Although the binding sites of MHC class II molecules can accommodate longer ligands, peptides of 15 to 20 residues are the primary form of processed Ag recovered from class II dimers isolated from living cells. These peptides are derived from intact Ags by proteolysis in endocytic organelles, where binding to class II dimers also occurs. Whether generation of these short peptides typically precedes association with class II molecules, or whether class II molecules initially bind to unfolded proteins or large protein fragments, followed by degradation of the unprotected regions, remains unknown. Here we report the identification of an SDS-stable, long-lived, 120-kDa complex composed of two class II dimers bound to a common large Ag fragment. This complex is produced within the endocytic pathway from newly synthesized MHC class II molecules following exposure of the cells to exogenous hen egg lysozyme. These data suggest that a major pathway of Ag processing involves the initial binding of class II heterodimers to large protein substrates upon exposure of regions with suitable motifs, followed by cleavage and/or trimming of the exposed protein around this bound region. This sequence of events during Ag processing may provide a partial molecular explanation for the immunodominance of certain determinants in protein Ags.
T lymphocytes with αβ receptors recognize antigenic peptides associated with MHC class I or class II molecules, with the latter preferentially displaying peptides derived from degradation of proteins accessing the endocytic pathway (1, 2). Studies of the peptides eluted from immunochemically isolated MHC class II molecules have found these ligands to be fairly short, predominantly 15 to 20 residues in length (3, 4, 5, 6, 7, 8). Yet one of the most striking structural differences between MHC class I and class II molecules is the open ends of the binding groove of the latter that allows them to readily associate with ligands substantially longer than these short eluted peptides (9, 10, 11, 12, 13, 14). Consistent with this structure, mature class II molecules on the cell surface have been found to bind intact unfolded proteins, proteins with an extended native conformation, or long synthetic peptides, and in cells lacking invariant chain, class II dimers can be found associated with newly synthesized intracellular proteins (15, 16, 17, 18, 19). Another striking example of this type of interaction involves the invariant chain itself, which associates as an intact protein with the class II binding region via a stretch of amino acids termed CLIP,2 class II-associated Ii peptide (20, 21, 22, 23, 24, 25). In addition, the peptides eluted from class II molecules often include nested sets involving the same core sequence flanked by a variable number of amino- or carboxyl-terminal amino acids (4, 5, 6, 26), suggesting that once bound to a class II molecule, ligands undergo additional trimming by amino- and carboxyl-peptidases, presumably while resident in endocytic organelles. This possibility agrees with pooled sequence data on eluted peptides showing a high frequency of proline at the second position from the amino terminus (27). A proline in this position interferes with the action of many amino-peptidases.
Despite these structural data and demonstrations of the ability of class II molecules to bind large protein ligands, current models of endocytic Ag processing do not focus on this property of MHC class II dimers. Instead, it is generally assumed that proteins accessing the endocytic pathway are denatured and cleaved to relatively short fragments that are then captured in accordance with their affinity for the polymorphic structure of available class II molecules and the influence of DM or H-2M on their stable association with this class II dimer (28, 29, 30, 31, 32). Some trimming of the ends of those captured peptides just slightly larger than the MHC binding domain then presumably occurs, giving rise to the nested sets of peptides seen in elution experiments. This view derives in part from the binding of short exogenous peptides to MHC class II molecules in in vitro experiments (33, 34) and the ability of such synthetic peptides to effectively stimulate T cells (35, 36), and in part from the characteristics of the terminal products of the processing pathway seen in the elution studies (3, 4, 5, 6, 7, 8).
While no biochemical data exist that directly challenge this model, some functional results suggest an alternative view. Deng et al. (37) have described studies of hen egg lysozyme (HEL) presentation in cells coexpressing two different class II molecules. The results of these functional experiments have been interpreted as suggesting that Ag pieces substantially larger than terminal peptides are the substrate for binding to class II in endosomes. A prediction of this model is that during active Ag processing, class II molecules newly freed of invariant chain should be found associated with intact proteins or large protein fragments, rather than short peptides. Such intermediates, however, would be very hard to identify in the very proteolytically active environment of late endosomes and lysosomes in which most Ag processing and binding to MHC class II molecules occur (38, 39, 40, 41, 42, 43) because they are likely to be produced asynchronously and have very short lifetimes. In addition, even if they formed SDS-stable complexes with metabolically labeled class II molecules (44) or were labeled themselves at sites not trimmed away during early processing steps (45), they would create a heterogeneous array difficult to visualize by autoradiography. The study of unlabeled eluted peptides from purified class II proteins does not help in this regard, as >90% of these peptides derive from mature class II dimers present on the cell membrane. These molecules have resided for several hours in endocytic organelles before initial surface display and then have been exposed to either extracellular peptidases or endocytic enzymes during recycling events (46, 47, 48), providing ample opportunity for any bound ligand to be trimmed to a short length.
