The peroxiredoxins are a ubiquitous family of proteins involved in protection against oxidative stress through the detoxification of cellular peroxides. In addition, the typical 2-Cys peroxiredoxins function in signalling of peroxide stress and as molecular chaperones, functions that are influenced by their oligomeric state. Of the human peroxiredoxins, Prx IV (peroxiredoxin IV) is unique in possessing an N-terminal signal peptide believed to allow secretion from the cell. Here, we present a characterization of Prx IV in human cells demonstrating that it is actually retained within the ER (endoplasmic reticulum). Stable knockdown of Prx IV expression led to detrimental effects on the viability of human HT1080 cells following treatment with exogenous H2O2. However, these effects were not consistent with a dose-dependent correlation between Prx IV expression and peroxide tolerance. Moreover, modulation of Prx IV expression showed no obvious effect on ER-associated stress, redox conditions or H2O2 turnover. Subsequent investigation demonstrated that Prx IV forms complex structures within the ER, consistent with the formation of homodecamers. Furthermore, Prx IV oligomeric interactions are stabilized by additional non-catalytic disulfide bonds, indicative of a primary role other than peroxide elimination.
Exposure to ROS (reactive oxygen species) has long been recognized as a major problem for aerobic organisms, and numerous biological mechanisms exist that facilitate removal of ROS within cells. Imbalances between these protective processes and those generating ROS have been implicated in pathogenesis of cancers and important degenerative disorders including Alzheimer's disease and Parkinson's disease .
ROS also impart significant benefits. For example, H2O2 is important as a cytotoxic agent during microbial engulfment by phagocytic immune cells . Such cells generate H2O2 catalytically from NADPH oxidase-derived superoxide anions (O2−), and there is increasing evidence that a wide range of mammalian cytokines and growth factors also stimulate H2O2 production via NADPH oxidases for second messenger signalling purposes . Despite its defined roles, H2O2 remains a reactive molecule that can directly modify lipids, proteins and nucleic acids. This is particularly relevant, given that substantial H2O2 is generated as a by-product of metabolic processes such as electron transport ‘leakage’ releasing O2− from the mitochondria . Such reactions highlight the importance of peroxide detoxification pathways.
Peroxide degradation can occur via several routes including direct reaction with glutathione, breakdown by catalase (free H2O2) or glutathione peroxidases (H2O2 and lipid hydroperoxides), or reaction with vitamins and other non-enzymatic antioxidants . In addition, the peroxiredoxin family of proteins has combined peroxidase activity with other functions, including the ability to communicate peroxide stress in the cell.
Mammals possess six peroxiredoxin isoforms [5,6] with cellular locations including the cytosol [Prx I (peroxiredoxin I), II and VI], nucleus (Prx I), mitochondria (Prx III and VI) and peroxisomes (Prx V) and potentially secreted (Prx IV). Each is characterized by a redox-active ‘peroxidatic’ cysteine residue that attacks peroxides, becoming oxidized to a cysteine sulfenic acid in the process [7–9]. The best-characterized peroxiredoxin family members are the typical 2-Cys peroxiredoxins (including Prx I, II, III and IV in humans), which contain an additional ‘resolving’ cysteine near the C-terminus. Typical 2-Cys peroxiredoxins function as obligate homodimers and, following peroxide elimination, the peroxidatic cysteine sulfenic acid reacts with the resolving cysteine of its partner to form a stable intermolecular disulfide [7,10]. This disulfide may be reduced by a cell-specific disulfide reductase regenerating the active-site thiols.
Most 2-Cys peroxiredoxins undergo further fluid transition from dimers to toroid decamers and back again [11–14]. This process optimizes active-site arrangement for efficient catalysis, with active-site disulfide formation subsequently destabilizing the complex. Several studies have shown that hyperoxidation of the peroxidatic cysteine to a sulfinic acid (SO2H) derivative can occur in high peroxide concentrations [15,16], which in turn stabilizes peroxiredoxin decamers by preventing the formation of the resolving disulfide . In mouse lung cells, this was demonstrated to lead to aggregates of Prx II decamers whose appearance and subsequent breakdown correlated with arrest and eventual resumption of the cell cycle . Hence, Prx II's oligomeric state may be used as a monitor of cytosolic H2O2. Further oligomeric state-dependent peroxiredoxin functions have been implicated, most notably a chaperone activity associated with high-molecular-mass fractions of Saccharomyces cerevesiae cytosolic peroxiredoxins following stress  and with decameric human Prx I in vitro . In the latter case, the decamer is covalently stabilized by non-catalytic disulfides preventing dimer–decamer transitions. This rigidity appears to reduce peroxidase activity and increase prevalence of chaperone activity .
