find that this is an intrinsic property of Mia40 and not of the substrate, as Cox17* is efficiently oxidized by DsbA and PDI. Possibly, Mia40 forms long-lived mixed disulfides because it contains a CPC and not a CXXC active site motif as the thioredoxin-like thiol oxidases. Their redox properties are strongly influenced by the two intervening residues XX.29−31 To examine this, we replaced the Pro of Mia40 by Ala (CPC → CAC) or inserted two residues between the catalytic cysteines to create a CGHC motif as in PDI or a CPHC motif as in DsbA. All three variants share with the wild-type protein the common principle of forming long-lived mixed disulfides with Cox17*. In Mia40 CGHC, the covalent intermediate is even more stable than in Mia40, and for Mia40 CPHC, release of oxidized Cox17* could not be detected at all (Figure S1, Supporting Information). This rules out that the unusual CPC motif of Mia40 determines the long lifetime of the mixed disulfides. Active-Site Cysteines of Mia40 Are Only Moderately Reactive. The pK values of active-site cysteines are of utmost importance for catalytic activity because they determine the extent of deprotonation at a particular pH and thus the reactivity in thiol−disulfide exchange processes. We measured the pK values of the active-site cysteines of Mia40 with the aromatic disulfide DTNB, which reacts with deprotonated

work, we replaced these cysteines by serines in our pseudo wild-type protein (termed Cox17*), to avoid interference with disulfide bond formation in the CX9C domain. The solution structures of both human and yeast holo and apo Cox17 show that the overall helix−loop−helix structure is virtually not affected by copper removal.26,27 PDI from the yeast ER reacted very rapidly with Cox17* (Figure 2a). In SDS-PAGE, the band of the reduced form vanished within the dead time of manual mixing (5 s). A band for the mixed disulfide could not be detected, and the formation of native, fully oxidized Cox17* was complete within about 30 s. PDI is thus an excellent generic folding enzyme that oxidizes also a mitochondrial protein without forming long-lived reaction intermediates. DsbA from the periplasm of E. coli also reacted extremely fast with Cox17* without accumulating mixed disulfides, and the reaction was also complete in less than 30 s (Figure 2b). In this case, however, only about one-third of the product migrated at the position of native, correctly oxidized Cox17*. The remainder migrated at the position of incorrectly disulfidebonded protein (scCox17*), as produced by oxidizing Cox17* under unfolding conditions. Four cysteines (as in Cox17*) can form two disulfides in three different arrangements, in the native and in two incorrect pairings. The results in Figure 2b indicate that DsbA introduces disulfide bonds into Cox17* almost at random. DsbA does not show disulfide isomerase activity3,28 and therefore cannot resolve wrong disulfides. Incorrectly oxidized Cox17* was also detectable within the first seconds of oxidative folding by PDI (Figure 2a), but rapidly isomerized to native Cox17* by the efficient disulfide isomerase activity of PDI. Mia40 from the yeast IMS forms mixed disulfides with Cox17* also in a very fast reaction (Figure 2c). However, 2050

