In autoimmune diseases, the immune system targets molecules to which tolerance is normally maintained by a variety of mechanisms. A number of hypotheses explain how this may occur, including, for example, previous “unawareness” of the antigen because of its protected location, haptenization, a change in autoantigen sequence or folding, molecular mimicry, altered thresholds for proinflammatory signaling, or impaired normal tolerance mechanisms. The last two mechanisms are likely to predispose to autoimmunity to multiple autoantigens and so are less likely to be relevant to organ-specific diseases.
Goodpasture’s (anti–glomerular basement membrane [anti-GBM]) disease is the best understood renal organ–specific autoimmune disease. It is characterized by antibody and T lymphocyte (T cell) reactivity to the NC1 domain of a tissue-specific isoform of type IV collagen, α3(IV)NC1 (1). Patients with recent-onset Goodpasture’s disease have in their peripheral blood CD4+ T lymphocytes that proliferate and make IFN-γ when incubated with the Goodpasture autoantigen (2,3). Recent work has demonstrated that the fine specificity of patients’ α3(IV)NC1-reactive T cells is highly consistent between patients and that the predominant α3(IV)NC1-reactive T cells are specific for peptides that have high affinity for the disease-associated HLA DR15 class II molecule (3–5). These data raise the important question of how such T cells avoid deletion or immunoregulation by the normal tolerance mechanisms.
It is possible to rule out that the restricted tissue distribution and very slow turnover of α3(IV) collagen might lead to a relative lack of immunoregulatory mechanisms. We and others have demonstrated that the Goodpasture antigen is expressed in human thymus (6,7). Therefore, any failure to develop tolerance to high-affinity peptides from this antigen is likely to be a consequence of the failure of antigen-presenting cells (APC) to generate them or a failure to generate them in a context that leads to immunoregulation.
There is clear evidence that processing factors exert a powerful influence on the presentation of α3(IV)NC1 peptides. Biochemical analysis of the α3(IV)NC1 peptides that bound to HLA-DR15 on the surface of antigen-pulsed APC found that the most abundantly presented α3(IV)NC1 peptides were not those with highest affinity for HLA-DR15 (8,9). Therefore, processing factors must prevent peptides with higher affinity for DR15 from gaining access to MHC class II molecules within APC. Similar conclusions have been reached in experiments with other antigens (10).
We examined the hypothesis that the disease-associated peptides in Goodpasture’s disease are particularly susceptible to proteolytic destruction in APC. We focused on the four α3(IV)NC1 peptides that contain epitopes that are recognized by T cells from all six patients (the two disease-associated peptides dap131–150 and dap71–90) and from four of six patients (dap1–20 and dap31–50) (3). The results confirmed that the disease-associated peptides are destroyed early and furthermore showed that cleavages of two of them were obligate early steps in the sequential proteolysis of the antigen by lysosomes.
Materials and Methods
Antigens, Enzymes, and Cells
Cathepsin D that was purified from human liver was obtained from Athen Research & Technology (Athens, GA). Asparagine endopeptidase (AEP) was donated by Prof. C. Watts (University of Dundee, Dundee, UK). Production of recombinant α3(IV) NC1 [r-α3(IV) NC1] was as described before (8). The synthetic peptides were as described before (9). Lysosomal extracts were prepared from Epstein-Barr virus (EBV) transformed B cells as described previously (11). The DR15-expressing EBV-transformed B cell lines LCL013, 041, and 061 were donated by Prof. D. Crawford (University of Edinburgh, UK).
Digestion of r-3(IV) NC1
Lysosomal extracts were incubated at 37°C with r-α3(IV)NC1 in 10 mM dithiothreitol (DTT), 50 mM citrate (pH 4.6), and the indicated protease inhibitors at the following concentrations: Pepstatin A 10 μM, iodoacetic acid 14 mM, and PMSF 14 mM. Digests were analyzed by SDS-PAGE. For digests with purified enzymes, r-α3(IV)NC1 (100 μg) was dissolved in 8 M urea and coated onto S-Sepharose 6B beads (Sigma, St. Louis, MO) in 8 M urea and 50 mM Tris (pH 7.8) carried in 1.5 ml of Mobicol (VH Bio, Gateshead, UK) spin columns. Adsorbed protein was reduced by incubation in 10 mM DTT at 56°C for 45 min; washed with UHQ water (Millipore, Billerica, MA); and then suspended either with 1 μg of human liver cathepsin D (Athen Research & Technology) in 50 μl of tetramethylammonium formate (25% wt, pH 4.5; Sigma-Aldrich, St. Louis, MO) or with 2 mU of AEP in 50 μl of 80 mM citrate, 240 mM Na2HPO4, 1 mM EDTA, and 2 mM DTT (pH 5.8). Peptides that were released by digestion after 5 min to 2 h were recovered from the supernatant by centrifugation and from the beads by washing with 20% acetonitrile 1% tri-fluoroacetic acid (Romil, Cambridge, UK) and 1 M NH4OH. Recovered peptides were combined, then divided into aliquots, dried under vacuum, and stored at −80°C.
