The invariant chain (Ii) shows promiscuous binding to a great variety of MHC class II allotypes. In contrast, the affinities of the Ii-derived fragments, class II-associated Ii peptides, show large differences in binding to class II allotypes. The promiscuous association of Ii to all class II polypeptides therefore requires an additional contact site to stabilize the interaction to the polymorphic class II cleft. We constructed recombinant molecules containing the class II binding site of Ii (CBS) and tested their association with HLA-DR dimers. The CBS fused to the transferrin receptor mediates binding of transferrin receptor-CBS to class II dimers. Within the CBS, deletion of a sequence N-terminal to the groove-binding motif abolished binding of Ii to DR. A promiscuous class II binding site was identified by reinsertion of the N-terminal residues, amino acids 81–87, of Ii into an Ii mutant that lacks the groove-binding segment. DR allotype-dependent association of Ii was achieved by insertion of antigenic sequences. The promiscuous association, in contrast to the class II allotype-dependent binding of Ii, is important to prevent interaction of class II dimers to nascent polypeptides in the endoplasmic reticulum.
The first indication that the invariant chain (Ii)3 plays a functional role in Ag processing was obtained with Ii-deficient APCs (1). Analysis of Ii-deficient and Ii31/Ii41 transgenic mice supported this observation and, in addition, revealed the importance of Ii for the development of CD4+ T cells (2, 3, 4, 5, 6). In recent years, several molecular functions of Ii have been identified. A major function of Ii is the targeting of MHC class II dimers to the endocytic pathway (7, 8, 9). On this route, Ii degradation controls the acquisition of peptides by class II molecules in specialized loading compartments (reviewed in 10 . In addition, association of class II molecules with Ii supports the assembly of MHC class II polypeptides in the endoplasmic reticulum and is essential for certain allotypes to form functional αβ dimers (2, 11, 12). Peptide-free class II molecules are stabilized by association with Ii, and aggregation is prevented (13, 14, 15, 16). A segment encoded by exon 3 of the Ii gene is indispensable for the formation of the MHC class II/Ii complex (17, 18, 19). Fragments of processed Ii are called class II-associated Ii peptides (CLIP), the sequences of which are encoded by exon 3. The sequence of CLIP is contained in the class II binding site (CBS) of Ii. CLIP were found associated with various MHC class II allotypes (reviewed in 20 . The x-ray crystal structure of CLIP bound to HLA-DR3 revealed that a number of residues in the sequence 91–99 (groove binding site) interact with specific pockets of the MHC class II peptide-binding groove, whereas the N terminus of the Ii fragment is disordered in the crystal (21). Studies on proteolytic fragments of Ii suggest that the CBS also binds to the Ag-binding groove of class II dimers and that CLIP is a naturally occurring degradation intermediate (22, 23, 24, 25). The affinities of CLIP for different MHC allotypes can vary over several orders of magnitude (26). This could indicate that the affinity of CLIP regulates the binding of antigenic peptides. Release of CLIP from MHC class II dimers for which it has a high affinity requires the accessory molecule HLA-DM (27, 28, 29). Allotypes showing a high off rate for CLIP apparently do not depend on the catalytic function of DM for dissociation, although the acquisition of antigenic peptides might be facilitated by DM (30, 31, 32). In these cases, the acidic conditions in the MHC class II loading compartments might be sufficient for dissociation. The strong binding of Ii to different class II allo- and isotypes even across species barriers suggests a highly conserved interaction between Ii and class II dimers (33). Thus, it is conceivable that contacts adjacent to the polymorphic class II groove are responsible for stabilizing the binding of Ii to MHC class II molecules.
Here, we demonstrate both allotype-dependent and promiscuous binding of rIi polypeptides mediated by different Ii sequences to MHC class II dimers. A membrane proximal region of Ii mediates binding to three DR allotypes, whereas the groove-binding segment of Ii replaced by an antigenic sequence leads to allotype-dependent association of class II dimers.
