The γ-secretase complex, which consists of at least four proteins-Presenilin (PS), Nicastrin, Aph-1, and Pen-2, serves as a transmembrane aspartyl protease that plays a critical role in Alzheimer’s disease (AD) and Notch signaling pathway (Bai et al., 2015; De Strooper, 2003; Fortini, 2009). Amyloid βpeptides (Aβs), which are the main constituents of senile plaques present in the brain affected with AD, are generated by sequential cleavages of the β-amyloid precursor protein (APP) by β-secretase (BACE) and γ-secretase (Esler and Wolfe, 2001; Goedert, 2015). Another well-known substrate for γ-secretase is Notch, whose signaling controls a large number of cell fate decisions during development (Fortini, 2009; Louvi and Artavanis-Tsakonas, 2006). The cleavage of both APP and Notch within their single-pass transmembrane domains by γ-secretase releases cytosolic fragments and allows them to enter the nucleus, thereby regulating gene transcription (Brown et al., 2000; Cao and S?dhof, 2001; Louvi and Artavanis-Tsakonas, 2006; Wang et al., 2014; Figs. 1A and 1B). This γ-secretase-dependent regulated intramembrane proteolysis (RIP) is tightly controlled (Brown et al., 2000), and thus alterations in the RIP of APP and Notch result in developmental defects and diseases (Esler and Wolfe, 2001; Louvi and Artavanis-Tsakonas, 2006).

AD is the most common form of senile dementia characterized by the deposition of aggregation-prone Aβs and neurofibrillary tangles in the brain (Selkoe, 1998). Interestingly, early-onset familial Alzheimer’s disease (EOAD) is caused by rare mutations in APP, PS1, and PS2 (Tanzi, 2012). This is consistent with the fact that PS is the catalytic subunit of the γ-secretase complex whereas APP is a substrate for γ-secretase (De Strooper, 2003). In addition, γ-secretase activity and Aβ production were not detected in PS1/PS2 double-knockout cells (Herreman et al., 2000; Zhang et al., 2000). Therefore, γ-secretase serves as a therapeutic target to reduce Aβ production and development of γ-secretase inhibitors can provide an effective therapy for AD. However, there is still one major concern that γ-secretase inhibitors might cause deleterious side effects because they also affect the RIP of numerous other substrates including Notch (Esler and Wolfe, 2001; McCarthy et al., 2009). One possible approach to overcome this problem is to identify new therapeutic target genes that specifically affect the γ-secretase-dependent processing of APP rather than other substrates. To this end, we built a Drosophila reporter system for in vivo detection of γ-secretase activity by taking advantage of the Notch signaling pathway in the wing.

Canonical Notch signaling, which is highly conserved in the animal kingdom, is initiated by the binding of ligands including Delta, Serrate, and LAG-2 (Fortini, 2009; Louvi and Artavanis-Tsakonas, 2006). Ligand binding induces sequential cleavages of the Notch receptor by ADAM family metalloproteases and γ-secretase, releasing the Notch intracellular domain (NICD) (Fortini, 2009; Louvi and Artavanis-Tsakonas, 2006). NICD then enters the nucleus and forms an active transcriptional complex with the DNA-binding protein Suppressor of Hairless (Su(H)) and Mastermind (MAM), triggering expression of Notch target genes (Fortini, 2009; Louvi and Artavanis-Tsakonas, 2006; Fig. 1B). In the absence of Notch signaling, Su(H) proteins associate with various core-pressors to actively repress the transcription of Notch target genes (Fortini, 2009; Louvi and Artavanis-Tsakonas, 2006; Fig. 1B). These findings demonstrate that Su(H) is the key effector of Notch signaling.

The Drosophila homolog of human APP is known as β-amyloid protein precursor-like (APPL) (Martin-Morris and White, 1990). Flies homozygous for Appl deletion allele showed a fast phototaxis defect that can be restored by human APP as well as fly APPL, revealing their functional conservation (Luo et al., 1992). Furthermore, APPL is proteolytically cleaved by a BACE-like secretase (dBACE) and γ-secretase, producing a neurotoxic Aβ-like fragment in Drosophila (Carmine-Simmen et al., 2009). However, dBACE does not appear to recognize and cleave human APP at the β-secretase (BACE) cleavage site, suggesting the lack of a clear homolog of human BACE in Drosophila (Carmine-Simmen et al., 2009; Greeve et al., 2004). In contrast, each component and the activity of the γ-secretase complex appear to be evolutionarily conserved across invertebrates and vertebrates (De Strooper, 2003; Selkoe and Wolfe, 2007). In support of this claim, Drosophila γ-secretase can cleave human APP and thus produce Aβ peptides in combination with human BACE (Greeve et al., 2004).