Over the past several years, this laboratory has used the Ak-HEL combination as a model system for biochemical analysis of Ag processing in a variety of cell types (43, 44, 49). During the course of these studies, a change in the protocol used to culture B lymphoblasts with Ag led to the appearance of an SDS-stable 120-kDa structure in metabolic labeling, pulse-chase experiments. Because of its possible relationship to the crystallographic class II “dimer of dimers” (12, 13) and the putative class II superdimers reported by several groups (50, 51), we attempted to identify the components of this class II molecule-containing complex. We report here that this material is a trimer of two isotypically distinct class II molecules (AαkAβk and EαkEβk) bound to a large HEL fragment. This complex is detectable only when splenocytes are fed native HEL, it forms intracellularly in late endocytic organelles, and it persists for several hours. These biochemical data support the conclusion of functional studies reported by Deng et al. (37) and argue that a major pathway of active Ag processing involves the binding of newly synthesized class II molecules to unfolded proteins or large protein fragments in late endosomal/lysosomal organelles.
Materials and Methods
Cell labeling and immunoprecipitation
Single cell suspensions were prepared from the spleens of CBA/J (H-2k) mice. Unless otherwise indicated, cells (15 × 106/ml) were cultured at 37°C for 1 h in medium lacking leucine (RPMI medium kit, Life Technologies, Gaithersburg, MD), pulsed with [3H]leucine (New England Nuclear, Boston, MA; 168 Ci/mmol) at 0.5 mCi/ml for 30 min, and chased at 37°C in the presence of 10× cold leucine for the specified time. In all the experiments shown, 2-ME was omitted from the culture medium. All the incubations were performed in the presence of 10% FBS; where specified, during the prelabeling, labeling, and chase periods, the spleen cells were cultured in the additional presence of 2 mg/ml HEL (Sigma, St. Louis, MO) or 100 μM peptide (HEL46–61, HEL1–18, and PCC86–104, synthesized by F-moc chemistry and >95% pure; National Institute of Allergy and Infectious Disease Peptide Synthesis Facility, Rockville, MD). To examine possible binding of HEL to cell surface MHC class II molecules, cells were labeled and chased for 4 h before addition of 2 mg/ml HEL for an additional 2 h in the presence or the absence of 300 μM chloroquine to inhibit endosomal acid proteases.
Cells were lysed using standard lysis buffer containing 0.5% Nonidet P-40, and the samples were precleared twice with Pansorbin (Calbiochem, La Jolla, CA) and once with protein A-Sepharose (Pharmacia, Piscataway, NJ) (44). MHC class II molecules were immunoprecipitated from the precleared lysates using mAbs 10-2.16 (52), Y17 (53), and 14.4.4S (54) prebound to protein A-Sepharose beads. Immunoprecipitated material was eluted from the beads in SDS sample buffer containing either 2-ME (reducing conditions) or 10 mM iodoacetamide (nonreducing conditions). Samples from individual immunoprecipitates were divided into two aliquots and eluted either without sample boiling or with boiling. The eluted samples were analyzed by SDS-PAGE using 10% polyacrylamide gels. Gels were fixed, treated with EnHance (Amersham, Arlington Heights, IL), dried, and exposed to Kodak XAR-5 film (Eastman Kodak, Rochester, NY) at −70°C.
Cell labeling and fractionation
Cells (300 × 106 in 50 ml) were cultured and metabolically labeled as described above in the presence of 2 mg/ml HEL. The labeled cells were then disrupted by nitrogen cavitation in 5 ml of homogenizing buffer (1× = 0.25 M sucrose and 1 mM EDTA, pH 6.8) and a postnuclear supernatant prepared by low speed centrifugation (1,000 × g for 5 min at 4°C). Twenty-five milliliters of Percoll (27%), prepared as previously described (43), were layered onto a cushion of 5 ml of 2.5 M sucrose. Five milliliters of postnuclear supernatant were layered onto the Percoll, and centrifugation was conducted in a VTi50 vertical rotor (Beckman Instruments, Palo Alto, CA) for 1 h at 34,500 × g at 4°C. Fractions of 1 ml were collected from the bottom of the gradient using a Beckman fraction recovery system. The marker enzymes β-hexosaminidase (characteristic of endocytic pathway organelles, especially lysosomes) were assayed in each fraction using published procedures (43). Fractions comprising the light density peak of β-hexosaminidase activity were pooled and refractionated on a 10% Percoll gradient using the same methods. Percoll was removed by high speed centrifugation (100,000 × g for 30 min), and the recovered organelles and membranes were lysed in lysis buffer. These lysates were then used for immunoprecipitation and SDS-PAGE. The resulting autoradiographs were analyzed using a Molecular Dynamics densitometer (Sunnyvale, CA) to determine the total amount of compact SDS-stable dimer and 120-kDa complex in the total set of fractions and in subsets of fractions known to contain distinct endocytic organelles, as determined by marker analysis (43).
