Phosphoinositides Suppress γ-Secretase in Both the Detergent-soluble and -insoluble States*

γ-Secretase is an aspartic protease that hydrolyzes type I membrane proteins within the hydrophobic environment of the lipid bilayer. Using the CHAPSO-solubilized γ-secretase assay system, we previously found that γ-secretase activity was sensitive to the concentrations of detergent and phosphatidylcholine. This strongly suggests that the composition of the lipid bilayer has a significant impact on the activity of γ-secretase. Recently, level of secreted β-amyloid protein was reported to be attenuated by increasing levels of phosphatidylinositol 4,5-diphosphate (PI(4,5)P2) in cultured cells. However, it is not clear whether PI(4,5)P2 has a direct effect on γ-secretase activity. In this study, we found that phosphoinositides directly inhibited CHAPSO-solubilized γ-secretase activity. Interestingly, neither phosphatidylinositol nor inositol triphosphate altered γ-secretase activity. PI(4,5)P2 was also found to inhibit γ-secretase activity in CHAPSO-insoluble membrane microdomains (rafts). Kinetic analysis of β-amyloid protein production in the presence of PI(4,5)P2 suggested a competitive inhibition. Even though phosphoinositides are minor phospholipids of the membrane, the concentration of PI(4,5)P2 within the intact membrane has been reported to be in the range of 4–8 mm. The presence of PI(4,5)P2-rich rafts in the membrane has been reported in a range of cell types. Furthermore, γ-secretase is enriched in rafts. Taking these data together, we propose that phosphoinositides potentially regulate γ-secretase activity by suppressing its association with the substrate.

observation points to the possibility that a change in the composition of the lipid bilayer could have a significant impact on the enzymatic activity of ␥-secretase. The level of secreted A␤ from cultured cells was reported to be attenuated by increasing levels of phosphatidylinositol 4,5-diphosphate [PI(4,5)P 2 ], one of the phosphoinositides (19). The phosphoinositides play pivotal roles in numerous biological processes, such as ion channel regulation, membrane trafficking, cell polarity, and actin rearrangement (20 -25).
PI(4,5)P 2 is estimated as 0.3-2.0% of the total cellular lipid. McLaughlin et al. (26) estimated that the intracellular concentration of PI(4,5)P 2 , if uniformly distributed inside the cell, was in the range of 2-30 M, based on the dissociation constants between PH domains and PI(4,5)P 2 . Bunce et al. (27) reported that the concentration of PI(4,5)P 2 was in the range of 32-159 M in several cell species. Even though the phosphoinositides are a minor component of cellular lipids, one can assume that their concentrations in the two-dimensional intact membrane would be higher than that reported for the total three-dimensional cell volume. If the volume of the membrane is estimated as ϳ10 -20% of the cell volume, the concentrations of PI(4,5)P 2 in the membrane are at least five times higher than those in the total cell volume. The local concentration of PI(4,5)P 2 at the inner leaflet of neutrophil membrane was reported to be ϳ5 mM in the steady state (28). Further, Sheetz et al. (29,30) showed that the concentration of PI(4,5)P 2 was 4 -8 mM in a 50-Å area of the inner leaflet of erythrocyte membrane. In addition, a number of reports showed that PI(4,5)P 2 localizes in the detergent-insoluble microdomains (rafts) of the membrane (31)(32)(33). It has been proposed that there is a spatially confined pool of PI(4,5)P 2 in the membrane (34 -37). Thus it is reasonable to consider that the concentration of PI(4,5)P 2 in the microdomains of the membrane is much higher than previously thought. ␥-Secretase is also enriched in lipid raft microdomains (38 -40). It is likely that the phosphoinositides and ␥-secretase localize in the same membrane microdomains.
