Our findings reveal an autoinhibitory switch in ZAP-70 and Syk and suggest that, in contrast to the previously held view, Y315 and Y319 in ZAP-70 do not play a requisite scaffolding function in order to recruit substrates. Rather, they are primarily responsible for an autoinhibitory switch that regulates kinase function. Our findings suggest that Lck or another kinase activates ZAP-70 by phosphorylating Y319 (and possibly Y315), which, by analogy to the juxtamembrane region of several RTKs (
15), relieves a hydrophobic autoinhibitory interaction with the N-terminal lobe and the activation loop of the kinase domain. This likely induces a conformational change that enhances the kinase activity and/or accessibility of the activation loop tyrosines for phosphorylation.
This hypothesis is supported by experiments in 293 and Cos cells as well as in the P116 Jurkat T-cell line. The YYAA mutants of ZAP-70 and Syk appeared dysregulated in 293 cells, as measured by tyrosine phosphorylation of the ZAP-70 activation loop and the ability of both of these YYAA mutants to phosphorylate their substrates. In contrast to the YYFF mutant, YYAA ZAP-70 substantially reconstituted a number of signaling events in P116 cells, including tyrosine phosphorylation, calcium flux, transcriptional activation of NFAT, and activation of the Erk1/2 MAP kinase pathway. Also of note is the fact that we had difficulties generating wild-type expression levels of the interdomain B deletion or YYAA mutants in stable P116 transfectants, suggesting some toxicity of these potentially unregulatable mutants. This may be explained by the fact that activation of the TCR can induce apoptosis in Jurkat cells via Fas-dependent pathways (
9).
Although our experiments with the YYAA mutant of ZAP-70 suggest that binding of signal transducing molecules to ZAP-70 tyrosines Y315 and Y319 has only partial amplifying effects on several major TCR-dependent signaling pathways, the scaffolding function of these tyrosines cannot be entirely dismissed. Our observation that some of the signaling events, most notably full tyrosine phosphorylation of LAT, SLP-76, and PLC-γ1, could not be completely rescued by YYAA ZAP-70 in P116 cells suggests that there still may be some, perhaps only quantitative, function for Y315 and Y319 in the recruitment of downstream signaling molecules.
Apart from directly affecting the downstream signaling pathways, the molecules interacting with phosphorylated Y315 and Y319 potentially can be involved in the regulation of ZAP-70 itself. Binding of SH2 domain-containing proteins to phosphorylated interdomain B tyrosines may stabilize the active conformation of ZAP-70 and protect Y315/Y319 from dephosphorylation. In addition, the effector functions of these molecules could directly influence ZAP-70 kinase activity. Indeed, Pelosi et al. suggested that binding of Lck via its SH2 domain to phosphorylated Y319 followed by Lck-mediated phosphorylation of Y493 in the activation loop of ZAP-70 is an important positive feedback mechanism in ZAP-70 activation (
26).
Up until now, Lck has been viewed as the major PTK responsible for the phosphorylation of ZAP-70 Y493. This was primarily based on mass spectrometry and two-dimensional analysis of tryptic peptides generated from purified recombinant ZAP-70 protein phosphorylated by Lck (
5,
34) as well as the data of Pelosi et al. (
26). These studies however, did not investigate the role of other kinases, including Fyn and c-Abl, in ZAP-70 phosphorylation. The first two studies also failed to examine phosphorylation of Y315/Y319 and thus did not take into account the influence of these sites on ZAP-70 activity and the ability to autophosphorylate at Y493. Our data from the 293 cell overexpression system revealed that, unlike WT ZAP-70, mere overexpression of the ZAP-70 YYAA mutant led to its phosphorylation at Y493, suggesting that YYAA was able to autophosphorylate at Y493, perhaps via an intermolecular
trans-phosphorylation mechanism (Fig. ). Moreover, with Lck as well as with Fyn and c-Abl cotransfection, we observed high levels of Y493 phosphorylation in WT ZAP-70 but only marginal phosphorylation of Y493 in the kinase-inactive mutant, despite comparable phosphorylation at Y319. These results suggest that these kinases are not very efficient at phosphorylating Y493, even when Y319 is phosphorylated and potentially available for binding. This seems to suggest that in 293 or Cos cells phosphorylation of interdomain B Y315/Y319 is the only major event dependent on kinase(s) other than ZAP-70 itself, which in turn results in an increase of ZAP-70 kinase activity and its
trans-autophosphorylation of the activation loop. These data also seem to rule out the possibility that phosphorylation of Y315/Y319 alone substantially enhances accessibility of Y493 for phosphorylation by Lck, Fyn, or c-Abl. In P116 Jurkat cells, however, the KA mutant was hyperphosphorylated at both Y319 and Y493, and its phosphorylation was further increased upon TCR stimulation. This apparent discrepancy could result from differential regulation of ZAP-70 in the P116 Jurkat T-cell line versus 293 or Cos cells, which are of epithelial origin. T cells likely have multiple regulatory mechanisms to control their key signaling molecules, including ZAP-70. Some of these mechanisms may be missing in 293 or Cos cells because they do not express endogenous ZAP-70 and therefore would have no need to specifically control its activity. The observed hyperphosphorylation of KA ZAP-70 in P116 cells indeed suggests the existence of a ZAP-70-driven negative feedback loop in T cells. By impairing ZAP-70 activity by the KA mutation, this putative regulatory mechanism is turned off and nonphysiological hyperphosphorylation of Y319 and Y493 occurs. A similar regulatory mechanism might also explain the incomplete reconstitution of several tyrosine phosphorylation events in P116 cells expressing the potentially hyperactive YYAA ZAP-70 mutant, which would enhance the activity of the same negative feedback loop. However, considering the difficulties we had in obtaining the cells stably expressing YYAA ZAP-70, it is also possible that only the clones able to compensate for increased ZAP-70 activity by reinforcing some of the negative regulatory pathways were selected.