For N-terminal sequencing of fragments that were visualized by SDS-PAGE, gels were blotted onto polyvinylidene difluoride membranes and selected fragments were cut out and submitted for Edman degradation. Peptides were sequenced by a battery of mass spectrometric techniques, including matrix-assisted laser desorption ionization time of flight with Mass and Composition analysis (12) and daughter ion generation in a Finnigan LCQ mass spectrometer (ThermoFisher Scientific, Waltham, MA) followed by database and manual interpretation of daughter ion spectra.
Generation of Hybridomas
DR15-transgenic Balb/c mice were immunized with 50 μg of dap131–150 in complete Freund adjuvant with appropriate UK Home Office approvals. Splenocytes and lymph nodes were recovered after 7 to 10 d and expanded in vitro with three cycles of re-stimulation with dap131–150, expansion with rIL-2 at 20U/ml, and resting with rIL-2 at 2 U/ml. Surviving T cells were fused BW5147 [T cell recptor αβ] (13) cells (a donation from Dr. John Robinson, Newcastle University, Newcastle, UK) using the polyethylene glycol method and cloned by limiting dilution. Clones that expressed T cell receptor αβ and CD4 were screened for IL-2 production in response to dap131–150 presented by mitomycin C–treated DR15-expressing splenocytes. IL-2 was measured by ELISA (R&D DuoSet, Minneapolis, MN).
To identify clones that recognized α3(IV)NC1 epitopes that are naturally processed from intact α3(IV)NC1, DR15 B cells that were intact or fixed by incubation in 0.5% paraformaldehyde for 30 min at room temperature were pulsed with 7 μM AS346 or α3(IV)NC1 for 4 to 6 h, then washed and added to 0.5 × 105 of selected T hybridomas. IL-2 production at 24 h was measured by ELISA as above. The clone ha3p132.2 was selected as representative of a number of clones that produced larger amounts of IL-2 when incubated with dap131–150 or intact α3(IV)NC1 and used for all of the assays reported herein.
Fine Specificity of Patients’ T Cells
The harvesting and initial preparation of T cells from patients with recent-onset Goodpasture’s disease was as we have described before (3). For assessment of fine specificity of α3(IV)NC1-specific T cells, 2 × 105 purified peripheral blood mononuclear cells were incubated with 10 μg/ml selected α3(IV)NC1 peptides for 72 h at 37°C, then IFN-γ and IL-2 were measured in the culture supernatant by BD cytometric bead array (BD Bioscience, San Jose, CA) and proliferation measured from incorporation of 3H thymidine during an 18-h period.
Antigen Presentation Assays
DR15+ve EBV transformed B cells (1 × 105) or phorbyl myristate acetate–differentiated IFN-γ–activated THP1 cells were pretreated with pepstatin A (Sigma) for 30 min at 4°C, then 3.5 μM dap131–150 or α3(IV)NC1 was added and the cells were incubated at 37°C for 2 to 12 h. At the end of the incubation periods, processing was terminated by fixation with 0.5% paraformaldehyde for 30 min, the APC were extensively washed, 0.25 × 105 ha3p132.2 T hybridoma cells were added, and the cultures were incubated for an additional 24 h, at the end of which IL-2 production was measured by ELISA as described.
α3(IV)NC1 Is Rapidly Cleaved by Lysosomal Proteases
To examine endolysosomal processing of α3(IV)NC1 under controlled conditions, we used the system described by Watts et al. (11,14). Antigen was incubated at 37°C with freshly disrupted lysosomes that were purified from human EBV-transformed B cells and degradation followed by SDS-PAGE and mass spectrometric peptide analysis.