Materials and Methods
DNAs for transferrin receptor (TFR), rIi, and DR chains were expressed under SV40 promoter control in the pcEXV3 or pSV51 expression vector (7, 34). For construction of TFR-CBS, bp 237–416 of the murine Ii31 cDNA, encoding Ii amino acids (aa) 80–139, was excised by AluI digestion. Maintaining the reading frame, this Ii fragment was inserted into the Eco47III restriction site at the 585-bp position of the human TFR cDNA.
For construction of the deletion mutant ΔIi80-93, bp 233 to 272 of the human Ii33 cDNA (pSV51-huIi) was removed by digestion with FspI, followed by religation. In ΔIi105-157, the segment bp 309 to 465 was excised using NcoI. The cDNA/genomic DNA fusion construct for ΔIi81-127 has been described previously (17). QASLALSYRLNMFTP is a peptide derived from the major outer membrane protein of Chlamydia trachomatis (MOMP). The aa 81–87 (PKSAKPV) of Ii were designated the promiscuous binding site (PBSite). To obtain rIi MOMP, rIi spacer 1, rIi PBSite, and rIi spacer 2, the aa shown in Figure 1 were introduced in ΔIi81-127. For construction of rIi MOMP and rIi PBSite, the two oligonucleotides 5′AGCTTCAAGCAAGTTTGGCTCTCTCTTACAGACTGAATATGTTCACTCCCA and 5′AGCTTCCGAAATCTGCCAAACCTGTGCTGCAGA were hybridized with their complementary strands, thereby generating HindIII overlaps, and ligated into the HindIII restriction site of ΔIi81-127 (at position bp 4717 of the genomic sequence; Fig. 2). Due to the cloning procedure, there is an additional C-terminal KL and LQKL that connects MOMP and Ii aa 81–87, respectively, to Ii aa 128. In the constructs rIi spacer 1 and rIi spacer 2, the oligonucleotides coding for MOMP and for PBSite were inserted in the inverse orientation, thereby encoding a nonrelated sequence with the same number of aa. Oligonucleotides encoding aa 17–31 of the influenza virus matrix protein (MAT) sequence were introduced into genomic Ii DNA. The small fragment between the restriction sites HindIII (bp 4717 in exon 2) and NcoI (bp 4945 in exon 3) was replaced by the oligonucleotides AGCTTTCGGGCCCGCTGAAGGCGGAGATCG/ACGCAGCGCCTCGAGGACGTGTC. This resulted in replacement of aa 81–101 of Ii by MAT aa 17–31, with alanine or threonine at residue 89 of rIiMAT.
Cell lines and transfections
COS1 (CRL-1650; American Type Culture Collection (ATCC), Rockville, MD) and COS7 (CRL-1651; ATCC) cells were cultivated in high glucose DMEM supplemented with 10% FCS, 100 μg/ml penicillin, 100 U/ml streptomycin, 1 mM sodium pyruvate, 10 mM HEPES, and 2 mM l-glutamine. Transient transfections were performed by DEAE-mediated DNA transfer (10 μg DNA/3 × 106 COS1 cells; 24 h), electroporation (210 V/1.2 mF, 25 μg DNA/6 × 106 COS1 cells; 72 h), and liposome-aided transfer (1 μg DNA/5 × 105 COS7 cells; 48 h) as described (35, 36, 37).
