Structural basis for clonal diversity of the human T-cell response to a dominant influenza virus epitope

Influenza A virus (IAV) causes an acute infection in humans that is normally eliminated by CD8+ cytotoxic T lymphocytes. Individuals expressing the MHC class I molecule HLA-A2 produce cytotoxic T lymphocytes bearing T-cell receptors (TCRs) that recognize the immunodominant IAV epitope GILGFVFTL (GIL). Most GIL-specific TCRs utilize α/β chain pairs encoded by the TRAV27/TRBV19 gene combination to recognize this relatively featureless peptide epitope (canonical TCRs). However, ∼40% of GIL-specific TCRs express a wide variety of other TRAV/TRBV combinations (non-canonical TCRs). To investigate the structural underpinnings of this remarkable diversity, we determined the crystal structure of a non-canonical GIL-specific TCR (F50) expressing the TRAV13-1/TRBV27 gene combination bound to GIL–HLA-A2 to 1.7 Å resolution. Comparison of the F50–GIL–HLA-A2 complex with the previously published complex formed by a canonical TCR (JM22) revealed that F50 and JM22 engage GIL–HLA-A2 in markedly different orientations. These orientations are distinguished by crossing angles of TCR to peptide–MHC of 29° for F50 versus 69° for JM22 and by a focus by F50 on the C terminus rather than the center of the MHC α1 helix for JM22. In addition, F50, unlike JM22, uses a tryptophan instead of an arginine to fill a critical notch between GIL and the HLA-A2 α2 helix. The F50–GIL–HLA-A2 complex shows that there are multiple structurally distinct solutions to recognizing an identical peptide–MHC ligand with sufficient affinity to elicit a broad anti-IAV response that protects against viral escape and T-cell clonal loss.

CD8 ϩ T cells play a critical role in the immune response to viruses by recognizing and eliminating infected cells (1). Recognition is mediated by ␣␤ T-cell receptors (TCRs), 2 which bind viral peptides presented by major histocompatibility complex (MHC) class I molecules on infected cells. TCRs engage peptide-MHC (pMHC) through their six complementarity-determining region (CDR) loops, three from the variable ␣ (V␣) domain, and three from V␤. The first and second CDRs (CDR1 and CDR2) are encoded within the V␣ and V␤ gene segments. CDR3 is formed by DNA recombination involving juxtaposition of V␣ and J␣ segments for the ␣ chain genes and of V␤, D, and J␤ segments for the ␤ chain genes. Estimates of TCR diversity in humans have placed the number of unique structures in the range of 10 5 -10 8 (2)(3)(4)(5).
The human CD8 ϩ T-cell response to influenza A virus (IAV) has been studied extensively (6 -11). The dominant epitope in individuals expressing the MHC class I molecule HLA-A*0201 (HLA-A2) corresponds to residues 58 -66 of matrix protein M1 (GILGFVFTL; referred to as GIL) (12). Initial studies of GIL-specific CD8 ϩ T-cell responses in HLA-A2 ϩ subjects indicated that the V␤ repertoire is highly biased toward usage of the TRBV19 gene (up to 98%) with a highly conserved CDR3␤ motif, 97 XRSX 100 (6 -9). The V␣ gene segment, although not as strongly selected as V␤, showed a strong preference for TRAV27 (up to 80%) (6,7,13). This canonical TRAV27/ TRBV19 gene combination is observed in multiple HLA-A2matched but otherwise genetically unrelated individuals. More recently, however, high-throughput sequencing and single-cell TCR analysis have revealed substantially greater repertoire diversity than previously realized, with only ϳ60% of GIL-specific TCRs expressing the TRBV19 gene and ϳ20% expressing TRAV27 (14). These non-canonical TCRs utilize a wide variety of TRAV and TRBV genes, including TRAV13-1, TRAV17, TRAV29, TRAV38-2, TRBV14, TRBV24-1, TRBV27, and TRBV29-1, among others. Indeed, 2,406 unique TCR␣ and 2,437 TCR␤ sequences have been identified for TCRs recognizing the GIL peptide presented by HLA-A2 (14). These sequences include 461 V␣-J␣ and 359 V␤-J␤ combinations, as well as dozens of distinct CDR3␣ and CDR3␤ consensus motifs. Broad TCR repertoire diversity has also been documented for other defined viral antigens, including ones from cytomegalovirus, Epstein-Barr virus, and HIV (14 -17). Such diversity ensures robust T-cell responses to single viral epitopes that would not be possible if a single epitope could only elicit a few   (18). Moreover, TCR diversity provides protection against viral escape (19 -21).
In most pMHC structures, one or more residues in the central portion of the antigenic peptide (P4 -P6) feature solventexposed side chains that facilitate TCR binding (22). However, the GIL peptide is unusual in that only the side chain of P8 Thr, at the C-terminal end of the peptide, is substantially exposed to solvent, making GIL a challenging target for TCR recognition (23). The crystal structure of a canonical GIL-specific TRAV27/ TRBV19 TCR (JM22) bound to GIL-HLA-A2 revealed what appears to be the most efficient solution to recognizing the featureless GIL peptide with sufficient affinity to permit selection of engaging T cells. In the JM22-GIL-HLA-A2 complex, the side chain of the conserved ␤Arg 98 residue of the 97 XRSX 100 CDR3␤ consensus motif fills a notch between the peptide and the HLA-A2 ␣2 helix (23). This structural solution is also adopted by other GIL-specific TCRs expressing TRBV19 and the 97 XRSX 100 CDR3␤ motif, even when TRBV19 is paired with ␣ chains different from TRAV27 (14,24). However, ϳ40% of GIL-specific TCRs do not use the canonical TRBV19 gene, and not all TRBV19-expressing TCRs contain the 97 XRSX 100 CDR3␤ motif (14,24). Hence, multiple solutions exist for binding the featureless GIL peptide. Here we investigated the structural basis for the surprising diversity of the TCR response to this dominant IAV epitope by determining the structure of a GIL-specific TCR (F50) expressing a completely different, non-canonical TRAV/TRBV gene combination (TRAV13-1/ TRBV27) in complex with GIL-HLA-A2.

