Presentation of Ag by class II MHC is regulated by lysosomal proteases that not only destroy the class II invariant chain (Ii) chaperone but also generate the peptide Ag that is loaded onto the class II MHC dimer. We sought to determine the extent to which asparagine endopeptidase (AEP) influences human Ag and Ii processing. Our data confirm the constructive function of AEP in tetanus toxoid processing, but they are discordant with findings that suggest a destructive role for AEP in processing of the immunodominant myelin basic protein epitope. Furthermore, we observed no effect on invariant chain processing following AEP inhibition for several distinct allelic variants of human class II MHC products. We find that cysteine and aspartic proteases, as well as AEP, can act redundantly to initiate Ii processing. We detected considerable variation in lysosomal activity between different EBV-transformed B cell lines, but these differences do not result in altered regulation of invariant chain catabolism. We propose that, as for bound peptide Ag, the identity of the lysosomal enzyme that initiates invariant chain cleavage is dependent on the class II MHC allelic variants expressed.
Class II MHC-restricted presentation of antigenic peptides drives activation of CD4+ T cells. Generation of stimulatory epitopes requires processing of intact protein into peptides, which are then loaded onto the class II MHC dimer and presented at the cell surface. Maturation of class II MHC products is regulated by invariant chain (Ii)4 (1), a surrogate substrate and trafficking chaperone that is degraded by proteolytic enzymes in endosomal compartments. These proteolytic enzymes are active throughout the endosomal–lysosomal pathway and cleave Ags destined for presentation by class II MHC protein cargo to yield the full repertoire of peptides available for binding. Productive proteolytic cleavages generate immunogenic epitopes, while destructive proteolysis may prevent presentation of antigenic peptides. The identity of the T cell epitope presented by class II MHC shapes the ensuing immune response and likely contributes to the establishment and maintenance of immunological tolerance.
The use of specific protease inhibitors and the study of gene knockout mice has been applied to the study of many of the proteases involved in Ag processing and presentation by class II MHC (2, 3). Blockade of the progressive cleavage of Ii results in the accumulation of Ii intermediates and a corresponding decrease in surface expression of class II MHC products (4, 5). Experiments using human B cells treated with the cathepsin S (CatS)-specific inhibitor leucine-homophenylalanine-vinyl sulfone (LHVS) (6, 7) and characterization of CatS knockout mice (2, 8, 9) have implicated CatS in the terminal cleavage of Ii to yield CLIP (10). Further studies have demonstrated cell-specific cleavage of p10 by cathepsin L (CatL) (11) in thymic epithelial cells (8) and cathepsin F (CatF) in macrophages (9). Earlier steps in Ii degradation have been more difficult to elucidate, although MHC-haplotype-dependent differences may well exist (12). Treatment of human EBV-transformed B cell lines with the broad-spectrum cysteine protease inhibitor leupeptin results in the generation of the largest characterized Ii intermediate, p22, the leupeptin-induced peptide (LIP) (13, 14). The protease(s) that mediate this intermediate cleavage remain to be identified. Also unknown are the proteases that mediate the earliest cleavage event in Ii degradation. Initial Ii cleavage has been ascribed to several candidates, including an aspartyl protease (15) and asparagine endopeptidase (AEP) (16). The most abundant aspartyl protease cathepsin D (CatD) (17) is dispensable for class II MHC presentation and degradation of Ii in CatD knockout mice (18). Likewise, AEP, which is required to activate CatL and has been proposed as a candidate for Ii cleavage, plays no immediate role in Ii degradation in AEP knockout mice of the H-2b haplotype (19).
Although lack of AEP does not impair invariant chain processing in the murine model, AEP inhibition has been shown to delay the initiation of Ii cleavage in some human EBV-transformed B cell lines and in monocyte-derived dendritic cells (16). The specific cleavage by AEP immediately C-terminal to asparagine residues—an action that, unlike that of other cysteine proteases, is insensitive to the inhibitors leupeptin and E-64 (20, 21)—has been implicated in Ii processing. Ii contains two C-terminal AEP cleavage sites that could potentially generate the p10 and p22 fragments of Ii, and in vitro Ii is a substrate for AEP (16). Although AEP may not be the only lysosomal enzyme capable of initiating Ii cleavage in vivo, AEP could dominate Ii processing or accelerate Ii degradation in human APC. The role of AEP in initiating Ii processing and in Ag processing more generally in humans is still unclear.
The cleavage specificity of AEP has led to hypotheses about, and identification of, intact proteins that contain suitably accessible asparagine residues and that are therefore potential substrates for AEP. For one such protein, tetanus toxoid (TT) Ag, proteolysis by AEP is required to generate peptides that bind to class II MHC products and so initiate a CD4+ T cell response. AEP initiates processing of TT Ag and promotes presentation of immunogenic peptides (1, 21, 22). Such gateway processing could ultimately regulate the availability of peptides for binding to class II MHC products. In addition to productive generation of peptide, AEP may also destroy peptide epitopes. An immunodominant peptide of myelin basic protein, myelin basic protein (MBP) peptide 85–99, contains an asparagine residue that constitutes a cleavage site for AEP, preventing its presentation (23). MBP peptide 85–99 has been implicated in multiple sclerosis (MS), and CD4+ T cells reactive to this peptide have been cloned from individuals with MS (24). It has been claimed, however, that cathepsin G, and not AEP, is responsible for destructive processing of MBP peptide 85–99 (25).
