Hard wiring of T cell receptor specificity for the major histocompatibility complex is underpinned by TCR adaptability

Abstract

αβ T cell receptors (TCRs) are genetically restricted to corecognize peptide antigens bound to self-major histocompatibility complex (pMHC) molecules; however, the basis for this MHC specificity remains unclear. Despite the current dogma, evaluation of the TCR–pMHC-I structural database shows that the nongermline-encoded complementarity-determining region (CDR)-3 loops often contact the MHC-I, and the germline-encoded CDR1 and -2 loops frequently participate in peptide-mediated interactions. Nevertheless, different TCRs adopt a roughly conserved docking mode over the pMHC-I, in which three MHC-I residues (65, 69, and 155) are invariably contacted by the TCR in one way or another. Nonetheless, the impact of mutations at these three positions, either individually or together, was not uniformly detrimental to TCR recognition of pHLA-B*0801 or pHLA-B*3508. Moreover, when TCR–pMHC-I recognition was impaired, this could be partially restored by expression of the CD8 coreceptor. The structure of a TCR–pMHC-I complex in which these three (65, 69, and 155) MHC-I positions were all mutated resulted in shifting of the TCR footprint relative to the cognate complex and formation of compensatory interactions. Collectively, our findings reveal the inherent adaptability of the TCR in maintaining peptide recognition while accommodating changes to the central docking site on the pMHC-I.

During thymic selection αβ T cell receptors (TCRs) are selected for weak reactivity with one or more self-peptides in complex with self-MHC (major histocompatibility complex), resulting in mature T cells being restricted to recognize processed peptides only when they are presented by self-MHC molecules (1). The exquisite specificity of TCR–pMHC binding must address the highly polymorphic nature of the MHC and variable peptide cargo. The antigen-recognition site of the TCR is made up of six complementarity-determining regions (CDRs), three each from the α and β chains (2). Currently, it is postulated that the CDR1 and 2 loops underpin the specificity or “bias” toward the MHC, whereas the CDR3 loops recognize the diverse range of bound peptides (3–5). Recent studies support the view that the T cell repertoire is intrinsically biased toward the MHC through contacts mediated by conserved TCR Vα and Vβ residues (6). However, the factors governing MHC restriction, a central paradigm of antigen-specific T cell immunity, remain unclear.

Presumably as a consequence of the polymorphic pMHC landscape, and the inherent diversity of the responding T cell repertoire, structural studies have shown that the TCR engages the pMHC-I and pMHC-II surfaces in a range of different docking modes (2). Nevertheless, conserved pairwise interactions between closely related Vβ8.2⁺ TCRs and pMHC-II molecules were identified, leading to the concept that there are defined interaction “motifs” or “codons” between the Vα and/or Vβ domains of a TCR and a given MHC allotype (3, 6–8). Central to the “interaction codon” hypothesis is that the CDR1 and CDR2 loops contact the MHC, whereas the CDR3 loops “read-out” the peptide leading to an inherent MHC bias of TCR recognition. Consistent with this perspective, mutations in the germline-encoded CDR2β region within Vβ8.2⁺ TCRs impact negatively on thymic selection reflecting a central role for these interactions in MHC recognition (9). However, associated biophysical, thermodynamic, and mutational studies have also shown that the relative energetic contributions from the CDR loops can vary between the various TCR–pMHC systems (10, 11). Accordingly, it is also considered that extrinsic factors, such as the CD8 coreceptor, play a principal role in MHC restriction, which is consistent with recent observations from CD8⁻ CD4⁻ mice (12, 13). Nevertheless, in the absence of the CD8 or CD4 coreceptor, a rough docking mode between the TCR–pMHC is preserved, so that for pMHC-I, the Vα and Vβ domains are positioned over the MHC α2- and α1-helices, respectively (2), suggesting some hidden logic in TCR–pMHC-I engagement. In searching for a central structural logic to TCR–pMHC-I bias we have observed that, in all known TCR–pMHC-I structures, the TCR makes contact with three MHC positions (65, 69, and 155—termed the “restriction triad”), suggesting that these may represent a minimal docking framework to enable MHC-I restriction (14).

