Allosteric activation of T cell antigen receptor signaling by quaternary structure relaxation

Summary The mechanism of T cell antigen receptor (TCR-CD3) signaling remains elusive. Here, we identify mutations in the transmembrane region of TCRβ or CD3ζ that augment peptide T cell antigen receptor (pMHC)-induced signaling not explicable by enhanced ligand binding, lateral diffusion, clustering, or co-receptor function. Using a biochemical assay and molecular dynamics simulation, we demonstrate that the gain-of-function mutations loosen the interaction between TCRαβ and CD3ζ. Similar to the activating mutations, pMHC binding reduces TCRαβ cohesion with CD3ζ. This event occurs prior to CD3ζ phosphorylation and at 0°C. Moreover, we demonstrate that soluble monovalent pMHC alone induces signaling and reduces TCRαβ cohesion with CD3ζ in membrane-bound or solubilised TCR-CD3. Our data provide compelling evidence that pMHC binding suffices to activate allosteric changes propagating from TCRαβ to the CD3 subunits, reconfiguring interchain transmembrane region interactions. These dynamic modifications could change the arrangement of TCR-CD3 boundary lipids to license CD3ζ phosphorylation and initiate signal propagation.

. βY291 contribution to TCR-CD3 quaternary structure cohesion, related to Figure 2 A Sensitivity of TCR-CD3 quaternary structure cohesion to DDM concentrations. CD8 -J76 1G4-WT were solubilised with increasing concentrations of DDM and analysed by β-HA PD and IB for β, ε and ζ (left). Mean ± SD of ε/βT (middle) and ζ/β2 (right), n = 3, unpaired t-test (ε/βT): p < 0.0001; (ζ/β2): 0.5 vs. 1 % p < 0.01, 0.5 vs. 2 % p < 0.0001, 0.5 vs. 3 % p < 0.001. Note that at ≤ 0.5 % DDM there is no detectable change in cohesion of CD3 subunits with TCRαβ, further supporting that in TCR-CD3 extracted at 0.5 % DDM the subunits' stoichiometry in the octamer remains intact (Swamy et al., 2008). At ≥ 1 % DDM there is progressive loss of ζ followed by γε and δε. B Graphic scheme describing the DSA. The entire pool of the TCRαβ is captured by anti-HA (β-HA) PD which includes partial αβγεδε complex assembled in the endoplasmic reticulum (ER) (Alcover et al., 2018), αβγεδεζζ complex resident in the trans-Golgi (Alcover et al., 2018) or recycling at steady state in the ER (Alcover et al., 2018) and the largest fraction of αβγεδεζζ present at the plasma membrane (PM). The IB scheme on the right shows expected band patterns for β, ε, δ, γ and ζ derived from the ER and PM. Arrows indicate the different isoforms of TCRβ (β1, β2, β3). To evaluate ζ recovery, ζ/β2 ratio was calculated and the value for ζ/β2 ratio from WT was set equal to one. To evaluate ε, δ, γ recovery, ε/βT, δ/βT and γ/βT ratios were calculated and the ratios for WT were set equal to one. These represented the recovery of intact TCR-CD3 complex. Ratios < 1 indicate a lower recovery of ζ, ε, δ and γ hence a reduced cohesion of TCR-CD3 quaternary structure. See STAR Methods for a detailed description of the experimental procedure. C J76 wtc51 treated or not with endo H and analysed by β-HA PD and IB for β. Arrows indicate β isoforms, βNG indicates non-glycosylated β isoform after removal of high-mannose