Prx IV is the least well characterized of the human 2-Cys peroxiredoxins and is unique in possessing an N-terminal secretory signal. Despite being identified a decade ago, some confusion exists as to the true nature of Prx IV in mammalian cells. Prx IV has been described as both a cytosolic protein attenuating activity of NF-κB (nuclear factor κB)  and as a secreted protein activating NF-κB . Later studies investigating rat Prx IV concluded that it was secreted and bound at the cell surface following transient overexpression in African green monkey cells [22,23]. The only consistent finding between these studies was the ability of Prx IV to act as a peroxidase in vitro. Consequently, many questions remain unanswered regarding the size, subcellular location and physiological relevance of Prx IV.
The possibility of Prx IV traversing the secretory pathway is intriguing, particularly given that oxidative protein folding in the ER (endoplasmic reticulum) has been recently proposed as another significant source of H2O2 within the cell [24,25]. H2O2 may be generated through electron transfer to molecular oxygen by Ero1 (endoplasmic reticulum oxidoreductin 1) proteins during the oxidation of PDI (protein disulfide-isomerase) [26,27]. Such situations may not be limited to the ER, as oxidative protein folding by other means may continue through the secretory pathway into the Golgi apparatus and perhaps beyond . Given the delicate redox balance required for native disulfide bond formation, we hypothesize that localized mechanisms might exist for the detection and removal of ROS generated during the oxidative folding process. Prx IV provides an attractive candidate given the described bifunctionality for other human Prx family members. To this end, we performed an investigation of Prx IV in vivo, in which endogenous Prx IV was demonstrated to be both translocated to and retained within the ER of human cells. Cell lines were created in which Prx IV was stably overexpressed or knocked down and the influence of altered expression on H2O2 turnover and ER homoeostasis was investigated. Subsequent structural analyses of Prx IV complexes formed within cells provided insights into the behaviour of Prx IV in comparison with the previously characterized human 2-Cys peroxiredoxins.
Chemicals and reagents
All reagents were acquired from Sigma–Aldrich (Poole, Dorset, U.K.) and enzymes from Promega (Chilworth, Southampton, U.K.) unless otherwise stated.
A rabbit polyclonal antibody to Prx IV was purchased from Lab Frontier (Seoul, Korea), whereas rabbit polyclonal antibody to Alix [ALG2 (asparagine-linked glycosylation 2 homologue)-interacting protein X] was donated by Professor Philip Woodman (Faculty of Life Sciences, University of Manchester, Manchester, U.K.). Mouse monoclonal antibodies recognizing α-tubulin and the KDEL ER-retrieval motif have been described previously [29,30] and were gifts from Professor Keith Gull (Sir William Dunn School of Pathology, University of Oxford, Oxford, U.K.) and Professor Stephen Fuller (The Structural Biology Programme, European Molecular Biology Laboratory, Heidelberg, Germany) respectively. A goat polyclonal antibody specific for the N-terminal region of BiP (immunoglobulin heavy-chain-binding protein) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.), whereas antibodies to the ER oxidoreductases PDI, ERp57 (57 kDa endoplasmic-reticulum protein) and ERp72 have been described previously .
Transcription and translation in vitro
A clone encoding human Prx IV in pOTB7 was obtained from Geneservice (Cambridge, U.K.) and in vitro transcription and translation were performed essentially as described previously . DNA was linearized with XhoI and transcribed using SP6 polymerase. The transcript was translated using rabbit reticulocyte lysate (Flexi-lysate; Promega) with SP cells (semi-permeabilized cells) added as required. Proteinase K treatment of SP cells was performed for 25 min on ice with or without 1% (v/v) Triton X-100, using 0.2 mg/ml proteinase K in the presence of 10 mM CaCl2, and terminated by 1 mM PMSF. When added, SP cells were isolated by centrifugation and resuspended in SDS/PAGE sample buffer [31.25 mM Tris/HCl, pH 6.8, 2% (w/v) SDS, 5% (v/v) glycerol and 0.01% (w/v) Bromophenol Blue]. Otherwise, reactions were mixed directly with SDS/PAGE sample buffer.