Furthermore, phosphoinositides were widely known to be modulated by their concentrations in membrane by extracellular stimulus in physiological condition (41). Studies of several PH domains fused with green fluorescent protein revealed that concentrations of PI(3,4)P 2 and PI(3,4,5)P 3 were elevated after stimulus in Dictyostelium cells and neutrophils in vivo (21,22,(42)(43)(44). In neutrophils, it has been reported that the local concentration of PI (3,4,5)P 3 at the inner leaflet of the plasma membrane is 5 M and that after extracellular stimulation it increases to 200 M (28). The concentration of PI(3,4)P 2 is estimated to increase from 10 -20 to 100 -200 M upon stimulation (28). Winks et al. (45) reported that concentration of PI(4,5)P 2 increased from 192-381 to 417-1153 M after expression of PI5K in superior cervical ganglia. Those observations suggest that concentration of phosphoinositides can be modulated in physiological conditions. Recently, increasing phosphoinositide (PI(4,5)P 2 ) levels alter A␤ production by cultured cells (19), implying a cross-talk between phosphoinositides and ␥-secretase. Here we examined whether there are direct effects of phosphoinositides on ␥-secretase in both CHAPSO-soluble and -insoluble states.
Preparation of Notch Substrate-For assessment of ␥-secretase-dependent Notch S3 cleavage in the presence of phosphoinositides, we expressed an artificial Notch substrate with shortened intracellular domain (designated as ⌬E Notch-FLAG) in Sf9 cells (48). Isolated Notch substrate was incubated with the CHAPSO lysate at 37°C for 4 h (15) (supplemental Fig.  S1A). ␥-Secretase-dependent S3 cleavage was visualized by detecting shortened Notch intracellular domain fused with FLAG tag (sNICD-FLAG) with ANTI-FLAG M2 monoclonal antibody (Sigma) (supplemental Fig. S1B). We confirmed that ⌬E Notch-FLAG was cleaved at the bona fide S3 cleavage site by ␥-secretase (see supplemental Fig. S1, C and D).
Isolation of CHAPSO-insoluble Rafts-The CHAPSO-insoluble fraction was obtained as described previously (40). The T-Rex-CHO stable cell line was cultured in the presence of tetracycline to induce expression of C99 (46). Microsomal fractions of the cells were homogenized in five volumes of 10% sucrose in MES-buffered saline (25 mM MES, pH 6.5, and 150 mM NaCl) containing 1% CHAPSO. After adjusting the sucrose concentration to 40%, the homogenate was centrifuged on a discontinuous sucrose gradient (5, 35, and 40%) at 39,000 rpm for 20 h at 4°C on an SW 41 Ti rotor (Beckman). The interface between 5 and 35% sucrose was collected and designated as CHAPSO-insoluble rafts. The CHAPSO-insoluble rafts were diluted with three volumes of dilution buffer (20 mM PIPES, pH 7.2, 140 mM KCl, 250 mM sucrose, 5 mM EGTA) and incubated for 45 min at 37°C in the presence or absence of PI(4,5)P 2 . For assessing A␤ production from exogenously added C99-FLAG, the CHAPSO insoluble rafts that were obtained from cells grown in the absence of tetracycline were incubated for 60 min with 100 nM C99-FLAG in the presence or absence of PI(4,5)P 2 .
Treatment of PLC Inhibitor-7WD10, CHO cells expressing APP751 were cultured in Dulbecco's modified Eagle's medium (Sigma) containing 10% fetal bovine serum (Invitrogen) and 200 g/ml G418 (49). The cells were treated with phosphatidylinositol specific PLC inhibitor, edelfosin (Calbiochem) at a concentration of 15 M in the absence of G418 for 6 h. Microsomal fraction of the cells were prepared as described previously (15). 500 l of the microsomal fraction (2.5 mg/ml protein concentration) was mixed with 1 ml of MeOH:CHCl 3 (2:1) and centrifuged at 15,000 rpm for 5 min at 4°C. Resultant pellet was mixed with the same solvent to complete neutral lipids extraction. 750 l of MeOH: CHCl 3 :12 N HCl (40:80:1) was added to the pellet and mixed for extraction of acidic lipids. Supernatant was transferred to a new 1.5-ml tube and mixed with 250 l of CHCl 3 and 450 l of 0.1 N HCl. After centrifugation, the organic phase was transferred to a new tube and dried up. The dried lipid sample was reconstituted with 80 l of CHCl 3 : MeOH:H 2 O (1:2:0.8) and spotted onto PI(4,5)P 2 Mass Strip (Echelon). PI(4,5)P 2 in extracted lipid sample was detected with PLC-␦1 PH domain glutathione S-transferasetagged protein (Echelon). Miocrosomal fraction prepared from cells treated with edelfosin was incubated at 37°C for 0.5 h in the presence of 15 M edelfosin and subjected to Western blot to assess effect of edelfosin on A␤ production from isolated membrane (10,13).
