Presentation of antigenic peptides to CTLs at the cell surface first requires assembly of MHC class I with peptide and β2-microglobulin in the endoplasmic reticulum. This process involves an assembly complex of several proteins, including TAP, tapasin, and calreticulin, all of which associate specifically with the β2-microglobulin-assembled, open form of the class I heavy chain. To better comprehend at a molecular level the regulation of class I assembly, we have assessed the influence of multiple individual amino acid substitutions in the MHC class I α2 domain on interaction with TAP, tapasin, and calreticulin. In this report, we present evidence indicating that many residues surrounding position 134 in H-2Ld influence interaction with assembly complex components. Most mutations decreased association, but one (LdK131D) strongly increased it. The Ld mutants, with the exception of LdK131D, exhibited characteristics suggesting suboptimal intracellular peptide loading, similar to the phenotype of Ld expressed in a tapasin-deficient cell line. Notably, K131D was less peptide inducible than wild-type Ld, which is consistent with its unusually strong association with the endoplasmic reticulum assembly complex.
Before binding peptides, MHC class I molecules are associated with several different endoplasmic reticulum (ER)3 proteins that are collectively referred to as the class I assembly complex. Members of this complex include TAP, tapasin, and calreticulin (1). MHC class I heavy (H)-chains detected in association with this ER peptide-loading complex are peptide empty, β2-microglobulin (β2m) assembled, and in an open conformation (2, 3). After peptide binding, the fully assembled class I heterotrimers are released to transit to the cell surface and display peptides to CD8+ T cells.
Although the selective roles of the members of the ER peptide-loading complex are unknown, these proteins facilitate assembly and surface expression of functional class I. In addition to class I, calreticulin is known to chaperone other glycoproteins in the ER (4). It has been suggested that tapasin and/or TAP may also serve class I as chaperones (5), and recent evidence supports this theory. Tapasin facilitates MHC class I surface expression, even apart from its interactions with TAP (6). Furthermore, it has been shown that TAP is either an active chaperone or a necessary accessory to a molecule that chaperones class I (7).
Recent evidence suggests that ER chaperones may interact differently with various allelic forms of MHC class I H-chain. In one study, TAP association has been reported to vary among HLA alleles. Specifically, several HLA alleles were evaluated for TAP interaction, and the majority of HLA-B alleles examined were found to associate very inefficiently with TAP (8). This observation might be due to competition among class I H-chains for TAP (9), or to differential association of class I H-chains with tapasin, which is required for class I/TAP association (2). Consistent with this latter explanation, allelic variation in tapasin dependence for assembly has been shown for class I molecules. Cotransfection of tapasin, along with B*2705, B8, or B*4402, into the tapasin-negative 721.220 cell line did not boost B*2705 surface expression; however, it did increase B8 expression by 5-fold and B*4402 expression by 25-fold (10). At this time, species or allelic variations in class I/calreticulin association have not yet been reported.
Three spatially distinct sites on the MHC class I H-chain have been shown biochemically to influence interaction with both TAP and calreticulin. These sites are indicated on the Ld three-dimensional structure (11) in Fig. 1. The sites are the N-linked carbohydrates at position 86 in the α1 domain (3), position 134 in the α2 domain (12, 13, 14), and an acidic residue at position 227 in the α3 domain (3). (We have also found that an Ld mutant with changes at both positions 227 and 229 loses TAP association (data not shown)). Of these three sites, position 134 would seem the most likely to be highly influenced by class I/peptide binding. Because calreticulin has been shown to bind glycosyl groups on other proteins (4), it is likely that calreticulin binds the carbohydrate at position 86 on class I. In a recent study, data were presented that suggested TAP binding was influenced, directly or indirectly, by residues in the α3 domain. In this case, amino acids 219–233 of the H-2Db α3 domain were exchanged for the same number of amino acids from the MHC class II chain. This mutant was not expressed at the cell surface, and its interaction with TAP was sharply reduced (15). Also, a recent publication suggested that a mutation at position 222 in the α3 domain could prevent association with tapasin and TAP (16). However, any influence of this amino acid substitution on calreticulin association cannot be clearly ascertained from this report. In total, several residues within a certain area of the α3 domain have been implicated in assembly complex interaction; however, at this time only one α2 domain residue (position 134) has been identified as important to such associations.