Interaction of TCR F50 with GIL-HLA-A2
The IAV GIL-specific TCR F50 was isolated from CD8 ϩ T cells from the peripheral blood of an HLA-A2 ϩ healthy male donor following in vitro stimulation with the GIL peptide as described under "Experimental procedures." F50 utilizes gene segments TRAV13-1 and TRAJ54 for the ␣ chain, and TRBV27 and TRBJ1-1 for the ␤ chain, whereas JM22 utilizes TRAV27 and TRAJ37 for the ␣ chain and TRBV19 and TRBJ7-2 for the ␤ chain. We used surface plasmon resonance (SPR) to measure the affinity of TCR F50 for HLA-A2 loaded with the GIL peptide (Fig. 1A). To characterize the interaction of F50 with GIL-HLA-A2, we expressed these proteins by in vitro folding from inclusion bodies produced in Escherichia coli. Biotinylated GIL-HLA-A2 was directionally coupled to a streptavidincoated biosensor surface, and different concentrations of F50 were sequentially flowed over the immobilized pMHC ligand. A dissociation constant (K D ) of 76 Ϯ 4 M was obtained by fitting equilibrium data to a 1:1 binding model (Fig. 1B). This affinity is ϳ25-fold weaker than that of JM22 for GIL-HLA-A2 (K D ϭ 3.2 M) (25), with the caveat that we did not independently measure the affinity of JM22 in our experimental system. It is, however, well within the range of 0.5-500 M for natural TCR-pMHC interactions (26). Moreover, it is consistent with the much lower representation among GIL-specific TCRs of the TRBV27 ␤ chain expressed by F50 than the dominant TRBV19 ␤ chain expressed by JM22 (14,24), in agreement with the concept of affinity-driven clonal expansion of T cell repertoires.

