Gleevec shifts APP processing from a β-cleavage to a nonamyloidogenic cleavage

Significance

Alzheimer’s disease (AD) is the leading cause of dementia in the elderly. Four approved drugs are used to treat AD cognitive symptoms, including loss of short-term memory. These drugs produce modest, temporary benefits at best and do not prevent or delay worsening of the disease. Recently, a mutation that protects elderly people from developing AD was discovered. The cellular process responsible for the mutation’s protective effect was also identified, suggesting that drugs targeting this process or pathway also might provide protection against the development of AD. In the present study, we discovered that the anticancer drug Gleevec and a related compound mimic the effects of the protective mutation and thus can act as models for the development of effective drugs to fight AD.

Abstract

Neurotoxic amyloid-β peptides (Aβ) are major drivers of Alzheimer’s disease (AD) and are formed by sequential cleavage of the amyloid precursor protein (APP) by β-secretase (BACE) and γ-secretase. Our previous study showed that the anticancer drug Gleevec lowers Aβ levels through indirect inhibition of γ-secretase activity. Here we report that Gleevec also achieves its Aβ-lowering effects through an additional cellular mechanism. It renders APP less susceptible to proteolysis by BACE without inhibiting BACE enzymatic activity or the processing of other BACE substrates. This effect closely mimics the phenotype of APP A673T, a recently discovered mutation that protects carriers against AD and age-related cognitive decline. In addition, Gleevec induces formation of a specific set of APP C-terminal fragments, also observed in cells expressing the APP protective mutation and in cells exposed to a conventional BACE inhibitor. These Gleevec phenotypes require an intracellular acidic pH and are independent of tyrosine kinase inhibition, given that a related compound lacking tyrosine kinase inhibitory activity, DV2-103, exerts similar effects on APP metabolism. In addition, DV2-103 accumulates at high concentrations in the rodent brain, where it rapidly lowers Aβ levels. This study suggests that long-term treatment with drugs that indirectly modulate BACE processing of APP but spare other BACE substrates and achieve therapeutic concentrations in the brain might be effective in preventing or delaying the onset of AD and could be safer than nonselective BACE inhibitor drugs.

Rare mutations in amyloid precursor protein (APP) occur in select families, where they are inherited as autosomal dominant genes that invariably cause early-onset Alzheimer’s disease (AD) and are referred to as familial AD (FAD) mutations (1, 2). Most APP mutations result in increased production of amyloidogenic forms of Aβ peptides, mainly Aβ42 or an increased Aβ42/40 ratio (3), which increases the propensity for Aβ42 to form oligomers, leading to the accumulation of toxic fold variants of Aβ (4) and insoluble Aβ species that form amyloid plaques (5, 6). Recently, an APP mutation (APP-A673T) that protects against the development of AD and age-related cognitive decline was identified in the Icelandic population (7, 8). The mutation exerts a protective effect by rendering APP a less efficacious substrate of β-secretase (BACE), possibly as the result of lower catalytic turnover (8). The subsequent reduction in BACE processing of APP results in reduced production of Aβ and soluble APPβ (sAPPβ) but, as expected, no reported effect on the processing of other BACE substrates. Thus, the mutation reduces the amount of APP processed through the amyloidogenic pathway and can be considered a natural proof-of-concept experiment demonstrating the efficacy of long-term, selective BACE inhibition in the prevention of AD.

Nearly 70 putative BACE substrates have been identified in cells (9). An inhibitor of this enzymatic activity that broadly disrupts BACE processing of its numerous substrates is unlikely to be ideal for the prevention or long-term treatment of AD. Indeed, both BACE1 and BACE1/2 knockout mice, as well as mice treated chronically with a nonselective BACE inhibitor drug, have exhibited neurologic deficits (10). In the present study, we found that Gleevec inhibits BACE cleavage of APP but does not inhibit BACE processing of several other of its substrates known to affect synaptic plasticity, learning, and memory. We also found that the inhibitory activities of these compounds depend on an acidic cellular compartment. In addition, we report that a related compound with an effect similar to Gleevec, DV2-103, reaches therapeutic concentrations inside the brain, and thus may have clinical potential.

In previous work, we demonstrated that the anticancer drug Gleevec lowers Aβ levels through indirect inhibition of γ-secretase activity (11, 12). This inhibition was shown to be selective, in that Gleevec did not inhibit γ-secretase processing of Notch1, thus avoiding one of the chief drawbacks of nonselective γ-secretase inhibitors. In the present study, we found that the effects of Gleevec on APP metabolism, including Aβ production, are exerted through additional pathways.