To explore the role of human AEP in the maturation of class II MHC products and in processing and presentation of class II MHC peptide cargo, we herein examine several distinct human class II MHC products expressed in EBV-transformed B cell lines. We find that, as in mice, AEP activity is not necessary for class II maturation. A number of lysosomal proteases, including both aspartyl and cysteine proteases, regulate early events in Ii degradation. We find that the repertoire of active proteases in EBV-transformed B cells differs between cell lines, suggesting that individuals or even clones of B cells within individuals may use distinct sets of lysosomal proteases to accomplish early events in invariant chain cleavage and Ag processing. These proteases also generate the peptide cargo loaded into the binding pocket of class II MHC. Leaving aside the well-established concept of minor histocompatibility Ags as the products of non-MHC genes that show allelic (or gender) variation, the repertoire of self- and antigenic peptides presented to CD4+ T cells differs between individuals, even if they possess identical MHC products. Enhanced presentation of immunogenic self-peptide may be a factor in determining susceptibility to and severity of autoimmune diseases such as MS. These findings underscore the complexity of the class II MHC-restricted pathway of Ag presentation. Although allelic variation in MHC products is a decisive factor, the identity and level of activity of processing proteases present in an APC contribute as well.
Materials and Methods
The EBV-transformed human B cell lines BM14 (DRB1∗0401), Boleth (DRB1∗0401), Cox (DRB1∗0301), DBB (DRB1∗0701), Deu (DRB1∗0401), Ducaf (DRB1∗0301), MGAR (DRB1∗1501), Priess (DRB1∗0401), QBL (DRB1∗0301), Steinlin (DRB1∗0301), Vavy (DRB1∗0301), and WT51 (DRB1∗0401) were obtained from the International Histocompatilibilty Work Group (Fred Hutchison Cancer Research Center, Seattle, WA). The lymphoblastoid B cell line PALA (DRB1∗0301) was a gift from P. Cresswell (Yale University, New Haven, CT). B cell lines were maintained in complete RPMI (RPMI 1640 supplemented with 2 mM of l-glutamine, 5 mM of HEPES, 100 U/ml penicillin/streptomycin (all from BioWhittaker), 0.5 mM of sodium pyruvate, and 0.5 mM of nonessential amino acids (from Life Technologies)) and 8% FBS in 150 ml (2) vented flasks (CoStar) at 37°C in 6% CO2.
Generation of Ag-specific T cell clones
The MBP-reactive T cell clone Ob.1A12 was generated as previously described (24). The TT-reactive clones 3B3 and 3B10 were generated from the peripheral blood of a healthy individual (DRB1∗0701, DRB1∗1501, DRB1∗1503, DRB1∗1507). To generate TT-reactive clones, PBMC were cultured with 0.05 Lf/ml TT (University of Massachusetts Medical School Biological Labs, Jamaica Plains, MA) and plated at 150,000 cells/well in complete RPMI with 5% heat-inactivated human AB male serum (Omega Scientific). On day 5, 20 U/ml recombinant human IL-2 (TECIN; NCI Biological Resources Branch) was added. On day 14, cryopreserved autologous PBMC were thawed, pulsed with TT, and added to restimulate the cultures, and IL-2 was added 48 h later. Wells with cell growth were split as necessary. When cultures were at rest, a split-well assay was performed. Each positive growth line was split into four wells. Two wells received irradiated (5000 rads), autologous PBMC pulsed with TT and two wells received irradiated, autologous PBMC without Ag. After 48 h, each well was pulsed with 1 μCi tritiated thymidine (New England Nuclear) and harvested after an additional 18 h (PerkinElmer). Stimulation index was calculated by dividing the mean cpm of a T cell line activated by TT-pulsed PBMC by the mean cpm of cells of the same line incubated with PBMC alone. Lines were considered positive if the stimulation index was greater than 5. Positive lines were cloned by limiting dilution at 0.3 cells/well on TT-pulsed irradiated autologous PMBC and expanded with IL-2. Clones were tested for TT reactivity as described above.
All clones were restimulated in 96-well U-bottom plates (Costar) with plate-bound anti-CD3 (clone UCHT1) and 0.5 μg/ml soluble anti-CD28 (clone 28.8) purchased from BD Biosciences. Plates were coated with 50 μl of anti-CD3 diluted to 2.5 μg/ml in PBS, incubated for 2 h at 37°C, and then washed once in PBS. Restimulated clones were expanded for 4 wk in X-VIVO 15 medium (Cambrex) containing 20 U/ml recombinant human IL-2. The T cells were then cryopreserved at a concentration of 1 × 107 cells/ml in freezing medium containing 92.5% FCS and 7.5% DMSO at −80°C until use. For Ag presentation assays, T cells were restimulated from aliquots of the same freeze date.
The T cell clone Ob.1A12 was plated in triplicate (5 × 104 cells/well) and cocultured at a 1:2 ratio with APC for 96 h at 37°C with 5% CO2 in 96-well U-bottom plates in X-VIVO 15 media without serum. The T cell clones 3B3 and 3B10 were plated in triplicate (2.5 × 104 cells/well) at a 1:2 ratio with APC. Assays were conducted in X-VIVO 15 medium containing 10 U/ml recombinant human IL-2. To determine proliferation, 120 μl of culture supernatant was removed and replaced with medium containing 1 μCi of [3H]thymidine (New England Nuclear) for the final 16 h of culture.
