Ero1-α and PDIs constitute a hierarchical electron transfer network of endoplasmic reticulum oxidoreductases

Abstract

Ero1-α and endoplasmic reticulum (ER) oxidoreductases of the protein disulfide isomerase (PDI) family promote the efficient introduction of disulfide bonds into nascent polypeptides in the ER. However, the hierarchy of electron transfer among these oxidoreductases is poorly understood. In this paper, Ero1-α-associated oxidoreductases were identified by proteomic analysis and further confirmed by surface plasmon resonance. Ero1-α and PDI were found to constitute a regulatory hub, whereby PDI induced conformational flexibility in an Ero1-α shuttle cysteine (Cys99) facilitated intramolecular electron transfer to the active site. In isolation, Ero1-α also oxidized ERp46, ERp57, and P5; however, kinetic measurements and redox equilibrium analysis revealed that PDI preferentially oxidized other oxidoreductases. PDI accepted electrons from the other oxidoreductases via its a' domain, bypassing the a domain, which serves as the electron acceptor from reduced glutathione. These observations provide an integrated picture of the hierarchy of cooperative redox interactions among ER oxidoreductases in mammalian cells.

Figure 1.

Ero1-α binds to ER-resident oxidoreductases and preferentially oxidizes PDI. (A, left) HEK293T cells (Mock) or HEK293T cells stably expressing Ero1-α–FLAG (Ero1-α(WT)–FLAG) were lysed and subjected to immunoprecipitation (I.P.) using antibodies against FLAG. (right) Resulting precipitates were examined by immunoblot analysis with the indicated antibodies. The black line on the right indicates the removal of intervening lanes for presentation purposes. (B) Association or dissociation rate constants (k_on or k_off) were determined with a two-state reaction model, and their first equilibrium constants are plotted. Diagonal lines represent dissociation constants (K_d). Data represent means from at least four individual experiments (see also Fig. S1, C and D). (C) Schematic models of oxidative relays (top) and electron transfer relays (bottom) between Ero1-α and PDI. (D) Assays were conducted in a sealed chamber starting with air-saturated buffer containing 10 mM GSH, which was regarded as the 100% oxygen level (∼250 µM oxygen). Control experiments are shown in Fig. S1 E. (D and E) Oxidation of reduced oxidoreductases was initiated by the injection of 2 µM Ero1-α (D) or Ero1-α(C104A/C131A) (E) and was monitored with an oxygen electrode. ROS, reactive oxygen species.

Figure 2.

PDI dominantly alters the activity of endogenous Ero1-α and increases conformational flexibility in a shuttle cysteine, Cys99, of Ero1-α. (A and B) HEK293T cells in which the series of oxidoreductases was overexpressed for 24 h (A) or knocked down for 72 h (B) were trapped by alkylation with NEM and solubilized in lysis buffer. The supernatant was subjected to precipitation with Con A–Sepharose, and the glycoprotein fraction was analyzed by immunoblot analysis with the anti–Ero1-α antibody under nonreducing condition. Black lines indicate the removal of intervening lanes for presentation purposes. As controls, cells were transfected with mock vector or two different siRNAs (L, low GC content; M, medium GC content). The activated states of Ero1-α (O_x1/ (O_x1 + O_x2)) were quantified as shown in the bottom graphs of Fig. 2 (A and B). Mutants of full-length PDI containing only the intact a or intact a′ catalytic thioredoxin domain are shown as PDI(a) and PDI(a′), respectively (see also Fig. 2 C). Data represent means ± SDs from three independent experiments (see also Fig. S2 A). Although we have reported that ERdj5 works as a reductase in the ER-associated degradation process, it had almost no significant effect on the redox states of Ero1-α (Ushioda et al., 2008). (C) Schematic representation of human PDI proteins with the CGHC active sites and the mutated AGHA sites indicated. x denotes a linker region between the b′ and the a′ domain. (D) Schematic and simplified model of Ero1-α and its intramolecular electron flow (Araki and Inaba, 2012). Four spheres show the catalytically essential cysteines, two of which form the shuttle disulfide (light gray, Cys94-Cys99) on the flexible loop. The other cysteines form active site disulfides (dark gray, Cys394-Cys397) located proximally to the cofactor (flavin adenine dinucleotide [FAD]). Dashed arrows indicate the electron transfer pathway. The four-helix core is shown as a cylinder. (E) ¹³C NMR spectra of the constitutively active Ero1-α(C104A/C131A) selectively labeled with ¹³C at the carbonyl carbons of cysteine residues (Ero1*). Spectra were measured in the absence (top) and presence of equimolar amounts of WT PDI (top middle), PDI(AA) mutant (bottom middle), or WT PDI together with 4.25 mM somatostatin (bottom). The spectrum of the unlabeled protein has been subtracted. The asterisk indicates the peak originating from Cys99. A.U., arbitrary unit.

Figure 3.

