A wealth of data has accumulated on the structure of mouse MHC class I (MHCI) molecules encoded by the H-2b and H-2d haplotypes. In contrast, there is a dearth of structural data regarding H-2k-encoded molecules. Therefore, the structures of H-2Kk complexed to an octameric peptide from influenza A virus (HA259–266) and to a nonameric peptide from SV40 (SV40560–568) have been determined by x-ray crystallography at 2.5 and 3.0 Å resolutions, respectively. The structure of the H-2Kk-HA259–266 complex reveals that residues located on the floor of the peptide-binding groove contact directly the backbone of the octameric peptide and force it to lie deep within the H-2Kk groove. This unprecedented mode of peptide binding occurs despite the presence of bulky residues in the middle of the floor of the H-2Kk peptide-binding groove. As a result, the Cα atoms of peptide residues P5 and P6 are more buried than the corresponding residues of H-2Kb-bound octapeptides, making them even less accessible to TCR contact. When bound to H-2Kk, the backbone of the SV40560–568 nonapeptide bulges out of the peptide-binding groove and adopts a conformation reminiscent of that observed for peptides bound to H-2Ld. This structural convergence occurs despite the totally different architectures of the H-2Ld and H-2Kk peptide-binding grooves. Therefore, these two H-2Kk-peptide complexes provide insights into the mechanisms through which MHC polymorphism outside primary peptide pockets influences the conformation of the bound peptides and have implications for TCR recognition and vaccine design.

Major histocompatibility complex-encoded molecules govern adaptive immune responses by presenting antigenic peptides to TCRs. MHC class Ia (MHCI)4 molecules are specifically devoted to present small peptide fragments (typically eight or nine amino acids in length) mostly generated as byproducts of aberrant protein synthesis (1). Approximately 35 residues contribute to the solvent-accessible surface of the peptide-binding groove, and ∼20 of them are polymorphic. Hydrogen bonds between conserved side chains of the peptide-binding groove and main chain polar atoms from the peptide provide a general, sequence-independent mechanism for binding many different peptides. These conserved hydrogen bonds are confined to both ends of the peptide-binding groove and involve peptide residues found at position 1 (P1), and at both the penultimate P(Ω-1) and C-terminal PΩ positions (numbering according to Madden (2)). This conserved bonding fixes the positions of the peptide N and C termini and forces the central part of peptides longer than eight amino acids to bulge out of the groove, to zigzag within the groove, or, more rarely, to extend from the termini of the groove (3).

Peptides capable of binding with high affinity to a given MHCI allele share two or three primary anchor residues, the side chains of which are specifically accommodated into MHC cavities or pockets. The polymorphic residues that line the pockets determine groove specificity and restrict the repertoire of peptides capable of binding to a given MHCI allele. Six pockets, denoted A through F, have been defined in MHCI peptide-binding grooves (2). In every MHCI allele, the side chain of the peptide residue found at the PΩ position constitutes a primary anchor and is deeply buried in the F pocket. The position of the second primary anchor is more variable and usually occurs at P2, P3, or P5.

Most studies on mouse MHCI molecules have focused on the H-2Kb allele and, to a lesser extent, on H-2Db, resulting in the elucidation of their structures in complex with various peptides (4, 5, 6, 7, 8, 9, 10) and with several TCRs (11, 12, 13, 14, 15). More limited structural data have been obtained for H-2Dd (16, 17) and H-2Ld (18, 19). Considering that no structural information is available yet for MHCI molecules encoded by the H-2k haplotype, we crystallized and solved the structure of the H-2Kk molecule in complex with an octapeptide derived from influenza A virus (HA259–266; FEANGNLI) (20) and with a nonapeptide derived from SV40 (SV40560–568; SEFLLEKRI) (21). These two H-2Kk-peptide complexes provide insights on the mechanisms by which MHC polymorphism outside primary peptide pockets influences the course of the bound peptides and have important implications for TCR recognition.

These procedures have been previously described for H-2Kk com-plexes (22).

The structure of the H-2Kk-HA259–266 complex was determined by molecular replacement with the Amore program (23). When used as a search model, a peptide-stripped H-2Kb molecule (PDB ID code 1LEG) failed to provide a solution when considered as a whole unit. To reach a solution, we had to separate it into the α1/α2 and α3/β2m domain sets. One single complex was identified in the asymmetric unit, and the HA259–266 peptide was then built in the resulting Fo-Fc electron density map (Fig. 1,D) using the Turbo Frodo program (24). Refinement was conducted using the Refmac 5.0 program (ccp4 package; 〈www.ccp4.ac.uk〉) with Translation Libration Screw refinement. Crystallographic statistics for each complex have been reported previously and are detailed in Table I (22). The Rfactor and Rfree values for the refined model are 21.2 and 24.5, respectively. Strong 2Fo-Fc densities were observed, in particular in a cavity between the α1/α2 and α3/β2m domains. These densities remained, however, noninterpretable and were attributed to water molecules for the sake of refinement. Furthermore, the region that links the α2 and α3 domains (aa 176–181) could not be built due to the lack of electron density.

FIGURE 1.

Binding of the octapeptide HA259–266 in the groove of H-2Kk. A, The H-2Kk binding groove secondary structure is represented as blue spirals (α1 helix) and arrows (β strands), and the α2 helix was removed for the sake of clarity. The HA259–266 peptide and Tyr99, Arg97, Tyr116, and Glu114 of H-2Kk are shown in ball-and-stick format, with carbon atoms in yellow, oxygen atoms in red, and nitrogen atoms in green. The trace of the HA259–266 peptide is compared with that of the dEV8 octapeptide backbone (in magenta) bound to H-2Kb. Lys66, Asn70, Ser73, Ser97, Val99, and Tyr116 of H-2Kb are shown in light magenta. B and C, Comparison of the HA259–266 peptide conformation to that of octapeptides bound to H-2Kb. The peptides are represented as Cα traces, except at P6. The orientation in B is from the side (the floor of the peptide-binding groove being perpendicular to the plan of the page), and that in C is from the top (the floor of the peptide-binding groove being parallel to the plan of the page). HA259–266 is in red, Dev8 in light violet, GNYSF in green, pBM1 in cyan, OVA in blue, pKB1 in orange, and vesicular stomatitis virus 8 in pink. D, 2Fo-Fc electron density map of HA259–266 after molecular replacement, contoured at 1.5ς.

FIGURE 1.