Cell surface biotinylation
Splenocytes (10 × 106) were cultured overnight in the presence of HEL (2 mg/ml). After culture the cells were washed extensively with cold PBS and incubated with PBS containing 10 mM NHS-biotin (Pierce, Rockford, IL) for 20 min on ice. The reaction was terminated by washing the cells four times with blocking buffer (PBS containing 50 mM NH4Cl, 1 mM MgCl2, and 0.1 mM CaCl2). Cells were then lysed using standard lysis buffer containing 0.5% Nonidet P-40, and the samples were precleared as previously described. MHC class II molecules were immunoprecipitated from the precleared lysates using mAb Y17. Immunoprecipitated material was eluted from the beads in SDS sample buffer under nonreducing conditions without sample boiling. The eluted samples were separated using 10% polyacrylamide gels and then transfered to nitrocellulose. The membrane was incubated with streptavidin-HRP (Vector), and the biotinylated proteins were detected by enhanced chemiluminescence (ECL, Amersham).
HEL iodination and detection in MHC class II molecules
HEL (1 mg) or HEL46–61 (100 μg) was dissolved in PBS to a final volume of 100 μl and mixed with Iodo-Beads (Pierce). Typically, 1 mCi of 125I was used for each reaction. Free 125I was separated from HEL or HEL46–61 by gel filtration through Bio-Gel P6 or Bio-Gel P2 (Bio-Rad, Hercules, CA), respectively. The average sp. act. of the labeled HEL was 7 μCi/nmol; that of the peptide was 700 μCi/nmol.
CBA/J spleen cells (25 × 106) were pulsed with [3H]leucine for 1 h and chased overnight in the presence of 2 mg/ml of cold HEL as described above. A similar number of spleen cells was incubated overnight with 2 mg of [125I]HEL or with 100 μg of [125I]HEL46–61 in a final volume of 3 to 4 ml of complete medium. After culture and cell lysis, MHC class II molecules were immunoprecipitated using the mAb 10-2.16. Samples from individual immunoprecipitates were divided into two aliquots and eluted either without sample boiling or with boiling. The eluted samples were analyzed by SDS-PAGE using 10% polyacrylamide gels or 10 to 20% Tris-tricine gels. Gels were fixed and either treated with EnHance (Amersham), dried, and exposed to Kodak XAR-5 film or dried, exposed in a PhosphorImager cassette, and analyzed using ImageQuant software (Molecular Dynamics).
MHC class II purification
Spleen cells (1 × 109) were incubated overnight in culture medium (RPMI plus 10% FBS) in the presence or the absence of 2 mg/ml HEL. Cells were lysed, and the lysates were precleared as described above. I-Ak or I-Ek class II molecules were then purified by immunoaffinity chromatography. Cleared lysates were passed over immunoaffinity columns consisting of either the Ak-specific mAb 10-2.16 or the Ek-specific mAb Y17 covalently bound to activated cyanogen bromide-Sepharose beads (Pharmacia, Piscataway, NJ). After extensive sequential washing with PBS containing octyl-β-d-thioglucopyranoside at 15 μg/ml and 0.45 M NaCl/octyl-β-d-thioglucopyranoside, the columns were eluted in 2 ml of 0.1% trifluoroacetic acid (TFA). Eluted fractions were further acidified by addition of 200 μl of glacial acetic acid and boiled for 5 min. No attempt was made to separate the eluted proteins and peptides by size filtration before further analysis. Samples were concentrated by lyophilization to a final volume of 250 μl before reverse phase HPLC separation using a narrow-bore Vydac C18 column (150 × 2.1 mm, 5 μm, 330Å pore size). Peptides were eluted by means of a gradient of 5% acetonitrile in 0.1% TFA for 10 min followed by a linear increase to 60% over 30 min. The flow rate was 250 μl/min, and the fraction size was 250 μl. Elution was monitored at 220 nm. Individual fractions were collected, dried, and stored at −20°C before mass spectral analysis.
Fractions 15 to 40 were analyzed by matrix-assisted laser desorption ionization mass spectrometry. Dried samples were resuspended in 5 μl of acetonitrile/0.1% TFA (1/1, v/v), and 0.3 μl was applied to the sample slide, mixed with an equal volume of matrix (α-cyano-4-hydroxycinnamic acid in acetonitrile/0.1% TFA (1/1, v/v)), and allowed to air-dry before data collection. Mass spectra were acquired on a Voyager RP mass spectrometer (Perseptive Biosystems, Wathertown, MA). Oxidized bovine insulin B chain (MH+3496.9) or human adrenocorticotropic hormone fragment 7 to 38 (MH+3660.2) were used as internal and external standards for mass calibration.
