Distinct Peptide Loading Pathways for MHC Class II Molecules Associated with Alternative Ii Chain Isoforms¹ Free

The highly polymorphic class II products of the MHC expressed by B cells, macrophages, dendritic cells, and thymic stromal elements guide CD4⁺ T cell responses toward selected peptide ligands. A complex biosynthetic pathway appears necessary to facilitate the formation and delivery of mature class II αβ/peptide complexes to the cell surface (1). In addition to class II α and β-chains, essential activities are contributed by the conserved Ii chain and DM, a nonconventional class II molecule, originally described as a facilitator of class II Ag presentation in mutant human cell lines (2, 3).

Since its discovery as the third polypeptide chain coassembled with class II α and β subunits, much has been learned about Ii chain actions as a class II chaperone (3). As a general rule, assembly of allelically matched α and β pairs does not require Ii chain coexpression, but in the exceptional case of Aα^bAβ^b molecules, Ii chain appears necessary for production or maintenance of αβ dimers (4). In the absence of Ii chain, empty class II molecules tend to aggregate, a property reversed upon occupancy of the peptide binding site (5, 6). The Ii chain protects the empty groove from association with molecular chaperones, such as BiP and calnexin, responsible for ER quality control (7, 8, 9, 10, 11). The Ii chain also promotes export of correctly folded αβ dimers past the cis-Golgi (12, 13) and directs their delivery to endocytic vesicles (14, 15). Selective Ii chain degradation then permits acquisition of peptide ligand (16).

DM function is also necessary for surface display of diverse peptide/class II complexes (2). Considerable data suggest that DM acts inside endocytic vesicles to cause dissociation of a relatively short proteolytic product of Ii chain corresponding to the so-called class II-associated Ii chain-derived peptide (CLIP)³ region, in exchange for tightly bound peptide ligand(s) (17, 18, 19, 20, 21, 22, 23). Mutant mice lacking DM function express class II molecules predominantly bound by CLIP peptide, and as a consequence they exhibit defective peptide loading, Ag presentation, and CD4⁺ T cell maturation (24, 25, 26). The inability of these class II/CLIP complexes to elicit CD4⁺ T cell responses strongly argues that DM-mediated CLIP release must be quite efficient under physiologic conditions in the intact animal. Recent studies demonstrate DM also interacts with empty class II molecules (27, 28, 29) and acting in this manner may function as a peptide editor serving to increase overall affinities of peptide/class II complexes (30, 31).

We have as yet only a partial view of the complex associations among MHC class II, Ii chain, DM, and peptide ligand(s). The formation of αβIi complexes is especially complicated considering that the Ii chain is oppositely oriented in comparison with class II α and β subunits, having its C terminus inside the endoplasmic reticulum lumen (32). The cytoplasmic N-terminal portion of the Ii chain contains its endosomal targeting signal(s) (14, 15). Similar targeting motif(s) have also been mapped to the cytoplasmic tails of DM (33, 34) and class II (35) β-chains. Peptide elution experiments (36, 37, 38, 39), transfection assays (40, 41, 42), and x-ray crystal studies (43) all consistently demonstrate that the Ii chain CLIP region is responsible for occupancy of the class II groove. As for conventional peptide, this sequence bound to class II has an extended rigid conformation due to its extensive contacts with both α- and β-chains (43). In contrast, on the intact Ii chain, the CLIP region comprises a highly disordered flexible domain accessible to proteases (44).

In comparison with this detailed picture of its membrane proximal regions, any physiologic contributions made by the other domains comprising most of Ii chain remain ill defined. The compact α-helical segment encoded by exons 5 and 6 appears necessary for assembly of Ii chain trimers (41, 44, 45), and may act to promote class II Ag presentation (45). Interestingly, the distal portion of Ii chain encoded by exons 7 and 8 probably also contacts the class II peptide groove (46). Thus, recombinant Ii chain spanning residues 118–208 has the ability to enhance peptide binding by empty class II αβ dimers (44). Similarly, this portion of the Ii chain blocks TSST binding to the class II α1 domain (42).