Two possible models for chaperone association with class I are: 1) separate sites on class I interact with different chaperones and each element of the assembly complex binds independently, and 2) class I/β2m must interact with all assembly components to produce full complex stability (i.e., a cooperative binding model). Because TAP is not associated with class I in tapasin-deficient 721.220 cells, tapasin may act as a bridge between class I and TAP, as suggested previously (2), or the binding of TAP may be in some other way dependent upon the presence of tapasin. Interestingly, Sadasivan et al. (2) also showed a reduction in the amount of class I and tapasin associated with calreticulin in the absence of TAP. Nevertheless, evidence from experiments in which a truncated form of tapasin was used showed that tapasin has a chaperone function for class I that is independent of physical association with TAP (6). The extent to which calreticulin and tapasin binding to class I is cooperative is an open question; certainly, if they interact cooperatively, the difficulty in discerning separate chaperone functions for each is increased.
To improve our understanding of the protein interactions vital to MHC assembly, we have examined the effect of multiple, individual point mutations in the MHC class I α2 domain on association with TAP, tapasin, and calreticulin. To be able to assess the relationship between class I H-chain folding status and the function of the assembly complex, we have used as our experimental model H2-Ld, for which Abs specific for both open and folded conformations are available. Our results show that not only position 134 but also a large number of amino acid residues around it are involved in interactions between the assembly complex and the open form of Ld. Within the 128–136 sequence, individual mutations at all but one position caused decreased association. Surprisingly, a Lys to Asp substitution at position 131 greatly strengthened the interaction. Furthermore, the lack of assembly complex binding was correlated with functional consequences, as exhibited by altered accessibility to exogenous peptides.
Materials and Methods
Cell lines, mutagenesis, and transfection
The 721.220 cell line is a human B lymphoblastoid cell line that does not express tapasin (2). The 721.221 cell line is a closely related cell line that does express tapasin (17, 18). L-Ld cells were made by introduction of the Ld gene into Ltk− DAP-3 (H-2k) fibroblast cells (19). All Ld mutants are named as follows: original amino acid residue, position, and substituted amino acid residue. The construction of the LdN86Q and LdD227K mutant cDNAs and their transfection into DAP-3 has been described previously (3, 20). For the generation of Ld mutants with amino acid substitutions in the 128–137 loop, the following approach was taken to make nonconservative mutations in such a way as to require the minimal DNA sequence changes. First, two unique blunt-end restriction sites, SmaI and EcoRV, were sequentially introduced into an Ld cDNA (21) in the vector RSV.5neo (22) using the Quik Change Mutagenesis Kit from Stratagene Cloning Systems (La Jolla, CA). The synthetic oligodeoxynucleotides that were used for mutagenesis to insert the SmaI site were as follows: 5′-GGCTGCGATTACATCGCCCGGGACCTGAAAACGTGG-3′ and 5′-CCACGTTTTCAGGTCTTCGTCCCGGATGTAATCGCAGCC-3′. For the mutagenesis required for insertion of the EcoRV restriction endonuclease site, the synthetic oligonucleotides used were 5′-GCGGCGGACATGGCGGCGGATATCACCCGACGCAAGTGG-3′ and 5′-CCACTTGCGTCGGGTGATATCCGCCGCCATGTCCGCCGC-3′. These two mutations would result in the following codon changes. For the SmaI mutation, residues 126 and 127 would be changed from Leu and Asn to Arg and Asp (GCC CTG AAC to GCC CGG GAC). For the EcoRV mutation, residue 141 would be changed from Gln to Asp (CAG ATC to GAT ATC). The unique SacII/Sse 8387I fragment encompassing these two mutations was sequenced completely and swapped with the corresponding fragment from RSV.5neo-Ld. The resultant mutant Ld construct was then digested with SmaI and EcoRV sequentially and ligated with pairs of annealed and phosphorylated oligodeoxynucleotides that would include each individual mutation on the loop while reverting the changes at residues 126, 127, and 141. The recombinants were first screened by restriction mapping and then confirmed by DNA sequencing for the correct orientation. For example, the pairs of oligodeoxynucleotides used to create the LdE128R mutation were 5′-TGAACAGAGACCTGAAAACGTGGACGGCGGCGGACATGGCGGCGCAG-3′ and 5′-CTGCGCCGCCATGTCCGCCGCCGTCCACGTTTTCAGGTCTCTGTTCA-3′. In the correct orientation, the nucleotide sequence corresponding to codons 126–142 would be CTG AAC AGA … . . CAG ATC.