Identification of a high m.w., SDS-stable complex containing recently synthesized MHC class II molecules
We have previously used as a model for the analysis of Ag processing the well-studied protein Ag HEL and normal CBA B lymphoblasts expressing the Ak and Ek class II molecules, each known to bind several distinct determinants within this small globular protein Ag (52). One reason for this choice was the formation of a large cohort of SDS-stable class II dimers containing metabolically labeled class II molecules in B lymphoblasts exposed to HEL (44, 53). This Ag-induced change in the biochemical behavior of class II dimers allowed us to track the formation of class II molecule-ligand complexes in intracellular organelles (43). In these previously reported studies, we observed only the α- and β-chains of SDS-unstable class II molecules and a 56-kDa stable form of peptide-loaded class II dimers (compact SDS-stable dimers) upon PAGE of immunoprecipitated labeled molecules.
When we modified our culture conditions, we noticed that addition of native HEL to CBA splenocytes leads to the appearance in pulse-chase experiments of a significant amount of 120-kDa material that is stable in SDS buffer without heating and can be precipitated by any of several mAbs to Ak or Ek molecules (Fig. 1,B). Boiling these samples reveals only labeled proteins corresponding to the α- and β-chains of MHC class II molecules (Fig. 1,B, lanes with stars). With CBA splenocytes cultured only in FBS-containing medium, we are unable to detect these 120-kDa forms (Fig. 1,A) even after overexposure of the autoradiographs (data not shown). The 120-kDa material can be observed as early as 45 min of chase after 30 min of metabolic labeling in the continuous presence of HEL; the amount of the 120-kDa form increases over the next 2 to 4 h of chase, and a large fraction persists after overnight chase in the absence of exogenous HEL (Fig. 2). This latter observation indicates that the complex is rather stable and is not subject to rapid degradation by intracellular or extracellular proteases.
The relevant change in condition leading to the appearance of these new forms appears to be the elimination of reducing agents from the cell cultures; the 120-kDa form is seen in cultures that lack 50 μM 2-ME, but not in cultures containing this agent or in cultures without 2-ME exposed to reduced, rather than native, HEL (data not shown). However, the 120-kDa forms are seen in SDS-PAGE run without sample heating regardless of whether the immunoprecipitation samples are eluted in the presence of reducing agents (Fig. 1 C), indicating that the effect of reduction on the appearance of this species involves events before its formation in the cell.
Characterization of the components of the 120-kDa complexes containing MHC class II molecules
Although several possible molecular models could explain the HEL-dependent appearance of a 120-kDa complex containing newly synthesized class II molecules, only a few seemed likely given the data presented above. Intact invariant chain is unlikely to be a part of the complex, as no labeled proteins corresponding to the 31- or 41-kDa form of Ii are seen in heated samples (Fig. 1, A–C), and invariant chain in newly assembled complexes with class II heterodimers labels to a greater extent than the class II chains, even using the [3H]leucine employed here (43, 44). In addition, elution of labeled 120-kDa material from a first gel, followed by reanalysis in a second gel after boiling, shows no labeled components other than the class II α- and β-chains, even after long exposures (data not shown). The 120-kDa complex is also unlikely to be an aggregate of two misfolded class II molecules produced early in biosynthesis, because the chains seen in samples from late chases have mobilities previously shown to correspond to class II subunits with mature N-linked glycans (43, 44, 54). Aggregated class II proteins tend to remain in the endoplasmic reticulum and not to show such glycan modification (55). The stability in SDS buffer also argues against such homoaggregation or the binding of a single class II heterodimer to aggregates of HEL as an explanation, as these would be unlikely to remain associated under such conditions. Finally, association of a single class II dimer with a single intact HEL molecule would not account for the mobility of this material in SDS-PAGE, as HEL is only 14 kDa.
The probable components of the complex are thus either two noncovalently associated class II molecules, each engaging a ligand of modest size, or a large, poorly labeled, cellular or unlabeled serum protein interacting with a single class II molecule via its binding domain, accounting in either case for the SDS stability of the complex. In both circumstances, the association would have had to be induced by or contain HEL or HEL-derived fragments. Distinguishing between these possibilities is difficult if the two class II molecules in the first case are identical or if the protein in the latter case is unknown. If, however, the class II molecules in a complex are of different isotypes, this can be determined by several means. To examine whether two distinct types of class II molecule might be present in and account for the properties of the 120-kDa complex, sequential immunoprecipitations were conducted. The labeled proteins from lysates of CBA lymphoblasts exposed to HEL were first precipitated with either anti-Ak or anti-Ek mAb, then reprecipitated with the same reagent to ensure all reactive molecules had been cleared, and material in the precleared lysate was precipitated using the reciprocal Ab (Fig. 3). When either anti-Ak or anti-Ek Abs are used for the initial immunoprecipitation, the 120-kDa band is removed and no longer visible in precipitates using anti-Ek or anti-Ak Abs for the secondary precipitation, respectively. In contrast, neither unstable Ak complexes nor HEL-induced compact Ak dimers running at 56 kDa are removed by the anti-Ek immunoprecipitation and vice versa. Control immunoprecipitations using an anti-class I Ab or an isotype-matched, irrelevant, anti-class II Ab have no effect on the subsequent detection of 120-kDa forms by anti-Ak or anti-Ek Abs, showing the specificity of the reciprocal anti-class II molecule preclearing (Fig. 3). Abs to the Aα-, Aβ-, Eα-, and Eβ-chains give the same results, indicating that all four chains are present in the complex (data not shown). Finally, Western blotting using polyclonal antisera to Aα- or Eα-chains confirms the presence of class II chains of the opposite isotype only in the 120-kDa material (data not shown). Although these data do not rule out the existence of a small number of homotypic class II dimer pairs in these cells, they argue that most, if not all, of the 120-kDa material consists of a pair of SDS-stable class II molecules (one Ak and one Ek) whose association depends on exposure of the cell to HEL.
