Research ArticleImmunology

Cooperativity Between T Cell Receptor Complexes Revealed by Conformational Mutants of CD3ɛ

See allHide authors and affiliations

Science Signaling  11 Aug 2009:
Vol. 2, Issue 83, pp. ra43
DOI: 10.1126/scisignal.2000402

Abstract

The CD3ɛ subunit of the T cell receptor (TCR) complex undergoes a conformational change upon ligand binding that is thought to be important for the activation of T cells. To study this process, we built a molecular dynamics model of the transmission of the conformational change within the ectodomains of CD3. The model showed that the CD3 dimers underwent a stiffening effect that was funneled to the base of the CD3ɛ subunit. Mutation of two relevant amino acid residues blocked transmission of the conformational change and the differentiation and activation of T cells. Furthermore, this inhibition occurred even in the presence of excess endogenous CD3ɛ subunits. These results emphasize the importance of the conformational change in CD3ɛ for the activation of T cells and suggest the existence of unforeseen cooperativity between TCR complexes.

Introduction

The mechanisms used by receptors for the transmission of information across the plasma membrane are not yet fully understood. Upon binding to ligand, these receptors transmit activation signals by clustering, undergoing conformational changes, or both. The T cell antigen receptor (TCR) is composed of the antigen-binding TCRα and TCRβ subunits noncovalently associated with the signaling subunits CD3γ, CD3δ, CD3ɛ, and CD3ζ (CD247) (1, 2). Although the structures of the ectodomains of the TCRα-TCRβ, CD3ɛ-CD3γ, and CD3ɛ-CD3δ dimers have been elucidated by x-ray crystallography and nuclear magnetic resonance (NMR) studies (36), the organization and structure of the whole TCR complex remains unclear. A conformational change was thought to explain how a monovalent Fab fragment of an antibody specific for the TCR could induce cocapping of the TCR with the CD4 co-receptor (7), although the importance of such a putative conformational change for the activation of T cells was not understood. Indeed, with the exception of the LC13 TCR (8), comparison of the crystallographic data of ligand-bound and unbound ectodomains of TCRα-TCRβ has failed to identify such a conformational change (9, 10). Moreover, it was argued that if the ectodomains of the ligand-binding α and β subunits of the TCR do not undergo a conformational change, such a mechanism could not participate in the activation of T cells. However, triggering of the TCR with stimulatory antibodies or antigen produces a conformational change, which is evident by the exposure of a proline-rich sequence (PRS) in the cytoplasmic tail of the CD3ɛ subunit (11, 12). Exposure of this PRS results in the recruitment of the cytoplasmic adaptor protein Nck through its N-terminal Src homology 3 (SH3) domain. Subsequently, the conformational change in the cytoplasmic tails of CD3ɛ and CD3ζ was witnessed by the use of a conformation-specific antibody and by the acquisition of resistance to digestion by a protease (1315). This conformational change in the tail of CD3ɛ is also proposed to involve a ligand-induced change in its association with membrane lipids (16).

The role of the recruitment of Nck to the PRS during the activation of T cells has remained unclear for some time (17), although the PRS appears to be necessary for the maturation of the immunological synapse (IS) and for the activation of T cells by weak peptide–major histocompatibility complex (pMHC) agonists (18). By contrast, data from studies of knock-in mutants in mice indicate that the PRS is important for the differentiation of thymocytes but not for the activation of mature T cells (19), although the mature cells in these mice could be an “adapted” population.

Regardless of the importance of the recruitment of Nck to the PRS for the activation of T cells, it is still unclear to what extent a conformational change in the TCR might participate in the activation and development of T cells. Here, we present a molecular dynamics (MD) model of how a conformational change is generated in the ectodomains of the CD3ɛ-CD3δ dimer upon binding of a stimulatory antibody specific for CD3. The model predicted that the β strands of CD3ɛ became more rigid, in a stiffening effect that was transmitted to the bottom of the ectodomain of CD3ɛ near the transmembrane domain. Furthermore, particular lysine and cysteine residues at the bottom of CD3ɛ seemed to participate in the transmission of the stiffening effect. By expressing variant CD3ɛ subunits containing mutations of either of these two amino acid residues in T cell lines and bone marrow hematopoietic cell precursors, we showed that these mutations prevented both the transmission of the conformational change to the cytoplasmic tail of CD3ɛ and the activation of T cells. Furthermore, differentiation of T cells was arrested at the pre–T cell stage, suggesting that the pre-TCR was in a constitutively stiffened (active) conformation. In addition, the conformational mutants of CD3ɛ had inhibitory effects even in the presence of an excess of endogenous, wild-type (WT) CD3ɛ. This dominant-negative effect of the CD3ɛ mutants suggests the existence of strong cooperativity between TCR complexes.

Results

Antibody binding to CD3ɛ-CD3δ and CD3ɛ-CD3γ promotes a stiffening effect that rearranges the bottom of the CD3ɛ ectodomain

Because the ligand-binding α and β subunits of the TCR have short cytoplasmic tails and because CD3ζ has a very short extracellular region, we hypothesized that the transmission of the ligand-dependent conformational change to the cytoplasmic tails of CD3 was likely to pass through the ectodomains of the CD3ɛ-CD3γ and CD3ɛ-CD3δ dimers. The structure of the complete TCR complex has not yet been characterized and it is still unknown how the TCRα-TCRβ heterodimer interacts with the ectodomains of the CD3 dimers. Nevertheless, we took advantage of the existing structural information to generate a MD model of how the conformational change could be originated in the ectodomains of the CD3ɛ-CD3δ dimer after the binding of a stimulatory antibody to CD3 (3, 6). Positional fluctuations of each amino acid residue after binding of the antibody were calculated, which showed that both CD3δ and CD3ɛ fluctuated less in the antibody-bound complex than in the unbound form. This effect was especially intense in a region of CD3ɛ that corresponds to the end of the “G” β strand and in the stalk sequence (Fig. 1A). Once the structures were equilibrated (fig. S1), a 9-ns MD simulation showed that the ectodomains of the CD3 dimer became more rigid upon antibody binding. This stiffening effect began in the β strands of CD3ɛ that are closer to the antibody-binding site and it was transmitted to the β strands of CD3δ and to the bottom of CD3ɛ (Fig. 1B and movie S1). A similar effect was detected in the MD model generated with existing structural information from the CD3ɛ-CD3γ dimer after the binding of the stimulatory antibody OKT3 (movie S2) (4, 5).