Immunoprecipitation-␥-Secretase complex was immunoprecipitated with anti-nicastrin polyclonal antibody (Sigma), as previously described (15). After thorough washing, the ␥-secretase complex bound to protein A-Sepharose beads was incubated in FIGURE 1. Effect of phosphatidylcholine and phosphatidylinositol 4,5-diphosphate on A␤ production. Microsomes from CHO cells were solubilized with 1% CHAPSO and cleared by centrifugation at 100,000 ϫ g for 1 h. The resultant supernatant was incubated with 100 nM C99-FLAG for 4 h in the presence of phospholipid at the concentrations indicated. A␤ production was visualized with 82E1, a monoclonal antibody specific for the amino terminus of human A␤. A, the addition of 1.3 mM (equivalent to 0.1%) PC enhanced A␤ production by the CHAPSO-solubilized membrane fraction. B, the addition of PI(4,5)P 2 attenuated A␤ production in a dose-dependent manner, with an IC 50 of ϳ141 M (equivalent to 0.016%). C, PI(4,5)P 2 inhibited A␤ production by the CHAPSO-soluble fraction even in the presence of 1.3 mM PC. The IC 50 was ϳ551 M (equivalent to 0.06%). Representative Western blots are shown. 0.25% CHAPSO buffer (50 mM PIPES, pH 7.2, 250 mM sucrose, 1 mM EGTA, 0.25% CHAPSO, 2 mM diisopropyl phosphorofluoridate, 20 g/ml antipain, 20 g/ml leupeptin, 20 g/ml TLCK, 10 mM phenanthroline, 2 mM thiorphan, and 0.1% phosphatidylcholine) with C99-FLAG substrate at 37°C for 4 h together with defined concentrations of PI(4,5)P 2 . To evaluate inhibitory effects of PI(4,5)P 2 on the interaction between ␥-secretase and C99-FLAG substrate, C99-FLAG prebound anti-FLAG M2 agarose beads (Sigma) were mixed with the CHAPSO-solubilized microsomal fraction of CHO cells and incubated at 4°C overnight in the presence or absence of 0.845 mM PI(4,5)P 2 . The agarose beads were washed three times and subjected to Western blotting to visualize ␥-secretase components, including nicastrin, carboxyl-terminal fragment (CTF) of presenilin 1, and Aph-1. Presenilin 1 CTF, Aph-1, and Pen-2 were detected with anti-presenilin 1 CTF antiserum (a gift from Dr. Iwatsubo, University of Tokyo), anti-Aph1 polyclonal antibody (Covance), and anti-Pen-2 polyclonal antibody (50), respectively.

RESULTS
Effects of PI(4,5)P 2 on ␥-Secretase-To demonstrate the effects of phosphoinositides on ␥-secretase activity, a CHAPSO-solubilized ␥-secretase assay was performed in the presence of various concentrations of PI(4,5)P 2 (15). The addition of PC to the CHAPSO-solubilized ␥-secretase reaction mixture at increasing concentrations up to 1.3 mM (equivalent to 0.1%) enhanced the production of A␤ as described previously (15,17,18) (Fig. 1A), whereas increasing concentrations of PI(4,5)P 2 dramatically reduced A␤ production with an IC 50 of 141 M (equivalent to 0.016%) (Fig. 1B). The inhibitory effect of PI(4,5)P 2 was observed even in the presence of 0.1% (1.3 mM) PC, with the IC 50 being ϳ551 M (equivalent to 0.06%) (Fig. 1C). We observed that PI(4,5)P 2 suppressed AICD and A␤ production in parallel (data not shown). To further confirm a direct effect of PI(4,5)P 2 on ␥-secretase activity, the ␥-secretase complex that was immunoprecipitated with anti-nicastrin antibody was evaluated for A␤ production in the presence of PI(4,5)P 2 . To enhance the detection of A␤, the PC was kept to 0.1% in the reaction mixture. As shown in Fig. 2, PI(4,5)P 2 inhibited A␤ production in a dosedependent manner, with the IC 50 being ϳ0.0085% (equivalent to 72.3 M). The values of IC 50 in three experimental conditions were different from each other; however, those are in the range of physiological variance of phosphoinositides concentration reported (27). We considered that the concentrations of phosphoinositide employed in this study were physiologically relevant. A␤40 and A␤42 produced in the reaction mixture were quantified with BA27 and BC05, respectively. No significant differential effects on A␤40 and A␤42 production were noted (Fig. 2B).
