Molecular Basis for the High Affinity Interaction between the Thymic Leukemia Antigen and the CD8αα Molecule¹

The T region of the mouse MHC encodes several nonclassical or nonpolymorphic class Ib proteins with diverse functions (1). The closely related T3 and T18 genes, with >95% sequence similarity, encode the thymic leukemia (TL)³ Ag, originally identified by serology as a determinant on thymic TLs (2). The TL Ag is also expressed on immature thymocytes of most mouse strains (3), activated peripheral T lymphocytes (4), activated dendritic cells (5), epithelial cells, and intraepithelial lymphocytes (IEL) of the small intestine (6, 7). Cell surface expression of the TL Ag is independent of the TAP but requires β₂-microglobulin (β₂m) (8). The crystal structure of the TL Ag shows that its binding groove is occluded by conformational changes that prevent the presentation of peptides (9). This corroborates the observation that no peptides have been identified in elution experiments with soluble TL expressed in insect cells (8) or bacteria (10). Despite the absence of Ag presentation, TL molecules can serve as transplantation Ags and can mediate a TCRγδ⁺ or even TCRαβ⁺ cytotoxic T cell responses (11, 12, 13, 14). This cannot be due to Ag presentation by TL, and in some cases it has been shown that TL Ag-reactive T cells respond to a conformational epitope in the TLα₁α₂ domains (15).

The homodimeric CD8αα molecule, expressed on the vast majority of IEL, was recently shown to bind the TL Ag. TL tetramers, encoded by either T18^d or T3^b constructs, selectively bind CD8αα, but not CD8αβ-expressing cells (16, 17). Moreover, direct binding measurements with monomeric TL protein demonstrated that TL binds CD8αα with an affinity ∼10-fold higher than the affinity for CD8αβ heterodimers (16). The TL Ag affinity for CD8αβ is low and comparable to those measured between other MHC class I molecules and CD8αβ or CD8αα (16, 18). TL is the only known MHC class I molecule that displays a significant difference in affinity favoring binding of CD8αα. Studies using IEL and CD8αα-expressing T cell hybridomas demonstrated that expression of TL on APCs modulates TCR-mediated cytokine release, cytotoxicity, and proliferation (16). Moreover, we have recently shown that a subpopulation of activated CD8αβ⁺ peripheral T lymphocytes transiently expresses CD8αα. The CD8αα interaction with TL, expressed on activated dendritic cells and perhaps other cell types, may promote the survival and differentiation of activated lymphocytes into memory T cells (5).

The molecular basis for the ability of the TL Ag to interact with CD8αα preferentially and with a relatively high affinity is not understood, especially as several of the major binding regions of classical MHC class I molecules for CD8αα are conserved in the TL Ag. To understand how the TL Ag binds preferentially to CD8αα to carry out its functions, and to gain insight into how other class I molecules might function similarly, we have conducted gain-and-loss of function experiments to define the mechanism for preferential interaction of the TL Ag with CD8αα homodimers.

