Regulation of Notch1 signaling by the APP intracellular domain facilitates degradation of the Notch1 intracellular domain and RBP-Jk

Notch is a highly conserved transmembrane receptor that performs a key role in the determination of cell fate, differentiation, adult cell self-renewal, cancer, neurodegenerative disease, wound healing, and inflammation (Artavanis-Tsakonas et al., 1995; Egan et al., 1998; Lai, 2004; Weinmaster, 1998). The Notch1 receptor plays the role of a membrane-bound transcription factor. Notch1 is processed by furin in the endoplasmic reticular Golgi complex (S1 cleavage) during transport to the cell surface, where it is expressed in heterodimeric form (Lieber et al., 2002; Pan and Rubin, 1997). Upon binding to the specific ligands, Jagged and Delta, the transmembrane C-terminal fragment of Notch is generated via proteolytic cleavage (S2 cleavage) (Brou et al., 2000; Mumm and Kopan, 2000). Cleavage of this fragment by γ-secretase (S3 cleavage) induces the release of the Notch intracellular domain (Notch-IC) from the membrane, and induces the nuclear translocation of Notch-IC, thus resulting in the formation of a complex with the CSL transcription factor family [CBF1/RBP-Jk/KBF2 in mammals, Su(H) in Drosophila and Xenopus, and Lag2 in Caenorhabditis elegans] (Capell et al., 2000; De Strooper et al., 1999; Mumm and Kopan, 2000; Ray et al., 1999; Steiner et al., 1999; Weinmaster, 1997). In the absence of Notch-IC, CSL interacts with the SKIP, SMRT, CoR and HDAC proteins, resulting in the formation of a transcriptional repressor complex (Espinosa et al., 2002; Kao et al., 1998; Zhou et al., 2000; Zhou and Hayward, 2001). Notch-IC dissociates the co-repressors, and Notch-IC interacts with co-activator complexes, including the Lag-3/mastermind, p300/CBP and P/CAF/GCN5, to form a transcriptional active complex and activates CSL-dependent transcription (Kurooka and Honjo, 2000; Oswald et al., 2001; Petcherski and Kimble, 2000; Schuldt and Brand, 1999; Wallberg et al., 2002). The RAM domain of Notch1, which mediates the interaction of RBP-Jk/Su(H) with the Notch1-IC, induces the activation of target gene transcription (Tamura et al., 1995; Tani et al., 2001). In addition to the enhancer of split [E(spl)] complex genes, and the mammalian homologues of the Hairy and E(spl) genes, Hes1, Hes5, Hes7, Hey1, Hey2 and Heyl are the downstream target genes of Notch signaling (Abu-Issa and Cavicchi, 1996; Bessho et al., 2001; de Celis et al., 1996; Fischer et al., 2004; Jennings et al., 1994; Jouve et al., 2000; Leimeister et al., 2000; Ligoxygakis et al., 1998; Maier and Gessler, 2000; Ohtsuka et al., 1999).

Following the transcriptional regulation of the target genes, Notch1-IC undergoes proteasomal degradation in the nucleus via the ubiquitin-proteasome system, including Fbw7, an E3 ligase relevant to the ubiquitylation of Notch1-IC (Gupta-Rossi et al., 2001; Lai, 2002; Minella and Clurman, 2005; Mo et al., 2007; Oberg et al., 2001; Wu et al., 2001). Several E3 ubiquitin ligases have been implicated in the half-life of Notch1-IC, including Fbw7, which promotes PEST-dependent Notch1-IC degradation in the nucleus, and Itch, which regulates the PEST-independent degradation of cytoplasmic Notch protein (Lai, 2002). We demonstrated previously that ILK downregulates the protein stability of Notch1-IC via the ubiquitin-proteasome pathway by means of Fbw7 (Mo et al., 2007).

The amyloid-β precursor protein (APP) is a type 1 integral transmembrane protein composed of a large extracellular sequence, a single transmembrane region and a short intracellular fragment, which is a cytotoxic 39–43 residue peptide that performs a crucial function in the pathogenesis of Alzheimer's disease (Beyreuther and Masters, 1991; Tang, 2005; Younkin, 1991). Under physiological conditions, APP is cleaved proteolytically by secretase activity. APP is cleaved in a fashion similar to that of Notch, which undergoes regulated intramembranous proteolysis induced by γ-secretase to release the APP intracellular domain (AICD), which modulates transcription (Tomita et al., 1998; Zhang et al., 2000). Indeed, AICD is capable of inducing transcriptional activation by interacting with the adaptor protein Fe65 and the acetyltransferase Tip60 (Cao and Sudhof, 2001). AICD was initially identified in the brains of patients with AD and was demonstrated to either sensitize or induce cells to undergo apoptosis. We demonstrated previously that Notch1-IC downregulates the AICD transcriptional activity through physical binding with AICD, Fe65 and Tip60 (Kim et al., 2007b). We have also demonstrated that Notch1-IC is a novel substrate for Tip60 and acetylation is one of the key factors in the regulation of the Notch1 signaling pathway (Kim et al., 2007a). D'Adamio's group has demonstrated that AICD binds to the cytosolic Notch inhibitors Numb and Numb-like, both of which can repress Notch activity (Roncarati et al., 2002). Despite the fact that AICD regulates Notch1 signaling, the precise mechanism underlying this control remains to be clarified.