Here, we have investigated the underlying basis of TCR–MHC-I specificity by testing the role of the restriction triad as a potential mandate for TCR–MHC-I interaction, and our results challenge some of the assumptions that currently shape our understanding of this interaction.

Results

Mining the TCR–pMHC-I Structural Database.

Presently, it is considered that most of the binding interface (>75%) of TCR–pMHC-I interactions involves contact between the germline-encoded CDR1/2 loops and the MHC and that the CDR3 loops primarily contact the peptide (3). Accordingly, we analyzed the 18 unique TCR–pMHC-I structures to calculate the relative contributions that the different CDR loops made in contacting the pMHC-I (Table 1). Our analyses included TCRs that recognized self and microbial peptides bound to MHC-I.

Table 1.

Relative contribution of the complementarity determining regions (CDR loops) to the binding of the MHC helices and the peptide

				% total BSA of TCR-MHC interaction						% total BSA of TCR-peptide interaction
PDB	MHC	Peptide	TCR	CDR1α	CDR2α	CDR3α	CDR1β	CDR2β	CDR3β	CDR1α	CDR2α	CDR3α	CDR1β	CDR2β	CDR3β
1AO7	HLA-A2	LLFGYPVYV	A6	35	14	25	0	0	26	28	0	21	5	0	46
1BD2	HLA-A2	LLFGYPVYV	B7	30	23	21	0	13	13	27	0	18	2	0	53
2BNR	HLA-A2	SSLMWITQC	1G4	4	19	24	11	30	12	14	0	36	10	0	40
3GSN	HLA-A2	NLVPMVATV	RA14	17	13	17	6	20	27	39	0	20	7	0	34
1LP9	HLA-A2	ALWGFFPVL	AHIII 12.2*	22	13	26	6	14	19	0	0	50	9	1	40
3HG1	HLA-A2	ELAGIGLTV	MEL5	21	15	19	5	14	26	25	7	8	0	0	60
1MI5	HLA-B*0801	FLRGRAYGL	LC13	16	18	23	3	21	19	23	4	27	0	0	46
1OGA	HLA-A2	GILGFVFTL	JM22	11	10	10	3	39	27	11	28	31	0	0	30
2AK4	HLA-B*3508	LPEPLPQGQLTAY	SB27	13	9	39	0	9	30	4	0	39	42	3	12
2ESV	HLA-E	VMAPRTLIL	KK50.4	20	5	19	3	36	17	6	0	45	8	11	30
2NX5	HLA-B*3501	EPLPQGQLTAY	ELS4	12	16	22	15	16	19	5	0	21	27	0	47
3DXA	HLA-B*4405	EENLLDFVRF	DM1	25	19	21	5	19	11	16	1	33	11	0	39
3KPS	HLA-B*4405	EEYLQAFTY	LC13	16	14	22	5	21	22	22	0	30	0	0	48
3FFC	HLA-B*0801	FLRGRAYGL	CF34	16	13	19	1	13	38	0	0	18	18	23	41
2CKB	H-2Kb	EQYKFYSV	2C	24	17	15	13	22	9	26	0	25	33	7	9
2OI9	H-2Ld	QLSPFPFDL	2C	19	27	15	10	25	4	19	0	18	24	9	30
1KJ2	H-2Kb	KVITFIDL	scKB5-C20	8	13	30	17	12	20	7	0	38	8	10	37
1NAM	H-2Kb	RGYVYQGL	scBM3.3	17	13	24	29	0	17	6	0	30	0	0	64
Average				18	15	22	7	18	20	16	2	30	12	4	41

BSA was calculated with the PDBePISA Web site, www.ebi.ac.uk.

*AHIII12.2 is a mouse TCR, xenoreactive complex.

