The activity of the αβ TCR is controlled by conformational switches. In the resting conformation, the TCR is not phosphorylated and is inactive. Binding of multivalent peptide-MHC to the TCR stabilizes the active conformation, leading to TCR signaling. These two conformations allow the TCRs to be allosterically regulated. We review recent data on heterotropic allostery where peptide-MHC and membrane cholesterol serve opposing functions as positive and negative allosteric regulators, respectively. In resting T cells cholesterol keeps TCRs in the resting conformation that otherwise would become spontaneously active. This regulation is well described by the classical Monod-Wyman-Changeux model of allostery. Moreover, the observation that TCRs assemble into nanoclusters might allow for homotropic allostery, in which individual TCRs could positively cooperate and thus enhance the sensitivity of T cell activation. This new view of TCR regulation will contribute to a better understanding of TCR functioning.
Life depends to a large extent on the precise control of molecular events. Proteins occupy a central place in molecular regulation. In the early 1960s, it became clear that different ligands can positively or negatively affect the function of a protein at the same time. This has raised the question of how such inputs are integrated. Moreover, in several cases switch-like responses of the proteins were seen, profoundly changing their activity over a relatively small range of ligand concentration. The landmark paper by Monod et al. (1) first presented an elegant and unified explanation of these phenomena by introducing a quantitative model of allosteric regulation; to date, it remains one of the most cited papers of molecular biology. Allostery is derived from the Greek words αλλο (different) and στερεο (relating to solid objects and having come to signify three-dimensionality). Thus, allosteric regulation indicates that binding of one ligand at one site of a protein, or protein complex, changes the binding of a second ligand, or a substrate, at a topographically distinct site of the protein (1, 2). The first ligand can enhance or reduce binding of the second ligand, being a positive or negative allosteric regulator, respectively. Hence, allosteric regulation is distinct from competitive inhibition, resulting from substrate and inhibitor binding to the same site. Switch-like responses typically arise from the interaction of multiple subunits in allosteric protein complexes and thus multiple ligand binding sites. The Monod-Wyman-Changeux (MWC) model, as it has come to be called, has stood the test of time remarkably well, explaining the regulation of hemoglobin, enzymes and membrane receptors (2–5).
The αβ TCR is an oligomeric protein complex that forms higher-order clusters in the plasma membrane of T cells (6). Although traditionally the TCR has not been associated with allostery, recent data strongly suggest it is regulated allosterically. In this Brief Review, we will discuss the evidence supporting this idea.
Following the concept introduced by Monod et al. (1), allostery is based on the premise that a regulated protein can exist in two different conformations, which can be defined by changes of the protein’s tertiary and/or quaternary structure. In one conformation, traditionally called the tense state, the protein is inactive and binds to its ligand with low affinity. In the other conformation, the relaxed state, the protein is active and binds with high affinity to the ligand. Usually, the tense state represents the resting state of the protein. Upon binding of the ligand, the protein switches to the relaxed state, the active state. Two different mechanisms have been proposed to explain the molecular mechanism leading to the conformational switch in oligomeric proteins.
On one hand, the Koshland-Nemethy-Filmer (KNF) model suggests the ligand binds to the protein in its tense conformation and by an induced-fit mechanism instructs the protein to switch to the relaxed conformation (7). Thus, only those proteins that are ligand-bound switch to the relaxed state. On the other hand, the MWC model proposes that the protein can switch spontaneously from the tense to the relaxed state, even in the absence of a ligand, in a process called allosteric transition (1). The ligand binds with higher affinity to the relaxed conformation of the protein rather than instructing the protein to change its conformation. This is called selection of a conformation. According to the MWC model, negative allosteric regulators bind to the protein in the tense state and thus shift the equilibrium to the inactive state. In contrast, positive allosteric regulators bind to the relaxed state and thus stabilize the protein in the active state (Fig. 1). In most proteins studied in detail, the MWC model provides an excellent fit to the experimental data (2, 3).
Moreover, two forms of allostery can be distinguished: heterotropic and homotropic. Heterotropic allostery works with mono- and multimeric proteins. Binding of a regulator to the protein regulates the binding of another ligand or substrate (Fig. 1A). This effect is reciprocal for thermodynamic reasons; the binding of the ligand/substrate also affects the binding of the regulator. Thus, at least two different binding partners regulate the protein’s function. Homotropic allostery occurs if the protein is an oligomer. The binding of the first ligand changes the affinity of the protein for the second, chemically identical, ligand (Fig. 1B). According to the MWC model, the conformational change occurs simultaneously for all binding sites (or subunits) of an oligomer or for none of them: this is an all-or-none molecular switch. Thus, the conformational transition is concerted and conserves the symmetry of the oligomer (Fig. 1B). More recently, the graded propagation of allosteric states has been suggested for large multiprotein complexes (8).
