0014-2980/00/0505-1480$17.50+.50/0 © WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000

Definition of polymorphic residues on killerIg-like receptor proteins which contribute to theHLA-C binding site

Jennifer Richardson1, Hugh T. Reyburn2, Isabel Luque1, Mar Vales-Gomez1 and JackL. Strominger1

1

1 Department of Molecular and Cellular Biology, Harvard University, Cambridge, USA2 Department of Pathology, Cambridge University, Cambridge, GB

Killer cell immunoglobulin-like receptors (KIR) bind HLA class I proteins in an allele- andlocus-specific manner. This report describes the use of transfectants expressing recombi-nant chimeric proteins, comprising the extracellular portions of KIR molecules and the trans-membrane and cytoplasmic tails of CD3- ´ , to create an in vitro system in which signaling isreadily measured and that preserves the specificity of the KIR/HLA-C interaction. The iden-tity of the amino acid residues on the KIR molecule important for binding to the HLA proteinis not well understood; although some KIR2D residues involved in HLA-C recognition havebeen identified, their relative importance and whether other amino acids contribute to bind-ing was unclear. This novel system was used to study, by site-directed mutagenesis, the roleof various amino acids in KIR binding to HLA-C ligand. The data presented here show thatwhile multiple polymorphic residues contribute to the HLA-C binding site on KIR proteins,two clusters of polymorphic residues define the group allotype specificity of HLA-C bindingto a KIR2D molecule.

Key words: KIR2D receptor / HLA-C / NK cell / CD3 zeta-chain fusion protein

Received 17/11/99Revised 2/2/00Accepted 9/2/00

[I 20213]

Abbreviation: KIR: Killer cell immunoglobulin-like recep-tors

1 Introduction

Target cell expression of HLA molecules results in pro-tection from lysis by NK cells. This recognition of HLAis mediated by a repertoire of inhibitory receptorsexpressed on NK cells which bind HLA protein in anallele- and locus-specific manner. One of the majorinhibitory pathways regulating the behavior of the NK cellbegins with the binding of a self-HLA-C molecule on thetarget cell by members of the p58 killer cells Ig-likereceptors (KIR) on the NK effector [1, 2]. NK cells dis-criminate between two groups of HLA-C allotypes thatdiffer at amino acids 77 and 80 in the § 1 helix of the HLA-C heavy chain [3]. HLA-C alleles with lysine 80 (includingHLA-Cw2, -Cw4 and -Cw5) are recognized by p58receptors of the KIR2DL1 group, whereas HLA-C allo-types with asparagine 80 (such as HLA-Cw1, -Cw3 and-Cw7) bind p58 KIR2DL2 and KIR2DL3 proteins.

The critical amino acid on the HLA molecule which con-trols the specificity of the KIR/HLA-C interaction is resi-due 80 [4]. However, although other residues also con-tribute to binding, the identity of the amino acids onthe KIR proteins important for binding to HLA-C hasnot been completely defined. Alignment of KIR2DL1sequences with KIR2DL2 and 2DL3 sequences revealsa high level of sequence identity with 13 dimorphic res-idues in the extracellular portion of the molecule. Twoclusters of dimorphic residues in the first Ig domain ofthe receptor represent a particular focus of variation[5–7]. A simple mutation in either of these clusters, resi-dues 44–46 [7–9] or residues 67–70 [10], affected bind-ing of KIR2D-Ig fusion proteins to HLA-C transfectants,but the relative importance of these regions and whetherother residues on the KIR molecule contribute to theHLA-C binding site remains unclear.

We have adopted an alternative strategy to assay KIR/HLA-C interaction. In this report we show that transfect-ants expressing fusion proteins comprising the extracel-lular domains of KIR molecules with the transmembraneand intracellular portions of CD3- ´ chain faithfully mimic

1480 J. Richardson et al. Eur. J. Immunol. 2000. 30: 1480–1485

Fig. 1. Schematic diagram of the chimeric KIR2D- ´ chainreceptor. The amino acid sequence through the fused regionis indicated.