Fig. 1

A MD model for the origination of a conformational change in the ectodomains of the CD3ɛ-CD3δ dimer. (A) Net changes (along the 9-ns simulation) in mass-weighted positional fluctuations between the values of each residue in the CD3ɛ-CD3δ dimer in the presence or absence of the anti-CD3 antibody UCHT1 (PDB entry 1xiw). (B) Color-coded compaction of the CD3ɛ-CD3δ dimer structure upon antibody binding. Highly motile regions are thicker and in red, whereas weakly motile regions are thinner and in blue; interacting UCHT1 loops are in gray. Time 0 corresponds to the fluctuation in the unbound dimer. (C) Effects of two mutant CD3ɛ subunits on fluctuations in the CD3ɛ-CD3δ dimer. (D) Sequence comparison of the ectodomains of mouse and human CD3ɛ and alignment with the β strand structure.

A comparison of the unbound and antibody-bound CD3 dimer showed that a large rearrangement took place at the bottom of CD3ɛ [the end of strand G and the stalk (fig. S2A)]. Two interesting features are evident in this region: a CXXC motif (C80xxC83) and three amino acid residues, Glu50, Arg76, and Arg78 (E50R76R78), which are in close proximity to each other (fig. S2B). Indeed, binding of antibody promoted the formation of a hydrogen bond between the sulfhydryl hydrogen of Cys80 and the carbonyl backbone group of Arg78 (fig. S2C), thereby linking the C80XXC83 motif with the E50R76R78 cluster. In addition, the carbonyl backbone group of Cys80 established hydrogen bonds with the guanidinium group of Arg70 in CD3δ. Together, these interactions accounted for a stabilization of ~3 kcal/mol. To investigate how this cluster of residues might affect the transmission of the stiffening effect at the bottom of CD3ɛ, MD simulations of a double Arg76→Thr, Arg78→Thr mutant (R76T-R78T) and of a single Cys80 mutant were performed. By the end of the simulation period, the R76T-R78T mutant showed intense fluctuations at the bottom and the Cys80→Gly mutation affected the condensation of the stalk sequence (Fig. 1C).

Mutation of Lys76 and Cys80 in the CD3ɛ ectodomain prevents transmission of the conformational change to the cytoplasmic tail

To determine whether the double arginine (Arg76, Arg78) and the double cysteine (Cys80, Cys83) clusters were indeed important for stabilizing the active state and to facilitate studies in animal models, single mutants of CD3ɛ were generated from FLAG-tagged murine CD3ɛ as the template. The structures of murine and human CD3ɛ-CD3δ dimers are comparable (fig. S3), and the primary sequences of both CD3ɛ proteins are conserved (Fig. 1D). Nevertheless, murine CD3ɛ contains a lysine residue in position 76 rather than an arginine, which is found in human CD3ɛ. Accordingly, the point mutants that were generated in murine CD3ɛ were Lys76→Thr (K76T), Arg78→Thr (R78T), Arg78→Lys (R78K), Cys80→Gly (C80G), Cys80→Ala (C80A), Cys83→Gly (C83G), and Cys83→Ala (C83A).

Murine 2B4 hybridoma cells were transduced with the appropriate plasmids whose expression was assessed by flow cytometric analysis of permeabilized cells to demonstrate that the abundance of each of the FLAG-tagged mutant proteins was similar (Fig. 2A). The transduced 2B4 cells were stimulated with an antibody specific for CD3, lysed, and subjected to pull-down assays with glutathione S-transferase (GST)–fused Nck(SH3.1). When the TCR is in an active conformation, the PRS of CD3ɛ can bind to the N-terminal SH3 domain (SH3.1) of Nck. Therefore, the GST-Nck(SH3.1) pull-down assay serves to estimate how much of the TCR is in the active conformation. Thus, when the material recovered was analyzed by Western blotting with an antibody specific for the FLAG tag, the R78T and R78K mutants did not appear to affect the transmission of the conformational change. By contrast, the K76T mutant fully impaired transmission (Fig. 2B). Furthermore, the mutants C83A and C83G inhibited the transmission of the conformational change to the PRS, whereas the C80G mutant completely abrogated it. These data indicated that Lys76 and Cys80 played an important role in the transmission of the conformational change from the ectodomains of the CD3 dimers to the cytoplasmic tail of CD3ɛ.

Fig. 2

Conformational change–defective mutants of CD3ɛ. (A) The presence of FLAG-tagged mutant CD3ɛ proteins in 2B4 cells was detected by flow cytometric analysis of permeabilized cells. (B and C) Dominant-negative effects on the conformational change induced by the presence of the K76T and C80G mutants of CD3ɛ in 2B4 cells as evident by pull-down assays (PD) (B) or through the exposure of the APA1/1 epitope at the IS (C). *P < 0.05; **P < 0.005 (two-tailed Mann–Whitney test); DIC, differential interference microscopy. Assembly (D) and surface expression (E) of the FLAG-tagged K76T and C80G mutants of CD3ɛ revealed by Western blotting (WB) analysis of surface biotin-labeled cells incubated with streptavidin-conjugated peroxidase (Strept-PO). IP, immunoprecipitation. (F) The transduced mutants of CD3ɛ represent only a minority of the total cellular CD3ɛ. Whole-cell lysates were analyzed by Western blotting with an antibody against CD3ɛ and the relative abundance of each FLAG-tagged CD3ɛ mutant is expressed as a percentage of the total cellular CD3ɛ. pHR, 2B4 cells transduced with the empty vector pHR-SIN. (G) The C80G mutant represents only a minor fraction of the total CD3ɛ found at the cell surface. 2D SDS-PAGE under nonreducing (NR) or reducing (R) conditions in which the inset shows the Western blot after being stripped and incubated with an antibody against the FLAG tag. Data presented are representative of three (C, E, and F), four (D and G), five (B), or eight (A) experiments.