Overview of the F50 -GIL-HLA-A2 complex
To understand how TCR F50 recognizes GIL-HLA-A2 and to compare its recognition mode with that of JM22, we determined the structure of the F50 -GIL-HLA-A2 complex to 1.7 Å resolution (Table 1 and Fig. 2a). The interface between F50 and GIL-HLA-A2 was in unambiguous electron density for the single complex molecule in the asymmetric unit of the crystal (Fig.  2b). Of note, the resolution of the F50 -GIL-HLA-A2 complex is one of the highest reported for any TCR-pMHC class I or II complex (Ͼ60 unique structures), whose resolutions seldom exceed 2.5 Å (22). Indeed, the 1.7 Å resolution of the F50 -GIL-HLA-A2 complex is second only to that of the JM22-GIL-HLA-A2 complex itself at 1.4 Å (23). The comparably high resolutions of these two structures justify detailed comparisons between them.
F50 docks symmetrically over GIL-HLA-A2 in a diagonal orientation, with a crossing angle of TCR to pMHC (27) of 29°c ompared with 69°for JM22 (Fig. 2c). Upon binding GIL-HLA-A2, F50 buries 73% (215 Å 2 ) of the peptide solvent-accessible surface. This percentage is significantly less than the 85% (258 Å 2 ) of peptide surface buried by JM22 (23), which may contribute to the lower affinity of F50 compared with JM22. As shown by the footprint of F50 on the pMHC surface (Fig. 2d), F50 establishes contacts with the N-terminal half of the GIL peptide where F c is the calculated structure factor. R free is as for R work but calculated for a randomly selected 5.0% of reflections not included in the refinement.
Overall, V␣ of F50 makes 27 van der Waals contacts and three hydrogen bonds with HLA-A2, compared with 25 van der Waals contacts and two hydrogen bonds by V␤. These interactions are mediated by six V␣ and six V␤ residues and involve twelve MHC residues, eight of which are also contacted by JM22 (Table 2). Of the total buried surface on HLA-A2, excluding the GIL peptide, CDR1␣, CDR2␣, and CDR3␣ contribute 6, 22, and 19%, respectively, compared with 9, 20, and 28% for CDR1␤, CDR2␤, and CDR3␤, respectively. Thus, MHC recognition by F50 involves both germline-encoded and somatically generated interactions. The contributions of CDR2␣ and CDR2␤ to the MHC buried surface in the F50 -GIL-HLA-A2 complex (22 and 20%, respectively) are substantially greater than the average for 34 other TCR-pMHC class I complexes (11 and 12%, respectively) (22). In particular, CDR2␣ of the canon-ical JM22 TCR accounts for only 7% of the buried surface on HLA-A2. Unlike JM22, F50 does not recruit water molecules to the interface with HLA-A2 (see below).

Human T-cell receptor recognition of influenza virus
hydrogen bonds with Thr 80H in the F50 -GIL-HLA-A2 complex, this same residue hydrogen bonds with Ala 69H and Thr 73H in the T36-5-HIV-HLA-A24 complex, in agreement with the idea of flexibility in evolutionarily selected contacts between TCR and MHC (29,30).