Our analyses showed that, on average, the CDR1/2 loops contributed only ≈60% of the buried surface area (BSA) upon contacting the MHC, indicating that the CDR3 loops, on average, contribute ≈40% of the BSA in contacting the MHC (Fig. S1 and Table 1). In only two systems (2C TCR-H2-K^b and 2C TCR-H2-L^d), the CDR1/2 loops played a principal role (>75%) in contacting the MHC, whereas in the other 16 TCR–pMHC-I complexes, the CDR1/2 loops played a lesser role (<75%) in contacting the MHC. For example, in six systems (LC13 TCR-HLA-B*0801, LC13 TCR-HLA-B*4405, ELS4 TCR-HLA-B*3501, scBM3.3 TCR-H2-K^b, AHIII 12.2 TCR-HLA-A2, and MEL5 TCR-HLA-A2) the CDR1/2 loops contributed 51–60% of the MHC-specific contacts, and in four systems (A6 TCR-HLA-A2, SB27 TCR-HLA-B*3508, CF34 TCR-HLA-B*0801, and scKB5-C20 TCR-H2-K^b), the contributions made by the CDR1/2 loops were only 50% or less, mandating a greater role of the CDR3 loops in contacting the MHC. For example, in the CF34 TCR-HLA-B*0801 complex, the CDR3 loops contribute 57% of the BSA when contacting the MHC-I. Conversely, in relation to the CDR-peptide interactions, the CDR3 loops, on average contributed ≈70% of the BSA when contacting the peptide. The CDR3 loops played a major role (>75%) in contacting the peptide in 6 complexes (AHIII 12.2 TCR-HLA-A2, 1G4 TCR-HLA-A2, KK50.4 TCR-HLA-E, LC13 TCR-HLA-B*4405, scKB5-C20 TCR-H2-K^b, and scBM3.3 TCR-H-2K^b), and thus in the remaining 12 TCR–pMHC-I complexes the CDR1/2 loops played a greater role (>25%) in contacting the peptide. For example, in three systems (SB27 TCR-HLA-B*3508, CF34 TCR-HLA-B*0801, and 2C TCR-H2-K^b), the contributions from the CDR1/2 loops in contacting the peptide were >40%. These observations support the notion that the CDR3 loops can contribute notably toward contacting the MHC, whereas the germline-encoded CDR1 and -2 loops can play an important role in contacting the peptide.

Restriction Triad.

Analysis of the full TCR–pMHC-I structural database showed that the three MHC positions, 65, 69, and 155 were virtually invariably involved in making TCR interactions (Table S1). These three MHC-I positions exhibit limited polymorphism across all human HLA-Ia molecules, lending support that they provide elements of a docking footprint (termed the restriction triad) that underscores the TCR–pMHC-I interaction (14). Although these three MHC-I positions are always contacted by the TCR, the details of the interactions vary in each case with different bonding properties and different CDR loop contacts (15). Nevertheless it was notable that residues located in the CDR3 loops make the vast majority of contacts with the restriction triad (Table S1).

We set out to examine the relative energetic contribution of the residues within the restriction triad toward TCR–pMHC-I binding. We generated a series of single (Q65A, T69A, and Q155A) and triple (Q65, T69, and Q155→A) alanine mutations within the MHC-I and analyzed their effect on TCR binding using surface plasmon resonance (SPR) in two different systems (LC13 TCR-HLA-B*0801 and SB27 TCR-HLA-B*3508) (16, 17) (Fig. 1 and Table S2). The LC13 TCR recognizes the EBV-derived peptide FLRGRAYGL (referred to as FLR) bound to HLA-B*0801 (18), and although the Q65A mutation had no significant effect on binding, the Q155A and T69A mutants resulted in a 2- to 10-fold decrease in affinity (Fig. 1A). In addition the binding of the LC13 TCR to HLA-B*0801^FLR was completely abolished when all three restriction triad residues were mutated to alanine (Fig. 1A). Next, we investigated the importance of these three MHC-I residues in the SB27 TCR interaction with the 13-mer viral peptide LPEPLPQGQLTAY (referred to as LPEP) bound to HLA-B*3508 (14). Mutating the residues at positions 65 and 69 had no significant effect on TCR binding affinity, whereas the Q155A mutant reduced the affinity to 50 μM (Fig. 1B). Paradoxically, the triple alanine mutation involving the complete restriction triad caused only a 2-fold decrease in affinity compared with wild type (Fig. 1B). The capacity of the SB27 TCR to tolerate mutations at these three positions was surprising, given that our earlier work had shown that this TCR made limited contacts with HLA-B*3508 including the three positions (65, 69, and 155) that represent conserved contact points in the TCR–pMHC-I interaction.

