Evaluation of the impact of antibody fragments on aggregation of intact molecules via size exclusion chromatography coupled with native mass spectrometry.

Aggregates are recognized as one of the most critical product-related impurities in monoclonal antibody (mAb)-based therapeutics due to their negative impact on the stability and safety of the drugs. So far, investigational efforts have primarily focused on understanding the causes and effects of mAb self-aggregation, including both internal and external factors. In this study, we focused on understanding mAb stability in the presence of its monovalent fragment, formed through hinge cleavage and loss of one Fab unit (referred to as "Fab/c"), a commonly observed impurity during manufacturing and stability. The Fab/c fragments were generated using a limited IgdE digestion that specifically cleaves above the IgG1 mAb hinge region, followed by hydrophobic interaction chromatographic (HIC) enrichment. Two IgG1 mAbs containing different levels of Fab/c fragments were incubated under thermally accelerated conditions. A method based on size exclusion chromatography coupled with native mass spectrometry (SEC-UV-native MS) was developed and used to characterize the stability samples and identified the formation of heterogeneous dimers, including intact dimer, mAb-Fab/c dimer, Fab/c-Fab/c dimer, and mAb-Fab dimer. Quantitative analyses on the aggregation kinetics suggested that the impact of Fab/c fragment on the aggregation rate of individual dimer differs between a glycosylated mAb (mAb1) and a non-glycosylated mAb (mAb2). An additional study of deglycosylated mAb1 under 25°C accelerated stability conditions suggests no significant impact of the N-glycan on mAb1 total aggregation rate. This study also highlighted the power of SEC-UV-native MS method in the characterization of mAb samples with regard to separating, identifying, and quantifying mAb aggregates and fragments.

A Highly Sensitive LC-MS/MS Method for Targeted Quantitation of Lipase Host Cell Proteins in Biotherapeutics.

Identification and accurate quantitation of host cell proteins (HCPs) in biotherapeutic drugs has become increasingly important due to the negative impact of certain HCPs on the safety, stability, and other product quality of biotherapeutics. Recently, several lipase HCPs have been identified to potentially cause the enzymatic degradation of polysorbate, a widely used excipient in the formulation of biotherapeutics, which can severely impact the stability and product quality of drug products. In this study, we identified three lipase HCPs that were frequently detected in Chinese hamster ovary (CHO) cell cultures using shotgun proteomics, including phospholipase B-like 2 (PLBL2), lipoprotein lipase (LPL), and lysosomal acid lipase (LIPA). A targeted quantitation method for these three lipase HCPs was developed utilizing liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) with high-resolution multiple-reaction-monitoring (MRMhr) quantitation. The method demonstrated good sensitivity with low limit of quantitation (LLOQ) around 1 ng/mL, and linear dynamic range of three orders of magnitude for the three lipase HCPs. It has been applied for the characterization of process intermediates from various in-house monoclonal antibody (mAb) production. In addition, the method has also been used to evaluate the robustness of clearance for one of the lipase HCPs, PLBL2, under different column purification process conditions.

Benchmarking and optimization of a high-throughput sequencing based method for transgene sequence variant analysis in biotherapeutic cell line development.

In recent years, High-Throughput Sequencing (HTS) based methods to detect mutations in biotherapeutic transgene products have become a key quality step deployed during the development of manufacturing cell line clones. Previously we reported on a higher throughput, rapid mutation detection method based on amplicon sequencing (targeting transgene RNA) and detailed its implementation to facilitate cell line clone selection. By gaining experience with our assay in a diverse set of cell line development programs, we improved the computational analysis as well as experimental protocols. Here we report on these improvements as well as on a comprehensive benchmarking of our assay. We evaluated assay performance by mixing amplicon samples of a verified mutated antibody clone with a non-mutated antibody clone to generate spike-in mutations from ∼60% down to ∼0.3% frequencies. We subsequently tested the effect of 16 different sample and HTS library preparation protocols on the assay's ability to quantify mutations and on the occurrence of false-positive background error mutations (artifacts). Our evaluation confirmed assay robustness, established a high confidence limit of detection of ∼0.6%, and identified protocols that reduce error levels thereby significantly reducing a source of false positives that bottlenecked the identification of low-level true mutations.

Sequence Variant and Posttranslational Modification Analysis During Cell Line Selection via High-Throughput Peptide Mapping.