Binding of the octapeptide HA259–266 in the groove of H-2Kk. A, The H-2Kk binding groove secondary structure is represented as blue spirals (α1 helix) and arrows (β strands), and the α2 helix was removed for the sake of clarity. The HA259–266 peptide and Tyr99, Arg97, Tyr116, and Glu114 of H-2Kk are shown in ball-and-stick format, with carbon atoms in yellow, oxygen atoms in red, and nitrogen atoms in green. The trace of the HA259–266 peptide is compared with that of the dEV8 octapeptide backbone (in magenta) bound to H-2Kb. Lys66, Asn70, Ser73, Ser97, Val99, and Tyr116 of H-2Kb are shown in light magenta. B and C, Comparison of the HA259–266 peptide conformation to that of octapeptides bound to H-2Kb. The peptides are represented as Cα traces, except at P6. The orientation in B is from the side (the floor of the peptide-binding groove being perpendicular to the plan of the page), and that in C is from the top (the floor of the peptide-binding groove being parallel to the plan of the page). HA259–266 is in red, Dev8 in light violet, GNYSF in green, pBM1 in cyan, OVA in blue, pKB1 in orange, and vesicular stomatitis virus 8 in pink. D, 2Fo-Fc electron density map of HA259–266 after molecular replacement, contoured at 1.5ς.

Close modal
Table I.

Data collection and refinement statistics

H-2Kk/HA(259–266)H-2Kk/SV40(560–568)
Data collection   
 Resolution (Å) 15.0–2.5 15–3.0 
 Space group P3221 P2 
 Cell dimension 111.76; 111.76; 109.42 85.13; 72.63; 88.79; β = 111.24 
 Number molecules/asymetric unit 
 Number of reflections 26,053 18,302 
 I/ς (I) 8.9 4.6 
 Completeness (%) 99.3 90.2 
Rsym (%) 6.9 (33.8) 15.8 (39.7) 
   
Refinement statistics   
Rfactor 21.2 (25.0)a 23.1 (26.7)b 
Rfree 24.5 (33.1)a 30.9 (37.4)b 
 Total number of atoms 3,314 6,480 
  from protein 3,132 6,334 
  from water 182 146 
 Rmsd from ideal geometry   
 Bond length (Å) 0.009 0.010 
 Angle angle (°) 1.16 1.21 
 Ramachandran plot   
 Most-favored regions (%) 91.4 82.5 
 Allowed regions (%) 8.0 14.8 
 Generously allowed regions (%) 0.6 2.7 
 B factors   
 α-chain 29.8 22.6 
 β-chain 27.7 43.8 
 Peptide 27.2 13.7 
 Average water molecules 47.8 11.4 
H-2Kk/HA(259–266)H-2Kk/SV40(560–568)
Data collection   
 Resolution (Å) 15.0–2.5 15–3.0 
 Space group P3221 P2 
 Cell dimension 111.76; 111.76; 109.42 85.13; 72.63; 88.79; β = 111.24 
 Number molecules/asymetric unit 
 Number of reflections 26,053 18,302 
 I/ς (I) 8.9 4.6 
 Completeness (%) 99.3 90.2 
Rsym (%) 6.9 (33.8) 15.8 (39.7) 
   
Refinement statistics   
Rfactor 21.2 (25.0)a 23.1 (26.7)b 
Rfree 24.5 (33.1)a 30.9 (37.4)b 
 Total number of atoms 3,314 6,480 
  from protein 3,132 6,334 
  from water 182 146 
 Rmsd from ideal geometry   
 Bond length (Å) 0.009 0.010 
 Angle angle (°) 1.16 1.21 
 Ramachandran plot   
 Most-favored regions (%) 91.4 82.5 
 Allowed regions (%) 8.0 14.8 
 Generously allowed regions (%) 0.6 2.7 
 B factors   
 α-chain 29.8 22.6 
 β-chain 27.7 43.8 
 Peptide 27.2 13.7 
 Average water molecules 47.8 11.4 
a

R factor and R free in the last shell (2.5–2.56).

b

R factor and R free in the last shell (3.0–3.075).

The structure of the H-2Kk-SV40560–568 complex was solved by molecular replacement using the structure of H-2Kk-HA259–266, the complex being treated as α1/α2 and α3/β2 independent parts. The nonapeptide was built unambiguously in the Fo-Fc density map (Fig. 3,B). The refinement was achieved using the CNS (25) and Refmac 5.0 programs. The final electron density map was of good quality for the α1, α2, and α3 domains and for the peptide, whereas β2m was poorly defined, probably due to internal mobility. The final Rfactor and Rfree values are 23.1 and 30.9, respectively (Table I). In contrast to that observed for the H-2Kk-HA259–266 complex (see above), no electron density was present in the cavity between the α1/α2 and α3/β2m domains. The relative positions of the α1/α2 and α3/α2m blocks differed in the two H-2Kk-peptide complexes and diverged from those observed in MHCI structures reported to date. The differences in the relative positions of the α1/α2 and α3/β2m domains are common and are probably due to packing differences in the crystals. The two H-2Kk complexes also showed small, but significant, deviations (range, 0.65–1.15 Å) in two strands of the β sheet (residues 95–98 of strand S4 and residues 116–118 of strand S5). The present analysis is limited to the structure of the α1/α2 domains of both H-2Kk-peptide complexes.

FIGURE 3.

Binding of the SV40560–568 nonapeptide into the groove of H-2Kk. A, The representation is the same as in Fig. 1 A. The trace of SV40560–568 is compared with that of the octapeptide HA259–266 (in magenta). B, 2Fo-Fc electron density map of SV40560–568 after molecular replacement, contoured at 1.5ς. C, Comparison of the Cα trace of SV40560–568 in H-2Kk to that of various nonapeptides. The orientation is from the side. For H-2Kk- and H-2Ld-bound peptides the P4 and P6 side chains are shown. Nonapeptides are colored as follows: SV40560–568 in H-2Kk in red, p29 in H-2Ld in orange, SEV9 in H-2Kb in green, Tax in HLA-A2 in dark blue, Ebna-3 in HLA-B8 in dark pink, DP46–54 in HLA-B44 in light blue, Km1 in HLA-B51 in yellow, and Gag in HLA-B53 in light violet.

FIGURE 3.

