Binding of the Natural Killer Cell Inhibitory Receptor Ly49A to Its Major Histocompatibility Complex Class I Ligand

CRUCIAL CONTACTS INCLUDE BOTH H-2Dd AND β2-MICROGLOBULIN*

  1. Jian Wang‡,
  2. Mary C. Whitman‡,
  3. Kannan Natarajan‡,
  4. José Tormo§,
  5. Roy A. Mariuzza and
  6. David H. Margulies‡‖
  1. From the ‡Molecular Biology Section, Laboratory of Immunology, NIAID, National Institutes of Health, Bethesda, Maryland 20892-1892, §Campus de la Universidad Autonoma de Madrid, 28049 Madrid, Spain, and Center for Advanced Research in Biotechnology, University of Maryland Biotechnology Institute, Rockville, Maryland 20850
     
    Next Section

    Abstract

    Ly49A, an inhibitory C-type lectin-like mouse natural killer cell receptor, functions through interaction with the major histocompatibility complex class I molecule, H-2Dd. The x-ray crystal structure of the Ly49A·H-2Dd complex revealed that homodimeric Ly49A interacts at two distinct sites of H-2Dd: Site 1, spanning one side of the α1 and α2 helices, and Site 2, involving the α1, α2, α3, and β2m domains. Mutants of Ly49A, H-2Dd, and β2-microglobulin at intermolecular contacts and the Ly49A dimer interface were examined for binding affinity and kinetics. Although mutations at Site 1 had little affect, several at Site 2 and at the dimer interface hampered the Ly49A·H-2Dd interaction, with no effect on gross structure or T cell receptor interaction. The region surrounding the most critical residues (in H-2Dd, Asp122; in Ly49A, Asp229, Ser236, Thr238, Arg239, and Asp241; and in β2-microglobulin, Gln29 and Lys58) of the Ly49A·H-2Dd interface at Site 2 includes a network of water molecules, suggesting a molecular basis for allelic specificity in natural killer cell recognition.

    NK1 cells provide a crucial arm of the innate immune system as they are poised to respond by cytolysis and lymphokine production to stimulation by neoplastic or infected cells (1-3). NK cell activation is regulated by a balance of activating and inhibitory signals delivered through receptors of the immunoglobulin and C-type lectin-like superfamilies, that recognize MHC-I or MHC-I-like molecules expressed on target cells (4). In resting NK cells the inhibitory signals dominate, and perturbation of normal expression of MHC-I, by viral infection or oncogenic dysregulation, leads to attenuation of the inhibitory signal and augmentation of the activating one. The best understood inhibitory NK receptor of the C-type lectin family is murine Ly49A, a type II membrane homodimer in which the structure in complex with the MHC-I ligand, H-2Dd, has been determined crystallographically (5). Direct interaction of Ly49A with H-2Dd has been demonstrated in cell adhesion and protein binding assays (6-9). In addition, biotinylated Ly49A has been employed in staining the naturally expressed H-2Dd ligand on normal cells (10, 11). Labeled tetrameric MHC-I molecules including H-2Dd bind specifically to various Ly49-expressing cells (12, 13). Although Ly49A is a member of the C-type lectin superfamily, Ly49A-mediated recognition is Ca2+- and carbohydrate-independent and is MHC-I-restricted but not peptide-specific (9, 14-16).

    The crystal structure of the Ly49A·H-2Dd complex (5) indicated that the Ly49A homodimer interacts with two topographically distinct sites on H-2Dd (Fig. 1 a). At Site 1, a single Ly49A subunit contacts one end of the MHC-I peptide-binding groove, involving the amino terminus of the α1 helix and the carboxyl terminus of α2. Site 2 is a concavity beneath the peptide-binding platform, bounded by amino acid residues of the floor of the β sheet of the α1 and α2 domains, as well as by the α3 domain and the MHC-I light chain, β2m. This site overlaps the region where CD8αα interacts with MHC-I (5). In attempts to understand the details of the interaction of Ly49 receptors with their MHC-I ligands, several groups have analyzed binding and recognition using cells transfected with recombinant Ly49 molecules (17, 18), MHC-I mutants in transfected cells (16, 19-22), and binding of recombinant Ly49A to cells derived from different mouse inbred strains (10, 11). These studies suggest that amino acids in the “neck” region of Ly49A (17) or Ly49C (18), residues of both Site 1 of H-2Dd (10) and Site 2 (22), as well as of its peptide-binding grove (20, 21) may all contribute either directly or indirectly to the Ly49A/MHC-I interaction. In addition, two groups have reported the effects of human β2m on the Ly49A/MHC-I interaction (22, 23). Thus, although some evidence supports Site 1 as a primary region of interaction between MHC-I and Ly49A, other experiments are more consistent with Site 2 playing the predominant role, and some studies are inconclusive.

    Figure 1
    View larger version:
    Figure 1

    Location of contacts between Ly49A and H-2Dd. a, a ribbon diagram of the structure of the Ly49A·H-2Dd complex (Protein Data Bank code 1qo3) showing sites of interaction. These are referred to as Site 1, Site 2, and Ly49A homodimer interface(labeled only on one dimer). The subunit designations are:A, H-2Dd heavy chain; B, β2m; C, Ly49A subunit at Site 1; D, dimeric partner of C; E, symmetry-related Ly49A subunit C; F, major Ly49A subunit interacting at Site 2. The Cα atoms of residues of the homodimer interface (orchid), Site 2 alone (magenta), and Sites 1 and 2 (orange-red) are indicated as space-filled spheres. b, Ly49A amino acid sequence and location of residues selected for mutagenesis. The amino acid sequence of Ly49AC57BL/6 is shown with the transmembrane region (amino acids 45–66) and the lectin-like domain (amino acid 141–262) enclosed in boxes. Secondary structure elements are labeled, and the positions of site-directed mutants are indicated: interface contacts of the homodimer (filled circles, orchid), Site 1 and Site 2 (orange-red downward arrows), and Site 2 unique (magenta upward arrows).