Here, we have created a Drosophila reporter system for in vivo monitoring of γ-secretase activity and it provides a genetic tool to discover novel genes that modulate γ-secretase-mediated RIP of Notch and APP-like (APPL), and APPL metabolism.

Drosophila strains

The w¹¹¹⁸Drosophila strain was used as a wild-type control. The following flies were obtained from the Bloomington Stock Center: UAS-psn RNAi (#27681), UAS-p35 (#5073), UAS-psn (#8310), e16E-GAL4 (#30557), and N^Ax-E2 (#51660). The e16E-GAL4 enhancer trap line was shown to drive expression of a reporter gene in the striped pattern of engrailed during embryonic development (Weiss et al., 2001) and also in the posterior compartment of the wing imaginal disc during third instar larval stage (Johnson et al., 1995).

Expression constructs

The UAS-N™-SV and UAS-APPL-SV constructs were generated by PCR and/or restriction enzyme-based strategies and subcloned into the pUAST vector (Brand and Perrimon, 1993). For the Litmus 28-Su(H)-VP16 (SV), a NcoI-PstI fragment from pGEX-Su(H) (Bailey and Posakony, 1995), which contains the DNA-binding domain (amino acids 109?457; GenBank: AAF53434.1) of Su(H), and a PstI-XbaI fragment from pAct-GAL4-VP16 (Han and Manley, 1993), which contains the transcriptional activation domain (amino acids 2?79; GenBank: AIZ65950.1) of VP16, were subcloned into the Litmus 28 vector (New England Biolabs, Inc.). For the Litmus 28-N™-SV, a SalI-NcoI fragment from a Notch minigene clone (Wharton et al., 1985), which contains the N™ (amino acids 1710?1891; GenBank: AAF45848.2), and a BglII-SalI fragment from Litmus 28-N^ΔECN (Ju et al., 2000), which contains the signal peptide region (amino acids 1?60) of Notch, were subcloned into the Litmus 28-SV. For the UAS-N™-SV, a BglII-XbaI fragment from the Litmus 28-N™-SV was subcloned into the pUAST vector. For the UAS-APPL-SV construct, a EcoRI-NcoI fragment from GH04413, which encodes the full-length APPL (amino acids 1?887; GenBank: AAF45520), and a NcoI-XbaI fragment from the Limus 28-SV, were subcloned into the pUAST vector. These construct sequences were confirmed by DNA sequencing. Transgenic flies carrying these constructs were created according to standard D. melanogaster transformation procedures (KAIST transgenic fly service). Multiple independent lines for the same transgenic construct are indicated by superscripts. For example, N™-SV^61f and N™-SV^42m are independent insertion lines for the construct UAS-N™-SV.

Immunohistochemistry

Expression patterns of transgenes were visualized using anti-VP16 monoclonal antibody (14-5, Santa Cruz Biotech.) and 3,3′-diaminobenzidine reaction, as described previously (Jeong et al., 2012; Kim et al., 1995).

Preparation of Drosophila wings and phenotypic characterization

Using two pairs of forceps, the wings of female and male adult flies (younger than 3 days) were carefully cut off and subsequently were arranged in the same orientation on a glass slide. A coverslip was applied and each corner of the coverslip was sealed with a regular nail polish. In wild-type wing, the average L4 to L3 vein length ratio was 0.97 and the average L5 to L3 vein length ratio was 0.58 (see Supplementary Fig. S1A). The average percentages of L4 and L5 vein truncations were calculated by the following equations (see Supplementary Fig. S1B):

Statistical analysis

All quantitative data are presented as mean ± standard error of the mean (S.E.M.). Statistical significance was determined using unpaired t-test (*p < 0.05, **p < 0.01, and ***p < 0.001).