Electrophoresis and Western blotting
Samples for SDS/PAGE were resuspended in SDS sample buffer and heated to 100 °C for 5 min. For reducing conditions, DTT (dithiothreitol) was added to a final concentration of 50 mM. Gels containing radioactive samples were fixed in 10% (v/v) acetic acid and 10% (v/v) methanol, dried and exposed to a Kodak Biomax MR film (Genetic Research Instrumentation Limited, Rayne, Braintree, Essex, U.K.). For Western blotting, gels were transferred on to nitrocellulose and blocked using 3% (w/v) non-fat dried skimmed milk in TTBS [Tris-buffered saline with Tween 20: 10 mM Tris, 150 mM NaCl, pH 7.5, and 0.1% (v/v) Tween 20]. Primary antibody incubations were performed for 1 h at room temperature (22 °C) with 3% (w/v) non-fat dried skimmed milk powder. As secondary antibodies, polyclonal goat anti-rabbit, rabbit anti-goat and rabbit anti-mouse immunoglobulins (each conjugated to horseradish peroxidase) were obtained from Dako (Ely, Cambs., U.K.). Secondary antibodies were diluted 1:2000 in TTBS and incubation was performed at room temperature for 1 h. Products were visualized using an enhanced chemiluminescent substrate [Perbio Science (UK) Ltd, Cramlington, Northumberland, U.K.] and a Fuji Super RX film (Fujifilm UK, Bedford, U.K.).
HT1080 human fibrosarcoma cells were suspended in buffer A (50 mM Tris/HCl, 0.25 M sucrose, 25 mM KCl, 0.5 mM MgCl2 and 1 mM EDTA) at 2×107 cells/ml and disrupted using a ball-bearing homogenizer with 10 μm clearance. Insoluble debris and nuclear material was removed at 500 g for 3 min and post-nuclear supernatant was centrifuged at 60000 rev./min for 10 min in a Beckman TLA 100.1 rotor to pellet the organelle membranes. Membranes were resuspended in buffer A and treated with proteinase K, when required, as described above.
Subconfluent HT1080 cells (107) were deprived of essential amino acids for 30 min, incubated with radioactive methionine/cysteine protein labelling mix (50 μCi/ml; NEN, Boston, MA, U.S.A.) for a further 30 min and then the medium was replaced with DMEM (Dulbecco's modified Eagle's medium) and 10% foetal calf serum. At required times, cells and medium were separated, and cells were lysed using IP (immunoprecipitation) buffer (50 mM Tris/HCl, 150 mM NaCl, 2 mM EDTA, 0.5 mM PMSF and 1% Triton X-100). Insoluble material was removed by centrifugation at 10000 g for 1 min, and lysates were mixed with SDS to 1% (w/v), boiled for 3 min and diluted 10-fold with lysis buffer. Pre-incubation with Protein A–Sepharose for 30 min preceded incubation with Protein A–Sepharose and anti-Prx IV antibody for 16 h at 4 °C. Beads were washed three times with 100 vol. of lysis buffer and resuspended in SDS/PAGE sample buffer. Immunoprecipitation was repeated for whole media samples.
Immunofluorescence was performed as described previously . Anti-Prx IV antibody was detected by an Alexa Fluor® 594 anti-rabbit antibody, whereas anti-KDEL antibody was detected by an Alexa Fluor® 448 anti-mouse antibody. Cells were visualized on an Olympus BX60 upright microscope at ×40 magnification.
Creation of stable cell lines
Human Prx IV was excised from the pOTB7 vector by using BamHI and XhoI and ligated with pcDNA3.1/Hygro(+) (Invitrogen, Paisley, Renfrewshire, Scotland, U.K.). The final construct was linearized with SspI for transfection. HuSH vectors encoding shRNA (small-hairpin RNA) for Prx IV knockdown were obtained from Origene (Rockville, MD, U.S.A.) and linearized with ScaI. All constructs were transfected into subconfluent HT1080 cells using FuGENE™ 8 (Roche, Indianapolis, IN, U.S.A.). Stable transfectants were selected using 250 μg/ml hygromycin B or 1 μg/ml puromycin. After 14 days growth, colonies were selected and screened for Prx IV expression by Western blotting.