    To resolve the issue as to whether Site 1 or Site 2 plays the major role, we have exploited a well defined biochemical system using only recombinant, highly purified molecules. Using a panel of site-directed mutants of Ly49A, H-2Dd, and β2m in binding assays, we have determined kinetic and equilibrium parameters quantitatively. These data, viewed in the context of the crystal structure of the Ly49A·H-2Dd complex, define not only the dominant site of interaction of Ly49A with its MHC-I ligand as Site 2, but also provide insight into the importance of homodimeric interaction in maintaining the integrity of Ly49A. Inspection of the region of intermolecular contact at Site 2 reveals important contributions from a network of water molecules coordinated by Ly49A, H-2Dd, and β2m residues and helps to explain the allelic specificity and peptide preference that particular members of the Ly49 family exhibit.

    Previous SectionNext Section

    EXPERIMENTAL PROCEDURES

    Mutagenesis of Ly49A, H-2Dd, and Mouse β2m

    Mutants were generated by site-directed mutagenesis of the codons of interest to encode alanine substitutions. Using Stratagene's QuickChange Mutagenesis Kit and following the manufacturer's instructions, mutations were introduced into: 1) a cDNA construct encoding the entire extracellular region (residues 67–262) of the C57BL/6 allele of Ly49A in a bacterial expression vector (pET21a; Novagen) (9); 2) a cDNA construct encoding the mature extracellular residues of H-2Dd in pET3a (24); or 3) a murine β2m-encoding construct based on the C57BL/6 allele of β2m in pET21d (25). Mutations were confirmed by automated DNA sequencing analysis using an ABI 377 DNA sequencer and accompanying sequencing analysis software (PerkinElmer Life Sciences).Escherichia coli strain BL21(DE3) was transformed with these mutant plasmid DNAs for protein expression.

    Protein Expression

    Bacteria harboring either the parental or mutant Ly49A-, H-2Dd-, or β2m-encoding plasmid vectors were cultured in LB broth containing 0.25 mg/ml carbenicillin and, following induction of exponentially growing cells for 2 h with isopropyl-1-thio-β-d-galactopyranoside, were lysed with 0.5 mg/ml of hen egg white lysozyme overnight at 4 °C and then with 0.1% deoxycholate for 30 min. The resulting inclusion bodies were washed first with 0.1% deoxycholate in TE buffer (0.1 mTris pH 8.0, 2 mm EDTA) and then extensively with TE, and they were then denatured in 6 m guanidine hydrochloride and reduced with 0.1 mm dithiothreitol. Ly49A was refolded following dilution into an arginine buffer (0.4 m l-arginine-HCl, 0.1 m Tris, pH 8, 2 mm EDTA, 3 mm reduced glutathione, and 0.3 mm oxidized glutathione). Following dialysis against 25 mm MES, pH 5.5, protein was further purified by ion exchange chromatography (Mono-S) followed by gel filtration on a Superdex 75 column. H-2Dd was prepared similarly, with refolding taking place in the presence of mouse β2m and either a motif (AGPARAAAL) peptide (26) or the HIV IIIB envelope gp160 peptide, P18-I10 (RGPGRAFVTI) (27, 28). Refolded proteins were further purified by size exclusion chromatography on Superdex-75 (Amersham Biosciences, Inc.).

    Monoclonal Antibodies and Recombinant Protein Ligands

    Monoclonal antibodies used in BIAcore binding assays were: A1, specific for Ly49AC57BL/6 (29, 30), YE1–32, and YE1–48 (derived in rat) (31, 32), and JR9-318 (derived from Mus spretus) (33), which binds all allelic forms of Ly49A. The anti-Ly49G2 mAb 4D11 is described by Mason et al. (34). Anti-Ly49C/I-reactive SW5E6 (35) was used as control for Ly49A binding. mAbs 34-2-12 (anti-H-2Dd α3 domain) and 34-5-8 (anti-H-2Dd α1/α2 domain) (36, 37) were from PharMingen (San Diego, CA). KP15, a mAb specific for H-2Dd bound to the HIV envelope glycoprotein-derived peptide Pro18-Ile10, has been described elsewhere (38). All antibodies were used as purified proteins. A single-chain TCR (scTCR) with specificity for H-2Dd/Pro18-Ile10 was expressed inE. coli and purified as described previously (39).

    Surface Plasmon Resonance Binding Assays

    Assays for binding of Ly49A, H-2Dd, and mAbs were based on kinetic progress curves obtained with the BIAcore 2000 using methods employed previously (9, 40). In general, one ligand (either a mAb, Ly49A, or scTCR) was immobilized on a CM5 biosensor chip using standard coupling procedures, and solution phase ligand (analyte) was offered at a flow rate of 10 μl/min. All binding experiments were performed at 25 °C. A mock-coupled surface that was activated and blocked but had no protein immobilized was always included as a control for nonspecific binding. When Ly49A mutants were evaluated, a Ly49A wild type-coupled surface was included on the same chip for direct comparison. Kinetic data were collected and analyzed using BIAevaluation 3.1 to calculate kinetic association and dissociation rate constants and equilibrium constants for dissociation. In all cases the potential for sequential deterioration of the integrity of the solid phase ligand was evaluated periodically with appropriate controls introduced throughout the run. Binding parameters obtained by kinetic analysis were compared with equilibrium parameters calculated in parallel and were always internally consistent.

    Previous SectionNext Section

    RESULTS

    Interface Mutants of the Ly49A Homodimer

    To evaluate the contribution of particular residues of Ly49A to recognition by H-2Dd, as well by specific monoclonal antibodies (mAbs), we made individual alanine substitutions at those positions that, in the crystal structure, appeared crucial to either the homodimeric interaction or the interaction between H-2Dd and Ly49A (Fig. 1). Six Ly49A interface residues, Tyr142, Trp143, Phe144, Tyr146, Leu188, and Val189 were mutated, and the bacterially expressed mutant proteins were tested for binding to bacterially expressed H-2Dd as well as to a panel of anti-Ly49 mAbs. Although several mutants showed decreased binding to some of the mAbs tested, all preserved the capacity to interact with several at levels at least 50% that of controls (Fig.2 a). Binding to 4D11, which primarily recognizes Ly49G2 and has weak activity on Ly49A, was lower for mutants at positions 142, 143, 146, and 189. Interestingly, 4D11 reactivity was greater for mutants at positions 144 and 188. Significantly, mutation at positions 142, 143, 146, and 189 each revealed decreased binding to either of the H-2Ddpreparations (Fig. 2 b). These results suggest that the integrity of the homodimer is important for binding to the MHC-I ligand and also influences binding by some mAbs.

