Oxidative protein folding by an endoplasmic reticulum-localized peroxiredoxin

Abstract

Endoplasmic reticulum (ER) oxidation 1 (ERO1) transfers disulfides to protein disulfide isomerase (PDI) and is essential for oxidative protein folding in simple eukaryotes such as yeast and worms. Surprisingly, ERO1-deficient mammalian cells exhibit only a modest delay in disulfide bond formation. To identify ERO1-independent pathways to disulfide bond formation, we purified PDI oxidants with a trapping mutant of PDI. Peroxiredoxin IV (PRDX4) stood out in this list, as the related cytosolic peroxiredoxins are known to form disulfides in the presence of hydroperoxides. Mouse embryo fibroblasts lacking ERO1 were intolerant of PRDX4 knockdown. Introduction of wild-type mammalian PRDX4 into the ER rescued the temperature-sensitive phenotype of an ero1 yeast mutation. In the presence of an H(2)O(2)-generating system, purified PRDX4 oxidized PDI and reconstituted oxidative folding of RNase A. These observations implicate ER-localized PRDX4 in a previously unanticipated, parallel, ERO1-independent pathway that couples hydroperoxide production to oxidative protein folding in mammalian cells.

Figure 1. A trapping mutant of PDI engages the known downstream electron acceptor, ERO1α, in mammalian cells

(a) Autoradiograph of immunoglobulin-M from metabolically-labeled wildtype (α^+/+;β^+/+) and ERO1 mutant (α^mut/mut;β^mut/mut) lipopolysaccharide blasts. Cells were labeled with ³⁵S methionine-cysteine for 15 minutes and chased for the indicated time with unlabeled media before lysis and immunoprecipitation. The upper panel is a radiograph of a non-reducing gel and the lower panel is a reducing gel. The migration of IgM monomers (μ) dimers (2μ) and pentamers of dimers (5μ) is indicated. (b) Immunoblots of FLAG-tagged proteins (left) or endogenous ERO1α (right) immunopurified with the FLAG-M1 antibody from lysates of HEK 293T cells that were untransfected or transfected with expression plasmids of the indicated proteins. The upper panels are of reducing and the lower panels non-reducing gels. The low mobility complex containing ERO1α immunoreactive material in complex with the FLAG-tagged PDI C-terminal active site trapping mutant (FLAG-PDI^C400S) is noted by an asterisks on both non reducing gels.

Figure 2. A PDI active-site trapping mutant engages PRDX4 in mammalian cells

(a) Coomassie stained non-reducing SDS-PAGE of proteins immunopurified in complex with FLAG-tagged PDI trapping mutant from untransfected and transfected cells. The boxes demarcate the binning for the mass spectrometric protein analysis. (b) List of proteins identified by LC MS/MS sequencing of tryptic peptide of endogenous proteins captured in a disulfide linked complex by a FLAG M2-tagged trapping mutant PDI^C400S expressed in HEK 293T cells (shown in Fig. 2a). Known cytosolic, nuclear and mitochondrial proteins were removed from the list and the remaining proteins were sorted by descending exponentially modified protein abundance index (emPAI) (Ishihama et al., 2005). The proteins are identified by their International Protein Index accession number (IPI) and their common name. (c) Immunoblot of endogenous PRDX4 and FLAG-tagged wildtype PDI and trapping mutant PDI^C400S immunopurified with the FLAG-M1 antibody from lysates of HEK 293T cells that were untransfected or transfected with expression plasmids for the indicated proteins. The lower two panels are of the same proteins in the lysates before the IP (“Input”). The proteins shown were resolved on reducing SDS-PAGE (d) FLAG and endogenous PRDX4 immunoblot of a non-reducing gel with samples as in panel “b”. The low mobility complex containing PRDX4 immunoreactive material in complex with FLAG-PDI^C400S is noted by an asterisks on both non reducing gels.

Figure 3. PRDX4 buffers the consequences of ERO1 deficiency in MEFs

(a) Immunoblot of endogenous PRDX4, ERO1α and Actin in wildtype (α^+/+;β^+/+) and ERO1 mutant (α^mut/mut;β^mut/mut) MEFs 4 days after transduction with a puro^r-marked lentivirus carrying a irrelevant insert (mock) or three different short hairpin RNAs directed to mouse PRDX4. (b) Crystal violet stained plates following two weeks puromycin selection after seeding with 8×10³ (Low density) or 4×10⁴ (High density) MEFs of the indicated genotype and transduced with puro^r-tagged lentivirus carrying the indicated shRNA. (c) Bar diagram of the ratio of cell mass accrued in the ERO1 double mutant versus wildtype MEFs from the experiment shown in “b”. The ratio was normalized to 1 in the cells transduced with the irrelevant shRNA. Shown are means ± SEM of a typical experiment reproduced three times (n=3, p<0.05) (d) Transmission electron micrographs of MEFs of the indicated genotype following transduction with a lentivirus containing an irrelevant shRNA or an shRNA directed to PRDX4. Note the dilation of the ER in the mutant cells transduced with the shRNA to PRDX4.

