T Cell Receptor Binding to a pMHCII Ligand Is Kinetically Distinct from and Independent of CD4.

Immune recognition of pMHCII ligands by a helper T lymphocyte involves its antigen-specific T cell receptor (TCR) and CD4 coreceptor. We have characterized the binding of both molecules to the same pMHCII. The D10 alphabeta TCR heterodimer binds to conalbumin/I-A(k) with virtually identical kinetics and affinity as the single chain ValphaVbeta domain module (scD10) (Kd = 6-8 microm). The CD4 ectodomain does not alter either interaction. Moreover, CD4 alone demonstrates weak pMHCII binding (Kd = 200 microm), with no discernable affinity for the alphabeta TCR heterodimer. Hence, rather than providing a major contribution to binding energy, the critical role for the coreceptor in antigen-specific activation likely results from transient inducible recruitment of the CD4 cytoplasmic tail-associated lck tyrosine kinase to the pMHCII-ligated TCR complex.

To date, biophysical parameters of CD4 interaction with pMHCII and TCR ectodomain components have not been meas-ured. Attempts to quantitate TCR-independent binding between CD4 and pMHCII in human and mouse systems using cell-based assays and soluble CD4 ectodomain constructs have suggested a weak interaction (K d Ͼ 100 M) (15,27,28). However, the precise affinity and kinetics of binding are unknown. Furthermore, whether CD4 ectodomain interaction with pMH-CII is modulated by the TCR or, alternatively, whether prior CD4-pMHCII interaction alters the MHC antigen-presenting platform, augmenting subsequent TCR affinity for its ligand is uncertain. In the present study, we have utilized highly purified recombinant TCR, pMHCII, and CD4 ectodomains to address these questions. The results offer insight into the fundamental nature of CD4-pMHCII interaction, further elucidate the role of CD4 in class II MHC-restricted TCR recognition, and indicate that if an interaction exists between CD4 and TCR extracellular segments it must involve components other than or in addition to the TCR ␣␤ heterodimer.

EXPERIMENTAL PROCEDURES
Construction and Expression of D10 TCR in Escherichia coli-Plasmids pEE14-D10␣B and pEE14-D10␤A were used as polymerase chain reaction templates to generate cDNAs encoding the extracellular domains of D10 ␣ and ␤ subunits. The 5Ј polymerase chain reaction primers contained a XbaI restriction site, a ribosome binding site, and a methionine initiation codon proximal to the first amino acid residue of each mature subunit. The 3Ј primers encoded a stop codon (TAA) and a BamHI restriction site. To avoid cysteine mispairing and incorrect disulfide bond formation, both ␣ and ␤ subunit sequences were terminated before their respective most-C-terminal cysteines utilized to form an interchain disulfide bond in the native protein. Moreover, the unpaired cysteine at position 109 in the ␣ chain was changed to serine (TGC to TCC) by overlapping polymerase chain reaction and restriction fragment replacement. The ␣ and ␤ subunit genes were inserted individually between XbaI and BamHI sites within the expression vector pET-11a (Novagen). After DNA sequences were confirmed, the expression plasmids were transformed into E. coli strain BL21 (DE3) separately.
Protein expression and inclusion body preparation of D10 ␣ and ␤ subunits were performed by the Cell Production and Recovery Facility, Rutgers University. Briefly, bacterial cells transformed with either pET-D10␣ or pET-D10␤ expression vectors were grown in a 50-liter bioreactor at 37°C in 4 ϫ YT medium (32 g/liter bactotryptone, 20 g/liter yeast extract, 5 g/liter NaCl) containing 50 g/ml ampicillin and 2% glycerol. 1 mM isopropyl-␤-D-thiogalactopyranoside was added to induce protein expression at A 600 ϭ 8 -12, and cells were harvested 4 h after the induction. Cells were resuspended in a buffer containing 50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride at 10 ml/g cell wet weight and lysed. Lysate was then incubated at room temperature for 1 h with 5 g/ml DNaseI and 4 mM MgCl 2 added. Inclusion bodies were spun down and washed twice with washing buffer (50 mM Tris-HCl, pH 8.0, 2 mM EDTA, 0.02% sodium azide) containing 0.5% Triton X-100 followed by two washes with washing buffer alone. The inclusion body pellets were frozen at Ϫ80°C and used for subsequent purification.