Generation of transgenic reporters for in vivo detection of the processing of Notch and APPL

We decided to take advantage of the Notch signaling pathway to generate transgenic reporters for in vivo detection of the activity of γ-secretase, which acts as a key player in Alzheimer’s disease (AD) (Esler and Wolfe, 2001; Selkoe, 1998). Notch is cleaved within its single-pass transmembrane domain by γ-secretase (Fortini, 2009; Louvi and Artavanis-Tsakonas, 2006). The cleavage of the Notch receptor releases its intracellular domain (NICD) and translocates it into the nucleus, thereby driving the expression of Notch target genes through the assembly of an active transcriptional complex with Su(H) and Mastermind (Fortini, 2009; Louvi and Artavanis-Tsakonas, 2006; Figs. 1A and 1B). Vein pattern formation in the Drosophila wing is controlled by Notch signaling (de Cellis, 1998). Reduction in Notch signaling activity leads to the formation of thicker veins, whereas increased Notch activity results in loss of veins (de Cellis and Garcia-Bellido, 1994a). These observations prompted us to examine whether Notch activity is correlated with vein phenotypes in a dose-dependent manner. Normal vein patterning was observed in wild-type females (Fig. 2A), whereas in females heterozygous for N^Ax-E2, a dominant gain-of-function (GOF) Abruptex (Ax) allele (de Cellis and Garcia-Bellido, 1994b), vein L5 consistently failed to reach the wing margin (5.4% L5 vein truncation; Figs. 2B and 2I). This Ax phenotype of vein L5 was significantly increased up to 26.1% vein truncation in females homozygous for N^Ax-E2 (p = 0.002, t-test; Figs. 2C and 2I). In addition, these homozygous mutants displayed a moderately truncated L4 vein (4.8% vein truncation; Figs. 2C and 2I). These findings suggest that the severity of Ax vein phenotype is proportional to Notch signaling activity. To further address whether γ-secretase activity is required for patterning of the Drosophila wing veins, we took advantage of the RNA interference (RNAi) technique. When the Drosophila presenilin (psn) gene was knocked down by overexpression of psn RNAi transgene using e16E-GAL4, which is expressed in the posterior compartment of the wing (Johnson et al., 1995), wing patterns showed robust vein thickening but no vein truncation phenotypes (Figs. 2D and 2I). This vein thickening phenotype, which is reminiscent of the phenotype found in Notch loss-of-function alleles (de Cellis and Garcia-Bellido, 1994a), was completely restored by coexpression of wild-type Psn (Fig. 2F). Overexpression of wild-type Psn alone hardly affected vein patterning (Figs. 2E and 2I). These results indicate target gene specificity of the psn RNAi transgene. Furthermore, knockdown of psn in the posterior compartment using e16E-GAL4 completely suppressed the vein truncation phenotype observed in males hemizygous for N^Ax-E2 (Figs. 2G and 2H). Taken together, these findings strongly suggest that γ-secretase activity contributes to vein pattern formation in the Drosophila wing through the Notch signaling pathway.

To monitor the proteolytic cleavage of Notch and APPL by γ-secretase within their transmembrane domains, we explored a sensitive reporter system, in which the chimeric transcriptional activator Su(H)-VP16 (SV) is fused to either a subfragment of Notch (N™) or full-length APPL (UAS-N™-SV and UAS-APPL-SV in Fig. 1A). The N™ subfragment possesses a transmembrane domain and the N-terminal region of NICD (Fig. 1A). The SV was created by fusion of the DNA-binding domain of Su(H) transcription factor (Bailey and Posakony, 1995) and the transcriptional activation domain of the virus VP16 (Sadowski et al., 1988) (Fig. 1A). Therefore, in case of nuclear translocation of the SV proteins, they are supposed to activate some of the Notch target genes, mimicking increased Notch pathway activity. Increased Notch activity in the wing can be visualized by the severity of the resulting vein phenotypes. Based on the fact that the N™ subfragment and APPL contain a γ-secretase-mediated cleavage site, we reasoned that the expression of both reporter transgenes under the control of e16E-GAL4 provides a sensitive assay for γ-secretase-dependent processing of Notch and APPL.

