Peptide presentation by MHC class II molecules plays a pivotal role in determining the peripheral T cell repertoire as a result of both positive and negative selection in the thymus. Homozygous I-Ag7 expression imparts susceptibility to autoimmune diabetes in the nonobese diabetic mouse, and recently, it has been proposed that this arises from ineffectual peptide binding. Following biosynthesis, class II molecules are complexed with class II-associated invariant chain peptides (CLIP), which remain associated until displaced by Ag-derived peptides. If I-Ag7 is a poor peptide binder, then this may result in continued occupation by CLIP to the point of translocation to the cell surface. To test this hypothesis we generated affinity-purified polyclonal antisera that recognized murine CLIP bound to class II molecules in an allele-independent fashion. We have found abnormally high natural levels of cell surface class II occupancy by CLIP on nonobese diabetic splenic B cells. Experiments using I-A-transfected M12.C3 cells showed that I-Ag7 alone was associated with elevated levels of CLIP, suggesting that this was determined solely by the amino acid sequence of the class II molecule. These results indicated that an intrinsic property of I-Ag7 would affect both the quantity and the repertoire of self-peptides presented during thymic selection.
Nonobese diabetic (NOD)2 mice develop autoimmune diabetes spontaneously and are a widely investigated model of insulin-dependent diabetes mellitus in humans (1). Insulin-dependent diabetes mellitus is an autoimmune disease caused by the T cell-mediated destruction of insulin-producing pancreatic islet β cells. While it is a polygenic disease, the most important associated genetic loci map to the MHC and the class II genes, in particular, have been implicated in disease susceptibility (2, 3, 4). Peptide-MHC complexes generated in the endosomal compartments of APC migrate to the cell surface for scrutiny by CD4+ T cells (5, 6).
The MHC class II molecules are integral membrane glycoproteins that assemble in the endoplasmic reticulum to form αβ heterodimers. The membrane-distal domains (α1, β1) pack together to form the highly polymorphic peptide-binding groove. Trimers of the invariant chain (Ii), a nonpolymorphic, integral membrane glycoprotein, interact with three pairs of αβ MHC class II subunits ensuring their correct dimerization (7). Chaperoned by the Ii, MHC class II complexes are targeted to endocytic vesicles (8, 9), and occlusion of the peptide-binding groove by residues Ii91–99 prevents premature peptide binding. Resident proteases in the acidic endosomal environment effect progressive proteolytic degradation of the Ii (10); however, the segment that occupies the peptide-binding groove remains protected from proteolytic attack. Study of peptides eluted from MHC class II molecules has revealed that fragments of varying lengths from the 81–104 segment of Ii are generated (11, 12, 13, 14, 15). Such class II-associated Ii peptides (CLIP) continue to protect the binding groove (16) until replacement by a peptide that provides more favorable binding interactions (reviewed in Refs. 17 and 18). In some, but not all, haplotypes the dissociation of CLIP requires assistance from another MHC-encoded molecule, HLA-DM in humans or H2-M in mice (15, 19, 20). As a result, most class II molecules at the surface of APC are occupied with Ag-derived peptides and not CLIP.
Most mouse APC express two types of MHC class II molecules, I-A and I-E; however, in NOD mice only molecules of the I-Ag7 isotype are expressed due to a deletion in the I-Eα promoter region (3). Compared with the I-Ad molecule, I-Ag7 has an identical α-chain, but a β-chain that differs at 17 aa residues. Of these differences, His56 and Ser57 in the class II β-chain are unique to mice of the NOD haplotype (21). Human diabetes susceptibility alleles also have a nonaspartate residue at position 57 of the β-chain.
The interaction of some mouse MHC class II molecules and CLIP has been studied in this laboratory (22, 23). We have shown previously, using transfected L929 fibroblast cells as model APC, that particular single-point mutations in Ii resulted in increased levels of CLIP expression on I-A molecules at the cell surface (24) and demonstrated that the APC with elevated CLIP expression were impaired in Ag presentation. Continuing this line of investigation, we have determined natural levels of cell surface occupancy of MHC class II dimers by CLIP fragments on APC of NOD and five other mouse haplotypes in the absence of antigenic challenge. We have raised two polyclonal anti-mouse CLIP antisera, one directed to the amino terminus and the other to the carboxyl terminus of the naturally occurring peptide CLIP81–104. Here we detail their characterization, and present data that reveal an enhanced occupancy of I-Ag7 by CLIP in both mouse splenic B cells ex vivo and B lymphoma cell lines in vitro, which were employed as model APC in this study.