Refolding and Purification of D10 TCR ␣␤ Heterodimer and Single Chain D10 (scD10)-D10 TCR ␣␤ heterodimer was refolded as described previously (29) with some modifications. The ␣ and ␤ inclusion bodies were dissolved separately in 50 mM Tris-HCl, pH 6.8, 8 M urea, 10 mM EDTA and 0.1 mM dithiothreitol and centrifuged at high speed. The protein concentration of the supernatant was determined by Bio-Rad protein assay. The denatured ␣ and ␤ inclusion bodies (1 mol of each) were added together to 10 -12 ml of 6 M guanidine-HCl, 10 mM sodium acetate, and 10 mM EDTA, pH 4.2. The mixture was then diluted dropwise into 1 liter of cold refolding buffer (100 mM Tris-HCl, pH 8.0, 400 mM L-arginine-HCl, 2 mM EDTA, 5 mM reduced glutathione, 0.5 mM oxidized glutathione, and 0.5 mM phenylmethylsulfonyl fluoride) with vigorous stirring. After the refolding solution was incubated at 4°C with slow stirring for 6 -12 h, another ␣/␤ mixture was added. A third mixture was added 6 -12 h later, and a further 24-h incubation was performed.
The refolded material was filtered (Corning, 0.22 m) and immunoaffinity-purified using mAb 3D3 covalently coupled ␥ bind plus Sepharose beads (Amersham Pharmacia Biotech) at 5 mg of mAb/ml of beads. The correctly refolded TCR protein was eluted with low pH buffer (0.5 M acetic acid, 10% glycerol, pH 3.0) and adjusted to pH 5.0 -6.0 immediately using 1 M Tris-HCl, pH 9.5. The protein was then concentrated using a Centriprep-10 concentrator (Amicon) and sized on a Superdex 75 gel filtration column (Amersham Pharmacia Biotech) equilibrated with 20 mM sodium acetate/acetic acid, pH 5.0, and 100 mM NaCl. The purified protein was finally concentrated, and the buffer was changed to 20 mM sodium acetate/acetic acid, pH 5.0.
The scD10 TCR consists of 237 residues and is organized from N to C terminus as follows: V␤8.2 (residues 3-116)-linker (GSADDAKK-DAAKKDG)-V␣2 (residues 1-117), with a mutation at C235S. Bacterial expression and inclusion body purification were the same as that of the D10 TCR (described above). An efficient refolding was achieved by diluting rapidly into a refolding buffer (50 mM Tris-HCl, pH 8.0, 400 mM arginine, 2 M urea, 2 mM EDTA, 4 mM reduced glutathione, and 0.4 mM oxidized glutathione). The refolded material was then applied to a 3D3 affinity column followed by gel filtration on Superdex 75, and buffer was changed to 20 mM sodium acetate, pH 5.0, with 0.025% sodium azide.
Eukaryotic Expression in Lec3.2.8.1 Cells and Purification of hCD4 and CA/I-A k Proteins-The human CD4 ectodomain (residues 1-371) was expressed in mammalian Chinese hamster ovary-derived Lec3.2.8.1 cells (30) using the glutamine synthetase vector pEE14 (31). To this end, 20 g of linearized plasmid DNA pEE14-hCD4 was transfected into Lec3.2.8.1 cells by a calcium phosphate precipitation method using a transfection MBS kit (Stratagene) as described (32). 48 h later, the transfected cells were harvested and cultured on 96-well plates in Glasgow minimal essential medium supplemented with 25 M methionine sulfoximine. After feeding the cells for 4 weeks, the growing clones were selected and assayed by enzyme-linked immunosorbent assay. 4 g/ml anti-D1 domain of hCD4 mAb, 19Thy5D7, was coated onto Immulon plates (Dynatech) overnight at 4°C and blocked by 1% bovine serum albumin at room temperature for 1 h. Then 50 l of supernatant from individual clones was added to each well and incubated overnight at 4°C. The reaction was followed by adding 0.1 mg/ml biotinylated mAb OKT4 (anti-D3 domain of hCD4) and developed by horseradish peroxidase-conjugated streptavidin. Once identified, positive clones were transferred to 24-well plates and then to 6-well plates and rescreened. Subsequently, the highest-expressing clone was picked and amplified for large scale production as described previously (31). Yields averaged 40 mg/liter culture supernatant. After each expression, the supernatant was filtered (Corning, 0.22 m) and immunoaffinity-purified by mAb 19Thy5D7-coupled Affi-Gel 10 beads (Bio-Rad). The hCD4 protein was eluted by 0.5 M acetic acid, pH 3.5, and neutralized to pH 7.0 immediately by adding 1 M Tris-HCl, pH 9.5. For BIAcore assay, the protein was further purified by gel filtration on a Superdex 75 column and concentrated to 20 -50 mg/ml.