dx.doi.org/10.1021/cb500408n | ACS Chem. Biol. 2014, 9, 2049−2057

ACS Chemical Biology

Articles

obtained for the single-Cys variants. The maximal reaction rates at high pH, where the cysteines are fully ionized, are also different (Figure 3). For the single-Cys variants rates between 1 and 2 s−1 were observed, whereas in the wild-type protein the first and the second cysteine show rates of 10 and 4 s−1, respectively. The apparent pK values and the reaction rates of the activesite cysteines of Mia40 thus depend on their environment, but unlike DsbA and PDI, Mia40 does not contain one cysteine with an unusually low pK and another one with an unusually high pK. Apparently, Cys296 shows the lowest pK and thus the best leaving group quality. Therefore, substrate cysteines should preferentially attack Cys298 to form the mixed disulfide with Mia40, which was also concluded from import experiments into mitochondria.32 Binding of the Substrate Has Only Minor Effects on the pK of Mia40. The extended lifetime of the mixed disulfide between Mia40 and its substrate proteins provides the opportunity to determine the reactivity of the remaining free active-site cysteine of Mia40. We mixed 5 μM oxidized Mia40 with 5 μM of a variant of Cox17* that contains only Cys36 (Cox17* C36). It rapidly forms the noncovalent complex, and the mixed disulfide with Mia40 accumulates within 100 ms (Figure 1b). The reaction with DTNB was then used to determine the titration curves of the free cysteines in the samples. Two modification reactions were observed: a slow reaction with a pK of 7.8 and a maximal rate of 0.12 s−1 and a fast reaction with a pK of 8.2 and a maximal rate of 2.1 s−1 (Figure 3c). These reactions originate from either the remaining free active-site Cys of Mia40 or from Cys36 of Cox17* C36, free or in the noncovalent complex (Figure 1b). In the absence of Mia40 or in the presence of a Mia40 variant without the active-site cysteines, Cox17* C36 shows a reaction profile with a pK of 8.2 (Figure S3a,b, Supporting Information) and a maximal rate of 2.4 s−1 (Table 1). The reaction profile with an apparent pK of 8.2 could thus be assigned to Cys36 of Cox17* in free form or noncovalently bound to Mia40. The other profile with a pK of 7.8 and a maximal rate of 0.12 s−1 thus reflects the remaining active-site Cys of Mia40 in the mixed disulfide with its substrate. We also produced the other three single-Cys variants of Cox17* and followed their reactions with DTNB in the presence or absence of Mia40 (Figure S3, Supporting Information). In the mixed disulfide with Cys47, the remaining free catalytic Cys of Mia40 showed virtually the same pK and the same reaction rate as in the mixed disulfide with Cys36 (Table 1). In the case of Cys57 of Cox17*, the control experiments indicate that now the fast phase corresponds to the reaction of the cysteine in Mia40 and the slow phase to Cys57 in Cox17*. The latter pK was decreased to 7.3, and the modification rate was increased about 50-fold to 9.8 s−1. Cys26 of Cox17* does not readily form a covalent mixed disulfide, presumably because it resides in an unfavorable position in the noncovalent complex with Mia40.19 Accordingly, Mia40 retains its catalytic disulfide, and only the modification reaction of Cys26 of Cox17* was observed, which coincided with the reaction measured for Cox17* C26 in the absence of Mia40 (Table 1). The apparent pK values of 7.3−7.8 for the free catalytic Cys of Mia40 in the various covalent mixed disulfides resemble the pK values of 7.0−7.9 obtained for Mia40 alone (Table 1), indicating that mixed-disulfide formation has only a minor effect on the pK and thus on the reactivity of the remaining free

cysteines in a thiol−disulfide exchange reaction. Its rate, measured as a function of pH, follows the titration curve of the respective cysteine and thus yields its apparent pK value. Single-Cys variants of Mia40 were produced by replacing either the first (Cys296) or the second (Cys298) Cys in the active site by Ser. Then the rates of their reactions with DTNB were measured (Figure S2a, Supporting Information) and plotted as a function of pH (Figure 3a). The fitting of titration

Figure 3. pK values of Mia40 and Cox17*. The reaction of 50 μM DTNB and (a) 5 μM Mia40 C298S (open circle) or Mia40 C296S (filled circle), (b) 5 μM reduced Mia40, and (c) a mixture of 5 μM Mia40 and Cox17* 36C, respectively, was measured by the change in absorbance at 412 nm. The open and filled symbols in panels b and c represent the rate constants of the two independent reaction phases. Exponential functions were fitted to the kinetics (cf. Figure S2, Supporting Information) and the apparent rate constants are plotted as a function of pH. The solid lines represent fits of the Henderson− Hasselbalch equation (eq 1) to the data; the results are given in Table 1. All measurements were performed at 15 °C in 100 mM potassium acetate, potassium phosphate, Tris-HCl, or glycine with appropriate pH values.

curves (eq 1) to the rates (kapp) gave apparent pK values of 6.3 for Cys296 and 7.1 for Cys298 (Table 1). These values are two and one units lower than the intrinsic pK of Cys, which is near 8.3. In wild-type Mia40, the two catalytic cysteines are in close proximity and presumably influence each other in their deprotonation behavior. We therefore determined the pK values also for the reduced wild-type protein with both activesite cysteines. Cys296 and Cys298 differ in the rate of their reaction with DTNB (Figure S2b, Supporting Information), and the increases of the two corresponding rate constants with pH could be measured individually. They followed the deprotonation of the two cysteines. The resulting titration curves (Figure 3b) gave apparent pK values of 7.0 and 7.9 (Table 1), which are about 0.7 pH units higher than those 2051

dx.doi.org/10.1021/cb500408n | ACS Chem. Biol. 2014, 9, 2049−2057

ACS Chemical Biology

Articles

Table 1. pK Values of Cysteines in Mia40 and Cox17*a protein Mia40 C298S Mia40 C296S Mia40 reduced Mia40 + Cox17* 26C Mia40 + Cox17* 36C Mia40 + Cox17* 47C Mia40 + Cox17* 57C Cox17* 26C Cox17* 36C Cox17* 47C Cox17* 57C Mia40 C296SC298S + Mia40 C296SC298S + Mia40 C296SC298S + Mia40 C296SC298S +