EBV-transformed B cells have been used previously as a model for Ag presentation (23, 26). To generate APC using EBV-transformed B cells, MGAR cells were washed twice in PBS to remove serum and then plated at 1.25 × 106 cells/ml in X-VIVO 15 media containing AEP inhibitor (AEPi; MV026630) or control amounts of DMSO. Cells were pretreated for 30 min before a pulse with recombinant human MBP or MBP peptide 85–99, as indicated. Final concentrations of AEPi or DMSO, as indicated, were maintained throughout the pulse. APC were incubated for 16 h at 37°C with 5% CO2 and then fixed with 0.05% glutaraldehyde (Sigma-Aldrich). Full-length, HPLC-purified, recombinant human MBP was a kind gift from K. Wucherpfennig (Dana-Farber Cancer Institute, Boston, MA). MBP peptide 85–99, amino acid sequence ENPVVHFFKNIVTPR, was synthesized by New England Peptide. PBMC isolated using a Ficoll-Paque gradient and frozen in freezing media were used to generate APC for TT presentation; these cells were pulsed using the protocol described above with TT.
Detection of cytokine by ELISA
Culture supernatants were harvested after 80 h of coculture to assess cytokine production. Concentration of IFN-γ and IL-13 in the supernatant were detected by capture ELISA, as previously described (27), with Abs and standards purchased from Pierce Biotechnology (Endogen).
Substrate-specific protease activity assays
EBV-transformed B cells were washed once in PBS and then resuspended in lysis buffer containing 50 mM of citrate, 0.1% CHAPS (pH 5.5), 5 mM of DTT, and 0.5% Triton X-100. After removal of cell debris and nuclei by centrifugation, 10 μg of postnuclear lysates (determined by Bradford assay, Bio-Rad Laboratories) were incubated with 100 μM of protease substrate in lysis buffer at 37°C for the times indicated. The protease substrate Z-Ala-Ala-Asn-AMC (AnaSpec) was used to assay for AEP activity, Z-Arg-Arg-AMC (cathepsin B substrate III) for cathespin B (CatB), and Z-Phe-Arg-AMC for CatB/CatL (both from EMD Biosciences). For AEP activity, reactions were measured in buffer containing 20 mM of citric acid, 60 mM of Na2HPO4, 0.1% CHAPS (pH 5.5), 1 mM of EDTA, and 1 mM of DTT. For CatB and CatB/CatL activity, reactions were performed in buffer containing 50 mM of Na2HPO4 (pH 6.25), 1 mM of EDTA, and 1 mM of DTT. All reactions were conducted in black-walled 96-well plates (Nunc) in a total volume of 200 μl. Cleavage of the substrate results in release of 7-amido-4-methyl coumarin (AMC), which was measured at 460 nm in a Wallac fluorescence plate reader. To test the effect of AEPi on protease activity, cells were incubated with 40 μM of AEPi for 30 min at 37°C and then placed on ice before lysis. The rate of substrate cleavage was calculated as the change in arbitrary fluorescence units per microgram protein during 1 h.
Pulse-chase and immunoprecipitation
For radiolabeling experiments, cells were washed once in PBS before incubation in methionine-free medium (Invitrogen: Life Technologies) for 1 h at 37°C. During the last 10 min of starvation, cells were treated with control amounts of DMSO, 1 mM of leupeptin, 10 μM of pepstatin A (all from Sigma-Aldrich), or 40 μM of AEPi (MV026630), synthesized as described (28). These concentrations of inhibitor were maintained throughout the remainder of the pulse and chase. Following starvation, cells were pulsed with 0.5 mCi of [35S]methionine (PerkinElmer) for 45 min and then chased with B cell maintenance medium for the times indicated. After the chase, cells were washed once with PBS and lysed in 50 mM of Tris (pH 7.4) containing 5 mM of MgCl2, 0.5% Nonidet-P40, and complete miniprotease inhibitor mix (29).
After removal of cell debris and nuclei by centrifugation, lysates were precleared with normal rabbit and mouse serum together with Staphylococcus aureus protein A or mouse anti-human IgG (Pierce Chemical) with protein G-agarose (29). The lysates were then incubated with the conformationally specific mAb Tü36 to immunoprecipitate class II MHC molecules. Samples were boiled in sample buffer before analysis on a 12.5% SDS-PAGE gel. Band OD was quantified using the software program ImageJ (Rasband, W. S. 1997–2007. ImageJ. National Institutes of Health, Bethesda, MD).
Active-site labeling and immunoblot
To label active cysteine proteases, 50 μg of postnuclear lysate, generated as described above, was incubated with 10 μM of DCG-04 for 1 h at 37°C. The synthesis of the active-site-directed probe DCG-04 has been described (30). Labeled proteins were resuspended in loading buffer and boiled before analysis on a 12% SDS-PAGE gel. Proteins were then transferred to polyvinylidene difluoride membranes (Millipore) for analysis of DCG-04 labeling and immunoblot. DCG-04 labeling was detected by incubating the membrane with HRP-coupled streptavidin (Sigma-Aldrich). To identify β-actin by immunoblot, membranes were incubated with rabbit anti-β-actin mAb (antibody 4967; Cell Signaling Technology) followed by HRP-conjugated goat anti-rabbit Ab (Cell Signaling Technology). To identify CatL by Western blot, membranes were incubated with rat anti-CatL (clone 204106; R&D Systems) followed by HRP-conjugated goat anti-rat (Biomeda). Bound HRP was visualized using the Western Lightning detection kit (PerkinElmer).