    Figure 2
    View larger version:
    Figure 2

    Binding of mAbs and H-2Dd to Ly49A and its site-directed mutants. a, from 4200 to 6000 resonance units of Ly49A and the indicated mutant proteins were coupled to biosensor chips and tested for binding to different specific mAbs. b, Ly49A mutants were tested for binding to H-2Dd/motif and H-2Dd·Pro18-Ile10 complexes at a concentration of 6.5 μm. The Relative Binding, corrected for background binding to a mock-coupled surface and for level of coupling of the solid phase reagent, was compared with binding to a wild type Ly49A surface and was calculated as follows.FormulaResonance unit (RU) levels are at steady state. Relative binding (RB) of Ly49A wild type therefore = 1. Open bar, Ly49A wild type (WT); solid bar, control mutant S113A;back-slashed bar, interface mutants; gray bar, Site 2 mutants alone; forward slashed bar, Site 1 and Site 2 mutants. c, relative binding of Ly49A mutants to mAbs and to H-2Dd was calculated as described above.JR9-318, a mAb that binds several Ly49 molecules, was derived from M. spretus, and we might expect the Ly49 molecules of this strain to lack Lys224. We also tested SW5E6, which recognizes Ly49C and Ly49I but not Ly49A (54). In surface plasmon resonance analysis SW5E6 failed to bind Ly49A and also showed no binding to the full panel of Ly49A mutants (data not shown).

    Site 2 and Site 1, 2 Mutants of Ly49A

    Next, we examined mutants of Ly49A at H-2Dd contact sites. However, no contact residues of Ly49A are unique to Site 1, and therefore any mutation would either include both Sites 1 and 2 or be unique to Site 2. We first studied mutants at Site 2, the interface between Ly49A and H-2Dd that involves Ly49A and all three domains of H-2Dd as well as its light chain, β2m (Fig.1 a). For H-2Dd binding, point mutations generally either improved binding (N203A and K224A) or impaired binding significantly (S236A, T238A, and R239A). S236A and R239A showed an undetectable level of binding to H-2Dd, and T238A quantitatively reduced binding by about 9-fold (K Dincreasing from 1.82 to 15.9 μm largely as a result ofk d increasing from 0.063 to 0.363 s−1(Fig. 3 b)). The behavior of the triad of Ser236, Thr238, and Arg239 clearly indicates the importance of Site 2 interactions in H-2Dd binding. Binding to a panel of mAbs indicated that the overall integrity of the mutants was preserved and also allowed mapping of epitopes recognized by particular mAbs (Fig.2). Mutants of N203A, K224A, S236A, T238A, and R239A revealed differences in binding to some of the mAbs. In particular, N203A showed augmented binding to A1, YE1/32, and YE1/48, had no effect on binding to JR9-318, and showed absolutely no binding to 4D11. K224A reduced binding to A1, YE1/48, and JR9-318 and increased binding to YE1/32. Mutants S236A, T238A, and R239A modestly decreased binding to A1, and R239A increased binding to YE1/32, YE1/48, and 4D11.

    Figure 3
    View larger version:
    Figure 3

    H-2Dd binding to Ly49A mutants. a, H-2Dd·motif and H-2Dd·P18-I10 were injected at five different concentrations ranging from 0.5 to 8 μm in 2-fold increments over biosensor surfaces coupled with different Ly49A mutants. Each row represents a binding assay performed on different flow cells of the same CM5 chip. Because the binding patterns of H-2Dd·motif and H-2Dd·Pro18-Ile10 preparations were similar, but the level of H-2Dd·Pro18-Ile10 binding was lower, we show only H-2Dd·motif-binding curves.b, experimental sensorgram data were obtained from binding curves of experiments similar to panel a, and kinetic evaluation was performed using BIAevaluation 3.1. Analysis employed a 1:1 binding model that included a correction for drifting base line. The different groups represent five independent experiments using different biosensor chips. t Formula was calculated according to the relationship t Formula = 0.693/k d. ND, nondetectable;WT, wild type.

    We then analyzed seven mutants that represented contacts at both Sites 1 and 2: D229A, D241A, N242A, D246A, Q247A, V248A, and F249A (Figs.1-3). D229A, although it revealed little or no effect on binding of any of the mAbs, eliminated interaction with H-2Dd. D241A, which profoundly reduced binding to A1 but had little effect on any other mAb, also affected binding to H-2Dd, most significantly with an increase in the k d from 0.054 to 0.363 s−1 (Fig. 3). N242A modestly reduced binding to YE1/32 and 4D11 and also impeded interaction with H-2Dd. D246A, which adversely affects binding of YE1/32 and 4D11, also showed quantitative effects on binding to H-2Dd. Other mutations of residues that structurally contribute at both Sites 1 and 2, Q247A and F249A, did not impede mAb binding and augmented it in some cases. V248A impaired recognition by YE1/32. These three mutants showed quantitative effects on binding to H-2Dd/P18-I10 but less pronounced effects when assayed with H-2Dd·motif complexes. The most consistent of the effects observed was with F249A, which showed a slight increase in k a, a marked (more than 5-fold) increase in the k d, and a 5-fold increase in K D (Fig. 3 b).

    Surface Representation of Ly49A Site 2 and Site 1, 2 Mutant Effects

    To visualize the location of those Ly49A residues that influence H-2Dd and mAb interaction, we generated a surface representation of their effect on binding (Fig.4; Ly49A interface mutants are not seen in such a display). Mutations that adversely affected binding are colored in red, orange, yellow, and green, whereas those that improved binding, presumably by eliminating a negative effect, are depicted in light blue andblue. The residues that most affected A1 binding (Lys224, Asp241, and Arg239) form a contiguous patch on the surface of the molecule (Fig. 4 a). Residues Asp229, Ser236, and Thr238neighbor this region, and it is straightforward to imagine that the antibody-combining site of A1 could overlay this set of residues on one monomer of the Ly49A homodimer. A1 is an alloantibody raised in BALB/c mice against Ly49AC57BL/6 (29). Allelic differences between Ly49ABALB/c and Ly49AC57BL/6 in the region where the Ly49AC57BL/6 structure is known are at positions 187 and 238 (where Ly49AC57BL/6 has Gln and Thr, respectively, and Ly49ABALB/c has His and Ile). Residue 187 lies adjacent to residue 238 on the surface of Ly49A. In contrast to those residues in which mutation adversely affected the binding of A1, mutants N203A, Q247A, V248A, and F249A improved binding by A1.