In this study, we demonstrate that signal crosstalk occurs between AICD and Notch1 signaling after their processing by γ-secretase. We have now evaluated the mechanism relevant to the AICD-mediated regulation of Notch signaling. Our data indicate that AICD inhibits the transcriptional activity of Notch1-IC by an induced reduction in the protein stability of Notch1-IC and RBP-Jk. Interestingly, the level of the Notch1-IC protein was downregulated markedly in the presence of AICD by the proteasomal degradation of Notch1-IC through Fbw7. Additionally, the level of RBP-Jk protein was downregulated markedly in the presence of AICD by the lysosomal degradation of RBP-Jk. Collectively, our findings demonstrate that AICD functions as a negative regulator of the protein turnover of Notch1-IC and RBP-Jk.

To evaluate the possible function of AICD in Notch1 signaling, a reporter assay was conducted with HEK293 cells, using luciferase reporter genes. HEK293 cells were transfected with 4×CSL-Luc, and either the active Notch1 mutant ΔEN1 or an empty vector. As anticipated, ΔEN1-mediated transcription activity was found to have increased in these samples. We determined that AICD attenuated the ability of ΔEN1 to stimulate transcription (Fig. 1A). The basic helix-loop-helix (bHLH) proteins, Hes1 and Hes5, both of which harbor several RBP-Jk-binding sequences on their promoters, were identified as essential targets of Notch in epithelial cells (Kageyama and Ohtsuka, 1999). Therefore, we confirmed the effects of AICD on the Notch1 signaling pathway, using the Hes1 reporter system. The expression of the active form of Notch1 significantly induced the activation of the Hes1 reporter system (Fig. 1B). Overexpression of AICD inhibits constitutively active Notch1-induced natural Hes1 promoter transcriptional activity (Fig. 1B). We then attempted to determine whether the overexpression of AICD influences Notch1-IC-mediated signaling. HEK293 cells were transfected with 4×CSL-Luc or Hes1-Luc and either the Notch1 intracellular domain (Notch1-IC) or an empty vector. Expression of Notch1-IC was found to significantly induce activation of the 4×CSL and Hes1 reporter systems (Fig. 1C,D). Overexpression of AICD inhibited Notch1-IC-induced 4×CSL and natural Hes1 promoter transcription activities (Fig. 1C,D).

The phorbol ester phorbol 12-myristate 13-acetate (PMA) has been shown to trigger γ-secretase-mediated APP cleavage and we attempted to determine whether PMA could modulate APP cleavage through γ-secretase, thereby modulating Notch1 signaling (Fukushima et al., 1993; Kume et al., 2004; Suh and Checler, 2002). We used APP–Gal4/VP16 fusion proteins to measure the γ-secretase-mediated cleavage of APP (Biederer et al., 2002; Herranz et al., 2006; May et al., 2002; Wiley et al., 2010). HEK293 cells were transfected with APP–Gal4/VP16 and Gal4–Luc, and treated with either PMA or DAPT, a γ-secretase inhibitor. As anticipated, PMA triggered the ability of APP–Gal4/VP16 to stimulate transcription in a dose-dependent fashion (Fig. 1E). Upon treatment with a γ-secretase inhibitor, γ-secretase activity was attenuated significantly (Fig. 1E). In an effort to delineate the possible role of AICD, which was generated from APP by γ-secretase, in the regulation of Notch1-IC signaling, we determined the effects of PMA on Notch1-IC transcriptional activity. Whereas Notch1-IC-mediated transcriptional activity was repressed in the PMA-exposed HEK293 cells, the PMA-induced suppression of Notch transcriptional activity was restored by treatment with DAPT or coexpression with APP siRNA (Fig. 1F,G). According to these results, it was assumed that PMA can reduce Notch1-IC-mediated transcriptional activity by the upregulation of γ-secretase-dependent APP cleavage.