Analysis of XBP1 (X box-binding protein 1) splicing
Total RNA was isolated from cells using TRI Reagent according to the manufacturer's instructions. Following DNase treatment, cDNA was prepared from 5 μg of RNA using Bioscript reverse transcriptase (Bioline, London, U.K.). XBP1 mRNA was amplified from 1 μg of cDNA with primers corresponding to XBP1 nt 450–469 and 671–690 (kindly provided by Dr Lisa Swanton, Faculty of Life Sciences, University of Manchester, Manchester, U.K.). The resulting PCR products were digested using PstI, separated by electrophoresis through 2% agarose, stained with 1 μg/ml ethidium bromide and visualized using a UVItec UVIDoc gel documentation system.
Crystal Violet viability assay
Adherent cells were supplemented with H2O2 and incubated for 24 h. Detached cells were removed by washing with PBS and the remaining cells were fixed in methanol at –20 °C for 5 min. Nuclei were stained with 0.2% (w/v) Crystal Violet for 5 min and washed three times with PBS. The stain was solubilized in 0.1% (w/v) SDS, diluted 5-fold and attenuance recorded at λ=540 nm.
H2O2 turnover assay
Adherent cells were trypsinized and resuspended in PBS or SP cells prepared and suspended in KHM buffer (110 mM potassium acetate, 20 mM Hepes and 2 mM magnesium acetate). H2O2 was added at a 20 μM final concentration and cells removed by centrifugation at appropriate times. The remaining H2O2 concentration was determined using Amplex Red reagent (Invitrogen) in accordance with the manufacturer's instructions.
Determination of oxidoreductase redox states
Redox states were determined exactly as described previously .
Sucrose gradient fractionation
Cells (107) were lysed by vortex-mixing in SG buffer (5 mM Tris/HCl, pH 7.5, 100 mM NaCl, 1 mM PMSF and 0.1% Nonidet P40) following treatment with 25 mM NEM (N-ethylmaleimide). Cleared lysates were applied to a continuous gradient of 9 parts of 5% (w/v) sucrose into 9 parts of 25% (w/v) sucrose, laid over 2 parts of 50% (w/v) sucrose in SG buffer. Gradients were centrifuged at 40000 rev./min for 16 h at 4 °C in a Beckman SW40Ti rotor and then separated into ten equal-volume fractions. Proteins were precipitated with 5% (v/v) trichloroacetic acid and resuspended in SDS/PAGE sample buffer. For denaturing gradients, cleared lysates were supplemented with 1% (w/v) SDS and boiled for 5 min prior to fractionation. Gradients contained 0.1% SDS. For denaturing and reducing gradients, NEM was added after boiling in the presence of 1% SDS and 10 mM DTT.
Site-directed mutagenesis of Prx IV
Mutagenesis was performed by the method of Hemsley et al.  by using pcDNA3.1/hygro(+); Prx IV as a template and primers encoding alanine in place of cysteine residues. Parental DNA was removed by DpnI digestion prior to transformation into Escherichia coli. Mutant constructs were transiently expressed for 24 h in HT1080 cells following transfection of 5×106 cells with 5 μg of DNA by using Lipofectamine™ 2000 (Invitrogen).
Prx IV is co-translationally translocated to the human ER
Prx IV was initially estimated to possess a secretory signal peptide consisting of 79 amino acids , leading to mature Prx IV with a predicted molecular mass of 22 kDa. In contrast, our predictive analysis using the SignalP 3.0 program  indicated signal peptide cleavage occurred between residues 37 and 38 generating a 27 kDa product. To clarify this issue, cDNA encoding the full-length protein was translated in vitro in the presence and absence of SP HT1080 human fibrosarcoma cells (Figure 1A). In addition to the pre-protein (upper band), a second species with enhanced mobility predominated when translated in the presence of SP cells (compare lanes 1 and 2). This was consistent with co-translational translocation of Prx IV to the ER and signal peptide cleavage. The cleaved product displayed an apparent molecular mass of 27 kDa, while protection of this fragment from proteinase K digestion confirmed that translocation was complete (lane 3).
