Semagacestat Is a Pseudo-Inhibitor of γ-Secretase Shinji Tagami, Kanta Yanagida, Takashi S. Kodama, Mako Takami, Naoki Mizuta, Hiroshi Oyama, Kouhei Nishitomi, Yu-wen Chiu, Toru Okamoto, Takeshi Ikeuchi, Gaku Sakaguchi, Takashi Kudo, Yoshiharu Matsuura, Akio Fukumori, Masatoshi Takeda, Yasuo Ihara, Masayasu Okochi  Cell Reports  Volume 21, Issue 1, Pages 259-273 (October 2017) DOI: 10.1016/j.celrep.2017.09.032 Copyright © 2017 The Authors Terms and Conditions

Cell Reports 2017 21, 259-273DOI: (10.1016/j.celrep.2017.09.032) Copyright © 2017 The Authors Terms and Conditions

Figure 1 Extracellular and Intracellular Aβ Levels in iPSC-Derived Human Neuronal Cells with or without Semagacestat Treatment (A) Immunohistochemistry of iPSCs generated from fibroblasts of a healthy 40-year-old woman, neural stem cells (NSCs) induced from the iPSCs, and neuronal cells differentiated from the NSCs (see details in Supplemental Experimental Procedures). (B) Levels of secreted Aβ in the conditioned medium of the neuronal cells following semagacestat treatment until 24 hr. Red squares and blue diamond indicate semagacestat and DMSO treatment, respectively. As is widely known, semagacestat decreases extracellular Aβ secretion. However, when using human neuronal cells, a 1-hr incubation was insufficient for observing a good enough size of the effect on Aβ secretion in the medium (see also Figure 3A), although it is sufficient for observing the effects on intracellular γ-byproducts (see below). (C) Aβ in the neuronal cells treated with or without semagacestat for 1 hr was immunoprecipitated (using 4G8) and then separated on a 10%–20% Tris-tricine gradient gel. ECL-prime chemiluminescence was used to detect the bands (using 82E1). In each lane, 2.6 mg of cell lysate protein were used for immunoprecipitation. By a simple assessment method (Figures S1F and S1G), it was found that neuronal cells contained 10 ± 3.2 fmol/dish or 2.3 ± 2.1 fmol/dish of total Aβ, upon treatment with or without semagacestat, respectively. (D) Immunoprecipitated Aβ species from cells treated with or without semagacestat for 1 hr used in (C) were separated on a Tris-tricine 8 M urea gel-1 (Qi-Takahara et al., 2005). Upon semagacestat treatment, the neuronal cells contained 2.4 ± 1.1 fmol/dish of Aβ40, 3.8 ± 1.1 fmol/dish of Aβ42+Aβ43, and 3.5 ± 1.5 fmol/dish of Aβ45+Aβ46. The asterisk indicates βAPP-CTF derivatives other than Aβ species (see Figure S2C). Note that Aβ47–49 species were undetectable on Tris-tricine 8 M urea gel-2 analysis (Figure S1D). Cell Reports 2017 21, 259-273DOI: (10.1016/j.celrep.2017.09.032) Copyright © 2017 The Authors Terms and Conditions

Figure 2 γ-Byproducts in HEK Cells with or without Expression of Exogenous sw βAPP or Endogenous PS1/2 Black and red asterisks indicate a significant increase and decrease, respectively (p < 0.01 by Welch’s t test). Error bars represent the SEM. (A) Theoretical small βAPP-derived peptides of 3 to 6 amino acid residues, which we measured in this study. The γ-byproducts are derived from serial γ-cleavage between the γ- and ε-sites (arrowheads) of βAPP. Cleavage starts at the ε-sites (Takami et al., 2009), and cleavage at the γ-sites eventually releases Aβ from the membrane. (B) Direct immunoblotting (for PS1/2, βAPP, and β-actin) and immunoprecipitation and immunoblotting (for Aβ) analysis of HEK cells with or without the stable expression of exogenous sw βAPP or endogenous PS1/2. Exogenous sw βAPP expression increased the βAPP-holo protein, βAPP-CTF stubs, and Aβ. Deletion of PS1/2 in sw βAPP/HEK cells further increased βAPP-CTF but led to the elimination of both intracellular and extracellular Aβ. (C) Immunoprecipitation and immunoblotting analysis for AICD in sw βAPP/HEK cells with or without endogenous PS1/2. (D) Levels of the tri-, tetra-, penta-, and hexa-peptides that can theoretically be produced from cleavage between the γ- and ε-cleavage sites of βAPP, measured in sw βAPP/HEK cells with (blue) or without (gray) endogenous PS1/2 expression. The y axis (fmol/dish) represents the peptide levels. The peptide levels (fmol/dish) in sw βAPP/HEK cells (values of the y axis) in (D) and (E) differed, because the total number of cells in the dishes could not be made exactly equal in each set of experiments. (E) Levels of the small peptides inside HEK cells with (blue) or without (gray) stable exogenous sw βAPP expression. See also Figure S2. Cell Reports 2017 21, 259-273DOI: (10.1016/j.celrep.2017.09.032) Copyright © 2017 The Authors Terms and Conditions