A construct encoding the T18^d cDNA in the pET23a vector has been described elsewhere (10). A K^b cDNA in the pET23a vector, an HLA-A2*0201 cDNA in the pHN1 vector, and a construct containing a human β₂m cDNA were a gift from Dr. J. Altman (Emory University, Atlanta, GA) (19). Constructs encoding T18^dα₁α₂/K^bα₃, K^bα₁α₂/T18^dα₃, and HLA-A2α₁α₂/T18^dα₃ chimeric class I molecules for recombinant protein production in bacteria were generated by overlapping PCR extension (20). In brief, DNA encoding the T18^dα₁α₂, Kb^bα₁α₂, HLA-A2α₁α₂, T18^dα₃, and K^bα₃ domains were amplified by PCR using the primers depicted in Table I and were gel purified. To create the T18^dα₁α₂/K^bα₃ chimeric molecule, the T18^dα₁α₂ PCR fragment was mixed at 1:1 ratio with K^bα₃ PCR fragments and subjected to overlapping PCR. DNA fragments were denaturated at 95°C for 5 min, annealed at 56°C for 2 min, and extended at 72°C for 5 min. Finally T18^dα₁α₂ sense primers and K^bα₃ antisense primers were added and the mixture was subjected to a 25-cycle run. The final PCR product was cloned into a TOPO-TA cloning kit (Invitrogen Life Technologies) and positive clones were sequenced. Finally, the T18^dα₁α₂/K^bα₃ DNA was NdeI-BamHI digested from the TOPO-TA vector and cloned in frame with a biotin-binding cassette (BSP) at the 3′ end terminus into the pET23a vector as previously described (10). A similar procedure was applied to generate K^bα₁α₂/T18^dα₃ and HLA-A2α₁α₂/T18^dα₃ chimeric molecules. For the generation of T18^d recombinant proteins bearing mutations, the T18^d cDNA in the pET23a vector was subjected to site directed mutagensis using the GeneEditor kit (Promega) with the mutagenic primers depicted in Table I. Finally, for the generation of K^b recombinant proteins bearing mutations, site directed mutagenesis was performed using the Quick Change mutagenesis kit (Stratagene) with the mutagenic primers depicted in Table I.

The MHC class I H chain-biotin-binding cassette and β₂m proteins were expressed as inclusion bodies in Escherichia coli BL21(DE3) (Novagen) after induction with 0.4 M isopropyl-β-d-thiogalactopyranoside. Inclusion bodies were purified from bacterial lysates and solubilized in 8 M urea. After further dilution with buffered 6 M guanidine hydrochloride, MHC class I H chains and β₂m were refolded in 200 ml or 500 ml of refolding buffer (pH 8.3) containing 400 mM l-arginine, 100 mM Tris-HCl, 2 mM EDTA, 5 mM reduced glutathione, 0.5 mM oxidized glutathione, and 0.2 mM PMSF. A total of 1 μM MHC class I H chain, equally distributed in six separated injections (every 12 h) and 2 μM β₂m, distributed in two separated injections (performed with the two first injections of MHC class I H chain) were added to the refolding buffer. For K^b and K^bα₁α₂/T18^dα₃ molecules, OVAp (SIINFEKL) was added at a concentration of 26 μM to the refolding solution before the H chains and β₂m injections. For the refolding of the HLA-A2 and HLA-A2α₁α₂/T18^dα₃ chimeric molecules, a peptide derived from hepatitis B core Ag (HBcAg) (FLPSDFFPSV) was used. The refolded proteins were then filtered through a 0.22-μm filter (Millipore) and concentrated in an Amicon chamber with a 10K-exclusion membrane and separated from aggregated H chain and free human β₂m on a S300 gel filtration column (Pharmacia Biotech). Refolded MHC class I molecules were concentrated using a Centricon Filter (Amicon) with a 30K-exclusion membrane. Sephadex columns (Amersham Biosciences) were used for desalting before the refolded β₂m/MHC class I complexes were subjected to an enzymatic biotinylation using AviTag technology (Avidity) for 12 h at 25°C. MHC class I tetramers were prepared by mixing the biotinylated protein with streptavidin-PE (Molecular Probes) at a molar ratio of 4:1 for 15 min at 4°C.

The concentrations of biotinylated, monomeric MHC class I molecules were measured using the BCA protein assay (Pierce). A 1/10 to 1/30,000 dilution of the streptavidin-PE-labeled tetramers was made and used to stain 10⁵ beef insulin (BI)-CD8αα cells for 30 min at room temperature. These cells are from the BI-reactive T cell hybridoma BI-141 transfected with a CD8α expression construct (a gift of Dr. C. Micelli, University of California, Los Angeles, CA). The mean fluorescence intensity (MFI) of the tetramer staining on BI-CD8αα was measured by flow cytometry using a FACScan instrument (BD Biosciences). As a control, BI-CD8αα cells were preincubated with an anti-CD8α mAb RM2200 (Caltag Laboratories), which blocks the T18^d-CD8αα interaction, before the tetramer staining. The specific MHC class I tetramer staining for CD8αα on the BI-CD8αα cells was calculated as follows: (MFI in the absence of anti-CD8α mAb) − (MFI in the presence of anti-CD8α mAb). The binding isotherm of the different MHC class I tetramers for CD8αα was obtained by plotting the specific MFI against the tetramer concentrations. Scatchard transformation of the binding isotherms was used to calculate the K_D.

