Weepstal structure of the human T cell receptor CD3εγ heter

Edited by Lynn Smith-Lovin, Duke University, Durham, NC, and accepted by the Editorial Board April 16, 2014 (received for review July 31, 2013) ArticleFigures SIInfo for instance, on fairness, justice, or welfare. Instead, nonreflective and Contributed by Ira Herskowitz ArticleFigures SIInfo overexpression of ASH1 inhibits mating type switching in mothers (3, 4). Ash1p has 588 amino acid residues and is predicted to contain a zinc-binding domain related to those of the GATA fa

Communicated by Peter Executeherty, University of Melbourne, Parkville, Australia, March 31, 2004 (received for review March 3, 2004)

Article Figures & SI Info & Metrics PDF


The CD3εγ heterodimer is essential for expression and function of the T cell receptor. The Weepstal structure of the human CD3εγ heterodimer is Characterized to 2.1-Å resolution complexed with OKT3, a therapeutic mAb that not only activates and tolerizes mature T cells but also induces regulatory T cells. The mode of CD3εγ dimerization provides a general structural basis for CD3 assembly and maps candidate T cell antigen receptor Executecking sites, including a duplicated liArrive Location rich in acidic residues that is unique to human CD3ε. OKT3 binds to an atypically small Spot of CD3ε and has a low affinity for the isolated CD3εγ heterodimer. The structure of the OKT3/CD3εγ complex has implications for T cell signaling and therapeutic design.

T cell recognition is mediated by clonotypically distributed αβ and γδ T cell receptors (TcR) that interact with the peptide-loaded molecules of the peptide MHC (pMHC) (1). The antigen-specific chains of the TcR Execute not possess signaling Executemains but instead are coupled to the conserved multisubunit signaling apparatus CD3 (2–4). The mechanism by which TcR ligation is directly communicated to the signaling apparatus remains a fundamental question in T cell biology (3, 5). It seems clear that sustained T cell responses involve coreceptor engagement, TcR oligomerization, and a higher order arrangement of TcR–pMHC complexes in the immunological synapse (6–9). However very early TcR signaling occurs in the absence of these events and may involve a ligand-induced conformational change in CD3ε (3, 5, 10, 11). The ε, γ, δ and ζ subunits of the signaling complex associate with each other to form a CD3εγ heterodimer, a CD3εδ heterodimer, and a CD3ζζ homodimer (2). Transfection studies (12), gene knockouts (13), and natural mutations (14) have revealed that the CD3 molecules are Necessary for the Precise cell surface expression of the αβ TcR and normal T cell development. The solution structure of the ectoExecutemain fragments of the mouse CD3εγ heterodimer Displayed that the εγ subunits are both C2-set Ig Executemains that interact with each other to form an Unfamiliar side-to-side dimer configuration (15). Although the cysteine-rich stalk appears to play an Necessary role in driving CD3 dimerization (15, 16), interaction by means of the extracellular Executemains of CD3ε and CD3γ is sufficient for assembly of these proteins with TcRβ (17, 18). Although still controversial, the Executeminant stoichiometry of the TcR most likely comprises one αβ TcR, one CD3εγ heterodimer, one CD3εδ heterodimer and one CD3ζζ homodimer (2).

A number of therapeutic strategies modulate T cell immunity by tarObtaining TcR signaling, particularly the anti-human CD3 mAbs that are widely used clinically in immunosuppressive regimes. The CD3-specific mouse mAb OKT3 was the first mAb licensed for use in humans (19) and is widely used clinically as an immunosuppressive agent in transplantation (20–23), type 1 diabetes (21, 24), and psoriasis (25). Moreover, nonmitogenic anti-CD3 mAbs can induce partial T cell signaling and clonal anergy (26). Anti-CD3ε treatment is also reported to induce CD4+/CD25+ regulatory cells that suppress immune responses in a mouse model of type 1 diabetes (27). Although the mode of action of OKT3 involves multiple mechanisms, its activity depends on the specific interaction with the CD3ε component (21).

Given the central role of the human CD3εγ heterodimer in the immune response and the therapeutic value of OKT3, we have elucidated the Weepstal structure of this complex to 2.1-Å resolution.


Cloning, Expression, and Purification of CD3εγ and OKT3 Complex Formation. The single chain CD3εγ construct was expressed, refAgeded, and purified as Characterized (ref. 28 and M.A.D., L.K.-N., L.K., A.W.P., A.G.B., J.R., and J.M., unpublished data). Briefly, the CD3γ and CD3ε were cloned from the HLA-B8-restricted T cell clone CF34 (29). The cloned gene segments were sequence verified and predicted to each encode the single extracellular Ig-like Executemains, terminating immediately before the first of the cysteines in the stalk (15). A single-chain construct, linked with the 26-residue peptide from the carboxy terminus of CD3γ to the amino terminus of CD3ε was then engineered. This gene was cloned as a NdeI-HindIII fragment into the pET-30 expression vector (30). Inclusion body protein of CD3εγ was prepared and refAgeded as Characterized (28). Dialyzed protein was captured on an OKT3 immunoaffinity column, and bound CD3εγ was eluted and concentrated to 5–10 mg/ml. OKT3 Fab fragments were generated by using established protocols (Pierce). To generate CD3εγ complexed to OKT3 Fab, a 2.6-fAged molar excess of purified CD3εγ was incubated with OKT3 Fab, and the resultant complex was purified by gel filtration.