Fig. 2. BW cells transfected with KIR2DL1- ´ were cultured,at the indicated ratios, with untransfected 721.221 cells,HLA-Cw6-transfected 721.221 cells or HLA-Cw7-trans-fected 721.221 cells. Supernatants were collected afterovernight culture and IL-2 content measured using prolifera-tion of CTLL cells as the assay.

the patterns of KIR recognition of HLA-C seen in in vitroassays of NK cell function, and that signaling via thesechimeric proteins can be readily measured. Further, wehave used this system to assay the effects of mutationsof polymorphic residues in the KIR on binding to HLA-Cand show that multiple amino acids influence the KIR/HLA-C interaction.

2 Results and discussion

2.1 Specific recognition of HLA-C by KIR- ‘ chainfusion proteins

Fusion proteins comprising the cytoplasmic domain ofthe TCR ´ chain and the extracellular portion of a heterol-ogous protein can induce cellular activation when cross-linked by antibody or ligand [11–13]. This observationhas been exploited to test whether cells expressingchimeric KIR2D- ´ receptors could be used as a modelsystem to study the binding of KIR molecules to HLAclass I proteins. Fig. 1 is a schematic of the chimericreceptor molecule. Transfection of a KIR2DL1- ´ fusionprotein into the mouse thymoma BW5147 generated acell line which expresses the KIR fusion protein at the cellsurface (data not shown) and which could be activatedto produce IL-2 when cultured with 721.221 cells trans-fected with the HLA-Cw*0602 molecule (Fig. 2). This rec-ognition was titratable and specific; culture with neitherMHC class I negative 721.221 cells nor 721.221 cellstransfected with HLA-Cw*0702 (that binds to KIR2DL2)triggered IL-2 production by the KIR2DL1- ´ transfect-ants. These data establish that this system can repro-duce the pattern of specific recognition of HLA-C seen inin vitro assays of NK cell function [2, 14].

2.2 A mutational analysis of KIR2D binding toHLA-C

Having verified that a specific interaction of KIR2DL1 withHLA-Cw*0602 could be demonstrated in this system andthat the chimeric protein can signal, a mutational analysiswas undertaken to identify some of the residues in theKIR molecule important for binding to HLA-C.

Alignment of KIR cDNA revealed the existence of variouspatterns of sequence variation [15], including residuesthat are polymorphic between KIR2D proteins and resi-dues that are conserved within the KIR2D molecules, butvary with respect to KIR3D molecules. Inspection of thecrystal structures of the extracellular domains of the p58receptors [16–18] showed that a number of these vari-able residues cluster at the domain 1/domain 2 interface.Of these amino acids previous studies have suggestedthat residues 44, 45 and 70 are involved in KIR2D bindingto HLA-C [7–10]. In the light of these data four mutants,all of which include amino acids located in the inter-domain “elbow” region, were constructed initially. AKIR2DL1 gene (NKAT1) was mutated by changing vari-ous amino acids to the residues found at the equivalentposition in KIR2DL3. These mutations were: N for K atthe dimorphic residue 190, KFK for MFN at the cluster ofdimorphic residues 44–46 and GPMM for SRMT at resi-dues 67–70. In addition amino acid 183, an aspartateresidue in all KIR2D molecules, but a histidine in allKIR3D receptors, was mutated to introduce the KIR3Dresidue into KIR2DL1. These constructs were trans-fected into BW cells, screened by FCM for equivalentlevels of receptor expression (data not shown), and mul-tiple, independently generated transfectants assayed forrecognition of HLA-C. Representative data from theseexperiments are shown in Fig. 3.

These data show that while the asparagine for lysinechange at residue 190 had no effect on KIR2DL1 recog-nition of HLA-Cw6, mutation of either residues 44–46 or67–70 resulted in loss of recognition of HLA-Cw6. Simi-larly, the replacement of an aspartate by a histidine at

Eur. J. Immunol. 2000. 30: 1480–1485 The HLA-C binding site on KIR 1481

Fig. 3. BW cells expressing a KIR2DL1- ´ fusion protein or mutants thereof were cultured at a ratio of 1:10 with untransfected721.221 cells or 721.221 cells transfected with HLA-Cw6, HLA-Cw7 or HLA-B27. Supernatants were collected after overnightculture and IL-2 content measured by ELISA.

position 183 resulted in loss of recognition of HLA-C.

