Disulfide Formation in the ER and Mitochondria: Two Solutions to a Common Process

See allHide authors and affiliations

Science  05 Jun 2009:
Vol. 324, Issue 5932, pp. 1284-1287
DOI: 10.1126/science.1170653


The endoplasmic reticulum (ER) was long considered to be the only compartment of the eukaryotic cell in which protein folding is accompanied by enzyme-catalyzed disulfide bond formation. However, it has recently become evident that cells harbor a second oxidizing compartment, the mitochondrial intermembrane space, where disulfide formation facilitates protein translocation from the cytosol. Moreover, protein oxidation has been implicated in many mitochondria-associated processes central for human health such as apoptosis, aging, and regulation of the respiratory chain. Whereas the machineries of ER and mitochondria both form disulfides between cysteine residues, they do not share evolutionary origins and exhibit distinct mechanistic properties. Here, we summarize the current knowledge of these oxidation systems and discuss their functional similarities and differences.

The production of functional proteins in the cell involves at least two steps: the synthesis of nascent polypeptides on cytosolic ribosomes, followed by the folding of these polypeptides into their functional conformation. Both processes can occur simultaneously, and ribosomes specifically recruit a variety of dedicated chaperone systems to their polypeptide exit tunnels to facilitate efficient folding of nascent chains. In the case of oxidative protein folding, synthesis and folding generally occur in temporally and spatially distinct reactions (Fig. 1A). Protein synthesis occurs in the cytosol where a very high concentration (10 to 14 mM) of reduced glutathione together with numerous reducing enzymes counteracts the formation of disulfide bonds (1). Thus, oxidative folding normally takes place only after translocation to a different compartment in which disulfide formation is favored. This spatial separation of synthesis and folding contributes to compartmentalization of the cell, ensuring that newly synthesized proteins become active only after successful delivery to their correct destination. Moreover, disulfide bonds provide increased stability, which is especially important for secreted and cell-surface proteins because they are not under the surveillance of the cellular chaperone machinery. In addition, redox regulation by the reversible formation of disulfide bonds can be used to adapt protein activity directly to the redox conditions in a given environment. In most cases, disulfide formation is an enzymatically catalyzed process that occurs in the periplasm of prokaryotes and in the ER and mitochondrial intermembrane space (IMS) of eukaryotes (Fig. 1B). We will focus specifically on the two eukaryotic oxidation machineries and compare their structural and mechanistic properties.

Fig. 1

Protein synthesis and oxidation occur in distinct cellular compartments. (A) In vivo, the sites of disulfide bond formation and protein synthesis are separated by a membrane. (B) Proteins are maintained in a reduced state in compartments in which protein synthesis takes place. These locations are separated by membranes from compartments harboring machineries for protein oxidation such as the bacterial periplasm, the ER, or the IMS of mitochondria. This spatial separation of protein synthesis and oxidation contributes decisively to the compartmental identity of cells. (C) Protein oxidation generally follows the same principles. First, proteins are oxidized by oxidoreductases that merely shuttle disulfide bonds that are initially generated by sulfhydryl oxidases. The oxidizing power for the entire process is provided by oxygen.

Common Principles of Disulfide Bond Formation

The formation of disulfides in substrate proteins is a reversible reaction in which the thiol groups of two cysteine residues are oxidized to form a covalently linked disulfide. This can be achieved by thiol-disulfide exchange in which a disulfide bond present in an electron acceptor is reduced. Alternatively, the transfer of electrons to molecular oxygen leads to de novo disulfide bond formation (Fig. 1C). Despite the thermodynamically favorable nature of the oxygen-dependent protein oxidation, this reaction typically requires catalysts such as transition metals or the redox cofactor flavin adenine dinucleotide (FAD) to overcome the kinetic barrier caused by the incompatibility of one- and two-electron transfer processes.