Cells were maintained at 37°C, 5% CO2 in RPMI 1640 (Life Technologies, Gaithersburg, MD) containing 10% bovine calf serum (HyClone Laboratories, Logan, UT), glutamine, pyruvate, and penicillin/streptomycin, or were cultured at 37°C, 7% CO2 in DMEM (Life Technologies) containing 10% FCS (HyClone Laboratories), glutamine, pyruvate, and penicillin/streptomycin.
Ab, immunoprecipitations, and Western blotting
The mAb 64-3-7 is an IgG2 Ab specific for the α1 domain of open Ld (19, 23, 24, 25); mAb 30-5-7 is an IgG2 Ab that binds the α2 domain of folded Ld (24, 25, 26). Ld forms with the 64-3-7 epitope are not recognized by several conformation-sensitive mAbs (e.g., 30-5-7 and B22/249), which have epitopes that surround the peptide-binding groove. This serologic pattern suggests that the Ld 64-3-7+ form has an open and unfolded cleft (23). Peptide binding to 64-3-7+ Ld causes a conversion to the folded (30-5-7+) form, as demonstrated by titration of radioiodinated peptide ligand into cell lysates and immunoprecipitation of the Ld molecules. With increasing concentrations of peptide ligand, the amount of 64-3-7+ Ld decreased and the amount of 30-5-7+ Ld increased (25). Furthermore, the labeled peptide was coprecipitated in a dose-dependent fashion with the 30-5-7+ Ld form and could be simultaneously visualized on the autoradiograph (25). In vivo, a precursor-product relationship exists between 64-3-7+ Ld and 30-5-7+ Ld, as indicated by pulse-chase analysis. As the chase time increases, 30-5-7+ Ld levels were shown to rise and 64-3-7+ Ld levels to drop (25).
The anti-calnexin serum specific for a C-terminal segment from mouse and human calnexin (27) was donated by Dr. David McKean (Mayo Clinic, Rochester, MN). Anti-calreticulin serum (28) was purchased from Stressgen (Victoria, British Columbia, Canada). The murine anti-TAP-1 serum was generated against a TAP-1 peptide coupled to keyhole limpet hemocyanin (20). The rabbit anti-mouse tapasin serum was also made against a synthetic peptide (GPAIECWFVEDAGGGGLSKC) coupled to keyhole limpet hemocyanin.
For radiolabeling, cells were preincubated for 30 min at 37°C in culture media without methionine. Next, [35S]methionine (100 μCi/ml) was added, and the cells were radiolabeled for 30 min. The cells were then washed three times in PBS containing iodoacetamide (Sigma, St. Louis, MO) and lysed in 1% 3-([cholamidopropyl]dimethylammonio)-1-propanesulfonate (CHAPS) (Boehringer-Mannheim, Indianapolis, IN) in Tris-buffered saline (pH 7.4) with freshly added 0.2 mM PMSF (Sigma) and 20 mM iodoacetamide. The lysis buffer was supplemented with a saturating volume of mAb before its addition to pelleted cells. After incubation for 1 h on ice, nuclei were removed by centrifugation, and lysates were incubated with protein A-Sepharose beads (Pharmacia, Piscataway, NJ). The beads were washed four times with 0.1% CHAPS in Tris-buffered saline (pH 7.4), and the samples were eluted by boiling in 0.125 M Tris (pH 6.8)/2% SDS/12% glycerol/2% bromophenol blue. All immunoprecipitates were electrophoresed on 4–20% or 8% acrylamide/Tris-glycine or 7% acrylamide/NuPage gels (Novex, San Diego, CA). Gels were soaked in Amplify (Amersham, Boston, MA) plus 2% glycerol, dried, and exposed to BioMax MR film (Eastman Kodak, Rochester, NY) at −70°C for varied lengths of time.