Peptide recognition by TCR F50
Except for a few contacts made by CDR1␣ and CDR1␤ (5 of 36), all interactions between F50 and the GIL peptide are mediated by the somatically generated CDR3 loops, with CDR3␣ and CDR3␤ accounting for 18 and 13 contacts, respectively. Peptide specificity is conferred mainly by shape complementarity, because the F50-GIL interface features only two hydrogen bonds: F50 ␣Gln 101 O⑀1-N P6 Val and F50 ␤Trp 99 N⑀1-O P6 Val (Table 3 and Fig. 3, c and d). Of note, JM22 also makes two hydrogen bonds with the main chain of P6 Val, but using different CDR residues: JM22 ␤Gln 52 O⑀1-N P6 Val and JM22 ␤Ser 99 O␥-O P6 Val. F50 engages the central and C-terminal portions of the GIL peptide (P4 Gly, P5 Phe, P6 Val, P8 Thr, and P9 Leu) with the principal focus on P5 Phe and P6 Val, whereas N-terminal residues P1-P3 do not contact TCR.
In most unliganded pMHC structures, one or more residues in the central portion of the antigenic peptide (P4 -P6) feature protruding side chains that facilitate TCR binding (22). However, the GIL peptide presented by HLA-A2 is atypical in that only the side chain of P8 Thr, at the C-terminal end of the peptide, is substantially exposed to solvent. In the JM22-GIL-HLA-A2 complex (23), the guanidinium moiety of CDR3␤ Arg 98 , which is conserved in most GIL-specific TCRs express-ing the canonical TRBV19 gene segment (6 -9, 14, 24), inserts into a notch between the peptide and the HLA-A2 ␣2 helix. In the F50 -GIL-HLA-A2 complex, by contrast, this same notch is occupied by the aromatic side chain of CDR3␤ Trp 99 , which makes 10 hydrophobic contacts and one hydrogen bond (F50 ␤Trp 99 N⑀1-O P6 Val) with the GIL peptide (Table 3 and Fig.  3e). These interactions are reinforced by eight contacts with HLA-A2 ␣2 residues Ala 150H , Val 152H , and Gln 155H , further anchoring the ␤Trp 99 side chain in the mainly hydrophobic notch. These same MHC residues are also contacted by ␤Arg 98 of JM22, indicating a certain degree of structural mimicry. However, F50 ␤Trp 99 , unlike JM22 ␤Arg 98 ␤, cannot form hydrogen bonds with HLA-A2 Ala 150H or Gln 155H (Table 3). Indeed, mutating ␤Trp 99 to arginine completely abolished . c, interactions of CDR1␣ and CDR3␣ with the GIL peptide. A bridging water molecule is depicted as a red sphere. d, interactions of CDR1␤ and CDR3␤ with the GIL peptide, including bridging water molecules (red spheres). e, interactions between CDR3␤ (green) of F50 and GIL-HLA-A2. The side chain of ␤Trp 99 occupies a notch between the GIL peptide (magenta) and the HLA-A2 ␣2 helix (orange). Superposed onto CDR3␤ of F50 is CDR3␤ of JM22 (pink). In the JM22-GIL-HLA-A2 complex (23), the pocket between GIL and the HLA-A2 ␣2 helix is filled by the side chain of ␤Arg 98 , which makes hydrogen bonds with HLA-A2 Ala 150H and Gln 155H (beige).

Human T-cell receptor recognition of influenza virus
binding of F50 to GIL-HLA-A2, as determined by SPR (not shown), demonstrating that these residues are not functionally interchangeable.
The side chain of HLA-A2 Gln 155H adopts different rotamer conformations in the F50 -GIL-HLA-A2 and JM22-GIL-HLA-A2 structures (Fig. 3e). In the JM22-GIL-HLA-A2 complex, the Gln 155H side chain shifts by 3.6 Å compared with its position in unbound GIL-HLA-A2 to open the notch between the peptide and the ␣2 helix, into which ␤Arg 98 docks (23). In the F50 -GIL-HLA-A2 complex, the gatekeeper Gln 155H side chain shifts an additional 1.8 Å from its unbound position to further open this notch to avoid steric clashes with the bulky ␤Trp 99 side chain and optimize shape complementarity and hydrogen bonding with pMHC (Fig. 3e).

Comparison with GIL-specific TCRs LS01 and LS10
A recent crystallographic study of two GIL-specific TCRs (LS01 and LS10) expressing the canonical TRBV19 gene, but different TRAV genes showed that both TCRs maintain the same overall docking orientation on pMHC as JM22, because of the preservation of specific contacts between the conserved CDR1␤ and CDR2␤ loops and HLA-A2 (Fig. 4A) (24). This orientation is distinct from that found in the non-canonical F50 -GIL-HLA-A2 complex, which features a much more acute crossing angle, as described above.
The CDR3␤ motifs of LS01 ( 97 XFX 99 ) and LS10 ( 97 XGXY 100 ) differ from the dominant 97 XRSX 100 motif of other GIL-specific TCRs expressing TRBV19 (24). Consequently, LS01, LS10, JM22, and F50 employ different strategies to recognize the critical notch between GIL and the MHC ␣2 helix near P5 Phe (Fig.  4B). Whereas JM22 and F50 use Arg 98 and Trp 99 of CDR3␤, respectively, to fill this common pocket, LS01 uses Phe 98 . By contrast, binding of LS10 induces a conformational change in GIL, whereby the P5 Phe side chain moves into the notch. This shift creates a new notch that is occupied by ␣Ala 98 and ␣Gly 99 and covered by ␣Tyr 103 .