We, and others, have reported previously that AICD might interact with Notch1-IC (Fassa et al., 2005; Fischer et al., 2005; Kim et al., 2007b; Oh et al., 2005). To delineate more precisely the manner in which AICD prevents Notch1-IC and RBP-Jk-mediated transcription, we conducted a series of in vitro binding and coimmunoprecipitation experiments. Lysates from cells expressing Myc–Notch1-IC and FLAG–RBP-Jk were immunoprecipitated and subsequently incubated with either GST or with GST–AICD. The formation of the Notch1-IC and RBP-Jk complex was suppressed substantially by AICD in vitro (Fig. 2A). Notch1-IC harbors a CDC domain that incorporates a RAM domain, seven ankyrin repeats (ANK), an OPA domain, and a PEST domain within its structure. Therefore, we attempted to determine which, if any, of these domains are involved in the interaction between Notch1-IC and AICD. Our results showed that AICD bound to the RAM–ANK domain of Notch1, but not to the OPA and PEST domains (Fig. 2B). We also attempted to confirm that the physical association of endogenous AICD with endogenous Notch1-IC or RBP-Jk in intact cells. Coimmunoprecipitation assays with endogenous Notch1-IC and RBP-Jk revealed binding with endogenous AICD in intact cells (Fig. 2C,D). To evaluate the effects of AICD on the molecular interactions between Notch1-IC and RBP-Jk in intact cells, coimmunoprecipitation was conducted in HEK293 cells by cotransfection of Myc–Notch1-IC, FLAG–RBP-Jk and GFP–AICD. Notch1-IC and RBP-Jk were coimmunoprecipitated, but when they were cotransfected with AICD, the band of Notch1-IC that interacted with RBP-Jk disappeared (Fig. 2E). Surprisingly, the protein levels of both Notch1-IC and RBP-Jk were significantly downregulated upon cotransfection with AICD as determined by immunoblotting (Fig. 2E, lanes 4 and 6), which demonstrates that AICD might regulate the steady state levels of the Notch1-IC and RBP-Jk proteins. Because the AICD protein is known to accumulate upon treatment with PMA, we attempted to determine whether PMA could modulate endogenous protein levels of AICD, Notch1-IC or RBP-Jk (Fukushima et al., 1993; Kume et al., 2004; Suh and Checler, 2002). We observed that AICD protein accumulated in a dose-dependent manner upon PMA treatment. However, the steady state levels of Notch1-IC and RBP-Jk proteins were gradually decreased by PMA treatment (Fig. 2F). To delineate the possible role for AICD in the regulation of Notch1-IC protein stability, we assessed the effects of APP knockdown on Notch1-IC protein stability. We determined that APP siRNA was able to block the suppressive effects of PMA on downregulation of Notch1-IC protein level (Fig. 2G). Furthermore, as demonstrated in Fig. 2B, the PEST domain was just detectable when coexpressed with AICD, probably as a result of the rapid turnover of the protein. Otherwise, AICD modulated neither the Tip60 nor Fe65 protein levels (data not shown). Therefore, we could assume that attenuation of Notch1 transcription activity by AICD results from the downregulation of steady state Notch1-IC and RBP-Jk protein levels.

We attempted to determine whether Notch1-IC could be subjected to proteasome-mediated proteolysis, as previously reported (Gupta-Rossi et al., 2001; Lai, 2002; Minella and Clurman, 2005; Mo et al., 2007; Oberg et al., 2001; Wu et al., 2001). We transfected the HEK293 cells with Myc–Notch1-IC and GFP–AICD, and the quantity of remaining Notch1-IC was evaluated after various periods of cycloheximide treatment. We determined the protein stability of Notch1-IC in HEK293 cells by cycloheximide treatment with or without AICD. Cycloheximide interacts with the translocase enzyme and blocks protein synthesis in eukaryotic cells. After cycloheximide treatment, the level of Notch1-IC protein gradually dropped, with approximately half of the protein degraded after 3 hours without AICD (Fig. 3A). Upon cycloheximide treatment, the reduced level of Notch1-IC protein dropped rapidly, with approximately half of the protein being degraded after 1.5 hours in the presence of AICD (Fig. 3A). No reduction was noted in the level of actin used as a control (Fig. 3A). This result demonstrates that Notch1-IC is rapidly turned over in the presence of AICD.

To determine whether degradation of Notch1-IC proteins is mediated by the proteasome pathway, the proteasome inhibitor MG132 was used to treat Notch1-IC-expressing and AICD-expressing cells. MG132 can reversibly block all activities of the 26S proteasome (Rock et al., 1994). ALLN inhibits neutral cysteine proteases and the proteasome (Drexler, 1997). The cells were treated with proteasome inhibitors for 6 hours and Notch1-IC protein was detected via an immunoblot assay. The results of our studies demonstrated that the proteasome inhibitor significantly increased the level of Notch1-IC (Fig. 3B), which was reduced in the presence of AICD, but was restored by treatment with MG132 or ALLN (Fig. 3B,C).

