- Email: [email protected]
The P9 peptide sidechain specificity of I-Ad
Immunology Letters 70 (1999) 199 – 202
www.elsevier.com/locate/
The P9 peptide sidechain specificity of I-Ad Kristian Bartnes a,*, Xiaotong Li a, Jean Paul Briand b, Paul J. Travers c, Kristian Hannestad a a
Department of Immunology, Institute of Medical Biology, Uni6ersity of Tromsø, School of Medicine, N-9037 Tromsø, Norway b Institut de Biologie Moleculaire et Cellulaire, UPR 9021 CNRS, Strasbourg, France c Imperial Cancer Research Fund Unit for Structural Molecular Biology, Department of Crystallography, Birkbeck College, London WC1E 7HX, UK Received 28 July 1999; accepted 27 August 1999
Abstract The murine MHC class II variant I-Ad confers susceptibility to herpes simplex virus (HSV)-induced keratitis and relative protection against type 1 diabetes mellitus. The association to these autoimmune diseases appears to be largely determined by the peptide sidechain specificity of the P9 pocket, which we therefore have analyzed in detail. Assessment of T-cell responses and I-Ad binding capacity of position 446-substituted analogs of an IgG2a allotype b (IgG2ab) heavy chain peptide demonstrates that engagement of the P9 pocket is crucial for effective peptide presentation. Sidechain size rather than charge decides the capacity to engage the P9 pocket. Thus, small, uncharged sidechains are accepted, whereas acidic and aromatic amino acids as well as lysine and arginine are disfavored. The specificity of the P9 pocket of I-Ad (serineb57) is distinct from that of the diabetes-associated I-Ag7 (aspartic acidb57), supporting the contention that the polymorphism at residue b57 influences diabetes susceptibility via P9-specific effects on the repertoires of self peptides presented to T cells. Furthermore, the data rationalize the susceptibility to HSV-induced keratitis conferred by the a and the protection conferred by the b allotypes of the IgG2a heavy chain. Keratitogenic T cells, which cross-react with the viral UL6 protein and a corneal antigen, are silenced in IgG2ab mice because of antigenic mimicry with g2ab 435–451. Our finding that the lysine P9 residue of the corresponding g2aa allopeptide precludes high-affinity binding to I-Ad indicates that the susceptibility of IgG2aa mice reflects inefficient thymic presentation of autologous IgG2a and thus failure to purge the T-cell repertoire of the pathogenic clones. © 1999 Elsevier Science B.V. All rights reserved. Keywords: Diabetes mellitus; I-Ad; Keratitis; MHC class II; Motif; P9 pocket
1. Introduction The murine MHC class II allomorph I-Ad is the restriction element for T cells which mediate herpes simplex virus (HSV)-induced autoimmune keratitis [1] and confers relative protection against type 1 diabetes mellitus [2]. I-Ad-restricted keratitogenic T cells which recognize a corneal autoantigen crossreact with the HSV UL6 protein which mimics the IgG2a heavy chain allotype b (IgG2ab) peptide g2ab 435 – 451 [3]. IgG2ab * Correspondence author. Tel.: +47-776-26000; fax: +47-77628298. E-mail address: [email protected] (K. Bartnes)
mice constitutively present this g2ab epitope in complex with I-Ad [4] and are resistant to HSV-induced keratitis [1]. In contrast, IgG2aa mice are susceptible, most likely as a result of inefficient thymic presentation of the corresponding g2aa self peptide since the lysine which corresponds to relative position 9 (P9) in the I-Ad motif encompassed by the g2ab 438–446 sequence precludes high-affinity binding to I-Ad [5]. The specificity of the P9 pocket of the I-Ad peptide binding site appears to be crucial for the susceptibility to autoimmune diabetes mellitus as well, since the predisposition to develop diabetes conferred by I-A and HLA-DQ variants is strongly influenced by the polymorphism at residue b57 [6] which lines the P9 pocket [7]. Thus, I-Ad (aspartic
0165-2478/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 5 - 2 4 7 8 ( 9 9 ) 0 0 1 4 9 - 2
200
K. Bartnes et al. / Immunology Letters 70 (1999) 199–202
acidb57) confers protection when transgenically introduced in non-obese diabetic (NOD) mice (I-Ag7, serineb57) [2]. By assessing antigenicity and I-Ad binding capacity of single-residue substituted analogs of peptide g2ab 435–447, we have completed the analysis of the I-Ad P9 pocket specificity and compared it with that of I-Ag7.