In contrast to PC, PI(4,5)P 2 contains a large head group with two phosphates (26). We reasoned that this bulky head group caused inhibition of ␥-secretase activity and tested various phosphoinositides using the CHAPSOsolubilized ␥-secretase assay system. The addition of PI tended to reduce productions of A␤ and sNICD-FLAG; however, we could not detect statistic significance (Fig.  3, A and B, and supplemental Fig.  S2). The addition of 0.85 mM phosphatidylinositol monophosphate (phosphatidylinositol 3-phosphate and phosphatidylinositol 4-phosphate) altered A␤ and sNICD-FLAG productions (Fig. 3, A and B, and supplemental Fig. S2). Interestingly, the addition of PI(5)P failed to alter A␤ and sNICD-FLAG productions significantly. Phosphatidylinositol diphosphates (PI(3,4)P 2 and PI(4,5)P 2 ) and phosphatidylinositol triphosphate (PI(3,4,5)P 3 ) showed significant inhibition of the enzyme activity, which implies that an increasing number of phosphate groups on the inositol ring enhances its inhibitory effect on ␥-secretase, as well as the position of phosphate group on inositol ring (Fig. 3, A and B, and supplemental Fig. S2). It should be noted that these phosphoinositides inhibited AICD production in parallel with A␤ production (data not shown). However, inositol 1,4,5triphosphate did not alter A␤ production in our assay system ( Fig.  3C and supplemental Fig. S2). These data indicate that the phosphorylated inositol moiety in combination with a fatty acid is required for the inhibition of ␥-secretase activity.
Inhibition by PI(4,5)P 2 of ␥-Secretase Activity in CHAPSOinsoluble Rafts-We showed that phosphoinositides inhibited ␥-secretase activity in the CHAPSO-soluble fraction. However, the inhibitory effects of phosphoinositides in the soluble fraction may not apply to ␥-secretase that is embedded in membrane. Thus CHAPSO-insoluble rafts from CHO cell membranes were prepared by sucrose density gradient centrifugation and incubated with C99-FLAG substrate in the presence of PI(4,5)P 2 . We found that lipid rafts prepared from CHO membranes produced A␤ from added C99-FLAG (supplemental Figs. S3 and S4). Fig. 4A shows decreased A␤ production from exogenous C99-FLAG by raft ␥-secretase in the presence of 0.85 mM PI(4,5)P 2 . This was also the case with endogenous C99 substrate (Fig. 4B). These data indicate that phosphoinositides suppress ␥-secretase even in the membrane-embedded state. . Effects of phosphoinositides on A␤ and sNICD-FLAG productions. ␥-Secretase substrates were incubated with CHAPSO-solubilized CHO membrane in the presence of various types of phosphoinositides. A␤ production was quantified as described (14,15). A, PI and PI(5)P did not alter A␤ production; however, other phosphoinositides significantly attenuated A␤ production at a concentration of 0.85 mM. B, ⌬E Notch-FLAG substrate was incubated with the CHAPSO-solubilized fraction. As seen in C99-FLAG, PI and PI(5)P failed to attenuated sNICD-FLAG production from ⌬E Notch-FLAG. Interestingly, an increasing number of phosphate groups on the inositol ring tended to suppress ␥-secretase activity to a greater extent. Inositol triphosphate itself failed to alter A␤ production even at 1.69 mM (C). This indicates that phosphoinositol in combination with fatty acid is necessary to inhibit ␥-secretase activity. The data represent the means Ϯ S.D. of three independent experiments. *, p Ͻ 0.05; **, p Ͻ 0.01 (analysis of variance, Scheffe's post hoc test compared with no treatment).