Materials and Methods
BALB/c (I-Ad, I-Ed), NOD (I-Ag7, I-Eo), C3H.HeJ (I-Ak, I-Ek), PL/J (I-Au, I-Eu), C57BL/6 (I-Ab, I-Eβb), B10.A(5R)(I-Ab, I-Eb/k), and C57BL/6-derived MHC class II knockout (KO; I-Ao, I-Eo) (25) mice, bred under specific pathogen-free conditions, were obtained from the John Curtin School of Medical Research Animal Services Division (Canberra, Australia). Male mice, 6–8 wk old, were used.
Cell lines and culture conditions
All cells were cultured at 37°C in 5% CO2 and humidified air in RPMI 1640 medium supplemented with 10% FBS, 0.05 mM 2-ME, 10 mM HEPES, 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. Growth medium for transfected B cell lines was also supplemented with 200 μg/ml G418. Cells that were maintained in G418-containing medium were cultured in RPMI 1640 medium for at least 24 h before use. Mouse B cell lines, M12.C3 (26) and the transfected derivatives M12.d, M12.g7, M12.u (which were obtained by transfection of M12.C3 cells with the cDNA encoding I-Ad, I-Ag7, and I-Au, respectively), and A20 (27), were used as model APC.
Peptides were synthesized at the Australia National University Biomolecular Resource Facility (Canberra, Australia) using standard t-butoxycarbonyl chemistry and purified by reverse phase HPLC to achieve a purity of >95%. Structure and purity were confirmed by mass spectroscopy. The peptides used were CLIP 81LPKSAKPVSQMRMATPLL98, CLIP 92RMATPLLMRPMSM104, CLIP 81LPKSAKPVSQMRMATPLLMRPMSMDNMLL109, murine β1-globin 65GVKVITAFNEGLKNLLDNLKGT85, and sperm whale myoglobin 132NKALELFRKDIAAKYK147.
Generation of rabbit antisera and affinity purification of anti-CLIP Abs
Synthetic mouse CLIP, Cys81–98, and Cys92–104 coupled with keyhole limpet hemocyanin KLH (Pierce, Rockford, IL) were used to immunize rabbits. For coupling, KLH was activated with succinimidyl 4-(N-maleimido-methyl)cyclohexane-1-carboxylate (Pierce) according to the manufacturer’s instructions and coupled in a 1/1 ratio with peptide. Four milligrams of the coupled peptide-Ag emulsified with CFA was injected (s.c.) into a rabbit. Two booster immunizations consisting of 2 mg of peptide (not coupled to KLH) in emulsion with IFA were given 2–4 wk apart. Serum was harvested 15 days after the final booster dose. Resulting polyclonal Abs were sequentially purified on protein A and peptide affinity columns. Serum was diluted 10-fold in PBS, pH 8.0, and passed over a column of protein A-Sepharose (Amersham Pharmacia Biotech, Sydney, Australia). After six washes with PBS, pH 8.0, the Abs were eluted in six 1-ml fractions with 0.2 M glycine-HCl buffer, pH 3.0, and neutralized immediately following elution by addition of 50 μl/fraction of 1 M Tris, pH 9.5. Pooled protein A-purified Abs were extensively dialyzed against PBS, pH 7.4, and then concentrated to 1 ml using Centricon-30 concentrators (Amicon, Lexington, MA). Peptide affinity columns were made with each of the two CLIP peptides (CLIP81–98-Cys and CLIPCys92–104) using Sulfolink coupling gel (Pierce) according to the manufacturer’s instructions. Concentrated protein A-purified Abs were loaded onto the affinity gel and incubated at room temperature for 1 h. Columns were washed six times with PBS, pH 7.4, eluted in six 1-ml fractions with 0.1 M glycine-HCl, pH 2.8, and neutralized as before. Pooled fractions were concentrated to 1-ml volumes at ∼2 mg/ml concentrations and used directly. These Abs are referred to as affinity-purified CLIP Abs (ApCLIPAb) throughout this report.
Generation of rabbit antiserum to Murray Valley encephalitis (MVE) virus
MVE (107 PFU) emulsified with CFA were injected s.c. into a rabbit. Three booster doses consisting of 107 PFU in emulsion with IFA were delivered 2–4 wk apart, and serum was harvested after 2 wk.