SDS-PAGE (Figure 1) demonstrated that intact α3(IV)NC1 was progressively depleted during the 120-min incubation with early appearance (by 5 min) of two strongly stained fragments (at approximately 22 and 9 kD) and subsequent appearance of two additional distinct bands (at 4.5 and 6 kD) and numerous weaker bands. It was striking that several large fragments were momentarily very distinct in the digests because it suggested that processing might proceed to some extent along a consistent pathway such that fragments that were relatively resistant to further processing could accumulate.
Cathepsin D/E Activity Is Required to Unlock α3(IV)NC1 to Further Lysosomal Processing
To identify the responsible proteases, we determined the influence of class-specific protease inhibitors on the appearance of early fragments during lysosomal processing. Pepstatin A, an inhibitor of aspartate proteases, but not inhibitors of cysteine or serine proteases prevented formation of early fragments (Figure 2A). Indeed, in the presence of pepstatin A, no discernible digestion of α3(IV)NC1 had occurred by 120 min, even though lysosomes contain a rich array of proteases that are not sensitive to this protease inhibitor. Therefore, processing of α3(IV)NC1 by the host of proteases in human B cell lysosomes requires a critical aspartate protease activity that unlocks the otherwise protease-resistant molecule for further processing by lysosomal proteases.
Cathepsin D and cathepsin E are the major pepstatin A–sensitive aspartate proteases that are known to occur within lysosomes. We chose first to investigate cathepsin D because it is more prevalent than cathepsin E, and it had a similar but slightly wider specificity for substrates within α3(IV)NC1 when the two proteases were compared with α3(IV)NC1 peptides (manuscript in preparation). Incubation of α3(IV)NC1 with purified human cathepsin D yielded fragments with very similar sizes and time course of appearance as with lysosomal extracts (Figure 1). Moreover, Edman degradation of lysosome and cathepsin D–generated fragments that appeared similar by SDS-PAGE showed identical or probably identical N-terminal amino acid sequences (Table 1, Figure 3B). For example, the N-terminal sequence that was determined for the 22-kD band 1 that was recovered from the lysosomal and cathepsin D digests was in both cases FVQGN. Similarly, the sequences of the heavier components of the 9-kD band 2 (band 2a) were identical and those of the 6-kD (band 3) and 4.5-kD (band 4) bands probably identical from both types of digest. The N-terminal sequences of the lighter approximately 9-kD bands 2 (band 2b) were TSAGS from the lysosome digest and IMFTSAGS from the cathepsin D digest. Because the TSAGS fragment did not occur in lysosomal digests that were treated with pepstatin A, and neither cathepsin D nor E cuts α3(IV)NC1 peptides at IMFTSAG (manuscript in preparation), it is likely that the TSAGS fragment was generated by an initial cathepsin D/E cleavage generating IMFTSAGS, followed by removal of the IMF residues by one of the exopeptidases that are active in lysosomal extracts.
Therefore, lysosomal processing of α3(IV)NC1 proceeds only after initial unlocking proteolytic events that are mediated by a pepstatin A–sensitive lysosomal protease. Because the unlocking protease is either cathepsin D itself or a protease with indistinguishable substrate specificity, we hereafter refer to this activity simply as cathepsin D.
Major Epitopes Recognized by Patients’ Autoreactive T Cells Are Destroyed during Early Processing In Vitro
Next we inferred the sites of early processing of α3(IV)NC1 and examined their location relative to epitopes that are recognized by patients’ T cells (Figure 3, A and B). Cleavage at SFL38 FVQ was clearly identified as an early event that gives rise to the N-termini of two prominent early fragments and very likely the C termini of two other prominent fragments. Note that the C-termini of the larger fragments were not directly determined but located to within 5 to 10 amino acids from their known N-termini, and from their molecular weights estimated by SDS-PAGE. Similarly, cleavages at FSF147 IMF, LFC69 NVN, and SFW196 LAS were inferred to be early processing events. Strikingly, two of the early cleavages mapped within two of the four peptides that stimulated patients’ T cells. Specifically, the peptide bond SFL38 FVQ is within dap31–50 and FSF147 IMF within dap131–150, potentially destroying T cell epitopes.