    Figure 4
    View larger version:
    Figure 4

    Molecular surface representation of Ly49A mutations affecting binding by H-2Dd and mAb. Ly49A Site 2 and Site 1, 2 mutants with different binding effects were visualized in a surface representation of the structure of Ly49A. The binding of A1 (a), YE1/32 (d), YE1/48 (c), JR9-318 (f), and H-2Dd/motif (b and e) is based on the BIAcore binding analysis as shown in Fig. 2. This illustration was rendered with GRASP 1.3.6 (55). The extent of binding of each of the alanine mutants is color-coded according to the values given in the legend for Fig.2 c. The following color scale was used, with redrepresenting the lowest level of binding and blue the highest: red, relative binding ≤ 0.05;orange, 0.05 < relative binding ≤ 0.1;yellow, 0.1 < relative binding ≤ 0.5;green, 0.5 < relative binding ≤ 1.0; light blue, 1.0 < relative binding ≤ 1.5; blue, 1.5 < relative binding ≤ 2.0.

    The binding of three other antibodies, YE1/48, YE1/32, and JR9-318, to the panel of mutants reveals different characteristics. The only mutation that adversely affects binding of YE1/48 is K224A (Fig.4 c); YE1/32 is most affected by D246A (Fig. 4 d); and JR9-318 is most dependent on K224 (Fig. 4 f). These results suggest that YE1/32 interacts more with residues on the “side” of the Ly49A molecule than with those on the “top.”

    The surface region that most affects H-2Dd binding is a large area that extends from Asp246 across the top of the molecule that includes Gln247, Val248, and Phe249 as well as a distinct surface consisting of Asp229, Ser236, Thr238, Arg239, Asp241, and Asn242 (Fig. 4,b and e). In addition to the adverse effects of mutation of any of the residues in this patch, mutation of Asn203 and Lys224 to alanine improves the binding to H-2Dd molecules. Thus, the focus of H-2Dd binding is on Arg239 and the surrounding residues located along the top of Ly49A.

    The Importance of Site 2 Mutants of H-2Dd and β2m

    Although the behavior of Ly49A mutants S236A, T238A, and R239A strongly suggested that Site 2 was the major focus of the Ly49A/H-2Dd interaction, other data, including the analysis of polymorphic MHC-I residues in different mouse strains and the behavior of a single H-2Dd mutation of I52M, had suggested that Site 1 was important (10). To clarify this issue, we tabulated the number of atomic contacts between those residues that seemed to be most involved in binding to H-2Dd (TableI). We looked closely at the contacts made by Ly49A residues Asp229, Ser236, Thr238, Arg239, Asp241, Asn242, Gln247, Val248, and Phe249 at either Site 1 or Site 2. For Site 1 contacts, Gln247, Val248, and Phe249 made 18 atomic contacts with residue Gln54 of H-2Dd, and residue Asp229 made 10 contacts with H-2Ddresidue Arg169. For Site 2 contacts, the greatest number (23) were made by Ly49A residues Ser236, Thr238, and Arg239 to Asp122 of H-2Dd. These contacts are specifically between the “F” subunit of Ly49A and H-2Dd. Ly49A residues Thr238, Arg239, and Asp241 are ambiguous, however, in that those residues in the “E” subunit (Fig.1 a and Table I) also make contact with H-2Ddresidues Tyr85 (8) and Asn86 (10). Residues Gln247, Val248, and Phe249 of Ly49A made 21 contacts with β2m residue Gln29; and residues Asp229, Arg239, Asp241, and Asn242 of Ly49A made 20 contacts with β2m residue Lys58.

    View this table:
    Table I

    Atomic contacts between Ly49A and H-2Dd residues

    To test the contribution to binding of these potential contact regions, we produced alanine mutations at all seven positions: two at Site 1 of the H-2Dd heavy chain (Q54A and R169A); three at Site 2 (Y85A, N86A, and D122A); and two at the β2m contact (Q29A and K58A). In addition, a non-contact mutant in H-2Dd, S246A, was generated as a control. Each of these eight mutants in parallel with a parental H-2Dd·motif peptide complex was tested by surface plasmon resonance for binding to Ly49A. The sensorgrams and the steady state levels of binding are plotted as a function of analyte concentration in Fig.5. The mutational analysis of H-2Dd and β2m clearly implicates Site 2 as the major determinant of the Ly49A·H-2Dd binding interaction: H-2Dd mutant D122A and β2m mutants Q29A and K58A had profound effects on binding to Ly49A. (In addition, these two β2m mutants exerted no effect on the binding of a β2m-dependent mAb, S19.8, over a wide range of concentration. S19.8 also showed concentration-dependent inhibition of binding of H-2Dd to Ly49A even when the H-2Dd was complexed with either of the two β2m mutants (data not shown).) In contrast to the effects of Site 2 mutants of H-2Dd, those at Site 1, Q54A and R169A, showed little effect on binding. Two mutants at the periphery of Site 2, Y85A and N86A, and the S246A control showed only a small quantitative effect.

    Figure 5
    View larger version:
    Figure 5

    H-2Dd and β2m mutant binding to Ly49A. The interactions between mutants of H-2Dd, β2m, and wild type Ly49A were measured in the BIAcore binding assay. Ly49A and mAbs specific for H-2Dd were immobilized on a CM5 chip as described under “Experimental Procedures.” Wild type (WT) H-2Dd or its mutants refolded with motif peptide were measured at concentrations from 0.156 to 10 μm in 2-fold increments. Resonance unit (RU) values are corrected for background binding. a andb, data from two different experiments. c andd, dose-dependent response comparison of binding of different H-2Dd mutants and β2m mutants to Ly49A. All preparations were tested in parallel for binding to anti-H-2Dd mAb to confirm the concentration as determined by UV absorbance and the preservation of serological activity (data not shown).