It has been particularly challenging to dissect the possibly divergent functional roles provided by individual p41 and p31 Ii chain isoforms arising due to alternative mRNA splicing (47, 48, 49). The p41-specific segment encoded by exon 6b represents the most highly conserved portion of Ii chain, suggestive of an important function for this region (49). Moreover, these extra 64 amino acids contribute 2 additional conserved sites for attachment of N-linked glycans, and all 6 lumenal cysteines, probably joined via disulfide bonds (49). Recent studies demonstrate that this domain functions as a specific inhibitor of the lysosomal cysteine protease, cathepsin L (50, 51). Selective enhancement by p41 of class II Ag presentation has been documented in restricted circumstances (52). On balance, however, previous data strongly argue that alternative p31 and p41 isoforms provide equivalent chaperone activities (53, 54, 55, 56, 57, 58, 59, 60, 61). For example, Stockinger et al. (53) found that either p31 or p41 alone was sufficient to render class II-expressing fibroblasts capable of Ag presentation. Similarly, p31 and p41 both interact with calnexin (57), form trimeric complexes (55), and facilitate transport to endocytic vesicles (55, 56). Selective expression of individual Ii chain isoforms under physiologic conditions in vivo is sufficient to promote class II Ag presentation and CD4⁺ T cell development (58, 59, 60, 61). In contrast to conventional transgenic strains created by two other laboratories using cDNA constructs (58, 59), we previously used ES cell technology to generate mice expressing alternative p31 or p41 Ii chain isoforms under control of endogenous regulatory elements (60, 61). Our mutants strongly express either p41 or p31 Ii chain at levels equivalent to wild type, and in all functional experiments undertaken to date gave virtually identical results. Moreover, similar conclusions were reached when analyzing mutants expressing two independent MHC haplotypes (61).

To further explore possibly divergent roles contributed by alternative Ii chain isoforms, here we generated double mutant strains selectively expressing p31 or p41 Ii chain and also carrying a null mutation at the DM locus, and compared their class II functional capabilities. Both strains appear equally competent with respect to surface expression of Aα^bAβ^b/CLIP complexes and CLIP occupancy as judged by SDS-PAGE. Surprisingly, in functional assays, we observe strikingly different phenotypes. Thus, DM-deficient mice expressing wild-type Ii chain or p31 alone are both severely compromised in their abilities to present exogenous peptides. In contrast, double mutants expressing the p41 isoform display markedly enhanced peptide-loading activities, approaching those observed for wild-type mice. As for DM mutants expressing wild-type Ii chain, double mutants expressing individual isoforms fail to present native protein Ags, and exhibit partially defective CD4⁺ T cell maturation. These results extend evidence for divergent class II Ag presentation pathways and demonstrate for the first time that functionally distinct roles are mediated by alternatively spliced forms of the MHC class II-associated Ii chain in a physiologic setting.

DMα-deficient mice (24) and mutants expressing only p31 (60) or p41 (61) Ii chain were previously described. These mutations were originally established on a (129 × C57BL/6)F₂ genetic background, and both of these strains express the H-2^b haplotype. H-2^k/k 31/31 and 41/41 sublines were produced as described (61) and have been maintained by brother-sister matings. In all experiments, comparisons were made between age- and whenever possible sex-matched animals.

The PCR genotyping assay used to distinguish wild type and mutants exclusively expressing individual Ii chain isoforms has been described (61). Additionally, to screen for the DMα mutant allele, we used a three-primer system. The common primer (5′-TCTGGACACTGGGATTTGACCTTC-3′) lying at the 3′ end of exon 2 used in conjunction with a second primer upstream (5′-CACATTCCGGCACACTCTATTCTG-3′) in a portion of the gene deleted by the targeting event yields a 270-bp wild-type band. Additionally, a third primer (5′-ATCGCCTCCTATCGCCTTCTTCAC-3′) specific for the neo cassette in the targeting vector gives rise to the 362-bp mutant product. Reactions were conducted for 30 s at 94°C, 30 s at 59°C, and 45 s at 72°C for 30 cycles, with a final extension for 10 min at 72°C. The amplification products were resolved on a 2% agarose gel and visualized by ethidium bromide staining.