The Mitochondrial Disulfide Relay

Recently, it was discovered that proteins with structural disulfides are commonly present in the IMS of mitochondria (2). Since then, the critical players in mitochondrial protein oxidation have been identified and their functions characterized (27). The mitochondrial disulfide relay system (Fig. 2A) facilitates disulfide bond formation in substrate proteins and plays a crucial role in the transport of newly synthesized IMS proteins across the outer mitochondrial membrane. Two proteins are of central importance for this process: Mia40 and Erv1. Mia40 is a conserved IMS protein, which is soluble in mammals and plants and membrane-anchored in fungi (5, 7). Mia40 contains an essential redox-active disulfide bond in a cysteine-proline-cysteine signature that oxidizes cysteine residues of incoming polypeptide chains (8). The Mia40-dependent oxidation of incoming proteins locks them in a stably folded state in which they are unable to traverse the outer membrane, thereby leading to a directed transport of Mia40 substrates into the mitochondrial IMS (9). Mia40 is reoxidized by Erv1, which belongs to a family of FAD-containing sulfhydryl oxidases (2). Members of this family are present in the secretory pathway (Erv2, quiescin-sulfhydryl oxidase) or expressed by certain viruses (e.g., E10R) (10). Erv1 contains two essential redox-active cysteine-x-x-cysteine pairs (where x indicates any amino acid), which shuttle electrons from Mia40 to FAD. In vitro, Erv1 can be directly reoxidized by oxygen in a reaction that yields hydrogen peroxide (H2O2). However, in vivo Erv1 is oxidized by cytochrome c, which in turn passes its electrons via cytochrome c oxidase to molecular oxygen to produce water (11, 12). The connection to the respiratory chain increases the efficiency by which Erv1 is reoxidized, and also prevents the generation of H2O2 in the IMS. The small metal-binding protein Hot13 is an additional factor of the mitochondrial oxidation machinery. It improves the electron transfer between Mia40 and Erv1 by maintaining Mia40 in a zinc-free state that can be efficiently oxidized by Erv1 (3).

Fig. 2

The disulfide relay machineries of the IMS and the ER. (A) After their synthesis in the cytosol, IMS proteins are translocated across the outer membrane of mitochondria in an unfolded, reduced conformation. Oxidized Mia40 interacts with these proteins and facilitates their stable folding by the introduction of disulfide bonds. A cascade of redox-active proteins transfers electrons from Mia40 to cytochrome c oxidase, which converts oxygen to water. The flavoprotein Erv1 mediates the switch from two-electron to one-electron transfer. (B) Already during their translocation into the ER lumen, proteins are oxidized by PDIs. The PDIs are maintained in an oxidized state by the flavoprotein Ero1, which transfers electrons directly onto oxygen, thereby generating H2O2. Thus, protein oxidation in the ER and in mitochondria relies on distinct components, but the biochemical principles of the transfer cascades are similar.

The Oxidation Machinery of the ER

Disulfide bond formation in the ER typically occurs concomitantly with protein translocation, albeit without any functional coupling of both processes. In a first thiol-disulfide exchange reaction, electrons are shuttled from reduced substrate proteins to oxidized members of the protein disulfide isomerase family (PDIs; Fig. 2B). This results in the reduction of conserved cysteine-x-x-cysteine motifs in these PDIs. Subsequently, the PDIs are reoxidized by the flavoenzyme Ero1, which transfers electrons via FAD onto oxygen (1315). To allow proteins to fold to their native states, disulfides sometimes have to be reshuffled by isomerization or repeatedly broken and re-formed by cycles of reduction and oxidation. Both reactions are efficiently catalyzed by members of the PDI family. Reduced glutathione, which is transported from the cytosol to the ER lumen by an unknown mechanism, serves as reductant in this PDI-mediated folding reaction (16). The ER contains several PDIs, which differ considerably in size, redox potential, and substrate spectrum. Structurally, PDIs are characterized by one or several a- or b-type thioredoxin-like domains. Thioredoxin folds of the a-type are redox-active and comprise cysteine-x-x-cysteine motifs. The two residues that separate the cysteines substantially modulate the redox potential and, hence, influence the physiological function of the particular PDI protein (17, 18). Noncatalytic b-type thioredoxin-like domains lack redox-active cysteine residues and presumably participate in peptide-binding and/or chaperone-like activities (19). Whereas PDIs shuffle disulfide bonds, Ero1 generates disulfide bonds de novo by the transfer of electrons from protein substrates to its FAD cofactor. This is achieved by a stepwise intracellular transfer reaction: a redox-active cysteine-x4-cysteine motif in a flexible loop in Ero1 receives electrons from PDIs and passes them on to a redox-active cysteine-x-x-cysteine pair in the catalytic core of the protein, which consists of a four-helix bundle structure surrounding the FAD cofactor (20). Finally, electrons are shuttled via the redox-active isoalloxazine moiety of FAD to molecular oxygen, giving rise to the production of H2O2 (15, 20).