The mature I-A k ␤ chain was fused at its N terminus via a flexible linker with a 13-residue hen egg conalbumin (CA) peptide (residues 153-165) that is recognized by D10 TCR. The 37-residue leucine zipper (LZ) sequences were appended to both C termini of the ␣ (residues -3-192) and ␤ (residues 3-198) chains by flexible thrombin-cleavable spacers. The cDNA constructions were subcloned into the pEE14 vector and expressed in Lec3.2.8.1 cells as described above. The screening of secreted recombinant protein in the culture supernatant was performed by both Sandwich enzyme-linked immunosorbent assay and BIAcore using antibodies specific for either the CA/I-A k (10.2.16) or the LZ epitope (2H11 or 13A12). The yield was ϳ0.7 mg of CA/I-A k /liter of supernatant. Protein in production supernatant was purified by 2H11 affinity chromatography followed by Superdex 75 gel filtration.
Determination of Protein Concentration and N-terminal Sequence Analysis-Protein concentrations were determined by spectrophotometry at 280 nm using factors of 1.2 (scD10), 1.4 (D10), 1.5 (CA/I-A k ), and 1.4 ml/mg/cm (hCD4), respectively. The factors were acquired from extinction coefficients that were calculated based on the tryptophan, tyrosine, and cysteine contents of each protein (33). N-terminal sequence was performed on each protein after blotting to a polyvinylidene difluoride membrane using a 120A sequencer (Applied Biosystems).
Surface Plasmon Resonance Studies-All binding studies were performed on a BIAcore1000 surface plasmon resonance biosensor (BIAcore) in HBS buffer (10 mM HEPES, 150 mM NaCl, 3 mM EDTA, and 0.005% surfactant p-20, pH 7.4) at 25°C. For indirect binding assay, an anti-LZ mAb, 13A12 (34), was coupled to the CM5 sensorchip using a standard amine coupling kit (BIAcore), resulting in ϳ8,000 RU coupled. 30 l of 1 M CA/I-A k -LZ was injected over the surface at a flow rate of 10 l/min and followed by a dissociation period of 3 min. Different analytes (D10 or scD10 TCRs or hCD4) were then passed through the13A12-CA/I-A k -LZ surface either at a high flow rate (50 l/min for kinetic studies) or at a low flow rate (5 l/min for equilibrium studies). The specific binding was taken after subtracting the response on a control surface (the same 13A12 immobilized surface but without any CA/I-A k -LZ captured). CD4 protein was coupled on the CM5 chip directly using an amine coupling kit, ranging from 2,000 to 3,000 RU, and analytes were then injected over the flow cell at a flow rate of 5 l/min for affinity studies. An irrelevant anti-gp140 mAb 116 Fab fragment was immobilized at the same level as a control surface. BIAevaluation 3.0 software (BIAcore) was used for data analysis. Kinetic data fitting was performed using a Langmuir 1:1 binding model (BIAcore), and equilibrium data were analyzed by nonlinear curve fitting or Scatchard plotting RU/concentration versus RU followed by linear regression (Cricket-Graph software).

COS-7 Cell Transfection and Cell Binding Assay-
The COS-7 cell transfection of human CD4 and the inhibition effect of CD4 on human MHC class II B-lymphoblastoid Raji cells binding were performed as described previously (15). Briefly, 5 ϫ 10 4 COS-7 cells were plated into each well of Falcon 6-well plates and transfected by a calcium phosphate/chloroquine method (35) with 5 g of either full-length hCD4 DNA (residues 1-435 cloned in the CDM8 vector) or CDM8 vector DNA alone as a control. Two days later, the cell binding assay was performed. 3 ϫ 10 6 Raji cells were pre-incubated in 1 ml of medium containing various concentrations of affinity-purified hCD4 protein or bovine serum albumin (as negative control) at 37°C for 30 min and then added to transfected COS-7 cells. The 6-well plates were incubated at 37°C for 1 h, and the rosettes formed by B Raji cell-COS-7 cell binding were counted under microscopic magnification. As a positive control for inhibition, 15 g of 19Thy5D7 was added to transfected COS-7 cells and incubated at 37°C for 30 min before adding the B cells. The inhibition percentage was calculated as 100 ϫ [(R c Ϫ R I )/ R c ], where R c and R I represent the rosette numbers in the absence or presence of inhibitor, respectively.