Selection of high-producing lead and backup cell lines with high-fidelity primary structure is a major goal of cell line development of protein therapeutics. Conventional techniques for sequence variant analysis, such as mass spectrometry (MS) and next-generation sequencing (NGS) have limitations on the sample number and turnaround time, thus often are only applied at the final stages of development, where an undesired lead or backup clone could cause a significant delay in project timeline. Here we presented a high-throughput (HT) peptide mapping workflow which can be applied at early stages of cell line selection for testing of a batch of 30-40 clones within 2-week turnaround while reporting valuable information on sequence variants and posttranslational modifications (PTMs). The successful application of this workflow was demonstrated for two mAb programs. Multiple clones were removed from a total of 33 mAb-1 clones using various criteria: nine clones contained at least one >1% upregulated unknown peptide ions, 11 clones contained at least eight >0.1% upregulated unknowns, and six clones contained upregulated critical PTMs. For mAb-2, light chain (LC) sequence extension of approximately 30 amino acids were detected in 6 out of 36 clones at levels up to 11%. Besides, a Q to H mutation at ~30% was detected in the heavy chain (HC) of a single clone. Q to H mutation has mass change of 9.00 Da and failed to be detected by intact mass analysis. Rapid PTM quantitation also facilitated the selection of clones with desirable quality attributes, such as N-glycan profile. Hence, we demonstrated a risk-reducing strategy where abnormal clones could be detected at earlier stages of cell line selection, which should result in reduced and more predictable timeline of cell line development.

Application of a label-free and domain-specific free thiol method in monoclonal antibody characterization.

Characterization of free thiol variants in antibody therapeutics is important for biopharmaceutical development, as the presence of free thiols may have an impact on aggregate formation, structural and thermal stability, as well as antigen-binding potency of monoclonal antibodies. Most current methods for free thiol quantification involve labeling of free thiol groups by different tagging molecules followed by UV, fluorescence or mass spectrometry (MS) detection. Here, we optimized a label-free liquid chromatography (LC)-UV/MS method for free thiol quantification at a subunit level and compared this method with two orthogonal and conventional approaches, Ellman's assay and peptide mapping with differential alkylation. This subunit unit approach was demonstrated to be able to provide domain-specific free thiol quantification and comparable results with labeling approaches, using a relatively simple and efficient workflow.

Discovery and Characterization of Histidine Oxidation Initiated Cross-links in an IgG1 Monoclonal Antibody.

Novel cross-links between an oxidized histidine and intact histidine, lysine, or cysteine residues were discovered and characterized from high-molecular weight (HMW) fractions of an IgG1 monoclonal antibody (mAb). The mAb HMW fractions were collected using preparative size-exclusion chromatography (SEC) and extensively characterized to understand the mechanism of formation of the nonreducible and covalently linked portion of the HMWs. The HMW fractions were IdeS digested, reduced, and analyzed by size-exclusion chromatography coupled with mass spectrometry (SEC-MS). The nonreducible cross-links were found to be enriched in the fragment crystallizable (Fc) region of the heavy chain, with a net mass increase of 14 Da. Detailed peptide mapping revealed as many as seven covalent cross-links in the HMW fractions, where oxidized histidines react with intact histidine, lysine, and free cysteine to form cross-links. It is the first time that histidine-cysteine (His-Cys) and histidine-lysine (His-Lys) in addition to histidine-histidine (His-His) cross-links were discovered in monoclonal antibody HMW species. The histidine oxidation hot spots were identified, which include conserved histidine residues His292 and His440 in the Fc region and His231 in the hinge region of the IgG1 mAb heavy chain. Their cross-linking partners include His231, His292, His440, and Cys233 in the hinge region and Lys297 in the Fc region. A cross-linking mechanism has been proposed that involves nucleophilic addition by histidine, cysteine, or lysine residues to the carbonyl-containing histidine oxidation intermediates to form the cross-links.

A supporting role of Chinese National Immortalized Cell Bank in life science research.

