We obtained structural insights into the kinase-dead D461N mutant of ZAP-70 (YY_DN) by HDX–MS. In brief, deuterium in solution (D₂O) replaces amide hydrogens in ZAP-70′s backbone. Mass spectrometry of subsequent pepsin digests allows quantification and assignment of HDX to specific amino acid (aa) sequences. HDX rates are influenced by solvent exposure and hydrogen bonds. Changes in HDX patterns locate conformational changes, intra- and intermolecular interaction sites induced by complex formation and/or post-translational modifications (Fig. 1c–f). Thus, HDX-MS provides structural insights on different activation states of soluble ZAP-70 that have been unattainable to date. However, HDX–MS does not generate an actual structure and has lower resolution than crystallography. Conformational changes upon pITAM-binding were identified by comparison of HDX between YY_DN alone (used as reference standard for all HDX–MS measurements in this study) and in complex with the phosphorylated cytosolic domain of the TCR γ-chain, pCD3γ_CD (YY_DN^cplx). Differences in HDX were projected onto the known auto-inhibited structure of YY_DN (Fig. 2a)¹⁷ and compared by alignment (Supplementary Fig. 1). Reduced HDX reveals regions of interactions between the pCD3γ_CD and theZAP-70 SH2 domains (aa E174-F187, I203-L212 and L239-L250). Increased HDX in I-A (aa D116-L152), I-B (aa K251-L287) and the KinD (aa Y598-L600) indicates separation of tSH2 and KinD upon pCD3γ_CD binding (illustrated in Supplementary Fig. 2a). The binding pocket for Y319 in the N-terminal (N)-lobe of the KinD shows no changes, suggesting that this interaction between I-B and the KinD is maintained upon pITAM-binding. Additional hydrogen bond rearrangements were observed in the KinD through reduced HDX in its N-lobe (aa G350-V356; I368-M383) and C-terminal (C)-lobe (aa L469; Y535-L568). These data show that pITAM binding alone is sufficient to induce the open conformation of ZAP-70 and rearrangement of the KinD.

Based on the auto-inhibited structure, phosphorylation of Y315/Y319 has been proposed to induce an open conformation and consequently catalytic activity of ZAP-70^{16, 17}. All recombinant ZAP-70s were analyzed for phosphorylation of Y292, Y319, Y492, Y493 and total tyrosine-phosphorylation by immunoblotting to confirm phosphorylation of Lck-treated proteins (Supplementary Fig 3). Furthermore, no non-phosphorylated peptides containing Y292, Y315, Y319, Y492 and Y493 were detected by MS analyses after ZAP-70 phosphorylation. Phosphorylated ZAP-70 (pYY_DN) showed increased HDX in I-A (aa D116-L152) and I-B (aa Y315-L332) (Fig. 2a and Supplementary Fig. 1). These data indicate that phosphorylation of Y315/Y319 separates tSH2 and the KinD (illustrated in Supplementary Fig. 2b). Additional HDX increases in I-B (aa Y315-L332) and the N-lobe of the KinD (aa R359-D365) indicate dissociation of Y319 from its binding pocket. The phenylalanine mutant Y315/Y319F (FF_DN) was analyzed to exclude the possibility that phosphorylation outside of I-B induces the open conformation (Fig. 2b and Supplementary Fig. 1). HDX for non-phosphorylated FF_DN is reduced in I-A (aa D116-L118; Y126-G135; E139-L152), I-B (N288-T312) and the KinD (aa M384-Q388; T494-C510; E531-L568; Y598-L600). HDX–MS also detects shifts in conformational equilibria²⁸ and these reductions indicate ZAP-70 oscillates between closed/auto-inhibited and open conformations. FF_DN is more prone to the former, due to the destabilizing effects of the hydroxyl groups on the binding of Y315 and Y319 to I-A and the KinD, respectively. Phosphorylated FF_DN (pFF_DN) showed only minor HDX differences at the beginning of I-B (K251-A267). This confirmed that Y315/Y319 phosphorylation is essential for the release of Y319 from its binding pocket and ZAP-70 assuming an open conformation without pITAM binding. Next we tested if FF_DN in complex with pCD3γ_CD (FF_DN^cplx) is still able to assume an open conformation. HDX was increased in I-A (E136-A137; I142-L152) and I-B (K251-A267; N288-T312), suggesting that FF_DN^cplx changes into an open conformation upon receptor binding similar to YY_DN^cplx. These data show that opening of ZAP-70 through pITAM-binding and phosphorylation are independent processes with similar outcomes. To test for combinatorial effects, we analyzed phosphorylated YY_DN and phosphorylated FF_DN in complex with pCD3γ_CD (pYY_DN^cplx and pFF_DN^cplx) (Supplementary Figs. 1 and 4). HDX–MS shows that the effects of pITAM-binding and phosphorylation are additive.