Generation of rabbit antiserum to NS4B peptide
A rabbit was immunized by s.c. injection with a CFA emulsion containing 0.5 mg of a peptide (H-NEYGMLECEKPAFKR-OH) comprising the N- and C-terminal ends of the MVE NS4B protein that had been conjugated to the diphtheria toxoid through an internal cysteine. Three boosters using the same amount of peptide and IFA were given four times, 2–4 wk apart, and serum was harvested after 2 wk.
Assays were conducted on polyvinylchloride 96-well plates (Thermo Labsystems, Franklin, MA) with reaction volumes maintained at 100 μl/well throughout the assay. Peptide solutions were plated out in carbonate buffer (0.027 M Na2CO3 and 0.089 M NaHCO3, pH 9.1) at the indicated concentration and incubated overnight at 4°C. The solution was then removed, and 2% FCS in PBS was added to block nonspecific binding sites. After a 2-h incubation at room temperature the plates were washed three times with PBS-T (PBS/0.05% Tween 20). ApCLIPAb were added in PBS (1/1,000) and incubated for 2 h at room temperature followed by three washes as before. Biotinylated goat anti-rabbit Ab (Pierce; 1/25,000, v/v) and streptavidin-HRP (Pierce; 1/12,000, v/v) were added sequentially in PBS supplemented with 1% BSA and incubated for 30 min each at room temperature. Plates were washed three times, and Ab binding was measured colorimetrically by addition of the chromogenic substrate 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (Sigma, St. Louis, MO)/H2O2. OD was read at a wavelength of 405 nm against a reference wavelength of 490 nm on a Thermo-Max microplate reader (Molecular Devices, Menlo Park, CA).
Cells were stained as described previously (24), and all Abs employed were titrated and used at optimal concentrations unless otherwise stated. Briefly, 2 × 105 cells were incubated with primary Ab for 30 min in 100 μl of 1% BSA/PBS/0.01% NaN3 (FACS medium) at 4οC. Cells were washed twice between incubations in FACS medium and incubated with secondary Ab that had been conjugated to PE for 30 min. After the second incubation, cells were washed three times, and dead cells were stained by a final wash in FACS medium containing either 1 μg/ml propidium iodide (in experiments using staining with FITC-tagged Abs only) or 0.02 μg/ml 7-amino-actinomycin D (Pierce; for double staining with FITC- and PE-tagged Abs). Twenty thousand viable cells, based on their ability to exclude vital dyes, were then analyzed using a FACScan flow cytometer (Becton Dickinson, Mountain View, CA). Relative fluorescence intensity values recorded were expressed as the median fluorescence intensity, and data were displayed as frequency histograms, as percentage of positive cells (based on controls), or as bivariate plots or were plotted as graphs. Surface expression of MHC class II was determined using mAbs titrated for optimal dilutions before use: MKD.6-FITC (anti I-Ad) (27) or OX-6FITC (I-Ag7,u,k; Serotec, Oxford, U.K.). The measurement of CLIP expression on B cells stained with B220-FITC (PharMingen, San Diego, CA) employed ApCLIPAb followed by secondary Ab goat F(ab′)2 anti-rabbit IgG-RPE (Southern Biotechnology Associates, Birmingham, AL).
Before metabolic labeling, 1 × 107 log-phase cells were twice incubated for 30 min at 37°C in methionine-deficient RPMI 1640 medium (Life Technologies) containing 10% dialyzed FBS (CSL, Victoria, Australia). Cells were then incubated in 5 ml of methionine-deficient RPMI 1640 supplemented with 0.5 mCi of [35S]methionine (Amersham) for 30 min and then washed three times in 20 ml of ice-cold PBS.
Total cellular immunoprecipitation was achieved by cell lysis in 1% digitonin (v/v) in 50 mM lysis buffer (Tris-HCl (pH 7.5), 0.3 M NaCl, 5 mM EDTA, pepstatin, aprotinin, leupeptin, and 10 μg/ml of each of the three protease inhibitors). After gentle inversion for 15 min at 4°C, cell lysates were cleared of cellular debris by centrifugation (14,000 × g) for 15 min and precleared with 50 μl of protein A beads. Appropriate Abs (5 μg of MKD.6 (anti-I-Ad) or 5 μg of OX6 (anti-I-Ag7) were used to precipitate immune complexes, which were recovered by incubation for 2 h at 4°C with 50 μl of either protein A (MKD.6 Ab) or protein G beads (OX-6 Ab; Sigma). After several 1-ml washes with protein lysis buffer, proteins were eluted from the beads at room temperature in 10 μl of nonreducing 2% SDS-PAGE loading buffer, and immunoblot analysis was performed.