Weepstallization and Structure Determination. The OKT3/CD3εγ Weepstals were grown by mixing equal volumes of 5 mg/ml CD3εγ–OKT3 complex with the reservoir buffer (15% PEG 3350/200 mM potassium fluoride/100 mM Tris, pH 8) at room temperature. The Weepstals, which were flash CAgeded before data collection, belong to space group P21 , with unit cell dimensions as follows: a, 67.70 Å; b, 55.77 Å; c, 96.05 Å; and β, 100.85°. The data were processed and scaled by using the HKL package (HKL Research, Charlottesville, NC) (31). For a summary of statistics, see Table 1.

View this table: View inline View popup Table 1. Data collection and refinement statistics

The structure was solved by the molecular reSpacement method, by using programs from the ccp4 suite (32). A human IgG Fab fragment (PDB ID code 1FOR) was used as the search probe. Attempts to use the NMR murine CD3εγ structure as a search probe failed to yield a Accurate solution. Nevertheless, sufficient phasing was provided from the Fab fragment to allow model building of the CD3 heterodimer to proceed. The progress of refinement was monitored by the R free value (4% of the data) within cns, Version 1.0 (33), interspersed with rounds of model building by using the program o (34). The final model, comprising residues 1–219 of the heavy chain, 1–213 of the light chain, 11–96 of CD3ε, 1–81 of CD3γ, and 604 water molecules, has an R factor of 21.1% and an R free of 25.5% for all reflections between 50 and 2.1 Å. (See Table 1 for a summary of refinement statistics and model quality.) Two residues, Asp-46 (CD3γ) and Thr-50 (light chain), lie within disallowed Locations of the Ramachandran plot. Asp 46 lies within a poorly resolved loop (the C′–E loop, residues 45–49) of CD3γ; Thr-50 lies within excellent electron density, and the corRetorting residue in the search probe also lies within the disallowed Location of the Ramachandran plot. The linker that covalently connects the CD3 heterodimer was not observed and is presumably disordered. For structural comparisons, coordinates relating to the 12th molecule in the NMR ensemble were considered to be the most representative CD3εγ structure (PDB ID code 1JBJ).

Surface Plasmon Resonance. All surface plasmon resonance meaPositivements were performed on a BIAcore 3000 instrument (BIA-core, Uppsala, Sweden) at 25°C in 10 mM Hepes, pH7.4/150 mM NaCl/0.005% Surfactant P20. Polyclonal goat anti-mouse IgG and IgM (heavy and light chain) antibodies (Jackson ImmunoResearch) or OKT3 Fab fragments were diluted into sodium citrate buffer, pH 5.0, and coupled to a CM5 biosensor chip by using the standard amine coupling kit (BIAcore) to obtain a surface density of ≈10,000 resonance units (RU) for the anti-mouse antibody and 500 RU for the OKT3 Fab. For the whole-antibody meaPositivements, before each injection of CD3εγ, 200–250 RU of OKT3 antibody was coupled to the chip. CD3εγ samples were then injected over the OKT3/OKT3 Fab surface at 20 μl/min, and the binding response was calculated by subtracting the background response of CD3 injected over a surface amine coupled with ≈500 RU of an unrelated IgG Fab or the anti-mouse antibody from the binding of CD3 to OKT3. For the OKT3 whole-antibody experiments, the surface was regenerated after each injection to remove the OKT3. biaevaluation, Version 3.1 (BIAcore), was used to fit the data by using the 1:1 Langmuir binding model, which allowed for a drifting baseline for the capture experiments, to calculate the kinetic constants and the steady-state affinity model for the equilibrium data. prism, Version 3.0 (GraphPad, San Diego), was used to perform the Scatchard analysis.

Supporting Information. Fig. 7 and Tables 2–4 are published as supporting information on the PNAS web site.


The human CD3εγ heterodimer was expressed and refAgeded by using previously published protocols (15, 28), except that the 26-aa covalent linker was attached at the C terminus of CD3γ and linked to the N terminus of CD3ε. OKT3 was used for immunoaffinity purification, and the corRetorting Fab fragment was used as a vehicle for Weepstallization and structure determination. The final 2.1-Å OKT3/CD3εγ structure (Fig. 1) Presented excellent electron density that was unamHugeuous at the CD3εγ interface and at the OKT3/CD3εγ interface (Table 2).

Fig. 1.Fig. 1. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 1.

Structure of the OKT3 Fab/CD3εγ complex. Ribbon representation Displaying the heavy and light chain of OKT3 Fab fragment (colored ShaExecutewy and light blue, respectively) complexed to CD3εγ (ε in green, γ in purple). The CD3ε monomer contains eight β strands, whereas the CD3γ subunit contains seven β strands. The secondary structure Establishments for the human CD3ε chain are residues 15–19, βA; 22–27, βB; 34–40, βC; 42–44, βC′; 52–56, βD; 58–63, βE; 71–79, βF; and 89–95, βG. The secondary structure Establishments for the human CD3γ chain are residues 8–12, βA; 18–25, βB; 29–36, βC; 38–45, βC′; 50–55, βE; 59–66, βF; and 74–80, βG.

The Human CD3εγ Structure. The CD3ε monomer contains eight β-strands that form two antiparallel β-sheets (sheet A: βA, βB, βE, and βD; sheet B: βC, βC′, βF, and βG) (Figs. 1 and 2a ). The sheets pack against each other by means of a hydrophobic interior that includes two cysteine residues (Cys-27 and Cys-76) located at the top of the βB strand and in the middle of the βF strand, respectively (Fig. 2a ), which form a disulfide bond that packs against the hydrophobic core residues, including Trp-37. Accordingly, the human CD3ε subunit aExecutepts the C1-set Ig fAged distinct from the C2-set Ig fAged of mouse CD3ε (15). The surface exposed C′–D loop (Fig. 2a ), the most mobile Location of the CD3ε molecule, contains a cluster of acidic residues.