Mechanistic Differences in the Eukaryotic Oxidation Machineries

A much greater diversity of substrates are oxidized by the ER machinery than by the mitochondrial machinery. In eukaryotic cells, thousands of different proteins are transported into the ER and along the secretory pathway (21). These proteins adopt different folds and can form complex oligomers often linked by intra- and intermolecular disulfides. In contrast, the mitochondrial IMS contains a rather small number of proteins. Moreover, most substrates of the mitochondrial disulfide relay system are of low molecular mass (7 to 15 kD) and show one type of fold: helix-turn-helix conformations in which the helices are connected by two parallel disulfide bonds. Depending on whether the cysteine pairs are separated by three or nine amino acid residues, these specific arrangements are named “twin Cx3C” or “twin Cx9C” motifs. Proteins with twin Cx3C motifs constitute the group of small Tim proteins, which represent soluble chaperones that usher hydrophobic inner-membrane proteins across the IMS (22). Most proteins with twin Cx9C motifs are involved in the biogenesis of respiratory chain complexes and are particularly important for the incorporation of copper into cytochrome c oxidase (23). In addition, the mitochondrial disulfide relay system is, directly or indirectly, implicated in the import of IMS proteins with other cysteine arrangements, such as Erv1, copper-zinc superoxide dismutase (Sod1), and copper chaperone for Sod1 (24, 25). However, the precise role of Mia40 and Erv1 in the import and folding of these proteins is not known.

Multiple PDIs—One Mia40. Presumably as a consequence of the large diversity of substrate proteins, the ER contains numerous PDIs that ensure the oxidative folding of secreted proteins by performing oxidation, isomerization, and reduction reactions, and all contain one or more thioredoxin-like domains (Fig. 3A). However, except for certain PDIs such as ERp57 (26), the respective subset of endogenous substrates remains unknown. In contrast, Mia40 is the only protein known that directly oxidizes substrates in the IMS. However, Mia40 has a completely different structure than PDIs (Fig. 3A) and so far, it is not clear whether it has chaperone and/or isomerase activity.

Fig. 3

The key components for eukaryotic protein oxidation. (A) Structures of the oxidoreductases Mia40 (48) and PDI (49). Presumably due to their interaction with a variety of reduced and unfolded substrates, both proteins consist of two modules to exhibit a redox function (green) mediated by a CPC (Mia40) or a CGHC (PDI) motif and a chaperone/peptide-binding function (gray). Although the chaperone function has been shown for the b/b′ thioredoxin-like domains of PDI, it has only been proposed for the helix-turn-helix motif of Mia40. (B) Structures of the sulfhydryl oxidases Erv1 (27) and Ero1 (20). Both proteins facilitate the de novo generation of disulfide bonds via a four-helix-bundle structure (dark blue, helices 1 to 4) positioning a FAD cofactor (yellow) in proximity to a cysteine pair (orange). Both proteins show an almost identical architecture, although the primary sequence and the order of the helices are completely different. For example, in Ero1 additional intervening helices (light blue) loop out from the helix bundle and do not contribute to the formation of the FAD-binding site. The Protein Database accession numbers are presented in white next to the structures.