For Western blotting, immunoprecipitates electrophoresed by SDS-PAGE as described above were transferred to Immobilon P membranes (Millipore, Bedford, MA). After overnight blocking, membranes were incubated in a dilution of Ab for 2 h, washed three times with PBS/0.5% Tween 20, and incubated for 1 h with biotin-conjugated goat anti-mouse or anti-rabbit IgG (Caltag Laboratories, San Francisco, CA). Following three washes with PBS/0.05% Tween 20, membranes were incubated for 1 h with streptavidin-conjugated HRP (Zymed, San Francisco, CA), washed three times with PBS/0.3% Tween 20, and incubated with enhanced chemiluminescence Western blot developing reagents (Amersham). Membranes were exposed to BioMax MR or ML film for varied lengths of time.
To analyze peptide-pulsed cells by FACS, the cells were incubated overnight at a density of 1 × 106 cells/ml with defined concentrations of peptide in DMEM supplemented with glutamine, penicillin, streptomycin, and 10% bovine calf serum. After incubation, the cells were centrifuged and resuspended at 5 × 106/ml in PBS with 0.2% BSA and 0.1% sodium azide. The cell suspension, in 0.1-ml aliquots, was distributed to wells of a 96-well plate. The cells were incubated with saturating concentrations of Ab or PBS/BSA/azide alone for 30 min at 4°C, washed twice, and incubated with a fluorescein-conjugated, Fc-specific F(ab′)2 fragment of goat anti-mouse IgG for 30 min at 4°C. The cells were then washed twice, resuspended, and analyzed on a FACScan flow cytometer (Becton Dickinson, Mountain View, CA). Statistical analyses were performed with Cell Quest software.
Results and Discussion
We have explored chaperone influence on MHC class I assembly by thorough investigation of the interactions between the α2 domain and the assembly complex. Mutations over a considerable region surrounding position 134, as well as mutations at positions 86 and 227, were discovered to alter the interaction between Ld and TAP/tapasin/calreticulin. These individual amino acid changes were found to have consequences for class I/assembly complex interaction as measured by both biochemical and functional assays.
Individual substitutions between position 128 and 136 of Ld disrupt TAP and calreticulin interaction
It has been reported that a Lys for Thr substitution in HLA-A2 at position 134 (T134K) abrogates association with TAP and calreticulin (12, 13, 14). For the purpose of examining 1) whether this position influences association of a different MHC class I molecule, H-2Ld, with the assembly complex, 2) whether residues flanking 134 also affect interaction, and 3) what the consequences of any loss of interaction would be, a series of Ld mutants with individual amino acid changes between position 128 and position 136 were made (see footnote 4 and Fig. 2). The choice of Ld permits the use of mAbs specific for precursor (open) and product (folded) class I assembly states. As shown in Fig. 3, these Ld mutants were all expressed at similar levels within the cell. When interactions of the open form of Ld and Ld mutants with TAP and calreticulin were assessed, it was found that (unlike wild-type (wt) Ld) none of these mutants associated with either TAP or calreticulin (Fig. 3). Thus H-2Ld, like HLA-A2, requires the wt residue at position 134 for interaction with TAP and calreticulin. This indicates that position 134 is important for these interactions in murine as well as human MHC class I molecules. Furthermore, the whole length of the 128–136 region around this position contributes to these interactions in Ld. These results demarcate a sizable area of the class I α2 domain, rather than a single MHC class I amino acid residue, as interacting with the assembly complex. The size of the region on the α2 domain that is involved suggests that a component of the assembly complex shares a large area of its surface with class I. Also, the structural nature of some of the class I amino acids involved suggests potential biochemical interactions. For example, because an E to R mutation at position 128 prevents association, it can be postulated that there may be binding of a basic residue in the assembly complex to this glutamic acid on class I.