Water interactions at TCR-pMHC interfaces
Bound water molecules have been localized in the interfaces of many antigen-antibody (and other protein-protein) complexes, where they act as molecular adaptors to bridge the protein partners and improve the fit between them (31). However, the contribution of bound waters to mediating TCR-pMHC interactions has not been examined in detail, in large measure because relatively few TCR-pMHC crystal structures have been determined at sufficiently high resolution (Յ2.5 Å) to permit the identification of ordered waters with a reasonable degree of accuracy.
The high resolution of the F50 -GIL-HLA-A2 complex allowed the inclusion of many bound water molecules, includ-

Human T-cell receptor recognition of influenza virus
ing ones in the interface between TCR and pMHC. In particular, four waters are mostly or completely buried in the interface, where they mediate hydrogen bonding interactions between F50 and the GIL peptide (Table 4 and Fig. 3d). No buried waters were found between F50 and HLA-A2. By contrast, the JM22-GIL-HLA-A2 structure includes bridging waters both between JM22 and GIL and between JM22 and HLA-A2 (23). The shape correlation statistic (S c ) (32) for the F50 -GIL-HLA-A2 complex with interfacial waters is 0.65 (S c ϭ 1.0 for interfaces with perfect geometrical fits), which is the same as the S c without such waters (Table 5). Thus, bound water molecules do not contribute appreciably to the overall interfacial shape complementarity of this particular TCR-pMHC complex, although they do fill a gap between F50 and GIL. In marked contrast, the corresponding S c values for the JM22-GIL-HLA-A2 complex (23) are 0.77 with interfacial waters versus 0.64 without waters (⌬S c ϭ 0.13), indicating a substantial contribution to improving the overall fit. Similarly, for the F6 -GIL-HLA-A2 complex (14), in which F6 uses the same ␤ chain as JM22 but a different ␣ chain, the S c values are 0.61 and 0.54 with and without interfacial waters, respectively.
An analysis of 11 other TCR-pMHC structures of Յ2.5 Å resolution revealed little or no effect of interfacial waters on overall shape complementarity (⌬S c Ͻ 0.05), except in one case (LC13-EBV-HLA-A8) (⌬S c ϭ 0.06) ( Table 5). Nevertheless, these waters help correct imperfections in the interface by filling small cavities or channels between TCR and pMHC. More importantly, they contribute to complex stability by forming bridging hydrogen bonds to enhance polar interactions and neutralize unpaired hydrogen-bonding groups, as observed in other protein-protein complexes (33). Indeed, most TCR-pMHC complexes contain comparable numbers of water-mediated and direct hydrogen bonds (e.g. 5 versus 7 for F50 -GIL-HLA-A2, 14 versus 8 for JM22-GIL-HLA-A2 (23), and 11 versus 11 for F6 -GIL-HLA-A2 (14)). For the 14 TCR-pMHC structures analyzed here (Table 5), the number of interfacial waters ranged from as few as 2 to as many as 11, with an average of 6/complex, in line with antigen-antibody complexes (31).

Discussion
The application of high-throughput sequencing and singlecell TCR analysis to interrogate CD8 ϩ T cell repertoires to single viral epitopes has revealed far greater TCR sequence diversity than previously appreciated for immune responses to IAV, cytomegalovirus, HIV, and other viruses (14,15,17,24). Given this diversity, understanding the structural basis for recognition of an identical pMHC complex by thousands of different TCRs represents a considerable challenge.
To date, ϳ25 structures of TCRs with different V␣ and/or V␤ gene usage bound to an identical (or nearly identical) pMHC ligand have been reported, involving both MHC class I and class II molecules (14,22,24,34). These complexes may be divided into three categories: 1) those in which the TCRs use the same V␣ gene segment but different V␤s (30), 2) those in which TCRs use the same V␤ but different V␣s (14,24,(35)(36)(37)(38), and 3) those in which the TCRs use different V␣s and V␤s (34,39). The third category, which includes TCR F50, has considerably fewer examples than the first two categories, where the TCRs use the same V␣ or V␤ region to engage the same pMHC. Nevertheless, some general conclusions may be drawn regarding V␣/V␤ gene usage and TCR docking orientation. Thus, TCRs that use the same V␤ but different V␣s typically bind pMHC in the same overall orientation because of the preservation of most (but not necessarily all) germline-encoded interactions between the conserved CDR1␤ and CDR2␤ loops and the MHC ␣-helices. That is, use of a different V␣ has not been observed to reposition V␤ on pMHC in an appreciably different way, although small adjustments do occur. Conversely, TCRs that use the same V␣ but different V␤s engage pMHC in very similar orientations because most germline-encoded interac- Table 4 Water bridges between TCR and GIL peptide in the F50 -GIL-HLA-A2 and JM22-GIL-HLA-A2 complexes a Only TCR-pMHC class I structures of at 2.5 Å resolution or better were considered in this analysis. b The first value is the total number of water molecules in the corresponding TCR-pMHC interface. In parentheses are the numbers of waters bridging TCR and MHC or TCR and peptide in each complex. These numbers may in some cases add up to more than the total number of interfacial waters because a single water molecule can sometimes bridge TCR to both MHC and peptide.