Ero1 and Erv1 are unrelated twins. De novo disulfide bond formation in the IMS and the ER is driven by the sulfhydryl oxidases Erv1 and Ero1, respectively. Although evolutionarily unrelated, both proteins contain the same core structure: a four-helix bundle that coordinates the FAD cofactor and positions it adjacent to a pair of cysteines (Fig. 3B) (20, 27). These cysteines directly interact with FAD as well as with a second cysteine pair in a flexible region of the protein that swings in and out of the core to interact with either PDIs (in the case of Ero1) or Mia40 (in the case of Erv1). Given the different amino acid sequences of the proteins, the similarity of their three-dimensional structures and mechanism is surprising and a beautiful example of convergent evolution.

Spatial organization of disulfide bond formation. It has been proposed that disulfide bond formation in the ER as well as in the IMS proceed via a relay system, i.e., the oxidoreductases (Mia40 or PDIs) oscillate between substrates and their respective sulfhydryl oxidases. However, recently, a ternary complex between substrate, Mia40, and Erv1 has been observed (28). It was proposed that this substrate-Mia40-Erv1 complex persists during substrate oxidation. Whether such a tight spatial organization of the oxidation pathway in the ER exists is unclear.

Mitochondria avoid the production of H2O2. Ero1 transfers its electrons directly to molecular oxygen, giving rise to the production of H2O2 (15). To avoid damage to cellular components, this reactive compound has to be detoxified in close proximity to its site of generation. How this detoxification is facilitated is unclear. In the IMS, the production of H2O2 is avoided by coupling Erv1 directly to the respiratory chain, giving rise to the production of water. This coupling not only prevents the release of reactive oxygen species (ROS) but also increases the efficiency of Erv1 reoxidation (11, 12).

The ER is more “oxidizing.” The IMS is connected to the cytosol by porins in the outer membrane of mitochondria that allow the diffusion of small ions such as glutathione. Consequently, the IMS has a higher ratio of reduced to oxidized glutathione than the ER lumen (29, 30). Nevertheless, the IMS is clearly more “oxidizing” than the cytosol or the mitochondrial matrix. Whether the higher concentration of oxidized glutathione in the IMS is due to the activity of Erv1 or to the production of ROS in this compartment remains to be clarified. However, the chemical environment does not necessarily dictate the redox state of proteins as the introduction of disulfide bonds is often enzymatically controlled. Hence, as has been shown for the ER, reduction and oxidation processes can occur in parallel in the same compartment.

The oxidation machinery in the ER is redox-regulated. Because the action of the oxidative machinery in the ER yields one molecule of H2O2 per de novo–generated disulfide bond, it has to be tightly controlled to counteract hyperoxidizing conditions and reactive oxygen damage. Recently, an elegant feedback regulation control system was identified that keeps a check on Ero1 activity (3133): Ero1 contains several noncatalytic cysteine residues that form intramolecular disulfide bonds under hyperoxidizing conditions, which attenuate its oxidase activity.

Protein oxidation as regulator of protein function. The adaptation of Ero1 activity to the respective redox conditions in the ER is a wonderful example of redox control by disulfide formation. In the ER, several processes are regulated by the reversible formation of disulfide bonds. One example is the calcium homeostasis of the ER lumen where the uptake of calcium by the sarco(endo)plasmic reticulum calcium (SERCA) pump as well as the release by the inositol 1,4,5-trisphosphate (IP3) receptor are controlled by redox regulation of cysteine residues (34, 35). A redox control of Erv1 comparable to that of Ero1 has not been observed so far. However, the activity of Erv1 depends on the cellular oxygen concentration through its coupling to the respiratory chain. Thus, it appears conceivable that Erv1 could be involved in the adaptation of mitochondrial activities to the local oxygen concentration by introducing regulatory disulfide bonds (36). Although experimental evidence for a regulatory function of Erv1 in vivo is still missing, it is striking that many IMS proteins contain conserved cysteine residues in motifs deviating from the ones found in typical import substrates (Fig. 4). Potentially, processes such as the import of proteins into mitochondria, the biogenesis of the respiratory chain, respiratory activity, and apoptosis could be modulated by thiol-dependent redox processes, and it will be exciting to identify the relevance of these cysteine residues.