Metabolic radiolabeling of proteins and immunoprecipitation
After incubation of cells for 1 h at 37°C in methionine-free RPMI 1640, newly synthesized proteins were labeled for 10 min with 50 μCi [35S]methionine (Amersham, Braunschweig, Germany) in methionine-free RPMI (supplemented with 10% dialysed FCS, 1 mM sodium pyruvate, 2 mM glutamine, and 10 mM HEPES). The cells were lysed with 1% Nonidet P-40 in Tris-buffered saline in the presence of protease inhibitors (1 μM PMSF and 0.024 trypsin inhibitor units of aprotinin per ml). Lysates were precleared by a 2-h incubation with Sepharose CL4B. To reduce background in some experiments, the lysates were first adjusted to pH 5 by the addition of acetic acid. Acid-precipitated proteins were spun down by centrifugation. The supernatant was neutralized and further cleared by addition of Sepharose CL4B. mAbs used for immunoprecipitation were PA1 (αTFR, a gift from Dr. G. Moldenhauer, Heidelberg, Germany), P4H5 (α-Ii peptide RPMSMDNMLLGPVKNVTK; 38 , VicY1 (α-human Ii; 39 , In1 (α-murine Ii; 40 , ISCR3, and I251SB (α-HLA-DR chains; Refs. 41 and 42) For immunoprecipitation, mAbs were added to the cell lysates in the presence of 10 μl protein A-Sepharose CL4B. After overnight incubation, precipitates were washed three times in 0.25% Nonidet P-40 in Tris-buffered saline (pH 6.8).
Immunoprecipitates were boiled in reducing sample buffer for 4 min and analyzed on 10 to 15% SDS polyacrylamide gradient gels or 13% SDS polyacrylamide gels (17). For two-dimensional nonequilibrated pH gradient electrophoresis (NEPHGE), protein A-Sepharose pellets were incubated for 1 h at room temperature in NEPHGE sample buffer (9.5 M urea, 2% Nonidet P-40, 2% ampholines (pH 3.5–10), and 50 mM DTT), and the proteins were separated according to their charge using 4% polyacrylamide rod gels (first dimension). The rod gels were then incubated for 2 h in reducing sample buffer. In the second dimension, the proteins were separated according to molecular mass in 13% SDS-polyacrylamide gels.
Transfer of a class II binding Ii segment to the TFR
In an attempt to localize contact sites of Ii that stabilize the interaction of the groove-binding segment to class II polypeptides, we inserted a sequence of Ii that contains the previously identified CBS to the luminal domain of another type II membrane protein, the TFR. The segment of the Ii cDNA encoding aa 80–139 was ligated into the TFR cDNA (Fig. 3,A). This Ii sequence contains the CBS (aa 81–109) and an adjacent stretch of 30 aa that provides an epitope recognized by the mAb P4H5. This recombinant TFR-CBS forms S-S-linked dimers, and its N-linked glycan side chains acquire Endo H resistance upon intracellular transport (G. Reuter and N. Koch, unpublished observations). To explore the binding properties of the recombinant TFR to HLA-DR molecules, various combinations of TFR-CBS, TFR, and Ii constructs were transiently coexpressed with DRα and -β cDNAs in COS cells (Fig. 3,B). Since some DR allotypes were reported to bind TFR-derived sequences, we used the DR3 allele, because no TFR peptides had been eluted from this allotype (43). Cells were radiolabeled, lysed, and immunoprecipitated with mAbs specific for DR, TFR, or Ii. The mAb against TFR immunoprecipitates TFR and TFR-CBS (Fig. 3 B). The 60-aa Ii segment increases the size of TFR-CBS compared with TFR. The Ii-specific mAb P4H5 does not bind to TFR, but immunoisolates TFR-CBS. This demonstrates that the Ii segment is positioned in a site accessible for the mAb P4H5. Immunoprecipitates of DR molecules reveal that large amounts of TFR-CBS, similar to the amounts of Ii in the first lane, are coisolated. Precipitation with Ii or TFR Abs do not exhibit class II bands, because the TFR-CBS is in large excess. This result is also obtained with wild-type Ii (data not shown).
Coprecipitation of TFR-CBS with DR indicates that the Ii segment aa 80–139 endows class II binding properties. The rest of Ii appears to be dispensable for interaction to DR molecules. This demonstrates that the Ii segment aa 80–139 alone is sufficient for productive association with MHC class II molecules. The association may be due to the groove-binding sequence alone, or additional sequences may stabilize the Ii/class II groove interaction.