Results

We compared APP C-terminal fragments (APP-CTFs) produced in N2a Swe/ΔE9 cells (3, 13) after exposure to Gleevec or to a γ-secretase inhibitor (Fig. 1A and Fig. S1). Whereas the γ-secretase inhibitor resulted in increased levels of APP-βCTF and APP-αCTF, Gleevec increased APP-αCTF levels, but also induced two unique APP-CTFs of 10 and 16 kDa.

Fig. 1.

Gleevec induces formation of APP-CTFs. (A) N2a APP Swe/PS1ΔE9 cells were incubated with DMSO (1:1,000 dilution), 1 μM γ-secretase inhibitor L-685,458 or 10 μM Gleevec for 5 h, and APP-CTFs from cell lysates were compared by Western blot analysis using antibody RU369. (B) N2a APP Swe/PS1ΔE9 cells were preincubated with BACE inhibitor IV for 1 h and then incubated in fresh medium with Gleevec and BACE inhibitor IV for 5 h. APP-CTFs from cell lysates were compared by Western blot analysis using antibody RU369. The arrowhead denotes APP-βCTF. (C) N2a APP Swe/PS1ΔE9 cells treated as in B. Secreted Aβ 40 levels from culture medium were measured by sandwich ELISA. n = 3. *P < 0.05; **P < 0.01, Student’s t test. (D) N2a APP 695 cells treated as in B. APP-CTFs from cell lysates were analyzed by Western blot analysis using antibody 6E10.

Fig. S1.

APP-CTFs produced by N2a APP Swe/PS1 ΔE9 cells exposed to Gleevec or the \x{03b3}-secretase inhibitor L-685,458. (A) Western blot (WB) of cell lysates using antibody RU369. The middle band produced by cells exposed to 10 μM Gleevec runs slightly below the βCTF and corresponds to the 10-kD APP-CTF. (B) Diagram of APP secretase cleavage sites and epitopes. (C) APP-CTFs produced by N2a APP Swe/PS1 ΔE9 cells treated as in A with either DMSO to 10 μM Gleevec or 1 μM L-685,458. WB antibody 6E10 is used to emphasize APP-βCTF and APP-C140.

We considered the possibility that the increase in α-CTF levels might be accompanied by a decrease in BACE activity, given that BACE and α-secretase activity are known to be competitive (14). We also wished to know whether the 10- and 16-kDa APP-CTFs were caused by alternative BACE cleavages of APP. To test these possibilities, we performed an order-of-addition experiment consisting of exposing N2a Swe/ΔE9 cells to two concentrations of BACE inhibitor IV and then to three concentrations of Gleevec plus BACE inhibitor IV (Fig. 1B). Rather than decrease the APP-CTFs that were induced by Gleevec, the BACE inhibitor raised their steady-state levels, suggesting that the Gleevec-induced APP-CTFs are caused by reduced BACE cleavage of APP. The combination of Gleevec and BACE inhibitor IV also decreased Aβ levels (Fig. 1C); however, the inhibitory effect of 10 μM Gleevec was greater than that of Gleevec alone compared with controls, suggesting that Gleevec and BACE inhibitor IV may act synergistically to lower Aβ levels.

To rule out the possibility that an FAD mutation is required to produce these effects, we performed a similar experiment using N2a 695 cells, which express human APP but no FAD mutations. The drug combination also raised levels of the Gleevec-induced 10- and 16-kDa APP-CTFs (Fig. 1D), indicating that these APP-CTFs are caused not by BACE, but rather by decreased BACE cleavage of APP in N2a 695 cells. We then sought to determine the location of the APP cleavage site that gave rise to the 16-kDa APP-CTF as an approach to investigate how this APP cleavage occurs. To accomplish this, we measured the precise molecular weight of the 16-kDa APP-CTF by performing MALDI-MS after immunoprecipitation of cell lysate prepared from N2a 695 cells treated with Gleevec. We noted a major peak of molecular mass, 15,622.53 Da, corresponding to the predicted mass of a 141-aa APP-CTF (C141).