Human T-cell receptor recognition of influenza virus
tions between CDR1␣ and CDR2␣ and the MHC ligand are maintained, irrespective of the V␤ partner. By contrast, TCRs using different V␣ and V␤ regions can employ very different strategies to bind an identical pMHC, as seen here and in previous studies (34,39). For example, a comparison of two TCRs expressing unrelated V␣/V␤ gene combinations in complex with a bulged peptide from Epstein-Barr virus presented by HLA-B8 revealed two distinct binding modes: one in which the TCR straddles the bulged peptide but makes few contacts with MHC and one in which the TCR is positioned toward the N-terminal end of the peptide binding groove of HLA-B8, thereby largely bypassing the bulged peptide (39). In another case, human cytomegalovirus-specific TCRs C7 (TRAV24/TRBV7-2) and C25 (TRAV26-2/TRBV7-6) were found to dock over pMHC with crossing angles of 29°and 61°, respectively, with C7-HLA-A2 interactions dominated by V␣ and C25-HLA-A2 interactions dominated by V␤ (34). Here we have shown that F50 (TRAV13-1/TRBV27) engages GIL-HLA-A2 in a decidedly different orientation than do canonical TCRs JM22 (TRAV27/TRBV19), F6 (TRAV27/TRBV19), LS01 (TRAV24/TRBV19), and LS10 (TRAV38 -2/TRBV19) (14,23,24). This binding mode is characterized by a more acute crossing angle and focus on the C terminus rather than the center of the MHC ␣1 helix. In addition, the critical notch between the peptide and MHC ␣2 helix is occupied by a tryptophan rather than arginine residue as in most canonical GIL-specific TCRs. These TCR-pMHC structures, together with the remarkable diversity of TCRs expressing non-canonical TRAV/TRBV combinations in GIL-specific repertoires (14), demonstrate that there are many ways for TCRs to bind even a featureless peptide such as GIL with sufficient affinity to elicit broad antiviral responses that provide protection against T-cell clonal loss and viral escape.

Isolation of GIL-specific TCR F50
To obtain TCR F50, GIL-specific CD8 ϩ T cells were isolated from the peripheral blood of a HLA-A2 ϩ healthy male donor (71 years old) after 14 days of in vitro stimulation with GIL-HLA-A2 using an artificial antigen presenting system as previously described (14). Briefly, peripheral blood mononuclear cells were separated from leukopheresis cells by Ficoll gradient centrifugation. CD8 ϩ T cells were isolated from peripheral blood mononuclear cells by immunomagnetic enrichment (34,40). GIL-specific CD8 ϩ T cells were expanded in an artificial antigen-presenting system using GIL peptide (GILGFVFTL) (BioMer Technology) (41). The expanded cells were stained with APC-and FITC-conjugated GIL dextramer (Immudex). FITC and APC double-positive cells were sorted by flow cytometry. Single-cell analysis was used to identify the paired ␣ and ␤ chains of GIL-specific TCRs, including F50 (14,34).