Activation of αβ T cells initiates an adaptive immune response and is mediated by stimulation of the TCR by its ligand peptide-MHC (pMHC). The TCR comprises the variable TCRαβ dimer, which binds to pMHC, and the CD3εδ, CD3εγ, and CD3ζζ dimers, which contain intracellular signaling motifs (9–11). These motifs include tyrosines that are phosphorylated upon pMHC binding (12, 13) and a proline-rich sequence in the CD3ε tail (CD3ε PRS) that can associate with the adaptor protein Nck (14, 15). As a prerequisite for allosteric regulation, the TCR exists in (at least) two different conformations. High resolution crystal and nuclear magnetic resonance structures of certain isolated domains of TCRαβ and CD3 and a low resolution electron microscopy structure exist (16–18). Unfortunately, these have not helped generate a three–dimensional, high-resolution structure of the complete TCR complex. Using other methods, ample experimental evidence has been generated for two different TCR conformations, which might include quaternary changes within the spatial arrangement of the multiprotein complex. This evidence has been summarized (19, 20) and includes a biochemical effector-binding assay (14), accessibility of an epitope (21), acquisition of protease resistance (22), measurement of the distance between individual subunits within a TCR (23), and determination of lipid binding to the transmembrane regions of the TCR (24). In addition, it was suggested that the cytoplasmic tails of CD3ε and CD3ζ are aligned to the membrane in resting TCRs and released from the membrane upon TCR activation (17, 25–27). The molecular mechanism causing dissociation from the membrane is unknown and might be part of the allosteric conformational changes discussed in this article.
The two conformations of the TCR are: 1) the resting conformation, in which the CD3ε PRS cannot bind to Nck, the CD3ζ cytoplasmic tails are separated from each other, cholesterol binds to the transmembrane region of TCRβ, and the cytoplasmic tyrosines are not phosphorylated; and 2) the active conformation, which is stabilized after multivalent pMHC binding. In the active conformation, the CD3ε PRS is accessible for Nck binding, the CD3ζ tails are close to each other, cholesterol dissociates from TCRβ, and the cytoplasmic tyrosines are accessible to tyrosine kinases (14, 23, 24, 28, 29). The capacity of TCRs to switch to the active conformation is absolutely required for TCR phosphorylation and T cell stimulation. This was shown using artificial ligands (28) and TCR mutants that are trapped in the resting conformation (CD3εC80G and CD3εK76T) (25, 30, 31).
To stabilize the active conformation and induce TCR signaling, soluble anti-TCR/CD3 Abs and soluble pMHC need to bind multivalently to the TCRs (28, 32–34). Of note, in Fig. 2 we show pMHC as a dimer, but omitted to show the TCR as a dimer for simplicity. Whether binding of membrane-bound (in contrast to soluble) monomeric pMHC might be sufficient to stabilize the active TCR conformation is as yet unknown.
T cells can already be activated when only very few TCRs are stimulated by pMHC (35, 36). In this respect, some important features of TCR biology are still incompletely understood. First, how is it guaranteed that in the absence of a ligand most TCRs are in an inactive resting state? Second, which mechanisms contribute to the extremely high sensitivity of T cell activation? Allosteric regulation of the TCR could provide both answers.
Heterotropic allostery at the TCR
When we found the first evidence, to our knowledge, in favor of conformational changes in the TCR as an outside-in signaling mechanism, we wrote that “ligand engagement of TCR-CD3 induces a conformational change” (14). This postulate could have implied that pMHC binds to TCRs in the resting conformation, then leading the TCR to adopt the active conformation (Fig. 2A). This scenario would be in line with the KNF model. However, later data demonstrated that TCRs spontaneously switch to the active conformation even in the absence of pMHC (24, 37, 38). Thus, the MWC model seems to better reflect the conformational regulation of the TCR: the TCR can switch between two conformations by itself, and multivalent pMHC binds preferentially to the active conformation, thus stabilizing this state (Fig. 2B). As a result of ligand (pMHC) binding, the equilibrium between the resting and the active states is shifted toward the active one.