Because early processing seemed to correlate with the specificities of patients’ autoreactive T cells, we sought to identify more fully the earliest cleaved peptide bonds in α3(IV)NC1 by examining the digests for small α3(IV)NC1 fragments that are invisible to SDS-PAGE. For these experiments, α3(IV)NC1 was incubated with purified lysosomal proteases rather than entire lysosomal extracts because of the difficulty in discerning peptide fragments of α3(IV)NC1 in complex lysosomal digests. Cathepsin D was chosen because of its important role in early α3(IV)NC1 processing and AEP because it is crucial in early processing of other globular antigens, including tetanus toxoid c-fragment and myelin basic protein (MBP) (15).
To infer the earliest cut peptide bonds, we identified α3(IV)NC1 peptides that were released after just 5 min of digestion with cathepsin D. Sequence was determined for 23 α3(IV)NC1 peptides, shown in full in Supplementary Information and diagrammatically in Figure 3C. Twelve of the early-released peptides had NH2- or COOH-termini generated by the already-identified early cleavages (10 by SFL38 FVQ, one by FSF147 IMF, and one by SFW196 LAS). Ten had sequences that began FVQ (within dap31–50), re-confirming the SFL38 FVQ peptide bond as highly preferred by cathepsin D, but all had different C-termini, suggesting that cathepsin D has little preference among several potentially scissile peptide bonds in α3(IV)NC1 immediately C-terminal to SFL38 FVQ. In contrast, only one peptide fragment was found indicative of the early cleavages FSF147 IMF and SFW196 LAS, suggesting that cathepsin D has strong preference to make subsequent cuts at particular nearby peptide bonds, specifically, ISL141 WKG N-terminal to FSF147 IMF releasing the fragment WKGFSF and ERM205 FRK C-terminal to SFW196 LAS releasing the peptide LASLNPERM. The peptide bonds ISL141 WKGFSF147 IMF were never found intact in any fragment smaller than 22 kD, so it is likely that both peptide bonds are similarly highly susceptible to early proteolysis. Also, one of the prominent early-released peptides indicated that cleavage at RGF10 VFT, destroying dap1–20, is an early event and possibly indispensable because no peptide was found with RGF10 VFT intact.
Taken together, the data identified 11 peptide bonds that were cleaved in early processing (Figure 3D). Among them, they destroy all conceivable DR15-binding subsequences of all 4 α3(IV)NC1 peptides that stimulate most patients’ T cells. Importantly, no fragment of α3(IV)NC1 smaller than 6 kD was found to contain any of the T cell epitopes intact.
The techniques similarly were applied to investigate α3(IV)NC1 processing by AEP. Four peptides were released by AEP treatment of intact α3(IV)NC1, indicative of cuts after five of the nine asparagines in α3(IV)NC1 (Figure 4). Remarkably, all of the scissile asparagines were in the vicinity of (two of five) or within (three of five) the sequence of the T cell self-epitope that contained dap71–90. The other four asparagines are presumably less accessible to AEP within intact α3(IV)NC1 because all nine asparagines were substrates for AEP when presented in the form of short synthetic peptides (data not shown). Note that although AEP released fragments from intact α3(IV)NC1, its action was insufficient to unlock α3(IV)NC1 in the presence of pepstatin A (Figure 2). This contrasts with lysosomal processing of tetanus toxoid c-fragment, in which AEP action alone is sufficient to unlock the antigens to further processing (11).
Aspartate Protease Activity within Intact Human APC Affects Presentation of Goodpasture Antigen
Next we investigated whether the propensity of lysosomal aspartate protease activity to destroy α3(IV)NC1 T cell epitopes in vitro had discernible consequences in vivo. To detect presentation of α3(IV)NC1 peptides by human APC, we made α3(IV)NC1 peptide-specific DR15-restricted murine T cell hybridomas in mice that were engineered to express HLA DR15 (donated by Dr. Altmann and described in reference ). Nine hybridomas, including ha3p132.2, 132.3, and 133.6, had the properties of responding to dap131–150 with fine specificity that was indistinguishable (with 15mer peptides overlapping by 10) from patients’ T cells (Figure 5). DR15 restriction was confirmed by demonstrating responses to dap131–150 presented by only DR15-expressing B cell lines (data not shown). Analysis of all possible interactions between the common sequence of the stimulatory peptides and DR15 molecules found only one orientation with high predicted affinity (5), so the hybridomas and patients’ T cells almost certainly recognize the peptide α3(IV)NC1139–148 with sequence ISLWKGFSFI engaging pockets in the peptide binding groove as shown in Table 2.
