Bjorkman, P. J., Saper, M. A., Samraoui, B., Bennet, W. S.,. Strominger, J. L. ... Maloy, W. L., Coligan, J.F. & Biddison, W. E. (1988) J. Exp. Med. 168, 725-736. 7.
Proc. Nati. Acad. Sci. USA Vol. 88, pp. 11325-11329, December 1991 Immunology
Positioning of a peptide in the cleft of HLA-A2 by complementing amino acid changes FRANCE LATRONtt, ROBERT MOOTSt, JONATHAN B. ROTHBARD§, TOM P. J. JACK L. STROMINGERt II, AND ANDREW MCMICHAELt
tDanaFarber Cancer Institute, Harvard Medical School, Boston, MA 02115; tThe Institute of Molecular Medicine, University of Oxford, Oxford, England; 'Department of Biochemistry and Molecular Biology and lHoward Hughes Medical Institute, Harvard University, Cambridge, MA 02138; and *ImmuLogic, Palo Alto, CA 95304
Contributed by Jack L. Strominger, September 9, 1991
Several mutant HLA-A2 molecules have been ABSTRACT constructed and expressed in the mutant human B-cell line COR, which lacks HLA-A and HLA-B antigens, and examined for presentation of a previously defined peptide epitope derived from the influenza matrix protein to appropriate human cytotoxic T-lymphocyte lines. When leucine residue 66 in this matrix peptide containing residues 57-68 (matrix peptide 5768) was replaced by arginine, the resulting matrix peptide 57-68 R66 was not presented to HLA-A2, but the mutation Y116D (tyrosine to aspartic acid at residue 116) in the floor of the peptide binding cleft near its right end dramatically restored peptide presentation. A similar result was obtained by substitution of ornithine for leucine at residue 66. These data provide strong support for a model in which the peptide is orientated with its amino terminus at the left end of the cleft of HLA-A2 and its carboxyl terminus at the right.
Class I major histocompatibility complex (MHC) molecules present small peptides to cytotoxic T lymphocytes (CTLs) (1). The peptides are known to bind in a cleft of the MHC molecules (2-4), and several have been defined from a number of viral proteins (for review, see ref. 5). However, it is not yet known how any peptide is orientated in the cleft. Studies of peptide antigen presentation by human class I MHC molecules mutated at single positions have identified probable contact points on the a-helices that border the cleft and on the floor (6-11). In the present study, a change in the floor of the cleft of HLA-A2 at position 116 that creates a negatively charged pocket has been genetically engineered and found to rescue presentation of analogues of the influenza matrix peptide containing residues 57-68 (matrix peptide 57-68) that have a complementary change to a positively charged amino acid at position 66 in the peptide. If the peptide is positioned in the cleft such that these two residues make contact and previously defined contact residues are taken into account, the orientation of the peptide in the cleft can be deduced. Moreover, a model for the peptide bound in the cleft has recently been deduced based on a clear image of the peptide visualized by x-ray crystallography of HLA-B27 (12, 13). In that peptide model the amino and carboxyl termini are also orientated left and right, respectively, in the cleft of HLA-B27. The present data provide strong support for the orientation in that model. In addition, a recent study of complementary mutations in a human class II MHC molecule and a peptide presented by it has reached a similar conclusion regarding the orientation of the peptide in its binding cleft (14). However, the structure of a class II MHC molecule has not yet been determined, and in that case the orientation is, therefore, necessarily based on a hypothetical structure
deduced for class II from its homology to a class I MHC molecule (15).