The spontaneous shift of the TCR between its conformations was particularly obvious when cholesterol extraction in the absence of ligand binding led to the accumulation of active TCRs and initiated spontaneous TCR signaling (24, 39, 40). Thus, cholesterol heterotropically and negatively regulates the functioning of the TCR as described by the MWC model: cholesterol only binds to and stabilizes TCRs in the resting conformation (Fig. 2C). Preventing cholesterol binding to the TCR results in a shift of the resting–active state equilibrium to the active state as demonstrated by either extracting cholesterol from the membrane, by oxidizing cholesterol to cholestenone, or by mutating the TCRβ transmembrane region so that cholesterol can no longer bind (24). In all these approaches, TCRs became signaling active and led to a low level of T cell stimulation. Hence, these data demonstrate that it is important to keep the resting–active equilibrium shifted toward the resting state to prevent ligand-independent signaling. Furthermore, preventing cholesterol dissociation from the TCR by covalent coupling resulted in the inability of these TCRs to become signaling active (24), enforcing the view that cholesterol allosterically regulates the TCR according to the MWC model. In conclusion, cholesterol is a natural negative allosteric regulator of the TCR that guarantees that in the absence of a ligand most TCRs are in the resting state.
Homotropic allostery at the TCR
In effector and memory T cells as well as in T cells lines, the TCR is present in so-called nanoclusters, in which several TCRs cluster together (41–45). The exact nature of the nanocluster arrangement is not known and some studies were unable to detect TCR nanoclusters (46). Whether the TCR nanoclusters are stable and highly organized symmetrical assemblies, or individual TCRs transiently come together in small regions of the plasma membrane remains unresolved (6, 47). However, the presence of TCR nanoclusters opens the possibility for a further layer of TCR regulation by homotropic allostery (Fig. 1B) as previously suggested using the CD3εC80G or CD3εK76T mutants (30).
According to the KNF model, within a nanocluster only the TCRs that are engaged multivalently by pMHC would switch to the active conformation (Fig. 3A, left panel). In this scenario, the presence of small amounts of the mutant CD3ε molecules (CD3εC80G or CD3εK76T) that cannot adopt the active conformation would not inhibit signaling by the wild type (wt) TCR (Fig. 3A, right panel). However, we found that expression of 5% mutant CD3ε molecules significantly inhibited signaling by wt TCRs. This strong dominant negative effect of the CD3ε mutant suggested the existence of allosteric effects among TCRs within nanoclusters that adopt either the resting or active conformations in a coordinated fashion (30). These allosteric effects are again best explained by the MWC model, because in this model the presence of small amounts of CD3εC80G or CD3εK76T would inhibit also the wt TCRs to adopt the active conformation (Fig. 3B), as seen experimentally (30).
Upon pMHC stimulation, several TCR nanoclusters come together into microclusters (48, 49). Single molecule resolution of these TCR microclusters has also suggested the existence of positive allostery within TCR nanoclusters. The tyrosine kinase ZAP-70 binds with its tandem SH2 domains to phosphorylated TCRs, but not to resting unphosphorylated TCRs (50). A first work using the tyrosine kinase ZAP-70 as a probe of signaling TCRs showed an uneven distribution of ZAP-70 within TCR microclusters, resulting in concentration of ZAP-70 in specific areas (51). These data could be interpreted in a way that in some nanoclusters all TCRs are in the active state and in others all TCRs are in the resting state, and these two different nanocluster states coexist within a microcluster. Likewise, the use of phospho-specific anti-CD3ζ Abs has shown the existence of high-density TCR nanoclusters with abundant phosphorylated CD3ζ and low-density TCR nanoclusters with a low number of phosphorylated CD3ζ (52). These last data suggest the existence of organized, densely packed TCR nanoclusters, which can allow mechanisms of allostery versus disorganized, perhaps randomly associated, TCR clusters.
Together, these experimental data suggest there is a strong positive cooperativity between individual TCRs within one nanocluster. These nanoclusters might contain up to 20 TCRs (6, 43–45) and binding of only two pMHC (a pMHC dimer) might stabilize all 20 TCRs in the signaling competent state. Hence, this concerted switch to the active conformation might help to explain the high sensitivity of T cell activation.