Three regions of the MHC class I molecule are involved in chaperone interaction
Alterations in three widely separated areas (α1, α2, and α3 domains) of the MHC class I molecule have been shown to abrogate association with both TAP and calreticulin (Refs. 3, 14, 20 and this report). Furthermore, a mutation in the α3 domain has been suggested to abrogate the binding of tapasin (16). In Fig. 4, mutants with substitutions in each of those three areas are directly compared with wt Ld and to an additional mutant, LdD137H. The LdD137H mutant has an amino acid substitution just carboxy-terminal to the area (positions 128–136) that is shown in Fig. 3 to affect TAP and calreticulin association. As can be seen in Fig. 4, TAP and calreticulin interaction is maintained by LdD137H, defining the C-terminal boundary of the chaperone interaction site as position 136. Mutants LdW133T and LdT134K, like LdN86Q and LdD227K, were found to lack TAP, tapasin, and calreticulin association, although interaction with calnexin was not impaired (Figs. 4 and 5). Thus, mutations at sites in each of the three domains of Ld disrupt interaction with tapasin, as well as TAP and calreticulin. These results are consistent with a model in which calreticulin and tapasin exhibit cooperativity in their binding. Possibly the α2 and α3 sites (which lie in the same plane) both interact with tapasin, and the α1 domain oligosaccharide interacts with calreticulin.
Mutant LdK131D has stronger association with the assembly complex than does wt Ld
Surprisingly, a substitution at position 131, unlike the other individual site-directed mutations in this loop,4 was found to strengthen association with TAP, calreticulin, and tapasin. Fig. 5 compares the LdK131D mutant with wt Ld, E128R, W133T, and T134K. The E128R, W133T, and T134K mutants fail to bind to TAP, calreticulin, and tapasin. In contrast, the K131D mutant binds much more strongly to these proteins than even wt Ld. This result was verified with the use of a separate K131D transfectant that, despite having low intracellular expression, still exhibited very strong interaction with the assembly complex (data not shown). Therefore, introduction of an acidic amino acid residue at this position facilitates interaction with assembly complex components, suggesting a basic residue in the assembly complex may be juxtaposed to this class I position.
The phenotype of LdK131D may have some similarities to a reported HLA-A2 mutant, S132C (13). This A2 mutant presents endogenous peptides well, but not exogenous peptides. Furthermore, S132C has an extended period of association with TAP. Notably, in all murine MHC class I molecules except one, Lys is found at position 131 (29). The sole exception is Q1k, which has Glu at that position. Because our substitution of an Asp for the natural Lys in Ld greatly strengthened assembly complex association, it could be hypothesized that Q1k has unusually strong association with these ER proteins.
β2m association with the open form of MHC class I correlates with chaperone association
In their folded forms, all of the mutant Ld H-chains (listed in footnote 4) were found to be associated with β2m (data not shown). As demonstrated in Fig. 6, β2m can be detected by autoradiography coprecipitated with the folded form of wt Ld, E128R, K131D, and T134K. Very little W133T is present in the folded form (Fig. 6), and detection of β2m with the 30-5-7+ form of W133T required a long exposure of a Western blot probed with anti-mouse β2m serum (data not shown). The ability of the mutants other than W133T to adopt the folded form verifies that the mutations do not cause gross conformational changes. Also, association of β2m with the folded form of the Ld mutants indicates that the lack of interaction with the assembly complex is not secondary to the complete inability of these Ld mutants to bind β2m. β2m interacts weakly with the open form of mutants E128R, D129R, L130D, T132K, W133T, T134K, A135H, and A136H compared with wt Ld (Fig. 6 and data not shown). The observation that β2m is poorly associated with the open form of these α2 domain mutants suggests that stable assembly of H-chains with β2m before peptide binding requires association with the assembly complex.