Selective cell surface immunoprecipitation was performed by radiolabeling cells as described above followed by a 6-h chase with 10× methionine-containing 1640 RPMI. Cells were washed twice with PBS/2% FBS and then incubated with primary Ab (20 μl of ApCLIPAb or 5 μg of MKD.6 mAb) for 30 min at 4°C, following which CLIP81–109 peptide was added to the reaction (final concentration, 0.5 mM) to remove residual free CLIP Ab. Several washes were given in PBS at 4°C before the cells were lysed, the Ag-Ab complexes were recovered, and analysis was performed by immunoblotting, as before.
Immunoprecipitated material was separated by SDS-PAGE on 10–20% gradient Tricine gels (NOVEX, San Diego, CA). Proteins were electrotransferred to nitrocellulose membranes at 40 V overnight at 4°C (NOVEX) in 39 mM glycine, 48 mM Tris, and 20% methanol (v/v). Membranes were dried before autoradiography by exposure to Kodak BioMAX MR film overnight (Eastman Kodak, Rochester, NY). Dried membranes were blocked for 2 h at 25°C in milk buffer (5% nonfat milk powder, PBS, and 0.05% Tween-20) and probed with the specific Ab (1/500 dilution in milk buffer of 5 mg/ml MKD.6, 1 mg/ml OX6, or 1B9A (anti-H2-Mb) and 1/1000 of no. 104 (anti-H2-Ma; anti-H2-M Abs were provided by Dr. L. Karlsson (San Diego, CA) and Dr. J. Trowsdale (Cambridge, U.K.)) and incubated for 1 h at room temperature. After three successive washes in milk buffer and one wash in PBS-T, HRP-labeled secondary Ab was added (0.01 mg/ml) in PBS-T followed again by four washes as described above, and binding was detected using an enhanced chemiluminescence kit (Amersham).
Enrichment of splenic B cells
Spleens were taken from each mouse and cut into fine pieces before being passed through steel mesh to prepare single-cell suspensions. Cells were pelleted by centrifugation at (600 × g) and water (4.5 ml) was added to the cell pellet (2 × 108 cells) with gentle vortexing, thereby lysing erythrocytes. 0.5 ml of 10× Hank’s medium and 5 ml of RPMI 1640 medium were added successively, and the cells were pelleted as before. B cells were purified on nylon wool according to the manufacturer’s instructions (Robbins Scientific, Mountain View, CA). Briefly, 0.5 g of nylon wool presoaked with 10 ml of RPMI 1640 and equilibrated to 37°C was placed in 10-ml syringes. Cells were resuspended in 2 ml of RPMI 1640 and passed over the nylon wool columns, which were placed in a 37°C incubator for 45 min. The fibers were extensively flushed with warm 10% FBS/RPMI 1640 medium to deplete nonadherent cells, after which B cells were eluted with 10 ml of cool (∼10°C) RPMI 1640 medium free of FBS. Cells were washed once in RPMI 1640 and used immediately.
Acid elution of peptides from intact cells
Acid elution of peptides was conducted by an adaptation of methods described previously (28, 29). Enriched B cells (2 × 108) were washed twice in PBS and suspended in 3 ml of acid elution buffer (500 mM NaCl and 200 mM acetic acid, pH 2.4) for 2.5 min at 20°C to extract MHC-associated peptides from the cell surface. Acid-treated cells were pelleted by centrifugation (600 × g), and the low m.w. fraction of the acid extract was recovered from this supernatant by centrifugal ultrafiltration using Centricon-10 extraction (Amicon). The eluate was transferred directly to an ELISA plate for detection.
Characterization of ApCLIPAb
Interaction of the immunizing peptides, either CLIP81–98 or CLIP92–104, with the ApCLIPAb in an ELISA (Fig. 1) demonstrated peptide specificity. The Abs did not react with two control peptides that lack linear sequence homology to either CLIP peptide, murine β1-globin65–85 and sperm whale myoglobin132–147. The ApCLIPAb also gave positive results against a panel of CLIP variant peptides, ranging in length from 9 to 29 residues (22), in an ELISA (data not shown).