RESULTS
Expression and Purification of Recombinant D10 TCR, pMHCII, and CD4 Ectodomain Proteins-To study the interaction between a MHC class II-restricted TCR, its ligands, and CD4 coreceptor, recombinant ectodomains corresponding to each were expressed and purified. An ␣␤ TCR heterodimer derived from the mouse T cell clone D10 (D10 TCR), specific for the hen egg CA foreign peptide (residues 153-165) associated with the self-MHC class II I-A k molecule (pMHCII), was constructed (Fig. 1a). The ␣ and ␤ subunits of the D10 TCR were truncated at residues 200 and 238 within their respective extracellular segments proximal to the cysteines involved in the interchain disulfide bond formation. Furthermore, to avoid cysteine mispairing, ␣ chain cysteine 109 was replaced by a serine. The ␣ and ␤ subunits were expressed separately as inclusion bodies in E. coli and refolded together at a 1:1 molar ratio as described under "Experimental Procedures." The refolded ␣␤ TCR was purified by immunoaffinity chromatography using the 3D3 anti-D10 TCR clonotypic mAb followed by gel filtration on a Superdex 75 column. A yield of 6 -8 mg of purified protein/ liter of starting material (ϳ150 mg of inclusion bodies) was obtained. The purified protein can be stored at 4°C at high concentration (up to 1 mM) with little tendency to form aggregates.
SDS-PAGE analysis of the purified protein shows that the ␣ and ␤ chains are present in equal molar amounts (Fig. 1b, lane  1 and 2). Since the interchain disulfide bond was mutated, no covalently linked heterodimer band is seen under nonreducing conditions (Fig. 1b, lane 2). The faster mobility of the ␣ and ␤ bands on SDS-PAGE under nonreducing relative to reducing conditions (Fig. 1b, lane 2 versus 1; 23 versus 26 kDa for ␣ and 27 versus 30 kDa for ␤) is a consequence of the two intra-chain disulfide bonds in each subunit leading to a more compact structure. Expected N-terminal sequences of both ␣ and ␤ chains were confirmed by Edman degradation after transfer to a polyvinylidene difluoride membrane. To assess the integrity of the ␣ and ␤ constant domains, the TCR protein was immunoprecipitated with TCR C␣-specific mAb H28 and C␤-specific mAb H57. SDS-PAGE showed that the TCR was precipitated by both mAbs. The TCR-mAb complexes were readily observed by native gel and equivalent to D10 expressed in the mammalian Chinese hamster ovary derivative Lec 3.2.8.1 cells as a glycosylated protein with the intact interchain disulfide bonds (data not shown). Given that the 3D3 mAb used for purification recognizes a conformational epitope dependent on the proper juxtaposition of V␣ and V␤ domains, the above results indicate that the purified D10 TCR assumes a correctly folded conformation in both V-and C-region modules.
To investigate any possible function of the C module in antigen recognition or CD4 interaction, we compared the above D10 ␣␤ TCR heterodimer with a scD10 TCR V␣V␤ module. As shown in Fig. 1a, the latter contains the D10 TCR V␤ domain connected from its C terminus by a flexible linker, a 15-amino acid residue peptide (GSADDAKKDAAKKDG), to the N termi-nus of the D10 TCR V␣ domain. Like the D10 TCR, the scD10 protein was expressed in E. coli, refolded in vitro, and purified by 3D3 affinity column followed by gel filtration on Superdex 75. The scD10 protein appears as a single band migrating at 31 kDa under reducing conditions (Fig. 1b, lane 3).
A cDNA encoding the four extracellular domains of human CD4 (Fig. 1a) in the pEE14 plasmid was transfected and expressed in the glycosylation mutant Lec3.2.8.1 derivative of Chinese hamster ovary cells, which produce glycoproteins with homogenous sugar adducts. In these cells, N-linked carbohydrates are of the Man 5 form, and O-linked carbohydrates are truncated to a single GalNac (30). The yield of hCD4 was more than 40 mg/liter culture supernatant, and the protein was readily purified by affinity chromatography on an anti-CD4 D1 mAb 19Thy5D7-coupled Affi-Gel 10 column followed by gel filtration on a Superdex 75 column. The protein shows high purity as well as size and charge homogeneity on SDS-PAGE (Fig. 1b, lane 5 and 6) and native gel (data not shown), respectively. As with the TCR constructs, the faster mobility of the protein under nonreducing conditions (45 kDa versus 47 kDa) indicates the formation of the expected intrachain disulfide bonds. The ability of other mAbs such as OKT4 (a CD4 D3specific mAb) to react with the protein was verified by enzymelinked immunosorbent assay (data not shown).