The LdW133T mutant bears some phenotypic similarities to the ethylnitrosourea-induced histocompatibility loss mutant Ddm6 (30). This mutation was found by sequencing to be DdW133R (31). The phenotype of this mutant was low surface expression, inability to be recognized by a conformationally sensitive Ab specific for the α2 domain, and lack of β2m association with the form precipitated by an anti-α3 domain Ab. Furthermore, dm6 cells were not lysed by Dd-specific alloreactive T cells (31), which is consistent with the original detection of Ddm6 as a loss mutation (30). Similarly, the LdW133T mutant characterized here was found to have very low surface expression after transfection into L cells. Indeed, of all the mutants characterized in this study, LdW133T was the most impaired in its surface expression (data not shown). Furthermore, a significant pool of intracellular LdW133T molecules were detected in our L cell transfectants, but very few mutant LdW133T molecules were detected in a folded (30-5-7+) conformation (Fig. 6). In addition, surface LdW133T molecules were found not to be inducible with exogenous peptide (data not shown), unlike all other mutants in this study, which were specifically inducible with Ld peptide ligands. These cumulative findings clearly indicate that substitution of the Trp at position 133 ablates H-chain interaction with peptide. It should also be noted that the W133 amino acid residue is conserved among all MHC class I H-chains (including the class Ib proteins), and that its side chain points inward (i.e., it lacks solvent accessibility) in the Ld three-dimensional structure (Ref. 11 and Fig. 2). Based on the more dramatic phenotype of the 133 mutation compared with other mutations in this region and based on the inward projection of the W133 side chain, we propose that the W133 residue may be critically involved in the overall folding of the class I H-chain. If this is the case, the severe phenotype of the substitutions at 133 extends beyond a lack of interaction with the assembly complex.
Quite interestingly, there is strong positive correlation between β2m association and assembly complex association for the open (64-3-7+) forms of the mutants (Figs. 3, 5, and 6). The open forms of mutants LdE128R, LdW133T, and LdT134K (all of which do not interact with the assembly complex) have extremely little coprecipitated β2m. In contrast, the amount of β2m coprecipitated with LdK131D (which strongly binds the assembly complex) exceeds the amount coprecipitated with Ld. Thus, assembly complex binding and β2m binding to class I H-chain are coordinated.
Lack of interaction with the assembly complex does not completely prevent adoption of the folded class I conformation in the ER (Fig. 6). This is consistent with the previous observation that folded, although not heat-stable, class I H-chain/β2m heterodimers are present in lysates of RMA-S cells (25). These folded forms are presumably occupied by low-affinity peptides. Despite the presence of some folded forms in the ER, the lack of association between these α2 domain mutants and the assembly complex has perceptible functional consequences, because it causes the class I molecule to become loaded with suboptimal peptides that are unusually susceptible to replacement (see below).
Cell surface Ld α2 domain loop mutants are more peptide accessible than wt Ld
The degree of cell surface stabilization by exogenous peptide is inversely correlated with the efficiency of intracellular loading. We postulated that inability to associate with the assembly complex might lead to production of class I molecules that have suboptimal peptides or that are functionally empty. To test the relevance of our observation that positions other than 134 in the α2 domain prevent assembly complex binding, we have examined the ability of exogenous peptide to induce folding of some of the α2 domain Ld mutants. In our comparison, we have also included mutants with substitutions in the other two areas that affect assembly complex binding (i.e., position 86 and amino acid 227). As shown in Fig. 7, Ld mutants unable to bind assembly complex proteins are more peptide inducible than wt Ld. After culture with either of two peptides, LdT134K surface expression rose by >7-fold, compared with <5-fold for Ld (Fig. 7,A and data not shown). Similarly, surface expression of wt Ld was induced only ∼4-fold by murine CMV (MCMV) peptide, but surface expression of the LdE128R, LdN86Q, and LdD227K mutants was induced ∼5- to 6-fold (Fig. 7,B and data not shown). Therefore, Ld mutants unable to bind tapasin and other assembly complex components obtain relatively low-affinity peptides in the ER. This suggests that the absence of direct binding to the complex denies the MHC class I H-chain access to the optimal selection of peptides. In line with our observation that LdK131D has unusually strong binding to TAP/calreticulin/tapasin, this mutant exhibited low peptide inducibility (only 2.8-fold, Fig. 7 B). LdK131D likely binds peptides with unusually good affinity due to its strong association with the assembly complex.