When presented in the peptide-binding groove of the class II MHC molecule CLIP may not be accessible for binding by the ApCLIPAb. Cell surface immunoprecipitation was used to test the ability of these Abs to bind MHC-bound CLIP. I-Ad expressing A20 cells were metabolically labeled with [35S]methionine for 30 min and chased in 10× methionine-containing medium for 6 h at 37°C to allow radiolabeled I-Ad molecules to undergo maturation and reach the cell surface. MKD.6 mAb was used to immunoprecipitate total cell surface I-Ad, and ApCLIPAb were used in a cell surface-selective immunoprecipitation of CLIP-I-Ad complexes. Fig. 2 A reveals radiolabeled proteins of between 65 and 69 kDa. To confirm that these bands represented MHC class II dimers, immunoblot analysis of the same membrane using the MKD.6 Ab was performed, which showed that the bands immunoprecipitated by the anti-CLIP Abs were class II MHC. Each of the ApCLIPAb precipitated 69-kDa proteins only, while the MKD.6 Ab precipitated 69- and 65-kDa proteins. This demonstrates the ability of the ApCLIPAb to bind to CLIP associated with class II on the cell surface.
If the Abs recognized cell surface CLIP, then the addition of free ligand (synthetic CLIP) would compete with cell surface CLIP-class II complexes for Ab binding. Surface-selective immunoprecipitation by ApCLIPAb in both the presence and the absence of free CLIP ligand and a subsequent immunoblot of the membrane with MKD.6 showed a decrease in density of the 69-kDa protein band in samples to which CLIP ligand had been added (Fig. 2 B). Together these results clearly demonstrate that the Abs recognized CLIP-MHC class II complexes at the cell surface.
CLIP expression on splenic B cells
Expression of CLIP in NOD mice was compared with that in five other mouse strains (BALB/c, C3H.HeJ, PL/J, C57BL/6, B10.A(5R), and an MHC class II KO). Double staining of B cells for CLIP ex vivo was followed by flow cytometric analysis (Fig. 3,A). Expression of CLIP on I-Ag7 was strikingly higher than that on all other haplotypes studied (Fig. 3 B). Only NOD, PL/J, and BALB/c B cells gave values above the background established by the MHC class II KO mice.
Addition of free peptide ligand near totally abrogated ApCLIPAb binding, which demonstrated that both ApCLIPAb were binding predominantly through their Ag binding site to the surface of NOD B cells (Fig. 3 C).
The most commonly distributed Fc receptor in mice, FcγRI (CD64), generally binds with high affinity to a single IgG subclass, IgG2a. However, NOD mice express a unique Fc receptor that also exhibits a high affinity for IgG3 and IgG2b (30, 31) that potentially may have contributed a false positive signal in this haplotype. We tested for binding of two other rabbit antisera, generated against MVE virus and the MVE NS4B protein, to splenic B cells isolated from NOD mice. Flow cytometric analysis staining with these sera, which would contain a range of IgG isotypes, revealed negligible binding (Fig. 4,A). Consequently, it is unlikely that the elevated immunostaining of NOD B cells by ApCLIPAb arose from binding Fc receptors. Furthermore, the binding of both ApCLIPAb to NOD B cells was diminished by >50% after incubation in a low pH buffer designed to elute peptides from MHC class II molecules (Fig. 4,A). Significant titers of CLIP in an ELISA were observed only in eluates from the NOD haplotype (Fig. 4,B), which correlated with the results presented in Fig. 3 B.
Expression of CLIP on M12.C3 cells expressing different class II MHC
Cultured M12.g7, M12.d, M12.u, and M12.C3 cells, which express I-Ag7, I-Ad, I-Au, and no I-A, respectively, were stained with ApCLIPAb. Flow cytometric analysis using Abs in saturating concentrations revealed that there was significantly higher anti-CLIP immunoreactivity on M12.g7 cells than on M12.d, M12.u, and M12.C3 cells (Fig. 5).
Interactions with H2-M
The physical association of I-Ag7 and I-Ad with H2-M molecules in the M12.d and M12.g7 cell lines was demonstrated by a coimmunoprecipitation experiment (Fig. 6). Mild detergent was used to disrupt cell membranes, allowing weak associations between protein molecules to be maintained. Both MHC class II molecules were found to interact similarly with H2-M.