Fig. 1.

The role of the “restriction triad” residues in TCR–pMHC-I binding as judged by SPR analysis and CD69 up-regulation. (A) Equilibrium binding of the LC13 TCR to HLA-B*0801^FLR (WT), or mutant HLA-B*0801^FLR with a Q65A, a T69A, or a Q155A substitution, or the triple restriction triad mutant of HLA-B*0801^FLR. (B) Equilibrium binding of the SB27 TCR to HLA-B*3508^LPEP (WT), or mutant HLA-B*3508^LPEP with a Q65A, a T69A, or a Q155A substitution, or the triple restriction triad mutant of HLA-B*3508^LPEP. Data are representative of at least four independent experiments. (C) CD69 up-regulation within Jurkat cells transfected with the LC13 TCR or (D) the LC13 TCR and CD8. The experiments were conducted at least twice with similar results.

T Cell Activation.

We next investigated the effect of these restriction triad mutations on T cell activation. We analyzed CD69 up-regulation on the Jurkat cell line, transfected with either the LC13 TCR (19) or the LC13 TCR and CD8, following stimulation with antigen-presenting cells treated with various concentrations of the FLRGRAYGL peptide. Stimulator cells were transfected derivatives of the antigen-processing mutant T2 cell line expressing wild-type and mutant HLA-B*0801. The mutants contained alanine substitutions at position 65, 69, or 155, or at all three of these positions. Surface expression of mutant HLA-B*0801 molecules was assessed by overnight peptide-stabilization assays, and cells were matched for MHC-I surface expression such that the T2 transfectants expressing HLA-B8 and its mutants were equivalent in their surface expression.

For the LC13 activation, the Q65A mutation had no significant effect on CD69 up-regulation, regardless of whether CD8 was present (Fig. 1C). T cell activation was impaired with the T69A and Q155A mutants (T69A > Q155A) and interestingly, the presence of CD8 substantially rescued the ability of the LC13 TCR to up-regulate CD69 when stimulated by the T69A or Q155A mutants of HLA-B*0801^FLR. Similarly, when all three restriction triad residues were mutated to alanine, T cell activation was dramatically reduced, and the presence of CD8 only partially rescued T cell activation (Fig. 1D). Accordingly, as judged by SPR and CD69 up-regulation, the LC13 TCR-HLA-B*0801^FLR interaction was dependent on one or more residues within the restriction triad; nevertheless, the impact of these restriction triad mutations was partially rescued by the CD8 coreceptor.

Cytotoxicity Assays.

To examine whether the restriction triad is required for T cell cytotoxicity, we performed chromium-release assays with a panel of independent HLA-B*0801- or HLA-B*3508-restricted T cell clones from different individuals and targeted to different epitopes. Firstly, we examined CTL recognition of HLA-B*0801^FLR by 12 different T cell clones, as well as two CTL clones directed against the HLA-B*0801-restricted RAKFKQLL (referred to as RAK) epitope (20). Secondly, we examined the effect of the mutations on a number of clones targeted to one of four HLA-B*3508-restricted viral epitopes of different length: a 13-mer (LPEPLPQGQLTAY), a 12-mer (CPSQEPMSIYVY), an 11-mer (HPVGEADYFEY), or a 10-mer (APQPAPENAY) (21–24).