We further discovered that Gleevec induced a pattern of APP-CTFs in N2a 695 cells (Fig. 2A), with the APP-C141 fragment as the predominant cleavage product. Importantly, the APP-CTFs induced by Gleevec or by BACE inhibitor IV were nearly identical, leading us to consider the possibility that the APP-CTF pattern is a signature of BACE inhibition, and that Gleevec (like the BACE inhibitor) inhibits BACE processing of APP. We tested this by treating N2a 695 cells with Gleevec or BACE inhibitor IV and then measuring levels of sAPPβ, the secreted N-terminal fragment of APP that is a direct product of BACE processing of full-length APP (FL-APP) (15) (Fig. 2B). Remarkably, Gleevec lowered levels of sAPPβ in a dose-dependent manner (mean ± SEM, 57 ± 6.2% of control at 10 μM and 90 ± 0.8% of control at 20 μM). However, in contrast to the BACE inhibitor, which caused a rise in sAPPα, 10 μM Gleevec did not affect sAPPα levels, whereas 20 μM Gleevec lowered sAPPα to 64.14 ± 1.96% of that in controls.

Fig. 2.

Gleevec inhibits BACE processing of APP. (A) Nearly identical APP-CTFs are formed in N2a 695 cells exposed to 5 μM BACE inhibitor IV or 10 μM Gleevec, as analyzed by Western blot analysis using antibody RU369. (B) BACE inhibitor IV and Gleevec inhibit formation of sAPPβ. Western blot antibodies, RU anti-sAPPβ, and 6E10. (C) N2a WT cells transiently transfected with APP 695 or APP-A673T were treated with vehicle or 10 μM Gleevec. Cells were lysed and analyzed by Western blot analysis using antibody RU369.

Given that the APP protective mutation (APP A673T) has been reported to reduce BACE cleavage of APP (7), we compared APP-CTFs generated by N2a cells transiently transfected with either WT APP695 or APP695-A673T exposed to vehicle or to Gleevec. Remarkably, expression of the protective mutation produced APP-CTFs similar to those formed by N2a cells expressing APP695 in response to Gleevec (Fig. 2C). In addition, N2a APP-A673T cells exposed to Gleevec produced even higher steady-state levels of these APP-CTFs.

Pharmacologic inhibition of BACE by a nonselective inhibitor has the potential to affect numerous signaling pathways, given the large number of putative BACE substrates present in cells (16). In contrast, the APP protective mutation, A673T, reduces BACE processing of APP but presumably of no other protein. To determine whether Gleevec’s effect on BACE processing of APP is selective, we tested whether Gleevec inhibits BACE processing of two physiological BACE substrates, L1CAM and Sez6 (Fig. 3A). Both of these proteins led to intracellular BACE cleaved CTFs (βCTFs), which are inhibited by BACE inhibitor IV but not by Gleevec, demonstrating that Gleevec shows some selectivity for inhibiting BACE processing of APP. To determine whether Gleevec inhibits BACE directly, we performed an in vitro assay testing BACE1 cleavage of a fluorogenic peptide substrate (Fig. 3B), which demonstrated that even at a concentration of 100 μM, Gleevec has no direct effect on BACE1 activity, in contrast to active-site BACE inhibitor IV. Nonetheless, to rule out the possibility that when an APP substrate is present, Gleevec inhibits BACE cleavage of APP either directly or by interacting with APP, we performed a second in vitro BACE assay testing BACE1 cleavage of APP-C140 (Fig. 3C). BACE inhibitor IV completely eliminated BACE1 cleavage of the APP-C140 substrate, in contrast to Gleevec, which had no effect at all concentrations tested, thus again ruling out a direct inhibitory effect of Gleevec on BACE1 or an indirect effect through interaction with APP. Interestingly, it has been reported that a series of benzofuran-containing compounds selectively inhibit BACE processing of APP in vitro by binding directly to APP around the BACE cleavage site (17), and another study involving two flavonoid compounds reported similar results (18). It is unlikely that Gleevec inhibits BACE cleavage of APP by the same mechanism, however. Our in vitro BACE1 assay uses APP C140 as a substrate, yet Gleevec fails to inhibit BACE1 cleavage within this APP C-terminal region, even at a concentration of 200 μM.

Fig. 3.

Gleevec does not inhibit BACE processing of L1CAM and Sez6 and does not directly affect BACE activity. (A) N2a cells are transfected with L1CAM and Sez6 and the effects of 5 μM BACE inhibitor IV or 10 μM Gleevec (G) on BACE processing of each substrate were resolved by comparing βCTFs generated for each protein. (B) In vitro BACE1 assay measuring recombinant BACE1 processing of a fluorogenic BACE substrate exposed to increasing concentrations of Gleevec or 5 μM BACE inhibitor IV. n = 3. ***P < 0.001, Student’s t test. pep, fluorogenic peptide. (C) In vitro BACE1 assay consisting of recombinant BACE1 and recombinant APP-C140 exposed to Gleevec or BACE inhibitor IV. (D) Comparison of secreted Aβ40 levels on N2a 695 and N2a APP-C99 cells exposed to increasing concentration of Gleevec. Student’s T-test, SEM **P < 0.01, n = 3.