Effect of Phosphatidylinositol Specific PLC Inhibitor on A␤
Production from Isolated Membrane-It was clearly shown that PI(4,5)P 2 affected ␥-secretase activity in the detergent-soluble and membrane-embedded state. However, it is not known whether elevation of physiological PI(4,5)P 2 levels in cells alters A␤ production. Thus we tested whether pharmacological treatments that increase PI(4,5)P 2 level in cells would reproduce our findings on ␥-secretase in the detergent-soluble states. Edelfosin is known as phosphatidylinositol-specific PLC inhibitor. Fig. 5A indicates increase of PI(4,5)P 2 level in isolated microsomal fraction of cells treated with edelfosin (see also Fig. 5B). As shown in Fig. 5 (C and F), edelfosin treatment decreased amount of secreted A␤ in the medium. We did not detect significant reduction of A␤ level in the isolated microsomal fraction of cells treated with edelfosin; however, de novo A␤ production from the isolated membrane after incubation at 37°C for 30 min was significantly reduced (Fig. 5, C-E). These results support the idea that membrane lipid composition alters ␥-secretase activity in cells.
Effect of PI(4,5)P 2 on the Association between ␥-Secretase Complex and C99-FLAG Substrate-We showed inhibition by PI(4,5)P 2 of ␥-secretase activity in both CHAPSO-soluble and -insoluble states. To gain further insight into the mech-anisms of the inhibitory effects of phosphoinositides, we evaluated the association between ␥-secretase complex and C99-FLAG in the presence of PI(4,5)P 2 . Purified C99-FLAG, which was recaptured with M2 anti-FLAG agarose beads, was incubated with the CHAPSO-soluble fraction of CHO membranes at 4°C overnight in the presence or absence of 0.85 mM PI(4,5)P 2 . The C99-FLAG prebound agarose beads were spun down to evaluate the amounts of co-immunoprecipitated ␥-secretase components. Substantial amounts of nicastrin, presenilin 1 CTF, Aph-1aL, and PEN-2 were detected bound to the M2 anti-FLAG agarose beads preincubated with C99-FLAG in the absence of PI(4,5)P 2 (Fig.  6A). In contrast, reduced amounts of these ␥-secretase components were found in the presence of PI(4,5)P 2 , but the interaction between the M2 anti-FLAG agarose beads and C99-FLAG was not affected by PI(4,5)P 2 (Fig. 6A, bottom  panel). These data suggest that PI(4,5)P 2 interferes with the association between ␥-secretase components and C99-FLAG. Interestingly, PI(4,5)P 2 markedly decreased the amount of Aph-1aL co-immunoprecipitated with C99-FLAG and concomitantly increased the level of Aph-1aL in the supernatant (Fig. 6A). This suggests that the association of Aph-1aL with C99-FLAG is more susceptible to phosphoinositide inhibition than that of other ␥-secretase components. We also tested whether PI(4,5)P 2 peeled away ␥-secretase from prebound C99-FLAG complex. As described above, purified C99-FLAG was recaptured with the M2 anti-FLAG agarose beads and then incubated with CHAPSO-soluble fraction of the CHO membrane at 4°C overnight in the absence of PI(4,5)P 2 . After sufficient washing, C99-FLAG beads complexed to ␥-secretase was reincubated at 4°C overnight in the presence or the absence of 0.85 mM PI(4,5)P 2 . Western blot analyses revealed that PI(4,5)P 2 had no effect on stripping of ␥-secretase from prebound C99-FLAG (Fig. 6B).