Consistent with the SPR and CD69 up-regulation assays, lysis by the LC13 CTL clone was greatly inhibited by the T69A, Q155A, and the triple restriction triad mutations (Fig. 2A). Moreover, six other HLA-B*0801^FLR-restricted CTL clones (SBD11, PP36, SBD12, RL29, BK10, and RL16) exhibited markedly diminished CTL killing when either one or all of the restriction triad residues were mutated to alanine (loss-of-function hierarchy triple Ala > 155 > 69). Notably however, some CTL clones displayed much less dependency on the residues comprising the restriction triad, even to the point that CTL killing of target cells expressing the triple Ala restriction triad mutant of HLA-B*0801^FLR still occurred for five CTL clones (JL9, CFA27, GL3, CFB40, and CFB14). To test whether this latter observation was peculiar to the HLA-B*0801^FLR epitope, we also examined CTL recognition of the HLA-B*0801^RAK determinant. One CTL clone (LC9) was shown to be dependent upon the restriction triad (triple Ala or 155 > 69 > 65), whereas cell lysis by another CTL clone (LC1) was only modestly impacted by single mutations and unaffected by the triple Ala mutations within the restriction triad (155 ≥ 65, Fig. 2B).

Fig. 2.

The impact of mutation at MHC α-helix positions 65, 69, and/or 155 on recognition by HLA-B*0801-restricted CTL clones. The antigen-processing mutant T2 cell line was used as a target in chromium-release cytotoxicity assays, either without transfection (see vertical axis: No HLA-B*0801) or transfected to express wild-type HLA-B*0801 (WT HLA-B*0801), or mutant HLA-B*0801 molecules with alanine substitution at position 65, 69, or 155, or at all three of these positions. These target cells were treated for 1 h with various concentrations (10⁻⁶ M, 10⁻⁷ M, 10⁻⁸ M, 10⁻⁹ M, and 10⁻¹⁰ M) of the HLA-B*0801-restricted EBV peptides FLRGRAYGL (A) or RAKFKQLL (B), and washed before CTL addition. CTL clones specific for each of these peptides were used as effectors (E:T = 1:1), and the concentration of peptide required for half maximal lysis was calculated from the dose–response curves. (A) The FLRGRAYGL-specific CTL clones were LC13 (34, 35), JL9, SBD11, CFA27, PP36, SBD12, GL3, RL29, CFB40 (34), CFB14, BK10 (36), and RL16 (34, 37). These experiments were independently performed twice for clones LC13, GL3, and CFB40, with analogous results on each occasion. Furthermore, the CFB34 CTL clone, which shares an identical TCR with CFB40, gave similar results to those shown for CFB40. The other clones were tested in a single experiment, using duplicate samples. (B) The RAKFKQLL-specific CTL clones were LC9 and LC1, which have not been described previously. These clones were tested in a single experiment, using duplicate samples.

Just as the SPR data showed that the SB27 TCR could still bind well to the triple Ala restriction triad mutant of HLA-B*3508^LPEP, SB27 CTL killing of APC-expressing triad mutations was indistinguishable from killing of wild-type cells. Additionally, two other CTL clones specific for this 13-mer peptide (SB9 and CA5) with different TCRs (14, 16, 23) were insensitive to the restriction triad mutations (Fig. S2). Similarly, some CTL clones directed against the 12-mer, 11-mer, and 10-mer peptides could tolerate mutations in the restriction triad. However, dependency on the triad residues was notable in some of the CTL clones (SB10, SB14, SB8, and SB16) that were targeted toward the shorter HLA-B*3508-restricted epitopes and that were particularly sensitive to mutation of residue 155 (Fig. S2 B–D).

Accordingly, against the panel of 24 CTL clones, there was large variation in the dependency on the restriction triad in enabling CTL killing.

Compensatory Interactions.