C57BL/6J (C57BL/6) were purchased from The Jackson Laboratory. Inguinal lymph nodes were isolated and prepared as single cell suspension. IEL were isolated as follows: small intestine was removed and cleaned, opened longitudinally, and fecal content was removed. Intestines were then cut into 0.5-cm pieces, transferred into a 50-ml Falcone tube and shaken two times at 250 rpm for 20 min at 37°C in HBSS without Ca²⁺ or Mg²⁺ and containing 1 mM DTT and 2% FCS. The cell suspensions were passed through a 60-μl nylon mesh and pelleted by centrifugation. The cell pellets were resuspended in 40% Percoll layered over a 70% Percoll and centrifuged at 900 × g for 20 min. Cells from the interface were collected and washed.

A construct encoding a T18^d full-length cDNA in the pCDNA3.1 vector (Invitrogen Life Technologies) was described elsewhere (8). The K^b full-length cDNA in the pCDNA3.1 vector was a gift from Dr. P. Creswell (Yale University, New Haven, CT). DNA constructs encoding T18^dα₁α₂/K^bα₃ and T18^dα₁α₂/K^bα₃-D197G/K198D/M228T for generating mouse cell transfectants were generated by overlapping PCR using the same protocol as for generating recombinant proteins (see above), and using the primers described in Table I. Constructs encoding T18^dα₁α₂/K^bα₃ chimeric molecules, mutated T18^d and mutated K^b were sequenced and electroporated into RMA-S and P815 cells using the Gene Pulser II (Bio-Rad) electroporator and a 0.4-cm cuvette (Invitrogen Life Technologies). Neither cell line expresses TL based on FACS analysis. Transfected cells were then kept in culture for 10 days in complete RPMI 1640 with 100 μg/ml geneticin (Invitrogen Life Technologies) and then FACS sorted using anti-T18^δ mAb 18/20 (21) or anti-H-2K^b mAb, clone MM3604 (Caltag Laboratories), to enrich for transfectants.

Activation of the CD8αα-expressing T cell hybridomas B3Z and 30NX was evaluated as described previously (22). Briefly, transfected RMA-S or P815 cells were incubated with mitomycin C and after extensive washing, were resuspended in HL-1 serum-free medium (Cambrex) and distributed in 96-well plates (2 × 10⁵ cells/well). B3Z or 30NX TCR agonist peptides, SIINFEKL and SSVVGVWYL, respectively, were added to the cells and the plate was incubated for 30 min at 37°C. Finally, 7 × 10⁴ B3Z or 30NX T hybridoma cells (a gift of Dr. N. Shastri, University of California, Berkeley) were added to each well and the plate was incubated for 12 h at 37°C. The plate was washed two times with PBS and 100 μl of Z buffer (100 mM 2-ME, 9 mM MgCl₂, 0.125% Nonidet P-40, 15 mM chlorophenol red-β-d-galactopyranoside in PBS) was added to each well. The plate was incubated for 1–4 h at 37°C and β-galactosidase enzymatic activity was measured by spectrophotometry (Spectra Max; Molecular Devices) at 595 nm.