Previous reports show that the alanine mutant Y315/Y319A (AA) has reduced activity in T cells, but increased activity in TCR-independent activation systems^{17, 18, 23}. HDX in I-A and I-B of AA_DN are the highest observed (Fig. 2c and Supplementary Fig. 1) and HDX in I-B and KinD include regions previously not detected (aa K333-L338; I342-M359; I368-E382; R385-Q388; L412-M414; S483-Y492; A601-G606). AA_DN binding to pCD3γ_CD and/or phosphorylation of AA_DN by Lck slightly revert the effects of the alanine mutations (Fig. 2c and Supplementary Fig. 1 and 4). However, for all forms of AA_DN, the Y319 binding pocket is unoccupied and no interaction between I-B and KinD can be formed. These data suggest that the phenyl groups of Y315/Y319 are required to assume an auto-inhibited and structurally stable conformation. Consequently, AA_DN is constitutively in an open conformation with the tSH2 and KinD separated.

We tested if the tSH2–KinD interaction is only loosened¹⁷ or completely disrupted by pITAM-binding or ZAP-70 phosphorylation. A dissociated tSH2 would have the same solvent accessibility and therefore, the same HDX patterns as a recombinant tSH2 module (aa 1–256). Therefore, we compared HDX for pCD3γ_CD-bound and/or phosphorylated tSH2 with the corresponding forms of YY_DN, FF_DN and AA_DN (Fig. 2d and Supplementary Fig. 1 and 4). Comparable HDX increases and patterns are consistent with complete dissociation of tSH2 and the KinD. In addition, an expected effect of the open conformation is an increased volume. Such an effect is demonstrated by reduced retention times in size exclusion chromatography, which indicate increased diameters for the ZAP-70 forms with increased HDX in I-A and I-B (Fig. 3a–c).

While all pITAMs of the TCR bind ZAP-70 with similar binding affinities^{29, 30, 31, 32}, it is possible that the conformational changes seen here in response to pCD3γ_CD binding might be specific to the CD3γ subunit. Therefore, HDX–MS was performed on YY_DN bound to six phosphorylated peptides comprising the sequences of all TCR ITAMs (i.e. CD3γ, δ, ε, ζ1, ζ2 and ζ3). Increased HDX in I-A was observed in all complexes, demonstrating that TCR pITAMs cause ZAP-70 to open via a general mechanism (Fig. 3d). These data are supported by the reduced retention times of all ZAP-70–pITAM complexes in analytical size exclusion chromatography (Fig. 3e). Taken together, the HDX–MS data shows opening of ZAP-70 after pITAM-binding, phosphorylation or mutation of Y315/Y319 to alanine. We established that the phenylalanine mutant is preferentially in the closed conformation, while the alanine mutant is always open. Analyses of these mutants enable us to determine how distinct ZAP-70 conformations control T cell activation.

Multiple reports have shown effects of Y315/Y319 mutations on TCR signal transduction^{17, 18, 23}. Wildtype (YY) and alanine (AA) mutants transduce TCR signals and trigger T cell activation; AA has increased background activity, but reduced activity upon TCR ligation. In TCR-independent activation systems AA shows higher activity than wildtype ZAP-70 (YY). The phenylalanine mutant (FF) cannot transduce signal or activate T cells. We activated ZAP-70 negative Jurkat T cells (P116) stably expressing ZAP-70–GFP fusion proteins (^GFPYY, ^GFPFF and ^GFPAA) with increasing concentrations of stimulatory antibodies (Fig. 4a). At high antibody concentration ^GFPYY showed phosphorylation on Y493 (pY493) in the activation loop. In contrast, ^GFPAA and ^GFPFF showed little or no increase in pY493. Only ^GFPYY- and ^GFPAA-expressing P116 cells showed LAT phosphorylation on Y191 (pY191). ^GFPAA requires >4-fold more antibody than ^GFPYY to achieve similar LAT phosphorylation. These data confirm the published studies and further suggest that the reduced signal transduction of the Y315/Y319 mutants is due to lower numbers of activated kinase molecules and not to reduced catalytic activity.