Surprisingly, analyzing the same membrane with a pan-CD3ɛ antibody indicated that inhibition of the conformational change affected all CD3ɛ proteins, including endogenous CD3ɛ (Fig. 2B). Transmission of the conformational change to the PRS in antigen-stimulated 2B4 cells was also examined with the conformation-specific antibody APA1/1 (13). For confocal microscopic analysis, 2B4 cells were stimulated with the murine fibroblast DCEK cell line, transfected with I-Ek and CD80, as antigen-presenting cells (APCs) loaded with moth cytochrome C (MCC) antigen. These results showed that exposure of the APA1/1 epitope at the IS was strongly inhibited in cells that contained the K76T or C80G mutant CD3ɛ subunits (Fig. 2C). Furthermore, the concentration of the TCR at the IS was also diminished, as indicated by staining with an antibody specific for CD3ζ.

The K76T and C80G mutants were assembled and expressed on the cell surface at a similar abundance to that of the WT control (Fig. 2, D and E); however, the transduced WT, K76T, and C80G CD3ɛ proteins only represented a minor fraction of the total cellular CD3ɛ (Fig. 2F). Furthermore, at the cell surface, only a small fraction of the total CD3ɛ corresponded to the CD3ɛ mutants (Fig. 2G). The C80G mutant CD3ɛ protein was resolved by two-dimensional (2D) polyacrylamide gel electrophoresis (PAGE) as a single spot corresponding to isolated, monomeric CD3ɛ (Fig. 2G, inset), indicating that the Cys80→Gly mutation did not lead to aberrant covalent binding to other proteins, including other CD3 subunits. Together, these results suggest that the presence of the K76T and C80G mutants exerted a dominant-negative effect on the transmission of the conformational change within the TCR complex, which also affected the endogenous, WT CD3ɛ.

The presence of CD3ɛ mutants that cannot undergo conformational change exerts a dominant-negative effect on T cell activation

To assess the effects of the conformational mutants of CD3ɛ on the activation of T cells by antigen, several activation markers were analyzed. 2B4 cells transduced with the C80G mutant of CD3ɛ were less sensitive to stimulation by antigen and secreted less interleukin 2 (IL-2) than did cells transduced with WT CD3ɛ (Fig. 3A). In addition, these cells were resistant to activation-induced cell death (Fig. 3B) and cell-surface CD25 was poorly increased in abundance after activation (Fig. 3C). The K76T mutant had a milder effect than did the C80G mutant and only inhibited the activation-induced increase in CD25 abundance, which correlated with its weaker dominant-negative effect on the transmission of the conformational change to the tail of CD3ɛ as detected in the pull-down assay (Fig. 2B).

Fig. 3

Dominant-negative effects of the conformational change–defective mutants of CD3ɛ on the activation of T cells. Dose-response effect of MCC antigen on the secretion of IL-2 (A) and activation-induced cell death (B) after stimulation of transduced 2B4 cells with MCC-loaded DCEK cells. (C) Histograms showing the abundance of surface CD25 after stimulation with MCC (10 μg/ml). Filled histogram, WT CD3ɛ; green trace, K76T mutant CD3ɛ; red trace, C80G mutant CD3ɛ. Antigen-dependent phosphorylation of ERK (D) and tyrosine phosphorylation of CD3ζ (E) after 5 min of stimulation (Stim) with MCC-loaded DCEK. (F) Dose-response effect of the C80G mutant on the phosphorylation of ERK. 2B4 cells were transduced with culture supernatants of HEK 293T cells transfected with mixtures of the lentiviral vector expressing the C80G mutant and the empty pHR-SIN vector. A C80G dose of 1 means that the HEK 293T cells were transfected with 1.5 μg of C80G vector alone; a dose of 0.6 means that the HEK 293T cells were transfected with 1.0 μg of the C80G vector and 0.5 μg of the pHR-SIN vector; whereas a dose of 0.3 means that the cells were transfected with 0.5 μg of C80G vector and 1.0 μg of pHR-SIN vector. A dose of 0 means that HEK 293T cells were transfected with 1.5 μg of the pHR-SIN vector alone. The percentage of the total CD3ɛ corresponding to the FLAG-tagged C80G mutant CD3ɛ protein was calculated by densitometry from Western blotting analysis of FLAG-tagged proteins and total CD3ɛ. Phosphorylation of ERK was measured in lysates of cells stimulated with 145-2C11 (10 μg/ml; antibody against CD3) for 5 min. A plot is shown of the percentage of phosphorylated ERK versus the proportion of total CD3ɛ protein that corresponded to the C80G mutant. Data presented are representative of four (A to E) or three (F) experiments.

To study how a deficient conformational change in the TCR affected events more proximal to TCR activation than secretion of IL-2 or increases in the abundance of CD25, we first studied the phosphorylation of the mitogen-activated protein kinase (MAPK) extracellular signal–regulated kinase (ERK). Antigen-induced phosphorylation of ERK was strongly inhibited by the K76T and C80G mutants (Fig. 3D), as was the phosphorylation of tyrosine residues within the TCR itself, which was evident through analysis of the phosphorylation of CD3ζ (Fig. 3E). Hence, the presence of the K76T or C80G mutants appeared to inhibit antigen-dependent activation of T cells, which suggests that the conformational change in the TCR is necessary for tyrosine phosphorylation of CD3ζ and for full activation of T cells.

To study the strength of the dominant-negative effect of the conformational mutants, we transduced 2B4 cells with different amounts of the plasmid encoding the C80G mutant of CD3ɛ. A dose-dependent effect of the C80G mutant on the phosphorylation of ERK was observed: Complete inhibition of the phosphorylation of ERK occurred when the abundance of the C80G mutant represented 30% of the total endogenous CD3ɛ, whereas 50% inhibition of phosphorylation occurred when the abundance of C80G was only 3% of that of endogenous CD3ɛ (Fig. 3F). Therefore, the C80G mutant of CD3ɛ exerted a potent dominant-negative effect even when 97% of the total CD3ɛ in the cell was the WT form.