Quantitating TCR allostery
The litmus test for any model of allosteric regulation is whether it can quantitatively account for experimental data and make valid predictions for new experiments. This has previously been achieved for classical examples such as hemoglobin (1, 2, 53), phosphofructokinase (54, 55), and aspartate transcarbamoylase (56). Also, for the transcription factor NFAT, which is pivotal in T cell activation, experimental data on multisite phosphorylation have been explained by an allosteric conformational switch model, where the individual phosphorylation sites take the role of the ligand binding sites (57). For the heterotropic regulation of the TCR by cholesterol and pMHC, Swamy et al. (24) showed the MWC model best accounted for the experimental data (see 4Heterotropic allostery at the TCR and Fig. 2). The conformational equilibrium constant of the unligated TCR (neither cholesterol nor pMHC interactions) was shown to be skewed strongly in favor of the resting state [L = (resting TCR)/([spontaneously] active TCR) = 14] (Fig. 2C). Physiological cholesterol levels enhance the occupancy of the resting state by a further factor of four, showing the quantitative impact of this negative allosteric regulator on the activation threshold. A critical feature of the MWC model is the dependence of ligand affinity on the conformational state of the protein, predicting that the TCR in the resting state should bind pMHC with a lower affinity than TCRs in the active state. In line with this idea, mechanical pulling on the TCR, which might have caused the TCR to switch to the active state, enhanced the TCR’s affinity toward pMHC as measured by a prolonged dwell time of the pMHC-TCR interaction (58). Hence, the databased model of heterotropic TCR allostery will inform further analysis of possible homotropic allosteric effects in TCR nanoclusters.
In this Brief Review we have presented a new view on the functioning of the TCR by taking into account the different conformations it can adopt. We have discussed ample experimental evidence that suggests the TCR is allosterically regulated, and that the MWC concerted allostery model best explains these observations. This refers to heterotropic allostery in which cholesterol serves as a negative allosteric regulator of the TCR and by homotropic allostery within TCR nanoclusters in which positive cooperativity was observed. Thus, TCRs can sense their environment, such as the concentration of cholesterol and pMHC, and respond accordingly. This new view on the TCR suggests five novel considerations:
Cholesterol is a negative regulator of the spontaneous conformational switch of the TCR, reducing TCR signaling (see the section headed Heterotropic allostery at the TCR). At the same time, cholesterol contributes to TCR nanoclustering, because cholesterol extraction leads to disassembly of the nanoclusters. In this respect cholesterol is a positive regulator of TCR functioning (59, 60). Thus, it would be important to study under which conditions these opposing functions dominate and which other lipids are involved in TCR regulation.
Homotropic allostery implies that TCRs communicate within a nanocluster. Whether this is done by protein–protein interactions (such as CD3ζ–CD3ζ contacts) or mediated by lipids and whether there is symmetry in the nanoclusters, as often suggested for homotropic allostery (2), would be important areas of research. Certainly, the highest goal would be to obtain a three-dimensional structure of the TCR and TCR nanoclusters.
Likewise, structural determination of the resting and active conformations of the TCR would allow atomic insight into the conformational change and help explain how multivalent pMHC binding stabilizes the active state. Most likely the conserved diagonal orientation of the pMHC–TCRαβ interaction, in which the N terminus of the peptide interacts with TCRα and the C terminus with TCRβ, plays a role. This has been elaborated by the permissive geometry model (19).
Besides the TCR, the coreceptors CD8 and CD4 also bind to pMHC. Because the coreceptors might also interact with the TCR, it is possible that they directly impact on the allosteric switch of the TCR (61, 62). Furthermore, differential glycosylation of the coreceptors or the TCR in different developmental or differentiation stages (63) could alter the TCR’s resting–active equilibrium and thus influence the sensitivity of T cell activation.
One important consequence of homotropic allostery would be that multivalent pMHC binding to some TCRs of a T cell would enhance the binding of the following pMHC (Fig. 3C). This would enhance T cell activation and is an important point to experimentally address. Homotropic allostery could also have important implications for ligand discrimination by the TCR. T cells are able to accurately measure the affinity/avidity and/or the dwell time of a pMHC–TCR interaction and respond accordingly (64–69). However, considering TCR nanoclusters, multivalent binding of the first pMHC would enhance the avidity of following pMHC. Thus, allosteric regulation of the TCR might not only impact on the sensitivity of a T cell response, but also on the quality.
We expect that this new view on the TCR will stimulate novel experiments and unravel a number of new mechanisms that regulate TCR functioning.
This work was supported by funds from the Deutsche Forschungsgemeinschaft (Cluster of Excellence Project EXC294 through BIOSS Center for Biological Signaling Studies [to W.W.A.S.], Cellular Networks Cluster of Excellence Project EXC81 [to T.H.], Spemann Graduate School Programme Grant GSC-4 [to W.W.A.S.], Grant SCHA 976/2-1 [to W.W.A.S.], and Grant MI 1942/1 [to S.M.]), as well as European Union–European Research Council 2013-Advanced Grant 334763, Novel Properties of Antigen Receptors and Instruments to Modulate Lymphoid Function in Physiological and Pathological Conditions Project (to B.A.).
Abbreviations used in this article:
The authors have no financial conflicts of interest.