Catalytic activity of ZAP-70 is reported to require the open conformation^{16, 17}. We quantified both the potential of ZAP-70 to be activated by Lck and its catalytic activity towards its natural substrate, LAT, in a two-step in vitro kinase assay. Increasing amounts of Lck phosphorylated Y493 similarly in recombinant wildtype and mutant ZAP-70s (Fig. 4b). None of the kinases had catalytic activity without initial phosphorylation (Fig. 4c). In contrast, prior studies reported catalytic activity for purified or over-expressed AA in the absence of Lck^{17, 18, 33}. We found that this activity is due to background phosphorylation/activity that becomes amplified upon addition of ATP. Evidently, the open conformation of ZAP-70 is more susceptible to non-specific phosphorylation during production/purification. To remedy this problem, YY and AA were expressed fused to the phosphatase domain of SHP-1 (which was removed during purification) to eliminate any background phosphorylation (Supplementary Fig. 5); this process yielded completely inactive ZAP-70s. Therefore, the open conformation (AA) is not sufficient to induce kinase activity or auto-phosphorylation. All forms of ZAP-70 had similar capacities for activation by Lck. While YY activity increased slightly faster, all forms of ZAP-70 reached comparable activities. Increasing ZAP-70 concentrations (70nM-1.4μM) had no effect on basal or relative activities, and only accelerated the kinetics of LAT phosphorylation. According to these data, all fully phosphorylated forms of ZAP-70 have equivalent molar activities. Our HDX–MS data shows that pFF_DN, despite being phosphorylated on Y493, is in the closed conformation. This suggests that both the closed/auto-inhibited and open conformations of ZAP-70 have similar potential to become activated as well as comparable catalytic activity towards their substrates. Based on these data, the cellular signaling defects of FF and AA cannot be explained through differences in their catalytic activity.

Affects of ZAP-70 conformations on pITAM-binding capacities were tested in ELISA based competition assays (Fig. 5a). We found that FF requires ~3-fold higher concentrations than YY to compete similarly for pCD3γ_CD binding (Fig. 5b and Supplementary Table 1). In contrast, AA and tSH2 can easily compete with >3 fold excess YY. These data suggest that the closed form of ZAP-70 (FF_DN) binds pITAMs less efficiently than the wildtype, while the open form (AA_DN) and tSH2 have increased pITAM-binding capacities.

Bio-layer interferometry (BLI) was used to determine dissociation constants (K_D) as well as on- and off-rates (k_on and k_off) for ZAP-70 binding to pCD3γ_CD (Fig. 6 and Supplementary Table 2)³⁴. The binding curves for YY_DN and FF_DN required fitting to a 2:1 heterogeneous ligand-binding model yielding two sets of binding constants. This suggests that they exist in equilibria of two free- and two pITAM-bound conformations (Fig. 6b and Supplementary Table 2). AA_DN and tSH2 binding data fit a 1:1 binding model with one set of constants. All ZAP-70 had K_D1’s of ~10nM. The corresponding k_on1s for YY_DN and FF_DN were twice as fast as for AA_DN and tSH2, while the k_off1s were half that of AA_DN and tSH2. The second set of (weaker) binding constants for YY_DN and FF_DN showed 10-fold higher K_D2s, with doubled on-rates and ~20-fold increased off-rates compared to their K_D1s. These data suggest that FF associates less efficiently with pCD3γ_CD because its fraction of weak binders is higher than YY’s. Together with the HDX–MS findings, these data suggest that the closed conformation causes weaker binding. We tested the effects of ZAP-70 phosphorylation on pITAM-binding. Only phosphorylated pYY_DN had to be fitted with a different binding model (1:1) than its unphosphorylated form, consistent with a complete transition into an open high affinity conformation. The fact that the ratio of K_D1 and K_D2 is preserved for pFF_DN suggests the equilibrium is defined by the intra-molecular interactions of Y315/Y319, which are not altered by phosphorylation in the FF mutant. In addition, pYY_DN showed an ~1.5-fold decrease in k_on2, resulting in a higher K_D2. This suggests that other ZAP-70 phosphorylation sites can weaken pITAM-binding.