We aimed to gain an understanding of how a TCR could tolerate mutations in the three restriction triad MHC-I-contacts points. We reasoned that the SB27 TCR-HLA-B*3508^LPEP complex would represent an ideal system to investigate this, as the SB27 TCR made limited contacts with the cognate HLA-B*3508—namely, interacting with only two residues on the α1-helix (65 and 69) and a focused series of contacts on the α2-helix (150-151, 154-155, 157-159, and 163) (14). Accordingly, we solved the structure of the SB27 TCR bound to HLA-B*3508 in which the restriction triads were all mutated to alanine (referred to as HLA-B*3508-AAA^LPEP) (Table S3). We also solved the structure of HLA-B*3508-AAA^LPEP in the nonliganded state to 2.2 Å resolution (Table S3) to ascertain whether the mutations impacted on conformation of the peptide within the Ag-binding cleft.

The HLA-B*3508-AAA^LPEP structure and the wild-type HLA-B*3508^LPEP structure (16) were very similar, indicating that the restriction triad mutation does not affect the conformation of the bound peptide. Similar to the SB27 TCR-HLA-B*3508^LPEP complex, the SB27 TCR perched atop the bulged peptide of the HLA-B*3508-AAA^LPEP complex. However the SB27 TCR-HLA-B*3508-AAA^LPEP complex shifted 8° with respect to how the SB27 TCR docked onto HLA-B*3508^LPEP (see SI Materials and Methods), resulting in the SB27 TCR tilting more toward the α2-helix when bound to HLA-B*3508-AAA^LPEP (Fig. 3A) (Table S4). The SB27 TCR made very few contacts with the MHC helices of HLA-B*3508-AAA^LPEP, with only 3 direct hydrogen bonds and two salt bridges being made exclusively with residues on the α2-helix (Fig. 3B). Although a number of van der Waal (vdw) contacts were made with the α2-helix, the interactions with the α1-helix were limited to one vdw contact with Arg-62. In comparison with the SB27 TCR-HLA-B*3508^LPEP complex, in the HLA-B*3508-AAA^LPEP complex SB27 loses two direct hydrogen bonds to Gln-65 and Gln-155, and two vdw interactions with residues Gln-65 and Thr-69. However, these interactions are compensated in this complex, by additional H-bonds to Glu-166 and to Ala-158, as well as vdw interactions with Arg-62, Ala-149, Glu-161, and Gly-162 (Fig. 3 B and C). Consequently, when compared with the SB27 TCR-HLA-B*3508^LPEP complex, the SB27 TCR buries an additional ≈150 Ų when binding to the HLA-B*3508-AAA^LPEP.

Fig. 3.

The SB27 TCR compensates for mutations in the MHC via “slippage” of the TCR footprint on the MHC. (A) Superposition of SB27 TCR on WT MHC (orange) compared with the position of the TCR on the “restriction triad triple mutant” MHC (cyan). (B) Contacts made by the SB27 TCR to HLA-B*3508-AAA^LPEP. The LPEP peptide is shown in dark gray; H bonds are shown as dashed black lines; salt bridges are shown as dashed red lines. (C) Contacts made by the SB27 TCR to the WT HLA-B*3508^LPEP. CDR1α, -2α, -3α, and -3β are colored purple, green, yellow, and orange, respectively.

Collectively, this demonstrates how the SB27 TCR can adapt to productively engage a pMHC-I ligand in the absence of the restriction triad residues of HLA-B*3508.