The peptide-bound murine MHC class II molecule CA/I-A k , the D10 ligand, was produced in recombinant form from Lec3.2.8.1 cells as well. This pMHCII protein consists of the extracellular domains of the noncovalently associated ␣ and ␤ subunits, with a 13-residue peptide corresponding to residues 153-165 of hen CA attached to the N terminus of the ␤ chain and the leucine zipper sequences appended to the C termini of both chains (Fig. 1a). Although the linkage of the peptide to the class II MHC is distinct from that found under natural circumstances, crystallographic analysis of the pMHCII protein demonstrates a native structure (36). The yield of the CA/I-A k is ϳ0.7 mg/liter culture supernatant after affinity purification. On SDS-PAGE (Fig. 1b, lane 4), the purified CA/I-A k migrates as two closely spaced bands, with apparent molecular masses of ϳ35 kDa. N-terminal amino acid sequence determination verified that the upper band is the ␣ chain and the lower one is the CA peptide fused ␤ chain. Unlike with the other recombinant proteins, the apparent molecular mass mobility on SDS-PAGE of the I-A k subunits is inconsistent with those predicted from amino acid sequences (␣ ϭ 27,295 daltons and ␤ ϭ 30,908 daltons). These differences are due to post-translational modification (there are two N-linked glycosylation sites in the ␣ chain and one in the ␤ chain) as well as the highly charged leucine zipper sequences, resulting in anomalous migration on SDS-PAGE (32,34). All four proteins can be concentrated to 1-2 mM with no aggregates observed by either light scattering or gel filtration (data not shown).
Affinity Studies of D10 and scD10 TCRs Binding to CA/I-A k Using an Orientational Capture BIAcore Method-Surface plasmon resonance was used to examine the binding affinity between the D10 TCR ␣␤ heterodimer and CA/I-A k and, as a comparison, between the scD10 TCR and CA/I-A k . To this end, the anti-LZ mAb 13A12 was immobilized on a CM5 sensorchip to a level of ϳ8000 RU. The CA/I-A k protein was captured (ϳ800 RU) through binding of the C-terminal LZ appended to the pMHCII, facilitating an orientation that exposes the pMHCII antigen-presenting platform to the TCR. pMHCII accessibility to immune recognition was first verified by injecting 0.5 M anti-I-A k mAb 10.2.16 (data not shown). Subsequently, the D10 ␣␤ TCR and scD10 were individually injected as the analyte over the pMHC surface (Fig. 2, a and b, insets). Sensorgrams of the indicated concentrations of D10 and scD10 TCR binding to captured CA/I-A k are shown in Fig. 2, a and b (main plots), respectively. For kinetic studies, data were fitted using a Langmuir 1:1 binding model, and similar dissociation constants (K d ) of the two D10 TCRs binding to CA/I-A k were acquired (see below). Moreover, consistent results were obtained independently by equilibrium studies (Fig. 2, c and d). Both direct nonlinear curve fitting and Scatchard plot analysis gave similar K d values. Fig. 2e summarizes the fitted kinetic and equilibrium data. Both D10 and scD10 TCRs show relatively fast association and dissociation rates when binding to CA/I-A k . D10 binds to CA/I-A k with a K d of 6.7 M and a half-life of about 17.5 s. scD10 TCR shows a similar binding affinity to CA/I-A k with a K d of 7.8 M and a half-life of 12.5 s. The similarities in these binding parameters imply that the V domain module of the TCR alone is sufficient for pMHCII recognition. These values are within the range of those reported for other TCR-pMHCI and pMHCII interactions at 25°C (37) as well as different D10 TCR-related constructs (38,39).

