To Bind or Not to Bind: APP Phosphorylation and Fe65
With the growing realization that it is functionally important in APP biology, the APP cytoplasmic domain has been attracting much more scrutiny lately. Many cytoplasmic adapter proteins, including Fe65, interact with APP at the EYNPTY motif present in the APP cytoplasmic domain. This motif is present downstream of the T668 residue that can be phosphorylated by many proline-directed kinases. This paper by Nakaya and Suzuki, and the recent paper by Chang et al. in Mol. Cell Biol. (Chang et al., 2006) examined the effects of APP phosphorylation at T668 on its interaction with Fe65 and transcriptional activation by AICD (γ-secretase-cleaved, APP intracellular domain). Nakaya and Suzuki conclude that phosphorylation of APP or AICD at T668 suppressed their association with Fe65 but had no role in translocation of AICD to the nucleus or had any discernible effect on Fe65 mediated transactivation. By contrast, Chang et al. report that phosphorylation of AICD at T668 was...  Read more

To Bind or Not to Bind: APP Phosphorylation and Fe65
With the growing realization that it is functionally important in APP biology, the APP cytoplasmic domain has been attracting much more scrutiny lately. Many cytoplasmic adapter proteins, including Fe65, interact with APP at the EYNPTY motif present in the APP cytoplasmic domain. This motif is present downstream of the T668 residue that can be phosphorylated by many proline-directed kinases. This paper by Nakaya and Suzuki, and the recent paper by Chang et al. in Mol. Cell Biol. (Chang et al., 2006) examined the effects of APP phosphorylation at T668 on its interaction with Fe65 and transcriptional activation by AICD (γ-secretase-cleaved, APP intracellular domain). Nakaya and Suzuki conclude that phosphorylation of APP or AICD at T668 suppressed their association with Fe65 but had no role in translocation of AICD to the nucleus or had any discernible effect on Fe65 mediated transactivation. By contrast, Chang et al. report that phosphorylation of AICD at T668 was essential for its binding to Fe65, translocation to the nucleus and gene transactivation. What gives?

Fe65 binds with APP cytoplasmic domain, and this association is shown to modulate cell migration and activate gene transcription. Extending their earlier observations that phosphorylation of APP at T668 inhibited its binding to Fe65, Suzuki's group used a pT668-specific antibody to show that both APP and AICD are phosphorylated in vivo in mouse brain. Both phosphorylated and non-phosphorylated forms of AICD were present in the membrane, whereas the nuclear AICD was mostly non-phosphorylated, suggesting that AICD phosphorylation was not needed for nuclear entry. Indeed, transgenic mice expressing T668A mutant APP also showed AICD to be present in the nucleus. It is not known whether T668A AICD is transcriptionally active since they did not directly measure the transcriptional activity of T668A-AICD. Also, coexpression of Fe65 with APP caused Fe65 to associate with the membrane, and phosphorylation of APP with JNK disrupted the binding, confirming that phosphorylation at T668 inhibits the APP-Fe65 interaction. Interestingly, coexpression of APP with Fe65 inhibited AICD-Fe65 mediated transcriptional activation (presumably by recruiting Fe65 to membrane and preventing its entry into the nucleus), whereas T668A-APP failed to inhibit the transcriptional activity.

By contrast, the paper by Chang et al. concludes that phosphorylation of AICD is needed for its binding to Fe65. They expressed various GFP-AICD fusion proteins in NGF-differentiated PC12 cells and observed that wild-type AICD-GFP was present in the nucleus, but mutant AICD-GFP expressing T668A or lacking the Fe65 binding motif EYNPTY were absent from the nucleus. One way to reconcile these differences is to remember that the AICD produced in vivo is only 6 kDa in size and therefore small enough to pass through the nuclear pore, while GFP-AICD (~35 kDa) is not. Thus, it is unlikely that phosphorylation of AICD is required in vivo to facilitate its entry into the nucleus. Another apparent discrepancy is that Chang et al. conclude that T668 phosphorylation is required for Fe65 binding. This conclusion is based on the observation that wild-type AICD-EGFP showed FRET with Fe65 (suggesting a close interaction between the two proteins) whereas T668A-AICD-EGFP, which cannot be phosphorylated, did not show FRET. They interpret these results to suggest that phosphorylation at T668 is essential for interaction with Fe65. However, in the absence of direct evidence that phosphorylation at T668 is essential for Fe65 binding, a more likely interpretation of their finding is that the T668A mutation alters the conformation of APP-CTF and precludes its binding to Fe65.

An important aspect of the Chang et al. study is that it confirms an earlier observation of increased T668-phosphorylation in human AD brains (Lee et al., 2003) and demonstrates the same to be true for Tg2576 animal model of AD. While the relationship between increased APP phosphorylation and AD brain pathology is causal or correlative remains to determined, it is becoming clear that T668 phosphorylation significantly alters APP metabolism by dissociating Fe65 or recruiting Pin1 (Pastorino et al., 2006) or both. Future studies will establish the significance of APP phosphorylation in AD pathology. However, if conflicting reports and contradictory conclusions (as exemplified by Aβ peptides) are a sign of coming of age, then these two papers may be an indication that APP cytoplasmic domain has arrived!

View all comments by Sanjay W. Pimplikar

Comments by Dr. Pimplikar for our paper and those of Chang et al. implicate APP phosphorylation at Thr668 in APP (and/or AICD) function or amyloidogenic processing of APP. Increasing reports and reviews on the phosphorylation state of APP Thr668 indicated direct and indirect participations of APP phosphorylation in AD pathogenesis. We, as the original finder of APP Thr668 phosphorylation, welcome these attempts to clarify whether the APP phosphorylation participates deeply in the pathogenesis of AD or not, because some reports suggest that Thr668 phosphorylation alters APP metabolism and may present a novel druggable target for AD therapy.

As described in Dr. Pimplikar's comments, we generated Thr668Ala and Thr668Asp mutant mice (which, incidentally, are not transgenic as he suggests). The mutant mice were generated by a standard gene knock-in method. Some of our research findings, including the knock-in effects on APP metabolism, including Aβ generation, will be presented at ICAD conference in Madrid, symposium session S5-02, morning of July 20.

View all comments by Tadashi Nakaya