Binding of the SV40560–568 nonapeptide into the groove of H-2Kk. A, The representation is the same as in Fig. 1 A. The trace of SV40560–568 is compared with that of the octapeptide HA259–266 (in magenta). B, 2Fo-Fc electron density map of SV40560–568 after molecular replacement, contoured at 1.5ς. C, Comparison of the Cα trace of SV40560–568 in H-2Kk to that of various nonapeptides. The orientation is from the side. For H-2Kk- and H-2Ld-bound peptides the P4 and P6 side chains are shown. Nonapeptides are colored as follows: SV40560–568 in H-2Kk in red, p29 in H-2Ld in orange, SEV9 in H-2Kb in green, Tax in HLA-A2 in dark blue, Ebna-3 in HLA-B8 in dark pink, DP46–54 in HLA-B44 in light blue, Km1 in HLA-B51 in yellow, and Gag in HLA-B53 in light violet.

Close modal

The structures of the two H-2Kk-peptide complexes were compared with human (HLA-A2, HLA-B8, HLA-B27, HLA-B35, HLA-B44, HLA-B51, and HLA-B53) and mouse (H-2Db, H-2Dd, H-2Kb, and H-2Ld) MHCI structures. The amino acid sequences of these MHCI molecules were aligned (data not shown). A first round of superimposition of each molecule on the H-2Kk α1/α2 domains (residues 1–175) was performed using the Cα atoms of amino acids found at positions 50, 100, and 150. The superimposition was then refined using all the Cα pairs distant by less than a given cutoff distance. In our strategy, the initial 1-Å cutoff distance was decreased to 0.5 Å in a second step. After this last calculation step, the quality of the superimposition can be estimated by the number of Cα pairs distant of <0.5 Å. For instance, when H-2Kk was compared with H-2Kb, of the 175 Cα pairs that were considered, 90 of them were distant by <0.5 Å.

The structures of the H-2Kk-HA259–266 and H-2Kk-SV40560–568 complexes were determined by molecular replacement at 2.5 and 3.0 Å resolution, respectively. In both complexes, H-2Kk showed the typical MHCI canonical fold. Superimposition of the α1/α2 domains of H-2Kk-HA259–266 and of H-2Kk-SV40560–568 gives average root mean square deviations of 0.57 and 0.40 Å for 162 and 108 Cα atom positions, respectively. The α1 helices were well superimposed, and the α2 helices were shifted by 0.5–1 Å. Structural divergences between the two H-2Kk complexes were noted at residues 39–41 (corresponding to a solvent-exposed loop connecting strands S3 and S4), residues 53–58 (corresponding to the end of a 310 helix before the α1 helix), and residues 103–106 (corresponding to a loop connecting strands S4 and S5), with maximal deviations of 2.71, 2.64, and 2.96 Å at residues 41, 54, and 105, respectively.

The P1 residue at the N-terminal end of HA259–266 forms hydrogen bonds with residues Tyr7, Tyr159, and Tyr171 of H-2Kk. At the C-terminal end, the P(Ω-1) residue binds to H-2Kk Trp147, and the PΩ residue binds to H-2Kk Asn77, Tyr84, and Thr143 (Table II). This network of interactions is highly conserved among human and mouse MHCI molecules. The overall conformation adopted by the backbone of the HA259–266 octapeptide is reminiscent of that of H-2Kb bound-octapeptides (Fig. 1, A and B). The Cα trace of HA259–266 differs slightly from H-2Kb-bound peptides at P2, P5, and P6. For instance, the Cα atom of P2 is higher up by ∼1Å, whereas the P5 and P6 Cα atoms are more buried than those of H-2Kb-bound octapeptides. The protrusion of the Cα atom at P2 of H-2Kk might be explained by the presence of bulky residues (His9, Tyr45, and Tyr99) that decreases the volume of the B pocket (see below), and by the need to accommodate the long side chain of the Glu residue found at P2 of HA259–266).

Table II.

Interactions between H-2Kk residues and the octa/nonapeptides

Main Chain InteractionsSide-Chain Interactions
OctapeptideH-2KkNonapeptideH-2KkOctapeptideH-2KkNonapeptideH-2Kk
FEANGNLISEFLLGKRI
ResidueAtomResidueAtomResidueAtomResidueAtomResidueAtomResidueAtomResidueAtomResidueAtom
PheP1 Tyr7 OH SerP1 Tyr7 OH     SerP1 OG Trp167 NE1 
 Tyr171 OH   Tyr171 OH GluP2 OE1 His9 NE2 GluP2 OE1 Ser24 OG 
 Tyr159 OH  Tyr159 OH   Tyr99 OH   Tyr45 OH 
GluP2 Asn63 OD1 GluP2 Asn63 OD1  OE2 Ser24 OG  OE2 Tyr99 OH 
AlaP3 Tyr99 OH PheP3 Arg97a NH1   Tyr45 OH   His9 NE2 
 Asn70 ND2      OE1 Asn70a ND2     
 Arg97 NH1     AsnP4 OD1 Ile66a     
AsnP4 Arg97 NH2     AsnP6 ND2 Asp156 OD2 LysP7 NZ Glu114 OE1 
 Asp156b OD1     (A) ND1 Glu114b OE1   Tyr116 OH 
GlyP5 Asn70 OD1 LeuP5 Asn70 OD1  ND2 Tyr116b OH     
AsnP6 Tyr116 OH      OD1 Asp156c OD1     
 Asn77 ND2     AsnP6  Asp152 OD1     
LeuP7 Trp147 NE1 ArgP8 Trp147 NE1 (B)   OD2     
IleP8 Asn77 OD1 IleP9 Asn77 OD1         
 OXT Tyr84 OH  OXT Tyr84 OH         
 OXT Thr143 OG1   Thr143 OH         
Main Chain InteractionsSide-Chain Interactions
OctapeptideH-2KkNonapeptideH-2KkOctapeptideH-2KkNonapeptideH-2Kk
FEANGNLISEFLLGKRI
ResidueAtomResidueAtomResidueAtomResidueAtomResidueAtomResidueAtomResidueAtomResidueAtom
PheP1 Tyr7 OH SerP1 Tyr7 OH     SerP1 OG Trp167 NE1 
 Tyr171 OH   Tyr171 OH GluP2 OE1 His9 NE2 GluP2 OE1 Ser24 OG 
 Tyr159 OH  Tyr159 OH   Tyr99 OH   Tyr45 OH 
GluP2 Asn63 OD1 GluP2 Asn63 OD1  OE2 Ser24 OG  OE2 Tyr99 OH 
AlaP3 Tyr99 OH PheP3 Arg97a NH1   Tyr45 OH   His9 NE2 
 Asn70 ND2      OE1 Asn70a ND2     
 Arg97 NH1     AsnP4 OD1 Ile66a     
AsnP4 Arg97 NH2     AsnP6 ND2 Asp156 OD2 LysP7 NZ Glu114 OE1 
 Asp156b OD1     (A) ND1 Glu114b OE1   Tyr116 OH 
GlyP5 Asn70 OD1 LeuP5 Asn70 OD1  ND2 Tyr116b OH     
AsnP6 Tyr116 OH      OD1 Asp156c OD1     
 Asn77 ND2     AsnP6  Asp152 OD1     
LeuP7 Trp147 NE1 ArgP8 Trp147 NE1 (B)   OD2     
IleP8 Asn77 OD1 IleP9 Asn77 OD1         
 OXT Tyr84 OH  OXT Tyr84 OH         
 OXT Thr143 OG1   Thr143 OH         
a