An N-terminal deletion of the class II binding sequence of Ii abrogates association with class II molecules
We studied whether the sequences flanking the groove-binding segment (aa 91–99) influence Ii/class II association. With two available restriction sites, sequences adjacent to the groove-binding segment of Ii were deleted (Fig. 4,A). In ΔIi80-93, the truncated anchor positions Met91 and Met93, in addition to the deleted aa 80–90, could impair binding to class II dimers. The deletion in ΔIi105-157 starts exactly beyond the sequence of CLIP (aa 81–104). ΔIi80-93 or ΔIi105-157 was coexpressed with DR cDNAs and immunoprecipitated with anti-Ii or anti-DR mAbs. Since ΔIi80-93 comigrates with the DR β-chain in one-dimensional SDS-PAGE, the immunoprecipitates were analyzed in two-dimensional NEPHGE/SDS-PAGE (Fig. 4,B). The anti-Ii immunoprecipitates show that the recombinant Ii and wild-type Ii are expressed at a high level. Class II is not detected because of the excess of Ii. DR precipitates reveal that wild-type Ii is coprecipitated (Fig. 4,B, top), whereas ΔIi80–93 is not detected in DR precipitates (Fig. 4,B, middle). This result was verified by endoglycosidase H digestion and one-dimensional separation of the immunoprecipitates, in which ΔIi80-93 and DRβ have different m.w.s according to the differing number of digested N-glycan side chains (data not shown). In contrast to the recombinant Ii lacking aa 80–93, the mutant ΔIi105-157 is coisolated with MHC class II molecules from cotransfectants (Fig. 4 B, bottom).
A sequence adjacent to the groove-binding segment of Ii mediates promiscuous binding of Ii to class II molecules
The lack of association of ΔIi80-93 to DR molecules could suggest that residues N-terminal to the groove-binding motif of Ii are important for the association with MHC class II molecules. We investigated whether in the absence of the groove-binding segment reintroduction of N-terminal residues would restore binding to class II. Fig. 2 shows a schematic representation of a recombinant Ii (rIi PBSite) in which aa 81–87 were introduced into ΔIi81-127. These two deletion constructs and DR cDNAs were coexpressed in COS cells. rIi PBSites was coprecipitated with DR1; while consistent with our previous results, ΔIi81-127 could not be coisolated (Fig. 5,A, left) (17). Ii immunoprecipitates show that the Ii derivatives were strongly expressed (Fig. 5 A, right). Ii precipitates usually do not show class II bands. To examine the specificity of aa 81–87 binding, a construct with an unrelated spacer sequence of the same length (rIi spacer 2) was generated. Transient expression with DR1 and immunoprecipitation with anti-DR Abs revealed that rIi spacer 2 did not bind to class II dimers, which supports the specific binding of rIi PBSite to DR molecules. This result indicates that aa 81–87 mediate binding of an Ii mutant, which lacks the groove-binding segment, to class II molecules.
The important question now was whether rIi PBSite binds to other DR allotypes. rIi PBSite was coexpressed with DR3 or DR4 dimers. Class II immunoprecipitates show that rIi PBSite also associates with DR3 and DR4 molecules (Fig. 5, B and C). Again, ΔIi81-127 did not coprecipitate with DR. This result demonstrates promiscuous binding of rIi PBSite to three DR allotypes.
Class II binding of a mutant Ii is mediated by an antigenic sequence
Ii thus appears to harbor a promiscuous allotype-independent site in addition to an allotype-dependent class II binding site. Allotype-dependent binding of CLIP has been demonstrated, but a basis for the allotype-independent binding of Ii has not been defined. To monitor DR allele-specific binding of Ii, we introduced antigenic sequences into ΔIi81-127. At first, we studied whether association of a rIi, with the groove-binding segment replaced by an antigenic sequence, to DR dimers is possible. The sequence of MOMP, QASLALSYRLNMFTP, binds to DR3 (44). ΔIi81-127 lacking the CBS was used to introduce the sequence of MOMP (compare Figs. 1 and 2). Coexpression of this chimeric Ii with DR3 molecules and subsequent immunoisolation of DR molecules show coisolation of rIi MOMP (Fig. 6). As a control, rIi spacer 1, with an unrelated sequence of the same length as MOMP, does not coprecipitate with DR3. Immunoprecipitation of Ii indicates equivalent expression of both rIi. Class II is not detected in the Ii precipitates.