carbohydrates by endo-H treatment. Data representative of 2 experiments. D J76-1G4-ζKO expressing inducible ζTST were solubilised and analysed by β-HA (lanes 1, 3) or ζTST (lanes 2, 4) PD and IB for β and ζ. Arrows indicate β isoforms. IB representative of 2 experiments. E CD8 -J76 1G4-WT or 1G4-βA291 ± A770041 were solubilised and analysed by β-HA PD and IB for β, ε and ζ. Mean ± SD of ε/βT (left) and ζ/β2 (right), n = 3, unpaired t-test (ε/βT): 1G4-WT vs. 1G4-βA291 p < 0.05, 1G4-WT+A770041 vs. 1G4-βA291+A770041 p < 0.01; (ζ/β2): 1G4-WT vs. 1G4-βA291 p < 0.0001, 1G4-WT+A770041 vs. 1G4-βA291+A770041 p < 0.0001; ns = non-significant. F Representative FACS histograms of UCHT1-Fab' staining of J76-1G4WT-ζKO reconstituted with doxycycline inducible ζWT. Cells were induced or not for ζWT expression, labelled or not with CellTrace violet, mixed 1:1 and analysed for TCR-CD3 surface expression by FACS. G Normalised number of contacts of the WT β TMR with the rest of the TCR-CD3 TMRs in our all-atom molecular dynamics simulations (MDS) (related to Fig.   2D). The contacts in the cryo-EM structure (PDB: 6JXR) are shown in green for comparison with the WT allatom MDS (black). Magenta arrow indicates β291. Normalisation was done by dividing the number of contacts of each residue by the highest number of contacts. For all contacts analyses in Figs. S2G -S2J, a cut-off distance of 4 Å was used to define a contact and the red dotted line represents 70 % of the normalised contacts, a threshold used to measure the significance of contacts. H Normalised number of contacts of βY291 with the rest of the TCR-CD3 TMRs (related to Fig. 2D). Comparison of protein-protein interactions between our WT allatom MDS (black) and the cryo-EM structure (PDB: 6JXR) (green). Normalisation was done by dividing the number of contacts of each residue by the highest number of contacts. I Normalised number of contacts of βWT  Right, mean ± SD of ζ/β2, n = 3, unpaired t-test (ns). E CD8 -J76 1G4-WT or 1G4-βA303 were solubilised and analysed by β-HA PD and IB for β, ε and ζ. Left, IB: 1 of 3 experiments. The arrow indicates β2 isoform.
Middle, mean ± SD of ε/βT, n = 3, t-test (ns). Right, mean ± SD of ζ/β2, n = 3, unpaired t-test (ns).  were observed in the simulations carrying ζA38 substitution. This is likely to be the consequence of the increased ζζ loosening that allows ζ2 to wobble and to come in contact with β TMR. I Normalised number of contacts of ζ1 with CD3ε (δε) TMR (left) and of CD3ε (δε) with ζ1 TMR (right) in the WT (green) and ζA38 mutant (red) allatom MDS. Magenta arrow indicates ζ138 showing that this residue reduced its interaction with CD3ε (δε) TMR when mutated to alanine (ζA38, in red), with a net effect of increasing flexibility of both ζ1 and ζ2 subunits (ζ1>ζ2) relative to TCRαβ (see also Fig. 4F). J Normalised number of contacts of α TMR (left) and of CD3ε (δε) TMR (right) with ζ138 in the WT (green) and ζA38 mutant (red) all-atom MDS. A small number of contacts between ζ138 and α TMR were observed in both WT and ζA38 simulations while all three residues of CD3ε (δε) TMR (εV134, εI135 and εI138) that interacted in the WT simulations reduced their contacts with ζ1A38 (red).