2. Materials and methods The responses of the I-Ad-restricted T-cell hybridoma B5 [4] to a panel of single amino acid substituted analogs of the wildtype g2ab 435-447 (YFMYSKLRVQKST) allopeptide epitope were determined in a growth inhibition assay, as described [8]. Briefly, B5 cells (5×104 per well) were cultured together with irradiated (2000 rad) BALB/c spleen cells (5 ×105 per well) and peptide or medium alone for 18 h followed by a 6-h pulse with 3H-thymidine. After harvesting cultures onto glass fiber filter mats with a Micromate cell harvester (Packard Instrument, Meriden, CT), 3H-thymidine incorporation was determined with a Matrix96 Direct b Counter (Packard Instrument). The relative I-Ad-binding capacity of various peptide analogs was determined by flow cytometric assessment of their ability to compete with biotinylhexanoate-g2ab 436 – 451 for binding to I-Ad homozygous A20 cells [5], using an Epics Profile II flow cytometer equipped with a 488 nm argon laser (Coulter Electronics, Luton, UK). Peptide synthesis and purification has been described [5].
to I-Ad in many different registers [7] may complicate interpretation of direct binding data. Notably, however, the responses of the g2ab-reactive, I-Ad-restricted T-cell hybridoma B5 (Fig. 1, upper panel) correlated well with the relative binding capacities of the various analogs. Since the T-cell receptor generally accepts only a single binding register, it is unlikely that our analysis is confounded by multiple registers.
4. Discussion The present analysis demonstrates that the P9 pocket of I-Ad is less selective than that accommodating P4 [9], but more selective than pocket P1 [5]. Aromatic and acidic sidechains conferred a 1000-fold loss in antigenicity. Thus, the sidechain at P9 has a striking impact on the antigenic potency of peptides presented by I-Ad. Most likely, this exclusively reflects P9-mediated effects on the peptide/I-Ad interaction, since the P9 sidechain of hemagglutinin 126–138 (which is serine, as in the g2ab epitope) is buried within the I-Ad molecule [7] and thus inaccessible to
3. Results Serine446 (P9) of the g2ab 435 – 447 determinant was substituted with all natural amino acids that had not previously been analyzed except isoleucine and the labile cysteine, and the capacity of the various analogs to bind I-Ad and their antigenicity for a g2ab 435– 447/I-Ad-reactive T-cell hybridoma were determined (Fig. 1). Notably, via this approach we have recently arrived at a model for allele-specific peptide/I-Ad interactions [5] which is in agreement with subsequent crystallographic analysis [7]. The relative binding capacity varied 100-fold between the different analogs (Fig. 1, lower panel). Combined with previous direct binding analyses [5,9] the data demonstrate that I-Ad favors histidine and the small, uncharged sidechains of alanine, glycine, serine and threonine at P9, whereas arginine, lysine and acidic and aromatic amino acids preclude high-affinity binding (Table 1). The indications that the ovalbumin peptide 323 – 339 may bind
Fig. 1. Antigenicity (upper panel) and I-Ad binding capacity (lower panel) of analogs of the g2ab 435 – 447 peptide substituted at P9 (g2ab 446), determined by a T-cell hybridoma growth inhibition assay [8] and flow cytometric direct binding analysis [5], respectively. The data represent values relative to the wildtype g2ab 435 – 447 (serine446). Data from a representative experiment are shown in each panel.