We then attempted to determine whether lysosomal inhibitors exerted any detectable effect on the degradation of Notch1-IC. NH₄Cl is a very effective inhibitor of lysosomal function and inhibits the function of lysosomal proteases by an induced increase in the intralysosomal pH (Dean et al., 1984; Ohkuma and Poole, 1978). Owing to its relatively low cytotoxicity, NH₄C1 was primarily used in our study (Dean et al., 1984). We transfected HEK293 cells with Myc–Notch1-IC and GFP–AICD, and the quantity of remaining Notch1-IC was analyzed after various periods of NH₄Cl treatment. Our studies demonstrated that lysosomal inhibitor could not regulate the steady state level of Notch1-IC proteins (Fig. 3D). Moreover, compared with the proteasome inhibitor, the lysosomal inhibitor exerted no detectable effect on the AICD-induced downregulation of Notch1-IC (Fig. 3D). To confirm the role of MG132, NH₄Cl and chloroquine in the regulation of known target protein, we introduced c-Myc for proteasomal degradation and Notch3-IC for lysosomal degradation (Bahram et al., 2000; Gregory and Hann, 2000; Jia et al., 2009). Coincident with previous reports, we found the accumulation of those proteins in a dose-dependent manner (Fig. 3E–G). These results reveal that the stability of the Notch1-IC protein is downregulated by AICD through the proteasome-dependent pathway.

Nevertheless, despite the many studies of RBP-Jk conducted thus far, the mechanism underlying degradation of the protein remains incompletely understood. In an effort to evaluate the possible role of AICD in the regulation of RBP-Jk protein stability, we transfected HEK293 cells with FLAG–RBP-Jk and GFP–AICD, and the quantity of remaining RBP-Jk was analyzed after various periods of cycloheximide treatment. We determined the protein stability of RBP-Jk in HEK293 cells by cycloheximide treatment with or without AICD. After cycloheximide treatment, the level of RBP-Jk protein declined gradually with approximately half of the protein being degraded after 5 hours without AICD (Fig. 4A). The level of the RBP-Jk protein was reduced significantly without cycloheximide treatment; after treatment, the reduced level of RBP-Jk protein also declined rapidly, with approximately half of the protein being degraded after 2 hours in the presence of AICD (Fig. 4A). No reduction was detected in the level of actin used as a control (Fig. 4A). This result demonstrates that RBP-Jk is rapidly turned over in the presence of AICD.

In an effort to determine whether the degradation of RBP-Jk proteins is mediated by the proteasome pathway, the proteasome inhibitor MG132 was administered to cells expressing RBP-Jk and AICD. The cells were then subjected to 6 hours of treatment with proteasome inhibitor, and the RBP-Jk protein was detected using an immunoblot assay. Our studies demonstrated that the proteasome inhibitor induced a significant increase in the RBP-Jk protein level (Fig. 3B), which was reduced in the presence of AICD, but was not significantly restored by treatment with MG132 (Fig. 4B). However, we found a band shift of RBP-Jk: the upper band clearly remained in the presence of AICD (Fig. 4B). Several phosphorylation events might be required to induce a band shift to a new level. Thus, we attempted to determine whether the upper band was a phosphorylated form of RBP-Jk using the general phosphatase, CIP. In the same samples as shown in Fig. 4B, we determined that the upper band was downshifted in the presence of CIP, thereby suggesting that the two bands appear to correspond to the phosphorylation of RBP-Jk by an unknown kinase (Fig. 4C).

We then attempted to determine whether the lysosomal inhibitor exerted any effect on RBP-Jk degradation. Chloroquine is another lysosomotrophic agent, and is commonly used to inhibit lysosome function (Welman and Peters, 1977). Chloroquine is a weak base, which can disrupt lysosomal function by the blockade of acidification. We transfected HEK293 cells with FLAG–RBP-Jk and GFP–AICD, and the quantity of remaining RBP-Jk was analyzed after various periods of chloroquine treatment. The results of our studies revealed that chloroquine also induced an increase in RBP-Jk protein levels (Fig. 4D). The RBP-Jk protein levels were reduced in the presence of AICD, but were significantly restored after treatment with chloroquine (Fig. 4D) and NH₄Cl (Fig. 4E).

RBP-Jk is localized mainly in the nucleus, and the intracellular distribution of LAMP2 is in the lysosome. To determine the effect of siRNA encoding AICD or APP on RBP-Jk localization, cells were transfected with the respective vectors. The expression of AICD significantly facilitated accumulation of RBP-Jk in the LAMP2-positive lysosome. However, APP siRNA did not affect the cellular localization of RBP-Jk; however, the intensity of RBP-Jk fluorescence was moderately increased by transfection with APP siRNA compared with the control (Fig. 4F). These results demonstrated that the stability of the RBP-Jk protein is downregulated by AICD through the lysosomal-dependent pathway.