A biorepository of human samples is essential to support the research of life science. Lymphoblastoid B cell line (LCL), which is easy to be prepared and can reproduce indefinitely, is a convenient form of sample preservation. LCLs are established from human B cells transformed by Epstein-Barr virus (EBV). Chinese National Immortalized Cell Bank has preserved human LCLs from different ethnic groups in China. As there are many studies on the nature of LCLs and public available resources with genome-wide data for LCLs, they have been widely applied in genetics, immunology, pharmacogenetics/genomics, regenerative medicine, cancer pathogenesis and immunotherapy, screening and generation of fully human neutralizing monoclonal antibodies and study on EBV pathogenesis. Here, we review the characteristics of LCLs and their contributions to scientific research, and introduce preserved samples in Chinese National Immortalized Cell Bank to the scientific community. We hope this bank can support more areas in the scientific research.

Size-exclusion chromatography-mass spectrometry with m-nitrobenzyl alcohol as post-column additive for direct characterization of size variants of monoclonal antibodies.

Size-exclusion chromatography (SEC) is commonly used to monitor low molecular weight fragments and aggregates present in recombinant monoclonal antibody (mAb) biopharmaceuticals. It has been previously demonstrated that SEC could be coupled with mass spectrometry (MS) to directly measure the molecular weights of these protein species to aid in their identification. However, the use of certain mobile phase modifiers led to compromised sensitivity in MS detection. In this work, we demonstrate that the use of m-nitrobenzyl alcohol (m-NBA) as a post-column additive in an SEC-MS method is able to improve the ionization of antibody light chain and heavy chain approximately 7-fold and 2-fold, respectively, and thus allows the MS detection of low-abundance size variants present in mAb biopharmaceuticals. Application of the 15-min reducing SEC-UV/MS method enabled the direct identification of size variants present in an IgG1 mAb sample. One high molecular weight species observed under reducing conditions was identified to be a thioether-linked heterodimer of light chain and heavy chain. Multiple lower molecular weight species were found to result from cleavage of the heavy chain at a number of sites throughout the conserved sequence. The reducing SEC-UV/MS method provides a straightforward approach for identification of size variants present in mAb and may be applicable generally to all types of mAb biopharmaceuticals.

Brain-derived neurotrophic factor signaling rewrites the glucocorticoid transcriptome via glucocorticoid receptor phosphorylation.

Abnormal glucocorticoid and neurotrophin signaling has been implicated in numerous psychiatric disorders. However, the impact of neurotrophic signaling on glucocorticoid receptor (GR)-dependent gene expression is not understood. We therefore examined the impact of brain-derived neurotrophic factor (BDNF) signaling on GR transcriptional regulatory function by gene expression profiling in primary rat cortical neurons stimulated with the selective GR agonist dexamethasone (Dex) and BDNF, alone or in combination. Simultaneous treatment with BDNF and Dex elicited a unique set of GR-responsive genes associated with neuronal growth and differentiation and also enhanced the induction of a large number of Dex-sensitive genes. BDNF via its receptor TrkB enhanced the transcriptional activity of a synthetic GR reporter, suggesting a direct effect of BDNF signaling on GR function. Indeed, BDNF treatment induces the phosphorylation of GR at serine 155 (S155) and serine 287 (S287). Expression of a nonphosphorylatable mutant (GR S155A/S287A) impaired the induction of a subset of BDNF- and Dex-regulated genes. Mechanistically, BDNF-induced GR phosphorylation increased GR occupancy and cofactor recruitment at the promoter of a BDNF-enhanced gene. GR phosphorylation in vivo is sensitive to changes in the levels of BDNF and TrkB as well as stress. Therefore, BDNF signaling specifies and amplifies the GR transcriptome through a coordinated GR phosphorylation-dependent detection mechanism.

The pseudokinase domain of JAK2 is a dual-specificity protein kinase that negatively regulates cytokine signaling.

Human JAK2 tyrosine kinase mediates signaling through numerous cytokine receptors. The JAK2 JH2 domain functions as a negative regulator and is presumed to be a catalytically inactive pseudokinase, but the mechanism(s) for its inhibition of JAK2 remains unknown. Mutations in JH2 lead to increased JAK2 activity, contributing to myeloproliferative neoplasms (MPNs). Here we show that JH2 is a dual-specificity protein kinase that phosphorylates two negative regulatory sites in JAK2: Ser523 and Tyr570. Inactivation of JH2 catalytic activity increased JAK2 basal activity and downstream signaling. Notably, different MPN mutations abrogated JH2 activity in cells, and in MPN (V617F) patient cells phosphorylation of Tyr570 was reduced, suggesting that loss of JH2 activity contributes to the pathogenesis of MPNs. These results identify the catalytic activity of JH2 as a previously unrecognized mechanism to control basal activity and signaling of JAK2.