The 120-kDa complexes cannot be induced by loading class II dimers with short synthetic peptides
Because they are SDS stable and involve two intact class II molecules, the 120-kDa complexes could consist of two class II molecules loaded with short HEL-derived peptides that associate due to an intrinsic property of SDS-stable class II heterodimers (12). This is the structure presumed by several groups who have observed 120-kDa class II-containing material in pulse-chase labeling, immunoprecipitation experiments (50, 51). Alternatively, such complexes could represent two class II molecules bound simultaneously to distinct sites on a large fragment of HEL, which are then held in association due to the peptide bonds of the HEL fragment. The former possibility seemed unlikely due to the disruptive effect of reducing agents added before formation of the complexes but not after, but to directly examine this possibility, CBA B lymphoblasts were cultured in high concentrations of each of several different peptides known to bind avidly to either Ak (HEL46–61) or Ek class II molecules (HEL1–18 or PCC88–104) and to induce the formation of SDS-stable dimers. Compared with class II molecules recovered from cells cultured in medium alone, incubation with a given peptide results in the formation of a large cohort of labeled SDS-stable compact dimers involving either Ak or Ek molecules, but not in the generation of any 120-kDa class II-containing forms (Fig. 4). In contrast, intact HEL induces compact dimer formation to a similar extent as the peptides, and it is also able to stimulate the appearance of the 120-kDa complex. The inability of short peptides to generate the 120-kDa form is seen even when we compensate for the slightly different compact SDS-stable dimer amounts in the whole Ag vs the peptide-exposed preparations (Fig. 4 B). These data argue that the 120-kDa material does not represent noncovalently associated class II molecules containing small processed peptides suitable for cross-linking of the TCR on an individual T cell.
120-kDa complexes are formed in late endocytic organelles, not on the cell surface
Previous studies have shown that mature, surface-expressed or isolated class II molecules can bind directly to denatured intact protein ligands (15, 16, 17, 18, 19). Although reduction of HEL interferes with our ability to induce formation of the 120-kDa material, and this material becomes evident after less than a 1-h chase, a time when few labeled class II molecules in these cells have reached the cell surface (43), it is formally possible that the 120-kDa complex represents the product of surface cross-linking of mature class II dimers by unfolded HEL in the Ag preparations we use. To determine whether the 120-kDa complex forms intracellularly or at the surface, pulse-labeled B lymphoblasts were chased for 4 h in the absence of Ag. After the chase period, when most labeled class II molecules are on the plasma membrane (43), the B cells were incubated for an additional 2 h with HEL, chloroquine, or both. If the 120-kDa complex forms at the cell surface, it should be detected after immunoprecipitation of class II proteins from lysates of these cells. Chloroquine was included to interfere with any possible endosomal processing and binding to class II molecules either not yet chased out of the endosomal compartments or recycling through these organelles. No labeled 120-kDa forms can be detected under these conditions (Fig. 5), arguing strongly against surface generation of the complexes.
To identify the intracellular site at which HEL-induced association of class II Ak and Ek molecules occurs, B cells exposed to HEL were metabolically labeled, and the labeled proteins were chased for 2 h before cell disruption and fractionation on Percoll gradients (43). The organelles contained in the various gradient fractions were lysed in detergent and MHC class II molecules precipitated using anti-Ek mAb. Under these conditions, 80% of the 120-kDa complexes were found in late endosomes, and the remainder were in the dense MIIC/lysosome compartment. After 2 h of chase 120-kDa complexes were not detected in light membrane fractions containing secretory pathway components, such as endoplasmic reticulum and Golgi, or in early endosomes and plasma membrane (Fig. 6,A). Using a biotinylation approach, however, some surface expression of the 120-kDa material could be detected after an overnight chase in the presence of HEL (Fig. 6 B).