Our cumulative observations revealed that AICD activates GSK-3β in vitro (Ghosal et al., 2009) and in vivo (Ryan and Pimplikar, 2005; von Rotz et al., 2004) and that GSK-3β modulates Notch1 signaling and stability (Espinosa et al., 2003; Lai, 2002; Lee et al., 2008). Therefore, we attempted, using GSK-3β inhibitors and GSK-3β-knockout mouse embryonic fibroblast cells, to determine whether AICD downregulates Notch1 transcription activity and stability via GSK-3β. To confirm the role of lithium chloride in the regulation of GSK-3β, we introduced β-catenin, a specific target of GSK-3β. We found that β-catenin was accumulated in a dose-dependent manner upon treatment with lithium chloride (Fig. 5A). To evaluate any involvement of GSK-3β in the downregulation of the Notch1-IC protein by AICD, HEK293 cells were transfected with Myc–Notch1-IC and GFP–AICD with the GSK-3β inhibitors lithium chloride and SB216763. The results demonstrate that the downregulated level of Notch1-IC protein was not restored to a sufficient degree by the inhibition of GSK-3β (Fig. 5B,C). Additionally, the transcriptional activation of the Notch1-IC target gene was suppressed by cotransfection with AICD in GSK-3β wild-type mouse embryonic fibroblast cells and to a similar extent in GSK-3β-knockout cells, thus demonstrating the GSK-3β-independent negative regulation of Notch1 by AICD (Fig. 5D). These results show that the negative regulation of the transcriptional activity and protein stability of Notch1-IC by AICD occurs via a GSK-3β-independent pathway.

The phosphorylation of AICD at T668 by JNK3 regulates APP signaling by accumulation in the cytoplasm, amyloidogenic processing, and the destabilization of AICD (Chang et al., 2006; Colombo et al., 2009; Lee et al., 2003; Santos et al., 2010; Shin et al., 2007; Sodhi et al., 2008). We subsequently attempted to characterize the involvement of JNK3 in the regulation of Notch1 signaling by using a reporter assay. HEK293 cells were transfected with 4×CSL-Luc and either the Notch1 intracellular domain (Notch1-IC) or an empty vector. The expression of Notch1-IC was shown to induce significant activation of the 4×CSL reporter systems (Fig. 6A). Whereas Notch1-IC-mediated transcriptional activity was repressed in the AICD-expressing HEK293 cells, the AICD-induced suppression of Notch1 transcriptional activity was restored by the coexpression of JNK3, but not JNK1 (Fig. 6A). The Notch1-IC and RBP-Jk protein levels were reduced in the presence of AICD, but were restored to a moderate degree by coexpression with JNK3, but not with JNK1 (Fig. 6B).

We then evaluated the involvement of APP phosphorylation at T668 in the regulation of Notch1-IC transcriptional activity and protein stability using the T668A mutant. Whereas AICD inhibited Notch1-IC induced transcriptional activity and was restored by JNK3, AICD (T668A) did not prevent Notch1-IC-induced transcriptional activity (Fig. 6C). The Notch1-IC and RBP-Jk protein levels were reduced in the presence of AICD, but were not influenced by coexpression with AICD (T668A) (Fig. 6D). These results reveal that the JNK3-mediated phosphorylation of AICD is critically relevant to its ability to regulate Notch1 signaling.

Notch1-IC is ubiquitylated by the F-box protein Fbw7, which was initially isolated in a genetic screening process for negative regulators of Notch in C. elegans (Oberg et al., 2001; Wu et al., 2001). We anticipated that Fbw7 might function as a mediator for the negative regulation of Notch1-IC by AICD. Therefore, we evaluated the involvement of Fbw7 using the F-box-deleted, and hence the dominant-negative mutant form of Fbw7 (Fbw7ΔF). We also attempted to determine whether AICD could regulate the level of Notch1-IC ubiquitylation through Fbw7. Using dominant-negative Fbw7, the immunoblot analysis of the Notch1-IC precipitated with anti-Myc antibodies demonstrated that the levels of polyubiquitylated Notch1-IC were increased upon coexpression of AICD (Fig. 7A). The AICD-mediated upregulation of Notch1-IC ubiquitylation was inhibited by coexpression with Fbw7ΔF (Fig. 7A).