Allotype-dependent class II binding of chimeric Ii with the CBS replaced by MAT
To determine allotype-dependent binding, we introduced the sequence of MAT into Ii. MAT is a peptide comprising aa 17–31 of influenza virus matrix protein. The MAT peptide 17SGPLKAEIAQRLEDV31 has been shown to bind to DR1, while the A in position 25 replaced by a T leads to binding to DR4 (45). Both sequences were cloned into ΔIi81-101 (Figs. 1 and 2) and expressed in COS cells with either DR1 or DR4 (Fig. 7). Immunoprecipitation demonstrates that rIi MAT A is coprecipitated with DR1, whereas the mutation of Ala25 to Thr almost completely abrogates binding to DR1. Conversely, rIiMAT T binds to DR4, whereas rIi MAT A does not. This shows that the binding of the rIi MAT constructs is DR allele specific and that an allotype-independent binding sequence is missing in these constructs. The expression of the recombinant Ii is similar in all samples (Fig. 7, bottom). Class II is not visible in Ii precipitates.
Several pieces of data indicate that Ii and CLIP bind with the same sequence to MHC class II dimers. The observation that both CLIP and an Ii degradation product containing the CBS can stimulate the same T cell clone argues that in the intact protein, Ii interacts with class II in much the same way as the peptide (25). However, the binding characteristics of CLIP and the CBS of Ii show one major difference. The strength of the interaction of CLIP with individual allotypes varies up to four orders of magnitude (26). Affinity measurements for the association of Ii with various MHC class II alleles are not available. However, if binding differences of this order existed, they should have been detected in biochemical studies of class II/Ii interaction. Thus, the nonameric complex of class II and Ii probably bears additional contact sites that stabilize the interaction between Ii and class II molecules. We report here that the sequence comprising aa 81–87 (PKSAKPV) is important for binding of a recombinant Ii protein to class II polypeptides. Despite the absence of the groove-binding segment, rIi PBSite shows promiscuous binding to three DR alleles. In particular, HLA-DR4, which possesses only moderate binding capacities for CLIP (22), binds efficiently to rIi PBSite. The PBSite of Ii may compensate for the allele-regulated interaction of aa 91–99 with polymorphic residues of the class II pockets. The proximity of the PBSite to the groove-binding motif could be of particular importance for stabilization of the class II cleft. A stabilizing effect of elongations of sequences accommodated within the class II groove has previously been shown for antigenic peptides. The elongation of synthetic HEL peptides at either end of the core region increases the stability of the class II/peptide complex, an N-terminal elongation being particularly effective (46).
Consistent with our findings, almost identical binding characteristics were found for the Ii76-91 peptide to four murine alleles, whereas the binding of CLIP differed according to the individual allotypes (47). These authors in addition reported that the groove-binding peptide inhibits binding of antigenic peptides. In contrast, the Ii76-91 peptide enhances binding of two different HEL peptides.
Peptides containing aa 81–87 of Ii possibly bind to the α1 domain of MHC class II dimers, as their binding site seems to overlap with the Staphylococcus enterotoxin B contact region (48, 49). The interactions of these Ii-derived peptides and of intact Ii with class II seem to involve the same class II domain, as the peptide binding can be inhibited by soluble Ii (49). Possible contact residues on the α1 domain were suggested to be Asp17, Glu21, and/or Asp35 of DR (48). These residues were proposed to interact with Lys83 and/or Lys86 of human Ii. These residues are highly conserved between mice and humans, and there are putative acidic counterparts for the lysine residues in the α1 domain of H2-A, H2-E, DR, DP, and DQ. The interaction of these class II residues with the lysine-rich PBSite could thus have a stabilizing effect on Ii binding irrespective of the class II allele. A binding domain containing residues 81–87 of Ii might not be formed by a peptide of the same sequence. This could explain why a strong binding of the free peptide to class II molecules was not observed (48). The aa 81–87 sequence could be part of a larger motif that is destroyed by Ii degradation.