Y3P (62) and Y-Ae (63) hybridomas were provided by Charlie Janeway, Jr. (Yale University School of Medicine, New Haven, CT.), 30-2 (64) was the kind gift of Sasha Rudensky (University of Washington School of Medicine, Seattle, WA), BP107 (65) and 10-2-16 (66) were from the American Type Culture Collection (Rockville, MD), and H116-32 (67) was provided by G. Hammerling (German Cancer Research Center, Heidelberg, Germany). The Eα 56-73 (ASFEAQGALANIVDKA), OVA 323–339 (ISQAVHAAHAEINEAGR), and HEL 74–88 (NLANIPASALLSSDI) peptides were purchased from Quality Controlled Biochemicals, Inc. (Hopkinton, MA).

Biosynthetic labeling, immunoprecipitations, and SDS-PAGE were conducted as described (68). Briefly, spleen cells were washed with warm HBSS containing 2% FCS and antibiotics and resuspended (2 × 10⁷/ml) in warm methionine-free DMEM supplemented with 4 mM glutamine and 5% dialyzed FCS. After 1 h at 37°C, [³⁵S]methionine was added (250 μCi/ml) for 40 min. The cells were subsequently resuspended in a fivefold excess volume of warm DMEM containing 15% FCS and 10 × excess cold methionine, incubated at 37°C for 4 h, harvested, and then washed twice with ice-cold PBS. The cell pellet was lysed in buffer containing 1% Nonidet P-40, 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM EDTA, 1 mM PMSF, and 10 μg/ml aprotinin. After incubation on ice for 15 min, extracts were cleared of nuclei and debris by centrifugation for 30 min at 15,000 rpm. Lysates were precleared once with rabbit anti-mouse IgG (heavy chain plus light chain (H + L)) Abs (Zymed, South San Francisco, CA), twice with rabbit anti-rat IgG (H + L) Abs (Zymed), and twice with protein A-agarose (Life Technologies, Gaithersburg, MD) before the addition of specific Abs. Immunoprecipitates were washed three times with buffer containing 0.05 M Tris-HCl (pH 8), 0.45 M. NaCl 0.5% Nonidet P-40, 0.05% sodium azide, and 1 μg/ml aprotinin and then solubilized in Laemmli buffer containing 2% SDS and 2-ME by treatment either for 60 min at room temperature or by heating at 100°C for 10 min as indicated in the figure legends. Samples were analyzed by SDS-PAGE, subsequently treated with EnHance (DuPont-NEN, Wilmington, DE), dried, and exposed to x-ray film.

For single-color analysis, spleen cell suspensions depleted of erythrocytes by ammonium chloride-Tris treatment were incubated on ice with saturating amounts of biotin-conjugated Abs followed by FITC-labeled avidin D. Fluorescence was analyzed using a FACScan flow cytometer (Becton Dickinson, Mountain View, CA), and the data were displayed as cell number vs log fluorescence. Dead cells and erythrocytes were eliminated from the analysis by appropriate gating. For three color analysis, suspensions of thymocytes or spleen cells were incubated on ice with anti-CD8 FITC, anti-CD4 PE, biotinylated anti-TCR (cat #01044D, 01065B, and 01302D, respectively, PharMingen, San Diego, CA) followed by Streptavidin-Red 670 (Life Technologies). CD4 vs CD8 dot plots are shown.