We subsequently evaluated the involvement of AICD in the physical association between Fbw7 and Notch1-IC in a coimmunoprecipitation experiment. HEK293 cells were cotransfected with vectors encoding Myc–Notch1-IC, FLAG–Fbw7 and GFP–AICD, and were then subjected to coimmunoprecipitation analysis (Fig. 7B). Immunoblot analysis using the anti-FLAG antibody on anti-Myc immunoprecipitates from the transfected cells showed that AICD facilitates the physical association between Fbw7 and Notch1-IC in the cells (Fig. 7B). These results demonstrate that the downregulation of the Notch1-IC protein by AICD occurs in an Fbw7-dependent pathway. At this point, we evaluated the formation of a trimeric complex between AICD and Notch1-IC or Fbw7, in an effort to define more precisely the role of AICD in the negative regulation of Notch1 signaling. We detected binding between Notch1-IC and AICD, but not between AICD and Fbw7, although the trimeric complex was detected in this case (Fig. 7C). Therefore, the results demonstrate that AICD interacts with Fbw7 in the presence of Notch1-IC, thereby forming a trimeric complex.

MyoD is basic helix-loop-helix transcription factor that has key but redundant roles in myogenesis (Li and Olson, 1992; Trouche et al., 1993). Notch1 signaling inhibits muscle cell differentiation through inhibition of MyoD expression (Anant et al., 1998; Baker and Schubiger, 1996; Kopan et al., 1994; Shawber et al., 1996). Because AICD suppressed the Notch1 signaling, we decided to examine the transcription activity of MyoD. As expected, Notch1-IC inhibited the expression of a MyoD promoter-luciferase reporter gene (MyoD-Luc) in C2C12 cells. We also found that AICD restored the suppression of MyoD reporter activity by Notch1-IC (Fig. 8A).

We next analyzed MyoD and Notch1-IC levels in the presence and absence of AICD at different time points during muscle differentiation in C2C12 cells. MyoD expression was detected in early muscle differentiation and gradually disappeared. Compared with the control, AICD ectopic expression modulated the expression pattern of MyoD proteins during muscle differentiation. Furthermore, AICD ectopic expression reduced the expression levels of Notch1-IC proteins during muscle differentiation (Fig. 8B). These results indicate that decreased Notch1-IC protein levels correlate with AICD and MyoD protein levels. Therefore, it is likely that the induction of AICD is involved in the regulation of Notch1 signaling.

In this study, we have demonstrated that AICD promotes the degradation of Notch1-IC and RBP-Jk by different pathways. AICD inhibits Notch1 transcription activity by dissociating the Notch1-IC–RBP-Jk complex. Furthermore, Notch1-IC is capable of forming a trimeric complex with Fbw7 and AICD; AICD thereby enhances the protein degradation of Notch1-IC in an Fbw7-dependent proteasomal pathway. AICD-mediated degradation is involved in the preferential degradation of the non-phosphorylated RBP-Jk through the lysosomal pathway.

In our recent report, we showed that expression of Notch1-IC downregulates transcriptional activity mediated by the AICD–Fe65–Tip60 (AFT) complex, ROS generation and cell death (Kim et al., 2007b). Our results also represent the functional crosstalk between Tip60 and Notch1 through acetylation (Kim et al., 2007a). Fe65 has recently been determined to be involved in the regulation of Notch1 signaling in an AICD- or Tip60-independent manner (our unpublished results). AICD harbors several internalization and trafficking motifs and might possess transcriptional activity that resembles the Notch1-IC of Notch1 (Baek et al., 2002; Cao and Sudhof, 2001; Gao and Pimplikar, 2001). AICD regulates phosphoinositide-mediated calcium signaling in vitro (Leissring et al., 2002) and also induces apoptosis and cytotoxicity in neurons (Lee et al., 2000). Previous reports have suggested the possibility of crosstalk between the Notch and APP signaling pathways, which would manifest as γ-secretase substrate competition (Berezovska et al., 2001; Lleo et al., 2003). However, we and other groups have demonstrated that negative crosstalk occurs between Notch and APP signaling, in a γ-secretase-independent fashion (Kim et al., 2007b; Petit et al., 2002). That is, Notch1-IC-mediated gene expression is regulated negatively by AICD, by some currently unknown mechanism (Roncarati et al., 2002). The functional involvement of AICD in Notch1 signaling, therefore remains a matter of some controversy. Our results demonstrate that Notch1-IC transcriptional activity is attenuated in the presence of AICD, which suggests that AICD is also involved in suppression of Notch1-IC transcriptional activity.