    Although H-2Dd mutant D122A and β2m mutants Q29A and K58A severely reduced binding to Ly49A, they retained the ability to bind to H-2Dd-specific mAbs 34-35-8 and 34-2-12 (data not shown). To eliminate the possibility that these mutations may have distorted either the peptide binding site or the TCR binding site of the molecule, we tested their ability to interact with both an H-2Dd/Pro18-Ile10-restricted specific recombinant scTCR and with an MHC-restricted, peptide-specific mAb, KP15 (38). As shown in Fig. 6, the three mutants and the parental H-2Dd/Pro18-Ile10 bind both the scTCR and KP15 equivalently. These results reinforce the view that the Ly49A binding site on H-2Dd is distinct from that of the TCR.

    Figure 6
    View larger version:
    Figure 6

    Binding of different H-2Dd·peptide/β2m complexes to scTCR. H-2Dd mutant D122A and β2m mutants Q29A and K58A, which were assembled with parental H-2Dd and with peptide Pro18-Ile10, were analyzed in a dose-dependent manner for binding to: a, scTCR, 0.6–10 μm; b, mAb KP15, 0.015–0.5 μm; c, Ly49A, 0.6–20 μm (these were also compared with H-2Dd/motif binding in all cases). All preparations were also tested for binding to mAbs 34-2-12 and 34-5-8 in a similar fashion (data not shown).

    To explain the profound effects exerted by mutation of H-2Dd D122A, as well as β2m Q29A and K58A, we returned to a close inspection of the interface of H-2Ddwith Ly49A at Site 2 (Fig. 7). In particular, we have included the water molecules at this interface. The region that surrounds residue Asp122 of H-2Ddreveals not only the proximity of the side chains of residues Ser236, Thr238, and Arg239 of Ly49A but also at least seven water molecules that are involved in this interface (Fig. 7 b). The β2m residue Lys58, which is highly conserved among species, interacts with Asp229, Asp241, and Asn242 of Ly49A amid a network of water molecules that also coordinate with the side chain of H-2Dd residue Arg6 as well as Arg239 (Fig. 7 c). Continuing this general trend, β2m residue Gln29 (the Gln is unique to mouse β2m) not only interacts with its contact residues Gln247, Val248, and Phe249 on Ly49A but also interacts with a number of waters in this vicinity (Fig.7 d).

    Figure 7
    View larger version:
    Figure 7

    H-2Dd residue Asp122and β2m residues Gln29and Lys58 interact with Ly49A at Site 2. Interactions of residues in H-2Dd·Ly49A complex (Protein Data Bank code 1qo3) were illustrated with MOLSCRIPT (56) and rendered with Raster3d (42). a, interaction between Ly49A (F domain), H-2Dd, and β2m mutants at Site 2.b, side-by-side stereo view of network of interactions of H-2Dd residue Asp122 and Ly49A residues Ser236, Lys237, Thr238, and Arg239 in the presence of water molecules. Asp122 of H-2Dd contacts directly with Ser236, Thr238, and Arg239 of Ly49A through hydrogen bonds. c, side-by-side stereo view of network of β2m residue Lys58 with Ly49A residues Asp229, Asp241, and Asn242in the presence of water molecules. d, side-by-side stereo view of network of β2m residue Gln29 with Ly49A residues Gln247, Val248, and Phe249 in the presence of water molecules. Dashed lines indicate hydrogen bonds. Other atomic contacts are not explicitly shown. Yellow, H-2Dd;blue, Ly49A; lavender, β2m;orange spheres, water molecules.

    Previous SectionNext Section

    DISCUSSION

    Our mutational analysis of the binding of Ly49A, H-2Dd, and β2m mutants leads to two major conclusions: that the primary binding site of H-2Ddemployed by Ly49A is at Site 2, a large surface area that involves contacts from the floor of the peptide-binding groove of the MHC-I and residues from β2m; and that integrity of the Ly49A homodimer is critical for preservation of MHC-I binding. The effect of mutation of the homodimeric interface residues Tyr142, Trp143, Tyr146, and Val189 in reducing binding to H-2Dd as well as the augmentation of binding by mutations of Phe144 and Leu188 suggest that the structure of the surface that contacts H-2Dd is influenced by the nature of the homodimerization. These results are of particular importance when we consider the three-dimensional structure and mode of dimerization of NK receptors and related molecules. As found previously (41), the dimer interfaces of Ly49A, NKG2D, and CD69 are highly conserved, although in Ly49A it is somewhat asymmetric, whereas NKG2D and CD69 show clear symmetry. The recently determined structure of Ly49I indicates that even within the Ly49 family there is considerable flexibility in the dimerization, because in this molecule the α2 helix is not involved in the dimer interface.2 These NK receptor domains dimerize differently from other more distantly related members of the C-type lectin family such as tunicate lectin and coagulation factor IX binding protein. In an evaluation of the binding of Ly49CC57BL/6 and Ly49IC57BL/6 to various MHC-I molecules in a cell agglutination assay, Lian et al. (18) observed that mutation of Tyr146 (the position equivalent to Tyr142 of Ly49A) of Ly49C to the histidine found in the closely related Ly49I resulted in reduced binding to H-2sand H-2b cell lines as well to an H-2Ddtransfectant. These results suggest that variability in the mode of dimerization influences allelic specificity of Ly49 family members.