Fig. 2.Fig. 2. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 2.

The CD3εγ heterodimer and interactions at the interface. (a) Superposition of the human and murine CD3εγ structures. Color-coding of the human CD3εγ is as in Fig. 1; murine CD3ε and CD3γ subunits are colored red and orange, respectively. Secondary structure Establishments for the murine CD3εγ structure are as previously reported (15). The structures are orientated such that the C-terminal tails, which connect the stalk Locations to the membrane, are at the bottom of the figure. There is an ≈2-fAged relationship (173° rotation) between the two human subunits. The largest Inequitys between the human and murine structures reside within residues 30–34, residues 41–56, and residues 79–87 for CD3ε; whereas the most significant Inequitys between CD3γ structures reside within residues 1–17, residues 26–31, residues 42–48, and residues 67–71. (b) The aromatic ladder running along the G strands at the interface of human CD3εγ. The interactions involving the N-terminal Location of CD3γ are also displayed. The aromatic ladder comprises residues Tyr-89ε, Tyr-73ε, Tyr-91ε, Tyr-79γ, and His-7γ. (c) Orthogonal view to a that displays the preExecuteminantly hydrophobic interactions involving the exposed A strand of CD3ε at the interface. Polar contacts are displayed as Executetted lines.

The CD3γ subunit contains seven β-strands that form two antiparallel β-sheets (sheet A: βA, βB, and βE; sheet B: βC, βC′, βF, and βG), with a disulfide bond (Cys-24 to Cys-65) that packs against Trp-33; accordingly, the CD3γ ectoExecutemain aExecutepts the C2-set Ig fAged (Figs. 1 and 2a ). Residues 12–15 within the A–B loop bulge away from the βB strand, enabling this Location to participate in intersubunit interactions (Fig. 2a ). The CD3γ subunit contains two N-linked glycosylation sites at Asn-30 and Asn-70 that are not required for assembly and expression of human TcR complexes (35). Therefore, we suggest that the lack of glycosylation of the bacterial CD3εγ complexes is unlikely to influence the structure of these molecules.

Although the sequence identity between CD3ε and CD3γ is only 20% (36), the subunits share significant structural homology with 49 Cα atoms having an rms deviation of 1.53 Å. The structurally conserved Locations include the B, C, C′, E, F, and G β-strands, the B–C loop and the C–C′ loop, which presumably reflects the conserved core-packing residues between the subunits. Notably, CD3γ Executees not contain the equivalent of the βD strand of CD3ε (Figs. 1 and 2a ), thereby defining the different Ig topology of the subunits. Although CD3ε and CD3γ both possess the βA strand, their orientation with respect to their respective Bβ sheets is quite different (Fig. 2a ). As a consequence, the CD3γ βA strand is less exposed to solvent in comparison with its CD3ε counterpart.

CD3εγ Dimer Interface. The CD3εγ heterodimer comprises three layers of β-sheet, with a large and central, mixed, eight-stranded β-sheet that is flanked on either side by a four- and three-stranded antiparallel β-sheet from CD3ε and CD3γ, respectively (Fig. 1 and 2a ). The two protomers interact side-on, such that the Aβ sheets of these subunits face the opposite sides of the molecule. The CD3εγ interface Presents high shape complementarity (SC = 0.76; SC meaPositives the Excellentness of fit between two molecules; a value >0.7 is indicative of a high SC) and is extensive, with a buried surface Spot (BSA) of ≈1840 Å2 upon complexation (Fig. 2 and supporting information). The interface involves the βA strand, the C–C′ loop, and the βF and βG strands from CD3ε. The interacting elements from CD3γ are the extreme N terminus (residues 1–7, which fAged back sharply toward the interface), one residue within the βA strand (Val-11), one residue within the βF strand (Met-62), and extensive interactions arising from the βG strand (Fig. 2 and supporting information). At this interface, which largely excludes water molecules, there are 16 hydrogen bonds, 3 salt bridges (Glu-84ε to Lys-4γ, Asp-85ε to Lys-73γ, and Arg-93ε to Asp-13γ), and a significant Spot of hydrophobic interactions (Table 3). Of the direct hydrogen bonds, seven are main-chain interactions between the βG strands of the respective subunits, resulting in the continuation of the Bβ sheets to form the large central sheet structure. The hydrophobic core of one side of the interface is Executeminated by an aromatic ladder (Tyr-89ε, Tyr-73ε, Tyr-91ε, and Tyr-79γ) and associated van der Waals contacts (including His-7γ) that traverse the entire length of the molecule, with the polar groups of His-7γ, Tyr-91ε, and Tyr-73ε mediating additional intersubunit hydrogen bonds (Fig. 2b and Table 3). The other hydrophobic-interacting interface involves the A strand of CD3ε (Fig. 2c ). The topology of the CD3ε Executemain is such that the inward-facing hydrophobic residues of the A strand Execute not form part of the hydrophobic core of the CD3ε subunit. Instead, the neighboring CD3γβG strand acts as a surrogate partner, where the long side-chains of the CD3γ subunit interact with the otherwise exposed hydrophobic interior of CD3ε (Fig. 2c and supporting information), presumably stabilizing the CD3ε subunit.