Normalisation is performed such that the number of contacts of TCRα TMR is compared to that of CD3ε (δε) residues. K Normalised number of contacts of ζ1 and ζ2 with α TMR (left) and of α TMR with ζζ TMR (right) in the WT (green) and ζA41 mutant (red) all-atom MDS. Magenta arrow indicates ζ41 showing no significant contacts with α TMR in both WT and ζA41 simulations. However, one contact of ζ2 (ζ2Y33) with α TMR and two contacts of α TMR (αL247, αV249) with ζζ TMRs were reduced in the simulations carrying ζA41 substitution. However, one contact of ζ2 (ζ2R52) with α TMR increased on ζA41 substitution. L Normalised number of contacts of ζ1 and ζ2 with β TMR (left) and of β TMR with ζζ TMR (right) in the WT (green) and ζA41 mutant (red) all-atom MDS. Magenta arrow indicates ζ41 showing no significant contacts of ζ141 and ζ241 with β TMR in both WT and ζA41 simulations. However, two contacts of ζ2 with β TMR (ζ2L27 and ζ2F40) and one of β TMR with ζζ TMRs (βL284) were increased in the simulations carrying ζA41 substitution. However, one residue of the β TMR (βS269) reduced contact with ζζ TMR during the ζA41 simulations. M Normalised number of contacts of ζ2 with CD3γ TMR (left) and of CD3γ with ζ2 TMR (right) in the WT (green) and ζA41 mutant (red) all-atom MDS. Magenta arrow indicates ζ41 showing that the mutated residue ζ2A41 still maintained contact with CD3γ during the simulations compared to the WT. However, γV124, which strongly interacted with ζ2I41 in the WT, reduced its interaction with ζ2A41. One contact of ζ2 (ζ2R52) with CD3γ TMR increased in the simulations carrying the ζA41 substitution. N Normalised number of contacts of CD3γ TMR (left) and of β TMR (right) with ζ241 in the WT (green) and ζA41 mutant (red) all-atom MDS. Two contacts of CD3γ (γV124 and γF127) with ζ241 were reduced and one contact (γV128) was increased during ζA41 simulation. No significant contacts of β TMR with ζ241 were observed in both WT and ζA41 simulations. A J76-1G4WT-ζKO expressing 1G4 ζWT or ζA38 or ζA41 were lysed with 0.5 % DDM and analysed by IB for β, ε, ζ and actin. Left, IB of input lysates of the experiment shown in Fig. 5B (1 of 4 experiments). Right, mean ± SD of βT/actin, ε/actin and ζ/actin normalized to WT, n = 4, unpaired t-test (ns). B (6V-A2)4 binding to J76-1G4WT-ζKO expressing ζWT or ζA38 related to Fig. 5C Plot shows mean ± SD of 3 experiments measured in triplicates. Figure S6. pMHC tetramer binding loosens αβ association with ζ, related to Figure 6 A Graphical scheme describing the experimental procedure used to compare the cohesion of unliganded receptor (top) and (pMHC)4 ligated TCR-CD3 (liganded receptor, bottom). The unliganded receptor was captured by anti-HA (β-HA) PD described in Fig. S2B. To pull down the liganded receptor, cells were stimulated with tetramerised His-tagged streptavidin (His SA) and ligand excess was removed. Cells were solubilised with 0.5 % DDM and post-nuclear lysate was incubated with HisPur Cobalt beads to pull down pMHC4-engaged TCR-CD3. The IB schemes on the right show expected bands pattern for β and ζ. Arrows indicate the different isoforms of TCRβ (β1, β2, β3). To evaluate ζ recovery, ζ/β2 ratio was calculated and the value for ζ/β2 ratio from non-stimulated samples was set equal to one. This value represented the recovery of intact TCR-CD3 complex and was compared to ζ/β2 ratio from (pMHC)4 stimulated samples. A ratio < 1 indicates a lower recovery of ζ revealing a reduced cohesion of TCR-CD3 quaternary structure. See STAR Methods for a detailed description of the experimental procedure. B Structure and modelling of 1G4-WT and affinity enhanced 1G4 mutants (QM-α and wtc51) used in this study. Top, table shows binding affinities of the 1G4 and affinity enhanced TCR mutants (QM-α and wtc51) with sequence alignment highlighting the CDR loops with mutated residues (bold and underlined). a, structural overview of the 1G4 TCR (green)-A2 (grey)-SLL (red sticks) tri-molecular complex structure (PDB: 2BNR). Positions of the TCR residues (green sticks) in relation to A2-SLL that are mutated in the affinity enhanced TCRs are shown below. b, structural overview of the QM-α TCR (blue)-A2-SLL tri-molecular complex structure (mutations modelled using PDB: 2BNR).