Synergistic effects of oxidoreductases (ERp46, ERp57, and P5) on oxygen consumption in the presence of PDI and constitutively active Ero1-α. (A) Schematic model of intermolecular electron flow among oxidoreductases. (B–E) Oxygen consumption was assayed in the presence of 10 mM GSH and 10 µM ERp46 (B), ERp57 (C), P5 (D), and ERp72 (E) with or without 5 µM PDI. Calc represents calculated data from individual consumption rates of the oxidoreductases. Mix represents actual consumption rates under conditions in which 10 µM of each oxidoreductase and 5 µM PDI were mixed in the presence of 10 mM GSH. (F) 10 µM P5 and 10 µM ERp57, without PDI, were mixed in the presence of 10 mM GSH. ROS, reactive oxygen species.

Figure 4.

Synergistic effect is mediated by the a′ domain of PDI. (A) Schematic model of electron transfer relays between Ero1-α and PDI. Ero1-α oxidizes the a′ domain of PDI, which in turn oxidizes the a domain internally. The oxidized a domain is rereduced by GSH. (B) Kinetics of oxygen consumption by 2 µM constitutively active Ero1-α(C104A/C131A) during the reaction with 5 µM human PDI variants, as depicted in the figure, in the presence of 10 mM GSH. (C–G) Oxygen consumption was assayed in the presence of 10 mM GSH and 10 µM ERp46 (C) with or without 5 µM PDI(a) or 10 µM ERp46 (D), ERp57 (E), P5 (F), or ERp72 (G) with or without 5 µM PDI(a′). Calc shows the calculation data from individual consumption rates of these oxidoreductases. Mix represents the actual consumption rates under conditions in which 10 µM of each oxidoreductase and 5 µM PDI were mixed in the presence of 10 mM GSH. (H) Schematic model of electron transfer relays among Ero1-α, the a′ domain of PDI, and oxidoreductases (ERp46, ERp57, and P5). ROS, reactive oxygen species.

Figure 5.

ERp46 substitutes for the a domain of PDI and increases conformational flexibility in a shuttle cysteine, Cys99, of Ero1-α. ¹³C NMR spectra of the constitutively active Ero1-α(C104A/C131A) labeled with ¹³C selectively at the carbonyl carbons of cysteine residues (Ero1*). (A–E) Spectra were measured in the absence (A) and presence of equimolar amounts of PDI (B), PDI(a′) (C), ERp46 (D), or PDI(a′) and ERp46 (E). Spectrum of the unlabeled protein has already been subtracted. PDI(a′) or ERp46 had no apparent effect on the mobilization of Cys99. The asterisk indicates the peak originating from Cys99. A.U., arbitrary unit.

Figure 6.

PDI works as a penultimate electron acceptor in the Ero1-α–mediated oxidation cascade. (A) Oxidoreductase mutants (CA or AA) and Ero1-α(WT) with FLAG tag were expressed in HEK293T cells, and anti-FLAG immunoprecipitates were analyzed by direct nanoflow liquid chromatography coupled with tandem mass spectrometry. Reproducibly identified oxidoreductases from four independent trials are listed. Each number indicates the identified peptide number of each protein in an individual experiment. Light and dark shading indicate identified prey and bait peptides, respectively. (B) Free sulfhydryl groups of the cysteine residues were modified with mPEG2000-mal after incubation with different [GSH]²/[GSSG] ratios in a buffer containing 0.1 mM GSSG and varying concentrations of GSH (0.05–10 mM) under a nonoxidative atmosphere at 25°C followed by SDS-PAGE and CBB staining. The apparent equilibrium constants between oxidoreductases and glutathione were determined by the nonlinear least square fitting of the data (Fig. S4, A–D). K_eq values were determined from at least three independent trials as follows: 3.02 ± 0.14 (PDI, correlation coefficient: 0.985), 2.73 ± 0.10 (ERp46, 0.994), 1.69 ± 0.12 (ERp57, 0.991), 1.67 ± 0.08 (ERp72, 0.994), and 0.91 ± 0.04 (P5, 0.995). A.U., arbitrary unit.

Figure 7.

Model of inter- and intramolecular electron transfer cascade among ER oxidoreductases. (A) Proposed model for the synergistic effect. Details are described in the Discussion. ERp57 and P5 could be oxidized in a similar fashion to that of ERp46. (B) The intracellular amounts of several ER oxidoreductases in HEK293T or HeLa cells were roughly estimated by immunoblot analysis using the appropriate antibodies, with parallel loading of recombinant proteins on the same gel. The intracellular amounts of PDI, ERp57, P5, and ERp72 were roughly similar, but those of ERp46 and ERp44 were about one fifth of the former. The amount of Ero1-α was only one tenth that of PDI. Black lines indicate the removal of intervening lanes from some of the gels for presentation purposes. (C) Proposed model for electron transfer cascades among the ER oxidoreductases. Red and purple arrows indicate the directions of electron transport. Broken line or arrows indicate a nonredox-based interaction or presumed electron transport, respectively. The apparent redox equilibrium constants are from Fig. 6 B. The relative intracellular amounts of these oxidoreductases are indicated by the box sizes.