K. Bartnes et al. / Immunology Letters 70 (1999) 199–202
201
Table 1 Compilation of P9 peptide sidechain specificities reported for I-Ad and I-Ag7a P9 residue
I-Ad binding Sette et al., 1989 [9]
gly ala val leu lys arg his asp glu asn glnb ser thr met phe tyr trp pro
2 3 2
I-Ag7 binding Bartnes et al., 1997 [5]
This study
3 3 1 1
1 3 1 1 2 3 3
3 2 1
1 1 2
Harrison et al., 1997 [10] 1 1 1 2 3 3 2 2 1 2 2 1 1 2 3 3 2 1
a Relative affinities are indicated by numbers from 1–3, where 3 represents the best binders. For I-Ad, the values are assigned to the various substitutions in accordance with [10]. For I-Ag7, the classification of Harrison et al. [10] is used with respect to well-tolerated, weakly tolerated, and nontolerated substitutions. Only studies in which IC50% values were obtained by direct binding T cell-independent analysis are included. b Although IC50% values have not been determined with I-Ad, there is evidance that gln is accepted [14].
the T-cell receptor. The preference for small P9 sidechains most likely reflects the fact that the corresponding pocket is relatively shallow [7]. Notably, thespecificity of peptide/class II MHC interactions is not solely determined by the ability of peptide sidechains to productively engage specificity pockets, as anchor substitutions can exert position-specific, dominant negative effects which substantially contribute to the overall peptide/class II affinity [12,13]. Accordingly, since I-Ad accepts glycine at P9 [5,9], the specificity at this position may predominantly reflect repulsion of large P9 sidechains. However, this must be qualified by the fact that also the relatively large histidine allows good binding ([14], this study). Among the amino acids which line the P9 pocket of I-Ad, only aspartic acidb57 and tryptophaneb61 represent differences from I-Ag7 (serineb57, tyrosineb61) [7,15]. Since residue b61 is aromatic in I-Ad as well as in I-Ag7, b57 appears to represent the main structural distinction between the P9 pockets of these allomorphs. In class II variants with aspartic acid at b57 this residue interacts with the conserved argininea76 via a salt bridge [7,16]. In other class II variants the positive charge of this arginine is uncomplemented, potentially free to interact with negatively charged peptide sidechains. This has led
to the hypothesis that the P9 of diabetes-associated sidechain charge as a feature distinguishing highaffinity ligands of the diabetes-associated I-Ag7 variant from those of the protective I-Ad. Thus, both allomorphs disfavour acidic and accept some basic P9 residues. Rather than charge, the data identify P9 sidechain size as the property which most consistently distinguish peptides with high affinity for I-Ag7 and I-Ad, respectively. Possibly, the absence of the b57-a76 interchain salt bridge in class II variants which confer susceptibility to diabetes may render their P9 pocket more capacious and therefore allow larger sidechains to interact with the class II molecule. In conclusion, the P9 sidechain contributes substantially to the overall affinity of I-Ad peptide ligands, and the specificity of this interaction ([5,9,14], this study) is distinct from that of I-Ag7 [10,11,17]. This supports the contention that the polymorphism at residue b57 influences diabetes susceptibility via P9-specific effects on the repertoires of self peptides presented to T cells.
Acknowledgements This work was supported by the Norwegian Cancer Society, the Norwegian Research Council, and Aakre’s Fund.