Indirect interaction.

b

Water-mediated interaction involving the same water molecule (Asp156).

c

Indirect interactions involving the same water molecule.

The deep course adopted by the central region of HA259–266 is in part determined by its interaction with three residues (Arg97, Tyr99, and Tyr116) that belong to the β sheet that constitutes the floor of the H-2Kk peptide-binding groove. As observed in H-2Ld and in most HLA class I molecules, Tyr99 of H-2Kk is hydrogen bonded to the amide nitrogen found at P3 of HA259–266 (Fig. 1,A). The Arg97 residue lying directly underneath the central region of the peptide is of particular importance. Its NH1 and NH2 amines interact directly with the P3 and P4 carbonyl oxygens, anchoring both of them tightly in the H-2Kk groove (Fig. 1,A and Table II). The conformation of the side chain of Arg97, and thus of the contacting amines, is tightly maintained by two hydrogen bonds with H-2Kk Glu114 (Fig. 1,A), and by a π cation interaction involving the Arg97 guanidinium group and the neighboring Tyr99 aromatic ring. The main chain of residue P6 interacts with H-2Kk through two hydrogen bonds involving both Asp77 and Tyr116 (Fig. 1,A and Table II). This last feature is rather unusual, because in the vast majority of MHCI-octamer complexes described to date, the main chain of P6 either does not contact the MHC or indirectly interacts with it through a water-mediated hydrogen bond. Interestingly, the lack of contact between H-2Kb and the P6 main chain of the octapeptides correlates with the horizontal (Fig. 1,B) or vertical (Fig. 1 C) positional variability manifested by the Cα atom found at P6 of H-2Kb-octamer complexes. An absence of contact with the MHC is also found for the P6 main chain of peptides bound to H-2Db, thereby favoring some conformational malleability, and in HLA-A2-peptide complexes, where Cα displacements can even occur across residues P4 to P6 (26). Therefore, in the H-2Kk-HA259–266 complex, the bidentate anchoring of the P6 main chain through both Asp77 and Tyr116 probably prevents any wobbling of the P6 Cα atom and constitutes a distinctive attribute among the MHCI molecules studied to date.

The B pocket of H-2Kk has a hydrophobic entrance made of residues 7 and 66 and a hydrophilic bottom made of residues 9, 24, 45, and 99. The apolar stalk of GluP2 packs against the Tyr7 aromatic moiety and is close (4 Å) to Ile66. Accommodation of the negative charge of P2Glu is achieved by four direct hydrogen bonds involving His9 and Tyr99 on one side of the pocket and Ser24 and Tyr45 on the other side (Fig. 2). The indirect interaction between Asn70 and P2Glu found in the H-2Kk-HA259–266 complex is absent in the H-2Kk-SV40560–568 complex. In Fig. 2, the H-2Kk B pocket is compared with that of HLA-B44, because both show the same specificity for Glu at P2 (27). HLA-B44 uses the same residues as H-2Kk to contact the side chain of P2Glu. In particular, the hydrophobic entrance is similar. Moreover, Tyr99, which is conserved in human MHCI molecules, is used by both H-2Kk and HLA-B44 to contact the P2 carboxylate through a hydrogen bond. Although Ser24 of H-2Kk makes a direct hydrogen bond, the one involving Thr24 of HLA-B44 is indirect. The contact between P2 Glu and residue 45 is either a hydrogen bond (involving Tyr in H-2Kk) or a salt bridge (involving Lys in HLA-B44). Residue 9 is a His in H-2Kk, and its amine (NE2) achieves a direct hydrogen bond with the carboxylate of Glu. In HLA-B44, it is a Tyr that makes a hydrogen bond with the carboxylate of Glu. Therefore, the structural environment of the B pocket is largely conserved between mouse H-2Kk and human HLA-B44 and is consistent with their preference for a long, negatively charged side chain as found in Glu.

FIGURE 2.

Architecture of the B pocket of H-2Kk and comparison with that of HLA-B44. The Glu residue found at P2 of the HA259–266 peptide and the His9, Ser24, Tyr45, and Tyr99 residues of H-2Kk are shown in ball-and-stick format, with carbon atoms in yellow, oxygen atoms in red, and nitrogen atoms in blue. Hydrogen bonds between H-2Kk and the side chain of GluP2 are represented as blue dotted lines. For the sake of clarity, the hydrophobic interactions and an indirect interaction between Asn70 and P2 Glu are omitted. For HLA-B44, the residues of the B pocket that contact Glu found at P2 of the DPα46–54-bound peptide are specified in parentheses and colored in pink. Hydrogen bonds between GluP2 and HLA-B44 are shown in pink dotted lines (the one involving Thr24 occurs through one molecule of water).

FIGURE 2.

Architecture of the B pocket of H-2Kk and comparison with that of HLA-B44. The Glu residue found at P2 of the HA259–266 peptide and the His9, Ser24, Tyr45, and Tyr99 residues of H-2Kk are shown in ball-and-stick format, with carbon atoms in yellow, oxygen atoms in red, and nitrogen atoms in blue. Hydrogen bonds between H-2Kk and the side chain of GluP2 are represented as blue dotted lines. For the sake of clarity, the hydrophobic interactions and an indirect interaction between Asn70 and P2 Glu are omitted. For HLA-B44, the residues of the B pocket that contact Glu found at P2 of the DPα46–54-bound peptide are specified in parentheses and colored in pink. Hydrogen bonds between GluP2 and HLA-B44 are shown in pink dotted lines (the one involving Thr24 occurs through one molecule of water).