Prx IV is co-translationally translocated to the ER in vitro and in vivo.
A comparison between Prx IV translated in vitro and HT1080 whole-cell lysates clearly demonstrated that the translocated product in SP cells corresponded to the prevalent form of Prx IV in vivo (Figure 1B). Pre-Prx IV was not detected in whole cells, although a smaller immunoreactive species of 22–24 kDa was apparent (marked by *). This species was judged to be cytosolic by virtue of its co-localization with the cytosolic marker Alix following HT1080 homogenization and preparation of organelle membranes (Figure 1C, lane 1). In contrast, the predominant form of Prx IV co-localized mainly with calnexin in the organellar fraction (lane 2, representing the membranes of the secretory pathway) and, like the calnexin luminal domain, was again protected from proteinase K digestion (lane 3). Thus human Prx IV exists primarily as an ER-translocated peptide of approx. 27 kDa.
The most probable explanation for the presence of the cytosolic species is that it is an artefact of antibody cross-reactivity during Western blotting. This is based on the observation that its levels remained unaffected in cells stably overexpressing Prx IV and also in cell lines with up to 90% stable knockdown of Prx IV expression (see Figure 3A). Furthermore, unlike ER-translocated Prx IV, this species could not be immunoprecipitated from cell lysate (Figure 2A). Nonetheless, the possibility remains that it may constitute some hitherto unknown Prx IV splice variant.
Prx IV is not secreted and remains ER-associated
Prx IV is retained in the ER
To evaluate secretion of Prx IV into the extracellular environment, pulse–chase time courses were performed for HT1080 cells (results not shown) and HT1080 cells transfected to overexpress Prx IV (Figure 2A). Transfection was stable rather than transient, ensuring Prx IV expression was consistent and physiologically tolerable. In each case, the labelled pool of Prx IV was immunoprecipitated exclusively from cell lysates throughout the following 5 h (lanes 1–3), with none detected in culture supernatant samples (lanes 4–6). In addition, Western-blot analyses of protein precipitates from HT1080, HeLa, HEK-293 cells (human embryonic kidney cells) and HepG2 culture supernatants showed no detectable secretion of Prx IV (results not shown).
As Prx IV clearly remains associated with cells, an experiment was conducted to determine its final cellular localization. Endogenous Prx IV was visualized by immunofluorescence microscopy of HT1080 cells following cycloheximide treatment to inhibit de novo protein synthesis (Figure 2B). Interestingly, Prx IV remained co-localized with ER markers for a full 5 h after cycloheximide addition, mirroring the distribution seen in untreated cells and in cells overexpressing Prx IV (results not shown). The results indicate that native Prx IV is ER-retained in human cells even when overproduced, despite lacking a recognized retention/retrieval motif.
Altering Prx IV expression does not induce ER stress
To investigate the function of Prx IV in the ER, cell lines were created in which Prx IV expression was stably modified (Figure 3A). In addition to the aforementioned cell line overexpressing Prx IV (lane 2), two HT1080 cell lines were established displaying stable expression of shRNA directed against the Prx IV coding sequence (lanes 3 and 4). Densitometry performed for multiple Western blot exposures, within the linear range of chemiluminescent substrate response, allowed quantification of Prx IV knockdowns as indicated (Figure 3A).
Stable alteration of Prx IV expression does not induce ER stress
To determine whether modulated Prx IV expression resulted in ER stress induction, components of the UPR (unfolded protein response) were examined in each of the cell lines created. XBP1 is a transcription factor involved in the expression of UPR-induced genes following accumulation of protein aggregates within the ER . Production of active XBP1 requires IRE1-dependent mRNA splicing, which in turn eliminates a PstI restriction enzyme recognition site from the mRNA sequence. XBP1 splicing was induced in HT1080 cells by incubation with DTT (Figure 2B, lane 2). For each of our cell lines however, the status of XBP1 mRNA reflected that seen for the HT1080 parent cells (compare lanes 3–5 and lane 1), indicating no induction of UPR. This was confirmed by Western blotting of whole-cell lysates (Figure 2C) using antibodies to BiP, an ER chaperone up-regulated during UPR [37,38].