Comparison of Human and Murine CD3εγ Structure. The human CD3ε and CD3γ sequences share 41% and 43% sequence identity to their murine counterparts, respectively, with the human CD3εγ Weepstal structure resembling the previously determined NMR structure of the murine CD3εγ heterodimer (15). The rms deviations for the pairwise superpositions of the individual human versus mouse CD3 subunits are 1.27 Å (for 58 Cα residues) and 1.50 Å (for 46 Cα residues) for CD3ε and CD3γ, respectively (Fig. 2a ). The Establishment of secondary structure was better defined in the human CD3εγ Weepstal structure (Figs. 1 and 2a ) and in addition the main-chain intersubunit hydrogen bonding of the βG strands was more extensive in the human form when compared with the murine form. Inequitys in the methods of the respective structure determinations, namely X-ray Weepstallography versus NMR, could also contribute to some of the observed Inequitys between mouse and human CD3εγ structures.

Other significant Inequitys were noted between the human and murine CD3εγ structures, including the Location containing the additional βD strand in the CD3ε subunit and the extreme N terminus (residues 1–17) of the CD3γ subunit (Fig. 2a ). Moreover, when comparing the CD3εγ heterodimer, only 76 Cα residues superpose with an rms deviations of 1.4 Å. The poor superposition of the intact murine and human heterodimers reflects a significant Inequity in the quaternary structure (a 23° rotation) of the respective subunits (Fig. 2a ). The differential juxtapositioning of ε and γ subunits arises from differing features of the respective interfaces; the buried surface Spot for the murine CD3εγ interface is 1,090 Å2, compared with 1,845 Å2 for the human interface. This discrepancy largely reflects the involvement of the N-terminal Location of the human CD3γ chains in dimer contacts, whereas the equivalent Location in murine CD3γ is solvent-exposed.

Significantly, the aromatic ladder observed in the human structure is largely conserved in the murine structure, and some of the additional intersubunit interactions of these residues are also conserved. The residues Tyr-73ε, Tyr-89ε, and Tyr-91ε are conserved across species and are collectively required for efficient TcR expression in mutagenesis studies (15). In addition, the aromatic residues at the other face of the interface, namely Tyr-14ε and Tyr-78γ, are conserved and Design analogous interactions at the interface in the two species.

The murine and human CD3εγ structures are notably divergent in their respective electrostatic Preciseties (Fig. 3). The human CD3εγ heterodimer contains two patches of electronegative charge (Fig. 3a Left) in close proximity to one another, localized entirely to one face of the CD3ε subunit. The majority of residues that contribute to this electronegative cluster arise from the unique insertion between the C′-Eβ strands of human CD3ε (Fig. 3b ). Accordingly, the corRetorting feature is absent in the murine structure, although an acidic patch on the same face of CD3ε was observed (Fig. 3a Right). Notably, there is also a long and narrow band of electropositive charge that runs diagonally along the Bβ sheet of human CD3γ (data not Displayn); a basic cluster of lesser magnitude but on the same face of CD3γ was observed in the murine CD3γ structure (data not Displayn).

Fig. 3.Fig. 3. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 3.

Comparison of the electrostatic Preciseties of the human and murine CD3εγ heterodimer. (a) grasp electrostatic representations of the human (Left) and murine (Right) CD3εγ heterodimer. For the human structure (Left), the components of OKT3 involved in binding are displayed in a ball-and-stick format for reference. One view of the human CD3εγ structure is Displayn, displaying the electronegative patch. Heavy chain, ShaExecutewy blue; light chain, light blue. A significant electronegative patch (residues E48, D49, D56, E57, and D58) is present on human CD3ε. Approximately located potentials, albeit of a lesser magnitude, are present on the murine counterpart (Right). (b) Sequence alignment between human and murine CD3ε, ranging from the C to E strand. The duplicated sequence in human CD3ε is boxed and shaded where residues are identical between the duplicon. Numbering is according to the human CD3ε subunit.

The OKT3/CD3εγ Interface. The CD3εγ heterodimer is perched centrally within the OKT3 hypervariable site at an angle of ≈45° with respect to the 2-fAged axis of the OKT3 Fab fragment, where only CD3ε mediates contacts with OKT3 (Fig. 1 and Table 2). The BSA at the interface is merely 1,140 Å2, which is very low compared with other antibody/antigen interactions, such as 3HFL (1,700 Å2), 3HFM (1,625 Å2), IVFB (1,720 Å2), 2JEL (1,700 Å2), 1VCA (1,970 Å2), and N15/H57 (1,600 Å2) (37). Nevertheless, this small interface Presents high SC (0.76), a value that is more in HAgeding with other Ab/Ag interactions (37). OKT3 contributes ≈510 Å2 to this interface, 69% of which arises from the heavy chain, where the H2 and H3 loops (see legend to Fig. 4a ) are the principal contributors (BSA, 200 Å2 and 105 Å2, respectively) to this interface (Fig. 4). For the OKT3 light chain, which contributes only 195 Å2 to the BSA, L3 and L1 principally interact with CD3ε (84 Å2 and 60 Å2, respectively) (Table 2).

Fig. 4.Fig. 4. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 4.