The observation that KA ZAP-70 could be phosphorylated at Y493 also shows that ZAP-70 is not the only PTK able to phosphorylate Y493 and that other PTKs, most likely Lck (
26), participate in phosphorylating this site. Additional studies are needed to dissect the contribution of different kinases to Y493 phosphorylation. Nevertheless, based on currently available data, we favor the possibility that Lck and ZAP-70 are the major kinases phosphorylating Y493 in vivo. It is possible that Lck is responsible only for the initial but rather inefficient phosphorylation of Y493 on several ZAP-70 molecules. These activated ZAP-70 molecules may in turn
trans-autophosphorylate other ZAP-70 molecules with which they are dimerized by paired ITAMs in the TCR-ζ and CD3 chains, much like dimerized RTKs. Since there certainly is rapid turnover of ZAP-70 binding to ITAMs (
3), it is likely that one active ZAP-70 molecule can activate many other ZAP-70 molecules leading to substantial amplification of the initial signal. Consistent with this model, it has been demonstrated that phosphorylated TCR-ζ chain dimers potentiate ZAP-70 in vitro catalytic activity substantially more than do monomers (
21). Also, this mechanism may explain why ITAMs are almost always found as pairs in homo- or heterodimeric non-ligand-binding chains within the receptors of the hematopoietic lineage involved in antigen recognition: i.e., to allow for induced dimers of ZAP-70 or Syk to
trans-autophosphorylate.
The importance of the activation loop phosphorylation of Syk family kinases has recently been questioned by the studies describing the structures of the isolated kinase domains of ZAP-70 and Syk (
1,
19). In both kinase domains, the unphosphorylated activation loop assumes an extended conformation, typical for an activated kinase. In ZAP-70, this might have been driven by the interactions between the activation loop and the kinase domain of neighboring molecules in the crystal, but it also might suggest that phosphorylation of the activation loop is not as critical for ZAP-70/Syk activation as previously was thought. Interdomain B phosphorylation or YYAA mutation may be sufficient to increase ZAP-70/Syk activity enough to autophosphorylate and/or phosphorylate its downstream substrates. By analogy with EphB2, interdomain B in fact may regulate the activation loop conformation and the observed structure may reflect the lack of interdomain B in the construct used for structure determination. In such a circumstance, phosphorylation of the activation loop probably would have mainly stabilizing effects.
It appears that ZAP-70 activation is a complicated multistep process requiring proper spatial as well as temporal coordination of several key events (Fig. ). Upon TCR stimulation and phosphorylation of TCR ITAMs by Src family kinases, ZAP-70 is recruited to phosphorylated ITAM sequences. Subsequently, Y315 and Y319 in interdomain B of ZAP-70 become phosphorylated by Src family or Abl/Arg family PTKs. This results in the release of ZAP-70 from an autoinhibited state, the upregulation of kinase activity, and the stabilization of the interaction between ZAP-70 and phosphorylated TCR chains. Subsequent steps are not clear yet, but they likely involve phosphorylation of the ZAP-70 activation loop, possibly via the cooperative action of Lck and ZAP-70 itself resulting in the fully active enzyme.
Collectively, our findings suggest that the cytoplasmic PTKs ZAP-70 and Syk share with RTKs a regulatory mechanism involved in inhibiting catalytic activity that involves two tyrosine residues N terminal to their kinase domains. The paired autoinhibitory tyrosines are localized within the juxtamembrane regions of RTKs, whereas they are within interdomain B of ZAP-70 and Syk. Thus, one could view ZAP-70 and Syk essentially as RTKs in which membrane localization is just another means of regulation. However, once localized to the membrane, these kinases are activated through mechanisms similar to those involved in RTK activation.