T hybridomas used in this study include BDK11.1 specific for I-A^b/KLH (69) and BO97.1 specific for I-A^b/OVA (70), provided by Philippa Marrack (Howard Hughes Medical Institute, National Jewish Center, Denver, CO), 1H3.1 specific for I-A^b/Eα 52-68 (71), the kind gift of Sasha Rudensky (University of Washington), and BO4H.9 specific for I-A^b/HEL 74–88 (72), given to us by Nilabh Shastri (University of California at Berkeley, Berkeley, CA). IL-2 production was assessed by incubating T cells (5 × 10⁴/well) with irradiated (3300 R) spleen cells (2 × 10⁵/well) in 200 μl of complete RPMI 1640 supplemented with 15% FCS, 10% NCTC 109, 100 U/ml penicillin, 100 μg/ml streptomycin, 1 mM sodium pyruvate, 15 mM HEPES (pH 7.2), 0.1 mM nonessential amino acids, 5 × 10⁻⁵ M 2-ME, 2 mM glutamine, and increasing concentrations of Ag as indicated in Figure 5. Supernatants were collected after 20 h and assayed for IL-2 content in a secondary culture using CTLL indicator cells. [³H]TdR incorporation was measured in the presence of 50% primary supernatant. Responses were measured after a 48 h culture period by a 16–18 h exposure to 1 μCi of [³H]TdR. All results are expressed as mean counts per minute of triplicate cultures.

The CLIP residues responsible for occupancy of the class II groove potentially represent a structurally independent region on intact Ii chain. Alternatively, interactions with other domains may also affect CLIP conformation. To test for possible constraints imposed by exon 6b sequences, we decided to compare individual Ii chain isoforms for their functional abilities in a DM-deficient background. To this end, we set up matings between DM-deficient animals (24) and Ii chain mutants selectively expressing either p31 or p41 Ii chain (60, 61). Their double heterozygous offspring were subsequently intercrossed to produce (eventually) DM mutants expressing only p31 or p41 Ii chain. Consistent with earlier results, we also observed for DM double mutants expressing p31 or p41 alone, virtually identical cytoplasmic staining profiles using In-1 mAb specific for an epitope located near the Ii chain N terminus (data not shown). Thus, these mutant strains produce equal total amounts of Ii chain protein, and differ only by their expression of exon 6b sequences.

As shown in Figure 1, DM mutants expressing wild-type Ii chain, or double mutants expressing either the p31 or p41 isoform, gave similar patterns in surface-staining experiments. As judged by their reactivity with Y3P (α + β) mAb, DM mutants expressing wild-type Ii chain, or double mutants expressing p31 or p41 alone, and wild-type control strains, all express the same total amount of mature Aα^bAβ^b at the cell surface (Fig. 1 A). In comparison with their wild-type counterparts, DM mutants expressing wild-type Ii chain and double mutants selectively expressing the p31 or p41 isoform gained reactivity with the 30-2 mAb specific for Aα^bAβ^b/CLIP (64). Consistent with earlier results, we also observe here that Aα^bAβ^b surface molecules expressed by DM mutant spleen cells show no reactivity toward BP107 (β-specific) mAb (25, 26). The double mutants similarly lack expression of BP107 epitopes. Thus, DM mutants expressing either p31 or p41 Ii chain display remarkably similar surface staining profiles. We conclude that CLIP regions present on individual isoforms are equally competent to promote formation of Aα^bAβ^b/CLIP complexes and direct their transport to the cell surface.