MATERIALS AND METHODS Mutations at predefined residues in HLA-A2 (HLA-A*0201) were made by site-directed mutagenesis (16) of genomic HLA-A2 DNA and transfected into the HLA-A- and -Bnegative monocytic cell line HMy2.ClR (C1R) (17), followed by selection for high-level expression. Residues 95, 97, 114, and 116, which are polymorphic in class I molecules and located on the central /3-strands of the a 1-a 2 /3-sheet (2, 3) were substituted to recreate allelic variability at these positions. The changes made were Y116D (tyrosine changed to aspartic acid at position 116) as in HLA-B*2705, H114D (HLA-B*2706), H114D/Y116D (HLA-B*4401), H114R/ Y116D (HLA-A*6801), V95I (HLA-A*6801 and A*6802), H114R (HLA-A*6801), and R97M (HLA-A*6801) (Table 1). The two substitutions in the HLA-A*6801 at residues 114 and 116 are known to result in a pocket in the floor of this molecule that is not found in HLA-A2 (4). The mutated HLA-A2 genes were subcloned into the pSV2neo plasmid containing the Escherichia coli neomyocinresistance gene, which confers Geneticin (G418 sulfate) resistance to transfected human cells. After linearization with BamHI, the plasmid carrying the mutated HLA-A2 gene was transfected, using the Lipofectin reagent (GIBCO) into the HLA-A- and -B-negative C1R cell line (1 x 107 CiR cells plus 10 jig pSV2neo DNA in serum-free medium). After incubation at 37°C in 5% C02/95% air for 24 hr, cells were pelleted and resuspended in 10 ml of RPMI 16 medium plus 10% (vol/vol) fetal bovine serum (RPMI/10). G418 (800 ,g/ml) was added 24-48 hr later. Cell surface expression was assessed by analytical flow cytometry with the HLA-A2specific monoclonal antibodies BB7.2 (18), MA2.1 (19), 4B3 (20), and PA2.1 (21) by indirect fluorescence. Stable highexpressor transfected cells of each mutated HLA-A2 gene were selected by cell sorter. All of the transfected cells used in this study expressed equivalent high levels of HLA-A2 as assayed in this way. CTL lines specific for the influenza virus type A matrix peptide 57-68 were generated as described (9) using Hypaque/Ficoll-purified peripheral blood mononuclear cells from three normal HLA-A2-positive donors. The CTL lines were maintained in vitro by weekly stimulation with irradiated autologous B-lymphoblastoid cells and pulsed with the peptide and interleukin 2. All three CTL lines gave identical results, and data from only one, JM-31, are shown. Cells were used in a standard 4-hr cytolytic assay (9) after they had been in culture for approximately 20 days. They retained cytolytic activity for at least a month. Target cells were labeled for 1
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Abbreviations: CTL, cytotoxic T lymphocyte; MHC, major histocompatibility complex. 11325
Immunology: Latron et al.
Proc. Natl. Acad. Sci. USA 88 (1991)
Table 1. Amino acid differences of HLA-A2, -A68, -B44, -B27, and related molecules in the floor of the peptide-binding groove at residues 95, 97, 114, and 116 Amino acid residue 116 97 114 95 Gene Allele Tyr His Arg Val A2 A*0201 Asp Met Ile A*6801 A68 Arg Ile A*6802 Asp Asp Ile B*4401 B44 Asp Asn B27 B*2705 Leu Asn Asp B*2706 Leu Sequences are compared to HLA-A2, and identities with HLA-A2 are given by dashes.
hr with 51Cr and mixed with CTLs at effector/target ratios of 2:1. The influenza A matrix peptide 57-68 was added to these cells at a range of concentrations of 0.001 to 10 jtg/ml immediately prior to the addition of CTLs. Supernatants were harvested after 4 hr and specific lysis was calculated from the formula: 100 x [(E - M)/(D - M)], where E is experimental release, M is release in the presence of culture medium, and D is release in the presence of 5% (vol/vol) Triton X-100. CTL assays included COR cells that had not been transfected as a negative control and CiR cells that expressed transfected normal genomic HLA-A2 DNA as a positive control. Target cells were CiR cells that express transfected genomic HLA-A2 DNA with substitutions at positions- 95, 97, 114, and/or 116. They are referred to as V95I, R97M, H114-D, Y116D, H114D/Y116D, H114R, and H114R/Y116D; the first letter represents the single letter code for the amino acid in HLA-A2, the number is the residue in HLA-A2, and the last letter is the amino acid code for the mutation. The influenza type A matrix peptide 57-68 H2N-Lys-GlyIle-Leu-Gly-Phe-Val-Phe-Thr-Leu-Thr-Val-CO2H M was synthesized on a solid-phase peptide synthesizer (Applied Biosystems). The purity was determined to be >95% by HPLC on a C18 reverse-phase column.