Blocking transmission of the conformational change arrests T cell differentiation at an early stage

To study the importance of the conformational change in the TCR for T cell differentiation, bone marrow cells from CD3ɛ-deficient mice were transduced with lentiviral vectors encoding either the K76T or the C80G mutants of CD3ɛ. In the absence of CD3ɛ, T cell differentiation in the thymus was arrested at the double negative (DN) stage (CD4CD8) (Fig. 4A) (20). Differentiation to the next stage (double positive, DP, CD4+CD8+) requires a functional pre-TCR, a version of the TCR that lacks the TCRα subunit and has a surrogate pTα subunit (21). When adoptively transferred into CD3ɛ-deficient mice, CD3ɛ−/− bone marrow cells transduced with WT CD3ɛ reestablished the capacity of the pre-TCR to promote the DN-to-DP transition (Fig. 4, A and B). Surprisingly, the C80G mutant CD3ɛ, and to a lesser extent the K76T mutant, almost completely blocked thymic differentiation at the DN stage. This inhibition was not caused by a reduction in the abundance of the pre-TCR, as indicated by flow cytometric analysis of the DN population with an antibody against CD3 (Fig. 4C). However, the abundance of the TCR in the DP and in the CD4 and CD8 single positive (SP) populations diminished in thymocytes that contained the K76T or C80G mutant CD3ɛ proteins, which indicated that both mutations arrested DP thymocytes at a preselection stage (Fig. 4C). These data suggest the pre-TCR uses the conformational change in CD3ɛ to transduce outside-in signals and to promote differentiation and proliferation of thymocytes.

Fig. 4

The dependence of the function of the pre-TCR on the conformational change in CD3ɛ. (A) Phenotype of thymocytes from CD3ɛ−/− mice reconstituted with bone marrow cells from CD3ɛ−/− mice that were transduced with the indicated CD3ɛ constructs. The percentage of DN, DP, CD4 SP, and CD8 SP thymocytes is shown in the corresponding quadrants. Quantitative data representing the mean ± SD of groups of six mice per condition are shown (B). (C) The abundance of surface pre-TCR is not affected by the K76T and C80G mutations. The abundance of surface pre-TCR in DN thymocytes, and of surface TCR in DP, CD4 SP, and CD8 SP thymocytes in (A) were measured by flow cytometry with an antibody against CD3. MFI, mean fluorescence intensity. The thin gray line indicates the fluorescence intensity of the DN population from CD3ɛ−/− mice, which were used as a negative control. Data presented are representative of nine (A and B) or five (C) experiments.

Conformational mutants of CD3ɛ exert a dominant-negative effect on pre-TCR function and antigen-dependent activation of primary T cells

Because the presence of either of two mutants of CD3ɛ in T cell lines exerted a dominant-negative effect to prevent the TCR from adopting the active conformation (Fig. 2) and to prevent activation of T cells (Fig. 3), we examined whether this dominant-negative effect also occurred in vivo and in primary T cells. Accordingly, bone marrow cells from CD3ɛ+/+ mice transgenic for the OT1 TCR were transduced with either the K76T or the C80G mutants and then adoptively transferred into nontransgenic Ly5.1+ mice. The OT1 TCR recognizes an ovalbumin-derived peptide antigen presented by the class I H-2Kb isotype and drives T cell differentiation to CD8 SP cells in the thymus. Adoptively transferred cells bore the Ly5.2 surface marker, which allowed them to be distinguished from the Ly5.1+ cells of the recipient mice. Intracellular staining of thymocytes from reconstituted mice with an antibody against the FLAG tag indicated that all donor-derived cells contained the corresponding CD3ɛ construct (fig. S4). In the presence of endogenous WT CD3ɛ and the OT1 TCR transgene (thymocytes Vα2+), the presence of either the K76T or the C80G mutant inhibited thymic differentiation at the DN stage (Fig. 5A). This inhibition was not due to a reduction in the abundance of the OT1 TCR at the cell surface, because the transgene was expressed to a similar extent in the DN population irrespective of the transduced CD3ɛ sequence (Fig. 5B).

Fig. 5

Dominant-negative effects of conformational change-defective mutants on both the activation and the differentiation of T cells. (A) Expression of the K76T and C80G mutants of CD3ɛ inhibits the maturation of OT1+ (Vα2+) thymocytes to CD8 SP cells and of OT1 (Vα2) thymocytes to DP cells. Vα2+ thymocytes were analyzed by flow cytometry for the presence of surface CD4 and CD8. The Vα2CD4CD8 DN thymocyte population was analyzed for the presence of the markers CD44 and CD25. Quantitative data showing the mean ± SD (n = 5 mice per group) are presented on the right. (B) The abundance of the OT1 TCR in thymic populations is not affected by the K76T and C80G mutant CD3ɛ proteins. The abundance of the TCR in DN, DP, and CD8 SP thymocytes expressing WT (shaded histogram), K76T (green), or C80G (red) CD3ɛ subunits was measured by flow cytometry with an antibody against Vα2. The thin gray line indicates the fluorescence intensity of the DN population of thymocytes from nontransgenic C57BL/6 mice, which were used as a negative control. (C) The presence of the K76T and C80G mutants of CD3ɛ inhibits the activation of mature T cells. OT1+CD8+ T cells from lymph nodes were stimulated with 100 pM antigen and the activation of the cells was analyzed by measuring the abundance of surface CD69 and CD25 or of intracellular IFN-γ. Proliferation of T cells was calculated by dilution of the PKH26 dye. Thin line, unstimulated cells; shaded histogram, WT CD3ɛ; green trace, K76T mutant CD3ɛ; red trace, C80G mutant CD3ɛ. (D) The abundance of the OT1 TCR in populations of peripheral T cells is not affected by the K76T and C80G mutations. The abundance of the TCR in the CD8+ T cell population was measured by flow cytometry with an antibody against Vα2. LN, lymph nodes; shaded histogram, WT; green trace, K76T mutant CD3ɛ; red trace, C80G mutant CD3ɛ. The thin gray line indicates the fluorescence intensity of the CD8+ T cell population from nontransgenic C57BL/6 mice, which were used as a negative control. Data presented in (A) to (D) are representative of four experiments.

Differentiation of thymocytes within the transferred Ly5.2+ OT1 population (Vα2) was inhibited or arrested in the presence of the K76T or C80G mutants, respectively. Furthermore, DN cells were arrested at the DN3 stage (CD44CD25+), which typically occurs when there is a defect in the function of the pre-TCR (Fig. 5A) (21). These results show that the mutant CD3ɛ proteins that cannot undergo a conformational change exerted a dominant-negative effect on the function of the pre-TCR. The differentiation of OT1+ thymocytes to CD8 SP cells was inhibited, but not blocked, by the presence of the conformation-defective mutants (Fig. 5A). This enabled us to study the response of peripheral mature CD8+ T cells to antigen. Primary OT1+ CD8+ T cells containing either of the mutant CD3ɛ subunits proliferated less, produced less interferon-γ (IFN-γ), and had a lower abundance of the surface markers of activation CD69 and CD25 in response to antigen stimulation than did cells containing WT CD3ɛ (Fig. 5C). As before, the inhibitory effect of the K76T and C80G mutant CD3ɛ subunits on the activation of T cells was not caused by reduced expression of the OT1 transgene in peripheral T cells (Fig. 5D). Hence, the presence of the K76T and C80G mutants of CD3ɛ again appeared to exert a dominant-negative effect on T cell activation.