Inhibition of AEP is not sufficient to prevent Ii processing in human B cells
To assess the role of AEP in Ii processing, we inhibited AEP activity in human EBV-transformed B cells and analyzed the formation of Ii cleavage intermediates in pulse-chase experiments. The formation of protease inhibitor-induced cleavage intermediates is diagnostic of the different stages of Ii processing. Full-length Ii is cleaved to generate the p21–22 kDa (LIP; p22) N-terminal fragment, which is visible following leupeptin treatment and is thought to be cleaved by cysteinyl proteases to form the intermediate p10–14 kDa (small LIP; p10) and the terminal Ii degradation product CLIP (2). We treated the EBV-transformed B cell line PALA (DRB∗0301) with the AEP inhibitor MV026630 (AEPi) in the presence or absence of leupeptin, pulsed the cells with radiolabeled [35S]methionine, and then chased the cells with unlabeled amino acids for 3 h. This combination of protease inhibitors was used to maximize the likelihood of detecting unique cleavage intermediates larger than p22. Radiolabeled class II MHC molecules were immunoprecipitated from these cells using the Ab Tü36, which binds properly folded class II αβ dimers. In cells treated with leupeptin alone, the p22 Ii intermediate (13) is clearly detectable after 2 h of chase (Fig. 1). The block of Ii cleavage at p22 is complete, as generation of the p10 Ii fragment is not observed in leupeptin-treated cells. Addition of AEPi to these cells neither delays the formation of SDS-stable dimers nor blocks Ii processing, as the p10 fragment is detectable in these cells. When treated with AEPi and leupeptin combined, there is a slight delay in formation of p22, which is clearly present after 2 h of chase, although to a lesser extent than in cells treated with leupeptin alone. In the PALA cell line, we were unable to inhibit formation of p22 with AEPi. We did find, however, that treatment with AEPi and leupeptin together results in the formation of an intermediate Ii cleavage product at ∼24 kDa, similar to that detected in human monocyte-derived dendritic cells (16). We also detected this slight delay in Ii processing, which generated a p24 cleavage fragment, in a second, haplotype-matched, EBV-transformed B cell line, QBL (DRB1∗0301) (Fig. 1). In this cell line, a delay in p22 formation and appearance of the p24 fragment was visible after 2 h of chase (see Fig. 3 A). QBL treated with AEPi and leupeptin formed 18% less p22 and 10% more p24 after 3 h of chase than cells treated with leupeptin alone. These data confirm the ability of AEP to cleave Ii in the presence of leupeptin, as AEPi and leupeptin treatment resulted in formation of an intermediate fragment slightly larger than the leupeptin-induced p22. Inhibition of AEP alone, however, was not sufficient to block Ii cleavage. We think that this intermediate cleavage is not required to initiate Ii processing, as class II MHC molecules mature with normal kinetics even in the presence of AEPi.
Class II MHC allelic variants bind the Ii CLIP fragment with variable stability (10). To determine the impact of MHC haplotype on early Ii processing, we subjected HLA-DR2 and HLA-DR4 B cell lines to pulse-chase and immunoprecipitation analysis during AEPi and leupeptin treatment. The Ii fragments detected in the EBV-transformed B cells lines MGAR (DRB1∗1501) and Priess (DRB1∗0401) were identical to those visualized using the HLA-DR3 lines PALA and QBL under control (DMSO) or leupeptin treatment (Fig. 1). Inhibition of AEP alone did not prevent class II maturation or promote Ii fragment generation in either of these cell lines (Fig. 2) or in an HLA-DR7 cell line (DBB; data not shown). We could not determine a requirement for AEP enzymatic activity in maturation of any of the HLA-DR allelic variants assayed.
AEPi treatment alone did not reveal any differences in class II MHC processing between class II MHC variants, but when treated in combination with leupeptin, a subset of these cell lines produced increased levels of the Ii p24 fragment. Generation of the leupeptin- and AEPi-induced p24 fragment was pronounced in leupeptin- and AEPi-treated MGAR cells, as compared with identically treated Priess cells (Fig. 1). In the Priess cell line, p24 was only faintly detectable in the presence of both AEPi and leupeptin after 3 h of chase, whereas in MGAR cells, inhibition of AEP in the presence of leupeptin not only yields p24, but it delays the formation of p22 (Fig. 1). The Ii fragmentation we observed in the MGAR and Priess cell lines was different from that visualized in the HLA-DR3 MGAR and Priess cells, which, as demonstrated above, generated equivalent amounts of p22 and p24 after 3 h of treatment with AEPi and leupeptin.
