Glutathione is required to regulate the formation of native disulphide bonds within proteins entering the secretory pathway.

The formation of native disulphide bonds is an essential event in the folding and maturation of proteins entering the secretory pathway. For native disulphides to form efficiently an oxidative pathway is required for disulphide bond formation and a reductive pathway is required to ensure isomerisation of non-native disulphide bonds. The oxidative pathway involves the oxidation of substrate proteins by PDI, which in turn is oxidised by endoplasmic reticulum oxidase (Ero1). Here we demonstrate that overexpression of Ero1 results in the acceleration of disulphide bond formation and correct protein folding. In contrast, lowering the levels of glutathione within the cell resulted in acceleration of disulphide bond formation but did not lead to correct protein folding. These results demonstrate that lowering the level of glutathione in the cell compromises the reductive pathway and prevents disulphide bond isomerisation from occurring efficiently, highlighting the crucial role played by glutathione in native disulphide bond formation within the mammalian endoplasmic reticulum.


Introduction
The endoplasmic reticulum (ER) provides an environment that allows the oxidative folding and post-translational modification of proteins entering the secretory pathway. The compartmentalisation of the ER away from the cytosol allows the correct redox conditions to be established (1), which in turn enables a distinct set of folding catalysts to facilitate the formation of native disulphide bonds. A growing family of ER oxidoreductases is thought to be responsible for catalysing the formation, isomerisation and reduction of disulphide bonds (2). These oxidoreductases contain active sites homologous to the active site found in the cytosolic reductase thioredoxin, characterised by a pair of cysteine residues (CxxC) that shuttle between the disulphide and dithiol form (3). The reactions that these enzymes catalyse require the individual active sites to be maintained in either the oxidised disulphide form, for disulphide bond formation, or the reduced dithiol form, for isomerisation or reduction of disulphide bonds (4). How the active sites are maintained in either their reduced or oxidised state and how the ER maintains an environment conducive to disulphide bond formation, isomerisation and reduction has been the subject of intense speculation over the past 40 years (5,6). Recently, components of the oxidative pathway have been identified, however we know very little about the reductive pathway even though it is clear that disulphide bond isomerisation and reduction are essential for cell viability (7).
It is now firmly established that the oxidoreductase PDI catalyses the formation of disulphide bonds within the eukaryotic ER (8,9). During disulphide bond formation an intra-chain disulphide bond between the cysteine residues within the active site is able to accept two electrons from the polypeptide chain substrate resulting in the reduction of the PDI active site. Ero1p has been shown to be responsible for the oxidation of PDI in yeast, a defect in Ero1p leads to a lack of disulphide bond formation demonstrating the crucial role played by Ero1p in the oxidative folding pathway (5,6). There is growing evidence to suggest that Ero1p is an FADdependent oxidase that is able to pass electrons from PDI to the ultimate electron acceptor oxygen (10). However, Ero1p can also catalyse the oxidation of PDI under anaerobic conditions suggesting the possibility that alternative electron acceptors could substitute for oxygen under these conditions (11).
Although a clear mechanism exists to oxidise PDI, in mammalian cells most of the ER oxidoreductases, including PDI appear to be in a predominantly reduced form at steady state (12). This would suggest that a pathway exists to maintain these proteins in a reduced state within the cell. That PDI is predominantly reduced in mammalian cells contrasts with the situation in yeast where PDI is clearly predominantly oxidised shown to act as a reductase, at least in vitro, to allow the breaking of non-native disulphide bonds within MHC-class I heavy chain (13). Hence there is a requirement for both an oxidative pathway for disulphide bond formation and a reductive pathway to allow reduction and isomerisation of non-native disulphides and for these pathways to co-exist in the same intra-cellular compartment.
The mechanism for maintaining ER oxidoreductases in a reduced state could involve a protein-mediated process such as exists in the E.coli periplasm where DsbD maintains the main enzyme catalysing disulphide isomerisation DsbC in a reduced state (14). Alternatively a glutathione buffer may be involved, eliminating the requirement for a separate protein-mediated pathway for their reduction. Maintaining a pool of GSH within the ER could be brought about by continuous transport from the cytosol where glutathione reductase maintains a high concentration of GSH. Indeed some evidence exists to suggest the presence of a transport system that allows the selective passage of GSH rather than GSSG (15). Also, elimination of glutathione from yeast cells by removal of the enzyme involved in the first step of synthesis, γglutamylcysteine synthetase, did not prevent disulphide bond formation but did render the cells more prone to hyperoxidation, suggesting a direct or indirect role for glutathione in maintaining a redox balance in the yeast ER (16). Hence the possibility remains that reduction of ER oxidoreductases can be bought about either directly by glutathione or by a separate enzyme catalysed pathway.
To address this point we have chosen to study the folding and disulphide bond formation of human tissue type plasminogen (tPA). This protein contains 17 disulphide bonds and is secreted from recombinant cell lines as a mixture of two glycoforms that differ in their extent of core N-linked glycosylation. We have previously shown that conditions preventing disulphide bond formation, such as addition of the reducing agent DTT to culture medium of living cells, lead to complete glycosylation of a sequon that would otherwise undergo variable glycosylation in untreated cells (17). The extent of glycosylation of tPA is therefore intimately linked to the rate of protein folding and disulphide bond formation. The close temporal relationship between glycosylation and folding of tPA allows us to evaluate the affect of altering the redox conditions on both the rate of disulphide bond formation and also the rate of protein folding. Here we present evidence establishing that a rate-limiting step in the oxidative pathway and protein folding is the oxidation of PDI. We also show that a high concentration of intra-cellular glutathione is required to ensure the formation of native disulphide bonds.
6 stably transfected cells were maintained in the selection growth medium and overexpression of Ero1-L±myc was examined by Western blot analysis.