The Binding of the D10 ␣␤ Heterodimer to CA/I-A k Is Not
Influenced by the CD4 Ectodomain-To determine whether CD4 could influence the binding between the D10 TCR and its pMHCII ligand, the D10 TCR ␣␤ heterodimer protein was mixed with different concentrations of soluble hCD4 and incubated overnight at 4°C. In these experiments, the concentration of D10 TCR ␣␤ protein was constant (20 M), whereas the hCD4 concentration was varied from 0 to 100 M, generating a series of "D10/CD4" molar ratios. The individual D10 or hCD4 components or alternatively, a mixture of the two at a given molar ratio was injected separately over the same 13A12 mAbcaptured CA/I-A k surface as described above. As shown in Fig.  3a, the response of the 20 M D10 plus 50 M hCD4 mixture (D10 ϩ CD4) binding to CA/I-A k simply appears to be the addition of the individual D10 and hCD4 binding responses. Furthermore, the dissociation phases of the D10 ϩ CD4 mixture and the D10-alone sample overlay quite well, indicating that the addition of CD4 does not affect the kinetics of D10 TCR binding to CA/I-A k . Fig. 3b shows that the specific D10 binding

CD4-MHC Class II Affinity
responses in the presence of a range of hCD4 concentrations (computed by subtracting the hCD4 response from the response of the corresponding D10 ϩ CD4 mixture) were virtually identical to that in the absence of hCD4. These data suggest that the affinity of the D10 TCR ␣␤ clonotype for its ligand CA/I-A k is not altered by CD4. Such identity would not be observed if interaction between CD4 and pMHCII conformationally affected the antigen-presenting platform, thereby modulating TCR binding. Similar results were observed at 37°C (data not shown).
CD4 Binds to pMHCII with Low Affinity-The binding of CA/I-A k , D10 TCR ␣␤ heterodimer, or scD10 TCR to CD4 was studied by passage of these individual protein analytes over a CM5 sensor chip onto which the hCD4 coreceptor ectodomain was directly immobilized. Successive injections of mAb 19Thy5D7 (anti-D1 domain) and OKT4 (anti-D3 domain) verified the orientation and immunoreactivity of the coupled hCD4. Both mAbs showed identical binding responses, indicating that the overall exposure of the various segments of the CD4 molecules is equivalent (data not shown). The insets of Fig. 4, a, c, and d, show the schematic interaction of CA/I-A k (Fig. 4a), D10 (Fig. 4c), or scD10 (Fig. 4d) to the immobilized hCD4 (only one hCD4 orientation is shown here). The sensorgrams of CA/I-A k binding to CD4 and Fab116 (an anti-SIV gp140 mAb fragment used as the control surface) at indicated concentrations are shown in Fig. 4a, main plot (solid and  dashed lines, respectively). The binding of CA/I-A k to CD4 shows weak affinity with on-and off-rates too rapid to measure. The sustainable responses on the control surface are mainly due to the bulk effect from the high CA/I-A k protein concentrations (up to 250 M). The specific CA/I-A k binding to CD4 is taken as the difference between responses in RU on the CD4 and Fab116 surfaces. These data are plotted using Scatchard analysis as shown in Fig. 4b, giving an affinity of ϳ200 M (K d ) for the hCD4-CA/I-A k interaction. The affinity between hCD4 and the D10 or scD10 TCR is much lower than that between hCD4 and CA/I-A k (Fig. 4, c and d). In fact, no specific binding could be detected between CD4 and the D10 TCR ␣␤ heterodimer or between CD4 and the scD10. The responses on the Fab control surface (Fig. 4, c and d) and the intact, unrelated 13A12 IgG control surface (data not shown) were slightly higher than that on the CD4 surface. Comparison of the sensorgrams (Fig. 4, c and d) shows very similar results, except that the bulk effect differs, resulting from the distinct molecular masses of D10 and scD10 TCRs. Hence, we detect no binding between the TCR and CD4 ectodomains, implying that either the interaction, if it exists, is too low to measure or involves other TCR components in addition to or distinct from the ␣␤ heterodimer.
In the above BIAcore experiment, the interaction between mouse pMHCII and hCD4 was examined. Given that hCD4 is known to interact with mouse pMHCII in a fashion comparable with mCD4 in vitro and in vivo (6, 40 -43), it was unlikely that the low affinity was due to species differences. Nonetheless, to confirm the CD4-pMHCII binding affinity that we obtained from BIAcore assay and exclude any effects arising from the proteins derived from heterologous species, we examined the inhibitory effect of soluble hCD4 on the binding between class II MHC-expressing human Raji B cells and hCD4-transfected COS-7 cells. Immunoaffinity-purified hCD4 was added into this hCD4-MHCII-dependent cell-cell binding system at concentrations ranging from 0 to 875 M. Cell-cell adhesion was then quantitated by enumerating rosettes formed between hCD4-transfected COS cells and Raji B cells. As shown in Fig.  5, the number of rosettes was minimally decreased when CD4 protein concentration reached 100 M but was abolished at 400 M hCD4. The data are plotted using percentage inhibition versus inhibitor concentration. The concentration of CD4 inhibiting 50% of rosette formation (157 M) was taken as the dissociation constant (K d ). This value is consistent with the K d calculated from BIAcore data (ϳ200 M). Note that the positive inhibition control, anti-CD4 mAb 19Thy5D7, completely inhibits the B cell binding at a concentration Ͻ0.1 M, whereas the negative control, bovine serum albumin, does not affect the binding even at a 1 mM concentration.