    The importance of Site 2 is indicated by the effect of mutations in Ly49A unique to Site 2 (in particular residues Ser236, Thr238, and Arg239) as well as by mutation of H-2Dd (Asp122) (Fig. 7 b) and β2m (Gln29 and Lys58 (Fig. 7,d and c)). Ly49A is unique among NK receptors of either the Ig-like or C-type lectin-like families in its utilization of the β2m subunit as part of its binding site. A mutational analysis of H-2Dd, based on transfection and agglutination and functional assays, indicated an important role for residues Arg6, Asp122, and Lys243of H-2Dd (22), and the importance of species differences in β2m has been reported (22, 23). Our results add unequivocal data indicating the critical importance of the two major contact residues of β2m. This analysis also implicates β2m as a spatial transducer of subtle structural changes that occur either in the peptide-binding groove or in other parts of the α1α2 unit of the MHC heavy chain. The disposition of β2m with respect to the heavy chain varies markedly in different crystal structures (24, 43, 44), and the stability of the MHC/β2m interaction is affected by amino acid substitutions in the peptide-binding groove (45). The exquisite sensitivity of Ly49A binding to substitution of either of the two major β2m contact residues, Gln29 or Lys58, explains the observed effects of substitution of bovine or human β2m in NK recognition (22, 23), because position Gln29 is unique to murine β2m, whereas human and bovine variants have glycine. The effect of β2m dislocation at the Site 2 interface explains the behavior of two mutations of H-2Dd, D29N and R35A, that affect Ly49A binding without direct involvement of the Site 2 interface (21). Mutant D29N would be expected to have an indirect effect on β2m binding because it interacts primarily with H3 of H-2Dd and only indirectly with β2m residue Arg228. Residue Arg35 is involved in hydrogen bonds to β2m M54 backbone oxygen as well as to H-2Dd E32. Thus, mutation of Arg35 to Ala would be expected to change the affinity of the MHC heavy chain for β2m and distort the disposition of this subunit, thereby altering the interaction with Ly49A. Several other substitutions of H-2Dd that do not make direct contact with either Ly49A or β2m have been studied, and their behaviors may be explained by similar distant effects. In particular, the double mutant S73W,D156Y, comprising residues that line the peptide-binding groove, has been shown to have functional effects on NK recognition (20), and the double mutants D77S,A99F and N30D,A99F as well as the triple mutant N30D,D77S,A99F showed reduced binding in an Ly49A tetramer staining assay (21). Residues Ser73, Asp77, Ala99, and Asp156 play a role in peptide interaction and thus are likely to contribute to the stability of the interaction with β2m, thereby affecting the interaction with Ly49A. The nature of the N30D substitution, which seems synergistic with substitution of peptide binding cleft residues, probably is due to the role that N30D plays in the interaction with Ly49A via distant interaction with residue R228 of β2m. In addition to these experiments with mutant H-2Dd and β2m, our earlier studies examining the behavior of a single-chain β2m/H-2Ddmolecule in NK cell recognition are also consistent with Ly49A interacting primarily with a site distinct from that of the TCR and subject to inhibition by the peptide linker extending from the carboxyl terminus of β2m to the amino terminus of H-2Dd (10, 46). Because Ly49G2, a molecule related to Ly49A in both sequence and in its ability to interact with H-2Dd, also is incapable of interacting with the single-chain β2m/H-2Dd molecule (46), it is likely that Ly49G2 binds H-2Dd at the same site as Ly49A. (Comparison of amino acid sequences of Ly49AC57BL/6 and Ly49G2C57BL/6 reveals that the residues that contact Asp122 of H-2Dd and Lys58 of β2m in the Ly49A complex are conserved, whereas those that interact with the conserved Gln29 of β2m are polymorphic.)

    The analysis of the regions of importance in Site 2 recognition clearly implicates the organization of water molecules in this large and imperfect interface in mediating the interaction between Ly49A and H-2Dd (Fig. 7). We speculate that the binding specificity of the Ly49 molecules for their MHC/peptide ligands is influenced by a water network exemplified in the Ly49A·H-2Ddstructure by residues Asp122 of H-2Ddwith Ly49A·residues Ser236, Lys237, Thr238, and Arg239; Lys58 of β2m along with Arg6 of H-2Dd and Asp229, Arg239, Asp241, and Asn242 of Ly49A; and Gln29 of β2m with Asp59 of β2m, and Gln247, Val248, and Phe249 of Ly49A. Since the description of an important role for water in the carbohydrate specificity of the l-arabinose-binding protein (47), there has been considerable interest in the contribution of water to ligand specificity (48). Recently water has been shown to be important in the specific binding of peptides to MHC molecules (49-53).

    The accumulation of the mutational evidence clearly reveals effects of H-2Dd residue Asp122 and β2m residues Gln29 and Lys58 on Ly49A binding. These effects seem to be local rather than global, because the TCR binding site of the H-2Dd·Pro18-Ile10 complex was preserved both for direct interaction with a cognate TCR and for binding by an MHC-restricted, peptide-specific mAb.

    In summary, our mutational analysis of Ly49A has permitted the localization not only of the sites bound by each of a panel of mAbs but has also demonstrated conclusively the importance of the homodimer interface and has localized the critical site of interaction between Ly49A, H-2Dd, and β2m to Site 2, a region that potentially overlaps with the CD8 binding site. The strength of this interaction and the lack of more than a 3-fold effect of any of the mutants at Site 1 tested to date suggest that the Site 2 interaction functions in the recognition between the NK cell and the antigen-presenting cell, a trans mode, although it does not eliminate the possibility of cis interactions between molecules expressed on the NK cell. Further understanding of the quantitative contributions of the redistribution of water as binding takes place will require careful thermodynamic measurements under conditions that vary the activity of water as solvent.

    Previous SectionNext Section

    ACKNOWLEDGEMENTS

    We thank W. M. Yokoyama (Washington University, St. Louis, MO) for providing hybridoma cells and encouragement. We thank J. Dorfman and M. Lenardo for their comments on the manuscript.

    Previous SectionNext Section

    Footnotes

    • ↵* This work was supported in part by National Institutes of Health Grants AI47900 and GM52801 (to R. A. M.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

      This paper is dedicated to the memory of José Tormo, colleague and friend.

    • ↵‖ To whom correspondence should be addressed: Molecular Biology Section, LI/NIAID, Bldg. 10, Rm. 11N311, National Institutes of Health, Bethesda, MD 20892-1892. Tel.: 301-496-6429; Fax: 301-496-0222; E-mail: dhm@nih.gov.

    • Published, JBC Papers in Press, November 5, 2001, DOI 10.1074/jbc.M110316200

    • ↵2 N. Dimasi, M. W. Sawicki, L. N. Reineck, Y. Li, K. Natarajan, D. H. Margulies, and R. A. Mariuzza, manuscript in preparation.