Cox17* Cox17* Cox17* Cox17*

26C 36C 47C 57C

pK1

kapp, S1− (s−1)

± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

1.13 ± 0.02 1.84 ± 0.07 9.9 ± 0.3 1.12 ± 0.03 2.09 ± 0.03 2.66 ± 0.09 9.8 ± 0.5 1.50 ± 0.03 2.39 ± 0.04 2.37 ± 0.06 4.00 ± 0.06 1.40 ± 0.03 2.63 ± 0.06 2.47 ± 0.05 3.62 ± 0.06

6.28 7.14 7.00 8.35 8.24 8.08 7.30 8.49 8.22 8.21 8.11 8.41 8.26 8.30 8.20

0.04 0.07 0.07 0.05 0.03 0.07 0.10 0.03 0.03 0.05 0.03 0.08 0.08 0.04 0.03

kapp, S2− (s−1)

pK2

7.9 ± 0.2

3.8 ± 0.5

7.81 ± 0.08 7.8 ± 0.2 8.0 ± 0.1

0.12 ± 0.01 0.11 ± 0.01 2.7 ± 0.2

a pK values were determined as described in Figure 3. The apparent rate constants of the reaction of the fully deprotonated cysteines (kapp, S1−) were derived from the final value of the data fit. The standard deviation is given; all data were reproduced in independent experiments.

Cys of Mia40. It resembles normal, exposed cysteines in its ionization properties and is not highly reactive at physiological pH. Therefore, it attacks the mixed disulfide with the substrate only slowly or not at all. Because this back reaction would lead to release of the substrate, this might contribute to the unusually long lifetime of the covalent reaction intermediate. Corresponding data are not available for the other thiol oxidases because their mixed disulfides are unstable. Conformational Stability of Mia40 Increases upon Mixed Disulfide Formation. The conformational stability of a protein and the stability of its disulfide bonds are linked. Usually, disulfide bonds stabilize the folded conformation, and reciprocally, the folded conformation stabilizes the disulfide bonds. DsbA is a unique exception because it gains in conformational stability upon reduction of its catalytic disulfide bond.33,34 This explains why DsbA is such a powerful oxidase. Mia40 shows high thermal stability. In the oxidized form, it remains folded up to about 70 °C, and the unfolding transition does not go to completion in the accessible temperature range (Figure 4a). We estimated a transition midpoint TM of about 88 °C (Table 2) from the maximum of the first derivative of the transition curve (Figure S4, Supporting Information). Thermal

The stability against thermal denaturation was measured by following the CD at 222 nm as shown in Figure 4. TM values were estimated from the maxima of the first derivatives, and the uncertainty range depends on the width of the peak. The reduction potentials were determined as described in Figure 5. The mean and the standard error of at least 3 measurements are given. The reduction potentials of oxidized Mia40 and the mixed disulfide between Mia40 and glutathione could only be estimated. bValue was taken from ref 17.

Figure 4. Thermal unfolding curves of Mia40 and the mixed disulfides between Mia40 and Cox17*. Thermal denaturation of (a) 1 μM Mia40 (black) and Mia40 C296S C298S (red) and (b) 1 μM Mia40 (black) and the mixed disulfides between Mia40 and Cox17* 36C (green), Cox17* 47C (blue), and Cox17* 57 C (orange), respectively, was followed by the change in ellipticity at 222 nm. All measurements were performed in 100 mM potassium phosphate, pH 7.4. The transition midpoints as derived from the first derivative, as shown in Figure S5, Supporting Information, are given in Table 2. Unfolding was irreversible after heating of the proteins to 100 °C.

unfolding of the reduced form could not be followed because the free Cys residues become too reactive at high temperature. Instead, Mia40 C296S C298S was employed as a model for the reduced form. Its TM value is near 83 °C (Figure 4a) and thus about 5 °C lower than the TM of the oxidized wild-type protein (Table 2). This suggests that the catalytic disulfide stabilizes Mia40, in a similar fashion as structural disulfides. Thermal unfolding was also followed for Mia40 in the mixed disulfide intermediates with the cysteines 36, 47, or 57 of Cox17*. They were produced by a 2 min coincubation of the two reactants, subsequent blocking of the remaining free Cys of Mia40 with iodo acetate, and purification by size exclusion chromatography. The one-Cys Cox17* variants are unfolded.19 Therefore, they do not contribute to the change in ellipticity, and the measured transitions can be assigned to Mia40 alone. For all three mixed disulfides, a strong shift of the unfolding transitions toward higher temperature compared to Mia40 was observed (Figure 4b). For the mixed disulfides with Cys47 or