A novel transcription complex that selectively modulates apoptosis of breast cancer cells through regulation of FASTKD2.

We previously reported that expression of NRIF3 (nuclear receptor interacting factor-3) rapidly and selectively leads to apoptosis of breast cancer cells. DIF-1 (also known as interferon regulatory factor-2 binding protein 2 [IRF-2BP2]), the cellular target of NRIF3, was identified as a transcriptional repressor, and DIF-1 knockdown leads to apoptosis of breast cancer cells but not other cell types. Here, we identify IRF-2BP1 and EAP1 (enhanced at puberty 1) as important components of the DIF-1 complex mediating both complex stability and transcriptional repression. This interaction of DIF-1, IRF-2BP1, and EAP1 occurs through the conserved C4 zinc fingers of these proteins. Microarray studies were carried out in breast cancer cell lines engineered to conditionally and rapidly increase the levels of the death domain (DD1) region of NRIF3. The DIF-1 complex was found to repress FASTKD2, a putative proapoptotic gene, in breast cancer cells and to bind to the FASTKD2 gene by chromatin immunoprecipitation. FASTKD2 knockdown prevents apoptosis of breast cancer cells from NRIF3 expression or DIF-1 knockdown, while expression of FASTKD2 leads to apoptosis of both breast and nonbreast cancer cells. Thus, regulation of FASTKD2 by NRIF3 and the DIF-1 complex acts as a novel death switch that selectively modulates apoptosis in breast cancer.

Phosphorylation of the PRC2 component Ezh2 is cell cycle-regulated and up-regulates its binding to ncRNA.

Ezh2 functions as a histone H3 Lys 27 (H3K27) methyltransferase when comprising the Polycomb-Repressive Complex 2 (PRC2). Trimethylation of H3K27 (H3K27me3) correlates with transcriptionally repressed chromatin. The means by which PRC2 targets specific chromatin regions is currently unclear, but noncoding RNAs (ncRNAs) have been shown to interact with PRC2 and may facilitate its recruitment to some target genes. Here we show that Ezh2 interacts with HOTAIR and Xist. Ezh2 is phosphorylated by cyclin-dependent kinase 1 (CDK1) at threonine residues 345 and 487 in a cell cycle-dependent manner. A phospho-mimic at residue 345 increased HOTAIR ncRNA binding to Ezh2, while the phospho-mimic at residue 487 was ineffectual. An Ezh2 domain comprising T345 was found to be important for binding to HOTAIR and the 5' end of Xist.

Dok-7 regulates neuromuscular synapse formation by recruiting Crk and Crk-L.

Agrin, released by motor neurons, promotes neuromuscular synapse formation by stimulating MuSK, a receptor tyrosine kinase expressed in skeletal muscle. Phosphorylated MuSK recruits docking protein-7 (Dok-7), an adaptor protein that is expressed selectively in muscle. In the absence of Dok-7, neuromuscular synapses fail to form, and mutations that impair Dok-7 are a major cause of congenital myasthenia in humans. How Dok-7 stimulates synaptic differentiation is poorly understood. Once recruited to MuSK, Dok-7 directly stimulates MuSK kinase activity. This unusual activity of an adapter protein is mediated by the N-terminal region of Dok-7, whereas most mutations that cause congenital myasthenia truncate the C-terminal domain. Here, we demonstrate that Dok-7 also functions downstream from MuSK, and we identify the proteins that are recruited to the C-terminal domain of Dok-7. We show that Agrin stimulates phosphorylation of two tyrosine residues in the C-terminal domain of Dok-7, which leads to recruitment of two adapter proteins: Crk and Crk-L. Furthermore, we show that selective inactivation of Crk and Crk-L in skeletal muscle leads to severe defects in neuromuscular synapses in vivo, revealing a critical role for Crk and Crk-L downstream from Dok-7 in presynaptic and postsynaptic differentiation.

Myristoylation of the dual-specificity phosphatase c-JUN N-terminal kinase (JNK) stimulatory phosphatase 1 is necessary for its activation of JNK signaling and apoptosis.