    • Abbreviations:
      NK

      natural killer

      β2m

      β2-microglobulin

      mAb

      monoclonal antibody

      MHC

      major histocompatibility complex

      TCR

      T cell receptor

      scTCR

      single-chain T cell receptor

      HIV

      human immunodeficiency virus

      MES

      2-(N-morpholino)ethanesulfonic acid

      • Received October 26, 2001.
    Previous Section
     

    REFERENCES

    1. ↵
      1. Biron C. A.,
      2. Nguyen K. B.,
      3. Pien G. C.,
      4. Cousens L. P.,
      5. Salazar-Mather T. P.
      (1999) Annu. Rev. Immunol. 17:189–220.
    2. ↵
      1. Lanier L. L.
      (1998) Annu. Rev. Immunol. 16:359–393.
    3. ↵
      1. Paul W. E.
      1. Yokoyama W. M.
      (1999) in Fundamental Immunology, ed Paul W. E. (Lippincott-Raven Publishers, Philadelphia), 4th Ed. pp 575–603.
    4. ↵
      1. Kärre K.
      (1997) Immunol. Rev. 155:5–9.
    5. ↵
      1. Tormo J.,
      2. Natarajan K.,
      3. Margulies D. H.,
      4. Mariuzza R. A.
      (1999) Nature 402:623–631.
    6. ↵
      1. Daniels B. F.,
      2. Karlhofer F. M.,
      3. Seaman W. E.,
      4. Yokoyama W. M.
      (1994) J. Exp. Med. 180:687–692.
    7. ↵
      1. Brennan J.,
      2. Mager D.,
      3. Jefferies W.,
      4. Takei F.
      (1994) J. Exp. Med. 180:2287–2295.
    8. ↵
      1. Kane K. P.
      (1994) J. Exp. Med. 179:1011–1015.
    9. ↵
      1. Natarajan K.,
      2. Boyd L. F.,
      3. Schuck P.,
      4. Yokoyama W. M.,
      5. Eilat D.,
      6. Margulies D. H.
      (1999) Immunity 11:591–601.
    10. ↵
      1. Chung D. H.,
      2. Natarajan K.,
      3. Boyd L. F.,
      4. Tormo J.,
      5. Mariuzza R. A.,
      6. Yokoyama W. M.,
      7. Margulies D. H.
      (2000) J. Immunol. 165:6922–6932.
    11. ↵
      1. Matsumoto N.,
      2. Tajima K.,
      3. Mitsuki M.,
      4. Yamamoto K.
      (2001) Int. Immunol. 13:615–623.
    12. ↵
      1. Mehta I. K.,
      2. Smith H. R.,
      3. Wang J.,
      4. Margulies D. H.,
      5. Yokoyama W. M.
      (2001) Cell. Immunol. 209:29–41.
    13. ↵
      1. Hanke T.,
      2. Takizawa H.,
      3. McMahon C. W.,
      4. Busch D. H.,
      5. Pamer E. G.,
      6. Miller J. D.,
      7. Altman J. D.,
      8. Liu Y.,
      9. Cado D.,
      10. Lemonnier F. A.,
      11. Bjorkman P. J.,
      12. Raulet D. H.
      (1999) Immunity 11:67–77.
    14. ↵
      1. Orihuela M.,
      2. Margulies D. H.,
      3. Yokoyama W. M.
      (1996) Proc. Natl. Acad. Sci. U. S. A. 93:11792–11797.
    15. ↵
      1. Correa I.,
      2. Raulet D. H.
      (1995) Immunity 2:61–71.
    16. ↵
      1. Matsumoto N.,
      2. Ribaudo R. K.,
      3. Abastado J. P.,
      4. Margulies D. H.,
      5. Yokoyama W. M.
      (1998) Immunity 8:245–254.
    17. ↵
      1. Brennan J.,
      2. Mahon G.,
      3. Mager D. L.,
      4. Jefferies W. A.,
      5. Takei F.
      (1996) J. Exp. Med. 183:1553–1559.
    18. ↵
      1. Lian R. H., Li, Y.,
      2. Kubota S.,
      3. Mager D. L.,
      4. Takei F.
      (1999) J. Immunol. 162:7271–7276.
    19. ↵
      1. Sundbäck J.,
      2. Nakamura M. C.,
      3. Waldenstrom M.,
      4. Niemi E. C.,
      5. Seaman W. E.,
      6. Ryan J. C.,
      7. Kärre K.
      (1998) J. Immunol. 160:5971–5978.
    20. ↵
      1. Waldenstrom M.,
      2. Sundbäck J.,
      3. Olsson-Alheim M. Y.,
      4. Achour A.,
      5. Kärre K.
      (1998) Eur. J. Immunol. 28:2872–2881.
    21. ↵
      1. Matsumoto N.,
      2. Yokoyama W. M.,
      3. Kojima S.,
      4. Yamamoto K.
      (2001) J. Immunol. 166:4422–4428.
    22. ↵
      1. Matsumoto N.,
      2. Mitsuki M.,
      3. Tajima K.,
      4. Yokoyama W. M.,
      5. Yamamoto K.
      (2001) J. Exp. Med. 193:147–158.
    23. ↵
      1. Michaelsson J.,
      2. Achour A.,
      3. Rolle A.,
      4. Kärre K.
      (2001) J. Immunol. 166:7327–7334.
    24. ↵
      1. Li H.,
      2. Natarajan K.,
      3. Malchiodi E. L.,
      4. Margulies D. H.,
      5. Mariuzza R. A.
      (1998) J. Mol. Biol. 283:179–191.
    25. ↵
      1. Shields M. J.,
      2. Moffat L. E.,
      3. Ribaudo R. K.
      (1998) Mol. Immunol. 35:919–928.
    26. ↵
      1. Corr M.,
      2. Boyd L. F.,
      3. Padlan E. A.,
      4. Margulies D. H.
      (1993) J. Exp. Med. 178:1877–1892.
    27. ↵
      1. Kozlowski S.,
      2. Corr M.,
      3. Takeshita T.,
      4. Boyd L. F.,
      5. Pendleton C. D.,
      6. Germain R. N.,
      7. Berzofsky J. A.,
      8. Margulies D. H.
      (1992) J. Exp. Med. 175:1417–1422.
    28. ↵
      1. Takeshita T.,
      2. Takahashi H.,
      3. Kozlowski S.,
      4. Ahlers J. D.,
      5. Pendleton C. D.,
      6. Moore R. L.,
      7. Nakagawa Y.,