Our data demonstrate that pITAM-binding dynamics of ZAP-70 are determined by the weak binding of the closed conformation and strong binding of the open conformation, as well as their equilibrium. The closed conformation is more likely to associate with pITAMs. The propensity of the kinases to be in the closed conformation (FF_DN > YY_DN) correlates with increased dissociation rates. Stabilization of the open conformation by either phosphorylation or alanine mutation of Y315/Y319 strengthens the interaction of ZAP-70 with pITAMs and extends its TCR dwell-time. The latter increases the likelihood of ZAP-70 becoming activated by either CD4- or CD8-associated Lck, or another TCR associated ZAP-70.

ZAP-70 recruitment to microclusters (MCs) was analyzed to test if the different binding characteristics we observed in vitro exist in vivo. MCs are membrane structures that are formed around ligand-engaged TCR (Supplementary Fig. 5a)^{35, 36}. MCs are crucial for signal initiation and contain most of the signaling molecules involved in early T cell activation. The interaction kinetics of wildtype ZAP-70 with the TCR were previously studied by fluorescence recovery after photo-bleaching using confocal microscopy (FRAP^Conf)³⁵. ZAP-70 showed high exchange rates with the MCs during early (<5min) T cell activation and strongly reduced exchange rates at later stages (10–20min). Therefore, we analyzed early ZAP-70 recruitment to MCs by FRAP^Conf. GFP-tagged ZAP-70s (^GFPYY, ^GFPFF, GFPAA and ^GFPtSH2) or GFP-tagged TCR subunit CD3ζ (^GFPCD3ζ) were expressed in ZAP-70 deficient (P116) (Fig. 7a) and wildtype (E6.1) Jurkat T cell lines (Supplementary Fig. 5b). Cells were activated on antibody-coated glass surfaces to form MCs (Supplementary Fig. 5a) and the mobile fraction of MC-associated ZAP-70 or CD3ζ as well as their halftimes of recovery (t_1/2) were determined (Fig. 7b, Supplementary Fig 5c and Supplementary Table 3). It should be noted that binding to all ten pITAMs of the TCR/CD3-complex contributes to ZAP-70 clustering. Fluorescence recovery of ^GFPCD3ζ is minimal due to TCR immobilization on the antibody surfaces. Wildtype ZAP-70 (^GFPYY) showed 80–90% mobile fraction and a t_1/2 of 19 seconds. ^GFPFF cannot compete with endogenous ZAP-70 for TCR binding and could only be analyzed in P116 T cells (Supplementary Fig 5a); ^GFPFF recovers completely (~100%) and has a t_1/2 of 14 seconds in P116 cells. The alanine mutant ^GFPAA only recovers to ~50% and has a slower t_1/2 of 28 seconds. ^GFPtSH2 showed similar values as ^GFPAA with ~60% recovery and t_1/2 of 26 seconds. Additionally, we analyzed ZAP-70 exchange rates in primary T cells from 5c.c7 TCR transgenic mice. These T cells recognize agonist peptide derived from moth cytochrome c (MCC; amino acid 88–103) presented by MHC class II molecules (specifically I-E^k). T cells were activated on I-E^k/MCC- and co-stimulatory CD80-coated glass surfaces. Smaller MCs and increased cytosolic background required exchange to be measured using total internal reflection microscopy (FRAP^TIRF). Mobile fractions and halftimes [^GFPYY (~80%; 11s), ^GFPFF (~75%; 3.5s), ^GFPAA (~45%; 9s) and ^GFPtSH2 (~60%; 11s)] showed similar results for the murine ZAP-70 mutants as was observed for human ZAP-70s in Jurkat T cells (Fig. 7c+d and Supplementary Table 3).

These results corroborate our in vitro binding studies; they show that the propensity of FF to be in the closed conformation causes high exchange rates and thus a low probability of activation. In contrast, the constitutively open AA is more stably associated with MCs and can initiate T cell activation. The intermediate mobile fraction of YY suggests that only a fraction of TCR-bound ZAP-70 becomes phosphorylated on Y315/319 at any given time and then binds similarly to AA. The fraction that does not encounter an active kinase is released like FF. The absent (FF) and reduced (AA) activity of the mutants in T cells suggests an optimized TCR dwell-time and phosphorylation sequence for ZAP-70 is required to achieve maximal kinase and cellular activity.