Activation of the c-JUN N-terminal kinase (JNK) pathway is implicated in a number of important physiological processes, from embryonic morphogenesis to cell survival and apoptosis. JNK stimulatory phosphatase 1 (JSP1) is a member of the dual-specificity phosphatase subfamily of protein tyrosine phosphatases. In contrast to other dual-specificity phosphatases that catalyze the inactivation of mitogen-activated protein kinases, expression of JSP1 activates JNK-mediated signaling. JSP1 and its relative DUSP15 are unique among members of the protein tyrosine phosphatase family in that they contain a potential myristoylation site at the N-terminus (MGNGMXK). In this study, we investigated whether JSP1 was myristoylated and examined the functional consequences of myristoylation. Using mass spectrometry, we showed that wild-type JSP1, but not a JSP1 mutant in which Gly2 was mutated to Ala (JSP1-G2A), was myristoylated in cells. Although JSP1 maintained intrinsic phosphatase activity in the absence of myristoylation, the subcellular localization of the enzyme was altered. Compared with the wild type, the ability of nonmyristoylated JSP1 to induce JNK activation and phosphorylation of the transcription factor c-JUN was attenuated. Upon expression of wild-type JSP1, a subpopulation of cells, with the highest levels of the phosphatase, was induced to float off the dish and undergo apoptosis. In contrast, cells expressing similar levels of JSP1-G2A remained attached, further highlighting that the myristoylation mutant was functionally compromised.

The target of the NSD family of histone lysine methyltransferases depends on the nature of the substrate.

The NSD (nuclear receptor SET domain-containing) family of histone lysine methyltransferases is a critical participant in chromatin integrity as evidenced by the number of human diseases associated with the aberrant expression of its family members. Yet, the specific targets of these enzymes are not clear, with marked discrepancies being reported in the literature. We demonstrate that NSD2 can exhibit disparate target preferences based on the nature of the substrate provided. The NSD2 complex purified from human cells and recombinant NSD2 both exhibit specific targeting of histone H3 lysine 36 (H3K36) when provided with nucleosome substrates, but histone H4 lysine 44 is the primary target in the case of octamer substrates, irrespective of the histones being native or recombinant. This disparity is negated when NSD2 is presented with octamer targets in conjunction with short single- or double-stranded DNA. Although the octamers cannot form nucleosomes, the target is nonetheless nucleosome-specific as is the product, dimethylated H3K36. This study clarifies in part the previous discrepancies reported with respect to NSD targets. We propose that DNA acts as an allosteric effector of NSD2 such that H3K36 becomes the preferred target.

Isoflurane inhibits cyclic adenosine monophosphate response element-binding protein phosphorylation and calmodulin translocation to the nucleus of SH-SY5Y cells.

BACKGROUND: Calmodulin (CaM) activation by Ca(2+), its translocation to the nucleus, and stimulation of phosphorylation of cyclic adenosine monophosphate response element-binding protein (CREB) (P-CREB) are necessary for new gene expression and have been linked to long-term potentiation, a process important in memory formation. Because isoflurane affects memory, we tested whether isoflurane interfered with the translocation of CaM to the neuronal cell nucleus and attenuated the formation P-CREB. METHODS: SH-SY5Y cells, a human neuroblastoma cell line, were cultured. Cells were depolarized with KCl and the phosphorylation of CREB examined by Western blotting, enzyme-linked immunosorbant assay, and immunocytochemistry. The translocation of CaM from the cytosol to the nucleus was also examined after depolarization. Cells were depolarized and lysed and fractionated by centrifugation to determine the amount of CaM translocated to the nucleus. CaM was localized by immunocytochemistry and quantitated by Western blotting and imaging. Before and during KCl depolarization, cells were exposed to isoflurane, isoflurane plus Bay K 8644, nitrendipine, and omega-conotoxin GVIa, respectively. RESULTS: P-CREB increased after KCl depolarization. The increase of P-CREB peaked at depolarization duration of 30 s. The increase in P-CREB formation was inhibited by nitrendipine, but not omega-conotoxin, and by isoflurane in a concentration-dependent fashion. Pretreatment with the L-type Ca(2+) channel agonist, Bay K 8644, attenuated the inhibition of P-CREB formation by isoflurane. CaM presence in the nucleus occurred after KCl depolarization. CaM translocation was inhibited by nitrendipine and attenuated by isoflurane. Bay K 8644 pretreatment decreased the isoflurane inhibition of CaM translocation to the nucleus. CONCLUSIONS: Our data demonstrate that isoflurane inhibits CaM translocation and P-CREB formation. This most likely occurs through isoflurane inhibition of Ca(2+)entry through L-type Ca(2+) channels.