Discussion

The germline-encoded V genes of the TCR are thought to determine bias toward the MHC (4, 5), and consistent with this view there is evidence for preserved “interaction motifs” between the V-gene encoded regions of a TCR and a given MHC-II allotype (6, 7). This has only been reported within a narrow slice of the Vβ repertoire, where consensus Vβ8.2⁺ CDR2β-mediated interactions were observed with closely related pMHC-II molecules (6, 7). It is also suggested that the same V region can potentially interact with disparate pMHC molecules in a different register (3), as exemplified by the divergent docking modes of the Vβ8.2 2C TCR with H2-K^b and H2-L^d (25, 26). Further, given the influence of the peptide-TCR contacts in any given TCR–pMHC system, the binding energy contributed by the interaction motif can be critically supplemented to enable the TCR to bind in another “MHC register” (3, 27). Accordingly, the proposed role of interaction codons or motifs occurs on a backdrop of considerable “noise” in relation to the generality and exclusivity of their function in defining the TCR–pMHC orientation. Collective structural analyses raise additional questions about the generalization that the CDR1/2 loops dictate MHC-I bias. Namely, the current TCR–pMHC-I structural database demonstrates that the germline-encoded CDR1/2 loops frequently play an important role in contacting the peptide, but perhaps more critically, the CDR3 loops can and do play a principal role in contacting the MHC-I. This interpretation is further supported by mutagenesis experiments that demonstrated an important energetic role for the CDR3 loops in contacting the MHC-I and for the CDR1/2 loops in contacting the peptide (8, 11).

Whereas the TCR-MHC contacts are typically extensive, binding studies have shown that these interactions are governed by “energetic hotspots” (10, 11). Consistent with this finding, our analyses showed that three (65, 69, and 155) positions, on the MHC were invariably contacted in essentially all TCR–pMHC-I systems reported to date (14). Accordingly, we set out to establish whether these three positions represented the minimal docking framework required to enable MHC-I restriction in two HLA families, which differ from each other by 19 amino acids. Surprisingly, the dependency of the restriction triad varied among the different CTL clones examined. For example, whereas the LC13 TCR-HLA-B*0801^FLR interaction was wholly dependent on the restriction triad, 5 of another 12 CTL clones targeted toward HLA-B*0801^FLR tolerated mutations of the restriction triad. Similarly, the HLA B*3508-restricted CTL clones recognizing the 10-mer, 11-mer, and 12-mer peptides exhibited varied dependency on the restriction triad. However, all three CTL clones that recognize the HLA-B*3508^LPEP complex were somewhat independent of the restriction triad. The latter finding suggests that the peptide-centric recognition of CTL clones that recognize lengthy peptides could potentially outweigh any energetic contribution by the HLA-B*3508 molecule.

The structure of the SB27 TCR ligated to HLA-B*3508-AAA^LPEP showed how the SB27 TCR retained specificity for this altered pMHC-I landscape. Namely, the slippage of the SB27 TCR footprint enabled compensating interactions to be made with the MHC-I, simultaneously highlighting the inherent cross-reactivity of TCRs in interacting with different MHC ligands during alloresponses. These observations are consistent with other evidence that the TCR is finely tuned to alterations in the peptide Ag, yet remarkably tolerant to alterations in the MHC (28). It is conceivable that the energetic contributions from the restriction triad residues play a more consistent and critical role in TCR binding to weak affinity ligands during thymic positive selection. The impact of this dependency could then persist to a variable degree in mature T cells selected by specific antigenic peptides.

Whereas the αβ T cell repertoire is vast, many virus-specific immune responses are characterized by biased TCR usage (29, 30). Nevertheless, biases in the overall peripheral Vβ repertoire between unrelated individuals who share the same HLA alleles are rare (31), which argues against a general role for preferential pairwise TCR-MHC interaction motifs. Perhaps this is partly explained by the complexity of repertoire selection against up to 12 distinct MHC allotypes (assuming 6 heterozygous loci); however, one might still expect a general bias in Vβ selection across specific MHC allotypes and this has only been evident in twin studies (31). On the other hand, repeated selection of particular V regions and/or D/J and N-region genetic elements in antigen-specific responses indicates that these regions must encode exquisite interactions to enable TCR interaction with self-MHC complexed to specific peptides (30, 32). It is notable that, in the examples where TCRs that exhibit biased gene usage have been examined structurally (32), germline-encoded regions can often play a central role in interacting with the peptide determinant (Table 1). The evolution of the TCR V/D/J genes has undoubtedly been driven by infection by pathogens and in particular, the requirement to respond rapidly to viral peptides. It is therefore not unexpected that the CDR1/2 germline-encoded regions of the TCR might also frequently bind peptide as well as the presenting MHC-I molecule and that biased TCR V-gene use has frequently been described for particular pMHC complexes but rarely for particular MHC allotypes.