To assess directly the importance of the T18^dα₃ domain in the interaction with CD8αα, we generated T18^d, K^b, and HLA-A2 chimeric recombinant proteins from bacterial inclusion bodies (10) by swapping their constituent α₃ domains to form T18^dα₁α₂/K^bα₃, K^bα₁α₂/T18^dα₃, and HLA-A2α₁α₂/T18^dα₃ molecules, respectively. Binding of the chimeric proteins to CD8αα was evaluated by flow cytometry with PE-labeled tetramers that were used to stain CD8αα-transfected BI-141 T hybridoma cells (BI-CD8αα) (23) (Fig. 1,A). As previously published, wild-type (wt) T18^d tetramers stain BI-CD8αα (16) with an avidity of 0.07 μM, measured by a Scatchard transformation of the binding isotherm (Fig. 1,B) and staining could be blocked by preincubating the BI-CD8αα cells with an anti-CD8α mAb before adding the tetramer (Fig. 1 A). As expected, only very weak staining was detectable when K^b and HLA-A2 tetramers were tested on the BI-CD8αα cell line. Because of the weak staining, the avidity of these two proteins for mouse CD8αα could not be measured by a binding isotherm. T18^dα₁α₂/K^bα₃ chimeric tetramers were likewise unable to stain the BI-CD8αα cells, whereas K^bα₁α₂/T18^dα₃ and HLA-A2α₁α₂/T18^dα₃ chimeric tetramers were able to do so, with calculated binding constants of 0.10 μM and 0.53 μM, respectively. These results demonstrate that the T18^dα₃ domain is necessary for the high affinity interaction with CD8αα.

To determine which amino acids might be important, we focused on those located in the contacting α₃ domain for CD8α. Structural studies (9) clearly delineated that the CD loop and AB loop of the α₃ domain play a pivotal role in CD8α binding (Fig. 2). There are differences in the amino acid composition of these loops between the TL Ag and classical MHC class I. In the CD loop, position 228 is a threonine (T228) in T18^d and a methionine (M228) in K^b. We have previously suggested that an extra hydrogen bond between the hydroxyl oxygen of T228 to the carbonyl group of L224 within the CD loop of TL might further rigidify the loop, favoring CD8 binding from an entropic point of view (9). In the AB loop of the T18^d molecule, positions 197 and 198 are a glycine (G197) and an aspartic acid (D198) whereas in K^b they are an aspartic acid (D197) and a lysine (K198). One key factor that may augment CD8α-binding affinity is the D197G substitution. In TL, the G197 has its carbonyl oxygen hydrogen bonding to the R194 side chain atom NH1, inducing a readjustment of the R194 side chain so that its NH2 atoms now form a hydrogen bond to the main chain carbonyl oxygen of N61 of CD8α₂. The resulting overall consequence is to bring the AB loop of T18^d closer to the neighboring C′C″ loop of CD8 as compared with the K^b loop, enhancing the binding. The net gain of this closer approach is four hydrophobic contacts in the interaction of CD8αα with T18^d as compared with K^b.

Based upon these observations, we made a series of mutations in which K^b amino acids were substituted into the T18^d molecule or vice versa. A substitution of T18^d aspartic acid at position 198 with a lysine (D198K) reduced the tetramer staining on BI-CD8αα (Fig. 3,A), and the K_D from 0.07 μM for the wt T18^d (Fig. 1,B) to 0.58 μM for the T18^d-D198K mutant (Fig. 3,B). A T18^d molecule in which positions 198 (D198K) and 197 (G197D) were substituted to give the double mutant T18^d-G197D/D198K had an equilibrium binding constant of 0.2 μM (Fig. 3,B). The substitution of the threonine at position 228 by a methionine (T18^d-T228M) did not significantly affect the binding to CD8αα (K_D = 0.11 μM) (Fig. 3 B). In contrast, a triple substitution replacing three T18^d residues with K^b residues (G197D/D198K/T228M) caused a 16-fold decrease in CD8αα binding (K_D = 1.1 μM).

In a second set of mutants, we determined whether there would be a gain of CD8αα-specific affinity by substituting the α₃ domain TL amino acids, identified above, into the K^b molecule. A K^b MHC tetramer with a single substitution M228T had greatly reduced binding to COS7 cells expressing high levels of CD8αα compared with wt K^b tetramer (data not shown). Therefore, we did not continue studies with the single substitution. We created a D197G/K198D double substitution and a D197G/K198D/M228T triple substitutions in the K^b molecule and made affinity measurements. The double substitution did not lead to a detectable gain of affinity. Remarkably, however, a K^b molecule with the triple T18^d substitutions displayed a relatively high avidity K_D of 0.2 μM for CD8αα (Fig. 3 B).