The simplest interpretation of these data is that the iden-tity of the amino acids at residues 44–46, 67–70 and 183is critical for binding of KIR to HLA-C. That the substitu-tion of lysine 190 by an asparagine had no effect on rec-ognition might reflect either that this amino acid has norole in the KIR/HLA-C interaction or that its role is onethat can be satisfied by both lysine and asparagine, e.g.that of a hydrogen bond donor. An obvious explanationfor the drastic effect of the D183H mutation on KIR2Dbinding to HLA-C is that this is related to the change ofcharge involved in this substitution. Residue 183 is oneof a number of charged residues located in the putativeligand-binding region of the KIR [16–18] and it has previ-ously been suggested that the charge distribution in thisregion may be of great importance for recognition [7, 18].Another, not exclusive, interpretation of this result is thatthis mutation induces, either directly or indirectly, a localconformational change which abrogates the KIR/HLA-Cinteraction. In any event, substitution of the aspartateconserved in KIR2D molecules with the histidine found inKIR3D molecules also does not result in any gain of reac-tivity with a Bw4-expressing HLA-B molecule such asHLA-B27 (Fig. 3).

The results obtained with the KIR2DL1 mutants at resi-dues 44–46 and 67–70 are of particular interest (Fig. 3).Substitution of either KIR2DL1 residues 44–46 or 67–70with those from KIR2DL3 resulted in loss of recognition

of HLA-Cw6 (NK1 allele) without gaining the ability tobind either HLA-Cw7 (NK2 allele) or HLA-B27 (NK3allele), i.e. loss of interaction, but not gain or change ofspecificity. These data imply that both of these clustersof polymorphic residues contribute to the HLA-C bindingsite on a KIR molecule, a suggestion in agreement withthe previous observations that mutation at any one ofresidues 44, 45 or 70 was sufficient to lose KIR bindingto HLA-C [7–10]. The logical extension of this experi-ment was therefore to prepare a mutant KIR2DL1 con-taining both the 44–46 and 67–70 clusters fromKIR2DL3 and to test this chimera for recognition of HLA-Cw6 and -Cw7. The results of such an experiment areshown in Fig. 4 and indicate that transfer of both clustersis necessary and sufficient for transfer of the HLA-Cbinding specificity of KIR2DL3 to KIR2DL1. That the effi-ciency of recognition of HLA-Cw7 by the chimeric KIR isnever quite equal to recognition of HLA-Cw7 by wild-type KIR2DL3 probably indicates that other additionalsites within the receptor also have some influence on theinteraction of KIR2D molecules with HLA-C. This is con-sistent with the available structural data showing varia-tion between KIR2DL1 and KIR2DL3 in the angle of thehinge between the D1 and D2 domains and a 25°–30°rotation in the orientation of the D1 domain of KIR2DL3compared to that of KIR2DL [16–18]. These data havebeen interpreted as suggesting either that there aredistinct differences in the domain-domain orientationbetween these two KIR [17] or that this significant degreeof flexibility in the KIR molecule may be necessary to

1482 J. Richardson et al. Eur. J. Immunol. 2000. 30: 1480–1485

Fig. 4. BW cells expressing a KIR2DL1- ´ fusion protein, a KIR2DL3- ´ fusion protein or mutants of KIR2DL1 were cultured at aratio of 1:10 with untransfected 721.221 cells or 721.221 cells transfected with either HLA-Cw6 or HLA-Cw7. Supernatants werecollected after overnight culture and IL-2 content measured by ELISA.

allow a degree of “induced-fit” on ligand binding [18].In either case the mutagenesis data are consistent withthe idea that there may be subtle differences in thedetails of how these two KIR interact with their HLA-Cligands.

