Immune cells called T cells have T-cell receptors (TCRs) on their cell membrane that recognize dysfunctional cells expressing abnormal protein fragments. Such abnormalities can arise in cells if, for example, cancer develops or infection occurs. When TCRs recognize these unusual peptides, the receptors become activated and stimulate T cells to destroy or inhibit the abnormal cells. Such T-cell responses are being harnessed for anti-cancer clinical therapies. TCRs are also of interest because their dysfunction can lead to autoimmunity or immunodeficiency diseases.
Writing in Nature, Dong et al.1 present the structure of a human TCR, at a resolution of 3.7 ångströms, obtained using an imaging technique called single-particle cryogenic electron microscopy (cryoEM). Such a high-resolution structure of the entire TCR was previously lacking, and it provides a wealth of detail about this receptor.
For more than 35 years2, it has been known that each TCR of the type called an αβTCR is a protein complex. Eight proteins form the TCR: six of these are collectively known as CD3, which acts in a signalling capacity when a TCR is activated. CD3 comprises a heterodimer of CD3ε and CD3δ (CD3εδ), a heterodimer of CD3ε and CD3γ (CD3εγ) and a homodimer of CD3ζ (CD3ζζ). The other two proteins that form the TCR are TCRα and TCRβ. They create the ligand-binding heterodimer (TCRαβ) that recognizes a peptide bound to a major histocompatibility complex molecule (MHC) on the surface of another cell — a combination called a pMHC. TCRα, TCRβ, CD3δ, CD3ε and CD3γ contain extracellular regions (called ectodomains), connecting-peptide regions, transmembrane regions and cytoplasmic tails.
Each person’s many T cells have the same set of CD3 proteins, but TCRα and TCRβ have what are called variable domains that differ between T cells such that different T cells have a unique capacity to recognize specific peptides — called antigens — in pMHCs. Much is already known3–5 about the structure and function of individual TCR components, but an understanding of the complete TCR structure promises to provide fresh insights.
Dong and colleagues co-expressed all the TCR proteins in cultured cells, where they assembled into the TCR complex. The authors isolated the complex and stabilized it for cryoEM by chemical cross-linking, to form permanent bonds between adjacent proteins. This approach enabled the authors to obtain structural data. Given the limited interaction surface between the various CD3 dimers and the TCRαβ ectodomains shown by Dong and colleagues, it is possible that alternative TCR conformations exist. The cross-linking used to obtain cryoEM data imposes constraints on structural variability.
The paired variable domains formed by the TCRαβ heterodimer are located at the centre of the TCR, in the extracellular region farthest from the cell membrane (Fig. 1). The variable domains’ surface for binding a pMHC is oriented in a way that is consistent with the emerging idea that the TCR acts in a direction-selective manner as a mechanosensor during antigen recognition6. It binds its ligand in one direction only, owing to a physical property termed anisotropy.
Another striking structural feature of the TCR is its amalgam of the ectodomains, which consist of what are termed the constant domains of TCRα and TCRβ, as well as the ectodomains of CD3εγ and CD3εδ, plus the short extracellular segments of CD3ζζ. Connecting peptides link ectodomains and the bundle of eight transmembrane helices, one helix for each protein of the TCR. In Dong et al.’s structure, the transmembrane helices are mainly in a parallel orientation among the digitonin detergent that is used to replace cell-membrane lipid for protein solubilization and imaging. The authors could not obtain structural information for the cytoplasmic tails of TCR proteins, which include CD3 regions that are vital for signalling during T-cell activation. Presumably, this was because such regions have high conformational flexibility in the absence of normal cell-membrane lipids.
Dong and colleagues’ structure is compelling for many reasons. The authors docked structures of TCR components obtained previously using X-ray crystallography onto their structure. The locations of the ectodomains of CD3εδ and CD3εγ relative to those of TCRα and TCRβ are consistent with previous data5. Moreover, in the transmembrane region, the close juxtaposition of basic amino-acid residues in the TCRαβ heterodimer and the acidic amino acids in the CD3 proteins support an earlier proposal that such interactions might help to orient the transmembrane helices7.
An evolutionarily conserved protein motif called CxxC, found in connecting-peptide regions of CD3εγ and CD3εδ, has a crucial role in CD3 signalling8. Dong and colleagues report molecular connections called disulfide bridges in each of these four proteins’ CxxC linkages, and also describe the protein architecture that surrounds these key linkages.