Co-immunoprecipitation analyses suggest that PI(4,5)P 2 inhibits the association between ␥-secretase complex and C99-FLAG (Fig. 6A). However, it is possible that PI(4,5)P 2 mediates dissociation of the ␥-secretase complex itself, rather than of the interaction between ␥-secretase complex and C99-FLAG. To rule out this possibility, the effect of PI(4,5)P 2 on the stability of the ␥-secretase complex was investigated. The CHAPSO-soluble fraction was subjected to immunoprecipitation with antinicastrin antibody in the presence or absence of PI(4,5)P 2 to isolate the ␥-secretase complex. If PI(4,5)P 2 induces disassembly of the ␥-secretase complex, its components other than nicastrin would not be co-immunoprecipitated. We did not detect a significant difference between the signals for presenilin 1 CTF, Aph-1aL, and Pen-2 in the presence or absence of PI(4,5)P 2 , suggesting that PI(4,5)P 2 has no effect on the interaction between nicastrin and other components in our assay system (Fig. 6C). Overall, our data suggest that PI(4,5)P 2 mediates its inhibitory effect by suppression of the association of ␥-secretase with substrate.
Kinetics of the Inhibition of ␥-Secretase in the Presence of PI(4,5)P 2 -We showed that PI(4,5)P 2 inhibited activity of ␥-secretase through inhibition of substrate binding. This observation suggests that PI(4,5)P 2 potentially functions as a FIGURE 4. Inhibition of A␤ production by CHAPSO-insoluble rafts in the presence of PI(4,5)P 2 . CHO cells were solubilized with 1% CHAPSO and subjected to sucrose gradient centrifugation to isolate a detergent-insoluble, floating fraction (see "Experimental Procedures" for details). Isolated CHAPSO-insoluble rafts were incubated with 100 nM C99-FLAG (exogenous) substrate in the presence of 0.85 mM PI(4,5)P 2 , and A␤ production was quantified by the Western blotting. PI(4,5)P 2 was found to inhibit A␤ production (A). The rafts were prepared from CHO cells overexpressing C99 substrate. Isolated CHAPSO-insoluble fraction of the cells was incubated at 37°C in the presence of 0.85 mM PI(4,5)P 2 (B). A␤ production from endogenous C99 was markedly affected by PI(4,5)P 2 . The data represent the means Ϯ S.D. of three independent experiments. *, p Ͻ 0.005; **, p Ͻ 0.001 (t test compared with no treatment). 0Ј and 60Ј represent samples before and after 60 min of incubation at 37°C.
competitive inhibitor of ␥-secretase. Thus we tested whether the kinetics of A␤ production in the presence of PI(4,5)P 2 showed a double reciprocal plot typical of competitive inhibition. Various concentrations of C99-FLAG were incubated with CHAPSO-solubilized fractions of CHO membrane and defined amounts of PI(4,5)P 2 . After incubation, the reaction mixtures were subjected to Western blotting (Fig. 7A). As shown in Fig. 7B, double reciprocal plots of ␥-secretase activity in the presence of PI(4,5)P 2 generally displayed a pattern characteristic of competitive inhibition. This indicates that PI(4,5)P 2 acts principally as a competitive inhibitor of ␥-secretase in this reaction system.

DISCUSSION
Phosphoinositides are minor components of the phospholipids in the biological membrane. Nevertheless, they are pivotal signaling molecules involved in a number of biological processes, such as ion channel regulation, vesicle trafficking, actin polymerization, and cell migration (20 -25). Recently it has been reported that the turnover of PI(4,5)P 2 is affected in cells expressing a familial AD mutant of presenilin 1 (19). It was also shown that the PI(4,5)P 2 level was inversely correlated with the levels of A␤42 produced by cultured cells (19). PLC is known to hydrolyze PI(4,5)P 2 , and PLC-1␦ is abundant in neurons from AD brain (51). These observations suggest that phosphoinositides including PI(4,5)P 2 potentially regulate A␤ production and that modulation of phosphoinositide levels could offer a therapeutic approach for AD (19). Because A␤ is produced in the hydrophobic environment surrounded by the lipid bilayer and its secretion is mediated by vesicular trafficking, it is reasonable to postulate that the composition of the lipid bilayer has an influence on A␤ production and secretion. In fact, the production of A␤ was altered by lipids in in vitro ␥-secretase assay systems (15,17,18,52). In the present study, we have shown that phosphoinositides exhibited inhibitory effects on A␤ production by the CHAPSOsolubilized membrane fraction of CHO cells. This was also the case with presenilin 1 mutants (M146L and M233T) (data not shown). Moreover, we have shown that phospho-FIGURE 5. Effect of PI(4,5)P 2 elevation on A␤ production from isolated membrane. 7WD10 cells were treated with phosphatidylinositol specific PLC inhibitor, edelfosin at a concentration of 15 M in the absence of G418 for 6 h. Acidic lipids were extracted from microsomal fraction of the cells and spotted on PI(4,5)P 2 Mass Strip. PI(4,5)P 2 was visualized with PI(4,5)P 2 specific PH domain fused with glutathione S-transferase tag (A). The PI(4,5)P 2 level in isolated microsomal fraction was increased after edelfosin treatment (B). The microsomal fraction was incubated at 37°C for 0.5 h in the presence of 15 M edelfosin and subjected to Western blot (C). Edelfosin treatment failed to alter intracellular A␤ level (D); however, de novo A␤ production from the fraction after incubation at 37°C for 30 min was significantly reduced (E). Edelfosin treatment also reduced amount of secreted A␤ into medium (F). The data represent the means Ϯ S.D. of three independent experiments. The p values are indicated in graphs (t test compared with mock).