Hence, the TCR bias for binding to pMHC molecules is built on a suite of hard-wired interactions between the TCR and conserved MHC residues at positions 65, 69, and 155. Nonetheless, the clonal dependence upon these interactions can vary in the mature T cell repertoire such that TCR flexibility can still underpin MHC-recognition by substituting new compensatory contacts. We conclude that the conundrum of how TCRs restrict their interactions to pMHC (and MHC-like) molecules and not other molecular surfaces is surprisingly subtle.

Materials and Methods

Mining of the TCR–pMHC-I Database.

There are currently 33 TCR–pMHC-I structures in the Protein Data Bank database, which includes a series of very closely related TCR–pMHC-I complexes. Accordingly, one representative TCR–pMHC-I from each of these systems was used in the analysis, which amounted to 18 nonredundant TCR–pMHC-I complexes. The BSA interfaces for each TCR–pMHC-I complex was determined using PISA and, for verification, the BSA was also determined using AREAIMOL from the CCP4 program suite. Both PISA and AREAIMOL gave comparable results.

Protein Production and Crystallization.

The HLA-B*3508-AAA^LPEP and the SB27 TCR were expressed and purified and crystallized essentially as previously described (14, 33).

Data Collection, Structure Determination, and Refinement.

Data were collected using an in-house radiation source and/or the Industrial Macromolecular Crystallography Association–Collaborative Access Team beamline at the Argonne Photon Source Synchrotron in Chicago and/or at the MX2 beamline at the Australian Synchrotron. Data were processed using standard crystallographic software packages (Table S3). The structure of HLA-B*3508-AAA^LPEP was determined by molecular replacement using the HLA-B*3508^LPEP structure as the search model (16). For the SB27-HLA-B*3508-AAA^LPEP complex, the crystals belonged to the space group P2₁ with four complexes in the asymmetric unit. The structure was using the SB27-HLA-B*3508^LPEP complex as the search model. For both structures, subsequent model building was conducted using the COOT software followed by maximum-likelihood refinement (CCP4 suite). All structures were validated using standard software packages.

Surface Plasmon Resonance Measurement and Analysis.

All surface plasmon resonance experiments were conducted at 25 °C on the BIAcore 3000 essentially as previously described (11). All experiments were conducted in duplicate or quadruplicate.

T Cell Activation Assay.

Ag presenting cells (T2-B*0801 or T2-B*3508 transfectant cells, 10⁵/well) were cultured with peptide FLR or LPEP at the indicated concentration in 100 μL of AimV serum free-medium overnight at 26 °C for stablization of pMHC-I surface expression. Jurkat.LC13 or Jurkat.CD8.LC13 (10⁵/well) were added and cocultured for 4 h at 37 °C. The cells were then pelleted and assayed for CD69 up-regulation by staining and fluorescence-activated cell sorting. T cell activation was measured by the increase of the mean channel fluorescence in CD69 staining of the gated GFP positive Jurkat cells.

Cytotoxicity Assays.

CTL clones (Table S5) were tested in duplicate for cytotoxicity in a standard 5-h chromium-release assay. CTLs were assayed against ⁵¹Cr-labeled target cells at an effector:target ratio of 1:1. The target cells were the antigen-processing mutant T2 cell line, either without transfection or transfected to express either wild-type HLA-B*0801, or B*3508, or mutant HLA-B*0801/B*3508 molecules with alanine substitution at position 65, 69, or 155, or at all three of these positions. These target cells were treated for 1 h with various concentrations of HLA-B*0801/B*3508-restricted viral peptide epitopes and washed three times before CTL addition. Percent-specific lysis was calculated, and the concentration of peptide required for half maximal lysis was determined from the dose–response curves. Peptides were synthesized by Mimotopes. The mean spontaneous lysis for target cells in the culture medium was always <20%, and the variation about the mean-specific lysis was <10%.