3 Concluding remarks

The major conclusion from this work is that the specific-ity of KIR2D interaction with HLA-C is controlled in largepart by the two clusters of polymorphic residues at posi-tions 44–46 and 67–70 in domain 1. These experimentshave also provided evidence that other residues close tothis binding site also contribute to this interaction, but donot allow discrimination between their playing a directrole in binding to HLA-C or an indirect role influencingthe conformation of the KIR, e.g. the orientation of resi-dues 44–46 and 67–70. Resolution of this question willonly come from detailed structural analysis of a KIR2D/HLA-C complex. The mutagenesis data also confirm andextend previous observations suggesting that KIR resi-dues 44, 45 and 70 are important for HLA-C binding[7–10], but are in apparent contrast to a previous reportwhich suggested that transfer of only residue 44 be-tween KIR2DL1 and KIR2DL2 was sufficient to transferthe HLA-C specificity between these two KIR [7]. Subse-quent studies by this group, however, revealed thatKIR2DL2, although nominally specific for NK2 alleles ofHLA-C, is actually cross-reactive with certain NK1 alleles

[8]; perhaps for this reason transfer of residue 44 alonewas sufficient to transfer the specificity of binding. In anyevent the suggestion that both clusters 44–46 and67–70 contribute to the HLA-C binding site on KIR isstrongly supported by structural data clearly indicatingthat both clusters contribute significantly to the putativebinding site [16–18] and further, that alteration of residue68 is likely to alter the main chain conformation of theKIR and thus alter the binding site structure around resi-dues 44–46 on the CC’ loop and residues 67–72 on theEF loop [17]. It is interesting to relate this conclusion tothe observed differences between activating and inhibi-tory KIR in binding affinity for HLA-C. p50 activating KIR,if they bind HLA-C at all, bind with a markedly loweraffinity than p58 inhibitory KIR [8–10] and this despitethe fact that some activating receptors differ from aninhibitory counterpart by as little as two or three aminoacids in the extracellular portion. Strikingly, however,inspection of sequence alignments comparing theknown inhibitory and activating KIR reveals that, withoutexception, the activating KIR contain sequence changesin either one or both of the two clusters of polymorphicresidues 44–46 and 67–70 [9]. Further, mutation of theactivating KIR to restore the “cluster-consensus se-quence” of the inhibitory receptors results in restorationof binding to HLA-C [8–10]. These observations stronglysupport the suggestion that both residues 44–46 and67–70 are important for KIR binding to HLA-C. Moreover,while these data leave open the question as to what isthe true physiological function of the activating KIR, they

Eur. J. Immunol. 2000. 30: 1480–1485 The HLA-C binding site on KIR 1483

would seem to imply that their having no or very lowbinding affinity for HLA is important for that function.

4 Materials and methods

4.1 Cloning, mutagenesis and transfection

Constructs encoding KIR2DL1- ´ and KIR2DL3- ´ fusion pro-teins were generated by overlap PCR using the oligonucleo-tide primers 5'-ACGCGTCGACCCGCCACCATGTYGCTCWYGGTCRTCAKCATGGC-3' and 5'-CAAGTAGCAGAGGTGCAGGTGTCGGGGGTTACC-3' and NKAT1 and NKAT2cDNA [19] as templates for the KIR genes and 5'-CCTGCACCTCTGCTACTTGCTAGATGGA-3' and 5'-CGGAATTCAGCGAGGGGCCAGGGTCTG-3' for CD3 ´ , using asa template a plasmid encoding a 2B4 TCR g chain- ´ fusion[11]. The products of these first rounds of PCR were gel puri-fied, annealed and a second PCR performed using the 5'KIR primer and the 3' ´ primer. These KIR- ´ fusion proteinswere then cloned into pCI-Neo (Promega) and sequenced.