Fig. 4

IMS proteins with conserved cysteines. Numerous IMS proteins contain cysteine residues (indicated as dots) that are conserved among metazoa (p66shc, Pink1, Smac) or even among metazoa and fungi (all other proteins). Some of these cysteine residues coordinate cofactors such as heme groups or metal ions and therefore are maintained in a reduced state (yellow dots). In many cases, the presence of disulfide bonds between these residues has been proved (red dots). Further studies are needed to address the physiological relevance of thiol groups in the IMS, especially of those not characterized so far (orange dots).

Protein Oxidation in Health and Disease

The oxidative folding of proteins in the ER is vital for cellular functionality. Dysfunctions in protein oxidation contribute to a variety of diseases (3739), and the ER redox machinery plays a crucial role for the infection process of many viruses [for reviews, see, e.g., (40)]. The physiological relevance of mitochondrial protein oxidation is less understood. Mitochondria are a major source of cellular ROS production, particularly under pathological conditions. The oxidative stress caused by mitochondrial dysfunctions is intimately linked to the pathology of many neurodegenerative diseases such as Alzheimer’s, Parkinson’s, or Huntington’s disease (4143). However, it remains difficult to determine whether oxidative stress leads to, or is a consequence of, neurodegenerative processes.

Even under physiological conditions, the accumulating oxidative damage of mitochondrially produced ROS presumably contributes directly to the aging process of animals (44). ROS levels certainly influence the redox states of protein thiols, but it is not clear whether the aging effect of ROS is due to the increased formation of disulfide bonds in proteins or is caused by more pleiotropic oxidative damage. Recently, it was shown that the mitochondrial redox relay is critical for the translocation of Sod1 into mitochondria (2, 25). Because this enzyme counteracts mitochondrial ROS, Erv1 activity might be directly linked to the protection from ROS. Mutations in Sod1 are a well-described cause of amyotrophic lateral sclerosis, a progressive, fatal neurodegenerative disease caused by the degeneration of motor neurons. Most mutations interfere with the folding of Sod1, which, at least for some mutations, leads to an accumulation of the mutated Sod1 in mitochondria (45). A dynamic redox-dependent distribution of proteins between the cytosol and the IMS was also observed for other proteins such as p66Shc. This critical determinant of individual life span (46) is translocated under stress conditions from the cytosol into the IMS of mitochondria in an oxidation-mediated manner (47). Critical for this translocation reaction is a cysteine-dependent homodimerization of the protein. It is not known how this oxidation is catalyzed, but it was suggested that p66Shc is oxidized by the mitochondrial disulfide relay system in order to trap it in the IMS (47).


The formation of disulfide bonds in and between proteins is an efficient way to stabilize protein structures and to regulate protein activity in a redox-dependent manner. Recent studies on the oxidation machineries in the ER and in mitochondria have clarified the structures and mechanisms of both disulfide relay systems. Nevertheless, we only have a limited understanding of the physiological and pathological relevance, as well as of the regulatory functions, of these systems. This is particularly true for the mitochondrial disulfide relay. However, the large number of conserved cysteines in proteins of the IMS points to a critical role of thiol groups in this cellular compartment.

References and Notes

  1. We thank B. Anstett for help with Fig. 4 and M. Bien and L. Ellgaard for critical reading of the manuscript. Research in the author’s laboratories is supported by fellowships of the Deutsche Forschungsgemeinschaft and the Stiftung fur Innovation in Rheinland-Pfalz.
View Abstract

Navigate This Article