inositides inhibited the activity of immunoprecipitated ␥-secretase in the CHAPSO-solubilized assay system. Thus phosphoinositides have a direct effect on the activity of solubilized ␥-secretase. In addition, PI(4,5)P 2 suppressed A␤ production by CHAPSO-insoluble rafts. Thus inhibitory effects of phosphoinositides were exerted not only on ␥-secretase in the soluble state but also on ␥-secretase embedded in the membrane. PI(4,5)P 2 amounts to 0.3-2.0% of the total cellular lipid. The concentration of PI(4,5)P 2 was determined to be in a range of 32-159 M in a hypothetical cell sphere (27). However, phosphoinositides are generally distributed in the two-dimensional membrane within the cell, not in the three-dimensional space within the cell body. Theoretically, the concentration of PI(4,5)P 2 in the lipid bilayer should be higher than that described above. It has been reported that the theoretical local concentration of PI (3,4,5)P 3 at the inner leaflet of the plasma membrane of neutrophils is 5 M (28) and that after extracellular stimulation it increases to 200 M. The concentration of PI(3,4)P 2 is estimated to increase from 10 -20 to 100 -200 M upon stimulation. It has been reported that the concentration of PI(4,5)P 2 increases from 192-381 to 417-1153 M after expression of PI5K in superior cervical ganglia (45). PI(4,5)P 2 is a major phosphatidylinositol diphosphate, and its local concentration was estimated to be ϳ5 mM in the steady state level (28). Sheetz et al. (29,30) determined PI(4,5)P 2 concentration to be in a range of 4 -8 mM in intact erythrocyte membranes scaffolded by the spectrin network. Thus the concentrations of PI(4,5)P 2 employed in this study have considerable physiological relevance. In addition, PI(4,5)P 2 is reported to be concentrated at the inner leaflet of cholesterol-rich microdomains (rafts) (31)(32)(33). Furthermore, it has been reported that ␥-secretase is also enriched in the rafts (38 -40). We thus presumed that phosphoinositides including PI(4,5)P 2 and ␥-secretase colocalize in the same microdomain in living cells. It would be important in future studies to examine the colocalization of ␥-secretase and phosphoinositides in rafts.