In addition to investigating the influence of Prx IV expression on the UPR, the levels of Prx IV itself were evaluated under conditions of UPR induction. Prx IV expression in HT1080 cells was unaffected by a 12 h incubation with DTT (2 mM), tunicamycin (10 μg/ml) or thapsigargin (2 μM), indicating that it does not respond to ER stress (results not shown).
Prx IV knockdown does not alter ER redox balance
We hypothesized that if Prx IV has important peroxidase activity, knockdown of Prx IV expression would limit the ability of cells to remove H2O2 from the ER and increase susceptibility of cells to H2O2-induced cell death. To test the latter, our Prx IV knockdown cell lines were treated with increasing concentrations of H2O2 and viability was recorded after a 24 h incubation period (Figure 4A). Compared with the HT1080 parent, shRNA2 cells displayed reduced survival at all H2O2 concentrations tested (compare black and grey bars). This differed from the response of the shRNA1 cell line, which, while demonstrating some attenuation of viability at 1–2 mM H2O2, matched that of the parent at higher concentrations (black bars compared with white bars). Disparity in response of the knockdown cell lines may possibly reflect the respective levels at which Prx IV is expressed in each.
Prx IV knockdown affects viability but not H2O2 turnover or ER redox balance
To establish whether the attenuated viability could be a manifestation of reduced peroxidase activity, Prx IV knockdown cell lines were evaluated for their ability to decompose exogenously supplied H2O2. Surprisingly, both shRNA1 and shRNA2 cell lines were able to remove H2O2 at least as efficiently as unmodified HT1080 cells (Figure 4B). Similar results were obtained using SP cells (lacking any cytosol) in place of intact cells (Figure 4C), confirming that the results observed were not due to cytosolic factors masking any underlying ER activity.
As we could detect no decrease in peroxide turnover following Prx IV knockdown, an approach was taken to examine more general effects on the ER redox environment. Several ER-resident oxidoreductases are known to exist in a reduced steady state in HT1080 cells, the maintenance of which can be influenced by altering redox conditions . We therefore postulated that if Prx IV was a major ER peroxidase, substantially reducing its expression could result in a more oxidizing environment within the ER. As markers, we investigated the redox states of PDI, ERp57 and ERp72 in the Prx IV-knockdown cell lines. In both shRNA1 (results not shown) and shRNA2 cells (Figure 4D), each oxidoreductase was in a reduced state exactly mirroring that of the HT1080 control cells (compare lanes 3 and 6). The redox states of these enzymes therefore appear to be unaffected by reduced Prx IV expression, indicating no gross change in ER redox homoeostasis.
Prx IV forms oligomeric complexes containing non-catalytic disulfide bonds in vivo
As mentioned previously, human Prx I and Prx II display distinct oligomeric behaviours that potentially influence function . While each forms a pentadimeric decamer via hydrophobic interactions, Prx I decamers may be stabilized by additional intermolecular disulfide bonds at the dimer–dimer interface. Relative to Prx II, Prx I displays reduced peroxidase activity and increased propensity to inhibit thermal aggregation of malate dehydrogenase in vitro . Replacement of the Cys83 residue with alanine however, preventing formation of the interdimer disulfide, resulted in elevated and diminished Prx I peroxidase and chaperone activities respectively. Thus increased oligomeric stability potentially diverts typical 2-Cys peroxiredoxins from operating principally as peroxidases. As we observed no detectable peroxidase activity attributable to Prx IV in the preceding experiments, we investigated whether there could be a structural basis for these findings similar to those responsible for disparities between Prx I and Prx II.
A crystal structure for a Prx IV fragment comprising amino acid residues 84–271 has recently been submitted to the RSCB PDB (Research Collaboratory for Structural Bioinformatics Protein Data Bank) (PDB code: 2PN8). The structure confirmed that human Prx IV purified from E. coli forms decamers characteristic of typical 2-Cys peroxiredoxins in vitro. To establish whether this occurs in vivo, HT1080 cells overexpressing Prx IV were alkylated to preserve protein thiol-disulfide status. Cell lysates were subsequently fractionated through a sucrose gradient to separate native protein complexes principally on the basis of molecular mass. Prx IV predominantly sedimented towards the high-sucrose end of the gradient (Figure 5A, top two panels, lanes 4 and 5), as opposed to the fractionation patterns obtained for a mixture of dimeric and monomeric Prx IV (denaturing gradient), or monomeric Prx IV alone (denaturing and reducing gradient). Comparable results were obtained for endogenous Prx IV in HT1080 cells under reducing conditions, although expression levels limited non-reducing examination (results not shown). These results indicate that Prx IV does exist in an oligomeric state in human cells.