Next we compared these strains for expression of Aα^bAβ^b/CLIP complexes in immunoprecipitation experiments. DM double mutants were analyzed alongside their wild-type counterparts. As before (61), we also observe here equal amounts of mature compact Aα^bAβ^b dimers in the presence of either p31 or p41 Ii chain produced by control strains carrying the wild-type DM locus (Fig. 2). A significant fraction of mature Aα^bAβ^b molecules coexpressed with p41 only, like the floppy Aα^bAβ^b dimers produced by Ii chain-deficient mutants, display reduced mobility in SDS-PAGE (61) (Fig. 2), but in contrast to floppy Aα^bAβ^b molecules expressed by Ii chain-deficient mutants the behavior of which in functional assays suggests that they are empty or occupied by an easily displaced peptide (4, 73, 74), the floppy Aα^bAβ^b conformers coexpressed with p41 Ii chain apparently lack enhanced peptide-binding capabilities. As shown in Figure 2 A, mature Aα^bAβ^b molecules produced by DM mutants in the presence of p41 are similarly comprised of two discrete species, including a small proportion of floppy Aα^bAβ^b conformers, and predominantly the relatively compact population of Aα^bAβ^b dimers identical to those produced by DM mutants expressing wild-type or p31 Ii chain. As judged by the presence of equal amounts of CLIP peptide in boiled samples, DM mutants expressing wild-type Ii chain, or double mutants expressing p31 or p41 alone, all efficiently produce indistinguishable Aα^bAβ^b/CLIP complexes. The additional p12 product found selectively coexpressed with p41 Ii chain is discussed in detail below. The present results strongly suggest that the CLIP regions present on individual Ii chain isoforms are equally capable of promoting occupancy of the class II groove.

Recent observations suggest that individual Ii chain isoforms are processed via distinct cleavage pathways (61, 75). Consistent with this, the results above demonstrate selective expression of p12 with p41 Ii chain (Fig. 2,A). To learn more about possibly divergent Ii chain-processing pathways and the Aα^bAβ^b/CLIP complexes expressed by double mutant strains, low m.w. Aα^bAβ^b-associated products were further resolved using 15% gels. As above, we found that mature Aα^bAβ^b dimers produced by DM mutants expressing wild-type Ii chain or individual isoforms all contain indistinguishable CLIP peptides (Fig. 3,A). We found here as before (61) that the p25 cleavage product, representing the C-terminal portion of p31 beginning at Met98 inside CLIP (76), is selectively coexpressed with p31 Ii chain (Fig. 3). This p25 fragment is present in mutants carrying either the H-2^b or H-2^k haplotype (Fig. 3).

Ii chain N-terminal 10- to 12-kDa cleavage product(s) have been extensively documented (51, 75, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87). Interestingly, in Aα^kAβ^k-transfected L cells, generation of p12 requires expression of the p41 isoform (51). Consistent with this, we also observe here markedly enhanced p12 expression in the presence of p41 Ii chain (Figs. 2,A and 3A). Unlike transfected B lymphoma cells (85), thymic epithelial cell lines (86) or leupeptin-treated spleen cells (87), we found p12 production was not associated with the appearance of any larger complexes such as p70 in the nonboiled samples (Fig. 2,A). Recent results suggest the p12 fragment is released by DM (88). In contrast here, p12 expression is independent of DM activity (Figs. 2 A and 3A).

Surprisingly, we found that mutant spleen cells fail to express any p12 associated with mature Aα^kAβ^k dimers. Similar results were obtained using different mAbs including 10-2-16 (Aβ^k specific), H116-32 (Aα^k specific) (Fig. 3 B) and 11-5-2 (Aα^k-specific) (data not shown). The lack of Aα^kAβ^k-associated p12 does not simply reflect its rapid release, since similar observations were made after short periods of chase (data not shown). These findings extend earlier evidence for class II allelic differences affecting CLIP associations (89, 90, 91) and DM activities (29, 88, 92) and confirm recent data suggesting that Ii chain in association with diverse class II polymorphic residues is processed via alternative cleavage pathways (87). Taken together, these complex findings demonstrate allele- and isotype-specific Ii chain degradation intermediates.