For comparison, we also examined the peptide inducibility of the wt Ld expressed in 721.220 and 721.221. The 721.220 cell line does not express the tapasin protein (2). The 721.221 line is closely related to 721.220 (17, 18), but expresses a normal level of tapasin protein. As shown in Fig. 8, peptide induces Ld expression on 721.221 by 3.7-fold, but it induces Ld expression on 721.220 by 7-fold. Thus, complete absence of cellular tapasin also causes class I to be suboptimally loaded or functionally empty, and to a similar (although perhaps slightly greater) extent than prevention of tapasin/class I interaction by mutation of the class I H-chain. Ld expressed on a tapasin-negative cell could be slightly more peptide inducible than an Ld mutant unable to bind tapasin due to additional activities of tapasin that do not involve direct binding to class I. For example, tapasin has been shown to increase the stabilization of TAP (6), which likely improves the pool of available peptides. Notably, the influence of tapasin on Ld expression is relative and not absolute. For example, tapasin is not required for functional Ag presentation by Ld (data not shown). However, Ld expressed in the absence of tapasin is more peptide accessible at the cell surface; thus, tapasin must contribute to optimization of peptide binding in the ER. Our 721.220-Ld results suggest that the phenotype of the α2 domain loop mutations could possibly be due to a deficiency in tapasin interaction.
In summary, our results indicate that mutations in amino acid residues 128–136 of the class I H-chain affect association with the assembly complex of TAP/tapasin/calreticulin. In the three-dimensional structure of the folded Ld H-chain, residues 128–136 are on a loop connecting β sheets under the peptide-binding groove with an α-helix extending above the peptide-binding groove. Thus, the 128–136 loop could be a hinge region susceptible to conformational change when a peptide binds. If true, chaperones could monitor the conformational change in the 128–136 loop such that they would only bind to class I when the loop is in the open conformation. Tapasin may be the most likely candidate to monitor peptide binding because it is a specific chaperone for class I. Interestingly, others have speculated that tapasin interacts with the α3 domain (16). Indeed, as shown in Fig. 1, residues 128–136 in the α2 domain and residues 227–229 in the α3 domain are in the same plane along the side of the H-chain. Thus tapasin could interact with both of these determinants. In this model, the interaction between the membrane proximal Ig fold of tapasin (32) and the α3 domain of the H-chain could be similar to the interaction of the complementarity-determining regions of CD8 with H-chain residue 227 (33), and the C terminal portion of tapasin could interact with the α2 loop region of the H-chain. The other site on the H-chain shown to influence association with the assembly complex is N86, the glycosylation site in the α1 domain of all mouse and human class I molecules. Based on its lectin properties, calreticulin likely binds to the α1 domain oligosaccharide (2, 3). Consistent with this model, the location of N86 is spatially separated from the aforementioned, putative tapasin interaction sites (residues 128–136, and around 227) (Fig. 1). Thus tapasin and calreticulin could both simultaneously bind H-chain. The fact that mutation at each of these sites ablates TAP association could then be explained if calreticulin, like tapasin (2, 34), is required for class I to associate stably with TAP.
We thank Dr. David McKean for the antisera to calnexin and Drs. Daved Fremont and Daniel Quinn for helpful discussions and critical reading of the manuscript.
This work was supported by Grants AI-19876 (to T.H.H.) and GM-57428 (to J.C.S.) from the National Institutes of Health and by National Science Foundation Grant OSR-9452894 (to J.C.S.) and the South Dakota Future Fund (to J.C.S.).
Abbreviations used in this paper: ER, endoplasmic reticulum; H-chain, heavy chain; β2m, β2-microglobulin; MCMV, murine CMV; CHAPS, 3-([cholamidopropyl]dimethylammonio)-1-propanesulfonate; wt, wild type.
Ld mutant cell lines: L-LdN86Q, L-LdE128R, L-LdD129R, L-LdL130D, L-LdK131D, L-LdT132K, L-LdW133T, L-LdT134K, L-LdA135H, L-LdA136H, L-LdD137H, and L-LdD227K.