In our earlier studies, we demonstrated that Gleevec lowers Aβ production through a γ-secretase–dependent pathway (11, 12). To determine the extent to which Gleevec contributes to an Aβ decrease through selective inhibition of BACE activity vs. inhibition of γ-secretase, we performed a Gleevec dose–response experiment comparing Aβ40 secretion in N2a 695 cells and N2a βCTF (C99) cells, which require γ-secretase, but not BACE, for Aβ production (Fig. 3D). Thus, we demonstrated that Gleevec is 3.5- to 5-fold more potent at inhibiting production of Aβ in N2a 695 cells compared with N2a βCTF cells at concentrations ranging from 10 to 40 μM, suggesting that BACE is a rate-limiting factor in the inhibition of Aβ production by Gleevec.

We previously demonstrated that both Gleevec and the Abl and Src kinase inhibitor PD173955 (“inhibitor 2”) inhibit Aβ production independent of Abl kinase activity (11). In the present study, to extend this analysis to other kinases, we used PD173955 as a template and chose a kinase inhibitor-defective derivative of that compound, DV2-103 (19) (Fig. 4A). We then compared the effects of DV2-103, Gleevec, and PD173955 on the kinase activities of Abl, Src, and MAP kinase 1 (Fig. 4B). Gleevec inhibited only Abl kinase, whereas PD173955 inhibited both Abl and Src, as expected, and showed some inhibition of MAP kinase 1 as well. DV2-103 did not inhibit any of these kinases, and also failed to show significant kinase inhibition in a screen consisting of 211 additional kinases (Fig. S2). Remarkably, we found that DV2-103, although not a kinase inhibitor, inhibited the production of Aβ in N2a 695 cells (Fig. 4C), as did Gleevec, and produced the same pattern of APP-CTFs, including C141 (Fig. 4D). These results indicate that kinase inhibition is not necessary for DV2-103 to have a Gleevec-like effect on APP metabolism. We previously demonstrated that Gleevec does not inhibit γ-secretase–catalyzed Notch cleavage (11). Therefore, we tested DV2-103 in a similar assay, and found that like Gleevec, it did not inhibit the γ-secretase–catalyzed S3 cleavage of Notch (Fig. 4E).

Fig. 4.

The effects of a kinase-inhibitor-defective compound (DV2-103) and Gleevec on APP metabolism. (A) Chemical structures of Gleevec, the Src/Abl kinase inhibitor PD173955, and its kinase inhibitor inactive derivative DV2-103. (B) Effects of each inhibitor on kinase activities of Abl, Src, and Map kinase in vitro, Student’s t test. There were no significant differences between means for DV2-103 and vehicle control for any of the kinases tested. (C) DV2-103 inhibits production of secreted Aβ40 in N2a 695 cells. n = 4. *P < 0.05, Student’s t test. (D) DV2-103 induces a unique pattern of APP-CTFs in N2a cells. (E, Left) The γ-secretase inhibitor L-684,458 inhibits the formation of NICD in N2a 695 cells. n = 3. **P < 0.01, Student’s t test. (E, Right) DV2-103 does not inhibit formation of the NICD in N2a 695 cells, as analyzed by Western blot analysis using an anti-NICD antibody. (F) DV2-103 was administered to 3xTg mice by i.p. injection, and brain Aβ levels were analyzed 4 h later. n = 10 for controls, 5 for the 15 mg/kg group, and 10 for the 30 mg/kg group. **P < 0.01, ***P < 0.001, Student’s t test.

Fig. S2.

Additional kinases in the DV2-103 Z-lyte screen (Invitrogen). Positives have white backgrounds and are statistically significant. Blue background denote no statistical significance and thus are negatives. The positives may be exaggerated because ATP concentrations for the kinases scoring positive were 100 μM during the assay, which is 10–20 times lower than the ATP concentrations found in cultured cells. That would tend to exaggerate the potency of an ATP-competitive ligand (kinase inhibitor) tested at 10 μM. Specific inhibition of Raf, Braf (positives), and several mutations of these kinases were later tested in APP 695 cells, and none inhibited Aβ production or altered APP metabolism. Therefore, inhibition of these kinases could not be associated with the effects of DV2-103 on Aβ or APP metabolism.