Positions of the TCR residues (blue sticks) in relation to A2-SLL that are mutated in the affinity enhanced TCR are shown below. c, structural overview of the wtc51 TCR (orange)-A2-SLL tri-molecular complex structure (mutations modelled using PDB: 2BNR). Positions of the TCR residues (orange sticks) in relation to A2-SLL that are mutated in the affinity enhanced TCRs are shown below. d, overlay of the mutated residues in the CDR loops comparing 1G4 (green sticks), QM-α (blue sticks) and wtc51 (orange sticks) TCRs. e, overlay of the positions of the CDR loops comparing 1G4 (green ribbon), QM-α (blue ribbon) and wtc51 (orange ribbon). f, analysis of the TCR crossing angles for each TCR. C J76 QM-α treated with A770041 and stimulated or not with (9V-A2)4-His. Left, β-HA (lanes 1, 3) or His (lanes 2, 4) PD and IB for β and ζ (1 of 3 experiments). The arrow indicates β2 isoform. Middle, mean ± SD of ζ/β2, n = 3, unpaired t-test p < 0.001. Right, pErk IB: 1 of 3 experiments. D J76 1G4 ± A770041 were incubated with or w/o UCHT1-Fab' or UCHT1-Ab. pErk IB of the experiment shown in Fig. 6G (1 of 3 experiments). E J76 1G4 were cooled on ice for 20 min and incubated for 5 min on ice with UCHT1-Fab' or UCHT1-Ab. pErk IB of the experiment shown in Fig. 6H (1 of 3 experiments).

Figure S7. Monovalent pMHC in solution triggers TCR-CD3 untying and intracellular signalling, related to Figure 7
A Graphical scheme describing the experimental procedure used to compare the cohesion of unliganded receptor (top) and soluble, monovalent, mono-dispersed (sm)-pMHC ligated TCR-CD3 (liganded receptor, bottom). The unliganded receptor was captured by anti-HA (β-HA) PD described in Fig. S2B. To pull down the liganded receptor cells were stimulated with biotinylated sm-pMHC and ligand excess was removed. Cells were then solubilised with 0.5 % DDM and post-nuclear lysate was incubated with His-tagged streptavidin (His SA) followed by pull down with HisPur Cobalt beads or post-nuclear lysate was incubated with monomeric Avidin beads (mAv). The IB schemes on the right show expected bands pattern for β and ζ. Arrows on the left indicate the different isoforms of TCRβ (β1, β2, β3). To evaluate ζ recovery, ζ/β2 ratio was calculated and the value for ζ/β2 ratio from the non-stimulated sample was set equal to one. This value represented the recovery of intact TCR-CD3 complex and was compared to ζ/β2 ratio from sm-pMHC stimulated samples. Dox-induced (+ dox) CD8 -J76 wtc51 stimulated or not with 200 nM sm-9V-A2 for 5 minutes at 37 ⁰C served as control. Left, pErk IB (1 of 2 experiments). Right, mean ± SD of pErk of mixed dox-induced and not doxinduced ± sm-9V-A2 cells (mix ± dox), n = 2 experiments in duplicates, unpaired t-test: ns (mix ± dox), p < 0.05 (+ dox). K Graphical scheme describing the "TCR-CD3 allosteric relaxation" mechanism uncovered in this study. Briefly we suggest that in absence of force, co-receptor, clustering or PTPs exclusion, monovalent pMHC binding to TCR-CD3 allosterically regulates a cascade of conformational changes that relaxes the quaternary structure of TCR-CD3 TMRs. In the proposed model, conformational changes occurring at the pMHC binding site propagate to CαCβ ECDs at the site where they contact the CD3 subunits. These rearrangements are transmitted to the CD3 TMRs, resulting in a reduced cohesion of TCR-CD3 TMRs, ITAMs exposure and their phosphorylation by active Lck. Moreover, we envisage the possibility that pMHC-induced reconfiguration of the octamer's TMRs leads to a local redistribution of negatively-charged lipids which could reduce hydrophobic and electrostatic forces that help holding CD3 tails within the plasma membrane.  Inner leaflet  10  20  40  -20  8  2  Outer leaflet  50  -10  20  20  --