TCR ␣␤ Heterodimer Interaction with pMHCII Does Not Augment Subsequent CD4 Binding-The presence of CD4 does not influence the interaction between D10 ␣␤ and CA/I-A k ectodomains, indicating independent binding of TCR and CD4 to CA/I-A k . We next examined whether the D10 ␣␤ heterodimer ligation to CA/I-A k could affect binding of each component to the immobilized CD4. Since the D10 TCR binds to CA/I-A k at a 1:1 molar ratio (36), the two ectodomain proteins were mixed in equivalent amounts and injected over the CD4 surface. As shown in Fig. 6, a-c,

CD4-MHC Class II Affinity
D10-CA/I-A k pre-mixture to CD4 are clearly no greater than the sum of the individual components in each pre-mixture. Furthermore, no change is observed in either association or dissociation phases. Collectively, these data show that the D10 TCR and CA/I-A k do not affect each other's binding to CD4. DISCUSSION The present study provides the first direct quantitation by BIAcore equilibrium analysis of the monomeric affinity of CD4 for pMHCII (200 M K d ). For these measurements, we utilized highly purified, aggregate-free human CD4 and mouse I-A k ectodomains. We also observed a ϳ160 M IC 50 for human CD4 in blocking the CD4-MHCII-based adhesion between COS cells transfected with human transmembrane CD4 and human Raji B cells expressing class II MHC. Since the Raji cells coexpress endogenous HLA-DR, -DP and -DQ molecules, it is unlikely that greater binding affinities exist for any of the polymorphic alleles present on the surface of Raji cells. These results are also consistent with earlier measurements by cell-based assays reported for hCD4-hpMHCII and mCD4-mpMHCII suggesting a K d Ͼ100 M (15,28).
In several regards, the interaction between CD4 and MHCII needs to be compared and contrasted with the interaction between CD8 and MHCI. CD8 and CD4 are both coreceptors for MHC molecules that bind to the membrane proximal exposed loop on the ␣3 and ␤2M domains of class I MHC and homologous ␤2 and ␣2 domains of class II MHC molecules, respectively (11,12,14,44). Furthermore, to mediate physiologic interactions, each coreceptor binds to the same MHC as the TCR- ligated MHC molecules. In the case of murine CD8␣␣ or CD8␣␤ interaction with H-2K b molecules, the affinity is ϳ30 -75 M (32,45). A severalfold lower affinity has been reported for the interaction between human CD8␣␣ and HLA-A2 (46).
Despite the similarities of CD4 and CD8 in terms of their target ligands and TCR coreceptor function, substantial differences exist. For example, the CD8 coreceptor is an Ig domain dimer that projects from the T cell surface on a long, heavily O-glycosylated stalk (reviewed in Ref. 47 and references therein). Interaction with pMHCI is via the six antibody-like CDR loops, three from each subunit Ig domain (47,48). In contrast, CD4 interaction with pMHCII involves the D1-D2 module including residues outside of the CDR regions (13,49,50). Perhaps more importantly, the membrane proximal CD4 segment is not a flexible stalk but rather is composed of two Ig-like domains. Given the structured nature of the CD4 D3-D4 module, it has been suggested that the membrane proximal domain region may be involved in protein-protein interaction. In this regard, prior chimeric studies have shown that although D3-D4 is not involved in pMHCII binding, the presence of this module is necessary for mediating proper pMHCII ligation via D1-D2 (15,51). Thus, its replacement with either the related CD4 D1-D2 module or alternatively, CD2 D1-D2, fails to result in a CD4 chimera with class II MHC binding activity. The potential for CD4 self-oligomerization via D3-D4 might explain this result, accounting for the ability of the non-MHC class II binding CD4 variant F43I to function as a dominant negative mutant when cotransfected with wild-type CD4. Oligomerization or other mechanisms of CD4 clustering offers a means to augment avidity, thereby offsetting the weak monomeric CD4-pMHCII affinity.