Several previous reports have suggested that Notch might directly interact with APP in intact cells (Fassa et al., 2005; Fischer et al., 2005; Kim et al., 2007b; Oh et al., 2005). We have determined that Notch1-IC interacts directly with AICD through the RAM–ANK domain. Intriguingly, our results demonstrate that the inhibitory mechanism suppresses the interaction of Notch1-IC and RBP-Jk, as a result of the downregulation of Notch1-IC and RBP-Jk protein stability. Several groups have demonstrated that Sel-10/Fbw7, through its WD40 domains, binds to phosphorylated Notch1-IC and mediates its ubiquitylation and subsequent rapid degradation (Gupta-Rossi et al., 2001; O'Neil et al., 2007; Wu et al., 2001). In this study, we determined that AICD stimulates the proteasomal degradation of Notch1-IC and the lysosomal degradation of RBP-Jk. AICD-mediated degradation is involved in the preferential degradation of non-phosphorylated RBP-Jk through the lysosomal pathway. Collectively, our findings demonstrate that the phosphorylation of RBP-Jk by unknown kinases regulates its half-life in either a positive or negative manner.

Thus AICD activates GSK-3β in vitro (Ghosal et al., 2009) and in vivo (Ryan and Pimplikar, 2005; von Rotz et al., 2004) and GSK-3β modulates Notch1 signaling and stability (Espinosa et al., 2003; Lai, 2002; Lee et al., 2008). Our results demonstrated that the AICD-mediated degradation of the Notch1-IC protein occurs independently of GSK-3β. The phosphorylation of AICD at T668 by JNK3 contributes to the neuronal degeneration inherent to Alzheimer's disease (AD) by regulating its translocation into the cytoplasm, amyloidogenic processing and destabilization of AICD (Chang et al., 2006; Colombo et al., 2009; Lee et al., 2003; Santos et al., 2010; Shin et al., 2007; Sodhi et al., 2008). For that reason, it was anticipated that JNK3 might function as a possible regulator for Notch1 signaling by the deregulation of AICD. The Notch1-IC transcriptional activity and protein levels were reduced in the presence of AICD, but were restored by coexpression with JNK3, but not with JNK1. The phosphorylation-deficient mutant of AICD (T668A) was not influenced by the regulation of Notch1-IC transcriptional activity and protein stability. Moreover, we determined that AICD negatively regulates the transcriptional activation of the Notch1-IC target genes and the stability of the Notch1-IC protein in an Fbw7- and proteasome-dependent manner. Notch1-IC, Fbw7 and AICD form a trimeric complex, and enhancement of the interaction occurring between Notch1-IC and Fbw7 might be a possible mechanism underlying the AICD-mediated proteasomal degradation of Notch1-IC (Fig. 9).

In summary, our results demonstrate that AICD performs the function of a negative regulator in Notch1 signaling by the promotion of Notch1-IC and RBP-Jk protein degradation. Henceforth, the findings of this study might begin to shed some light onto what may be a signal crosstalk mechanism of Notch1 and APP, or might point to the existence of γ-secretase-independent crosstalk.

HEK 293 and GSK-3β wild-type and GSK-3β-knockout mouse embryonic fibroblasts cells and mouse skeletal muscle C2C12 myoblast cells were cultured in Dulbecco's modified Eagle's medium (DMEM-Gibco) containing 10% fetal bovine serum and 1% penicillin-streptomycin, in a humidified incubator with an atmosphere containing 5% CO₂. The cultured cells were transiently transfected using calcium phosphate for HEK293, GSK-3β wild type and GSK-3β-knockout mouse embryonic fibroblast cells. Cells were grown to ~80% confluence and transfected with the plasmids (Chen and Okayama, 1987; Mo et al., 2007). C2C12 cells were grown to 80–90% confluence, and induced for differentiation by switching from growth medium to differentiation medium with DMEM containing 2% horse serum. Differentiation medium was replenished every day. The cultured cells were transiently transfected using the calcium phosphate method or Lipofectamine-plus reagent. For plasmid DNA transfection, the cells were grown to ~80% confluence and transfected with the plasmids (Chen and Okayama, 1987; Mo et al., 2007).

HEK293 cells were co-transfected with 4×CSL-Luc (a repeat section of the RBP-Jk target sequence, CGTGGGAA, with the luciferase gene) and β-galactosidase coupled with the indicated vector constructs. PMA was added 2 hours before the addition of DAPT, which was present for the final 6 hours. After 48 hours of transfection, the cells were lysed in chemiluminescent lysis buffer (18.3% of 1 M K₂HPO₄, 1.7% of 1 M KH₂PO₄, 1 mM phenylmethylsulfonyl fluoride and 1 mM dithiothreitol), and were analyzed using a Luminometer (Berthold). The luciferase reporter activity in each sample was normalized in relation to the β-galactosidase activity in the same lysate (Kim et al., 2007a).