Gleevec does not accumulate in brain at therapeutic concentrations, owing to its rapid efflux from the brain by the p-glycoprotein and other efflux pumps that form the blood-brain barrier (20–22). We tested i.p. administration of DV2-103 in C57BL/6 mice to determine whether it might achieve the same or higher brain concentrations than Gleevec. In contrast to Gleevec (23), DV2-103 accumulated in mouse brains at concentrations >10 μM by 4 h after i.p. administration of 30 mg/kg (Table S1). We then administered DV2-103 (15 and 30 mg/kg) via i.p. injection to 3xTg AD mice (24) to test for effects on Aβ levels. Remarkably, DV2-103 lowered brain Aβ levels in 3xTg mice to 60 ± 4.4% of those in controls by 4 h after administration of 15 mg/kg (Fig. 4F).

Table S1.

Plasma and brain concentrations of DV2-103 in six C57BL/6 mice at 4 h after i.p. injection of 30 mg/kg

Plasma, μM	Brain, μM
12.76	39.26
10.55	46.03
9.07	36.92
6.67	34.04
4.77	26.92
5.57	25.09

We next addressed the questions of what proteases are responsible for formation of the Gleevec- and DV2-103–induced APP-CTFs, and whether inhibition, or, conversely, stimulation, of such proteases contributes to the effects of these compounds on APP metabolism. Almost all (95%) of Gleevec entering cells (regardless of cell type) becomes sequestered in lysosomes (25, 26) through ion trapping, because Gleevec is a weak base (27–29). To examine the possibility that Gleevec might inhibit the activities of lysosomal proteases by raising lysosomal pH, we exposed N2a 695 cells to vehicle or to Gleevec for 4 h and then to the lysosomotropic dye Lysotracker Red (Thermo Fisher Scientific), which sequesters in acidified lysosomes (30). Gleevec pretreatment increased the amount of Lysotracker Red in lysosomes and also appeared to increase lysosomal volume (Fig. S3), indicating maintenance of acidic pH in the presence of Gleevec. Therefore, proteases that require low pH for activity could remain active. It also is unlikely that Gleevec lowers the pH of lysosomes.

Fig. S3.

N2a 695 cells incubated with DMSO control (Left) or 10 μM Gleevec (Right) for 4 h, followed by 5 min incubation with LysoTracker Red. Punctate brightness shows acidic organelles in areas of LysoTracker Red concentration (mainly lysosomes). Each cluster of pucta reside within a single cell. Live cell imaging, 40× magnification. Each field contains equal numbers of cells photographed using same optical settings with a DeltaVision Image Restoration Microscope (Applied Precision) at The Rockefeller University Bioimaging Resource Center.

To examine the possibility that lysosomal proteases mediate or contribute to the effects of Gleevec and DV2-103 on APP metabolism, we exposed N2a 695 cells to bafilomycin A1, an inhibitor of the vacuolar ATPase (31), to deacidify lysosomes and other acidic cellular compartments, thereby preventing their proteases from functioning. After incubation with bafilomycin A1, no LysoTracker Red fluorescence was visualized as punctate structures in N2a 695 cells. Remarkably, bafilomycin A1 also resulted in nearly complete inhibition of formation of C141 and other APP-CTFs except for the APP-αCTF (Fig. 5A). We next measured sAPPβ levels to test whether bafilomycin A1 also affected BACE processing of APP. Remarkably, bafilomycin A1 did not prevent BACE (or a BACE-like) processing of APP (Discussion), but did prevent inhibition of BACE processing by Gleevec and DV2-103, as measured by secreted levels of sAPPβ (Fig. 5A, Lower). Furthermore, the ability of Gleevec and DV2-103 to lower Aβ40 levels was greatly reduced in the presence of bafilomycin A1 (Fig. 5 B and C).

Fig. 5.

An acidic intracellular compartment is required for the production of APP-C141 and for inhibition of BACE processing of APP by Gleevec and DV2-103. (A, Upper) Lysate: N2a 695 cells were treated with vehicle or bafilomycin A1 and exposed to Gleevec, DV2-103, or BACE inhibitor IV. Shown are Western blots of cell lysate, using RU369 antibody. (A, Lower) Media: sAPPβ was resolved by Western blot analysis using RU anti-sAPPβ antibody. G, 10 μM Gleevec; DV, 20 μM DV2-103; IV, 1 μM BACE inhibitor IV. (B) Effects of inhibitors on secreted Aβ40 ± bafilomycin A1, on Aβ40 sandwich ELISA. n = 3. **P < 0.01; ***P < 0.001, Student’s t test. (C) Gleevec dose–response ± bafilomycin A1 (x axis is μM Gleevec) by Aβ40 sandwich ELISA. n = 3. **P < 0.01; ***P < 0.001, Student’s t test. (D) N2a 695 cells treated with 10 μM Gleevec and/or 5 mM NH₄Cl. Cell lysate, Western blot antibody RU369. (E) N2a 695 cells treated with 10 μM Gleevec and/or 5 mM NH₄Cl. N2a 695 cell media, Aβ40 sandwich ELISA. n = 3. ***P < 0.001, Student’s t test. (F) Lysosomal fractions (LF) from N2a 695 cell lysate treated with vehicle or 10 μM Gleevec. APP-CTFs were resolved by Western blot antibody RU369, and Lamp 2 was resolved by anti-mouse Lamp 2.