Next we examined DM mutant spleen cells for their peptide-binding capabilities. We used the Y-Ae mAb (63) to assess formation of Aα^bAβ^b/Eα 56-73 peptide complexes in surface-staining experiments. As before (24), we also observe here DM mutant spleen cells carrying the wild-type Ii chain allele display severely compromised peptide-loading abilities, reflecting their expression of stable Aα^bAβ^b/CLIP complexes (Fig. 4). DM mutants expressing p31 alone gave virtually identical results. Surprisingly, in contrast, DM mutant spleen cells expressing p41 Ii chain efficiently bound exogenously added Eα 56-73 peptide. This striking difference cannot simply be explained by changes affecting B cell percentages or levels of Aα^bAβ^b surface expression, since both these parameters are the same for DM mutants expressing individual Ii chain isoforms (Fig. 1 and data not shown).

To learn more about distinct functions mediated by alternative Ii chain isoforms in a DM-deficient context, we also tested mutant spleen cells for their Ag presentation activities in T cell stimulation assays. As before (61), we also found here equivalent functional capabilities for individual Ii chain isoforms in the presence of DM activity (Fig. 5). Consistent with recent results (24, 25), DM mutant spleen cells expressing wild-type Ii chain appear relatively ineffective for presentation of already processed peptides and intact protein Ags (Fig. 5). As predicted by data above, DM mutant spleen cells selectively expressing p31 Ii chain gave similar results. In contrast, DM mutant spleen cells expressing the p41 isoform display markedly enhanced peptide presentation abilities in the presence of several I-A^b-restricted T cells specific for diverse epitopes. Interestingly, their Ag presentation capabilities are only partially reconstituted, since these mutant spleen cells selectively lack the ability to present intact protein Ags. These observations strengthen evidence for divergent class II Ag presentation pathways and serve to demonstrate distinct functional roles for individual Ii chain isoforms in a physiologic setting.

DM-deficient mutants exhibit partially defective CD4⁺ T cell maturation (24, 25, 26). Given the differences described above affecting peptide presentation capabilities, we were especially curious to compare the extent of CD4⁺ T cell maturation in DM double mutants expressing alternative Ii chain isoforms. As expected, we found reduced numbers of mature CD4⁺ T cells in the thymus and periphery of DM mutants carrying the wild-type Ii chain locus (Fig. 6). CD4⁺ T cell maturation was decreased to the same degree in DM mutants expressing individual Ii chain isoforms. Thus, selective peptide presentation via an exogenous pathway fails to rescue CD4⁺ T cell maturation.

It has been known for many years that structurally distinct p31 and p41 Ii chain proteins are encoded by alternative transcripts (47, 48, 49). The p41 isoform contains an extra 64-amino acid segment located immediately adjacent to the C-terminal domain responsible for trimer formation (41, 44, 45). This portion of the Ii chain contributes two extra sites for attachment of N-linked glycans and six conserved cysteines, perhaps serving a special function (49). Experiments to date strongly argue that both Ii chain isoforms function equally as class II chaperones (53, 54, 55, 56, 57, 58, 59, 60, 61). In contrast, we observe here distinct capabilities contributed by the Ii chain p41 isoform in normal APC.

We recently produced mutant mice expressing either p31 or p41 Ii chain under control of endogenous regulatory elements via homologous recombination in ES cells. A “hit and run” approach was used to introduce a specific deletion of exon 6b sequences, creating mice expressing only p31 Ii chain (60). Alternatively, to generate animals expressing p41 alone, a short cDNA fragment spanning exons 5, 6, and 6b was introduced in place of the corresponding genomic fragment (61). In contrast to conventional transgenic strains only partially reconstituted for Ii chain functions (58, 59), our mutants strongly express individual isoforms at levels equivalent to wild-type Ii chain. RNase protection assays, protein analysis, and cytoplasmic staining experiments all demonstrate that these mutant strains express the same total amount of Ii chain protein (60, 61). Both isoforms equally promote class II surface expression and peptide occupancy in the context of two different MHC haplotypes. In all functional assays, mutant strains expressing individual Ii chain isoforms gave virtually identical results (60, 61).