Close modal

Although both H-2Kk and H-2Kb bind predominantly octapeptides, they are also capable of accommodating nonapeptides (see 〈http://syfpeithi.bmi-heidelberg.com〉). Studies of the binding of a series of nonapeptides to H-2Kb revealed that no canonical binding mode exists, except for the conformational constraint imposed by the presence of a central primary anchor at P5 or P6, that is accommodated by the deep pocket that characterizes the groove of H-2Kb (4, 6, 8). Considering that H-2Kk lacks such a deep recess, we analyzed next how it accommodates the SV40560–568 nonameric peptide (Fig. 3,A). The quality of the electron density map obtained for the H-2Kk-SV40560–568 complex enabled the unambiguous building of the nonapeptide (Fig. 3,B). Residues at P1, P2, and PΩ of both SV40560–568 and HA259–266 contact H-2Kk in a similar way (Table II). In contrast, the Cα trace of the central part of SV40560–568 and HA259–266 differs markedly (Fig. 3,A). Compared with the HA259–266 octapeptide, the extra residue present in the SV40560–568 nonapeptide is accommodated through a prominent bulge in its central part. Comparison of SV40560–568 with nonapeptides bound to human and mouse MHCI molecules shows that it adopts a Cα trace reminiscent of that of peptides bound to H-2Ld (Fig. 3,C) (18, 19). This structural convergence occurs despite the totally different architectures of the H-2Kk and H-2Ld binding grooves, the latter being typified by aromatic residues at positions 73, 97, 155, and 156. Accommodating the additional amino acid of SV40560–568 distorts the dense network of hydrogen bond interactions noted with the octamer backbone (Table II). For instance, no interaction occurs between the main chain of PheP3 and Tyr99. Moreover, the two hydrogen bonds observed between Arg97 and the main chain of AlaP3 and AsnP4, in the case of H-2Kk HA259–266 are replaced by one water-mediated hydrogen bond involving PheP3. Finally, the interaction present between Tyr116 and HA259–266 is also lost. This differs from HLA-A2, HLA B8, HLA-B51, and H-2Kb, which all use Tyr116 to make an indirect hydrogen bond with the P7 carbonyl oxygen of nonapeptides. Therefore, the binding of the SV40560–568 nonamer to H-2Kk involves fewer hydrogen bonds than that of the HA259–266 octamer.

The quality of the electron density map enabled us to build unambiguously the HA259–266 octapeptide except at P6 where the asparagine shows a double conformation (Fig. 1,D and Table II). In conformer A, the side chain of AsnP6 is clamped by two hydrogen bonds involving Asp156 and by two indirect bonds with Asp114 and Tyr116. As a consequence, AsnP6 is buried within the cleft and oriented toward the α2 helix. In conformer B, the side chain of AsnP6 is hydrogen-bonded to Asp152 and points outside the groove. Superimposition of HA259–266 with H-2Kb-bound octapeptides shows that conformer B of AsnP6 lies deeper in the groove by 1–2.2 Å (for the Cα atom) and 2–5 Å (for the apex of the side chain; Fig. 1 B).

The degree of solvent exposure of HA259–266 bound to H-2Kk and of several octapeptides bound to H-2Kb was calculated for each position using the Turbo-Frodo program (Fig. 4). Peptide side chains at positions 2, 3, 5, and 8 are deeply buried in both H-2Kk and H-2Kb, whereas side chains at P4, P6, and P7 are pointing up and show variable degrees of solvent exposure (P4, 53–104 Å2; P6, 16–148 Å2; P7, 22–92 Å2). Among the P6 residues examined, AsnP6 of HA259–266 is the least accessible. Thus in the H-2Kk-HA259–266 complex, there are only two residues, AsnP4 and LeuP7, that remain available for TCR contact.

FIGURE 4.

Solvent exposure for HA259–266 octapeptide in complex with H-2Kk and for H-2Kb-bound octapeptides. The y-axis represents the accessible surface in Å2 for each residue. The x-axis refers to the residue number of each peptide displayed.

FIGURE 4.

Solvent exposure for HA259–266 octapeptide in complex with H-2Kk and for H-2Kb-bound octapeptides. The y-axis represents the accessible surface in Å2 for each residue. The x-axis refers to the residue number of each peptide displayed.

Close modal

In the case of the SV40560–568 nonamer bound to H-2Kk, the side chains found at P1, P2, P5, and P7 of are not available for TCR contact, as shown by their low solvent exposure (Table III). In contrast, residues P4, P6, and P8 of SV40560–568 are exposed to solvent. Moreover, the side chain of GluP6 of SV40560–568 shows a similar exposure as the TyrP6 of the H-2Kb-bound, Dev8 octapeptide and the IleP6 of the H-2Ld-bound, p29 nonapeptide (Table III and Fig. 3 C). Therefore, the prominent accessibility of GluP6 of SV40560–568 contrasts with the almost buried conformation of AsnP6 of HA259–266.

Table III.

Solvent accessibility of peptide residues in the H-2Kk-HA(259–266) and H-2Kk-SV40(560–568) complexesa

Residue of HA(259–266)Solvent Accessible Surface (Å2)Residue of SV40(560–568)Solvent Accessible Surface (Å2)
PheP1 32 SerP1 
GluP2 GluP2 
AlaP3 10 PheP3 10 
AsnP4 104 LeuP4 78 
GlyP5 LeuP5 
  GluP6 114 
AsnP6 16/25 LysP7 
LeuP7 92/82 ArgP8 108 
IleP8 IleP9 
Residue of HA(259–266)Solvent Accessible Surface (Å2)Residue of SV40(560–568)Solvent Accessible Surface (Å2)
PheP1 32 SerP1 
GluP2 GluP2 
AlaP3 10 PheP3 10 
AsnP4 104 LeuP4 78 
GlyP5 LeuP5 
  GluP6 114 
AsnP6 16/25 LysP7 
LeuP7 92/82 ArgP8 108 
IleP8 IleP9 
a

The values were determined using the program Turbo-Frodo. The water molecules were omitted for the calculation. For positions P6 and P7 of HA(259–266), the two values correspond to conformers A and B.