Class II MHC molecules have been implicated in susceptibility to type 1 diabetes by genetic association (1, 32). In particular, the homozygous expression of non-Asp residues at position β57 has been implicated strongly in susceptibility to diabetes in both mice and humans (2, 33). The mechanism by which the structure of MHC class II protein might affect the peripheral T cell repertoire remains to be determined (32, 34), and currently there are differing views (35). Nepom (36) attributed this to high affinity binding of diabetogenic peptides by I-Ag7, but more recently it has been reported that this class II molecule may bind peptides poorly, suggesting that potentially autoreactive T cells might escape thymic selection (37). Efficient peptide loading in the endosomal compartments leads to displacement of CLIP from the binding groove of MHC class II molecules. Accordingly, we propose that if I-Ag7 is indeed ineffectual in sequestering peptides, this would result in the retention of CLIP and the expression of CLIP-class II complexes at the cell surface. It has been reported that CLIP has a rapid rate of dissociation from such complexes at the cell surface (15), and consequently, we required a method enabling their detection directly on cells. To determine whether CLIP occupancy of class II MHC proteins at the cell surface differs in a haplotype-dependent manner, we required a reagent that was able to recognize mouse CLIP independently of the complexing partner. Given the length and sequence differences observed between typical CLIP eluted from mouse as opposed to human class II MHC molecules (13, 14), we chose not to rely on the cross-reactivity of mAbs raised against human CLIP for this present purpose. Rather, we elected to generate polyclonal Abs against peptides spanning the mouse CLIP81–104 sequence in an attempt to maximize avidity and minimize discrimination between different complexing mouse class II MHC molecules concomitantly. This strategy was successful and permitted the use of viable cells, providing a “snap-shot” view of the cell surface.
Fortuitously, A20 cells provided a special opportunity to test the avidity of ApCLIPAb for class II-associated fragments of Ii by virtue of the high expression levels of I-Ad and the discrete bands obtained from different conformations of this class II molecule in SDS-PAGE. In SDS-PAGE analysis, class II heterodimers may be observed in both floppy and compact forms that differ in their electrophoretic mobility, provided that the sample has not been heat-denatured (38). Dornmair et al. showed that mild denaturation may lead to loss of bound peptide, resulting in the more slowly migrating floppy dimer, while more extreme denaturation leads to dissociation into free α- and β-chains (38). Both in vitro and in vivo work has demonstrated the critical role played by the peptide in generating SDS-stable, compact MHC class II dimers (39, 40, 41). The sample from the MKD.6 immunoprecipitate that had not been heat treated included both the floppy (69-kDa) and compact (65-kDa) forms of the I-Ad class II MHC molecule (Fig. 2,A). The presence of a minor proportion of the I-Ad molecules in the floppy form revealed that not all were complexed with cognate peptides at the cell surface (17, 42). By contrast, the corresponding heat-treated sample yielded none of the floppy forms but, rather, a complex mixture of compact forms that were sufficiently stable so as to survive and a general shift of material to higher relative molecular masses, indicative of aggregation (Fig. 2,B). Significantly, comparison of the MKD.6-immunoreactive band patterns obtained with and without heat treatment of the ApCLIPAb immunoprecipitates (Fig. 2) revealed that only floppy forms of the I-Ad dimer were precipitated. These floppy CLIP-I-Ad complexes formed aggregates upon heat treatment, in contrast to the heat-stable compact forms precipitated by MKD.6. This demonstrated not only the ability of the ApCLIPAb to bind to CLIP-associated I-Ad dimers, but also that such complexes are SDS-stable, albeit not able to assume the compact conformation. This is consistent with the view proposed by Cresswell and coworkers (43), who have previously suggested that CLIP may be present in the pool of unstable heterodimers.
Free ligand (synthetic CLIP peptide) competed with MHC-bound CLIP for Ab recognition at the cell surface (Fig. 2 C) and decreased the intensity of the protein band obtained by immunoprecipitation, confirming the specificity of both Abs for CLIP-class II complexes.