The present experiments tested Ii chain isoforms for their functional capabilities in a DM-deficient background. DM mutant spleen cells expressing either p31 or p41 alone gave indistinguishable surface-staining patterns using 30-2 mAb specific for Aα^bAβ^b/CLIP complexes. Similarly, immunoprecipitation experiments demonstrate that in the presence of wild-type Ii chain or individual isoforms, DM mutant spleen cells contain indistinguishable CLIP peptides. Thus, we found that alternative Ii chain isoforms act equally well as chaperones for assembly and transport of Aα^bAβ^b/CLIP complexes. On the other hand, functional experiments strongly suggest that p41 Ii chain has a distinct mode of class II occupancy.

As the simplest scenario to explain these results, perhaps alternative Ii chain isoforms give rise to structural variants of CLIP. Class II-associated Ii chain-derived peptides originally described in peptide elution studies were found to have ragged ends (36, 37, 38, 39, 93, 94, 95). Bound to the class II groove, CLIP has an extended rigid conformation (43), but in contrast on the intact Ii chain, the CLIP region comprises a highly disordered flexible domain accessible to proteases (44). The question of how the conserved Ii chain via its CLIP region binds polymorphic class II during assembly and is eventually released by DM has stimulated lively discussion (2, 3, 96, 97, 98, 99). Peptide binding studies suggest CLIP release occurs by an allosteric DM-independent mechanism involving adjacent residues outside the groove (100). Recent studies demonstrate this N-terminal extension enhances spontaneous CLIP release, but in the presence of DM activity it has little impact on CLIP dissociation rates (31). Little is known about contributions made by this portion of Ii chain toward class II/Ii chain/DM interactions under physiologic conditions in normal APC. The present experiments demonstrate indistinguishable CLIP peptides derived from both isoforms, but of course we cannot yet address possibly subtle structural differences such as an extension contributing a few extra residues. Perhaps p41-derived CLIP is slightly longer at its N terminus, has a faster spontaneous off rate, and thus selectively permits peptide capture in the absence of DM. According to this way of thinking, the extra p41-specific segment may selectively protect residues adjacent to CLIP creating a distinct DM substrate(s). Peptide elution studies should allow direct evaluation of this model.

Mature Aα^bAβ^b molecules exclusively expressed with the p31 isoform appear identical to those produced by wild-type mice (60). In contrast in the presence of p41, we found a substantial fraction of floppy Aα^bAβ^b conformers expressed alongside the predominant population of mature compact Aα^bAβ^b dimers (61). Our previous experiments failed to demonstrate any exceptional peptide-loading capabilities contributed by these floppy Aα^bAβ^b molecules in the presence of DM activity (61). However, selective peptide presentation abilities by p41 were readily detectable in a DM-deficient context. Perhaps loss of DM function permits peptide occupancy by this discrete subpopulation of floppy Aα^bAβ^b dimers. In support of this hypothesis, recent experiments demonstrate DM associates with empty class II molecules (27, 28, 29). DM interacts with class II bound to CLIP and N-terminal cleavage fragments, but not intact Ii chain (27). The formation of DM/class II complexes thus seems to coincide with removal of the C-terminal portion of Ii chain. Interestingly, this portion of Ii chain also enhances peptide binding by empty class II αβ dimers (44). The simplest possibility is its class II association masks DM contact site(s). Perhaps the extra segment encoded by exon 6b selectively confers distinct binding properties to the C-terminal portion of p41 Ii chain. The selective appearance of floppy Aα^bAβ^b conformers thus potentially reflects enhanced stabilities of unoccupied dimers due to their association with p41-specific cleavage intermediate(s).