Detection of a large HEL fragment uniquely associated with 120-kDa complexes obtained from cells exposed to intact Ag
Taken together, the above results argue that the 120-kDa complexes consist of two isotypically distinct class II heterodimers simultaneously bound to a single fragment of Ag, most likely a portion of HEL itself. It is known that HEL contains several high affinity determinants for A and E molecules of various allelic origin (52, 56). In (H-2g7 × H-2d)F1 cells, the relevant determinants overlap, and binding of I-Ed to its site seems to interfere with Ag7 binding to its partially overlapping site (37). In the present case, we considered the possibility that simultaneous binding is sterically possible, and that such binding actually prevents protease attack on the intervening region, leading to a long lived complex even in an endocytic environment. Two observations argue that the ligand is a fragment of the added HEL. The first is the requisite addition of intact, nonreduced HEL to CBA lymphoblasts for the appearance of these complexes. The second is our inability to find such 120-kDa forms upon exposure of cells to short peptides that bind avidly to the same class II molecules. To look for direct biochemical evidence that exogenously supplied HEL or a fragment of this protein is involved in formation of the 120-kDa complexes, B splenocytes were exposed overnight to [125I]HEL, the cells were lysed, and class II molecules were immunoprecipitated, then analyzed for the presence of bound radioactive material by SDS-PAGE (Fig. 7, C–E). Because the appearance of the 120-kDa complex is dependent on the native conformation of HEL, the iodination was performed without intentionally denaturing the protein. In parallel, Ak class II molecules were immunoprecipitated from either B splenocytes metabolically labeled with [3H]leucine and chased overnight in the presence of cold HEL (Fig. 7, A and B) or with unlabeled splenocytes incubated overnight in the presence of [125I]HEL46–61 peptide (Fig. 7, D–F). As expected, the class II molecules isolated from metabolically labeled B cells exposed to unlabeled intact HEL migrate as 120-kDa class II complexes, compact class II dimers, and free class II chains upon SDS-PAGE without sample heating (Fig. 7,A); both the high molecular mass class II complexes and the compact dimers dissociate upon heating of the sample (Fig. 7,B). When B splenocytes are cultured overnight in the presence of [125I]HEL, only a single labeled band with an apparent molecular mass of 120 kDa is detectable (Fig. 7,C). This species comigrates, under the same electrophoretic conditions, with the high molecular mass class II complex identified by metabolically labeling the cells (Fig. 7,A). This result establishes the presence of exogenous HEL-derived material in the high molecular mass class II complex. Furthermore, if instead of intact HEL, HEL46–61 is labeled and B splenocytes are cultured overnight in its presence, only very intensely labeled compact dimers, but no labeled 120-kDa complexes, are observed after anti-Ak immunoprecipitation (Fig. 7,D). These data further support the conclusion that the 120-kDa class II complexes are not formed by the association of class II dimers containing short peptide ligands (Figs. 4 and 7 D).
To begin to characterize the HEL-derived material present in the 120-kDa complexes, B splenocytes were grown overnight in the presence of either intact [125I]HEL or [125I]HEL46–61, the cells were lysed, the class II molecules were immunoprecipitated, and half of the sample was processed without heating for standard SDS-PAGE (Fig. 7, C and D), while the other half was boiled and separated in a 10 to 20% Tris-tricine SDS gel (Fig. 7, E and F). When class II molecules were immunoprecipitated from cells exposed to intact [125I]HEL, the latter procedure revealed a 125I-labeled band with an apparent molecular mass of approximately 7 kDa (Fig. 7,E). Because the SDS-PAGE shows that upon exposure of splenocytes to [125I]HEL, only 120-kDa class II complexes, and not compact dimers, contain labeled material that would be released by sample heating (Fig. 7,C), these data indicate that this fragment of HEL with an apparent molecular mass of 7 kDa is derived from the high molecular mass class II complexes. A minor band with the mobility of intact [125I]HEL was also detectable in some experiments (data not shown). When class II molecules were immunoprecipitated from splenocytes cultured overnight in the presence of [125I]HEL46–61, boiled, and separated in a Tris-tricine SDS gel, a unique band migrating at a molecular mass of approximately 5 kDa was observed (Fig. 7 F). As previously reported (19) the HEL46–61 peptide migrated with an unexpectedly high apparent molecular mass, but its position did not correspond to that of the [125I]HEL fragment eluted from the 120-kDa complex, which migrated even more slowly. All the other HEL peptides tested migrated, as expected, with apparent molecular masses less than the smallest marker (4 kDa).
HEL contains three tyrosine residues (Y20, Y23, and Y53); Y20 and Y23 are located superficially in the intact protein, whereas Y53 is buried inside the folded molecule and therefore is difficult to access without denaturation. Y53 is contained in HEL46–61, which is the most abundant HEL determinant found associated with Ak class II molecules after endocytic processing of HEL and which is known to induce SDS-stable class II dimers (53). The absence of labeled compact dimers when B splenocytes are fed intact [125I]HEL (Fig. 7 C) therefore suggests that without unfolding the protein, the Y53 is not accessible for iodination. The absence of iodination of Y53 also argues that the labeled fragment of HEL we find bound to class II molecules in the 120-kDa complex contains Y20, Y23, or both.