202
K. Bartnes et al. / Immunology Letters 70 (1999) 199–202
References [1] A.C. Avery, Z.-S. Zhao, A. Rodriguez, et al., Nature 376 (1995) 431 – 434. [2] S.M. Singer, R. Tisch, X.-D. Yang, H.O. McDevitt, Proc. Natl. Acad. Sci. USA 90 (1993) 9566–9570. [3] Z.-S. Zhao, F. Granucci, L. Yeh, P.A. Schaffer, H. Cantor, Science 279 (1998) 1344–1347. [4] K. Bartnes, K. Hannestad, Eur. J. Immunol. 21 (1991) 2365 – 2371. [5] K. Bartnes, F. Leon, J.P. Briand, P.J. Travers, K. Hannestad, Int. Immunol. 9 (1997) 1185–1193. [6] J.A. Todd, J.I. Bell, H.O. McDevitt, Nature 329 (1987) 599 – 604. [7] C.A. Scott, P.A. Peterson, L. Teyton, I.A. Wilson, Immunity 8 (1998) 319–329. [8] K. Bartnes, K. Hannestad, G. Guichard, J.P. Briand, Eur. J. Immunol. 27 (1997) 1387–1391. [9] A. Sette, S. Buus, E. Appella, et al., Proc. Natl. Acad. Sci USA
.
86 (1989) 3296 – 3300. [10] L.C. Harrison, M.C. Honeyman, S. Trembleau, et al., J. Exp. Med. 185 (1997) 1013 – 1021. [11] B. Reizis, M. Eisenstein, J. Bockova´, et al., Int. Immunol. 9 (1997) 43 – 51. [12] W.-H. Boehncke, T. Takeshita, C.D. Pendleton, et al., J. Immunol. 150 (1993) 331 – 341. [13] J. Hammer, C. Belunis, D. Bolin, et al., Proc. Natl. Acad. Sci. USA 91 (1994) 4456 – 4460. [14] R.D. England, M.C. Kullberg, J.L. Cornette, J.A. Berzofsky, J. Immunol. 155 (1995) 4295 – 4306. [15] E.A. Kabat, T.T. Wu, H.M. Perry, K.S. Gottesmann, C. Foeller, Sequences of Proteins of Immunological Interest, fifth ed, National Institutes of Health, Bethesda, 1991. [16] J.H. Brown, T.S. Jardetzky, J.C. Gorga, et al., Nature 364 (1993) 33 – 39. [17] E.-P. Reich, H. von Grafenstein, A. Barlow, K.E. Swenson, K. Williams, C.A. Janeway Jr, J. Immunol. 152 (1994) 2279–2288.
www.elsevier.com/locate/
The P9 peptide sidechain specificity of I-Ad Kristian Bartnes a,*, Xiaotong Li a, Jean Paul Briand b, Paul J. Travers c, Kristian Hannestad a a
Department of Immunology, Institute of Medical Biology, Uni6ersity of Tromsø, School of Medicine, N-9037 Tromsø, Norway b Institut de Biologie Moleculaire et Cellulaire, UPR 9021 CNRS, Strasbourg, France c Imperial Cancer Research Fund Unit for Structural Molecular Biology, Department of Crystallography, Birkbeck College, London WC1E 7HX, UK Received 28 July 1999; accepted 27 August 1999
Abstract The murine MHC class II variant I-Ad confers susceptibility to herpes simplex virus (HSV)-induced keratitis and relative protection against type 1 diabetes mellitus. The association to these autoimmune diseases appears to be largely determined by the peptide sidechain specificity of the P9 pocket, which we therefore have analyzed in detail. Assessment of T-cell responses and I-Ad binding capacity of position 446-substituted analogs of an IgG2a allotype b (IgG2ab) heavy chain peptide demonstrates that engagement of the P9 pocket is crucial for effective peptide presentation. Sidechain size rather than charge decides the capacity to engage the P9 pocket. Thus, small, uncharged sidechains are accepted, whereas acidic and aromatic amino acids as well as lysine and arginine are disfavored. The specificity of the P9 pocket of I-Ad (serineb57) is distinct from that of the diabetes-associated I-Ag7 (aspartic acidb57), supporting the contention that the polymorphism at residue b57 influences diabetes susceptibility via P9-specific effects on the repertoires of self peptides presented to T cells. Furthermore, the data rationalize the susceptibility to HSV-induced keratitis conferred by the a and the protection conferred by the b allotypes of the IgG2a heavy chain. Keratitogenic T cells, which cross-react with the viral UL6 protein and a corneal antigen, are silenced in IgG2ab mice because of antigenic mimicry with g2ab 435–451. Our finding that the lysine P9 residue of the corresponding g2aa allopeptide precludes high-affinity binding to I-Ad indicates that the susceptibility of IgG2aa mice reflects inefficient thymic presentation of autologous IgG2a and thus failure to purge the T-cell repertoire of the pathogenic clones. © 1999 Elsevier Science B.V. All rights reserved. Keywords: Diabetes mellitus; I-Ad; Keratitis; MHC class II; Motif; P9 pocket