Prx IV forms oligomeric complexes in the ER
In addition to whole-cell extracts, protease-treated organelle membranes were subjected to sucrose gradient analyses confirming that Prx IV oligomeric complexes formed within the ER (Figure 5B). Moreover, incubation of membranes with H2O2 at concentrations substantially greater than those previously determined to cause ‘stacking’ of Prx II decamers , led to negligible effects on the fractionation profile for Prx IV.
Prx IV oligomers contain two prevalent species with apparent molecular masses of 27 and 54 kDa (Figure 5A, top panel), corresponding to Prx IV monomers and disulfide-linked homodimers respectively. Intriguingly, mutation of the catalytic cysteine residues to alanine did not abolish the formation of Prx IV dimers (Figure 6B, lanes 2–4), indicating that additional intermolecular disulfide bonds could form independently of the active sites.
Cys51 is required for forming a non-catalytic disulfide in Prx IV
Prx IV contains two non-catalytic cysteine residues at positions 51 and 148 in the primary structure (Figure 6A). To evaluate the involvement of each in disulfide bond formation, cysteine to alanine mutants were created for Cys51 and Cys148 and expressed in HT1080 cells (Figure 6B, lanes 5–8). Mutation of Cys148, individually and in combination with the peroxidatic and resolving cysteine residues, had no detectable effect on Prx IV dimerization (lanes 7–8). The same was true of a Cys51 mutant in which active-site disulfides were still intact (lane 5).
However, mutation of Cys51 in conjunction with the catalytic residues led to total depletion of covalently linked dimers. We therefore conclude that Cys51 facilitates formation of an intermolecular disulfide bond independently of the Prx IV active-site cysteine residues.
The finding that Prx IV resides within the human ER clarifies an issue that has remained clouded for the last 10 years. Our observation of a second anti-Prx IV immunoreactive species within the cytosol may help to explain previous disagreement regarding the cellular localization of Prx IV. Similarly, the detection of this product within human cell lysates may shed some light on the original prediction of such a large signal peptide . Previously, it has been suggested that the larger form of Prx IV detected in cells could be unprocessed pre-protein, anchored to membranes via the uncleaved hydrophobic sequence . However, in the present study, we clearly show that pre-Prx IV is not present at any substantial level in human cells and that the larger, predominant form is mature ER-localized Prx IV. Furthermore, while the implied Prx IV signal peptide may seem unusually long (37 residues), no obvious anomalies were visible within the amino acid sequence. All positions within the first 34 residues scored poorly as potential cleavage sites during SignalP 3.0 analysis, with the only potential alternative presented as a possible cleavage site existing between positions 34 and 35.
The mechanism by which Prx IV is retained within the ER remains undetermined as no KDEL-type retrieval motif is present within the primary structure. This situation is not without precedent however. Human Ero1 proteins lack an ER-retrieval motif, yet they remain within the ER. Recently, this has been indicated to occur through competitive interactions in the ER lumen with both PDI (containing a KDEL motif) and ERp44 (RDEL) in a largely thiol-dependent fashion .
Prx IV may also have a requirement for thiol-mediated interaction with ER proteins. Recycling of the peroxidatic cysteine, by reduction of the resolving disulfide, can occur in the presence of both thioredoxin and glutathione in vitro . Whereas glutathione is found within the ER, thioredoxin is cytosolic. Numerous thioredoxin-like proteins exist within the ER however, including the PDI family members. It is therefore possible that Prx IV may be retained in the ER via interactions akin to those observed for the Ero1 family. If so, identification of partners involved in redox turnover of Prx IV may in turn provide insights into the maintenance of its ER localization.