      8. Yokomuro K.,
      9. Fox B. S.,
      10. Margulies D. H.,
      11. et al.
      (1995) J. Immunol. 154:1973–1986.
    29. ↵
      1. Nagasawa R.,
      2. Gross J.,
      3. Kanagawa O.,
      4. Townsend K.,
      5. Lanier L. L.,
      6. Chiller J.,
      7. Allison J. P.
      (1987) J. Immunol. 138:815–824.
    30. ↵
      1. Yokoyama W. M.,
      2. Jacobs L. B.,
      3. Kanagawa O.,
      4. Shevach E. M.,
      5. Cohen D. I.
      (1989) J. Immunol. 143:1379–1386.
    31. ↵
      1. Takei F.
      (1983) J. Immunol. 130:2794–2797.
    32. ↵
      1. Takei F.,
      2. Brennan J.,
      3. Mager D. L.
      (1997) Immunol. Rev. 155:67–77.
    33. ↵
      1. Roland J.,
      2. Cazenave P. A.
      (1992) Int. Immunol. 4:699–706.
    34. ↵
      1. Mason L. H.,
      2. Ortaldo J. R.,
      3. Young H. A.,
      4. Kumar V.,
      5. Bennett M.,
      6. Anderson S. K.
      (1995) J. Exp. Med. 182:293–303.
    35. ↵
      1. Sentman C. L.,
      2. Hackett J., Jr.,
      3. Kumar V.,
      4. Bennett M.
      (1989) J. Exp. Med. 170:191–202.
    36. ↵
      1. Ozato K.,
      2. Mayer N.,
      3. Sachs D.
      (1982) Transplantation 34:113–120.
    37. ↵
      1. Margulies D. H.,
      2. Evans G. A.,
      3. Ozato K.,
      4. Camerini-Otero R. D.,
      5. Tanaka K.,
      6. Appella E.,
      7. Seidman J. G.
      (1983) J. Immunol. 130:463–470.
    38. ↵
      1. Polakova K.,
      2. Plaksin D.,
      3. Chung D. H.,
      4. Belyakov I. M.,
      5. Berzofsky J. A.,
      6. Margulies D. H.
      (2000) J. Immunol. 165:5703–5712.
    39. ↵
      1. Plaksin D.,
      2. Polakova K.,
      3. McPhie P.,
      4. Margulies D. H.
      (1997) J. Immunol. 158:2218–2227.
    40. ↵
      1. Corr M.,
      2. Slanetz A. E.,
      3. Boyd L. F.,
      4. Jelonek M. T.,
      5. Khilko S.,
      6. al-Ramadi B. K.,
      7. Kim Y. S.,
      8. Maher S. E.,
      9. Bothwell A. L.,
      10. Margulies D. H.
      (1994) Science 265:946–949.
    41. ↵
      1. Sawicki M. W.,
      2. Dimasi N.,
      3. Natarajan K.,
      4. Wang J.,
      5. Margulies D. H.,
      6. Mariuzza R. A.
      (2001) Immunol. Rev. 181:52–65.
    42. ↵
      1. Merritt E. A.,
      2. Bacon D. J.
      (1997) Methods Enzymol. 277:505–524.
    43. ↵
      1. Fremont D. H.,
      2. Matsumura M.,
      3. Stura E. A.,
      4. Peterson P. A.,
      5. Wilson I. A.
      (1992) Science 257:919–927.
    44. ↵
      1. Madden D. R.
      (1995) Annu. Rev. Immunol. 13:587–622.
    45. ↵
      1. Ribaudo R. K.,
      2. Margulies D. H.
      (1995) J. Immunol. 155:3481–3493.
    46. ↵
      1. Chung D. H.,
      2. Dorfman J.,
      3. Plaksin D.,
      4. Natarajan K.,
      5. Belyakov I. M.,
      6. Hunziker R.,
      7. Berzofsky J. A.,
      8. Yokoyama W. M.,
      9. Mage M. G.,
      10. Margulies D. H.
      (1999) J. Immunol. 163:3699–3708.
    47. ↵
      1. Quiocho F. A.,
      2. Wilson D. K.,
      3. Vyas N. K.
      (1989) Nature 340:404–407.
    48. ↵
      1. Ladbury J. E.
      (1996) Chem. Biol. 3:973–980.
    49. ↵
      1. Ceman S., Wu, S.,
      2. Jardetzky T. S.,
      3. Sant A. J.
      (1998) J. Exp. Med. 188:2139–2149.
    50. ↵
      1. Gurlo T.,
      2. Meng W. S.,
      3. Bui H. H.,
      4. Haworth I. S.,
      5. von Grafenstein H.
      (1999) Immunol. Lett. 70:139–141.
    51. ↵
      1. McFarland B. J.,
      2. Beeson C.,
      3. Sant A. J.
      (1999) J. Immunol. 163:3567–3571.
    52. ↵
      1. Wilson N.,
      2. Fremont D.,
      3. Marrack P.,
      4. Kappler J.
      (2001) Immunity 14:513–522.
    53. ↵
      1. Rudolph M. G.,
      2. Speir J. A.,
      3. Brunmark A.,
      4. Mattsson N.,
      5. Jackson M. R.,
      6. Peterson P. A.,
      7. Teyton L.,
      8. Wilson I. A.
      (2001) Immunity 14:231–242.
    54. ↵
      1. Stoneman E. R.,
      2. Bennett M., An, J.,
      3. Chesnut K. A.,
      4. Wakeland E. K.,
      5. Scheerer J. B.,
      6. Siciliano M. J.,
      7. Kumar V.,
      8. Mathew P. A.
      (1995) J. Exp. Med. 182:305–313.
    55. ↵
      1. Nicholls A.,
      2. Sharp K. A.,
      3. Honig B.
      (1991) Proteins 11:281–296.
    56. ↵
      1. Kraulis P. J.
      (1991) J. Appl. Crystallogr. 24:946–950.
    « Previous | Next Article »Table of Contents

    This Article

    1. First Published on November 5, 2001, doi: 10.1074/jbc.M110316200 January 11, 2002 The Journal of Biological Chemistry, 277, 1433-1442.
    1. AbstractFree
    2. » Full TextFree
    3. Full Text (PDF)Free

    Navigate This Article

    This Week's Issue

    1. November 6, 2009, 284 (45)
    1. Current Issue
    1. Alert me to new issues of JBC
    • Advertisement
    • Advertisement
    Advertisement