RESULTS AND DISCUSSION CTL lines specific for the influenza A matrix peptide 57-68
were generated from three normal donors. When HLA-A2 and several HLA-A2 mutants were tested with the matrix
peptide 57-68 over a range of concentrations, mutants V951, R97M, H114D, and Y116D (Figs. 1 and 2A) presented the peptide to the CTLs and to C1R cells transfected with wild-type HLA-A2. However, the mutant H114R and the double mutants H114D/Y116D and H114R/Y116D failed to present the matrix peptide (Fig. 1). These results demonstrate the importance of the polymorphic residues at 114 and 116 in presentation of the matrix peptide. Those of the mutants that had charge changes were tested for the ability to present analogues of the matrix peptide 57-68 with complementing charged amino acid substitutions. One of them, Y116D, showed an unusual dramatic response to the matrix peptide altered by a complementing positively charged amino acid. The influenza A matrix peptide 57-68 R66 (arginine substituted for leucine at position 66) was not presented to CTLs by HLA-A2-transfected C1R cells (Fig. 2B). However, presentation of this substituted peptide was restored to a level comparable to that of the native matrix peptide in OCR cells transfected with Y116D, the mutant that expressed HLA-A2 molecules with aspartic acid instead of tyrosine at position 116 (compare Fig. 2 A and B). Titration of the peptides showed that cells expressing this mutant HLA-A2 molecule (Y116D) were 1000 times wmre efficient in presenting the R66 peptide, although surface expression of HLA-A2 was equivalent in both cell lines as determined by analytical flow cytometry. Other charged substitutions at position 66 in the peptide, D66, E66, and K66, also abolished the presentation by the wild-type HLA-A2 molecule, which suggests that this residue in the peptide is very important for the binding to HLA-A2. However, none of these peptides was significantly better presented by the Y116D molecule. Other peptides with substitutions of lysine or argimne at residues 62, 63, or 65 were presented by neither wild-type HLA-A2 nor the mutant Y116D. The R60 substitution reduced recognition when presented by HLA-A2 by a factor of 100, but presentation was only minimally enhanced by the Y116D mutation (Fig. 2C). Similarly, the substitutions R57, K58, K59, D60, E60, K60, and K68 did not result in better presentation by the mutant Y116D than by HLA-A2 (data not shown). A shorter matrix peptide containing residues 59-68 (matrix peptide 59-68) was also used, with or without an ornithine substituted at position 66 (066). This substitution also abolished presentation by HLA-A2 expressing cells, but this was restored by cells expressing the mutant Y116D (Fig. 3B,
Y116D A2 ~ 0 C.)
Hi 14R/Y1 16D H1 14D/Y1 16D H 14R 0.001 0.01 0.1 Type A matrix peptide 57-68, ug/ml
FIG. 1. Lysis of cells expressing HLA-A2 or mutant HLA-A2 genes by HLA-A2-restricted matrix peptide 57-68-specific CTLs in the presence of various concentrations of the peptide.
Immunology: Latron et al.
Proc. Natl. Acad. Sci. USA 88 (1991)
Gene Transfected -_ A2 0 Y116D
Gene Transfected * A2
Zcn -> 30
0 a) CL Co 20
Y 0.... V116D
Type A matrix peptide 57-68,
Gene Tra nsfected _,oYl 1 6D
Type A matrix peptide 59-68, pg/ml 30
Y 16D 40 U,
.cn -> 30 C)
Type A matrix peptide 57-68 (R66),
Type A matrix peptide 59-68 (066), ,ug/ml Gene Transfected
,,OY1 I 6D
*>30-} Co ,,20
Type A matrix peptide 57-68 (R60),
FIG. 2. Effect of influenza A matrix peptide with a leucine > arginine substitution at position 60 or 66 on presentation by HLA-A2 and by Y116D to the JM CTL line. JM CTL recognition of transfected cells expressing HLA-A2 and Y116D in the presence of the following peptides is shown. (A) The influenza A matrix peptide 57-68 (KGILGFVFTLTV). (B) Influenza A matrix peptide 57-68 with an arginine at position 66 (R66). (C) Same peptide with an arginine at position 60 (R60). The killer/target ratio was 2:1. Each point was the mean of four experiments made in duplicate. compare to Fig. 3A using the unsubstituted matrix peptide 59-68). In contrast, the peptide with ornithine substituted at position 60 was presented less efficiently by the mutant Y116D than by HLA-A2, although both were greatly reduced (Fig. 3C). These results show that CTLs recognized the matrix peptides with arginine substituted at position 66 when presented
Type A matrix peptide 59-68 (060),
FIG. 3. Recognition by the JM CTLs of the influenza A matrix peptide 59-68 and the influenza A matrix peptide substituted at position 66 or 60 with an ornithine by HLA-A2 and by Y116D expressing C1R cells. (A) Influenza A matrix peptide 59-68. (B) Influenza A matrix peptide with an ornithine at position 66 (066). (C) Same peptide with an ornithine at position 60 (060).
by the mutant Y116D to the same level as the matrix peptide-HLA-A2 combination and recognized matrix peptides with ornithine at position 66 to a lesser extent. This implies that R66 or 066 of the peptide and D116 of HLA-A2 are close together and may directly interact to form a salt bridge. Replacement of the leucine at position 66 in the peptide by these positively charged amino acids severely reduced presentation of these peptides by HLA-A2. Replacing tyrosine with the smaller aspartic acid at position 116 in HLA-A2 creates a negatively charged pocket in the floor of
Immunology: Latron et al.