To further demonstrate that an acidic cellular compartment is required for Gleevec’s effects on APP metabolism, we exposed N2a 695 cells to the alkalizing agent NH₄Cl to raise the pH of otherwise acidic intracellular compartments (32). Like bafilomycin A1, NH₄Cl prevented Gleevec from inducing APP-C140 or the 10-kDa APP-CTF (Fig. 5D) and from lowering Aβ40 levels (Fig. 5E). To test the possibility that the Gleevec-induced APP-CTFs are localized to lysosomes, we purified lysosomes from N2a 695 cells that had been exposed to vehicle or Gleevec, and found that the Gleevec-induced APP-CTFs were in fact localized to lysosomes (Fig. 5F).

Discussion

Gleevec and DV2-103 indirectly lower BACE processing of APP, reducing Aβ production and promoting the formation of multiple APP-CTFs, including a unique 16-kDa APP-CTF. These activities are kinase-independent, because DV2-103 in not a kinase inhibitor, and selective, because the inhibition of BACE processing of APP by Gleevec does not inhibit BACE processing of L1CAM and Sez6.

By comparing the Aβ-lowering effects of Gleevec in N2a cells that express either APP695 or APP-βCTF (C99), we provide evidence that Gleevec’s inhibition of BACE processing of APP is the major contributor to the decreased Aβ levels in N2a cells. This result does not conflict with previous reports of Gleevec lowering γ-secretase activity, but instead suggests that multiple pathways affecting Aβ levels are involved.

Selective inhibition of BACE processing of APP is also a hallmark of the APP-protective mutation APP-A673T, which was recently discovered in the Icelandic population and has been shown to protect against the development of AD and age-related cognitive decline. Thus, this mutation represents a natural experiment providing proof of principle that long-term, selective inhibition of BACE processing of APP reduces the risk of mental deterioration in elderly persons, and suggests that drugs producing a similar biochemical phenotype may be especially useful in preventing or delaying AD.

We also have demonstrated that Gleevec and DV2-103 induce formation of the same APP-CTFs that occur in response to treatment with a standard BACE inhibitor. More importantly, this commonality of APP-CTFs extends to the APP-protective mutation A673T, further suggesting that the CTF pattern is a signature of inhibition of BACE processing of APP and is likely to be nontoxic, considering that the APP-protective mutation is beneficial in humans (7). The model shown in Fig. 6 summarizes our results.

Fig. 6.

Diagram showing the mechanisms driving APP metabolism in control, with BACE inhibitor, an APP protective mutation, and Gleevec or DV2-103 compound.

Our analysis is particularly interesting, in that two other groups recently reported additional cleavages of APP in response to a BACE inhibitor that result in previously uncharacterized APP-CTFs, APP-NTFs (33), and N-terminally extended Aβ peptides (34). Both groups found that some of these APP metabolites inhibit long-term potentiation in mouse hippocampal slices and thus are synaptotoxic. We do not know whether any of the APP metabolites induced by Gleevec or DV2-103 are the same as those identified by the Haass and Selkoe groups, but we have shown that the Gleevec-induced CTFs are very similar to those produced by the protective APP mutation.

We also have demonstrated that Gleevec does not inhibit BACE activity directly, based on in vitro BACE assays. Thus, the inhibition of BACE cleavage by BACE inhibitor IV and the reduction of BACE cleavage of APP by Gleevec have different mechanisms of action. This finding is supported by our observation of the synergistic effect of the two inhibitors used together. Had the mechanisms been the same, we would have expected an additive effect.