Figure 3 Extracellular and Intracellular Aβ and γ-Byproducts in sw βAPP/HEK Cells with or without Semagacestat Treatment (A) Levels of secreted Aβ in the conditioned medium of cultured cells following semagacestat treatment. iPSC-derived human neuronal cells, sw βAPP/HEK cells, and SH-SY5Y neuroblastoma cells were cultured with or without semagacestat, and the levels of secreted Aβ at 1 hr, 4 hr, 8 hr, and 24 hr were measured by ELISA. (B) Aβ in the sw βAPP/HEK cells treated with or without semagacestat for 1 hr. (C) Immunoprecipitated Aβ species from cells treated with or without semagacestat for 1 hr used in (B). The asterisk indicates βAPP-CTF derivatives other than Aβ species (see also Figure S2C). (D) Levels of the small βAPP-derived peptides inside the sw βAPP/HEK cells with or without semagacestat treatment for 1 hr. (E) Levels of the small peptides inside the sw βAPP/HEK cells and PS1/2-double KO sw βAPP/HEK cells (clone 2411). (F) Levels of the small peptides inside PS1/2-double KO sw βAPP/HEK cells (clone 2411) with or without semagacestat treatment for 1 hr. See also Figure S3. Cell Reports 2017 21, 259-273DOI: (10.1016/j.celrep.2017.09.032) Copyright © 2017 The Authors Terms and Conditions

Figure 4 γ-Byproducts in MEF and Human Neuronal Cells with or without Semagacestat Treatment MEF cells in a 10-cm dish usually contain approximately 1.2 mg of protein for each cell lysate (using our extraction method). (A–C) The absolute levels (per dish) of the γ-byproducts after correcting the values by the protein levels (1.2 mg) are shown. Black asterisks indicate a significant increase (p < 0.01 by Welch’s t test). Error bars represent the SEM of 4 measurements. (A) Levels of the small βAPP-derived peptides inside MEFs with or without semagacestat treatment for 1 hr are shown. (B) Levels of the peptides inside PS1/2-double KO MEFs with or without semagacestat treatment for 1 hr are shown. Note that intracellular VVI in PS1/2-double KO MEFs was not increased by semagacestat. (C) Levels of the peptides inside βAPP KO MEFs with or without semagacestat treatment for 1 hr are shown. Note that semagacestat also did not increase the γ-byproducts in βAPP KO MEFs. However, the level of intracellular VVI peptide in βAPP KO MEFs was increased by semagacestat. Combined with (A) and (B), the data suggest that VVI can be produced by PS/γ-secretase cleavage of the transmembrane domain of any of its substrates other than βAPP that is endogenously expressed in the tested cells (i.e., MEF, HEK, and human neuronal cells). (D) Levels of the peptides in iPSC-derived human neuronal cells treated with or without semagacestat for 1 hr. See also Figure S4. Cell Reports 2017 21, 259-273DOI: (10.1016/j.celrep.2017.09.032) Copyright © 2017 The Authors Terms and Conditions

Figure 5 γ-Byproducts Produced in the In Vitro Reconstitution and Cell-free γ-Secretase Assay Black and red asterisks indicate a significant increase and decrease, respectively (p < 0.01 by Welch’s t test). Error bars represent the SEM of 4 measurements. (A) Relative levels of the small βAPP-derived peptides after the in vitro assay with or without semagacestat post-treatment. (B) Relative levels of the small peptides after the cell-free assay with or without semagacestat post-treatment. (C) Relative levels of the small peptides after the cell-free assay with or without semagacestat pretreatment. Note that semagacestat was added to the cells 1 hr prior to the cell collection and then crude membrane fraction was obtained. (D) Relative levels of de novo peptides in membrane fraction (membrane) or in supernatant fraction (soluble) after the cell-free assay with or without semagacestat post-treatment. (E) Relative levels of de novo peptides in membrane fraction (membrane) or in supernatant fraction (soluble) after the cell-free assay with or without semagacestat pretreatment for 1 hr. Cell Reports 2017 21, 259-273DOI: (10.1016/j.celrep.2017.09.032) Copyright © 2017 The Authors Terms and Conditions