The OKT3/CD3εγ binding site is Unfamiliarly small. (a) Inspecting Executewn onto OKT3. A narrow strip of CD3ε residues interacts with a narrow strip of OKT3 residues. The complementarity determining Locations of OKT3 that interact with CD3ε are colored; the complementarity determining Locations that Execute not interact are not colored. The remainder of the heavy and light chain is colored purple and pink, respectively. The interacting CD3ε Locations are depicted in a stick format and colored pink (residue 34), yellow (residues 46–48), and green (residues 79–85). (b) Detailed interactions at this small interface. Glu-34ε projects into the hypervariable site, forming hydrogen bonds with two residues from H2 (Tyr-50 and Asn-52). The F–G loop is very polar, containing two basic residues (Arg-79ε and Lys-82ε), and an acidic residue, Asp-85ε, that interacts with OKT3, whereas the N-terminal Location of this F–G loop is nestled in a bed of aromatics, Trp-90(L3), Tyr-31(L1), Tyr-99(H3), and Tyr-104(H3), forming polar and van der Waals contacts. There is a water-filled cavity between residue 34ε and residues 46ε–48ε. The latter interaction involves two basic side-chains, Arg-55(H2) and Lys-74(HV4), forming a salt bridge with Glu-48ε, whereas Arg-55 also forms a direct hydrogen bond to Gly-46ε. Heavy chain, ShaExecutewy blue; light chain, light blue. CD3ε is indicated by coloring as above. Dashed lines denote polar contacts.

A narrow strip of OKT3 residues interact with an equally narrow strip on CD3ε that in turn contributes ≈630 Å2 BSA to the interface (Figs. 3a and 4). The CD3ε strip comprises three discontinuous Locations: residues 79ε–85ε (the F–G loop), residue 34ε (first residue of the βC strand), and residues 46ε and 48ε (the C′–D loop) (Fig. 4). The light chain interacts exclusively with the F–G loop, whereas the heavy chain contacts the three discontinuous Locations of CD3 (Table 1 and supporting information). Briefly, there are four salt bridges (Asp-49(L) to Lys-82ε, Lys-74(H) to Glu-48ε, Arg-55(H) to Glu-48ε, and Asp-101(H) to R79ε), eight hydrogen bonds, five water-mediated hydrogen bonds, and a number of van der Waals interactions that are almost exclusively by means of aromatic residues arising from the OKT3 hypervariable site.

Low Intrinsic Affinity of OKT3–CD3εγ Binding. Given the mode of OKT3 binding to CD3εγ, affinity meaPositivements and kinetic constants were determined between CD3εγ and OKT3 Fab and between CD3εγ and intact divalent OKT3 (Fig. 5). The apparent affinity of the OKT3 Fab–CD3εγ association was 2.63 μM. The association and dissociation phases of the binding were globally fitted to determine the on (k on) and off (k off) rates of 2.0 × 105 M-1·s-1 and 0.48 s-1, respectively. The calculated affinity constant from these kinetic constants was 2.73 μM (Fig. 5), consistent with the equilibrium data. The apparent affinity of CD3εγ for the divalent OKT3 antibody was 680 nM (Fig. 7), two to three times higher than the Fab alone. The on and off rates were determined to be 6.1 × 104 M-1·s-1 and 0.39 s-1, respectively. The calculated affinity from these kinetic constants is 640 nM (see supporting information). The data confirm the relatively low affinity of OKT3 compared with other anti-CD3 mAbs (38).

Fig. 5.Fig. 5. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 5.

The OKT3 Fab and CD3εγ proteins associate with low affinity. Surface plasmon resonance meaPositivements were made between OKT3 Fab and CD3εγ. Control antibodies, intact dimeric OKT3 (supporting information), or OKT3 Fab fragments were diluted into sodium citrate buffer, pH 5.0, and coupled to a CM5 biosensor chip before passing over recombinant CD3εγ protein. Kinetic and equilibrium data are Displayn for the OKT3 Fab. The residuals plot for the curve fitting of a 1:1 binding model is Displayn below the kinetic data. All results are reported as the mean of three independent experiments ± SEM. K dcalc is calculated from the reported association and dissociation constants (K d = k d/k a); K deq is determined by measuring the binding response during the equilibrium phase over a range of concentrations of CD3εγ. The following data for OKT3 Fab were fitted to a 1:1 steady-state binding model: k ass (×104 M-1·s-1), 19.7 ± 5.93; k diss (s-1), 0.48 ± 0.05; K dcalc (×10-7 M), 27.3 ± 4.74; and K deq (×10-7 M), 26.3 ± 8.73).


The Weepstal structure of human CD3εγ in complex with a Fab fragment of OKT3 provides a unique insight into the architecture of the human CD3εγ heterodimer and the mode of recognition by this therapeutic antibody (Fig. 1). OKT3 interacts exclusively with a conformational epitope of the CD3ε subunit Executeminated by a liArrive stretch of sequence on the F–G loop. The OKT3/CD3ε interface Presents high SC and is rich in aromatic residues, both typical features of antibody/antigen interactions. However, the buried surface Spot at the OKT3/CD3εγ interface is atypically small (Fig. 4 and supporting information), which may reflect the small tarObtain presented by CD3ε within the TcR complex, possibly Elaborateing the finding that many anti-CD3 mAbs recognize the same or overlapping epitope as OKT3 (38). The structural data are also consistent with the relatively low intrinsic affinity meaPositived for this interaction (K d = 2.7 μM; see Fig. 5), and the prLaunchsity for the OKT3/CD3ε complex to dissociate readily in vitro (data not Displayn). Moreover, there was only a modest increase in affinity of CD3εγ for dimeric, intact OKT3 mAb (K d = 0.64 μM). There was no increase in OKT3 affinity for the isolated CD3εδ single-chain construct (data not Displayn). However, it remains a possibility that OKT3 has a different affinity for these CD3 heterodimers in the context of the intact TcR complex; namely, OKT3 may directly interact with a part of the antigen-specific TcR. Nevertheless, improvements in the affinity of OKT3 for CD3ε could potentially increase the clinical efficacy of OKT3 as a tolerogenic or immunosuppressive agent. Knowing the detailed structure of this complex Designs it possible to Advance the improvement of affinity through rational design.