Immunofluorescence
Cells were grown on coverslips to 50-70% confluency, washed twice in PBS and fixed in methanol for 10 minutes at -20 o C followed by a brief rehydration with PBS. minutes. All incubations were carried out at room temperature and the proteins were visualized by exposure to Fuji Medical X-ray film.

Radiolabelling and Immunoisolation.
For pulse chase experiments, 2 × 10 6 cells per 6 cm dish were washed twice with pre-warmed methionine-and cysteine-free DMEM (Sigma), and then pre- All samples, reduced and non-reduced, were boiled for 5 mins prior to electrophoresis. Samples were resolved by SDS-PAGE and visualized by either autoradiography using Kodak Biomax MR film or phosphorimage analysis using a Fuji Film Bas 1800 phosphorimager.

Glutathione assays
Total glutathione levels were measured using the DTNB [5, 5'-dithiobis(2nitrobenzoic acid), Sigma]-GSSG reductase recycling assay (32). CHO cells (2 × 10 6 ) were harvested by trypsinisation and cell pellet was lysed in 100 µl of ice cold 8 mM HCl / 1.3 % 5-sulfosalicylic acid (SSA, Sigma) by agitating with glass beads. The lysate was incubated on ice for 30 minutes. Glass beads, precipitated protein and cell debris was removed by centrifugation at 12,000g for 5 minutes. Total glutathione was measured by adding 10 µl of the lysate to 1 ml assay mixture pre-warmed to 30 0 C in a spectrophotometer cuvette. GSSG reductase was then added with mixing to initiate the assay and change in absorbance at 405 nm was measured over 4 minutes. Standard curves were generated using 0.5 -4 nmol GSH solubilised in 8 mM HCl / 1.3 % SSA.

PDI redox state
The in vivo redox state of PDI was assayed by modification with the thiol -reactive reagent acetamido-maleimidylstilbene-disulphonic acid AMS (Molecular Probes).
Cells were incubated for 10 min at 37°C with or without DTT (5 mM) or 4,4'dithiodipyridine (0.5 mM). At the end of the incubation period, the cells were transferred to ice and washed twice with ice cold phosphate-buffered saline (PBS) containing 20mM NEM to minimise disulphide bond rearrangements. The cells were lysed in 1 ml of lysis buffer (without NEM) for 10 minutes on ice. Cell lysates were centrifuged at 12,000g for 20 minutes to remove insoluble material. The supernantant was transferred into a fresh tube, 1% SDS and 10mM TCEP was added to all the samples. Samples were boiled for 2 min, cooled and treated with 25 mM AMS for 60 min at room temperature, in the dark.

Overexpression of Ero1-L± leads to enhanced oxidative protein folding
Our initial studies focused on the consequence of overexpressing Ero1 on the ability of tPA to form disulphide bonds, fold correctly and be secreted. Our approach was to generate stable cell-lines expressing both t-PA and wild type Ero1 or Ero1 containing single cysteine point mutations known to compromise function (18,19). A indicates the formation of intra-chain disulphide bonds as previously described for the endogenous protein (19). Some higher molecular weight bands can also be seen when the cell lysate was separated under non-reducing conditions. These higher molecular weight bands are likely to be mixed disulphides between exogenously expressed Ero1-L± and endogenous PDI, as characterised previously (12). No such high molecular weight bands were seen with either cysteine mutants demonstrating their inability to form functional mixed disulphides with PDI ( Fig. 1A, lanes 6 and 7).
These results demonstrate that the wild type Ero1-L± expressed in the CHO cells has folded correctly and is functional as indicated by its ability to form intrachain disulphides and mixed disulphides with other known ER oxidoreductases.
The intracellular location of exogenously expressed Ero1-L± protein in the stably The ability of cysteine mutants of Ero1 to bring about a similar effect was also investigated. Mammalian Ero1 contains 14 cysteine residues of which three (Cys 391, 394 and 397) have been shown, by single point mutation, to be involved in either stabilisation of structure or enzymatic function (12,19). An analysis of single cysteine mutations has also been carried out with yeast Ero1 and four cysteine residues (Cys100, Cys105, Cys352 and Cys 355) were found to be absolutely required for function (18). These correspond to Cys99, Cys104, Cys394 and Cys397 in the human protein. We created two Ero1 cysteine mutants with mutation in cys99 or both cys99 and cys394 and prepared separate stable cell-lines expressing these proteins (Fig.   1A). Both the cell-line expressing Ero1-cys99 ( Fig. 2A  In the Ero1-L± overexpressing cells, tPA migrated as a diffuse band of disulphide bonded protein within 5 minutes of the chase (Fig 3B, Ero1) and there was no further increase in mobility after 20 minutes of chase, indicating that in these cells disulphide bond formation was complete between 10-20 minutes into the chase. Thus overexpression of Ero1-L± accelerates the recovery from DTT treatment, allowing more rapid oxidative folding within mammalian cells.
The acceleration of the oxidative pathway would suggest that the rate-limiting step in the pathway is the oxidation of PDI by Ero1 as any increase in Ero1 causes acceleration in the rate of oxidation of the substrate.
An increase in the intensity of tPA immunoisolated during the chase period after labelling in the presence of DTT was observed at early time points in both cell lines ( Fig. 3A and B). This could reflect a lack of protein synthesis inhibition by cycloheximide. However, this is unlikely to be the case as only fully glycosylated tPA was detected during the chase and any protein synthesised during the chase would be variably glycosylated as the redox conditions within the ER are restored rapidly (24). Affinity of the polyclonal antibody for reduced and non-reduced tPA was determined (results not shown) and it was found that only 35% of the tPA immunoisolated from the non-reducing sample was precipitated from the reducing sample, demonstrating that the antibody has a reduced affinity for non-disulphide does not effect the rate of secretion but does accelerate the formation of native disulphide bonds giving rise to tPA that is secreted from the cell. The fact that the tPA is secreted from the cell and that it has escaped the ER quality control system for glycoproteins suggests that it has folded correctly.