Proc. Natl. Acad. Sci. USA 88 (1991)
FIG. 4. (A) Section through the cleft of HLA-A2 viewed from the right of B. HLA-A2 is shown in green with two alternate amino acids at position 116, tyrosine (yellow) and aspartic acid (red). The molecular surface is shown in yellow dots and the additional negatively charged pocket created by the Y116D substitution is in red dots. (B) Model positioning the influenza A matrix peptide 57-68 (KGILGFVFTLTV) in the antigen-binding site of the HLA-A2 molecule deduced from this study. The positions of the polymorphic a-sheet residue 116 on the cleft floor of the HLA-A2 molecule and residue 45 at the apex of the B pocket are shown. The arrows indicate the positions of the side chains that are proposed to fill the 45 (I59 or L60) and 116 (R66) pockets in the Y116D mutant.
the cleft (Fig. 4A), which complements the substitution in the peptide. Further support comes from the observation that a lobe from the extra electron density found in the cleft reaches to tyrosine 116 on the floor of the cleft (figure 14 in ref. 22). This lobe appears to be a side chain of a bound peptide and would correspond to leucine 66 in the matrix peptide 57-68 or to arginine in the R66 analogue. The fact that the mutant Y116D also presents the unsubstituted peptide containing leucine 66 (Fig. 2A) may be due to protonation of H114, which would neutralize the negative charge at D116. Another possibility is that R97 could be linked by a salt bridge to D116 instead of D77. These features may represent a way of constructing a pocket that can accommodate several different types of amino acid side chain. Notably, the data obtained correspond to the fact that endogenous peptides bound to HLA-A2 (containing Y116) have only leucine or valine (hydrophobic amino acids) at this position (23), whereas peptides bound to HLA-B27 (containing D116, as in the HLA-A2 mutant described here) have leucine (hydrophobic) or arginine (positively charged) and several other types of amino acid at the same position (24). Such a pocket may
increase enormously the range of peptide epitopes that can be bound by a given allele. Previous data have shown that substitutions in HLA-A2 at residues 9, 62-63, 66, 70, 152, 155, 156, and 161 can affect presentation of the peptide to CTLs (6-11). Recent experiments have also indicated that residue 45, at the bottom of the B, or 45, pocket (4, 22) is also critically important because a mutant with glutamic acid at residue 45 (as in several HLA-B locus molecules) instead of methionine failed to present matrix peptide or any ofits analogues (data not shown). None of matrix peptides R60, K60, and 060 or K59 restored presentation, however. This may be because creation of an appropriate negatively charged 45 pocket would require additional amino acid changes at the other polymorphic positions (4). Since peptide residue 66 (leucine or arginine) is near the carboxyl terminus of the matrix peptide, all of these data can only be accommodated if the carboxyl terminus of the peptide is in the "right" end of the cleft where residue 116 is located and the amino terminus in the "left." Moreover, the density found in the B pocket of HLA-A2 has the size and shape of a leucine or isoleucine side chain (figure 13 in ref.
Immunology: Latron et al. 22), and this model is compatible with isoleucine 59 or leucine 60 of the matrix peptide occupying the 45 pocket. Based on the above arguments, we conclude that the influenza matrix peptide 57-68 lies in the cleft of HLA-A2 from left (amino terminus) to right (carboxyl terminus), as shown in Fig. 4B. In this position it would make contact with previously defined critical contact residues in the cleft of the HLA-A2 molecule. Note Added in Proof. Recently, we have identified matrix peptide 58-66 as the optimal epitope in both cell mediated lysis and binding assays. When leucine at position 66 in this peptide nonamer was changed to arginine and tested with the Y116D mutant, similar results to those described here were obtained (J. Morrison, J. Elvin, F. Latron, F. Gotch, J. L. Strominger, and A. J. McMichael, unpublished data).