Mutagenesis was performed using the Quikchangemutagenesis system (Stratagene) using the oligonucleo-tides 5'-GCTTCGGCTCTTTCCATCACTCTCCATACGAGTGG-3' (D183H), 5'-CCATACGAGTGGTCAAACTCAAGTGACCCACTGC-3' (K190N), 5'-CCTTCTGCACAGAGAGGGGAAGTTTAAGGACACTTTGCGCCTCATTGG-3' (44–46, MNF G KFK) and 5'-GCCAACTTCTCCATCGGTCCCATGATGCAAGACCTGGCAGGG-3' (67–70, SRMTG GPMM) and their complements. Mutant proteins were

fully sequenced to check that only the desired changeshad been introduced. For transfection, BW cells werewashed into ice-cold PBS, mixed with 100 ? g linearizedplasmid DNA and transfected by electroporation using aBioRad electroporator set at 250V, 500 ? F. After electro-poration cells were cultured in RPMI medium supple-mented with 10 % FCS for 2 days and then the mediumwas supplemented to 2 mg/ml G418. Transfected cellswere screened by flow cytometry.

4.2 FCM

FCM analysis was done using the mAb EB6 and GL183(Coulter) [20, 21].

4.3 Assay for recognition of HLA-C

BW cells expressing KIR- ´ proteins were mixed, at variousratios, with MHC class I-negative 721.221 cells [22] and721.221 cells transfected with HLA-Cw6 and -Cw7 [4] in afinal volume of 200 ? l and cultured overnight. The next day,120 ? l supernatant were harvested and stored at − 80 °Cuntil assayed for IL-2 content.

4.4 Assay for IL-2 production

In some experiments IL-2 production by the transfected BWcells was assayed using proliferation of IL-2-dependentCTLL cells. However, in most of the experiments shown pro-duction of mouse IL-2 was measured using an IL-2 ELISA kit(Pharmingen).

References

1 Ciccone, E., Pende, D., Viale, O. et al., Involvement of HLAclass I alleles in natural killer (NK) cell-specific functions: expres-sion of HLA-Cw3 confers selective protection from lysis by allo-reactive NK clones displaying a defined specificity (specificity 2).J. Exp. Med. 1992. 176: 963–971.

2 Colonna, M., Borsellino, G., Falco, M., Ferrara, G. B. and Stro-minger, J. L., HLA-C is the inhibitory ligand that determinesdominant resistance to lysis by NK1- and NK2-specific natural-killer cells. Proc. Natl. Acad. Sci. USA 1993. 90: 12000–12004.

3 Colonna, M., Spies, T., Strominger, J. L., Ciccone, E., Moretta,A., Moretta, L., Pende, D. and Viale, O., Alloantigen recognitionby two human natural killer cell clones is associated with HLA-Cor a closely linked gene. Proc. Natl. Acad. Sci. USA 1992. 89:7983–7985.

4 Mandelboim, O., Reyburn, H. T., Vales-Gomez, M., Pazmany,L., Colonna, M., Borsellino, G and Strominger, J. L., Protectionfrom lysis by natural killer cells of group 1 and 2 specificity ismediated by residue 80 in human histocompatibility leukocyteantigen C alleles and also occurs with empty major histocompati-bility complex molecules. J. Exp. Med. 1996. 184: 913–922.

5 Reyburn, H., Mandelboim O., Vales-Gomez, M., Sheu, E. G.,Pazmany, L., Davis, D. M. and Strominger, J. L., Human NKcells: their ligands, receptors and functions. Immunol. Rev. 1997.155: 119–125.

6 Valiante, N. M., Lienert, K., Schilling, H. G., Smits, B. J. andParham, P., Killer cell receptors: keeping pace with MHC class Ievolution. Immunol. Rev. 1997. 155: 155–164.

7 Winter, C. C. and Long, E. O., A single amino acid in the p58killer cell inhibitory receptor controls the ability of natural killercells to discriminate between the two groups of HLA-C allotypes.J. Immunol. 1997. 158: 4026–4028.

8 Winter, C. C., Gumperz, J. E., Parham, P., Long, E. O. andWagtmann, N., Direct binding and functional transfer of NK cellinhibitory receptors reveal novel patterns of HLA-C allotype rec-ognition. J. Immunol. 1998. 161: 571–577.