Figure 6 Aβ and γ-Byproducts in Brains of Transgenic Mice with or without Semagacestat Treatment Four mice in each set of treatment were analyzed. (A) Aβ in the brains treated with or without semagacestat was immunoprecipitated (using 4G8) and then separated on a Tris-tricine 8 M urea gel-1. In each lane, 2.4 mg of brain protein was used for immunoprecipitation. Data for two out of four mice are shown. ECL-prime chemiluminescence was used to detect the bands (using 82E1). Short-exposure images (left panels) and long-exposure images (right panels) are shown. (B) Levels of the small peptides (fmol/10 mg protein) in the transgenic mouse brains with or without semagacestat treatment. We detected relatively large amounts of GVV and GGVV, possible degradation products of Aβ40 (Figure S5). Black asterisks indicate a significant increase (p < 0.01 by Welch’s t test). Error bars represent the SEM of 4 measurements. See also Figure S5. Cell Reports 2017 21, 259-273DOI: (10.1016/j.celrep.2017.09.032) Copyright © 2017 The Authors Terms and Conditions

Figure 7 Effects of Other Potential GSIs on the γ-Byproducts and Aβ Species Black and red asterisks indicate a significant increase and decrease, respectively (p < 0.01 by Welch’s t test). Purple, green, and yellow asterisks indicate a significant increase (p < 0.01 by Welch’s t test). Error bars represent the SEM of 4 measurements. (A–C) Effects of other potential non-TSA GSIs on the γ-byproducts and Aβ species in human neuronal cells. (A) Levels of secreted Aβ in the conditioned medium of cultured cells following treatment with each of the indicated potential GSIs (versus DMSO alone) until 24 hr are shown. The levels of secreted Aβ were measured by ELISA. (B) Immunoprecipitation/immunoblotting of intracellular Aβ species after treatment with the indicated compounds for 1 hr is shown. The asterisk indicates degradation products of βAPP-CTF stubs other than Aβ species (see Figure S2C). In each lane, 2.4 mg of cell lysate protein was used for immunoprecipitation. By a simple assessment method (Figures S1F and S1G), avagacestat-treated neuronal cells were shown to contain 0.59 ± 0.4 fmol/dish of Aβ40, 3.1 ± 0.66 fmol/dish of Aβ42+Aβ43, and 9.1 ± 3.2 fmol/dish of Aβ45+Aβ46. RO4979097-treated cells contained 0.72 ± 0.34 fmol/dish of Aβ40, 4.8 ± 1.4 fmol/dish of Aβ42+Aβ43, and 9.6 ± 2.3 fmol/dish of Aβ45+Aβ46. MK-0752-treated cells contained 5.2 ± 1.9 fmol/dish of Aβ40, 5.0 ± 1.7 fmol/dish of Aβ42+Aβ43, and 11 ± 2.1 fmol/dish of Aβ45+Aβ46. (C) Levels of the small βAPP-derived peptides after treatment for 1 hr with each of the indicated potential GSIs (versus DMSO alone) are shown. (D and E) Effects of L685,458, a TSA GSI, on the γ-byproducts and Aβ species in sw βAPP/HEK cells. (D) Levels of the small βAPP-derived peptides after treatment for 1 hr with L685,458 or semagacestat (versus DMSO alone) are shown. (E) Immunoprecipitation/immunoblotting of intracellular Aβ species after L685,458 treatment of cells for 1 hr is shown. Immunoprecipitation was performed using 4G8. Aβ and βAPP-CTF bands (upper panel), the Aβ40 band (middle panel), and the Aβ43 band (lower panel) were detected by immunoblotting using antibodies 82E1, 18580, and 18583, respectively. Using a simple assessment method (Figures S1F and S1G), without L685,458 treatment, the sw βAPP/HEK cells were found to contain 8.7 ± 2.1 fmol/dish of Aβ40, 2.8 ± 1.9 fmol/dish of Aβ42+Aβ43, and 2.2 ± 0.5 fmol/dish of Aβ45+Aβ46. The asterisk indicates degradation products of βAPP-CTF stubs other than Aβ species (see Figure S2C). In each lane, 1.2 mg of cell lysate protein was used for immunoprecipitation. See also Figure S6. Cell Reports 2017 21, 259-273DOI: (10.1016/j.celrep.2017.09.032) Copyright © 2017 The Authors Terms and Conditions