Discussion

The results presented here illustrate how point mutations in the ectodomain of CD3ɛ abolished the transmission of the ligand-dependent conformational change to the cytoplasmic tail of CD3ɛ. Thus, the effects of the K76T and C80G mutants on T cell activation fulfill the dual objective of demonstrating how important the conformational change in the TCR is for T cell differentiation and activation, as well as supporting the dynamic model of the origin of the conformational change in the ectodomains of the CD3 dimers. This model was based on the structure of the CD3 ectodomains in the presence and absence of stimulatory antibodies against CD3. Nevertheless, the inhibition of both antigen-driven T cell activation and the conformational change in the PRS by the K76T and C80G mutants indicates that the model of CD3 ectodomain compaction is also valid during transmission of the conformational change between the α and β chains of the TCR and CD3 upon engagement by the pMHC ligand. The effect of the binding of an antibody against CD3 to the CD3 dimers (that is, the compaction of the β strands of the CD3 dimer), may represent a laboratory-generated recreation of the effect of the TCR αβ dimer on CD3 after the binding of pMHC. In this regard, the position of the AB loop of the Cα domain of TCRα, shown by x-ray crystallography to undergo a conformational change after the binding of pMHC (8), could be proximal to the CXXC motif of CD3ɛ.

The CXXC motif that is present in the stalk sequences of the CD3ɛ, CD3γ, and CD3δ subunits was proposed to be involved in the formation of CD3 dimers (5, 22, 23). The CXXC motif lies in the N terminus of the stalk sequence, where it could help to stabilize an α helix, as has frequently been found in other proteins (24). The CXXC motif does not participate in the formation of interchain disulfide bridges, and, indeed, there is no clear evidence that an intrachain disulfide bond is formed. Given the effects of the C80 and C83 mutants on the transmission of the conformational change to the tail of CD3ɛ, it is tempting to speculate that the CXXC motif plays an important role as a transducer of the stiffening effect from the β strands of CD3ɛ to the transmembrane domain, reinforcing the propensity of the stalk sequence to form an α helix. It is noteworthy that the Cys80→Ala mutation had little effect on the transmission of the conformational change, whereas the conformational change was completely abolished by the Cys80→Gly mutation. Indeed, these data correlate with the propensity of Ala residues to stabilize α helices and of Gly residues to disrupt them (25). The CXXC motif contained in the stalk sequence is placed at the narrow part of a “functional funnel”; that is, the stiffening effect provoked in the β strands of CD3ɛ has to be funneled to the transmembrane and cytoplasmic domains of CD3ɛ through the CXXC sequence. This could explain why other positions placed upstream, such as Lys76, play a less important, redundant role in transmission of the conformational change.

We also found that the capacity of the C80G mutant to compete with endogenous WT CD3ɛ for interacting with the other TCR subunits was somewhat reduced, indicating that the C80G mutation affected the assembly of CD3ɛ to some extent. This effect of the C80G mutant on TCR assembly does not explain its dominant-negative effect in the presence of an excess of endogenous CD3ɛ; indeed, the amount of TCR at the cell surface was not affected and the ratio of mutant to WT CD3ɛ was low (3 to 30%). Furthermore, the abundance of the pre-TCR was not reduced when the C80G mutant was expressed in CD3ɛ−/− mice (that is, in the absence of competing WT CD3ɛ), indicating that inhibition of the function of the pre-TCR was not due to a defect in assembly but to impaired signal transduction within the pre-TCR. Indeed, the inhibitory effect of the K76T and C80G mutations on the function of the pre-TCR is one of our more puzzling results. The pre-TCR does not seem to require ligand binding; rather, its low abundance at the plasma membrane appears to be sufficient to drive the DN-to-DP transition (21). Therefore, if there is no ligand to induce the conformational change in the pre-TCR, the inhibitory effect of the K76T and C80G mutations might only be explained if the stiffened (active) conformation was constitutively adopted by CD3 dimers in the pre-TCR. This would explain why the presence of a mutant CD3ɛ in the pre-TCR that cannot adopt the stiffened conformation abrogated the function of the pre-TCR. Interestingly, retroviral expression of a C82SxxC85S double mutant of CD3γ in fetal thymic organ cultures did not affect the abundance of the pre-TCR, although it did inhibit its function (23). It would therefore be interesting to test whether the CXXC motif of CD3γ, and that of CD3δ, is also involved in transmitting the conformational change to the cytoplasmic tails of CD3 in the TCR. Similarly, transgenic mice expressing a mutant CD3ɛ that has all five proline residues in the PRS mutated to alanine exhibit a 10-fold reduction in the number of thymocytes compared to that of WT mice, suggesting that the PRS is necessary for the pre-TCR–mediated proliferation of DN thymocytes (26). According to these data, the deficient function of the pre-TCR in the presence of the K76T or C80G mutants is therefore consistent with the idea that the PRS of CD3ɛ mediates, at least in part, signal transmission by the active conformation.

In thermodynamic terms, the compaction of the β strands of the CD3 dimer reflects an increase in order and, therefore, a decrease in the entropy of the ectodomains of CD3. Interestingly, the acquisition of protease resistance when the cytoplasmic tail of CD3ɛ adopts the active state (14) is in keeping with a model wherein the compaction that occurs in the ectodomains of CD3 is transmitted to the cytoplasmic tails. The active and inactive states of the CD3 dimers could therefore parallel the “frozen” and “melted” states proposed for the chemotaxis receptors of Escherichia coli (27). The decrease in entropy in both the ectodomains and the cytoplasmic tails of the CD3 dimers would make the conformational change unfavorable, according to the standard thermodynamic expression ΔG = ΔHTΔS. Therefore, this decrease in entropy in the CD3 dimers must be compensated for by an increase in entropy (ΔS) or a decrease in enthalpy (ΔH) in other parts of the TCR complex, such as the TCR-pMHC interface (28).