Given these findings, we hypothesized that MHC allelic variation regulates the generation of the p24 fragment with AEPi and leupeptin treatment. Slight differences in Ii–MHC binding could result in increased exposure of the asparagine residues required for AEP-mediated proteolysis and thus produce an increased reliance on AEP for Ii proteolysis. Therefore, cell lines with increased use of AEP would form more p24 after AEPi and leupeptin treatment than cell lines primarily using other proteases. To test this hypothesis, we examined Ii processing in several additional DRB∗0301 and DRB∗0401 homozygous cell lines. As predicted, p24 and p22 banding patterns were consistent in the DRB∗0301 lines treated with AEPi and leupeptin during pulse-chase (Fig. 3,B). Furthermore, this banding pattern was distinctive from that generated in the identically treated DRB∗0401 lines (Fig. 3 C). After 3 h of chase, DR3 variant cells generated 62% p24 and 28% p22 (n = 6; SD = 6.97%, p = 0.001), while the DR4 haplotype lines generated 21% p24 and 79% p22 (n = 5; SD = 9.7%, p < 0.0001). These data suggest that cells expressing identical HLA-DR β-chains will process Ii similarly, resulting in the same relative amounts of p24 and p22 formation when treated with leupeptin and AEPi, and that the resulting Ii fragment formation is distinctive from cells of that class II MHC haplotype.
Inhibition of AEP enzymatic activity
The extent to which allelic variation in MHC products affects Ii cleavage has not been carefully studied in the human system, but polymorphisms in murine class II MHC can alter the proteolytic requirement for individual proteases in Ii cleavage (12). Because of the diverse genetic background of the human population, it is likely that EBV-transformed B cell lines generated from unique donors will exhibit differences in enzyme expression and activity as well as show polymorphisms in class II MHC products. We wanted to confirm that the minor differences in Ii cleavage detected during pulse-chase were due to donor variability rather than incomplete inactivation of AEP. We assayed AEP activity in MGAR and Priess postnuclear lysates with the AEP-specific substrate Z-Ala-Ala-Asn-AMC. Treatment with as little as 5 μM of AEPi was sufficient to reduce hydrolysis of the AEP substrate by 64% in Priess cells and 57% in MGAR cells. Treatment with 20–80 μM of AEPi resulted in complete inhibition of AEP activity (Fig. 4, and data not shown). AEP is completely inactivated by 40 μM of AEPi used in our pulse-chase experiments. Therefore, the minor differences we observed in the extent and rate of Ii cleavage cannot be attributed to residual AEP activity. When we assayed AEP substrate hydrolysis in the absence of AEPi treatment, we found that MGAR lysates cleaved the AEP substrate less efficiently than did Priess lysates, resulting in reduced release of fluorescence (Fig. 4). The rate of substrate cleavage in Priess was 2-fold that of MGAR (36.87 ± 12.12 as compared with 11.09 ± 4.826 (fluorescence units ×103) μg−1 h−1; n = 4; p = 0.0478). Priess cells thus have greater basal AEP activity than do MGAR cells. There could be an increased reliance on AEP to initiate Ii cleavage in cells that have high levels of AEP activity. Although we find that generation of AEPi-sensitive Ii fragments is largely regulated by MHC allelic variation, enzyme activity may still play some regulatory role in Ii processing.
Inhibition of AEP does not enhance presentation of MBP peptide 85–99
Lysosomal proteases mediate not only degradation of Ii, but also proteolytic generation of antigenic peptide for loading onto class II MHC products. Although we could not establish a significant role for AEP in the destructive processing of Ii, it remains likely that AEP plays a role in the processing of at least a subset of proteins into antigenic peptides. AEP has been implicated both in the productive generation of TT peptides and in the destructive processing of an immunodominant epitope from MBP (21, 23). Processing of the CatL precursor likewise completely depends on the presence of AEP (19). The role of AEP in MBP processing, however, is a subject of debate, as destruction of MBP peptide 85–99 has also been attributed to cathepsin G, a serine protease present in the lysosomal compartment of human B lymphocytes (25). AEP is dispensable for Ag presentation of epitopes from OVA and myelin oligodendrocyte glycoprotein in the AEP knockout mouse (19). The role of AEP in MBP processing, as well as the dispensability of AEP in the processing of other epitopes containing asparagine, necessitates a reappraisal of the role of AEP in Ag presentation. We treated MGAR cells with 40 μM of AEPi and then pulsed these cells with full-length, recombinant human MBP for 16 h. These Ag-loaded MGAR cells were then fixed with glutaraldehyde and used as APC for the CD4 T cell clone Ob.1A12, which is activated by the N-terminus of the MBP peptide 85–99 in the context of DRB∗1501 (31). We found that the presence of AEPi during Ag processing did not alter activation of Ob.1A12 (Fig. 5). T cell proliferation (Fig. 5,A) and secretion of the cytokines IFN-γ and IL-13 (Fig. 5, C and D) in response to MBP were not increased by AEPi treatment. As expected, inhibition of AEP did not affect the presentation of MBP peptide 85–99 (Fig. 5 B).
To confirm the efficacy of our AEP inhibitor in Ag-presentation assays, we tested the effects of this inhibitor in preventing the presentation of TT. Cleavage by AEP of the TT C fragment is reported to be required for optimal presentation of tetanus toxin to reactive T cells (1). We treated human PMBC with 40 μM of AEPi and then pulsed these cells with TT for 2 h before glutaraldehyde fixation. These cells were then used as APC for autologous tetanus-responsive clones. After 72 h of stimulation, we measured IFN-γ secretion by the Ag-specific T cells. AEP inhibition markedly reduced the efficacy of PBMC to present TT (Fig. 6). We conclude that AEP does play a role in the processing and presentation of some Ags as reported by Watts and coworkers (21), and that the used AEP inhibitor is effective in modulating this activity. Although AEP activity is not necessary for successful class II MHC maturation, the action of AEP can modulate the peptide repertoire available for class II MHC binding and will therefore determine, to some extent, the immunogenicity of the class II MHC: peptide complex presented by an APC.