Glutathione is required for isomerisation of non-native disulphide bonds
Genetic evidence in yeast has demonstrated that glutathione is dispensable for  (Fig. 6 A/B).
In the BSO untreated cells, the first correctly folded tPA molecules were observed after 45-60 minutes of chase (Fig. 6A). This result confirms our earlier observation that in the untransfected CHO-tPA cells, disulphide bond formation was complete between 30-45 minutes into the chase (Fig. 3). The fact that PDI was more oxidised in the Ero1 overexpressing cell-line and that PDI forms mixed disulphides with Ero1 strongly suggests that it is a substrate for Ero1. It has previously been shown that the oxidation state of ERp57 is not altered in cells overexpressing Ero1 and that no mixed disulphides are formed between ERp57 and Ero1, leading to the conclusion that ERp57 is not a substrate for Ero1 (12).
ERp57 is an abundant ER oxidoreductase that is a member of the PDI family of proteins, but is distinct in that it associates with calnexin and calreticulin and has been postulated specifically to be involved in disulphide bond formation in glycoproteins substrates (27). We also looked at the redox state of ERp57 and found that it is reduced both in the untransfected and Ero1 overexpressing cell-line (SC and NJB unpublished results). The protein we used in this study is a glycoprotein and interacts with calnexin (28), however the fact that Ero1 overexpression effects disulphide bond formation would suggest that PDI rather than ERp57 is involved in at least the oxidation of disulphide bonds in glycoprotein substrates. Clearly this does not rule out a role for ERp57 in the isomerisation of non-native disulphides a role that has been demonstrated in vitro with purified glycoproteins (13). The observation that in cells the active site disulphide in ERp57 is reduced and that this protein is not a substrate for Ero1 suggests a kinetic segregation of an oxidative pathway involving PDI and a reductive pathway involving ERp57 or one of the other ER oxidoreductases. If this is the case then clearly a pathway must exist to maintain a level of reducing equivalents in the ER lumen to facilitate reduction of ERp57.

The role of glutathione in maintaining the reductive pathway
The ability to eliminate glutathione biosynthesis in yeast cells has established that glutathione is not required for the oxidation of disulphide bonds (16). However, the depletion of glutathione from yeast cells causes oxidation of proteins within the ER and suppresses a temperature sensitive Ero1 mutant suggesting a role in the reduction of disulphide bonds. Our results in mammalian cells demonstrate that a normal level of glutathione is required within the cell to prevent the formation of aberrant disulphides during the posttranslational folding of tPA. The formation of native disulphide within the ER lumen is therefore dependent upon both an oxidative pathway and a reductive pathway to prevent the formation of non-native disulphides, the reductive pathway being compromised when the level of glutathione within the cell is decreased. The fact that glutathione can cross the ER membrane either through a specific transporter (15) or simply through pores in the membrane (29) suggests that this low molecular weight thiol could provide the necessary reducing equivalents to facilitate the reduction of folding proteins directly or via reduction of ER oxidoreductases such as ERp57. As glutathione is a poor substrate of Ero1 in vitro (26), it is unlikely to be directly oxidised by the Ero1 in vivo therefore providing segregation of the pathways of oxidation and reduction within the ER.
In general terms the ability to catalyse two thermodynamically opposed reactions within the same cellular compartment requires that the two reaction pathways be kinetically partitioned. In the prokaryotic periplasm this partitioning is solved by the