The dominant-negative effect produced by the K76T and C80G mutants in the presence of excess, endogenous WT CD3ɛ can only be explained in the context of oligomeric receptors. If the TCR complex consisted exclusively of monomeric complexes with an αβγɛδɛζ2 stoichiometry, the presence of two CD3ɛ subunits per complex could not explain why the C80G mutant (our most potent conformational mutant) could exert such a strong negative effect on the induction of the conformational change and on T cell activation when its abundance was as low as 3% of that of endogenous, WT CD3ɛ. Indeed, in such circumstances more than 99.9% [that is, 100 − (0.03 × 0.03 × 100)] of the TCR complexes would contain at least one WT CD3ɛ subunit and 94% (that is, 0.97 × 0.97 × 100) would contain two. Therefore, the majority of the TCRs should be competent for signaling. However, the TCR exists as linear arrays of up to 20 receptor complexes before the exposure of T cells to antigen (29). Expression of a mutant CD3ɛ subunit that cannot adopt the active conformation may prevent TCRs that exclusively contain WT CD3ɛ subunits from adopting the active conformation if both types of receptor form part of the same oligomer. Preassembled, oligomeric TCR complexes could, therefore, constitute a platform for the transmission of conformational changes from engaged to nonengaged receptors. Upon encounter with pMHC on the surface of APCs, the TCR forms clusters known as microclusters that are detectable by light microscopy (30). It would be interesting to assess whether the presence of a few conformational mutants of CD3ɛ subunits could still affect the function of these larger microclusters. Nevertheless, we believe that the dominant-negative effects of the conformational mutants must be exerted on oligomers of TCRs before stimulation rather than on microclusters of TCRs after stimulation.

The dramatic effects of the CD3ɛ mutants, even when they only represent a minor fraction of total CD3ɛ, suggest that the TCR follows a Monod-Wyman-Changeaux (MWC) model of allostery, whereby all TCR units in the oligomer adopt the same conformation at the same time (27, 31) (Fig. 6, A and B). Although we do not have definitive proof to refute alternative allosteric models, such as the Koshland-Néméthy-Filmer (KNF) or the conformational spread (CS) models, we believe the MWC model is the most adequate to explain the potent dominant-negative effects of the conformational mutants (Fig. 6). Although the MWC model was initially formulated for enzymes in solution and only later extrapolated to lattices of neuronal membrane receptors (31), the TCR has a peculiarity when compared to other membrane receptors. In addition to its agonistic pMHC complex, the TCR has some affinity for MHC bound to self-peptides. During the recognition of agonistic pMHCs, it is most likely that, rather than being freely available, the TCR units within an oligomer that are not occupied by agonistic pMHC make contact with MHC that is bound to self-peptides. Indeed, complexes of MHC with self-peptide contribute to the signal promoted by agonistic pMHC complexes (32). Therefore, the MWC model adapted to the TCR would be one in which the binding of agonistic pMHC to individual TCRs within an oligomer promotes a conformational change in the contacting TCR and in all of the other TCRs within the same oligomer. Such a phenomenon would thereby facilitate their binding to MHC–self-peptide complexes and promote their participation in TCR triggering. This cooperativity between TCR complexes could explain the paradoxical properties of high sensitivity and low affinity in the response of T cells to antigen.

Fig. 6

An adapted MWC model of allostery for the TCR. (A) Possible consequences of the presence of a conformational change–defective mutant of CD3ɛ within an oligomer of TCRs that contains WT CD3ɛ proteins. TCR oligomers are depicted as repetitions of TCR αβ heterodimers (white rectangles) bound to CD3 dimers (red) with “loose” cytoplasmic tails (inactive conformation). In the active conformation, the ectodomains of the CD3 dimers (in blue) are compacted (rectangles) and their tails are “locked” (the frozen state). The MHC ligand is depicted as a cluster of gray rectangles bound to either an agonistic peptide (yellow square) or a self-peptide (white square). The presence of a mutant CD3ɛ molecule (intense red) in the TCR oligomer in a forced inactive conformation influences the ability of the other TCRs to undergo the conformational change. According to the MWC model of allostery, essentially an all-or-none model, all TCR units in the oligomer adopt the same conformation at the same time, irrespective of the quality of the MHC-bound ligand that binds to each individual subunit (that is, if they are binding MHC bearing an agonistic peptide or a self-peptide). In the KNF model, according to a sequential induced-fit mechanism, the binding of agonistic pMHC to TCRs in the oligomer forces these TCRs to adopt the active conformation, which in turn causes neighboring TCR units to bind to other MHC-peptide complexes. The existence of preformed TCR oligomers bound to preformed MHC-peptide clusters would condition the availability of agonistic pMHC. Therefore, the most likely ligand for the neighboring TCR unit would be an MHC unit bearing a self-peptide. According to the induced-fit mechanism intrinsic to the KNF model, the conformational change cannot spread beyond the TCR containing the constitutively inactive mutant of CD3ɛ because the conformational change would not propagate to the neighboring TCR units. However, the conformational change could propagate within the TCR oligomer, to “all-WT” TCR units, in the opposite direction. In the Bray–Duke CS model, an individual TCR unit has a certain probability of being in the active or inactive conformation depending on whether it is bound to MHC bearing agonistic peptide or self-peptide and on the conformational states of its two neighbors. Again, the presence of a TCR unit containing the conformational mutant of CD3ɛ would make the spread of the conformational change beyond the mutant unlikely. However, the conformational change could propagate in the opposite direction. (B) Therefore, the potent dominant-negative effect of a mutant CD3ɛ would favor a MWC model in which all individual TCR units in the oligomer adopted the same conformation irrespective of the quality of the MHC ligand that is recognized by each individual unit within the oligomer and even of their binding to a ligand at all. In the MWC model, binding of agonistic pMHC to a few units in the oligomer forces all other TCR units to adopt the active conformation. The presence of a single TCR unit incompetent to adopt the active conformation in the oligomer would prevent the other TCR units from undergoing the conformational change even if they were bound to agonistic pMHC.

Materials and Methods

Mice

CD3ɛ knockout mice (CD3ɛ−/−) (20), C57BL/6 mice bearing the Ly5.1 hematopoietic cell marker, and TCR transgenic mice expressing the OT1 TCR (H-2Kb–restricted, specific for ovalbumin, OVAp, peptide SIINFEKL) (33, 34) were all bred and maintained free of pathogens at our animal facility.