Initiation of Ii cleavage is not blocked by AEPi or pepstatin
Aspartyl proteases have also been implicated in initiating Ii cleavage and may compensate for the activity of AEP in the presence of AEPi. To test this possibility, we conducted pulse-chase experiments in the presence of the broad-spectrum aspartyl protease inhibitor pepstatin A. Treatment of the PALA and MGAR cell lines with pepstatin A did not significantly affect Ii cleavage and maturation of class II MHC (Fig. 7). When we inhibited AEPi in the presence of pepstatin A, we also did not observe a significant contribution of AEP to Ii degradation. The p10 fragment was detectable as early as 1 h after chase in PALA and MGAR cells treated with both AEPi and pepstatin A (Fig. 7, and data not shown). These findings are similar to what we observed in cells treated with AEPi alone. To better detect early Ii cleavage fragments, we pretreated cells with leupeptin in addition to pepstatin A and AEPi before pulse-chase and immunoprecipitation analyses. In MGAR cells treated with this combination of enzyme inhibitors, Ii cleavage was blocked and p22 failed to form (Fig. 7). In the PALA cell line, we observed transient formation of p24 during the second hour of chase (data not shown). Our findings suggest that the ability of AEP to initiate cleavage of Ii is complemented by a pepstatin A-susceptible aspartyl protease, which can also initiate Ii cleavage to generate p22. Blockade with AEPi or pepstatin A is not sufficient to prevent formation of p22, but blockade of both prevents p22 generation even in the presence of leupeptin. It is likely that leupeptin-susceptible cysteine proteases are sufficient to mediate the initial cleavage of Ii in the presence of pepstatin A and AEPi. The initiating cleavage of Ii in humans appears to be a poorly regulated event, accomplished by the redundant action of several lysosomal proteases. Our findings suggest that leupeptin-sensitive enzymes, aspartyl proteases, and AEP can initiate Ii cleavage. These data support the hypothesis that multiple enzymes can initiate Ii processing (16); that is, cleavage of a region in Ii, suitably exposed to the action of any of a number of different proteases, can initiate Ii processing. MHC haplotype, subcellular localization of the class II MHC-Ii trimer, and lysosomal enzyme activity are all potential contributing factors in determining which protease(s) ultimately initiate(s) Ii destruction.
Variability in lysosomal enzyme activity among EBV-transformed B cell lines
Differential expression of active lysosomal proteases could account for observed variation in the generation of Ii cleavage products. Becaue we queried relatively few class II MHC-matched cell lines, preferences for an initial Ii degradation site cannot be attributed to MHC haplotype without excluding differences in lysosomal activity. Tissue type-specific expression of cathepsins within human B-lymphoblastoid cells has been described (32), but it may fluctuate from cell line to cell line, as in the case with monocytes (33).
When measuring the efficacy of AEP inhibition (Fig. 4), we noted surprising, but consistent, variability in baseline AEP activity between cell lines. To confirm these findings, we measured rates of hydrolysis of an AEP-specific substrate (Z-Ala-Ala-Asn-AMC) in a panel of EBV-transformed B cells (Fig. 8 A). In agreement with our previous findings, hydrolysis of the AEP substrate was greatest in the Priess cell line, which contained significantly more AEP activity than did the least active QBL cells (29.99 ± 4.46 vs 2.65 ± 0.85 (fluorescence units × 103) μg−1 h−1; n = 3; p = 0.0341). In addition to this range in enzyme activity, we noted that AEP activity did not correlate with observed delays in Ii cleavage following AEP inhibition. Although PALA and QBL cells generated nearly identical Ii p22 fragments when inhibited with both leupeptin and AEPi, the basal rate of AEP activity in PALA cells was approximately 5-fold greater than that of the QBL. Our results confirm that endogenous AEP activity cannot predict the extent to which AEP is required for Ii cleavage. A preponderance of catalytically active AEP does not predispose a cell toward AEP-mediated initiation of Ii degradation.
To determine whether AEP alone among lysosomal proteases was differentially expressed in EBV-transformed B cells, we also assayed cathepsins B, L, and S activity. By measuring hydrolysis of the fluorogenic substrates Z-Arg-Arg-AMC (CatB), Boc-Val-Leu-Lys-AMC (CatS), and Z-Phe-Arg-AMC (CatB/CatL), we determined that rates of hydrolysis for these substrates were also variable between B cell lines (Fig. 8 A, and data not shown). CatS plays a role in class II maturation in EBV-transformed B cells; we detected CatS activity consistent with that reported for the human B cell lines HOM3, WT51, and Jesthom (7, 32). Although all CatS hydrolysis rates measured fell within a 2- to 3-fold range, CatS activity was lower in QBL and Priess than in all other cell lines assayed. These minor differences in CatS activity provide a rationale as to why CatS inhibition has been shown to block late-stage Ii degradation in some B cell lines and not in others. It is possible that residual CatS activity in more active cell lines may be sufficient to cleave Ii, or, more likely, cells with low CatS activity may contain complementary proteases that compete with CatS for the Ii substrate. Complete selective inhibition or siRNA knock-down of CatS in multiple B cell lines, or in B cells ex vivo, would be required to conclusively establish the requirement for CatS in class II MHC maturation in this tissue.