As mentioned above, phosphoinositides inhibit A␤ production from C99-FLAG substrate by ␥-secretase in both CHAPSO-soluble and -insoluble fractions. ␥-Secretase cleaves not only APP but other type I membrane proteins after ectodomain shedding. It would be also important to evaluate inhibitory effects of phosphoinositides on cleavage of other substrates. We generated an artificial Notch sub-FIGURE 6. PI(4,5)P 2 inhibits the association of ␥-secretase with C99-FLAG substrate. Purified C99-FLAG that was recaptured with anti-FLAG agarose beads was incubated with CHAPSO-solubilized membrane fraction of CHO cells at 4°C in the presence or absence of 0.85 mM PI(4,5)P 2 . After thorough washing, the beads were subjected to Western blotting (WB) to visualize co-immunoprecipitated ␥-secretase components. The agarose beads incubated in the presence of 0.85 mM PI(4,5)P 2 showed a significant decrease in the signals for nicastrin, presenilin 1 CTF, Aph-1aL and Pen-2 bands (A). The level of C99-FLAG bound to the beads did not alter in the presence of PI(4,5)P 2 (A, bottom panel). Interestingly, PI(4,5)P 2 failed to peel away ␥-secretase complex prebound to C99-FLAG from the C99-FLAG substrate (B). To rule out the possibility of PI(4,5)P 2 -induced disassembly of ␥-secretase, the ␥-secretase complex was immunoprecipitated with anti-nicastrin antibody in the presence or absence of 0.85 mM PI(4,5)P 2 . The intensity of the coimmunoprecipitated ␥-secretase components was not affected even in the presence of PI(4,5)P 2 (C). It should be noted that Aph-1aL exhibits longer migration distance in Tris/Tricine gel (B and C), compared with that in Tris/glycine gel (A). These results suggest that ␥-secretase complex is stable in the presence of PI(4,5)P 2 and that PI(4,5)P 2 inhibits the association between ␥-secretase and the C99-FLAG substrate. The data are representative results of four independent experiments. sup and ppt indicate supernatant and precipitate, respectively, after immunoprecipitation. strate (⌬E Notch-FLAG) that mimicked an S2-cleaved fragment. ⌬E Notch-FLAG was incubated with the CHAPSOsoluble fraction of CHO cells in the presence of phosphoinositides. PI(4,5)P 2 was indeed found to inhibit cleavage of ⌬E Notch-FLAG in a similar fashion to that seen in C99-FLAG (Fig. 3B). This suggests that inhibitory effects of phosphoinositides on ␥-secretase are not substrate-specific.
Most interestingly, PI(4,5)P 2 was found to attenuate the association between ␥-secretase and the C99-FLAG substrate. This observation may help us to understand the mechanism of the inhibitory effects of phosphoinositides on ␥-secretase activity. Phosphoinositides are derivatives of phosphatidylinositol. The hexahydric inositol ring of phosphatidylinositol protrudes from the inner leaflet of the lipid bilayer into the cytoplasm. Such a large head group can be phosphorylated and become a landmark of the internal surface of the membrane to recruit potential binding proteins, such as AKT/PKB, spectrin, IRS, PLC-1␦, SOS, and dynamin (21,(53)(54)(55)(56)(57). Conversely, such a bulky head group may cause steric hindrance for assembly or binding of nearby molecules. However, the large head group alone was not enough to exert an inhibitory effect on ␥-secretase activity. Inositol triphosphate failed to alter A␤ production, suggesting that the fatty acid moiety in combination with a phosphoinositol ring is necessary to alter ␥-secretase activity. We postulate that the fatty acid of phosphoinositides escorts the bulky phosphoinositol into the interface between ␥-secretase and C99-FLAG substrate, leading to the inhibition of ␥-secretase.
The co-immunoprecipitation analyses support the idea that phosphoinositides function at least in part as a competitive inhibitor of ␥-secretase. Unexpectedly, in the presence of PI(4,5)P 2 the relative amount of Aph-1aL bound to the substrate was reduced compared with that of other components (Fig. 6A). Because we could not detect any difference in the amounts of co-immunoprecipitated Aph-1aL with anti-nicastrin antibody in the presence or absence of PI(4,5)P 2 (Fig. 6C), the reduction in Aph-1 bound to C99-FLAG may be accounted for by PI(4,5)P 2 -sensitive direct association of Aph-1aL to the substrate. It implies that even free Aph-1aL (that does not participate in a ␥-secretase complex) directly binds to C99-FLAG with PI(4,5)P 2 sensitivity. Kinetic analysis of A␤ production in the presence of PI(4,5)P 2 displayed a pattern consistent with competitive inhibition. This implies that phosphoinositides bind to substrate-binding site(s) of ␥-secretase complex. To our knowledge, PI(4,5)P 2 is the first natural compound to competitively inhibit ␥-secretase.