We also provide evidence that the Aβ-lowering effects of Gleevec and DV2-103 require one or more acidified cellular compartments. This is interesting, because BACE has an acidic pH optimum (35, 36) but APP is nevertheless metabolized to Aβ in the presence of bafilomycin A1. This phenomenon, which has been observed by others (37–39), could result from BACE-like cleavage of APP by cellular proteases that do not require an acidic milieu for activity. Candidate proteases include caspases that may become activated by cathepsins leaked from lysosomes after bafilomycin A1 treatment (40). The increased levels of Aβ that occur after bafilomycin A1 treatment also could result from inhibiting autophagy, thus increasing the pool of the Aβ precursor, APP-βCTF (41), or from increased BACE levels that could occur when BACE cannot be degraded in lysosomes (42).

We have further implicated lysosomes as mediating the effects of Gleevec and DV2-103 on APP metabolism by showing a strict requirement for an acidified cellular compartment for these drugs to induce the formation of APP-CTFs. It appears likely that except for the APP-αCTF, which is produced at the plasma membrane, the Gleevec-induced APP-CTFs are produced in lysosomes. If they were produced outside of the lysosome, we would expect to see a buildup of these CTFs when lysosomes were inhibited by bafilomycin. In contrast, we demonstrated that the Gleevec-induced APP-CTFs were greatly reduced under these conditions. Furthermore, deacidification of lysosomes by bafilomycin or NH₄Cl has been shown to prevent the sequestration of Gleevec within lysosomes (25). Thus, the concentration of Gleevec or DV2-103 within these lysosomes might become too low to engage a Gleevec or DV2-103 target, should a discrete target exist in lysosomes.

Based on our results, we propose that the selective reduction of BACE processing of APP is another mechanism by which Gleevec and DV2-103 lower Aβ production in cells. The effect of these drugs on lysosomal volume are consistent with the effects of other amphipathic, cationic molecules containing aromatic rings, which cause an increase in lysosomal volume, purportedly through intercalation of lysosomotropic drugs within lipid bilayers, but with no change in lysosomal pH (43). Such compounds also have been shown to increase the trafficking of proteins from the plasma membrane or endoplasmic reticulum to lysosomes (44, 45), and by doing so would be expected to reduce the quantity of APP exposed to amyloidogenic processing by BACE and γ-secretase.

Materials and Methods

Mice.

The 3xTg mice (24) used in this study express the human FAD mutations APP KM670/671NL (Swedish), MAPT (tau) P301L, and PSEN1 M146V (The Jackson Laboratory; JAX MMRRC stock no. 034830). The WT mice were C57BL/6J (The Jackson Laboratory; stock no. 000664). The 2-mo-old C57BL/6J mice were given DV2-103 dissolved in DMSO by i.p. injection at a dose of 30 mg/kg. Brains were dissected 4 h later. Hemibrains without cerebellum and olfactory bulbs were weighed and frozen in liquid N₂. Measured volumes of blood were harvested postmortem and frozen in liquid N₂. Brain and blood were transferred to Memorial Sloan Kettering Cancer Center’s Analytical Pharmacology Laboratory for measurement of DV2-103 concentration, with values normalized to brain wet weight or plasma volume. The 3xTg mice were administered DV2-103 i.p., and hemibrains were dissected after 4 h and analyzed for total Aβ by PAGE, followed by electrotransfer to PVDF membranes and Western blot analysis with antibody 6E10 (Signet Testing Labs).

Transfection of Cells.

APP 695, APP A673T, mNotch Δe. N2aWT cells were transfected using either Lipofectamine 2000 or Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. APP A673T DNA was constructed using human APP 695 DNA as a template.

Cell Cultures.

Gleevec (imatinib mesylate) was purchased from BioVision, and 10 mM stock solutions were prepared in DMSO (Sigma-Aldrich). DV2-103 was a generous gift from Drs. Darren Veach and B. Clarkson at MSKCC and was synthesized at the Organic Chemistry Core facility of Sloan Kettering Institute. stock solutions (10 and 20 mM) were prepared in DMSO (Sigma-Aldrich). Each Inhibitor was rapidly mixed in culture medium (1:1,000 dilution) by vortexing and then layered onto adherent cells (>95% confluent) that had been incubated in 6- or 12-well tissue culture plates (Corning) for 5 h at 37 °C in 5% CO₂. N2wt, N2 695, N2a-C99, and N2a Swe/Δe9 cell lines were used. In addition, inhibitors were tested on Cos cells and 3T3 fibroblasts transfected with APP 695 or APP Swe/Δe9 cells.

Elisa Assays.

Human Aβ 40/42 was measured in cell culture media by ELISA in accordance with the manufacturer’s instructions. Detailed information is provided in SI Materials and Methods.

Immunoprecipitation/Western Blot Analysis.