To formally demonstrate that the gain of affinity of the D197G/K198D/M228T K^b molecule was unique for CD8αα and not for CD8αβ, we stained CD8αβ lymph node cells from a wt mouse. wt T18^d and K^b tetramers, as well as T18^d and K^b triple mutant tetramers, were unable to stain CD8αβ lymph node T cells (Fig. 4). In contrast, K^b triple mutant tetramers were able to stain IEL, though the staining was less bright than wt T18^d tetramers (Fig. 4). This observation demonstrates that the increase in avidity of the mutated K^b tetramers for CD8αα is not restricted to cell lines but also applies for ex vivo isolated lymphocytes.

To test the functional consequences of the mutated α₃ domains, we used an in vitro system that is diagrammed in Fig. 5. The experiment measured the stimulation of a CD8αα-expressing T cell hybridoma, B3Z-CD8αα, when activated by its cognate OVA peptide, SIINFEKL (OVAp) bound to K^b. Hybridoma activation can be measured by an enzymatic assay, as the cells have been transfected with a LacZ reporter gene under the control of an IL-2 promoter element (22). As APCs, we used a H-2^b haplotype lymphoma cell line, RMA-S that is deficient for the TAP gene. Because of this deficiency, only a low level of MHC class I protein can be detected on the cell surface. By adding exogenous peptide, however, cell surface MHC class I-peptide complexes can be stabilized (24). In contrast to classical MHC class I Ag-presenting molecules, T18^d expression is TAP-independent (8). We previously showed that the IL-2 promoter activity and protein expression are increased when the B3Z-CD8αα hybridoma is activated by OVAp loaded on K^b APC transfectants that express T18^d (16).

We generated stable RMA-S transfectants expressing surface T18^d molecules bearing the same mutations as the recombinant proteins presented in Fig. 3. The level of surface expression of the mutated T18^d constructs in the RMA-S cell transfectants was approximately similar (Fig. 6,A). As expected, B3Z-CD8αα activated with OVAp loaded RMA-S/T18^d transfectants substantially increased the IL-2 promoter activity when compared with OVAp loaded untransfected (parental) RMA-S cells (Fig. 5). In accordance with the avidity measurements (Fig. 3 B), RMA-S-expressing T18^d with three K^b substitutions (G197D/D198K/T228M) displayed the most severe reduction in IL-2 promoter activity. Despite this reduction, the level of activation was still higher than the one observed with untransfected RMA-S (parental) cell line. In conclusion, the amino acid substitutions in T18^d that most affected the binding to CD8αα also affected the ability of the T18^d-CD8αα interaction to modulate immune function.

We also tested the mutated K^b molecules for a parallel gain of function. Three approaches were chosen. In the first approach, diagrammed in Fig. 7,A, we used the P815 mastocytoma cell line as APC. This cell line is of the d haplotype, and consequently is unable to present OVAp to the B3Z-CD8αα hybridoma. We generated P815 transfectants with wt K^b or the K^b triple mutant (K^b-D197G/K198D/M228T). Expression levels of the two transfectants were similar, as shown in Fig. 6,B. The capacity of the two P815 transfectants to activate B3Z-CD8αα in the presence of different concentrations of OVAp was evaluated. Untransfected control P815 cells were unable to activate B3Z-CD8αα (Fig. 7,A). wt K^b and K^b-D197G/K198D/M228T-transfected P815 cells were able to activate B3Z-CD8αα, although B3Z-CD8αα activation was stronger at every concentration of OVAp in the presence of the K^b-D197G/K198D/M228T-transfected P815 cells compared with the wt K^b transfectant (Fig. 7 A). This result is consistent with the notion that mutated K^b strongly interacts with CD8αα on the surface of the B3Z hybridoma, and that this interaction led to an increase in IL-2 promoter activity.