Dynamic movement of TCRα and TCRβ protein segments has previously been visualized using a method called NMR9. The model of the TCR as a directional mechanosensor proposes that forces exerted by T-cell motility and the motion of components of a T cell’s internal framework (the cytoskeleton) improve a T-cell’s recognition of pMHC more than 1,000-fold relative to recognition in the absence of force6. Consequently, when a TCR interacts with a pMHC that it recognizes, a physical load is placed on the bonding between the TCR and the pMHC, paradoxically resulting in an increase in bond lifetime6,10,11, and causing dynamic structural changes in the TCRαβ heterodimer that enhance the specificity and sensitivity of antigen recognition. The recognition of just one or two abnormal pMHCs among the roughly 100,000 normal pMHCs on a single cell suffices to trigger T-cell activation6. By contrast, the interactions between a typical receptor and ligand do not rely on biomechanical forces as a crucial recognition feature. Dong and colleagues’ structure provides insights into how the TCR is configured as a mechanosensor to meet its specificity and sensitivity requirements.
The orientation of the variable domains and the positioning of the antigen-binding site favours a directional interaction of the TCR with a pMHC that is consistent with T-cell scanning motions that occur tangentially to the surface of the surveyed cell. The FG loop structure5 of TCRβ’s constant region controls the lifetime of a TCR–pMHC bond and also the extension of the TCRαβ heterodimer when it transitions from a compact to an extended form under force during antigen recognition6,9. The FG loop is located above CD3ε of CD3εγ. If the TCR interacted with a pMHC on the right in Fig. 1, the resulting force would probably transfer, through a geometry like a lever and fulcrum, from the TCRαβ heterodimer’s variable domains downwards towards its constant regions and then to CD3 dimers on the same vertical plane below. The force would probably be amplified through reversible and repetitive transitioning of the TCRαβ heterodimer between its expanded and contracted forms, which rearranges the TCR components.
The CxxC motifs are located near the cell membrane, and the rigid connectivity of these regions on CD3 ectodomains and the arrowhead-shaped configuration of their respective heterodimer’s transmembrane segments suggests that pushing and pulling motions from TCRαβ to CD3εγ or CD3εδ, or to both, might transfer force through the membrane or deliver this energy as work. Such forces might cause changes to parts of the TCR structure, and affect neighbouring lipids to expose key amino-acid sites on CD3 cytoplasmic tails. The addition of phosphate groups to these tyrosine amino-acid residues would trigger signalling cascades and other cellular changes needed for T-cell activation6,9,12. The abundance of interactions between the various connecting-peptide segments of the TCR would probably aid signalling1,8,13.
When the authors superimposed an available structure of a pMHC bound to a TCRαβ heterodimer onto their TCR structure, the TCRαβ heterodimers were similar in both structures. This is unsurprising, because force application is probably the major cause of structural changes driving TCR subunit rearrangements, and these structures were obtained in the absence of force, and thus capture a compact state of the TCRαβ heterodimer. The force-based TCR–pMHC recognition process differs from typical receptor–ligand interactions such as antibody–antigen interactions, which are force-independent. Harnessing energy for mechanosensing from cellular motions could explain how, unlike in force-independent interactions, TCRs can discriminate so sensitively between very similar antigens, differing by just one amino acid.
It has been suggested that the subunit rearrangements that occur when force is applied to the TCR might foster CD3 dimer dissociation, starting with CD3ζζ, and that this contributes to T-cell activation8. The authors’ structure confirms that CD3ζζ dissociation would indeed cause changes to the TCR structure in the transmembrane region.
Dong and colleagues’ work provides a basis for future studies. Could structures of other αβ-type TCRs of defined antigen specificities, with or without the relevant pMHC, be obtained? Might it be possible to obtain high-resolution structures of the transmembrane segments of a TCR in a natural lipid-membrane environment to visualize the cytoplasmic tails of TCR proteins? Could conformations of the TCR complex under the application of force be imaged if new structural-analysis methods are developed?
Given the importance of the TCR for understanding immune-cell function and the use of T cells in immunotherapy to tackle cancer, information about TCR structure might bring improvements in TCR design for medical purposes. Dong and colleagues’ work is an urgent summons to immunologists interested in tumour biology and to others to consider bioforces when assessing T cells in vitro to gauge the potential of their TCRs in vivo. Great opportunities lie ahead to make more progress in developing high-quality TCRs for clinical use.
Nature 573, 502-504 (2019)