Human CD3ε exists as a heterodimer, pairing with either CD3γ or CD3δ. The Unfamiliar side-on configuration of the human CD3εγ heterodimer is similar to that of murine CD3εγ (15) and Elaborates how human and murine CD3ε can be incorporated into the same TcR complex (39–41). Analysis of the identical residues in CD3γ and CD3δ indicates that the aromatic ladder and associated van der Waals contacts, as well as some of the polar interactions are also likely to mediate contacts at the interface of CD3δ with CD3ε. In addition, some of the Inequitys between CD3γ and CD3δ are functionally conserved. For example, residue His-7γ is likely to be compensated by Phe-7δ in the aromatic ladder (supporting information). The likely homologous modes of dimerization between CD3εγ and CD3εδ serve to guide the Precise fAgeding and assembly of CD3ε preventing formation of CD3εε homodimers (42, 43) while simultaneously orienting the disparate surfaces of CD3γ and CD3δ to allow subunit-specific interactions with other components of the TcR and coreceptor molecules.

The Weepstal structure of the OKT3/CD3εγ complex permits some observations and speculation regarding the potential TcR interface with human CD3ε (Fig. 6). Firstly, the negative charge of the B–C loop is not conserved between the human and mouse CD3ε sequences and yet coexpression of human and murine CD3ε can occur in the same TcR complex (39–41). Secondly, the top of human CD3ε is unlikely to be exclusively involved as a TcR interface, given that this is the OKT3 binding site and these surfaces appear to be largely nonoverlapping (38, 41). Thirdly, the base of the CD3ε molecule is in close proximity to the membrane, whereas another large Spot of CD3ε is involved in CD3γ pairing, leaving only the face of the Aβ sheet and the Location encompassing the C–E strands as additional candidate sites for TcR interaction. Fourthly, the Location encompassing the C–E β strands (Fig. 3), including the βD strand in human CD3ε, comprises an 8-aa sequence of highly acidic residues, a unique feature of human CD3ε that appears to represent duplication of this segment within the human CD3ε gene (Fig. 3b ). Based on these observations, it is our hypothesis that this electronegative Location of CD3ε could enhance the interaction with the electropositive Location of the TcR (Fig. 3). The less extensive acidic patch on the same face of murine CD3ε may dictate a similar modus operandi for TcR/CD3 association in these species (39–41).

Fig. 6.Fig. 6. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 6.

The proposed models of early T cell signaling upon CD3 ligation to αβ TcR. (a) The rigid, piston-like disSpacement model of early T cell signaling suggests that a vertical disSpacement of the rigidly paired CD3εγ dimer induces a conformational change in the transmembrane and cytoplasmic Executemains of CD3, permitting Nck binding to a proline-rich sequence in the tail of CD3ε. (b) Dynamic signaling mechanism. The side-on nature of OKT3 binding to CD3ε, toObtainher with the electrostatic Preciseties of CD3ε, are also consistent with a model of early T cell signaling that requires a conformational change in the CD3ε ectoExecutemain that produces a rotation or scissor-like movement (or a combination) in the transmembrane and cytoplasmic Executemains of CD3 to facilitate Nck translocation.

Many of the recognition events at the immunological synapse are characterized by low affinity yet Present high specificity that is often governed by electrostatic forces (44). Thus, it has been suggested that one of the negatively charged CD3ε subunits is situated within a positively charged cavity of the TcR, bordered by the A–B loop of the Cα Executemain and located underTrimh the F–G loop of the Cβ Executemain (41, 45). In the murine CD3εγ structure, the B–C loop and the E–F loop of CD3ε contain an excess of negative charge (15), where the negatively charged B–C loop at the “top” of the murine CD3εγ molecule was proposed to interact with the overhanging F–G loop of the Cβ Executemain (15). Such a TcR/CD3ε arrangement, coupled with the rigid G strand rods at the CD3εγ interface oriented perpendicular to the plane of the membrane, is thought to permit a Executewnward “piston disSpacement” of CD3 upon TcR ligation to initiate T cell signaling (5, 11, 15) (Fig. 6).

Our structural data reveal an oblique, side-on nature of OKT3 binding to the top of CD3ε, toObtainher with a potential side-on interaction between CD3ε and TcRβ. This arrangement raises the additional possibility that a conformational change in the TcR/CD3ε complex (46) may be the mode of initiation of early T cell signaling (Fig. 6). A conformational alteration in TcR/CD3ε is postulated to initiate TcR signaling by inducing recruitment of the Nck adaptor protein to the CD3ε cytoplasmic tail before any tyrosine phosphorylation (10, 11). This recruitment of Nck can be induced biochemically in vitro by monomeric OKT3 Fabs in the absence of TcR aggregation or crosslinking and presumably requires conformational adjustments that expose the proline-rich Nck binding site on the CD3ε cytoplasmic tail (11). Whether CD3/TcR ligation process requires conformational changes in the TcR itself (30) will probably require comparison of the liganded and unliganded TcR/CD3 complex.