Presumably, there are several contact sites of Ii and class II in the (αβIi)3 complex. A recombinant, soluble Ii lacking the N-terminal part up to aa 117 augmented antigenic peptide binding of DR dimers, suggesting that the C-terminal part of Ii also interacts with class II molecules (24). Recently, it was shown that the membrane-proximal region and the sequence from which CLIP is derived are extended in the Ii trimer (50). This open structure could allow binding to the class II groove. Our result that aa 80–139 of Ii fused into the TFR sequence 107 aa from the transmembrane domain permits binding to class II polypeptides suggests that this open structure is retained in the TFR-CBS. Presumably other sequences of Ii are not essential for class II interaction.
The interaction between class II and aa residues 81–87 of the CBS suggests how Ii release could be controlled. The acquisition of peptides by class II dimers is postponed until a final step of Ii degradation, when the CLIP sequence is proteolytically separated from the highly conserved 22 aa between the transmembrane domain and the CBS. The stepwise release of Ii from MHC class II molecules is initiated by cleavage at a position C-terminal to the groove-binding motif (51). The Ii fragment LIP is still associated with class II dimers (52) and might dissociate from the nonameric complex. The trimer could be more accessible to proteases and to HLA-DM than the nonamer (28). If LIP is not removed by DM, further cleavage at residues 80–87 and C-terminal trimming of the fragment yields CLIP bound to MHC class II dimers. This degradation step might impair interactions of the proline-lysine-rich motif of CLIP with class II polypeptides. The allotype-dependent binding of aa 91–99 then could determine whether accessory molecules such as HLA-DM are necessary for CLIP release.
The promiscuous binding of Ii to class II molecules may facilitate the assembly of the nonameric class II/Ii complex in the endoplasmic reticulum and could postpone the binding of antigenic sequences to MHC class II vesicles. In this compartment, CLIP governs the acquisition of peptides by class II dimers as has been demonstrated with H2-M-deficient mice (53, 54, 55). It remains to be shown whether CLIP modulates an immune response as a consequence of the MHC polymorphism.
Positioning of an antigenic sequence into Ii might be useful for raising cellular vaccines. A similar strategy recently was described (56). In this report, a T cell epitope from hemagglutinin was used for insertion into Ii beyond aa 90. This construct contains the PBSite that we identified. A transfectant containing this chimeric Ii was efficiently recognized by a hemagglutinin-specific T cell clone. By comparison, we introduced the antigenic sequence of MAT into an Ii deletion mutant that lacks the PBSite and obtained DR allotype-dependent binding of MAT to class II molecules. This result suggests binding of the recombinant MAT to the class II peptide-binding groove. Our finding that an Ii sequence adjacent to the groove-binding segment stabilizes binding of Ii to class II molecules is consistent with a recent publication (57). In this report, antigenic sequences introduced into Ii mediate resistance of the DR1/Ii complex to SDS treatment. This result is consistent with binding of the antigenic sequence into the peptide binding cleft of class II molecules.
We thank Drs. O. Bakke, J. Miller, and H. Ploegh for providing cDNAs and vectors; A. König for excellent technical assistance; and Drs. P. Gleeson, R. Lindner, F. Sanderson, and J. Trowsdale for critical discussions.
This work was supported by the Sonderforschungsbereich 284, Sonderforschungsbereich 502 and by the Graduiertenkolleg “Funktionelle Proteindomänen.” N.K. was supported on his sabbatical leave by a grant from the Volkswagenstiftung.
Abbreviations used in this paper: Ii, invariant chain; CLIP, class II-associated Ii peptides; CBS, class II binding site of Ii; PBSite, promiscuous binding site of Ii; aa, amino acids; TFR, transferrin receptor; MOMP, peptide from major outer membrane protein of C. trachomatis; MAT, peptide from influenza virus matrix protein; NEPHGE, nonequilibrated pH gradient electrophoresis.