Expression of the notch- and APPL-based chimeric reporters in wings caused loss of wing veins

To determine whether both the N™-SV and APPL-SV chimeric reporter proteins transduce Notch signaling activity, we ectopically overexpressed these transgenes using e16E-GAL4 (Figs. 3E and 3H). Ectopic expression of N™-SV^61f in the posterior compartment of the wing resulted in moderate truncations of L4 and L5 veins in females (Fig. 3D). These vein truncation phenotypes were not observed in wild-type, e16E-GAL4/+, and N™-SV^61f/+ female wings (Figs. 3A?3C). Interestingly, in females expressing APPL-SV^85m, we observed severe truncations of L4 and L5 veins (Fig. 3G). However, vein pattern formation in females heterozygous for the APPL-SV^85m transgene was normal (Fig. 3F). These findings might suggest that both the N™-SV and APPL-SV reporter proteins were processed at least by γ-secretase, and they further demonstrate that the γ-secretase-cleaved C-terminal fragments of both reporters including the SV can transduce Notch activity in the wing.

To further address the difference in phenotypic severity between N™-SV and APPL-SV (Figs. 3D and 3G), we obtained a total of 5 independent transgenic lines for N™-SV and a total of 4 lines for APPL-SV. On average, ectopic expression in 5 different lines for N™-SV caused 43.1% L4 vein truncations and 43.8% L5 vein truncations (n = 84), while expression in 4 different lines for APPL-SV resulted in 90.4% L4 vein truncations and 96.8% L5 vein truncations (n = 62) (Fig. 3I). An almost 2-fold difference in the percentage of both vein truncations between N™-SV and APPL-SV (p < 0.001, t-test; Fig. 3I) strongly suggests that the N™-SV and APPL-SV proteins are differentially processed by γ-secretase and/or the γ-secretase-cleaved C-terminal portions of both reporter proteins generate different levels of Notch transducing activity. Another possibility is that the APPL-SV proteins are more abundantly expressed than the N™-SV proteins.

Reporter-mediated wing vein phenotypes are enhanced by Notch and suppressed by presenilin

Based on wing vein phenotypes, we deduced that the N™-SV and APPL-SV reporter proteins transduce Notch activity (Fig. 3). To further address this issue, we investigated genetic interactions among N™-SV, Notch, and psn. Ectopic expression of N™-SV^61f in the posterior compartment resulted in 38.3% L4 vein truncations and 33.6% L5 vein truncations (Figs. 4B and 4G). These N™-SV^61f vein phenotypes were significantly enhanced by one copy of N^Ax-E2 allele: from 38.3% to 69.2% for L4 vein truncations (p < 0.001, t-test) and from 33.6% to 41.0% for L5 vein truncations (p = 0.001, t-test) (Figs. 4C, 4D, and 4G). Similar patterns of genetic interactions were also observed with a different insertion line, N™-SV^42m (p < 0.001, t-test for both veins; Supplementary Figs. S2C, S2D, and S2G). Overexpression of N™-SV^42m in the wing led to 57.0% L4 vein truncations and 57.7% L5 vein truncations, which are greater than those obtained after ectopic expression of N™-SV^61f (Supplementary Figs. S2B and S2G). These findings support the claim that the N™-SV proteins transduce Notch activity in the wing. In contrast, the N™-SV^61f and N™-SV^42m vein truncation phenotypes were completely suppressed by RNAi-mediated knockdown of psn (Figs. 4E?4G and Supplementary Figs. S2E?S2G), indicating that Psn is required for the gain of N™-SV function in the wing. Interestingly, the psn RNAi-mediated vein thickening phenotype was also suppressed, but it was not completely suppressed by overexpression in either N™-SV transgenic line (Figs. 4E, 4F, Supplementary Figs. S2E and S2F). Therefore, knockdown of psn is epistatic to N™-SV overexpression with respect to vein pattern formation in the wing.