It is obviously important to learn more about the possible relationship between formation of floppy Aα^bAβ^b dimers and selective p12 expression. Previous reports describe similar p10-p12 cleavage product(s) arising from the N-terminal portion of Ii chain (75, 81, 82, 85, 86, 87). None of these fragments have been mapped with respect to their precise C termini, but their class II association and Ab reactivities demonstrate these Ii chain cleavage intermediates contain CLIP sequences. The Aα^bAβ^b-associated p12 fragment produced by our mutant spleen cells also shows reactivity with In-1 mAb in reprecipitation experiments, and its mobility is unaffected by N-glycosidase digestion (data not shown). By these criteria, p12 produced by mutant spleen cells represents the N-terminal portion of p41 Ii chain. In contrast to leupeptin-treated spleen cells (87), B lymphomas (85), and thymic epithelial cell lines (86), here we observe p12 efficiently expressed by mutant spleen cells in the absence of protease inhibitors. In our experiments, p12 production was not associated with the appearance of any higher order complexes such as p70. It is possible these larger products may in part arise due to class II loading with cell type-specific polypeptides in the endoplasmic reticulum (101, 102). In contrast to transfected fibroblasts (75, 88), here we found no evidence for p12 stably associated with mature Aα^kAβ^k dimers. These observations confirm and extend recent experiments by Villadangos et al. (87) suggesting that class II allelic differences influence Ii chain degradation. Clearly, additional work is needed to determine the precise structure of these and other Ii chain cleavage intermediates. This information together with detailed kinetic studies should provide insight into this complex proteolytic pathway.

Here we observe selective peptide presentation by p41 in a DM-deficient context. These findings are reminiscent of those reported by Peterson and Miller (52) suggesting p41 has superior Ag presentation abilities for selected T cell epitopes. On the other hand, contradictory results have been extensively documented (53, 54, 55, 56, 57, 58, 59, 60, 61). The efficiency of Ii chain expression is clearly an important factor determining the outcome of experiments analyzing class II functional activities (58, 59, 60, 61, 103). Such discrepancies in the literature may in part reflect different expression levels for class II α and β subunits, Ii chain isoforms, and DM in these diverse systems. Transfection recipients probably also differ in their content of organelles, proteases, and molecular chaperones. In contrast here, comparisons were made using novel mouse strains created via homologous recombination in ES cells. A strong argument can be made that these mutant spleen cell populations are identical in every respect except for their DM and Ii chain expression patterns.

It seems especially interesting that Ag presentation is partially defective due to the loss of DM and p31 Ii chain expression. Thus, mutant spleen cells lacking DM and p31 Ii chain efficiently present already processed peptides, but they lack the ability to present intact Ags. These characteristics closely resemble selective defects described for nonconventional APC distinguishing endogenous and exogenous class II pathways (54, 104). It is well known that the exogenous pathway requires both Ii chain and DM activities, consistent with peptide loading of newly synthesized class II en route to the cell surface (1). In contrast, Ag uptake by recycling class II provides an alternative Ii chain- and DM-independent pathway (35, 105, 106, 107, 108). This mode of presentation may facilitate efficient capture of partially denatured proteins and antigenic fragments released by pathogenic organisms at local sites of infection. Consistent with this suggestion, recent reports describe Ii chain-independent protective host responses to the intracellular parasite Leishmania (109) and selected viruses (110). The present experiments demonstrate that p41 Ii chain selectively promotes peptide presentation via the alternative pathway in the absence of DM function. Distinct functional activities contributed by individual Ii chain isoforms may serve to promote Ag capture via divergent class II routes in various types of APC in vivo.

We thank Liz Robertson, Ross Waldrip, and members of the laboratory for helpful discussions; Sasha Rudensky for the 30-2 mAb; Jennifer Lower for the DMα genotyping protocol; Jennifer Lower and Debbie Pelusi for valuable assistance screening mutant progeny; Patti Lewko and Mark O’Donnell for careful maintenance of the mouse colony; Pippa Marrack, Sasha Rudensky, and Nilabh Shastri for T cell hybridomas; Carol Plunkett for secretarial assistance; and Renate Hellmiss for preparing the figures.