1. Introduction The murine MHC class II allomorph I-Ad is the restriction element for T cells which mediate herpes simplex virus (HSV)-induced autoimmune keratitis [1] and confers relative protection against type 1 diabetes mellitus [2]. I-Ad-restricted keratitogenic T cells which recognize a corneal autoantigen crossreact with the HSV UL6 protein which mimics the IgG2a heavy chain allotype b (IgG2ab) peptide g2ab 435 – 451 [3]. IgG2ab * Correspondence author. Tel.: +47-776-26000; fax: +47-77628298. E-mail address: [email protected] (K. Bartnes)
mice constitutively present this g2ab epitope in complex with I-Ad [4] and are resistant to HSV-induced keratitis [1]. In contrast, IgG2aa mice are susceptible, most likely as a result of inefficient thymic presentation of the corresponding g2aa self peptide since the lysine which corresponds to relative position 9 (P9) in the I-Ad motif encompassed by the g2ab 438–446 sequence precludes high-affinity binding to I-Ad [5]. The specificity of the P9 pocket of the I-Ad peptide binding site appears to be crucial for the susceptibility to autoimmune diabetes mellitus as well, since the predisposition to develop diabetes conferred by I-A and HLA-DQ variants is strongly influenced by the polymorphism at residue b57 [6] which lines the P9 pocket [7]. Thus, I-Ad (aspartic
0165-2478/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 5 - 2 4 7 8 ( 9 9 ) 0 0 1 4 9 - 2
200
K. Bartnes et al. / Immunology Letters 70 (1999) 199–202
acidb57) confers protection when transgenically introduced in non-obese diabetic (NOD) mice (I-Ag7, serineb57) [2]. By assessing antigenicity and I-Ad binding capacity of single-residue substituted analogs of peptide g2ab 435–447, we have completed the analysis of the I-Ad P9 pocket specificity and compared it with that of I-Ag7.
2. Materials and methods The responses of the I-Ad-restricted T-cell hybridoma B5 [4] to a panel of single amino acid substituted analogs of the wildtype g2ab 435-447 (YFMYSKLRVQKST) allopeptide epitope were determined in a growth inhibition assay, as described [8]. Briefly, B5 cells (5×104 per well) were cultured together with irradiated (2000 rad) BALB/c spleen cells (5 ×105 per well) and peptide or medium alone for 18 h followed by a 6-h pulse with 3H-thymidine. After harvesting cultures onto glass fiber filter mats with a Micromate cell harvester (Packard Instrument, Meriden, CT), 3H-thymidine incorporation was determined with a Matrix96 Direct b Counter (Packard Instrument). The relative I-Ad-binding capacity of various peptide analogs was determined by flow cytometric assessment of their ability to compete with biotinylhexanoate-g2ab 436 – 451 for binding to I-Ad homozygous A20 cells [5], using an Epics Profile II flow cytometer equipped with a 488 nm argon laser (Coulter Electronics, Luton, UK). Peptide synthesis and purification has been described [5].
to I-Ad in many different registers [7] may complicate interpretation of direct binding data. Notably, however, the responses of the g2ab-reactive, I-Ad-restricted T-cell hybridoma B5 (Fig. 1, upper panel) correlated well with the relative binding capacities of the various analogs. Since the T-cell receptor generally accepts only a single binding register, it is unlikely that our analysis is confounded by multiple registers.