We next analyzed genetic interactions among APPL-SV, Notch, and psn. Females overexpressing APPL-SV^85m in the posterior compartment showed highly penetrant vein truncation defects: 84.8% L4 vein truncations and 90.0% L5 vein truncations (Figs. 5B and 5G). These APPL-SV^85m vein phenotypes were enhanced by one copy of N^Ax-E2 allele: from 84.8% to 94.7% for L4 vein truncations (p = 0.001, t-test) and from 90.0% to 98.2% for L5 vein truncations (p = 0.003, t-test), suggesting that the APPL-SV proteins can act as a Notch signal transducer in the wing (Figs. 5C, 5D, and 5G). Similar patterns of genetic interactions for both L4 and L5 vein phenotypes were not observed with the APPL-SV^82m transgene, a different insertion line (p > 0.3, t-test; Supplementary Figs. S3C, S3D, and S3G), probably due to high penetrance of vein truncation defects in females overexpressing APPL-SV^82m alone (92.8% for L4 veins and 100% for L5 veins; Supplementary Fig. S3G). In addition, we found that the APPL-SV^85m vein truncation phenotypes were robustly suppressed by RNAi-mediated knockdown of psn: from 84.8% to 51.2% for L4 vein truncations (p<0.001, t-test) and from 90.0% to 1.6% for L5 vein truncations (p < 0.001, t-test) (Figs. 5E?5G). Partial suppression especially in L4 veins, which is different from that observed in females coexpressing psn RNAi and N™-SV (Figs. 4F and 4G), might indicate that APPL-SV-induced Notch activity antagonizes the knockdown effects of psn to a greater extent than N™-SV. Furthermore, similar patterns of suppressive genetic interactions were also observed with the APPL-SV^82m transgene (p < 0.001, t-test for both L4 and L5 vein truncations; Supplementary Figs. S3E?S3G), demonstrating the requirement of Psn for the gain of APPL-SV function in the wing. Females coexpressing psn RNAi and APPL-SV exhibited large reduction in vein thickening phenotype compared to females overexpressing psn RNAi alone (Figs. 5E and 5F, Supplementary S3E and S3F), which is indicative of the antagonistic relationships between knockdown of psn and APPL-SV overexpression in patterning of the wing veins.

The role of apoptosis in reporter-induced wing vein truncations

Since Notch plays a role in apoptosis during Drosophila visual system development (Bertet et al., 2014; Brachmann and Cagan, 2003), we investigated whether Notch-mediated regulation of apoptosis is involved in patterning of the wing veins. Interestingly, females overexpressing the baculovirus pan-caspase inhibitor p35, which has been shown to prevent apoptotic death in Drosophila (Hay et al., 1994), exhibited a mild phenotype only in L5 vein, indicating a positive, but a minor role of apoptosis in vein pattern formation (3.5% vein truncation; Figs. 6D and 6G). In females expressing both p35 and N™-SV^61f in the posterior compartment, we observed a small but significant increase in vein phenotypes compared to females overexpressing N™-SV^61f: from 38.3% to 51.1% for L4 vein truncations (p < 0.001, t-test) and from 33.6% to 46.1% for L5 vein truncations (p < 0.001, t-test) (Figs. 6B, 6E and 6G). Similar enhancement in vein phenotype was also found in females coexpressing p35 and N™-SV^42m in the posterior compartment: from 57.0% to 60.5% for L4 vein truncations (p = 0.03, t-test) and from 57.7% to 76.6% for L5 vein truncations (p < 0.001, t-test) (Supplementary Figs. S4B, S4E, and S4G). Taken together, this seemingly additive genetic interaction between p35 and N™-SV suggests that N™-SV-induced vein truncation is not due to apoptosis.

To examine whether APPL-SV-induced vein phenotypes result from apoptosis, we coexpressed p35 and APPL-SV^85m in the posterior compartment of the wing. Unexpectedly, we observed that coexpression of p35 and APPL-SV^85m robustly suppressed L4 vein truncation phenotypes of APPL-SV^85m from 84.8% to 56.8% (p < 0.001, t-test), whereas coexpression of p35 and APPL-SV^85m enhanced L5 vein truncation phenotypes from 90.0% to 97.0% (p = 0.003, t-test) (Figs. 6C, 6F, and 6G). Suppressive genetic interactions were also observed in L4 veins, but not in L5 veins with the APPL-SV^82m transgene: from 92.8% to 54.8% for L4 vein truncations (p < 0.001, t-test) and no change was observed in L5 vein truncations (Supplementary Figs. S4C, S4F, and S4G). Strong suppressive interactions between p35 and APPL-SV, especially in L4 veins suggest that overexpression of APPL-SV leads to cell death, and they further demonstrate that APPL-SV-induced cell death partly contributes to L4 vein truncation phenotypes. However, phenotypic enhancement observed in L5 veins, which is similar to that observed in females coexpressing p35 and N™-SV (Fig. 6G), suggests that apoptosis is not attributable to L5 vein truncation phenotypes of APPL-SV.