Cells

The human Jurkat T cell line and the murine I-Ek–restricted and MCC-responsive 2B4 T cell hybridoma were maintained in complete RPMI 1640 supplemented with 5% fetal bovine serum (FBS, Sigma). The DCEK fibroblast cell line stably transfected with plasmids encoding I-Ek and CD80 was provided by R. Germain (National Institutes of Health, Bethesda, MD).

Plasmids

The pGEX-4T1 derivative GST-SH3.1, which contains the N-terminal SH3 domain of Nckα, was provided by R. Geha (Children’s Hospital, Harvard Medical School, Boston, MA). Point mutations in human and murine CD3ɛ were made with QuickChange-XL (Stragene). The FLAG epitope was added to regions encoding the C terminus, and the complementary DNA (cDNA) encoding human CD3ɛ was cloned into pSRα. The green fluorescent protein (GFP)–expressing HIV vector pHRSIN-WPRE (provided by J. A. Pintor) (35), in which murine CD3ɛ cDNA replaced GFP, was used to construct pHRSIN-mCD3ɛ.

Antibodies and other reagents

The APA1/1 mouse monoclonal antibody (mAb) against CD3ɛ has been described previously (36), as has rabbit antiserum 448 against CD3ζ (37). The other antibodies used are as follows: hybridomas producing the 145-2C11 mAb against mouse CD3ɛ and OKT3, a mAb against human CD3ɛ, were obtained from ATCC; 4G10, a mAb against phosphotyrosine residues, was purchased from Upstate Biotechnology; antibodies against ERK and phosphorylated ERK were obtained from Cell Signaling; M20, an antibody against CD3ɛ, was purchased from Santa Cruz; M2, a mAb against the FLAG tag was from Sigma; fluorescein isothiocyanate (FITC)–conjugated H57-597 antibody against mouse Cβ, phycoerythrin (PE)-conjugated Vα2 antibody against TCRα, allophycocyanin (APC)-conjugated antibody against Ly5.2, PE-conjugated antibody against mouse CD4, FITC-conjugated antibody against CD4, PE-conjugated antibody against CD8, biotin-conjugated antibody against CD8, PerCP-conjugated antibody against CD8, FITC-conjugated antibody against CD44, biotinylated antibody against CD25, APC-conjugated antibody against CD25, and FITC-conjugated antibody against CD69 were all purchased from BD-Pharmingen. The IFN-γ intracellular staining kit was obtained from Miltenyi, and the red PKH26 dye was purchased from Sigma. The OT1 TCR agonist (OVAp, SIINFEKL) and the 2B4 TCR agonist (MCC, ANERADLIAYLKQATK) (34) peptides were synthesized at our facility by the f-moc method.

Molecular dynamics

The initial models for CD3ɛ and CD3δ, both alone and bound to the single-chain antibody fragment UCHT1, were taken directly from the Protein Data Bank (PDB) [accession code 1XIW (3)]: chain A (ɛ monomer), chain B (δ monomer), chain C (UCHT1 light chain), and chain D (UCHT1, heavy chain). The missing loop between residues 53 and 62 (joining the F and G β sheets of the δ monomer) was modeled with the Swiss-Model Web server (http://swissmodel.expasy.org/SWISS-MODEL.html). The positions of hydrogen atoms and standard atomic charges and radii were assigned according to the ff99 force field (38). The two models were immersed in cubic boxes of transferable intermolecular potential 3 point (TIP3) water molecules (39) large enough to guarantee that the shortest distance between the solute and the edge of the box was greater than 15 Å. Counterions were also added to maintain electroneutrality. Three consecutive minimizations were performed: (i) involving only hydrogen atoms; (ii) involving only the water molecules and ions; and (iii) involving the entire system. The minimized starting structures were prepared as indicated above and they were simulated in the NPT (N, total number of atoms; P, pressure; T, temperature) ensemble with the periodic boundary conditions and particle mesh Ewald method to treat long-range electrostatic interactions. The systems were then heated and equilibrated in two steps: (i) 20 ps of MD heating the whole system from 100 to 300 K and (ii) equilibration of the entire system during 1.1 ns at 300 K. The equilibrated structures were the starting points for 9-ns MD simulations at constant temperature (300 K) and pressure (1 atm). The constraint algorithm SHAKE was used to keep bonds involving H atoms at their equilibrium length, allowing a 2-fs time step for the integration of Newton’s equations of motion. The ff99 and TIP3P force fields were used to describe the proteins and water molecules, respectively, as implemented in the AMBER (Assisted Model Building with Energy Refinement) 8 package (40). This last part of each trajectory (9 ns) was used to sample frames at 1-ps intervals, which were subsequently used for the analysis. Mass-weighted positional fluctuations were calculated with the ptraj program in AMBER 8 for the entire trajectories at each nanosecond, taking the differences between the values of a residue in each system in the presence or absence of the bound antibody. These differences were transformed into Z score values with a window of three residues. Finally, they were mapped onto each system structure not bound to the antibody and were color-coded from blue (loss of movement) to red (enhanced movement) before representing them as a sausage-style cartoon of the backbone in PyMOL [DeLano, The PyMOL Molecular Graphics System, DeLano Scientific, Palo Alto, CA (2002)]. The double (R76T and R78T) and single (C80G) mutants were prepared by the aforementioned models, and the mutation wizard was implemented in PyMOL. The same computational protocol was used in both cases. Configuration entropies were calculated with the ptraj program in AMBER 8. Briefly, the mass-weighted covariance matrix was obtained from the dynamics simulations and diagonalized. The eigenvalues were then converted into frequencies after quasiharmonic analysis (41, 42). Finally, standard statistical mechanics formulas were used to obtain the entropy values. A major drawback of the method is the dependence of the entropy on the length of the trajectory; however, in many cases this relationship can be fitted to an exponential function that allows the entropy at infinite simulation time to be estimated from finite simulations (43).