Unfortunately, this immunoprecipitation approach does not provide the required general means to estimate the relative levels of expression of CLIP-class II MHC molecules at the surface of undisrupted APC. In particular, it is expected that the method would underestimate less stable CLIP complexes that might not survive the conditions of cell lysis and SDS-PAGE. Instead, for this purpose we have employed the method of flow cytometry and the ApCLIPAb. Splenic B cells from the class II MHC KO and six mouse haplotypes (g7, d, u, k, b, b/k) were stained with both ApCLIPAb. Flow cytometric studies (Fig. 3) showed that staining intensities were not significantly above the class II KO background for splenocytes from k, b, and b/k haplotypes. By contrast, B cells from the g7 haplotype revealed CLIP expression levels that were remarkably high. BALB/c and PL/J mice also showed higher reactivity than KO background. In the case of I-Ad, this might be expected given the high affinity of I-Ad for CLIP (44) and our cell surface immunoprecipitation results. The affinity of I-Au for CLIP is currently unknown, but is predicted to be low (45).
It should be borne in mind that the level of expression of CLIP-class II complexes at the surface of APC has an indirect relationship with actual binding affinity for particular Ii fragments. The data from the b, b/k, d, k, and u mouse haplotypes may be thought of as establishing the normal range of surface CLIP immunoreactivity for splenocytes. This narrow range indicates that the binding affinity for CLIP per se has only a weak correlation with its presentation, as evidenced by our results. Indeed, this has been demonstrated previously by the differential expression of CLIP-class II complexes by thymic epithelial cells and peripheral APC (46). Thus, the data presented in Fig. 3 estimate the amount of CLIP at the cell surface that is the output of the total cellular process of class-II-mediated Ag presentation. One might expect low levels of CLIP presentation in an efficiently operating Ag presentation system. Indeed, where Ag presentation has been disrupted in certain haplotypes of H2-M−/− mice, high levels of CLIP expression by APC have been induced (47, 48, 49). It is thus not surprising that the expression of CLIP-class II complexes on splenocytes was low in haplotypes resistant to autoimmune diabetes, but it is surprising that the signal obtained from splenocytes derived from the g7 haplotype was so high.
Recently, it has been reported that a synthetic peptide, CLIP86–100, dissociated rapidly in vitro at endosomal pH from a soluble analogue of I-Ag7 that had been expressed in Drosophila cells (50). At first glance, these results may appear to contradict our finding that APC from NOD mice have abnormally high levels of CLIP immunoreactivity at the cell surface. It is not clear exactly how these data pertain to the presentation of CLIP immunoreactivity at the surface of unchallenged APC because the results were not obtained in situ from native endosomal I-Ag7. Hausmann et al. (50) reported that in the absence of detergent, bound CLIP86–104 remains detectable after 2 h of incubation at pH 7.4, even in the presence of a high affinity competitor peptide, but, by contrast, displacement with a high affinity peptide occurs rapidly at pH 5.0.
These observations in combination with the results presented here suggest strongly that I-Ag7, although it has the potential to bind high affinity peptides, sequesters peptides less efficiently in the endosomes of unchallenged APC than class II molecules expressed by autoimmune diabetes-resistant haplotypes. Additionally, provided that CLIP-I-Ag7 class II MHC complexes reach the cell surface, they should remain detectable for a protracted period.
The elevated presentation of CLIP by NOD B cells could be due to the intrinsic properties of the I-Ag7 MHC class II molecule itself, the influence of a variety of NOD background genes, or a combination of both. M12.C3 is a B lymphoma cell line that expresses neither the I-A nor the I-E class II MHC molecules at the cell surface (51). Transfection of these cells with cDNA that encodes both α- and β-chains of I-Ad, I-Ag7, or I-Au generated three cell lines differing only in their class II haplotype. This enabled a comparative study of levels of CLIP complexed with these different class II molecules at the surface of cells with an otherwise identical genetic background. The cells that expressed I-Ag7 (Fig. 5) displayed significantly more CLIP at the surface than those that expressed other class II molecules. We are currently investigating whether in addition to I- Ag7 other genes in the NOD background influence cell surface CLIP expression.
