An allosteric hot spot in the tandem-SH2 domain of ZAP-70 regulates T-cell signaling

T-cell receptor (TCR) signaling is initiated by recruiting ZAP-70 to the cytosolic part of TCR. ZAP-70, a non-receptor tyrosine kinase, is composed of an N-terminal tandem SH2 (tSH2) domain connected to the C-terminal kinase domain. The ZAP-70 is recruited to the membrane through binding of tSH2 domain and the doubly-phosphorylated ITAM motifs of CD3 chains in the TCR complex. Our results show that the tSH2 domain undergoes a biphasic structural transition while binding to the doubly-phosphorylated ITAM-ζ1 peptide. The C-terminal SH2 domain binds first to the phosphotyrosine residue of ITAM peptide to form an encounter complex leading to subsequent binding of second phosphotyrosine residue to the N-SH2 domain. We decipher a network of non-covalent interactions that allosterically couple the two SH2 domains during binding to doubly-phosphorylated ITAMs. Mutation in the allosteric network residues, for example, W165C, uncouples the formation of encounter complex to the subsequent ITAM binding thus explaining the altered recruitment of ZAP-70 to the plasma membrane causing autoimmune arthritis in mice. The proposed mechanism of allosteric coupling is unique to ZAP-70, which is fundamentally different from Syk, a close homolog of ZAP-70 expressed in B-cells. Significance T-cell and B-cell signaling is initiated by the same family of non-receptor tyrosine kinases, ZAP-70 and Syk, respectively. ZAP-70 and Syk share homologous sequence and similar structural architecture, yet the two kinases differ in their mode of ligand recognition. ZAP-70 binds cooperatively to its ligand, whereas Syk binds uncooperatively. Spontaneous mutation (W165C) in the regulatory module of ZAP-70 impairs T-cell signaling causes autoimmune arthritis in SKG mice, the mechanism of which is unknown. We showed that ZAP-70 regulatory module undergoes a biphasic structural transition while binding to its ligand, which is fundamentally different from Syk. We presented a molecular mechanism of cooperativity in ZAP-70 regulatory module that explains altered ligand binding by ZAP-70 mutant found in SKG mice.


51
The zeta-chain-associated protein tyrosine kinase, ZAP-70, is a non-receptor tyrosine kinase

81
Alternatively, biochemical analysis and molecular dynamics simulation suggest that the N-SH2 domain 82 may first bind to phosphotyrosine residue of ITAM peptide with low micromolar affinity followed by 83 cooperative binding of second phosphotyrosine to the C-SH2 domain (11,30,32,35).  from an open to a closed state upon binding to doubly-phosphorylated ITAM-z1 peptide. Using 97 molecular dynamics simulation, NMR spectroscopy, and biochemical analysis of different tSH2 domain 98 mutants, we show that the C-SH2 domain binds first to the phosphotyrosine residue of the ITAM 99 peptide. Following a plateau, the second phosphotyrosine residue of the ITAM peptide binds the N-SH2 100 phosphate-binding pocket. We deciphered an allosteric network, found only in ZAP-70, assembled by 101 threading aromatic stacking interactions that connect N-SH2 and C-SH2 phosphate-binding pockets.

102
The proposed model of allosteric network explained the molecular mechanism of altered interaction of 103 W165C mutant of ZAP-70 and doubly-phosphorylated ITAM peptide in SKG mice.

120
The two binding events are interleaved by a plateau where no conformational changes were observed.

121
We next tested the binding of ITAM-Y2P-z1 to the tSH2 domain of Syk by intrinsic tryptophan

132
In summary, we showed that the tSH2 domain of ZAP-70 binds to the doubly-phosphorylated 133 ITAM-z1 peptide in a biphasic pattern with three distinct binding events correspond to strong, medium 134 and weak dissociation constant regimes (

166
We analyzed the average root-mean-square fluctuations (RMSF) to elucidate the domain-167 specific dynamic behavior of the protein at the residue level. We observed an overall increase in RMSF 168 for the tSH2-apo, N-SH2 ITAM-YP , and C-SH2 ITAM-YP structures than the tSH2-holo state ( Figure 2C). Map 169 of residue-specific RMSF greater than 3.5 Å on the tSH2-holo structure ( Figure S3c  an overall increase in neighborhood connectivity for the tSH2-holo structure ( Figure 5A).