Proximal glycans outside of the epitopes regulate the presentation of HIV-1 envelope gp120 helper epitopes.

Glycosylation of HIV-1 envelope gp120 determines not only the proper structure, but also the immune responses against this Ag. Although glycans may be part of specific epitopes or shield other epitopes from T cells and Abs, this study provides evidence for a different immunomodulatory function of glycans associated with gp120 residues N230 and N448. These glycans are required for efficient MHC class II-restricted presentation of nearby CD4 T cell epitopes, even though they are not part of the epitopes. The glycans do not affect CD4 T cell recognition of more distant epitopes and are not essential for the proper folding and function of gp120. Data on CD4 T cell recognition of N448 mutants combined with proteolysis analyses and surface electrostatic potential calculation around residue N448 support the notion that N448 glycan near the epitope's C terminus renders the site to be surface accessible and allows its efficient processing. In contrast, the N230 glycan contributes to the nearby epitope presentation at a step other than the proteolytic processing of the epitope. Hence, N-glycans can determine CD4 T cell recognition of nearby gp120 epitopes by regulating the different steps in the MHC class II processing and presentation pathway after APCs acquire the intact gp120 Ag exogenously. Modifications of amino acids bearing glycans at the C termini of gp120 helper epitopes may prove to be a useful strategy for enhancing the immunogenicity of HIV-1 envelope gp120.

Crystal structure of a fibroblast growth factor homologous factor (FHF) defines a conserved surface on FHFs for binding and modulation of voltage-gated sodium channels.

Voltage-gated sodium channels (Nav) produce sodium currents that underlie the initiation and propagation of action potentials in nerve and muscle cells. Fibroblast growth factor homologous factors (FHFs) bind to the intracellular C-terminal region of the Nav alpha subunit to modulate fast inactivation of the channel. In this study we solved the crystal structure of a 149-residue-long fragment of human FHF2A which unveils the structural features of the homology core domain of all 10 human FHF isoforms. Through analysis of crystal packing contacts and site-directed mutagenesis experiments we identified a conserved surface on the FHF core domain that mediates channel binding in vitro and in vivo. Mutations at this channel binding surface impaired the ability of FHFs to co-localize with Navs at the axon initial segment of hippocampal neurons. The mutations also disabled FHF modulation of voltage-dependent fast inactivation of sodium channels in neuronal cells. Based on our data, we propose that FHFs constitute auxiliary subunits for Navs.

Thioredoxin-related Protein 32 is an arsenite-regulated Thiol Reductase of the proteasome 19 S particle.

Perturbation of the cytoplasmic protein folding environment by exposure to oxidative stress-inducing As(III)-containing compounds challenges the ubiquitin-proteasome system. Here we report on mass spectrometric analysis of As(III)-induced changes in the proteasome's composition in samples prepared by stable isotope labeling with amino acids in cell culture, using mammalian cells in which TRP32 (thioredoxin-related protein of 32 kDa; also referred to as TXNL1) was identified as a novel subunit of the 26 S proteasome. Quantitative genetic interaction mapping, using the epistatic miniarray profiling approach, identified a functional connection between TRP32 and the proteasome. Deletion of txl1, the Schizosaccharomyces pombe homolog of TRP32, results in a slow growth phenotype when combined with deletion of cut8, a gene required for normal proteasome localization. Deletion analysis in vivo, chemical cross-linking, and manipulation of the ATP concentration in vitro during proteasome immunopurification revealed that the C-terminal domain of mammalian TRP32 binds the 19 S regulatory particle in proximity to the proteasome substrate binding site. Thiol modification with polyethylene glycol-maleimide showed disulfide bond formation at the active site of TRP32 in cells exposed to As(III). Pulse-chase labeling showed that TRP32 is a stable protein whose half-life of >6 h is surprisingly reduced to 1 h upon exposure of cells to As(III). These findings reveal a previously undescribed thiol reductase at the proteasome's regulatory particle.