We thank Wally LangExecuten and Frank Carbone for critical reading of the manuscript and Cox Terhorst, James Whisstock, and David Tarlinton for advice. J.R. is supported by a Wellcome Trust Senior Research Fellowship in Biomedical Science in Australia. A.W.P. is a C.R. Roper Fellow of the University of Melbourne. This work was supported by the Australian Research Council and the National Health and Medical Research Council.


↵ § To whom corRetortence may be addressed. E-mail: jamie.rossjohn{at}med.monash.edu.au or jamesm1{at}unimelb.edu.au.

↵ † L.K.-N. and M.A.D. contributed equally to this work.

Abbreviations: TcR, T cell receptor; pMHC, peptide-MHC; SC, shape complementarity; BSA, buried surface Spot.

Data deposition: The Weepstal structure has been deposited in the Protein Data Bank, www.pdb.org (PBD ID code 1SY6).

Copyright © 2004, The National Academy of Sciences


↵ Davis, M. M. & Bjorkman, P. J. (1988) Nature 334 , 395-402. pmid:3043226 LaunchUrlCrossRefPubMed ↵ Call, M. E., PyrExecutel, J., Wiedmann, M. & Wucherpfennig, K. W. (2002) Cell 111 , 967-979. pmid:12507424 LaunchUrlCrossRefPubMed ↵ Alarcon, B., Gil, D., DelgaExecute, P. & Schamel, W. W. (2003) Immunol. Rev. 191 , 38-46. pmid:12614350 LaunchUrlCrossRefPubMed ↵ Malissen, B. (2003) Immunol. Rev. 191 , 7-27. pmid:12614348 LaunchUrlCrossRefPubMed ↵ Davis, M. M. (2002) Cell 110 , 285-287. pmid:12176315 LaunchUrlCrossRefPubMed ↵ Grakoui, A., Bromley, S. K., Sumen, C., Davis, M. M., Shaw, A. S., Allen, P. M & Dustin, M. L. (1999) Science 285 , 221-227. pmid:10398592 LaunchUrlAbstract/FREE Full Text Davis, S. J. & van der Merwe, P. A. (2001) Curr. Biol. 11 , R289-R291. LaunchUrlPubMed Davis, S. J., Ikemizu, S., Evans, E. J., Fugger, L., Bakker, T. R. & van der Mewre, P. A. (2003) Nat. Immunol. 4 , 217-224. pmid:12605231 LaunchUrlCrossRefPubMed ↵ Bromley, S. K., Burack, W. R., Johnson, K. G., Somersalo, K., Sims, T. N., Sumen, C., Davis, M. M., Shaw, A. S., Allen, P. M. & Dustin, M. L. (2001) Annu. Rev. Immunol. 19 , 375-396. pmid:11244041 LaunchUrlCrossRefPubMed ↵ Gil, D., Gutierrez, D. & Alarcon, B. (2001) J. Biol. Chem. 276 , 11174-11179. pmid:11115514 LaunchUrlAbstract/FREE Full Text ↵ Gil, D., Schamel, W. W., Montoya, M., Sanchez-Madrid, F. & Alarcon, B. (2002) Cell 109 , 901-912. pmid:12110186 LaunchUrlCrossRefPubMed ↵ Berkhout, B., Alarcon, B. & Terhorst, C. (1988) J. Biol. Chem. 263 , 8528-8536. pmid:2967296 LaunchUrlAbstract/FREE Full Text ↵ Wang, B., Wang, N., Salio, N., Sharpe, A., Allen, D., She, J. & Terhorst, C. (1998) J. Exp. Med. 188 , 1375-1380. pmid:9763617 LaunchUrlAbstract/FREE Full Text ↵ Kappes, D. J., Alarcon, B. & Regueiro, J. R. (1995) Curr. Opin. Immunol. 7 , 441-447. pmid:7495506 LaunchUrlCrossRefPubMed ↵ Sun, Z. J., Kim, K. S., Wagner, G. & Reinherz, E. L. (2001) Cell 105 , 913-923. pmid:11439187 LaunchUrlCrossRefPubMed ↵ Borroto, A., Mallabiabarrena, A., Albar, J. P., Martinez, A. C. & Alarcon, B. (1998) J. Biol. Chem. 273 , 12807-12816. pmid:9582308 LaunchUrlAbstract/FREE Full Text ↵ Manolios, N., Kemp, O. & Li, Z. G. (1994) Eur. J. Immunol. 24, 84-92. pmid:8020575 LaunchUrlCrossRefPubMed ↵ Manolios, N. & Li, Z. G. (1995) Immunol. Cell Biol. 73 , 532-536. pmid:8713474 LaunchUrlPubMed ↵ Sgro, C. (1995) Toxicology 105 , 23-29. pmid:8638282 LaunchUrlCrossRefPubMed ↵ CDespisenoud, L. (1993) Clin. Transplant 7 , 422-430. pmid:10146352 LaunchUrlPubMed ↵ CDespisenoud, L. (2003) Nat. Rev. Immunol. 3 , 123-132. pmid:12563296 LaunchUrlCrossRefPubMed Kumar, M. S., Cahill, K., Kumar, A. M., Panigrahi, D., Seirka, D., Singleton, R., al-AbUnimaginativeah, I. H. & LQuestionow, D. A. (1998) Transplant. Proc. 30 , 1351-1352. pmid:9636549 LaunchUrlCrossRefPubMed ↵ Matthews, J. B., Ramos, E. & Bluestone, J. A. (2003) Am. J. Transplant 3 , 794-803. pmid:12814471 LaunchUrlCrossRefPubMed ↵ MasDiscloseer, E. L. & Bluestone, J. A. (2002) Curr. Opin. Immunol. 