Table 2. Conformational Stability and Reduction Potentials of Mia40 and the Mixed Disulfides with Cox17*a protein

TM (°C)

Mia40 oxidized

88 ± 3

Mia40 reduced mixed disulfide between Mia40 and: glutathione Cox17* 36C Cox17* 36C 37A38A39A Cox17* 47C Cox17* 57C Cox17* 57C 37A38A39A Cox17* 57C 50A51A

83 ± 2

96 ± 1 89 ± 1 >102 >102 100 ± 2 92 ± 1

reduction potential (mV) ∼−160 −200 ± 5b

∼−210 −243 ± −182 ± −250 ± −277 ± −261 ± −225 ±

6 2 6 2 8 7

a

2052

dx.doi.org/10.1021/cb500408n | ACS Chem. Biol. 2014, 9, 2049−2057

ACS Chemical Biology

Articles

Cys57, the apparent TM value increased to above 100 °C, and for Cys36 to about 96 °C. Native Mia40 contains a hydrophobic site adjacent to its catalytic disulfide. Hence, the folded form of Mia40 should become stabilized upon the binding of nonpolar residues of a substrate protein to this site. Cys36 of Cox17 is followed by Ile37, Leu38, and Phe39, and after the substitution of these three hydrophobic residues by Ala, the TM of the corresponding mixed disulfide indeed dropped from 96 to 89 °C (Table 2). This provides strong evidence that hydrophobic interactions between folded Mia40 and nonpolar residues in its substrate stabilize the mixed disulfide intermediate. The substitution of the same residues in a Cox17* variant that used the remote Cys57 for forming the mixed disulfide had only a minor effect on the stability of Mia40 (Table 2). When Phe50 and Ile51 of the MISS/ITS24,35 sequence near Cys57 were replaced by Ala, the TM of the mixed disulfide with Cys57 dropped from above 100 to 92 °C (Table 2). This confirms that hydrophobic residues near a reacting Cys are indeed important for its interaction with Mia40. Energetic coupling thus exists between the noncovalent hydrophobic interactions of Mia40 with its substrate and the conformational stability of Mia40 in the mixed disulfide intermediate. This coupling seems to be strong for the mixed disulfides with Cys47 or Cys57 of Cox17*, which might explain why these mixed disulfides are particularly long-lived and frequently isolated after the import of Cox17 into isolated mitochondria. Conformational Stability of the Mixed Disulfide Correlates with Reduction Potential. The biochemical standard reduction potential (E0′) of a compound is a measure of the affinity for electrons of its oxidized form at pH 7.0. The higher the oxidizing power, the higher (less negative) the reduction potential. Accessible disulfide bonds often show E0′ values near −200 mV.36−39 The E0′ of DsbA is near −120 mV, which explains its high oxidizing power.33,40,41 The E0′ of glutathione is well-known, and the E0′ values of other dithiols are usually measured by incubating them with mixtures of GSH and GSSG at different concentration ratios and by determining the fractions of the oxidized and reduced forms. Previously, the increase in Trp fluorescence upon reduction has been employed to calculate a E0′ value of −200 mV for Mia40.17 Unfortunately, mixed disulfide formation between Mia40 and GSH leads to an increase in fluorescence as well,19 and therefore, this E0′ must be regarded as an apparent value. Oxidized and reduced Mia40 and the mixed disulfide with GSH migrate slightly different in native PAGE. We equilibrated Mia40 with 0.1 mM GSSG and 0.08−33 mM GSH, stopped thiol−disulfide exchange by adding iodo acetate, and analyzed the samples by native PAGE. In fact, Mia40 formed a stable mixed disulfide with GSH around 1 mM GSH (Figure 5a). A single E0′ value can therefore not be given for Mia40. As the bands on the gel were not well separated, only a semiquantitative analysis was possible. It gives E0′ estimates of −160 and −210 mV for the formation of the mixed disulfide with GSH and for its resolution by GSH, respectively; −210 mV is in the range of the formerly determined value, indicating that the change in Trp fluorescence in Mia40 mainly reflected the resolution of the mixed disulfide with GSH. The reduction potentials of the mixed disulfides between Mia40 and the cysteines 36, 47, or 57 of Cox17* were also determined (Figure 5b−d). All of them were more stable than the mixed disulfide between Mia40 and GSH. The extent of