9 Vales-Gomez, M., Reyburn, H. T., Erskine, R. A. and Stromin-ger, J. L., Differential binding to HLA-C of p50-activating andp58-inhibitory natural killer cell receptors. Proc. Natl. Acad. Sci.USA 1998. 95: 14326–14331.

10 Biassoni, R., Pessino, A., Malaspina, A., Cantoni, C., Bottino,C., Sivori, S., Moretta, L. and Moretta, A., Role of amino acidposition 70 in the binding affinity of p50.1 and p58.1 receptors forHLA-Cw4 molecules. Eur. J. Immunol. 1997. 27: 3095–3099.

11 Engel, I., Ottenhoff, T. H. M. and Klausner, R. D., High-efficiency expression and solubilization of functional T cell anti-gen receptor heterodimers. Science 1992. 256: 1318–1321.

12 Irving, B. A. and Weiss, A., The cytoplasmic domain of the T cellreceptor zeta chain is sufficient to couple to receptor-associatedsignal transduction pathways. Cell 1991. 64: 891–901.

13 Romeo, C. and Seed, B., Cellular immunity to HIV activated byCD4 fused to T cell or Fc receptor polypeptides. Cell 1991. 64:1037–1046.

1484 J. Richardson et al. Eur. J. Immunol. 2000. 30: 1480–1485

14 Colonna, M., Brooks, E. G., Falco, M., Ferrara, G. B. and Stro-minger, J. L., Generation of allospecific natural killer cells bystimulation across a polymorphism of HLA-C. Science 1993. 260:1121–1124.

15 Steffens, U., Vyas, Y., Dupont, B. and Selvakumar, A., Nucleo-tide and amino acid sequence alignment of human killer cellinhibitory receptors (KIR), 1998. Tissue Antigens 1998. 51:398–413.

16 Fan, Q. R., Mosyak, L., Winter, C. C., Wagtmann, N., Long, E.O. and Wiley, D. C., Structure of the inhibitory receptor forhuman natural killer cells resembles haematopoietic receptors.Nature 1997. 389: 96–100.

17 Maenaka, K., Juji, T., Stuart, D. I. and Jones, E. Y., Crystalstructure of the human p58 killer cell inhibitory receptor(KIR2DL3) specific for HLA-Cw3-related MHC class I. Structure1999. 7: 391–398.

18 Snyder, G. A., Brooks, A. G. and Sun, P. D., Crystal structure ofthe HLA-Cw3 allotype-specific killer cell inhibitory receptorKIR2DL2. Proc. Natl. Acad. Sci. USA 1999. 96: 3864–3869.

19 Colonna, M. and Samaridis, J., Cloning of immunoglobulin-superfamily members associated with HLA-C and HLA-B recog-nition by human natural killer cells. Science 1995. 268: 405–408.

20 Moretta, A., Tambussi, G., Bottino, C., Tripodi, G., Merli, A.,Ciccone, E., Pantaleo, G. and Moretta, L., A novel surface anti-gen expressed by a subset of human CD3− CD16+ natural killercells. Role in cell activation and regulation of cytolytic function.J. Exp. Med. 1990. 171: 695–714.

21 Moretta, A., Bottino, C., Pende, D. et al., Identification of foursubsets of human CD3− CD16+ natural killer (NK) cells by theexpression of clonally distributed functional surface molecules:correlation between subset assignment of NK clones and abilityto mediate specific alloantigen recognition. J. Exp. Med. 1990.172: 1589–1598.

22 Shimizu, Y. and DeMars, R., Production of human cells express-ing individual transferred HLA-A, -B, -C genes using an HLA-A,-B, -C null human cell line. J. Immunol. 1989. 142: 3320–3328.

Correspondence: Jack L. Strominger, Department ofMolecular and Cellular Biology, Harvard University, 7 DivinityAvenue, Cambridge MA 021 38, USAFax: +1-617-496-8351e-mail: jlstrom — fas.harvard.eduJ. Richardson’s present address: Millenium PharmaceuticalsInc., 75 Sidney Street, Cambridge, MA 02139, USA

Eur. J. Immunol. 2000. 30: 1480–1485 The HLA-C binding site on KIR 1485