The middle section of H-2Kb-bound peptides is deeply embedded in the peptide-binding groove as a consequence of the deep pocket that lies in the midportion of the H-2Kb peptide-binding groove and accommodates the primary anchor residue found at P5. In contrast, the corresponding pocket of H-2Kk is lined by residues bulkier than those found in H-2Kb. Despite this marked difference in midgroove architecture that should result in different peptide conformations, the central part of the HA259–266 octameric peptide also adopts a flat conformation that runs even deeper than that of H-2Kb-bound octapeptides. This unexpected finding is due to the binding of the main chain of the octamer by residues from the floor of the H-2Kk groove, in particular Arg97. The structure of H-2Kk-HA259–266 highlights the fact that bulky residues belonging to the β sheet of an MHCI peptide-binding groove can be used to anchor the central region of an octapeptide main chain. This is an unprecedented observation that differs, for instance, from H-2Kb-octamer complexes, where the central region of the peptide backbone is contacted exclusively by the side chains of amino acids belonging to α-helices. The contributions of Arg97, Tyr99, and Tyr116 to peptide binding have already been evoked for HLA-B35 (28) and HLA-B2705 (29). In both these instances, however, hydrogen bonds were made with the side chains and not the main chain of the bound peptides.

The crucial role played by residue Arg97 in the architecture of the H-2Kk peptide-binding groove can be emphasized by analyzing other MHC alleles with an Arg residue at position 97. This comparison shows that the side chain of Arg97 adopts four distinct conformations according to the nature of neighboring residues (Fig. 5). Two of these conformations are alternatively used by HLA-A2 depending on the sequence of the peptide to which it binds (30), one is used by the two closely related HLA-B35 and HLA-B53 alleles (31), and the last one is used by both HLA-B44 (27) and the two H-2Kk complexes reported in this paper. As pointed out by Madden et al. (30), alternate conformations of Arg97 and Tyr116 in HLA-A2 are used to bind to different peptide sequences. The flexibility of Arg97 is also involved in the structural reorganization that occurs within the Tax-HLA-A2 complex upon TCR binding (32, 33) (Fig. 5). Some conformational malleability has also been reported for Arg97 of HLA-B53 (31). Therefore, the coincident position noted for Arg97 in the two H-2Kk complexes reported in this study suggests that it is probably rigid and markedly contrasts with the malleability of Arg97 in HLA-B53 and HLA-A2. This unique feature has probably been selected to allow Arg97 of H-2Kk to directly bind octamers and nonamers.

FIGURE 5.

Four conformations of residue Arg97 in mouse and human MHCI molecules. The backbone of five nonapeptides is shown together with the side chains of MHC residues Arg97 and Tyr116. MHC residues and the nonapeptides of the complexes are colored as follows: H-2Kk, carbon atoms in yellow, oxygen atoms in red, and nitrogen atoms in blue; HLA-B44, red; HLA-B53, blue; HLA-A2-Tax complex, pink; and TCR-bound HLA-A2 Tax complex or HLA-A2-gp120, green. In each of the complexes, Arg97 uses either one or two water molecules to contact variable segments of the bound nonapeptides.

FIGURE 5.

Four conformations of residue Arg97 in mouse and human MHCI molecules. The backbone of five nonapeptides is shown together with the side chains of MHC residues Arg97 and Tyr116. MHC residues and the nonapeptides of the complexes are colored as follows: H-2Kk, carbon atoms in yellow, oxygen atoms in red, and nitrogen atoms in blue; HLA-B44, red; HLA-B53, blue; HLA-A2-Tax complex, pink; and TCR-bound HLA-A2 Tax complex or HLA-A2-gp120, green. In each of the complexes, Arg97 uses either one or two water molecules to contact variable segments of the bound nonapeptides.

Close modal

Taking into account the likely rigidity of Arg97, simulation studies were undertaken to determine the repertoire of peptide residues that can be accommodated within the H-2Kk groove, especially at position P5. These studies demonstrate that in the absence of major structural rearrangement of the backbone, it is impossible to fit an aromatic residue at P5 and also suggest that most H-2Kk-bound peptides will conserve a Cα trace similar to that of HA259–266 as long as their P5 residue is not bulkier than Leu/Ile. Consistent with this prediction, the syfpeithi database (〈http://syfpeithi.bmi-heidelberg.com〉) shows that 19 octapeptides of the 22 reported to bind H-2Kk have small side chains at P5. A minimal epitope of Mycobacterium tuberculosis (MT) recognized by CD8 T cells in the context of H-2Kk molecules has been recently reported (34). Docking of this MT-derived peptide into the H-2Kk binding groove can be achieved without steric clash, and this suggests the possibility of accommodating an Ala at P5. Together, these data suggest that the Cα trace adopted by the HA259–266 octamer constitutes a generic feature of octapeptides bound in the H-2Kk groove.

After the generation of an allele-neutral TCR repertoire in early thymocytes, a phase of molecular matching arises through positive selection of TCRs capable of coping with a composite surface made of self-peptide side chains and of MHC determinants that are both conserved and allele specific. There are probably fewer constraints linked to matching the few allele-specific residues found on the top of the MHC helices and available for TCR contacts than adapting to the diversity of peptide side chains and to the generic features that are imposed on the bound peptides by the architecture of a given MHC peptide-binding groove. For instance, TCR readout of peptides that follow a flat and deep course within the MHCI groove, as exemplified by octapeptides bound to H-2Kb, generates structural constraints distinct from the readout of H-2Db- and H-2Ld-bound peptides that protrude out of the C-terminal part of the peptide-binding groove. Such a challenge is in part met by the flexibility of the TCR Ag-binding site (35, 36). Conformational adjustments at the level of the bound peptides also maximize matching at the TCR-peptide-MHC interface (37, 38). We showed that in the case of H-2Kk, the HA259–266 octapeptide displays a low degree of solvent exposure. Moreover, our modeling studies suggest that an MT-derived and H-2Kk-bound peptide is even less exposed to TCR scrutiny than peptide HA259–266; the solvent accessibilities of the P4, P6, and P7 residues of the MT-derived peptide are 91, 17, and 68 Å2, respectively. No TCR structure bound to H-2Kk-octamer complex is yet available, and it remains to be determined how the TCR binding site manages to accommodate these peptides that lack prominently exposed residues and appear rather featureless to the TCR. Along that line, the crystal structure of a human TCR directed to a peptide derived from influenza virus and presented by HLA-A2 showed that very few docking possibilities exist to specifically recognize peptide lacking surface-exposed side chains (39).