The hybrids were first used to investigate whether B cells that were incubated with intact α3(IV)NC1 were able to present the ISLWKGFSFI epitope at all, in view of its containing two peptide bonds that are known to be cut in early processing by B cell lysosomes in vitro. As shown in Figure 5D, B cells that were incubated with abundant (200 μg/ml, approximately 7 μM) α3(IV)NC1 were able to present ISLWKGFSFI, as assessed by IL-2 production by ha3p132.2. Presentation was processing dependent because it was inhibited by fixation (Figure 5D) or chloroquine-treatment of the B cells (data not shown). Therefore, processing of α3(IV)NC1 within intact cells does generate some ISLWKGFSFI, which is presented on DR15 in the conformation that is recognized by ha3p132.2.
Next the hybrids were used to interrogate the level of ISLWKGFSFI presentation on the surface of living APC that were treated with pepstatin A. ISLWKGFSFI presentation from intact α3(IV)NC1 was enhanced two- to four-fold by 10 mM pepstatin A treatment of the APC. This was not an APC-specific or T hybrid–specific phenomenon because it was observed for all of the DR15-expressing human B cell lines that were studied and for a human macrophage cell line, using any of three of our ISLWKGFSFI-specific T hybridomas (Figure 6A). The effect of pepstatin A on presentation of peptide varied from no effect, as expected for processing-independent presentation of peptide, to up to two-fold enhancement. Enhancement was most striking for ThP1 cells and almost certainly related to the high levels of free cathepsin D that were detectable in medium that was conditioned by these cells (data not shown). The effects of pepstatin A were in striking contrast to previous reports of the effect of pepstatin A (17–19) that have shown epitope-specific effects that vary between strong inhibition of presentation and no discernible effect, thought to reflect the varying importance of aspartate proteases in generating particular epitopes (Figure 6A, bottom 8 pairs of bars). The effect of pepstatin A was further examined by determination of the time course and dosage dependence of ISLWKGFSFI presentation by the LCL061 B cell line interrogated with the ha3p132.2 T hybrid. Presentation of ISLWKGFSFI from intact α3(IV)NC1 was two- to three-fold less than from molar equivalent quantities of ISLWKGFSFI-containing peptide and exhibited dosage-dependent enhancement by pepstatin A (Figure 6, B and C). Taken together, the results indicate that the aspartate protease activity in two different human APC types is sufficient to reduce presentation of the ISLWKGFSFI epitope, at least under cell culture conditions.
The results demonstrate that the level of aspartate protease activity within human EBV-transformed B cells is sufficient to diminish substantially the presentation of a key epitope in Goodpasture’s disease, despite the epitope’s having high affinity for HLA DR15 molecules. This echoes the observation of Manoury et al. (20) that AEP activity within APC destroys a key epitope in multiple sclerosis and adds credence to the hypothesis that destructive processing may direct the specificities of autoreactive T cells by preventing constitutive presentation of self-epitopes at sufficient levels to drive secure self-tolerance (10,21). Whereas AEP activity seems critical in shaping autoimmunity to MBP, our data suggest that aspartate proteases, almost certainly cathepsin D, are critical in shaping autoimmunity to α3(IV)NC1. It is interesting that a prominent role for cathepsin D in destructively limiting antigen presentation was also found in a study of myoglobin processing (22). Thus, cathepsin D cleavage of α3(IV)NC1 and AEP cleavage of MBP are antigen-processing mechanisms that diminish presentation of potential epitopes within antigens, analogous to the processing mechanisms that are thought to account for certain epitopes’ behaving as poorly presented “cryptic epitopes” in animal models (23).
We suggest that the influence of destructive processing could run much deeper. In the case of α3(IV)NC1, all of the major self-epitopes contain highly scissile peptide bonds, and fragments that are indicative of cuts within all but one were detected at the earliest stages of processing. Therefore, destructive processing could diminish presentation of all of the important self-epitopes in α3(IV)NC1. However, some anomalies from our previous work that examined peptides that were eluted from DR15 molecules indicate that other processing factors must also be influential. For example, we eluted from α3(IV)NC1-pulsed B cells peptides that contained intact the highly scissile sequence SFL38 FVQ (8). The peptides comprised a nested set that seemed to bind to DR15 via VPLYSGFSF38. A possible explanation is suggested from consideration of how DR15 might interact with intact α3(IV)NC1. The DR15-binding VPLYSGFSF core sequence is almost entirely exposed on the surface of α3(IV)NC1, where it might be able to bind DR15 before substantial proteolysis, gaining steric protection for YSG34 FSF and SFL38 FVQ, and presumably directing processing via a different route (24).