In mice, the nonclassical MHC class II molecule, H2-M, is important in CLIP displacement and peptide loading of MHC class II molecules. This molecule and its human homologue, HLA-DM, act much in the manner of a catalyst, assisting the competitive exchange of peptides (52). It is therefore possible that elevated cell surface CLIP levels in NOD mice could arise from a dysfunctional interaction of H2-M with I-Ag7. However, we found no difference in the extent of physical interaction between I-Ag7 and H2-M or between I-Ad and H2-M in M12 cells (Fig. 6). Peterson and Sant (53) have also shown a normal functional interaction between H2-M and I-Ag7, which enables peptide exchange. It seems most likely, therefore, that elevated CLIP presentation arises from ineffectual sequestration of peptides in the endosomes. The molecular basis for this impaired presentation of peptides by I-Ag7 is currently under investigation.
Elevated presentation of CLIP by APC has not been reported previously either in I-Ag7-expressing cell lines in which the class II-mediated pathway remains undisrupted or in NOD mice. Autoradiography after SDS-PAGE resolution of anti- I-Ag7 immunoprecipitates from [35S]methionine-labeled cells that express in I-Ag7, but not H2-M, has revealed the presence of significant amounts of bound CLIP (53). In the same experiment these authors demonstrated that coexpression of H2-M abolished the low m.w. band from the autoradiograph. These observations demonstrated that exchange or loss of CLIP from I-Ag7 is catalyzed by H2-M to the point where the abundance of these peptides falls below the threshold of detection for the method.
Acid elution of peptides from I-Ag7, which had been isolated from NOD mice by immunopurification, did not reveal CLIP in subsequent analysis by mass spectroscopy (54). Because mass spectrometry provides a highly sensitive method for the detection of peptides, we reconcile the difference between that report and our data by considering the differences between the methods employed for CLIP detection. The indirect detection of CLIP-I-Ag7 complexes by immunopurification requires that they be stable to the conditions employed over a significant period of time or signal loss will result. Direct detection of these complexes on intact cells using ApCLIPAb does not require their survival through prolonged biochemical procedures. Mass spectrometry is a high resolution method that resolves CLIP into a nested set of molecular species, resulting in a set of signals proportional to their abundance. By contrast, the ApCLIPAb hardly discriminate between molecules of different molecular mass (data not shown), and this leads to signal summation over all immunoreactive species. The flow cytometric method detects cells with surface anti-CLIP immunoreactivity that is significantly greater than an empirical threshold, and thus while the percentage of positive cells may be relatively high, this may only derive from relatively low molar amounts of peptide. Together these factors may account for the sensitivity of CLIP detection using the method reported herein.
Abnormally elevated CLIP presentation by I-Ag7 at the cell surface, as demonstrated here for B cells, may also be a feature of thymic epithelial cells and the other APC that are involved in both positive and negative selection of developing thymic T cells (currently under investigation). In an avidity model, thymocytes are selected for maturation on the basis of the interactions between TCRs and combinations of MHC molecules and peptide complexes (55, 56). The presentation of I-Ag7-CLIP complexes on thymic APC could interfere with both positive and negative selection processes, thereby playing an important role in shaping the T cell repertoire. Indeed, a number of studies support this idea. Characteristically, in H2-M-dependent mouse strains that lack functional H2-M, APC display MHC class II predominantly occupied with CLIP at the cell surface and have been reported to have a less diverse CD4+ve peripheral T cell repertoire than wild-type mice (47, 48, 49). Mice that express the I-Ag7 molecule have a higher proportion of autoreactive T cells in the periphery (57), which is consistent with the view of altered thymic selection in these haplotypes.
The data presented here demonstrate that CLIP presentation is abnormally high in I-Ag7 molecules on the surface of unchallenged APC. This appears to be an intrinsic property of the structure of I-Ag7 itself. If the high level of CLIP presentation is a feature of other non-Asp β57 class II molecules (which are genetically linked with autoimmunity), then there could be a role for CLIP in the mechanism that links MHC structure to autoimmunity. The experimental approach described here may be a valuable tool in exploring this issue.
We thank Dr. Sarah M. Weenink for useful discussions during preparation of the manuscript, Sabine Grüninger and Geoff Osborne of the John Curtin School of Medical Research FACS facility for their excellent technical assistance and support, and Drs. Arno Müllbacher and Andrew Hapel for suggestions and experimental advice given during this research.
Abbreviations used in this paper: NOD, nonobese diabetic; Ii, invariant chain; CLIP, class II-associated Ii peptide; KO, knockout; KLH, keyhole limpet hemocyanin; ApCLIPAb, affinity-purified CLIP Abs; MVE, Murray Valley encephalitis; PBS-T, PBS/0.05% Tween 20.