In conjunction with the established literature regarding typical 2-Cys peroxiredoxin behaviour in vitro, the results presented here provide a new perspective for speculation of Prx IV function in vivo. Based on the available crystal data both for Prx IV and other closely related proteins [11,12,41,42], it seems highly likely that the oligomeric complexes observed in the ER of HT1080 cells correspond to toroid Prx IV decamers. Furthermore, it is clear that Prx IV intermolecular interactions can be stabilized by disulfide bonds formed independently of the peroxidatic and resolving cysteine residues. The missing piece to this puzzle remains the position of Cys51-mediated disulfides within the Prx IV quaternary structure. The current crystal structure for Prx IV provides no clarification on this matter, as Cys51 is absent from the crystallized fragment. Moreover, we cannot assume that these bonds are analogous to those at the Prx I dimer–dimer interface, as sequence alignments indicate that the Cys83 residue of Prx I does not correspond to Cys51 of Prx IV . Comparisons instead suggest that Cys148 of Prx IV is the residue that most closely matches Cys83 of Prx I. In the present study however, we have demonstrated that Cys148 is not required for disulfide bonding between Prx IV monomers. Consequently, we cannot currently discount the possibility that Cys51 disulfide bonds exist intradimerically, strengthening this particular interaction of Prx IV.
Irrespective of whether non-catalytic disulfides form between or within Prx IV dimers, it is clear that either situation may interfere with oligomeric transitions and therefore the peroxidatic cycle of Prx IV. Fluid conversion from dimeric>decameric>covalent-dimeric states are important for the efficient peroxidase activity of peroxiredoxins such as Prx II. This activity can clearly be manipulated by modulation of Prx I disulfide bonding capability at residue 83, although given the cytosolic/nuclear localization attributed to Prx I, it remains uncertain how prevalent this stabilizing disulfide may be in vivo. Prx IV, however, exists in an environment in which formation of disulfide bonds is positively encouraged. We would therefore predict the arrangement of Prx IV into disulfide-stabilized structures to be highly favoured, reflected in our inability to detect any significant peroxidase effects associated with Prx IV expression.
Despite no obvious defect in peroxidase activity, knockdown of Prx IV expression clearly caused some compromise in survival of HT1080 cells following H2O2 exposure. However, the differential responses of the two cell lines were not wholly consistent with a dose-dependent increase in H2O2 susceptibility relative to Prx IV's expression level. While Prx IV may not contribute significantly towards H2O2 elimination, the effects witnessed following Prx IV knockdown may be attributable to its presence helping to alleviate the after-effects of oxidative insult. As touched on previously, an alternative role already prescribed to typical 2-Cys peroxiredoxins is that of a molecular chaperone. Cytoplasmic peroxiredoxins from both yeast and humans have been demonstrated to inhibit thermal aggregation of substrates in vitro and to enhance heat-shock resistance in vivo [18,19,43]. Yeast TSA1 (thiol-specific antioxidant protein 1) has also been shown to prevent aggregation of ribosomal proteins following reductive stress . The prospect of Prx IV displaying such an activity within the mammalian ER is indeed attractive, particularly given that protein folding within the compartment is coupled with additional complexities such as introduction of native disulfide bonds. The possibility still remains, however, that Prx IV function may revolve around its active-site residues. While considered primarily as a peroxidatic mechanism, the dynamic redox status of peroxiredoxin active sites may necessitate thiol-dependent interaction with other proteins. Peroxiredoxins may therefore be considered as oxidoreductases of said proteins in their own right. Consequently, identification of interacting partners may provide the key to determine Prx IV function, and may yet indicate a role for Prx IV in thiol–disulfide exchange processes within the human ER.
This work was supported by grants from the Wellcome Trust (no. 74081) and BBSRC (Biotechnology and Biological Sciences Research Council) (no. D00769). In addition, we acknowledge the generosity of Professor Philip Woodman, Dr Lisa Swanton, Professor Keith Gull and Professor Stephen Fuller in providing reagents and antibodies as detailed in the text.
ALG2 (asparagine-linked glycosylation 2 homologue)-interacting protein X
immunoglobulin heavy-chain-binding protein
endoplasmic reticulum oxidoreductin 1
nuclear factor κB
reactive oxygen species
- SP cell
Tris-buffered saline with Tween 20
unfolded protein response
X box-binding protein 1