One caveat of this experimental setting is that the same transfected mutated K^b molecule, in addition to interacting with CD8αα, could simultaneously present OVAp to the B3Z-CD8αα TCR. Therefore, one cannot exclude the possibility that the simultaneous interaction of mutated K^b with the TCR and CD8αα could lead to a different outcome than for TL, which only interacts with CD8αα and not with conventional TCRs. To exclude this possibility, we used a second approach in which RMA-S cells were transfected with a chimeric construct encoding either T18^dα₁α₂/K^bα₃ or a T18^dα₁α₂/K^bα₃ mutant with substitutions in positions D197G, K198D, and M228T of the K^b α₃ domain (T18^dα₁α₂/K^bα₃-197/198/228). Expression levels of the transfectants are depicted in Fig. 6,C. The presence of the T18^dα₁α₂ domains in these two chimeric molecules renders them unable to present OVAp to B3Z-CD8αα cells. OVAp can still be presented to the B3Z-CD8αα hybridoma, however, through endogenously expressed K^b molecules produced by the RMA-S cells and stabilized on the cell surface by exogenous peptide. Therefore, as diagrammed in Fig. 7 B, in this experimental setting the TCR and CD8αα interactions are mediated by different MHC class I molecules. Endogenously expressed K^b will present OVAp to the TCR and the T18^dα₁α₂/K^bα₃-197/198/228 chimeric class I molecule should interact with CD8αα, whereas the T18^dα₁α₂/K^bα₃ chimera should not.

Analysis of B3Z-CD8αα activation showed that RMA-S cells expressing T18^dα₁α₂/K^bα₃-197/198/228 induced an activation level comparable to the one measured with RMA-S-expressing T18^d (Fig. 7 B). By contrast, RMA-S cells expressing the T18^dα₁α₂/K^bα₃ chimera induced a level of activation comparable to the untransfected RMA-S cells (parental). This result demonstrated that the T18^dα₁α₂/K^bα₃-197/198/228 chimeric molecule can recapitulate the ability of T18^d to deliver a signal through CD8αα, and moreover, it highlights the importance of the 197, 198, and 228 amino acids in the α₃ domain with regard to the high affinity interaction with CD8αα.

Because chimeric T18^d/K^b molecules were used as a T18^d surrogate molecule, we cannot completely rule out the possibility that amino acids in the T18^d α₁ and α₂ domains participate in the interaction with CD8αα, and that we are therefore measuring a cumulative effect of CD8αα interactions with the T18^dα₁α₂ domains in addition to the K^bα₃ domain mutated at positions 197, 198, and 228. To investigate this possibility, we used a third approach in which native K^b molecules bearing amino acid substitutions at positions 197/198/228 were not involved in Ag presentation. RMA-S cells were transfected with constructs encoding either wt K^b or the K^b-D197G-K198D-M228T α₃ domain triple mutant. A low level of K^b expression could be detected on the surface of transfected RMA-S cells, even in absence of exogenous peptide (Fig. 6 D). This probably reflects the overall increased level of K^b mRNA expression in the transfected RMA-S cells, compared with their endogenous level of K^b, perhaps combined with inefficient TAP-independent loading of peptides.

To measure the function of wt K^b vs the K^b-197/198/228 triple mutant in delivering a signal through CD8αα without involving the TCR, we used a CD8αα-expressing T hybridoma, 30NX/B10–1 (or 30NX), specific for the SVL9 peptide (SSVVGVWYL), derived from the H13 minor histocompatibility locus presented by D^b (25). Similar to B3Z, the 30NX hybridoma expresses the LacZ reporter construct. As schematically described in Fig. 7,C, the SVL9 peptide is presented to the 30NX hybridoma by endogenously expressed D^b molecules stabilized on the surface of the RMA-S cells by the added peptide, whereas the CD8αα interaction and signaling should be provided separately by the K^b-D197G-K198D-M228T mutant MHC class I molecule. wt K^b-transfected RMA-S cells were used as a control, and they did not deliver a high affinity signal through CD8αα. Activation of the 30NX hybridoma cells with RMA-S cells transfected with the K^b-197/198/228 triple mutant protein was increased compared with the control RMA-S cells transfected with wt K^b as measured by IL-2 promoter activity (Fig. 7 C).