Figure 5. Reduction potential of Mia40 and the mixed disulfides between Mia40 and Cox17*. (a) Ten micromolar Mia40 or 10 μM of the mixed disulfides between Mia40 and (b) Cox17* 36C, (c) Cox17* 47C, and (d) Cox17* 57C, respectively, were incubated with 0.1 mM GSSG and the indicated GSH concentrations for 6 h. The reaction was stopped by the addition of 200 mM iodoacetate, and the samples were analyzed by native PAGE (a) or SDS-PAGE (b−d). All measurements were performed in 100 mM potassium phosphate, pH 7.4, at 15 °C. # indicates the band of the mixed disulfide between Mia40 and GSH. (e) The intensities of the bands of the mixed disulfides between Mia40 and Cox17* 36C (circle), Cox17* 47C (triangle), and Cox17* 57C (square) are plotted as a function of the ratio of GSH2 and GSSG, and the reduction potential was determined with the Nernst equation (eqs 2 and 3). (f) Correlation between conformational stability and stability of disulfides in Mia40 and Cox17*. The reduction potentials are plotted against the TM values for Mia40 (black filled circle) and the mixed disulfides between Mia40 and Cox17* 36C (black open circle), Cox17* 36C 37A38A39A (blue oepn circle), Cox17* 47C (black open triangle), Cox17* 57C (black open square), Cox17* 57C 37A38A39A (blue open square), and Cox17* 57C 50A51A (purple open square). Values were taken from Table 2. The broken line indicates the linear dependency.

reduction was followed by the decrease of the intensities of the bands for the mixed disulfides in SDS-PAGE (Figure 5b−d) and plotted as a function of the ratio [GSH]2/[GSSG] (Figure 5e). Table 2 shows the E0′ values obtained from the analysis according to eqs 2 and 3. The mixed disulfides with Cys36 and Cys47 show similar E0′ values of −243 and −250 mV, respectively. The mixed disulfide with Cys57 shows the lowest E0′ value (−277 mV) and thus the highest stability toward reduction. It remained partially intact even at the highest accessible GSH concentration. The unfolding experiments had indicated that the thermal stability of Mia40 in the mixed disulfides with the different cysteines of Cox17* is modulated by the noncovalent hydrophobic interactions between Mia40 and nonpolar residues near the reacting cysteine. This holds true for the stability toward reduction by GSH as well. The substitutions of Ile37, Leu38, and Phe39 adjacent to Cys36 by Ala strongly reduced the stability of the mixed disulfide toward reduction (Figure S5, Supporting Information), and correspondingly, E0′ increased from −243 to −182 mV (Table 2). In a similar 2053

dx.doi.org/10.1021/cb500408n | ACS Chem. Biol. 2014, 9, 2049−2057

ACS Chemical Biology

Articles

fashion, the E0′ value of the mixed disulfide with Cys57 was increased from −277 to −225 mV (Figure S5, Supporting Information) when the MISS/ITS residues Phe50 and Ile51 were replaced by Ala (Table 2). This is equivalent to a 50-fold increase in the equilibrium constant of reduction and emphasizes the importance of hydrophobic residues for the interaction between Mia40 and Cox17*. Hydrophobic interactions are weakened by denaturants. The stability of the mixed disulfide between Mia40 and Cys36 toward reduction is in fact strongly reduced in the presence of 3.0 M urea. Its reduction profile (Figure S5h, Supporting Information) is shifted to lower GSH concentrations, toward the curve obtained for Cox17* 36C 37A38A39A, where the hydrophobic interactions were weakened by substitutions. As expected, the profile for the latter variant was hardly affected by 3.0 M urea. The curves measured in the presence of 3.0 M urea could not be analyzed further because they deviated from the Nernst equation at low GSH concentrations. The results from the unfolding experiments and from the determination of the reduction potentials are correlated (Figure 5f). They reveal the energetic coupling between the conformational stability of Mia40 in the mixed disulfides with the Cox17* variants and the stability of the intermolecular disulfide bond. The stabilization is caused by favorable hydrophobic interactions between the enzyme and the substrate, and this explains why mixed disulfides between Mia40 and its substrates are exceptionally long-lived.

cysteines in fact show heavily shifted pK values. The pK of the first catalytic Cys (Cys30) of DsbA is depressed to 3.5, and the pK of the second Cys (Cys33) lies above 10.43 The active disulfide bond is thus strongly polarized. PDI shows corresponding pK values of 4.5 and