Additional studies raise the possibility that the CD4 D3-D4 module can interact with the T cell receptor (16,17). In particular, CD4 mutants that fail to interact with pMHCII (D2 mutations) or p56 lck (cytoplasmic tail mutants) surprisingly restore antigen responsiveness in the 3A9 T cell hybridoma system. Such restoration is dependent on the D3-D4 module, as shown by additional mutagenesis studies (17). Nevertheless, BIAcore analysis failed to reveal any measurable affinity between the CD4 ectodomain and the TCR ␣␤ heterodimer. In fact, sensorgrams investigating the CD4 interaction with the D10 TCR ␣␤ heterodimer were equivalent to those of CD4 and the scD10 module. Retention of antibody epitopes on key TCR, pMHCII, and CD4 domains suggests that the recombinant proteins utilized herein are native. Although mAbs to all protein domains were not available for testing, the fact that each protein has been crystallized and diffracts to near-atomic res-olution argues for the native structures of the components analyzed above (36). Hence, we conclude that there is no TCR ␣␤-CD4 interaction. These findings suggest that either such a CD4-TCR interaction doesn't exist or, if it does, that the CD4 membrane proximal region must contact other TCR components in addition to or distinct from the ␣␤ heterodimer. Recent structural analysis of scD10 in complex with CA/I-A k indicates that the ␣-helix of the I-A k ␤-chain is contacted by the TCR V␣ domain (36). Furthermore, mAb mapping studies suggest that the CD3⑀␦ heterodimer is adjacent to the TCR ␣ subunit (52). Other potential contacts may involve CD3⑀␦ as shown in Fig. 7. CD3⑀␦ association with TCR ␣␤ may alter the structure of the outer face of the C␣ domain as well (53,54).
Aside from interactions between their respective ectodomains, additional mechanisms exist to juxtapose the TCR and CD4 molecules. Upon cross-linking, the TCR redistributes into lipid rafts where CD4 and p56 lck are resident (55,56). TCR ligation by pMHC results in phosphorylation of ITAMs in the cytoplasmic tail of the CD3␥, ␦, ⑀, and components (57)(58)(59)(60)(61). These segments serve as substrates for p56 lck -mediated tyrosine phosphorylation as well as p56 lck SH2 interaction, both of which contribute to TCR-based signaling (26,62). Coassociation of a ligated TCR and CD4 could result from an "adaptor" role for p56 lck given its noncovalent linkage to CD4, as demonstrated from the studies of Xu and Littman (26). Consistent with this suggestion, only partial CD3 phosphorylation is induced by antagonist peptide ligands versus more complete phosphorylation by agonist peptide ligands (63)(64)(65). CD4 selectively enhances T cell recognition of the agonists but not the antagonists (66).
In these current studies, we show that the binding of the D10 ␣␤ TCR heterodimer to its pMHCII ligand is not affected by prior contact with the CD4 ectodomain. Thus, no conformational change and/or alteration in the peptide antigen-presenting platform or pMHC interdomain disposition affecting TCR binding is induced. This result is analogous to earlier findings involving mouse as well as human CD8␣␣-pMHCI complexes (47,48). Coreceptor function per se does not modify monomeric TCR affinity for its pMHC ligand (32,46). Consistent with these functional results, crystallographic studies of unligated human and mouse MHC class I molecules show no detectable differences in the MHC antigen-resenting platform in the presence or absence of the CD8␣␣ interaction. The common theme for both CD4 and CD8 coreceptor is the bidentate interaction involving separate sites for TCR and coreceptor on the pMHC and the attendant recruitment of p56 lck .
Collectively, our results suggest that on helper T cells, a given TCR first interacts with the relevant pMHC class II ligand. The 30-fold higher affinity of the D10 TCR ␣␤ heterodimer relative to CD4 for the same pMHCII ligand argues that the TCR is the major contributor of the binding energy. TCR ligation then initiates a signaling cascade that includes and is critically dependent upon p56 lck in association with CD4. CD4 is directed toward the TCR-ligated pMHCII molecule, augmenting binding and perhaps facilitating TCR cross-linking via CD4 and/or TCR clustering. Consistent with this notion, anti-CD4 mAbs directed against a CD4 ectodomain fail to block pMHCII tetramer binding to antigen-specific T cells yet inhibit activation of those cells (67,68). Videomicroscopy results also support the notion of a primary binding role for the TCR rather than CD4 in T cell recognition of pMHCII on antigen-presenting cells (69). This bidentate interaction will increase contact of the T cell with pMHCII, permitting a sufficient dwell time to activate gene programs required for cytokine production and to mediate helper T cell activity.