The recombinant GST–AICD protein was expressed in Escherichia coli BL21 strain, using the pGEX system as indicated (Kim et al., 2007b). The GST fusion protein was then purified using glutathione–agarose beads (Sigma), in accordance with the manufacturer's instructions. An equal quantity of GST or GST–AICD fusion protein was incubated with lysates of HEK293 cells, which were transfected with combinations of expression vectors at 4°C on a rotator. The supernatants were subjected to immunoprecipitation with anti-FLAG antibody. After incubation for 1 hour, Protein-A–agarose was added, and the samples were incubated for 3 hours at 4°C on the rotator. After incubation, the beads were washed three times in ice-cold phosphate-buffered saline and boiled with 20 μl Laemmli sample buffer. The precipitates were then resolved by SDS-PAGE, and the immunoprecipitates were detected by immunoblotting with specific antibodies.

After 48 hours of transfection, the cultured cells were harvested and lysed in RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM PMSF, 1 mM DTT, and 2 g/ml each of leupeptin and aprotinin) for 30 minutes at 4°C. The cell lysates were then subjected to 20 minutes of centrifugation at 12,000 g at 4°C. The resultant soluble fraction was boiled in Laemmli buffer and subjected to SDS-PAGE. After gel electrophoresis, the separated proteins were transferred by electroblotting onto polyvinylidene difluoride (PVDF) membranes (Millipore). The membranes were then blocked with phosphate-buffered saline solution (pH 7.4) containing 0.1% Tween-20 and 5% non-fat milk. The blotted proteins were subsequently probed with anti-Myc antibody (9E10), anti-HA (12CA5) antibody, or anti-FLAG M2 antibody (Sigma), followed by incubation with anti-mouse horseradish-peroxidase-conjugated secondary antibodies (Amersham). The blots were then developed using enhanced chemiluminescence (ECL).

48 hours after transfection, the cells were lysed for 10 minutes in 1 ml of RIPA lysis buffer at room temperature. After 20 minutes of centrifugation at 12,000 g, the supernatants were subjected to immunoprecipitation with specific antibodies. After overnight incubation, Protein-A–agarose was added, and the samples were incubated for 3 hours at 4°C on the rotator. The beads were subsequently washed three times in ice-cold PBS, and any proteins that remained bound to the beads were eluted by boiling in 5× protein sample buffer. The samples were separated by SDS-PAGE, and visualized by immunoblotting.

Cells were treated with the proteasomal inhibitor MG-132 (Sigma), ALLN (Calbiochem) or the lysosomal inhibitors NH₄Cl (Sigma) and chloroquine (Sigma), or the translational inhibitor cycloheximide (Sigma). MG-132 was used at 0, 5 and 10 μM for 6 hours for the dosage assay of the proteasomal inhibitors. NH₄Cl was used at 0, 10, 20, and 50 mM for 6 hours for the dosage assay of lysosomal inhibitors. Chloroquine was used at 0, 50 and 200 μM for 6 hours for the dosage assay of lysosomal inhibitors. Cycloheximide was used at 1 mM for 0, 1, 2, 4 and 6 hours for the time-course assay of the translational inhibitors. Protein levels were analyzed by immunoblotting.

Half-life experiments using the cycloheximide-mediated inhibition of protein synthesis were conducted as previously described (Mo et al., 2007). Proteasome inhibitors were added 1 hour before cycloheximide treatment, and the cell lysates were subjected to SDS-PAGE and immunoblotting with the respective antibodies.

Assays were conducted as previously described with HEK293 cells plated at 1×10⁵ cells per well onto coverslips (Fisher). The cultured cells were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS), and then permeabilized with 0.1% Triton X-100 in PBS. Cells were blocked in 1% BSA in PBS. anti-LAMP2 antibody (Santa Cruz) and anti-RBP-Jk antibody (Santa Cruz) were used as the primary antibodies at a dilution of 1:100, washed three times in PBS. Mouse secondary antibodies conjugated to Alexa Fluor 488 or rabbit secondary antibodies conjugated to Alexa Fluor 546 (Invitrogen, 1:100) were added and the DNA dye ToPro3 was used for nuclear localization. The stained cells were evaluated for localization using confocal microscopy (LeicaTCS SPE). Each image is a single z section at the same cellular level. The final images were obtained and analyzed using confocal microscopy with LAS AF software (Leica).

We would like to thank Raphael Kopan (Washington University Medical School) for the Notch-related constructs, Thomas Südhof (University of Texas Southwestern) for providing us with the APP-related constructs, Yoo-Hun Suh (Seoul National University) for the AICD and AICD (T668A) constructs, Qubai Hu (Washington University) for the AICD construct and Bruce E. Clurman (Fred Hutchinson Cancer Research Center) for the Fbw7 construct. GSK-3β-knockout cells were kindly provided by James R. Woodgett (Ontario Cancer Institute). This study was supported by a grant from the Korea Healthcare Technology R&D Project, Ministry of Health & Welfare, Republic of Korea (A080441).