Drosophila melanogaster provides a powerful genetic tool to analyze gene function associated with human diseases including Alzheimer’s disease (AD) and evolutionarily conserved signaling pathways (Bier, 2005; Lu and Vogel, 2009). Indeed, many different Drosophila models of AD, which showed progressive degeneration through elevated production of amyloid β-peptides (Aβs), were developed (Fernandez-Funez et al., 2015). Some of these AD models are based on overexpression of wild-type human APP, EOAD-associated mutant APP, or a secreted form of human Aβ42 peptide (Fernandez-Funez et al., 2015). These findings demonstrate highly conserved mechanisms underlying Aβ-mediated neurotoxicity and neurodegeneration between Drosophila and humans. This is further supported by the fact that the fly genome contains functional homologs for most of the EOAD-associated genes including APP, PS1, and PS2 (Martin-Morris and White, 1990; Stempfle et al., 2010). In combination with human BACE, Drosophila γ-secretase is able to cleave human APP to generate Aβ peptides, indicating its conserved intramembrane proteolytic activity (Greeve et al., 2004). Although the Aβ sequence is not well conserved among APP homologs from different species, a growing body of evidence suggests that all APP homologs play an important and conserved role in neural development including neurite growth, axon guidance, and synaptogenesis (Nicolas and Hassan, 2014). However, little is known about how γ-secretase-mediated proteolytic processing of APP is regulated at the molecular and genetic levels.

One of the hallmarks of AD is senile plaques largely consisting of Aβs, which are produced by sequential cleavages of APP by β-secretase (BACE) and γ-secretase (Esler and Wolfe, 2001; Goedert, 2015). Given that all known mutations responsible for early-onset familial Alzheimer’s disease (EOAD) are localized in APP, PS1, and PS2 (Tanzi, 2012), γ-secretase represents a good therapeutic target for developing Aβ-lowering drugs. Therefore, brain-penetrant inhibitors of γ-secretase are emerging as one of the most effective therapies for AD (De Strooper et al., 2010). However, toxicity of γ-secretase inhibitors was found to be a major hurdle in preclinical studies since γ-secretase is also required for the RIP of numerous other substrates including Notch (De Strooper et al., 2010; Esler and Wolfe, 2001; McCarthy et al., 2009). Here, as an alternative approach to overcome this target-based toxicity, we report the development of a Drosophila reporter system for in vivo detection of γ-secretase activity. Based on several observations, we expect that this in vivo reporter system will provide a genetic tool to discover novel genes, which specifically modulate γ-secretase-mediated processing of Notch and APPL. First, overexpression of the N™-SV reporter proteins in the wing induced vein truncation phenotypes that are totally dependent on γ-secretase activity. Second, not only γ-secretase activity, but also apoptosis contributes to the vein truncation phenotypes of the APPL-SV reporter transgene. This appears to be consistent with previous observations suggesting that human APP induces APP intracellular domain (AICD)-dependent cell death in Drosophila (Gunawardena and Goldstein, 2001; Wang et al., 2014). Since the γ-secretase-cleaved C-terminal fragment of APPL-SV contains the APPL intracellular domain (dAICD) in addition to the SV, dAICD might be responsible for the contribution of apoptosis. Third, both N™-SV and APPL-SV transduce Notch signaling activity in a dose-dependent manner. In support of this claim, reporter-induced vein phenotypes are enhanced by one copy of N^Ax-E2, which is a Notch GOF allele. Fourth, APPL-SV-induced phenotype is almost 2-fold greater than the GOF phenotype of N™-SV, probably due to differences in their RIP and the contribution of dAICD-dependent apoptosis.

Several γ-secretase associated proteins (GSAPs), in addition to the four main components of the γ-secretase complex, have been discovered (Chen et al., 2006; He et al., 2010; Teranishi et al., 2015; Wakabayashi et al., 2009; Zhou et al., 2005). These GSAP proteins, including TMP21, pigeon homologue protein, and proton myo-inositol cotransporter, regulate substrate selectivity and Aβ production. Therefore, these GSAPs can serve as therapeutic targets for the treatment of AD. These observations might indicate the existence of unknown GSAP proteins regulating Aβ production without affecting proteolytic cleavage of other γ-secretase substrates including Notch. Future work using the in vivo reporter system developed in this study will determine whether known and additional unknown GSAP proteins are required for modulation of γ-secretase activity, and how substrate specificity of γ-secretase is regulated. Our current reporter system can also be applied to screening of chemical compound libraries.