Figure 4. Hypersensitivity to reducing agents, defective collagen secretion and a more reducing ER redox poise in ERO1-deficient cells lacking PRDX4

(a) Bar diagram of cell mass of wildtype (WT) or Ero1α^mut/mut;Ero1β^mut/mut compound mutant cells (MUT) transduced with mock or PRDX4 RNAi lentivirus (KD1) remaining 3 hours after a 30 minute pulse with DTT (5 mM) followed by recovery in normal media, normalized to the cell mass of a parallel culture of the same cells that had not be exposed to DTT. Shown are aggregate means ± SEM from a typical experiment conducted in triplicate (n=3; *p<0.05) and reproduced three times. (b) Bar diagram of soluble collagen secreted into the conditioned media from the cells described in panel “a”, normalized to the number of live cells in the culture. Shown are means ± SEM from three experiments conducted on different occasions in triplicate (n=9; * p<0.01). (c) Immunoblot of FLAG_M1_roGFP_iE immunopurified from the ER of ERO1 deficient MEFs with normal levels of PRDX4 or PRDX4 knockdown. The purified proteins were resolved by SDS-PAGE under non-reducing or reducing conditions. The position of the reduced and oxidized FLAG_M1_roGFP_iE in a typical experiment reproduced three times is shown. (d) Quantitative representation of the data from the experiment represented in panel C. Bar diagram of the ratio of reduced and oxidized roGFP in the Ero1α^mut/mut;Ero1β^mut/mut compound mutant cells transduced with mock or PRDX4 RNAi lentivirus (KD1) (n=3 *p<0.05, comparing oxidized fraction in the two genotypes)

Figure 5. ER localized, enzymatically-active PRDX4 rescues a lethal mutation of yeast ERO1

(a) Photomicrographs of serial dilutions of untransformed ero1-1 mutant yeast or transformed with yeast expression plasmids lacking (pRS316) or containing the PRDX4 coding sequence cultured at the permissive temperature of 24°C or the non-permissive temperature of 37°C. (b) Plot of absorbance at 600 nm of cultures inoculated with ero1-1 yeast transformed with the expression plasmid lacking (pRS316) or containing the PRDX4 coding sequence and cultured at the permissive temperature of 24°C or the non-permissive temperature of 37°C. Shown are mean ± SEM of a typical experiment conducted in triplicate (n=3; * the last four time points p<0.001 by two-way ANOVA). (c) Photomicrographs of serial dilutions of ero1-1 yeast transformed with pRS316 or the PRDX4 expressing plasmids and plated in the absence or presence of the toxin FOA at 24°C or 37°C. Note the inability of the ero1-1 yeast transformed with the PRDX4 expression plasmid to survive on FOA at the non-permissive temperature. (d) Photomicrographs of serial dilutions of ero1-1 yeast transformed with pRS316 or plasmids expressing wildtype or the indicated mutations in PRDX4 (ΔSP: lacks the signal peptide required for ER import of PRDX4). The lower panel is an immunoblot of FLAG-tagged PRDX4 from the four transformed strains and the parental ero1-1 yeast.

Figure 6. PRDX4 catalyzes H₂O₂ and PDI-dependent oxidative refolding of RNase A in vitro

(a) Coomassie stained reducing SDS-PAGE of purified PDI, wildtype and mutant PRDX4-GST tagged proteins used in the experiments shown below. (b) Time dependent change in absorbance of Ellman’s reagent reacted with reduced PDI (150 μM) after introduction of glucose (2.5 mM) and glucose oxidase (GO, 10 mU/mL) as a source of H₂O₂ and the indicated wildtype or mutant PRDX4 proteins shown in panel “a” (5 μM). The absorbance of each reaction mixture at t=0 is set at 100%. Shown are means ± SEM of a typical assay conducted in triplicate and reproduced 4 times (n=3, * P<0.05, **P<0.001). (c) Time dependent change in absorbance of Ellman’s reagent in reaction mixes containing 150 μM reduced glutathione in the presence of PDI (10 μM), PRDX4 (5 μM) or both enzymes. Glucose (2.5 mM) and glucose oxidase (GO, 10 mU/mL) were included as a source of H₂O₂ in all reactions. The absorbance of each reaction mixture at t=0 is set at 100%. Shown are means ± SEM of a typical assay conducted in triplicate and reproduced 3 times (n=3, **P<0.001). (d) Coomassie-stained SDS-PAGE of reduced and denatured RNAse A (25μM) after reacting for the indicated time with reduced PDI (7 μM) and PRDX4 (5 μM) in the presence of glucose (2.5 mM) and glucose oxidase (10 mU/mL) as a source of H₂O₂. Lane 13 in the top-most panel or lane 9 in the two lower panels contains a sample of native oxidized RNase A to serve as a reference. (e) Time-dependent change in RNase activity measured spectrophotometrically by absorbance at 296 nm using cCMP (4.5 mM) as substrate. The re-folding of RNase A was initiated at t=0 in reactions as in panel “C”. Enzymatic activity of an equal amount of native RNase A is set as 100%. Shown are means ± SEM of a typical experiment reproduced 4 times (n=3, * p<0.05, ** p<0.001 by two-way ANOVA).

Figure 7. Model for PRDX4-mediated oxidative protein folding in the ER

Disulfide bond formation (step I) leaves PDI in a reduced state. PDI is re-oxidized by reducing the active site disulfide (formed between two PRDX4 protomers) in step II. PRDX4 is re-oxidized by interaction of its peroxidic cysteine (C127) with H₂0₂, releasing one molecule of water. The sulfenic acid is resolved by C287, regenerating the disulfide and releasing the second molecule of water (step III). PRDX4 may utilize H₂O₂ produced by ERO1 as well as H₂O₂ produced by mitochondrial respiration or NADPH oxidases that may diffuse into the ER.