Numerous large protein fragments are found associated with MHC class II molecules in B lymphoblasts
Class II proteins were also isolated from cell lysates by immunoaffinity chromatography using mAb to Ak or Ek, and the bound peptides were eluted by acid extraction. These eluted peptides were then fractionated by HPLC without subjecting them to separation from the class II proteins on low m.w. cut-off filters as performed by most investigators. The peptide content of each HPLC fraction was then analyzed by mass spectrometry. The vast majority of the peptides eluted from Ak or Ek affinity-purified class II molecules derived from cells cultured with or without HEL have a molecular mass ranging from 1400 to 2000 Da (Fig. 8). There is a smaller diversity of peptides eluted from the Ek molecules compared with the Ak molecules. Interestingly, a substantial number of peptides are seen with molecular masses >3000 Da, with most of these ranging from 3000 to 7000 Da. Although, as expected, these larger protein species represent only a small fraction of the total pool of eluted material, their presence is consistent with the identification of a fragment of approximately 7 kDa as involved in the formation of the 120-kDa complex, and this supports the idea that many class II molecules bind to material substantial larger than 12 to 15 residue peptides that extend just beyond their binding sites.
Low pH and endosomal proteases are essential for effective MHC class II molecule presentation of most antigenic determinants for two reasons. First, they contribute to the cleavage of the invariant chain that associates with class II molecules during biosynthesis (57), resulting in the production of CLIP-class II heterodimer complexes that are subject to peptide exchange via the action of DM or H-2M in late endosomal/lysosomal compartments (28, 29, 30, 31, 32, 58). Second, they participate in the unfolding and degradation of proteins entering the endocytic pathway, including foreign proteins that give rise to ligands for CD4+ T cells (59, 60). Based on the typical endocytic digestion of most incoming proteins to their constituent amino acids (61), the short length of peptides eluted from immunoaffinity-purified MHC class II molecules (3, 4, 5, 6, 7, 8), and the ability of short peptides to serve as effective ligands for CD4+ T cells (35, 36), it has become commonplace to consider that Ag presentation by class II molecules involves their scavenging of late intermediates in the degradation process. In particular, it is generally assumed that their substrates are short peptides of 15 to 20 residues derived from extensively cleaved parent proteins, such that only a few residues may protrude from the binding groove and be subject to amino- or carboxyl-terminal trimming. This model is also in accord with very recent data on the exchange function of DM and H-2M that promotes replacement of CLIP (which ranges from 13–24 residues in length) with Ag-derived ligands (28, 29, 31, 32), which are presumed to be similar in size.
This model contrasts with the hypothesis of Sercarz and colleagues (37) that argues for large protein fragments as the ligands of class II molecules during endocytic processing, based on isotype-dependent determinant competition observed in functional Ag presentation experiments. It also does not explain one of the most striking structural differences between class I and class II molecules, namely, the function of the open ends of the binding groove of the latter that permit effective interaction with very long polypeptide chains (12, 13, 14, 62). It is therefore of substantial interest whether the major substrates of class II dimers during typical endosomal processing might be unfolded proteins/large antigenic fragments trimmed to final determinant-sized peptides only after class II binding. This has been a difficult area for investigation, as such processing intermediates are hard to detect biochemically due to their short half-life, proteolytic sensitivity, heterogeneity, and asynchronous formation within the cell. In at least one experimental system, the use of labeled Ags has allowed SDS-PAGE detection of some newly formed Ag-class II complexes and the demonstration of Ag fragments greater in size than the 1 to 2 kDa expected of fully processed and trimmed determinants (45), consistent with early class II interaction with larger antigenic species.
In the present study we report the identification of stable 120-kDa complexes formed within the endocytic pathway that consist of two isotypically distinct class II molecules (Ak and Ek) bound to a single HEL polypeptide chain of approximately 7 kDa (70 amino acids). Although we have been unable to obtain sequence information on this fragment due to limitations of the PAGE methods necessary to separate the 120-kDa ligand from other class II-containing materials, the results using iodinated Ag confirm its identity as a fragment of HEL and imply that it must contain residue Y20, Y23, or both. The N-terminus of HEL in which these residues reside contains several adjacent, well-characterized Ak and Ek determinants. The peptide HEL1–18 contains a determinant binding to Ek (56), whereas HEL residues 25 to 43 have been shown to contain a nondominant core peptide that binds to both Ak and Ek molecules, but that is preferentially recognized by T cells in association with Ek (56). HEL residues 48 to 61 contain the immunodominant Ak determinant (53, 63). A fragment of HEL extending through this region would match the size of the material we see present in 120-kDa species and provide suitably spaced sites for binding the one Ak and the one Ek class II dimer that sequential precipitation experiments indicate form the bulk of the 120-kDa complex.