If indeed destructive processing does suppress constitutive presentation of some of the epitopes that are recognized by patients’ T cells in autoimmune disease, how are those epitopes ever presented to drive pathogenesis? The results in the context of previous studies of α3(IV)NC1 processing suggest an explanation, at least for the ISLWKGFSFI epitope for which the data are most complete. Peptides that contain this epitope have the highest affinity of all of the α3(IV)NC1 peptides that we have studied but are not presented by DR15-bearing B cells at sufficient level for biochemical detection, whereas other lower DR15-affinity peptides are. The new data demonstrate that rapid destruction by endosomal aspartate proteases can account for the deficit of ISLWKGFSFI but, importantly, that sufficient ISLWKGFSFI to stimulate T cells can be presented by DR15-bearing B cells under particular conditions, such as culture in high concentrations of α3(IV)NC1, and that partial inhibition of the activity of endosomal aspartate proteases is sufficient to increase its presentation substantially. Therefore, presentation of ISLWKGFSFI is likely to be low level rather than absent and so could be a sufficient stimulus to drive T cells that are already activated by, for example, a cross-reactive epitope from an infecting organism. Moreover, the level of ISLWKGFSFI presentation will be modulated by the balance of positive factors, such as the abundance of α3(IV)NC1 and possibly the level of DR15 expression, and negative factors, including the activity of endosomal aspartate proteases. It is notable in this regard that many of the stimuli that are reported to trigger Goodpasture’s disease would be expected to increase turnover of basement membrane α3(IV)NC1 (25), and the levels of protease activity within APC are reported to vary with cell type and activation status (26).
This is the second report to associate destructive processing with the specificity of autoreactive T cells, so the question arises as to the general importance of the mechanism. The two autoantigens, MBP and α3(IV)NC1, share several features. Both are tightly folded, extensively disulfide-bonded, protease-resistant cationic globular molecules that are processed only by lysosomes after unlocking cleavages at particular peptide bonds. It is striking that in both cases, the unlocking cleavages destroy peptides that have high affinity for HLA DR15 and are major targets of autoimmune attack. This set of circumstances may not be rare, because many autoantigens are globular with disulfide bonds, and a requirement for unlocking cleavages has now been reported for several proteins. Clearly, evaluation of the general importance of destructive processing in defining the specificity of autoreactive T cells will require analysis of processing of more autoantigens, but of all of the ways processing could influence presentation, unlocking deserves earnest focus because it more than any other processing event is likely to be a consistent occurrence that could credibly influence the scope of self-tolerance that is built up during a life time and credibly be subverted, in a stimulating milieu, only infrequently to precipitate autoimmune disease at close to the low frequency with which disease occurs in human.
This work was supported by grants from the Medical Research Council of the UK.
DR15 transgenic mice were donated by Dr. D. Altmann (Imperial College, London, UK).
Published online ahead of print. Publication date available at www.jasn.org.
1. Saus J, Wieslander J, Langeveld J, Quinones S, Hudson B: Identification of the Goodpasture antigen as the alpha3(IV) chain of collagen IV. J Biol Chem 263: 13374–13380, 1988
2. Derry CJ, Ross CN, Lombardi G, Mason PD, Rees AJ, Lechler RI, Pusey CD: Analysis of T cell responses to the autoantigen in Goodpasture’s disease. Clin Exp Immunol 100: 262–268, 1995
3. Cairns LS, Phelps RG, Bowie L, Hall AM, Saweirs WWM, Rees AJ, Barker RN: The fine specificity and cytokine profile of T helper cells responsive to a glomerular autoantigen in Goodpasture’s Disease. J Am Soc Nephrol 14: 2801–2812, 2003