Unlike CatS, we detected significant variability in CatB activity. Notably, Priess lysates contained only background levels of CatB activity toward the substrate Z-Arg-Arg-AMC, but reacted considerably with the substrate Z-Phe-Arg-AMC, which is cleaved by both CatB and CatL. Primary human B cells and EBV-transformed B cell lines are known to contain active CatS and CatB, but apparently they lack CatL activity (7, 32, 34, 35). Priess cells contained 41 kDa pro-CatL and the 29/34 kDa form of mature CatL, which were not detectable in MGAR cells (Fig. 8 C). We could not, however, resolve a distinct, unique band corresponding to active CatL in DCG-04-labeled Priess lysates (see below, and Fig. 8 C). CatL is known to be up-regulated in certain human tumor cells and can be detected at low levels in the B cell lymphoma line Raji (36). Expression of CatL in the Priess line could be an unusual artifact of transformation. It is of interest that AEP and CatL activities are both high in the Priess B cell line, in light of evidence that AEP is required for CatL processing (19). Nevertheless, these data demonstrate that cathepsin activity is not necessarily equivalent among EBV-transformed B cells.
To extend our findings, we used the mechanism-based, active-site-directed probe DCG-04 to analyze the active cathepsin repertoire in lysates derived from different B cell lines. DCG-04, a biotintylated derivative of the broad-spectrum cysteine protease inhibitor E-64, has been successfully used as an activity-dependent affinity label for cysteine proteases in APC lysates (30, 37). To visualize DCG-04 binding, we incubated cell lysates in an acidic buffer (pH 5.5) containing 10 μM of DCG-04. We then resolved the lysates by SDS-PAGE and revealed active-site labeling with streptavidin blotting (19). Several distinct polypeptides were visualized, including a species at 25 kDa, which was identified by immunoblot as CatS (Fig. 8 B; data not shown). Consistent with our fluorescence release assay data, CatS activity was greatest in the DBB cell line and reduced in the other cell lysates analyzed. We detected variability in expression of active proteases at 50 and 37 kDa as well; for example, MGAR cells contain less of the active protease resolved at 50 kDa than do PALA cells, but both contain a similar amount of the 37 kDa protease. These findings confirm considerable variability in cathepsin activity, even among the few B lymphoblastoid cell lines tested. Because EBV-transformed cell lines contained in tissue culture for prolonged periods of time are likely to tend toward oligoclonality, this type of variation may contribute to differences in APC function, even when comparing sublines derived from the same original EBV-transformed cell line.
We undertook this study to clarify the role of human AEP in class II MHC processing and presentation. AEP, a lysosomal cysteine protease, is likely to play a role in class II MHC maturation and in shaping the peptide repertoire available for class II binding. AEP is unique among lysosomal proteases in possessing a substrate specificity that is well defined and potentially nonredundant (20), making this enzyme a promising target for directed immunomodulation. Investigations into the role of AEP in APC have been approached from several angles. AEP has been implicated in many aspects of the class II MHC presentation pathway: activation of other lysosomal proteases (19, 38), destruction of Ii (16), and both destruction and creation of peptide epitopes (21, 23, 39). The results obtained so far have been contradictory, and no publication has reconciled findings that suggest an important role for AEP in Ag processing and presentation with works that suggest a nonessential role for AEP or that attribute the proposed role of AEP to other enzymes (19, 25).
Using a cell-permeable, AEP-specific inhibitor, we first sought to confirm the possible role of AEP in Ii proteolysis in human EBV-transformed B cell lines. The use of protease inhibitors in human cell lines has suggested such a role for AEP in the initiation of Ii proteolysis (16). However, blockade of AEP activity (the success of which was verified by direct measurement of AEP activity) did not significantly impair Ii catabolism or prevent formation of mature class II αβ heterodimers in these cells (Fig. 1). Although we were unable to prevent the initial cleavage of Ii using AEPi, this inhibitor effectively and significantly decreased AEP activity in living cells (Fig. 4) and modulated peptide presentation efficacy for TT (Fig. 6), confirming the potency of the inhibitor and providing independent confirmation of the results of Watts and colleagues (21).
AEP inhibition prevented efficient “unlocking” of a cryptic tetanus epitope and resulted in reduced activation of a tetanus-specific CD4+ T cell clone. The requirement for AEP in tetanus presentation is a robust example of enzyme-restricted Ag processing. Similarly, AEP-mediated processing of MBP has been shown to destroy an asparagine-containing immunodominant epitope, MBP peptide 85–99 (6, 19). Because MBP-derived epitopes are presented intrathymically and can contribute to self-tolerance against MBP, the questions of how and whether such peptides are generated are of great importance. Hypothetically, loss of this antigenic epitope in the thymus could result in the survival of MBP-reactive T cells and result in export of these potentially pathogenic cells to the periphery.