4. Discussion The present analysis demonstrates that the P9 pocket of I-Ad is less selective than that accommodating P4 [9], but more selective than pocket P1 [5]. Aromatic and acidic sidechains conferred a 1000-fold loss in antigenicity. Thus, the sidechain at P9 has a striking impact on the antigenic potency of peptides presented by I-Ad. Most likely, this exclusively reflects P9-mediated effects on the peptide/I-Ad interaction, since the P9 sidechain of hemagglutinin 126–138 (which is serine, as in the g2ab epitope) is buried within the I-Ad molecule [7] and thus inaccessible to
3. Results Serine446 (P9) of the g2ab 435 – 447 determinant was substituted with all natural amino acids that had not previously been analyzed except isoleucine and the labile cysteine, and the capacity of the various analogs to bind I-Ad and their antigenicity for a g2ab 435– 447/I-Ad-reactive T-cell hybridoma were determined (Fig. 1). Notably, via this approach we have recently arrived at a model for allele-specific peptide/I-Ad interactions [5] which is in agreement with subsequent crystallographic analysis [7]. The relative binding capacity varied 100-fold between the different analogs (Fig. 1, lower panel). Combined with previous direct binding analyses [5,9] the data demonstrate that I-Ad favors histidine and the small, uncharged sidechains of alanine, glycine, serine and threonine at P9, whereas arginine, lysine and acidic and aromatic amino acids preclude high-affinity binding (Table 1). The indications that the ovalbumin peptide 323 – 339 may bind
Fig. 1. Antigenicity (upper panel) and I-Ad binding capacity (lower panel) of analogs of the g2ab 435 – 447 peptide substituted at P9 (g2ab 446), determined by a T-cell hybridoma growth inhibition assay [8] and flow cytometric direct binding analysis [5], respectively. The data represent values relative to the wildtype g2ab 435 – 447 (serine446). Data from a representative experiment are shown in each panel.
K. Bartnes et al. / Immunology Letters 70 (1999) 199–202
201
Table 1 Compilation of P9 peptide sidechain specificities reported for I-Ad and I-Ag7a P9 residue
I-Ad binding Sette et al., 1989 [9]
gly ala val leu lys arg his asp glu asn glnb ser thr met phe tyr trp pro
2 3 2
I-Ag7 binding Bartnes et al., 1997 [5]
This study
3 3 1 1
1 3 1 1 2 3 3
3 2 1
1 1 2
Harrison et al., 1997 [10] 1 1 1 2 3 3 2 2 1 2 2 1 1 2 3 3 2 1
a Relative affinities are indicated by numbers from 1–3, where 3 represents the best binders. For I-Ad, the values are assigned to the various substitutions in accordance with [10]. For I-Ag7, the classification of Harrison et al. [10] is used with respect to well-tolerated, weakly tolerated, and nontolerated substitutions. Only studies in which IC50% values were obtained by direct binding T cell-independent analysis are included. b Although IC50% values have not been determined with I-Ad, there is evidance that gln is accepted [14].