This work was supported by grants from the National Institutes of Health (CA47554), the Howard Hughes Medical Institute, the Medical Research Council, the International Union Against Cancer, the European Molecular Biology Organization, and the International Human Frontier Science Program Organization (HFSP). We are grateful to Cetus for the gift of recombinant interleukin 2. 1. Townsend, A. R. M., Rothbard, J., Gotch, F. M., Bahadur, G., Wraith, D. & McMichael, A. J. (1986) Cell 44, 959-968. 2. Bjorkman, P. J., Saper, M. A., Samraoui, B., Bennet, W. S., Strominger, J. L. & Wiley, D. C. (1987) Nature (London) 329, 506-512. 3. Bjorkman, P. J., Saper, M. A., Samraoui, B., Bennet, W. S., Strominger, J. L. & Wiley, D. C. (1987) Nature (London) 329,
512-518. 4. Garrett, T. P. J., Saper, M. A., Bjorkman, P. J., Strominger, J. L. & Wiley, D. C. (1989) Nature (London) 342, 692-6%. 5. Townsend, A. R. M. & Bodmer, H. C. (1989) Annu. Rev. Immunol. 7, 601-624. 6. Hogan, K. T., Shimojo, N., Walk, S. F., Engelhard, V. H., Maloy, W. L., Coligan, J. F. & Biddison, W. E. (1988) J. Exp. Med. 168, 725-736.
Proc. Natl. Acad. Sci. USA 88 (1991)
7. McMichael, A. J., Gotch, F. M., Santos-Aguado, J. & Strominger, J. L. (1988) Proc. Nat!. Acad. Sci. USA 85,9194-9198. 8. Hogan, K. T., Clayberger, C., Bernhard, E. J., Walk, S. F., Ridge, J. P., Parham, P., Krensky, A. M. & Engelhard, V. (1989) J. Immunol. 142, 2097-2104. 9. Gotch, F. M., McMichael, A. J. & Rothbard, J. (1988) J. Exp. Med. 168, 2045-2057. 10. Mattson, D. H., Shimojo, N., Cowan, E. P., Baskin, J. J., Turner, R. V., Shvetsky, B. D., Coligan, J. E., Maloy, W. L. & Biddison, W. E. (1989) J. Immunol. 143, 1101-1107. 11. Robbins, P. A., Lettice, L., Rota, P., Santos-Aguado, J., Rothbard, J., McMichael, A. J. & Strominger, J. L. (1989) J. Immunol. 243, 4098-4103. 12. Gorga, J. C., Madden, D., Prendergast, J., Wiley, D. C. & Strominger, J. L. (1991) Proteins, in press. 13. Madden, D., Gorga, J. C., Strominger, J. L. & Wiley, D. C. (1991) Nature (London) 353, 321-325. 14. Krieger, J. I., Karr, R. W., Grey, H. M., Yu, W. Y., O'Sullivan, D., Batovsky, L., Zheng, Z. L., Colon, S. M., Gaeta, F. C. A., Sidney, J., Albertson, M., Del Guercio, M. F., Chesnut, R. W. & Sette, A. (1991) J. Immunol. 146, 2331-2340. 15. Brown, J., Jardetzky, T., Saper, M. A., Samraoui, B., Bjorkman, P. J. & Wiley, D. C. (1988) Nature (London) 332, 848850. 16. Kunkel, T. A. (1985) Proc. Nat!. Acad. Sci. USA 82, 488-492. 17. Storkus, W. J., Howell, D. N., Salter, R. D., Dawson, J. R. & Cresswell, P. (1987) J. Immunol. 138, 1657-1659. 18. Parham, P. & Brodsky, F. M. (1981) Hum. Immunol. 3, 277299. 19. McMichael, A. J., Parham, P., Rust, N. & Brodsky, F. (1980) Hum. Immunol. 1, 121-129. 20. Yang, S. Y., Morishima, Y., Collins, N. H., Alton, T., Pollack, M. S., Yunis, E. J. & Dupont, B. (1984) Immunogenetics 19, 217-231. 21. Parham, P. & Bodmer, W. F. (1978) Nature (London) 276, 397-399. 22. Saper, M. A., Bjorkman, P. J. & Wiley, D. C. (1991) J. Mol. Biol. 219, 277-317. 23. Falk, K., R6tzschke, O., Stevanovic, S., Jung, G. & Rammensee, H.-G. (1991) Nature (London) 351, 290-296. 23. Jardetzky, T. S., Lane, W. S., Robinson, R. A., Madden, D. R. & Wiley, D. C. (1991) Nature (London) 353, 326-329.