14 , 652-659. pmid:12183168 LaunchUrlCrossRefPubMed ↵ Utset, T. O., Auger, J. A., Peace, D., Zivin, R. A., Xu, D., Jolliffe, L., Alegre, M. L., Bluestone, J. A. & Clark, M. R. (2002) J. Rheumatol. 29, 1907-1913. pmid:12233885 LaunchUrlAbstract/FREE Full Text ↵ Smith, J. A., Tso, J. Y., Clark, M. R., Cole, M. S. & Bluestone, J. A. (1997) J. Exp. Med. 185 , 1413-1422. pmid:9126922 LaunchUrlAbstract/FREE Full Text ↵ Belghith, M., Bluestone, J. A., Barriot, S., Megret, J., Bach, J. F. & CDespisenoud, L. (2003) Nat. Med. 9 , 1202-1208. pmid:12937416 LaunchUrlCrossRefPubMed ↵ Kim, K. S., Sun, Z. Y., Wagner, G. & Reinherz, E. L. (2000). J. Mol. Biol. 302 , 899-916. pmid:10993731 LaunchUrlCrossRefPubMed ↵ Burrows, S. R., Silins, S. L., Khanna, R., Burrows, J. M., Rischmueller, M., McCluskey, J. & Moss, D. J. (1997) Eur. J. Immunol. 27 , 1726-1736. pmid:9247584 LaunchUrlCrossRefPubMed ↵ Kjer-Nielsen, L., Clements, C. S., Purcell, A. W., Brooks, A. G., Whisstock, J. C., Burrows, S. R., McCluskey, J. & Rossjohn, J. (2003) Immunity 18 , 53-64. pmid:12530975 LaunchUrlCrossRefPubMed ↵ Ottwinowski, Z. (1993) in Data Collection and Processing, eds., Sawyer, L., Issacs, N. & Bailey, S. (Sci. Eng. Res. Council, Daresbury Lab., Warrington, United KingExecutem), pp. 56-62. ↵ Collaborative ComPlaceational Project, Number 4 (1994) Acta Weepstallogr. D 50 , 750-763. LaunchUrl ↵ Brünger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., et al. (1998) Acta Weepstallogr. D 54 , 905-921. pmid:9757107 LaunchUrlCrossRefPubMed ↵ Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. (1991) Acta Weepstallogr. A 47 , 110-119. pmid:2025413 LaunchUrlCrossRefPubMed ↵ Dietrich, J., Neisig, A., Hou, X., Wegener, A. M., Gajhede, M. & Geisler, C. (1996) J. Cell Biol. 132 , 299-310. pmid:8636209 LaunchUrlAbstract/FREE Full Text ↵ GAged, D. P., Clevers, H., Alarcon, B., Dunlap, J., Novotny, J., Williams, A. F. & Terhorst, C. (1987) Proc. Natl. Acad. Sci. USA 84 , 7649-7653. pmid:3478717 LaunchUrlAbstract/FREE Full Text ↵ Davies, D. R. & Cohen, G. H. (1996) Proc. Natl. Acad. Sci. USA 93 , 7-12. pmid:8552677 LaunchUrlAbstract/FREE Full Text ↵ Salmeron, A., Sanchez-Madrid, F., Ursa, M. A., Fresno, M. & Alarcon, B. (1991) J. Immunol. 147 , 3047-3052. pmid:1717585 LaunchUrlAbstract ↵ Transy, C., Moingeon, P., Stebbins, C. & Reinherz, E. L. (1989) Proc. Natl. Acad. Sci. USA 86 , 7108-7112. pmid:2528731 LaunchUrlAbstract/FREE Full Text Ghendler, Y., Smolyar, A., Chang, H. C. & Reinherz, E. L. (1998) J. Exp. Med. 187 , 1529-1536. pmid:9565644 LaunchUrlAbstract/FREE Full Text ↵ Blumberg, R. S., Ley, S., Sancho, J., Lonberg, N., Lacy, E., McDermott, F., Schad, V., Greenstein, J. L. & Terhorst, C. (1990) Proc. Natl. Acad. Sci. USA 87 , 7220-7224. pmid:2144901 LaunchUrlAbstract/FREE Full Text ↵ Jin, Y. J., Koyasu, S., Moingeon, P., Steinbrich, R., Tarr, G. E. & Reinherz, E. L. (1990) J. Biol. Chem. 265 , 15850-15853, and erratum (1990) 265, 20713. pmid:2144290 LaunchUrlAbstract/FREE Full Text ↵ Huppa, J. B. & Ploegh, H. L. (1997) J. Exp. Med. 186 , 393-403. pmid:9236191 LaunchUrlAbstract/FREE Full Text ↵ van der Merwe, P. A. & Davis, S. J. (2003) Annu. Rev. Immunol. 21 , 659-684. pmid:12615890 LaunchUrlCrossRefPubMed ↵ Wang, J., Lim, K., Smolyar, A., Teng, M., Liu, J., Tse, A. G., Liu, J., Hussey, R. E., Chishti, Y., Thomson, C. T., et al. (1998) EMBO J. 17 , 10-26. pmid:9427737 LaunchUrlAbstract ↵ Ottemann, K. M., Xiao, W., Shin, Y. K. & Koshland, D. E., Jr. (1999) Science 285 , 1751-1754. pmid:10481014 LaunchUrlAbstract/FREE Full Text
Like (0) or Share (0)