In conclusion, the two H-2Kk crystal structures reported in this paper together with data resulting from the sequencing of naturally processed peptides eluted from H-2Kk show that the H-2Kk binding motif is composed of two primary anchors at P2 and PΩ. Pocket B accommodates residue P2 and shows a strong preference for negatively charged residues with long aliphatic chain (Glu), whereas pocket F prefers aliphatic residues (Ile and Val) at the peptide C terminus. The dominant solvent-accessible residues of the HA259–266 octapeptide are found at P4 and P7, whereas P6 is poorly solvent accessible. In the SV40560–568 nonapeptide bound to H-2Kk, residues at position P4, P6, and P8 are fully solvent accessible. The structure of H-2Kk in complex with an octapeptide also shows that side chains from residues of the floor of the peptide-binding groove can be used to directly lock an octapeptide main chain. This constitutes a unique feature among the MHCI-peptide complexes studied to date and results in a central peptide region deeply embedded in the groove and poorly accessible to TCR contact. Given that the structures reported in this study represent a novel conformation of bound peptides not hitherto used for modeling TCR-peptide-MHCI interactions, they may help ab initio peptide epitope identification.

We thank C. Gregoire, A. Guimezanes, A.-M. Schmitt-Verhulst, and J. Ewbank for discussions.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

The H-2Kk-HA(259–266) and H-2Kk-SV40(560–568) structures have been submitted to the Protein Data Bank under accession codes 1ZT1 and 1ZT7, respectively.

2

This work was supported by grants from Centre d’Immunologie de Marseille-Luminy, Institut National de la Santé et de la Recherche Médicale-Centre National Recherche de la Scientifique, and the European Communities (Project EPI-PEP-VAC QLK2-CT-2002-00620).

4

Abbreviations used in this paper: MHCI, MHC class Ia; MT, Mycobacterium tuberculosis; P1, position 1.