When exposed to human serum, MBP becomes unstable, which may allow MBP breakdown products to form and enable APC to load these peptides at the cell surface (40). To prevent degradation of MBP by serum proteases, we conducted our experiments in serum-free media with highly purified, recombinant MBP. We tested the ability of AEP-inhibited B cells to activate a MBP-reactive T cell clone isolated from the periphery of a patient with MS (24). This MBP-restricted clone binds in an unconventional manner to the N-terminal segment of the 85–99 peptide and is unresponsive to HPLC-purified MBP in the absence of intracellular processing (41, 42). We found that inactivation of AEP did not result in significant disruption of MBP peptide 85–99 presentation by these cells (Fig. 5). We failed to detect any alteration in CD4+ T cell response following AEPi treatment, suggesting that AEP does not destroy whole, recombinant MBP presented by human B cell lines such that immunodominant peptides cannot be presented to T cells. These findings do not preclude a role for AEP in MBP destruction in vivo, but they do underscore the limitations of using B cell lines as a model of human APC. As previous reports have indicated, it is likely that both AEP and cathespin G (CatG) can cleave and destroy MBP peptide 85–99 (11, 16, 25). CatG has not been detected in human EBV-transformed B cells, but its presence in serum or bound to the cell surface of B lymphocytes may result in destructive MBP cleavage before internalization (25, 43). Furthermore, as AEP and CatG are expressed in the thymus (16, 25), it is possible that destructive processing by these enzymes contributes, in part, to deregulated negative selection of MBP-responsive T cells. Nevertheless, alternate evasion mechanisms for peptide 85–99-reactive T cells have been discussed (44), and, moreover, negative selection for MBP-reactive T cells is necessarily incomplete, as these cells can be cloned from the blood of healthy donors (24). In light of our data, it is unlikely that AEP destruction of MBP in vivo plays a major role in establishment or progression of MS.
In addition to shaping the peptide repertoire available for class II MHC binding, lysosomal enzymes regulate the proteolytic processing of Ii and thus affect the kinetics of peptide binding. Progressive catabolism of Ii prevents premature peptide loading while ensuring peptide sampling from the proper lysosomal–endosomal compartments. Several studies have established the roles of CatS and CatL in terminal Ii degradation, but uncertainty remains concerning the identity of the enzyme(s) required to initiate Ii cleavage. The inhibitor leupeptin does not prevent initiation of Ii degradation, which implies that a cysteine protease is not solely responsible for instigating Ii proteolysis. The leupeptin-insensitive proteases AEP and CatD are obvious candidates, but for both of these enzymes, data collected in human cell lines using protease inhibitors have not confirmed the analysis of AEP and CatD murine knockout models (18, 19). If mouse models are a proper representation of class II MHC proteolytic requirements, then neither AEP nor CatD is essential for Ii processing. We confirmed this observation and demonstrated herein that neither protease is absolutely required for Ii processing in human cells (Fig. 7). We find that successful blockade of early degradation of Ii requires inhibition of not only AEP and CatD, but also of cysteine endoproteases. Consequently, initiation of Ii cleavage is a highly redundant event that can be accomplished by any one of several lysosomal proteases.
Redundancy in early Ii processing events is not surprising, as AEP inhibition does not affect steady-state levels of surface class II MHC molecules, and Ii processing has been observed even in the presence of both AEP inhibitor and leupeptin (16). Despite the promiscuity of early Ii proteolysis, we observed variation between cell lines in the kinetics of class II MHC maturation following inhibition of AEP (Fig. 1). Formation of the intermediate p24 was not due to differences in enzyme activity (Fig. 8), but it correlated instead with the class II MHC haplotype carried by the cell line examined. Until now, variation in dominant Ii processing activity has been attributed primarily to APC type (16, 33), although early investigations that examined the effect of protease inhibitors such as leupeptin and pepstatin were careful to take HLA-DR haplotype into consideration when drawing broad biological conclusions about the mechanism of Ii degradation (4, 29, 45). Studies in mice have also determined unique proteolytic requirements for Ii degradation in strains with different class II MHC haplotypes (2, 46). We have not yet explored the mechanism by which class II MHC dictates enzyme preference, and further work is needed to determine the extent to which differentially regulated Ii proteolysis can alter the timing and location of class II MHC loading and thus bias the peptide repertoire. We speculate that the ease with which Ii breakdown is initiated is yet another mechanism by which susceptible MHC haplotypes can be predisposed toward disease.
In summary, we demonstrate that blockade of AEP activity in human APC does not significantly delay or impair the rate of Ii cleavage. We confirm the reported role of AEP in generating antigenic TT peptide, but we failed to observe an effect of AEP inhibition on processing of MBP or on the generation of mature αβ-peptide class II MHC molecules. Indeed, broad-spectrum blockade of cysteine and aspartyl proteases, as well as AEP inhibition, was required to more completely prevent Ii degradation. Despite the complementary action of lysosomal proteases on Ii, we did observe differences, albeit minor, in Ii hydrolysis in cell lines of different MHC haplotypes. These findings establish a significant regulatory role for the class II MHC products themselves during their own maturation.
We thank R. Maehr for support and technical advice. We also thank the J. Parvin, M. Boes, and M. Wolfe laboratories for use of their equipment; C. Baecher-Allan, D. Anderson, G. Bériou, and Y.-M. Kim for technical advice; and J. Pyrdol for assistance with the purified recombinant MBP. We also thank M. Hall for support.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health grants and by the National Multiple Sclerosis Society. H.C.H. was supported by a Damon Runyon Cancer Research Foundation postdoctoral fellowship. S.C.K. was supported with a research grant from the Juvenile Diabetes Research Foundation International.
Abbreviations used in this paper: Ii, invariant chain; AEP, asparagine endopeptidase; AEPi, AEP inhibitor; AMC, 7-amido-4-methyl coumarin; CatB, cathepsin B (form also for cathepsins D, F, G, L, and S); LIP, leupeptin-induced peptide; MBP, myelin basic protein; MS, multiple sclerosis.