4. Phelps RG, Jones V, Turner AN, Rees AJ: Properties of HLA class II molecules divergently associated with Goodpasture’s disease. Int Immunol 12: 1135–1143, 2000
5. Phelps RG, Rees AJ: The HLA complex in Goodpasture’s disease: A model for analyzing susceptibility to autoimmunity. Kidney Int 56: 1638–1654, 1999
6. Wong D, Phelps RG, Turner AN: The Goodpasture antigen is expressed in the human thymus. Kidney Int 60: 1777–1783, 2001
7. Salama AD, Chaudhry AN, Ryan JJ, Eren E, Levy JB, Pusey CD, Lightstone L, Lechler RI: In Goodpasture’s disease, CD4(+) T cells escape thymic deletion and are reactive with the autoantigen alpha3(IV)NC1. J Am Soc Nephrol 12: 1908–1915, 2001
8. Phelps RG, Turner A, N, Rees AJ: Direct identification of naturally processed autoantigen derived peptides bound to HLA-DR15. J Biol Chem 271: 18549–18553, 1996
9. Phelps RG, Jones V, Coughlan M, Turner A, N, Rees AJ: Presentation of the Goodpasture autoantigen to CD4 T cells is influenced more by processing constraints than by HLA class II peptide binding preferences. J Biol Chem 273: 11440–11447, 1998
10. Sercarz EE: Processing creates the self. Nat Immunol 3: 110–112, 2002
11. Antoniou AN, Blackwood SL, Mazzeo D, Watts C: Control of antigen presentation by a single protease cleavage site. Immunity 12: 391–398, 2000
12. Zou J, Turner AN, Phelps RG: Mass and Composition (MAC) MALDI TOF analysis enables inference of the sequence of most peptides where the protein of origin is known. Anal Chem 75: 2653–2662, 2003
13. Harwell L, Skidmore B, Marrack P, Kappler J: Concanavalin A-inducible, interleukin-2-producing T cell hybridoma. J Exp Med 152: 893–904, 1980
14. Watts C, Matthews SP, Mazzeo D, Manoury B, Moss CX: Asparaginyl endopeptidase: Case history of a class II MHC compartment protease. Immunol Rev 207: 218–228, 2005
15. Manoury B, Hewitt EW, Morrice N, Dando PM, Barrett AJ, Watts C: An asparaginyl endopeptidase processes a microbial antigen for class II MHC presentation. Nature 396: 695–699, 1998
16. Altmann DM, Douek DC, Frater AJ, Hetherington CM, Inoko H, Elliott JI: The T cell response of HLA-DR transgenic mice to human myelin basic protein and other antigens in the presence and absence of human CD4. J Exp Med 181: 867–875, 1995
17. Bennett K, Levine T, Ellis JS, Peanasky RJ, Samloff IM, Kay J, Chain BM: Antigen processing for presentation by class II major histocompatability complex requires cleavage by cathepsin E*. Eur J Immunol 22: 1519–1524, 1992
18. Musson JA, Walker N, Flick-Smith H, Williamson ED, Robinson JH: Differential processing of CD4 T-cell epitopes from the protective antigen of Bacillus anthracis. J Biol Chem 278: 52425–52431, 2003
19. von Delwig A, Hilkens CM, Altmann DM, Holmdahl R, Isaacs JD, Harding CV, Robertson H, McKie N, Robinson JH: Inhibition of macropinocytosis blocks antigen presentation of type II collagen in vitro and in vivo in HLA-DR1 transgenic mice. Arthritis Res Ther 8: R93, 2006
20. Manoury B, Mazzeo D, Fugger L, Viner N, Ponsford M, Streeter H, Mazza G, Wraith DC, Watts C: Destructive processing by asparagine endopeptidase limits presentation of a dominant T cell epitope in MBP. Nat Immunol 3: 169–174, 2002
21. Watts C, Moss CX, Mazzeo D, West MA, Matthews SP, Li DN, Manoury B: Creation versus destruction of T cell epitopes in the class II MHC pathway. Ann N Y Acad Sci 987: 9–14, 2003
22. Moss CX, Villadangos JA, Watts C: Destructive potential of the aspartyl protease cathepsin D in MHC class II-restricted antigen processing. Eur J Immunol 35: 3442–3451, 2005
23. Moudgil KD, Sercarz EE: The self-directed T cell repertoire: Its creation and activation. Rev Immunogenet 2: 26–37, 2000
24. Sundaramoorthy M, Meiyappan M, Todd P, Hudson BG: Crystal structure of NC1 domains. Structural basis for type IV collagen assembly in basement membranes. J Biol Chem 277: 31142–31153, 2002
25. Phelps RG: Immune recognition of renal autoantigens. Exp Nephrol 8: 226–234, 2000
26. Delamarre L, Pack M, Chang H, Mellman I, Trombetta ES: Differential lysosomal proteolysis in antigen-presenting cells determines antigen fate. Science 307: 1630–1634, 2005