1
Yewdell, J. W., E. Reits, J. Neefjes.
2003
. Making sense of mass destruction: quantitating MHC class I antigen presentation.
Nat. Rev. Immunol.
3
:
952
.-961.
2
Madden, D. R..
1995
. The three-dimensional structure of peptide-MHC complexes.
Annu. Rev. Immunol.
13
:
587
.-622.
3
Batalia, M. A., E. J. Collins.
1997
. Peptide binding by class I and class II MHC molecules.
Biopolymers
43
:
281
.-302.
4
Fremont, D. H., M. Matsumura, E. A. Stura, P. A. Peterson, I. A. Wilson.
1992
. Crystal structures of two viral peptides in complex with murine MHC class I H-2Kb.
Science
257
:
919
.-927.
5
Fremont, D. H., E. A. Stura, M. Matsumura, P. A. Peterson, I. A. Wilson.
1995
. Crystal structure of an H-2Kb-ovalbumin peptide complex reveals the interplay of primary and secondary anchor positions in the major histocompatibility complex binding groove.
Proc. Natl. Acad. Sci. USA
92
:
2479
.-2483.
6
Achour, A., J. Michaelsson, R. A. Harris, J. Odeberg, P. Grufman, J. K. Sandberg, V. Levitsky, K. Karre, T. Sandalova, G. Schneider.
2002
. A structural basis for LCMV immune evasion: subversion of H-2D(b) and H-2K(b) presentation of gp33 revealed by comparative crystal structure: analyses.
Immunity
17
:
757
.-768.
7
Apostolopoulos, V., M. Yu, A. L. Corper, L. Teyton, G. A. Pietersz, I. F. McKenzie, I. A. Wilson, M. Plebanski.
2002
. Crystal structure of a non-canonical low-affinity peptide complexed with MHC class I: a new approach for vaccine design.
J. Mol. Biol.
318
:
1293
.-1305.
8
Apostolopoulos, V., M. Yu, A. L. Corper, W. Li, I. F. McKenzie, L. Teyton, I. A. Wilson, M. Plebanski.
2002
. Crystal structure of a non-canonical high affinity peptide complexed with MHC class I: a novel use of alternative anchors.
J. Mol. Biol.
318
:
1307
.-1316.
9
Young, A. C., W. Zhang, J. C. Sacchettini, S. G. Nathenson.
1994
. The three-dimensional structure of H-2Db at 2.4 A resolution: implications for antigen-determinant selection.
Cell
76
:
39
.-50.
10
Zhao, R., D. J. Loftus, E. Appella, E. J. Collins.
1999
. Structural evidence of T cell xeno-reactivity in the absence of molecular mimicry.
J. Exp. Med.
189
:
359
.-370.
11
Garcia, K. C., M. Degano, L. R. Pease, M. Huang, P. A. Peterson, L. Teyton, I. A. Wilson.
1998
. Structural basis of plasticity in T cell receptor recognition of a self peptide-MHC antigen.
Science
279
:
1166
.-1172.
12
Degano, M., K. C. Garcia, V. Apostolopoulos, M. G. Rudolph, L. Teyton, I. A. Wilson.
2000
. A functional hot spot for antigen recognition in a superagonist TCR/MHC complex.
Immunity
12
:
251
.-261.
13
Reiser, J. B., C. Darnault, A. Guimezanes, C. Gregoire, T. Mosser, A. M. Schmitt-Verhulst, J. C. Fontecilla-Camps, B. Malissen, D. Housset, G. Mazza.
2000
. Crystal structure of a T cell receptor bound to an allogeneic MHC molecule.
Nat. Immunol.
1
:
291
.-297.
14
Reiser, J. B., C. Gregoire, C. Darnault, T. Mosser, A. Guimezanes, A. M. Schmitt-Verhulst, J. C. Fontecilla-Camps, G. Mazza, B. Malissen, D. Housset.
2002
. A T cell receptor CDR3β loop undergoes conformational changes of unprecedented magnitude upon binding to a peptide/MHC class I complex.
Immunity
16
:
345
.-354.
15
Reiser, J. B., C. Darnault, C. Gregoire, T. Mosser, G. Mazza, A. Kearney, P. A. van der Merwe, J. C. Fontecilla-Camps, D. Housset, B. Malissen.
2003
. CDR3 loop flexibility contributes to the degeneracy of TCR recognition.
Nat. Immunol.
4
:
241
.-247.
16
Li, H., K. Natarajan, E. L. Malchiodi, D. H. Margulies, R. A. Mariuzza.
1998
. Three-dimensional structure of H-2Dd complexed with an immunodominant peptide from human immunodeficiency virus envelope glycoprotein 120.
J. Mol. Biol.
283
:
179
.-191.
17
Achour, A., K. Persson, R. A. Harris, J. Sundback, C. L. Sentman, Y. Lindqvist, G. Schneider, K. Karre.
1998
. The crystal structure of H-2Dd MHC class I complexed with the HIV-1-derived peptide P18–I10 at 2.4 A resolution: implications for T cell and NK cell recognition.
Immunity
9
:
199
.-208.
18
Balendiran, G. K., J. C. Solheim, A. C. Young, T. H. Hansen, S. G. Nathenson, J. C. Sacchettini.
1997
. The three-dimensional structure of an H-2Ld-peptide complex explains the unique interaction of Ld with β-2 microglobulin and peptide.
Proc. Natl. Acad. Sci. USA
94
:
6880
.-6885.
19
Speir, J. A., K. C. Garcia, A. Brunmark, M. Degano, P. A. Peterson, L. Teyton, I. A. Wilson.
1998
. Structural basis of 2C TCR allorecognition of H-2Ld peptide complexes.
Immunity
8
:
553
.-562.
20
Gould, K. G., H. Scotney, G. G. Brownlee.
1991
. Characterization of two distinct major histocompatibility complex class Iκk-restricted T-cell epitopes within the influenza A/PR/8/34 virus hemagglutinin.
J. Virol.
65
:
5401
.-5409.
21
Rawle, F. C., K. A. O’Connell, R. W. Geib, B. Roberts, L. R. Gooding.
1988
. Fine mapping of an H-2Kk restricted cytotoxic T lymphocyte epitope in SV40 T antigen by using in-frame deletion mutants and a synthetic peptide.
J. Immunol.
141
:
2734
.-2739.
22
Kellenberger, C., S. Porciero, A. Roussel.
2004
. Expression, refolding, crystallization and preliminary crystallographic study of MHC H-2Kk complexed with octapeptides and nonapeptides.
Acta Crytallogr. D.
60
:
1278
.-1280.
23
Navaza, J..
1994
. AMoRe: an automated package for molecular replacement.
Acta Crysollogrt. A.
50
:
157
.-163.
24
Roussel, A., C. Cambillau.
1991
.
Silicon Graphics Geometry Partners Directory
85
. Silicon Graphics, Mountain View, CA.
25
Brunger, A. T., P. D. Adams, G. M. Clore, W. L. DeLano, P. Gros, R. W. Grosse-Kunstleve, J. S. Jiang, J. Kuszewski, M. Nilges, N. S. Pannu, et al
1998
. Crystallography & NMR system: A new software suite for macromolecular structure determination.
Acta Crystallogr. D
54
:
905
.-921.
26
Ciatto, C., A. C. Tissot, M. Tschopp, G. Capitani, F. Pecorari, A. Pluckthun, M. G. Grutter.
2001
. Zooming in on the hydrophobic ridge of H-2D(b): implications for the conformational variability of bound peptides.
J. Mol. Biol.
312
:
1059
.-1071.
27
Macdonald, W. A., A. W. Purcell, N. A. Mifsud, L. K. Ely, D. S. Williams, L. Chang, J. J. Gorman, C. S. Clements, L. Kjer-Nielsen, D. M. Koelle, et al
2003
. A naturally selected dimorphism within the HLA-B44 supertype alters class I structure, peptide repertoire, and T cell recognition.
J. Exp. Med.
198
:
679
.-691.
28
Menssen, R., P. Orth, A. Ziegler, W. Saenger.
1999
. Decamer-like conformation of a nona-peptide bound to HLA-B*3501 due to non-standard positioning of the C terminus.
J. Mol. Biol.
285
:
645
.-653.
29
Hulsmeyer, M., M. T. Fiorillo, F. Bettosini, R. Sorrentino, W. Saenger, A. Ziegler, B. Uchanska-Ziegler.
2004
. Dual, HLA-B27 subtype-dependent conformation of a self-peptide.
J. Exp. Med.
199
:
271
.-291.
30
Madden, D. R., D. N. Garboczi, D. C. Wiley.
1993
. The antigenic identity of peptide-MHC complexes: a comparison of the conformations of five viral peptides presented by HLA-A2.
Cell
75
:
693
.-708.
31
Smith, K. J., S. W. Reid, K. Harlos, A. J. McMichael, D. I. Stuart, J. I. Bell, E. Y. Jones.
1996
. Bound water structure and polymorphic amino acids act together to allow the binding of different peptides to MHC class I HLA-B53.
Immunity
4
:
215
.-228.
32
Garboczi, D. N., P. Ghosh, U. Utz, Q. R. Fan, W. E. Biddison, D. C. Wiley.
1996
. Structure of the complex between human T-cell receptor, viral peptide and HLA-A2.
Nature
384
:
134
.-141.
33
Ding, Y. H., K. J. Smith, D. N. Garboczi, U. Utz, W. E. Biddison, D. C. Wiley.
1998
. Two human T cell receptors bind in a similar diagonal mode to the HLA-A2/Tax peptide complex using different TCR amino acids.
Immunity
8
:
403
.-411.
34
Kamath, A. B., J. Woodworth, X. Xiong, C. Taylor, Y. Weng, S. M. Behar.
2004
. Cytolytic CD8+ T cells recognizing CFP10 are recruited to the lung after Mycobacterium tuberculosis infection.
J. Exp. Med.
200
:
1479
.-1489.
35
Housset, D., B. Malissen.
2003
. What do TCR-pMHC crystal structures teach us about MHC restriction and alloreactivity?.
Trends Immunol.
24
:
429
.-437.
36
Rudolph, M. G., I. A. Wilson.
2002
. The specificity of TCR/pMHC interaction.
Curr. Opin. Immunol.
14
:
52
.-65.
37
Miley, M. J., I. Messaoudi, B. M. Metzner, Y. Wu, J. Nikolich-Zugich, D. H. Fremont.
2004
. Structural basis for the restoration of TCR recognition of an MHC allelic variant by peptide secondary anchor substitution.
J. Exp. Med.
200
:
1445
.-1454.
38
Lee, J. K., G. Stewart-Jones, T. Dong, K. Harlos, K. Di Gleria, L. Dorrell, D. C. Douek, P. A. van der Merwe, E. Y. Jones, A. J. McMichael.
2004
. T cell cross-reactivity and conformational changes during TCR engagement.
J. Exp. Med.
200
:
1455
.-1466.
39
Stewart-Jones, G. B., A. J. McMichael, J. I. Bell, D. I. Stuart, E. Y. Jones.
2003
. A structural basis for immunodominant human T cell receptor recognition.
Nat. Immunol.
4
:
657
.-663.