291
We began our analysis with the tSH2-holo structure and searched for the shortest residue 292 interaction pathway involving a minimum number of steps (amino acids) connecting the two-phosphate 293 binding pockets through the allosteric hot-spot residues. As shown in figure 5b, the network initiates 294 with R192 at the phosphate-binding pocket of the C-SH2 domain, which is connected to the W235 and 295 W165. W235, in turn, is connected to F117 by p-p aromatic stacking interaction that finally converged 296 to R43 at the N-SH2 phosphate-binding pocket. In the holo-state W235 has the highest node degrees 297 is sandwiched between F117 and W165. Which suggests that W235 might function as an allosteric 298 switch (nodal hub) that couples the two SH2 domains during ITAM binding. In the apo-state, the F117-

299
W235 p-p aromatic stacking interaction is broken, which might uncouple the allosteric network between 300 C-SH2 and N-SH2 domains (Figure 5b).

305
To evaluate the importance of the allosteric-network in the tSH2 domain of ZAP-70 during 306 doubly-phosphorylated ITAM binding, we studied the overall structure and dynamics of four in-silico 307 mutants tSH2-holo W165C , tSH2-holo F117A , tSH2-holo R43A and tSH2-holo R43P (Figure 6a). All the tSH2 308 domain in-silico mutants used in the molecular dynamics simulations were prepared on the tSH2-holo

316
R192] coupling the two SH2 domains is present throughout the simulation trajectories, we measured 317 the time-dependent pairwise distance between the residues with the representative side-chain atoms 318 ( Figure S9). We observed that the network connectivity was maintained throughout the simulation 319 trajectory for the wildtype tSH2-holo structure (Table S3 and Figure S9a). However, we noted that 320 during the simulation, Q236 rearranges in a stacking position between F117 and W235, providing 321 stability to the network. In the network-mutants, tSH2-holo W165C , tSH2-holo F117A , tSH2-holo R43A and 322 tSH2-holo R43P , residue interaction-network connecting the two SH2 domains was significantly 323 destabilized and broken ( Figure S9b-e). The mutation increases the average pairwise distance between 324 the key amino acid residues in the network (Table S3)

328
We next investigate the strength of wildtype tSH2-holo and the mutated tSH2 structures to bind 329 the doubly-phosphorylated ITAMs from the simulation trajectories of the respective system (Table S2).

330
The average interaction energies were found to be -713.

335
To test the role of the proposed allosteric network, we made three mutation R43P, F117A, and 344 domain was preserved (Figure 7 and S10c, Table 1). We could not determine any medium or weak  (Table S5).

360
We evaluated the stability of each of the structures (holo and apo conformation) of wildtype and 361 mutated tSH2 domain by CD spectroscopy ( Figure S11). As expected, apo-and the holo-state of the 362 wildtype tSH2 domain represents the lowest and highest thermally stable conformation(32),

363
respectively. All the tSH2 domain mutants in the apo-state clustered together with wildtype tSH2-apo 364 structure, indicating the mutation did not change the overall stability of the proteins. The tSH2-holo R192A 365 mutant, which did not bind to doubly-phosphorylated ITAM-z1 peptide, exhibit similar thermal stability 366 as the tSH2-apo. The tSH2-holo R39A mutant that has a functional C-SH2 phosphate-binding pocket 367 showed intermediate stability ( Figure S11). The thermal stability of the holo-state for the wildtype and

386
resulting in a plateau during the intrinsic tryptophan fluorescence experiment (Figure 1c and b). To 387 adopt a stable tSH2-holo structure requires a reorientation of the aromatic residues F117, W165, and 388 W235 into a stacking interaction, which imposes a higher energetic penalty ( Figure S11 and Table S5).

423
Ethics Declaration:  determined from the curve-fitting to one-site specific binding model using Prism (Figure S1f and S1g).

579
The Hill-coefficient (nH) was calculated from the Hill-plot ( Figure S1c