The longevity of the 120-kDa complexes is unexpected for an oligomer that appears to represent what might be expected to be a transient intermediate in a protein-processing pathway. We suggest that the simultaneous binding of Ak and Ek molecules to adjacent determinants within the Ag sterically excludes proteases from the intervening region, preventing further cleavage of the complex. It is likely that individual class II dimers bind to this large HEL fragment more frequently than we measure based on the amount of 120-kDa complexes formed, but that cleavage between the Ak and Ek sites often takes place before both class II molecules interact and proteolysis in the intervening region can be inhibited. The distance between two or more adjacent determinants will affect the class II binding, surface presentation, and biochemical stability of complexes containing large Ag fragments. If the sites are overlapping (37) two class II dimers are sterically inhibited from binding, and only the determinant exposed first or with the highest affinity for class II will be displayed at the cell surface. If instead the distance between the determinants is too large, more than one class II molecule may bind, but endopeptidases can access and cleave the intervening region efficiently. In this case, both determinants will be displayed for Ag presentation, but the processing intermediate will not be biochemcially detectable. These considerations together suggest that MHC class II association with long polypeptide substrates may be a dominant, rather than a minor, pathway during active processing, even though the end products are primarily the trimmed 15 to 20 residue determinants seen in elution studies. In this regard, our observation of a significant number of peptide species of 3 to 7 kDa in size after elution from immunopurified class II molecules, as also seen using a distinct class II allele by Hunt et al. (6), agrees with this general conclusion.
The detection of these oligomeric complexes also depends on their resistance to SDS denaturation. This is frequently observed with optimal short peptides bound to class II molecules (53, 64), but has also been reported for several large, unidentified proteins associated with class II dimers in cells lacking invariant chain (18). Together these findings indicate that optimal class II molecule binding can often occur with large substrates, that H-2M/DM-mediated determinant selection based on relative affinity/off-rates is likely to take place with such ligands just as with short synthetic peptides (28, 31, 32), and that this is followed by additional proteolysis of the unprotected region outside the class II molecule binding groove (26, 65, 66, 67). Both N- and C-termini of natural class II ligands are ragged and frequently extend out of the class II groove (3, 4, 5, 6, 7, 8). In about one-third of the cases, there is a proline at position 2 (27). Both these latter observations are also consistent with this proposal that the trimming of Ag to its final size most often follows binding to class II.
The model supported by the data presented here involving initial class II molecule binding to target sequences within large substrates leading to determinant protection and followed by proteolysis of the unbound segments provides a simple explanation for why the class II molecule binding groove has evolved its present structure. The open ends facilitate rapid interaction with the unfolded forms of an Ag generated after its entry into the endosomal pathway and exposure to low pH and proteases. By quickly associating with the relevant regions within a polypeptide chain, class II dimers help preserve antigenic information from the terminal degradative events that otherwise await proteins entering lysosomes. Without the ability of class II proteins to bind directly to early intermediates in the degradative process, Ag capture would probably be very inefficient in the face of late endosomal/lysosomal protease activity.
The complexes we describe here are very similar in their SDS-PAGE properties to those reported by several groups (50, 51) as the cellular equivalents of the crystallographic “dimer of dimers” or “superdimers” seen with HLA-DR (12, 13, 62). Our attempts to identify such self-associated class II superdimers rather than the ligand-bound complexes described here, however, have been unsuccessful. We only have observed 120-kDa forms in metabolic labeling experiments under the Ag-feeding conditions described here, in which case the complexes clearly are not “dimers of dimers” with short ligands in the binding groove. Similar 120-kDa complexes containing class II chains have also been detected in lysates of spleen cells from different haplotypes grown only in 10% FBS (F. Castellino and R. N. Germain, unpublished observations). In the latter cases we cannot distinguish between the ligand-bound forms described here and the superdimers associated via class II structural elements, because identical, rather than different, class II dimers give rise to the complex. Elution of the class II ligands from such 120-kDa complexes would help in resolving this issue, as the finding of a large protein fragment as observed here might argue against the crystallographic “dimer of dimers.” Thus, our results do not support the idea of a substantial cohort of stably preassociated, peptide-occupied class II molecules on the cell surface, although none of the studies we have conducted rules out the presence of such structures as either transient forms stabilized upon TCR binding (1, 12, 68) or complexes unstable in the extraction and analysis conditions we have used to search for them. The evidence here for simultaneous association of multiple class II molecules with large antigenic substrates, however, must be taken into account in evaluating any studies looking for superdimers.
Finally, a major functional implication of these results is that immunodominance of certain determinants arises as a consequence of the intrinsic affinity of an amino acid sequence for available class II molecules, the order of exposure of the various regions of the Ag during endosomal processing, and the physical influence of adjacent determinants on class II binding (37). These three parameters together will determine whether a given region or an adjacent or overlapping segment of the Ag will bind to class II molecules and therefore will be displayed at the cell surface for presentation to T cells (52).
We thank A. Rinker, Jr., for help with iodinations, and Drs. Eric Long, Paul Roche, and Caetano Reis e Sousa for careful reading of the manuscript and many helpful suggestions for its improvement.
Abbreviations used in this paper: CLIP, class II-associated Ii peptide; HEL, hen egg lysozyme; PCC, pigeon cytochrome c; TFA, trifluoroacetic acid.