Lentiviral transduction

Lentiviral vectors were produced by cotransfecting human embryonic kidney (HEK) 293T cells with the pHRSIN-CD3ɛ, pCMVΔR8.1, and pMDG plasmids (44) with the JetPEI reagent (PolyPlus Transfection). The culture supernatants obtained were used to transduce target cells. For transduction of bone marrow cells, 8-week-old CD3ɛ−/− and OT-I TCR transgenic mice were intravenously inoculated with 5-fluorouracil and 2 days later they were killed. Bone marrow cells were collected from the femurs and cultured for 12 hours at 37°C in Iscove’s medium supplemented with 10% FBS, IL-7 (25 ng/ml), IL-6 (25 ng/ml), stem cell factor (SCF, 50 ng/ml), and Flt3-ligand (50 ng/ml). All cytokines were obtained from Peprotech. The bone marrow cells were then transduced by incubation with HEK 293T cell supernatants that were collected 48 hours after transfection with lentiviral vectors. Cells were centrifuged with the filtered supernatants for 90 min at 1500g at 32°C in six-well culture plates. The cells were harvested 24 hours later and injected in mice.

Bone marrow reconstitution

Transduced bone marrow cells (3 × 106 to 5 × 106 cells per mouse) were injected intravenously into sublethally (600 rad; 1 Gy = 100 rads) irradiated CD3ɛ−/− or C57BL/6 (Ly5.1+) recipient mice. Eight weeks after injection, the mice were killed and the thymus, lymph nodes, and spleen were removed for analysis and for stimulation experiments.

Pull-down assays

For the GST-Nck pull-down assay, cell lysates were first precleared with GST adsorbed to glutathione-Sepharose (Amersham Biosciences) before affinity chromatography was performed with the N-terminal SH3 domain of Nckα fused to GST and adsorbed to glutathione-sepharose (11).

Confocal microscopy

2B4-DCEK cell conjugates were incubated with an antibody against CD3ζ and with APA1/1 (13, 15). The proportion of conjugates in which antibody staining redistributed to the T cell–APC contacts was calculated by randomly selecting 100 different conjugates. T cell–APC conjugates were identified through their cell morphology under differential interference microscopy. Confocal microscopy was performed with a Zeiss Radiance 2000 with a 63× objective and analyzed with the ImageJ 1.33v software (NIH).

T cell stimulation and flow cytometry

To analyze cells for cell-surface CD25, 5 × 104 DCEK cells were loaded overnight with different concentrations of MCC peptide. The supernatant was discarded and 15 × 104 transduced 2B4 cells were added for 3 to 24 hours at 37°C. Cells were incubated with an APC-conjugated antibody against CD25 and analyzed on a FACSCalibur (Becton Dickinson) flow cytometer. To analyze the release of IL-2, the culture supernatant of 2B4 cells was collected 24 hours after stimulation and analyzed with an enzyme-linked immunosorbent assay kit (Bender Medsystems). Activation-induced cell death was measured 24 hours after stimulation by propidium iodide exclusion. Eight weeks after bone marrow reconstitution, lymph node cells were collected and stimulated by adding OVAp (100 pM). The abundance of CD69, CD25, and intracellular IFN-γ were analyzed 24 hours after stimulation. To measure antigen-dependent proliferation, splenic cells from mice reconstituted with bone marrow cells were labeled with PKH26 and stimulated by the addition of OVAp (100 pM). The dilution of PKH26 fluorescence in the CD8+ T cell population was recorded by flow cytometry 60 hours after stimulation.

Cell labeling and immunoprecipitations

For biotinylation of surface proteins, 50 × 106 2B4 cells were incubated for 45 min on ice with sulfo-N-hydroxysuccinimide-biotin (0.5 mg/ml; Pierce) in phosphate-buffered saline supplemented with 0.1 mM CaCl2 and 1 mM MgCl2. After washing, the cells were lysed in Brij96 lysis buffer [0.33% Brij96, 150 mM NaCl, 20 mM tris-HCl (pH 7.8), 10 mM iodoacetamide; 1 mM phenylmethylsulfonyl fluoride (PMSF), aprotinin (1 μg/ml); leupeptin (1 μg/ml)] and immunoprecipitation was carried out with antibodies against FLAG, CD3ζ (448), or CD3 (OKT3) bound to protein A–sepharose beads. The immunoprecipitated material was washed five times and subjected to 2D SDS-PAGE, with the first dimension under nonreducing conditions, and the second dimension under reducing conditions. The proteins recovered were then transferred by Western blotting to a nitrocelullose membrane that was incubated with streptavidin-conjugated horseradish peroxidase (Southern Biotechnology) and developed by ECL (Pierce).

Note added in proof: Since this paper was submitted, Rossjohn and colleagues used spectroscopic methods to demonstrate movement of the AB loop of the Cα domain upon engagement of soluble TCR αβ heterodimers by pMHC (45). This study adds further evidence to the notion of a conformational change being transmitted to the CD3 ectodomains by the AB loop of TCRα. Whether the AB loop is in close proximity to the CXXC motif of CD3ɛ remains to be determined.

Acknowledgments

We thank J. A. Pintor and C. Ardavín for providing reagents and H. M. van Santen, R. Blanco, and M. Sefton for critical reading of the manuscript. We are also indebted to C. Prieto and I. Arellano for their expert technical assistance. This work was supported by grants SAF2006-01391 from the Comision Interministerial de Ciencia y Technologia (Spain), SAL-0159/2006 from the Comunidad de Madrid, RD06/0020/1002 from Redes Temáticas de Investigación Cooperativa Sanitaria (Spain), and LSHC-CT-2005-018914 and FP7/2007-2013 from the European Union. Institutional support from the Fundación Ramón Areces to the Centro de Biología Molecular Severo Ochoa and the generous allocation of computer time at the Barcelona Supercomputing Center are also acknowledged. The authors declare no conflict of interest. We dedicate this paper to the memory of Angel R. Ortiz.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/2/83/ra43/DC1

Fig. S1. RMSD evolution of CD3ɛ-CD3δ dimers.

Fig. S2. Comparison of unbound and antibody-bound CD3ɛ-CD3δ dimers.

Fig. S3. Comparison of the structures of mouse and human CD3ɛ-CD3δ dimers.

Fig. S4. Expression of FLAG-tagged mutants of CD3ɛ in Ly5.2+ thymocytes from bone marrow chimeras.

Movie descriptions

Movies S1 and S2

References and Notes

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. Abbreviations for the amino acids are as follows: A, Ala; D, Asp; E, Glu; F, Phe; I, Ile; K, Lys; L, Leu; N, Asn; Q, Gln; R, Arg; S, Ser; T, Thr; and Y, Tyr.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
View Abstract

Stay Connected to Science Signaling

Navigate This Article