the T-cell receptor. The preference for small P9 sidechains most likely reflects the fact that the corresponding pocket is relatively shallow [7]. Notably, thespecificity of peptide/class II MHC interactions is not solely determined by the ability of peptide sidechains to productively engage specificity pockets, as anchor substitutions can exert position-specific, dominant negative effects which substantially contribute to the overall peptide/class II affinity [12,13]. Accordingly, since I-Ad accepts glycine at P9 [5,9], the specificity at this position may predominantly reflect repulsion of large P9 sidechains. However, this must be qualified by the fact that also the relatively large histidine allows good binding ([14], this study). Among the amino acids which line the P9 pocket of I-Ad, only aspartic acidb57 and tryptophaneb61 represent differences from I-Ag7 (serineb57, tyrosineb61) [7,15]. Since residue b61 is aromatic in I-Ad as well as in I-Ag7, b57 appears to represent the main structural distinction between the P9 pockets of these allomorphs. In class II variants with aspartic acid at b57 this residue interacts with the conserved argininea76 via a salt bridge [7,16]. In other class II variants the positive charge of this arginine is uncomplemented, potentially free to interact with negatively charged peptide sidechains. This has led
to the hypothesis that the P9 of diabetes-associated sidechain charge as a feature distinguishing highaffinity ligands of the diabetes-associated I-Ag7 variant from those of the protective I-Ad. Thus, both allomorphs disfavour acidic and accept some basic P9 residues. Rather than charge, the data identify P9 sidechain size as the property which most consistently distinguish peptides with high affinity for I-Ag7 and I-Ad, respectively. Possibly, the absence of the b57-a76 interchain salt bridge in class II variants which confer susceptibility to diabetes may render their P9 pocket more capacious and therefore allow larger sidechains to interact with the class II molecule. In conclusion, the P9 sidechain contributes substantially to the overall affinity of I-Ad peptide ligands, and the specificity of this interaction ([5,9,14], this study) is distinct from that of I-Ag7 [10,11,17]. This supports the contention that the polymorphism at residue b57 influences diabetes susceptibility via P9-specific effects on the repertoires of self peptides presented to T cells.
Acknowledgements This work was supported by the Norwegian Cancer Society, the Norwegian Research Council, and Aakre’s Fund.
202
K. Bartnes et al. / Immunology Letters 70 (1999) 199–202
References [1] A.C. Avery, Z.-S. Zhao, A. Rodriguez, et al., Nature 376 (1995) 431 – 434. [2] S.M. Singer, R. Tisch, X.-D. Yang, H.O. McDevitt, Proc. Natl. Acad. Sci. USA 90 (1993) 9566–9570. [3] Z.-S. Zhao, F. Granucci, L. Yeh, P.A. Schaffer, H. Cantor, Science 279 (1998) 1344–1347. [4] K. Bartnes, K. Hannestad, Eur. J. Immunol. 21 (1991) 2365 – 2371. [5] K. Bartnes, F. Leon, J.P. Briand, P.J. Travers, K. Hannestad, Int. Immunol. 9 (1997) 1185–1193. [6] J.A. Todd, J.I. Bell, H.O. McDevitt, Nature 329 (1987) 599 – 604. [7] C.A. Scott, P.A. Peterson, L. Teyton, I.A. Wilson, Immunity 8 (1998) 319–329. [8] K. Bartnes, K. Hannestad, G. Guichard, J.P. Briand, Eur. J. Immunol. 27 (1997) 1387–1391. [9] A. Sette, S. Buus, E. Appella, et al., Proc. Natl. Acad. Sci USA
.
86 (1989) 3296 – 3300. [10] L.C. Harrison, M.C. Honeyman, S. Trembleau, et al., J. Exp. Med. 185 (1997) 1013 – 1021. [11] B. Reizis, M. Eisenstein, J. Bockova´, et al., Int. Immunol. 9 (1997) 43 – 51. [12] W.-H. Boehncke, T. Takeshita, C.D. Pendleton, et al., J. Immunol. 150 (1993) 331 – 341. [13] J. Hammer, C. Belunis, D. Bolin, et al., Proc. Natl. Acad. Sci. USA 91 (1994) 4456 – 4460. [14] R.D. England, M.C. Kullberg, J.L. Cornette, J.A. Berzofsky, J. Immunol. 155 (1995) 4295 – 4306. [15] E.A. Kabat, T.T. Wu, H.M. Perry, K.S. Gottesmann, C. Foeller, Sequences of Proteins of Immunological Interest, fifth ed, National Institutes of Health, Bethesda, 1991. [16] J.H. Brown, T.S. Jardetzky, J.C. Gorga, et al., Nature 364 (1993) 33 – 39. [17] E.-P. Reich, H. von Grafenstein, A. Barlow, K.E. Swenson, K. Williams, C.A. Janeway Jr, J. Immunol. 152 (1994) 2279–2288.