The TL Ag is a mouse, MHC-encoded, nonclassical class I molecule with an unusual structure, expression pattern, and function. The function of this class I molecule is defined not by Ag presentation, but rather by its relatively high affinity binding to CD8αα compared with CD8αβ (16). The interaction of the TL Ag with CD8αα homodimers has an important influence on the behavior of different categories of T lymphocytes. Given the importance of this interaction for T cell biology, we set out to identify the molecular basis for the preferential binding of the TL Ag to CD8αα.

We demonstrated that the α₃ domain of the TL Ag is required for the relatively high avidity binding to CD8αα, and that amino acids 197 and 198 on the AB loop, and 228 on the CD loop, carry much of the specificity of TL Ag binding. There is less influence of the α₂ domain or other portions of the class I molecule. This is demonstrated most clearly by the ability of tetramers of the K^b molecule with T18^d substitutions at positions 197, 198, and 228 to bind to CD8αα-expressing cells with an avidity similar to the intact T18^d molecule itself. Tetramers of the HLA-A2/T18^d chimera bound CD8αα less avidly than T18^d, indicating that the α₁α₂ domains play a minor role.

The importance of a few amino acids in the α₃ domain for preferential CD8αα binding is consistent with several previous reports. Mutations in CD8 amino acids contacting the α₃ domain that greatly reduced binding of CD8 to classical class I molecules also greatly reduced CD8 binding to TL (26). Based on this, we concluded that CD8 likely contacted the α₃ domain of K^b and TL in a similar manner. This was confirmed by x-ray diffraction analysis of TL-CD8αα cocrystals, which demonstrated that the contacts of K^b for CD8αα are conserved in TL, and therefore that subtle changes in the TL α₃ domain might be responsible for the strengthened interactions of TL with CD8αα (9).

The existence of a human homologue for the TL Ag remains unresolved. Substantial numbers of human peripheral blood T cells express CD8αα homodimers in the absence of CD8β expression (27) or coexpressed CD4 and CD8αα (28), as do some human IEL. Moreover, biochemical evidence indicates that CD8αβ⁺ T cells also can coexpress CD8αα homodimers, especially when they are activated (Ref.29 , and H. Cheroutre and M. Kronenberg, unpublished data). Therefore, it is plausible that a human homologue of the TL Ag adapted for preferential CD8αα binding exists. One potential candidate is HLA-G, which has a restricted expression pattern and which binds to CD8 (30), but has unusual substitutions at positions 197 (tyrosine for histidine) and 228 (valine for threonine). Moreover, HLA-G has been reported to influence T cell function by binding to CD8 (31–37). Despite these similarities, HLA-G does not bind human CD8αα with an exceptionally high affinity (33). Another potential candidate for a functional human homologue of the TL Ag is the gp180 molecule, expressed on the intestinal epithelial cells, which interacts with CD8 when gp180 is associated with the nonclassical MHC class I molecule CD1^d (34). The CD1^d-gp180 complex activates the CD8-associated kinase p56^lck. Interestingly, interaction of the CD1^d-gp180 complex with CD8 is mediated by gp180 and not by CD1^d molecule (35). Therefore, gp180 could be a functional homologue of TL in humans, although it is not a structural homologue.

The set of nonclassical class I molecules have a strikingly diverse set of functions despite sharing a similar fold. In addition to recognition by TCRs and NK cell receptors, their functions include roles in iron transport, Ig transport, and Ig serum half life, and they may function as molecular chaperones (36). This functional plasticity encompasses the more membrane proximal α₃ domain, which has an Ig fold, as well as the α₁ and α₂ domains, which fold to form the Ag-binding groove. In this study, we demonstrate that small changes in the highly conserved α₃ domain can alter the selective binding for the different isoforms of CD8.

The authors have no financial conflict of interest.

We thank Stéphane Sidobre for help in isotherm binding data analysis, Peter Jensen for the bacterial T18^d construct, Carrie Micelli for providing the BI-CD8αα hybridoma, and Praveena Kasarabada for secretarial assistance.