Expression and purification of TCR F50
Soluble TCR F50 for affinity measurements and structure determination was prepared by in vitro folding from inclusion bodies produced in E. coli. The V␣ and V␤ regions of F50 (residues 1-209 and 1-244, respectively) were cloned into the expression vector pET26b (Novagen) containing C␣ and C␤ regions. An interchain disulfide (C␣Cys 163 -C␤Cys 171 ) was engineered to increase yield of the TCR ␣␤ heterodimer (42). The F50 ␣ and ␤ chains were expressed separately as inclusion bodies in BL21(DE3) E. coli cells (Agilent Technologies). Bacteria were grown at 37°C in LB medium to A 600 ϭ 0.6 -0.8 and induced with 1 mM isopropyl-␤-D-thiogalactoside. After incubation for 3 h, the bacteria were harvested by centrifugation and resuspended in 50 mM Tris-HCl (pH 8.0) containing 0.1 M NaCl and 2 mM EDTA; cells were disrupted by sonication. Inclusion bodies were washed extensively with 50 mM Tris-HCl (pH 8.0) and 5% (v/v) Triton X-100 and then dissolved overnight in 8 M urea, 50 mM Tris-HCl (pH 8.0), 10 mM EDTA, and 10 mM DTT. For in vitro folding, the TCR ␣ and ␤ chains were mixed in a 1.2:1 molar ratio for 30 min prior to dilution into ice-cold folding buffer containing 5 M urea, 0.4 M L-arginine HCl, 100 mM Tris-HCl (pH 8.0), 5 mM EDTA, 3.7 mM cystamine, and 6.6 mM cysteamine to a final protein concentration of 80 mg/liter. The folding mixture was dialyzed against 10 mM Tris-HCl (pH 8.0) for 72 h at 4°C. The mixture was concentrated 20-fold and dialyzed against 25 mM Tris-HCl (pH 8.0). Disulfide-linked TCR F50 heterodimers were purified using sequential Superdex 200 GL and Mono Q columns (GE Healthcare).

Crystallization and data collection
TCR F50 was mixed with GIL-HLA-A2 in a 1:1 molar ratio and concentrated to 10 mg/ml. Crystals of the F50 -GIL-HLA-A2 complex grew in 10 -15% (w/v) polyethylene glycol 3350, 0.1 M imidazole (pH 8.0), and 0.2 M sodium malonate. For data collection, crystals were transferred to a cryoprotectant solution of mother liquor containing 25% (v/v) glycerol prior to flash cooling in a nitrogen stream. X-ray diffraction data for the F50 -GIL-HLA-A2 complex were collected at Beamline 24ID-E of the Advanced Photon Source of the Argonne National Laboratory with an ADSC Q315 CCD detector. Diffraction data were indexed, integrated, and scaled with the program HKL2000 (43). The data collection statistics are summarized in Table 1.

Structure determination and refinement
The structure of the F50 -GIL-HLA-A2 complex was solved by molecular replacement with the program Phaser (44). An HIV-specific TCR (Protein Data Bank accession code 3VXU) (28) and GIL-HLA-A2 (Protein Data Bank accession code 1OGA) (23) were used as search models with the CDRs and peptide removed, respectively. One complex molecule was located in the asymmetric unit. Structure refinement was performed using rigid body and simulated annealing via Phenix (45). The model was further refined by manual model building with Coot (46) based on 2F o Ϫ F c and F o Ϫ F c maps with the GIL peptide omitted in the initial refinement. The final R work and R free values for the F50 -GIL-HLA-A2 complex are 19.1 and 21.3%, respectively. Refinement statistics are presented in Table 1. Stereochemical parameters were evaluated by PRO-CHECK (47).

Surface plasmon resonance analysis
The interaction of TCR F50 with GIL-HLA-A2 was assessed by SPR using a BIAcore T100 biosensor at 25°C. Biotin-tagged GIL-HLA-A2 (NIH Tetramer Core Facility) was immobilized on a streptavidin-coated BIAcore SA chip (GE Healthcare) at 1000 resonance units (RU), followed by blocking the remaining streptavidin sites with 20 M biotin solution. An additional flow cell was injected only with free biotin to serve as a blank control. For analysis of TCR binding, solutions containing different concentrations of F50 were flowed sequentially over the chips immobilized with GIL-HLA-A2 and the blank. Injections of TCR were stopped at 30 s after SPR signals reached a plateau. Equilibrium data were fitted with a 1:1 binding model using BIAevaluation 3.1 software to obtain the K D .

Protein Data Bank accession code
Coordinates and structure factors for the F50 -GIL-HLA-A2 complex have been deposited in the Protein Data Bank under accession code 5TEZ.
Author contributions-X. Y. determined the crystal structure. G. C. isolated TCR genes. X. Y., G. C., N.-P. W., and R. A. M. analyzed the data and wrote the manuscript.