Antibody 6E10 (Covance) was used to immunoprecipitate APP-C141 from cell lysates. Primary antibodies used for Western blot analyses included antibody 6E10 (Covance) to identify sAPPα, APP-βCTF, and FL-APP; antibody RU369 to identify APP-CTFs, RU anti-sAPPβ to identify sAPPβ, and anti-Lamp 2 (Abcam) to identify Lamp 2; and anti-NICD (Cell Signaling Technology) to identify cleaved Notch1. Western blots analyses were otherwise carried out as reported previously (11) using SDS/PAGE and electrotransference to PVDF membranes (EMD Millipore).

Notch-1 Cleavage Assay.

mNotchΔe lacking most of the Notch extracellular domain was transiently transfected into N2a WT cells as reported previously (11). mNotchΔe processing by γ-secretase was detected by Western blot analysis using anti-NICD antibody.

Lysosome Purification.

Lysosomes were purified using the Pierce Lysosome Enrichment Kit (Thermo Fisher Scientific) in accordi the manufacturer’s instructions. Detailed information is provided in SI Materials and Methods.

In Vitro BACE Activity: Fluorogenic Peptide Assay.

To measure BACE activity in vitro, 1 μM fluorogenic peptide substrate Mca-SEVNLDAEFRK(Dpn)RR-NH₂ (R&D Systems) was incubated for 1 h at 37 °C with 100 nM rhBACE1 (R&D Systems). After incubation, fluorescence (excitation at 320 nm, emission at 405 nm) was monitored using an Envision 2104 multilabel reader (PerkinElmer). A specific concentration of Gleevec constituted each test condition. The positive control was 5 μM BACE Inhibitor IV, and the negative control (vehicle) was DMSO.

In Vitro BACE Activity: BACE Cleavage of APP-C140.

APP C140 was purified from Escherichia coli using affinity purification. Then 1 μM APP-C140 was incubated with 100 ng of rhBACE in the presence of Gleevec, BACE Inhibitor IV (positive control), or DMSO (negative control) for 1 h at 37 °C. BACE processing of APP-C140 was assayed by Western blot analysis using antibody RU369 to detect APP-βCTF, the immediate BACE cleavage product of APP.

Live Cell Imaging.

N2a 695 cells were used for live cell imaging. Detailed information is provided in SI Materials and Methods.

SI Materials and Methods

ELISA Assays.

Human Aβ 40/42 was measured in cell culture media and homogenized mouse brain extracts using Aβ 40/42 sandwich ELISA kits (Thermo Fisher Scientific) in accordance with the manufacturer’s instructions. Brain lysates were made by sonicating preweighed hemibrains in 3% SDS with EDTA-Free Protease Inhibitor Mixture (Roche) on ice. Samples were normalized to total protein and diluted 10-fold in buffer containing 2% Triton-X100 (Sigma-Aldrich), and then centrifuged at 100,000 × g for 1 h in a TLS-55 rotor (Beckman Coulter).

Lysosome Purification.

N2a695 cells grown in 15-cm dishes to 95% confluence were exposed to 10 μM Gleevec or vehicle (DMSO) (dilution of drug and vehicle, 1:1,000). Dishes were placed on ice for 10 min, after which media was removed and cells were washed three times with ice-cold PBS with protease inhibitor mixture (Roche). Cells were lifted off of the dish using a cell scrapper and then processed for lysosomes using the Pierce Lysosome Enrichment Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions.

MALDI-MS.

N2a 695 cells were incubated with vehicle or 10 μM Gleevec for 6 h. Cell lysates were prepared in 3% SDS with Protease Inhibitor Mixture (Roche) at 0 °C. Lysates were diluted 10-fold into buffer (20 mM Tris⋅HCl pH 7.4, 300 mM NaCl, 5 mM EDTA, 1 mM MgCl, containing 1% Triton-X100) and immunoprecipitated using antibody 6E10 (Covance). Immunoprecipitates were run on 10–20% Tris-Tricine SDS gels (PAGE). Sections of gels corresponding to the position occupied by the 16-kDa APP-CTF were cut out and transferred to The Rockefeller University Proteomics Core facility for MALDI analysis.

Live Cell Imaging.

N2a 695 cells were cultured on poly-d-lysine–coated glass-bottom culture dishes (MatTek). Cells were maintained in 1:1 DMEM/Opti-MEM (Thermo Fisher Scientific) containing 5% FBS for 24 h. After preincubation with 10 μM Gleevec or vehicle control for 4 h, the medium was changed, and cells were incubated for 5 min with LysoTracker Red in the absence of Gleevec, after which cultures were immediately transferred to a DeltaVision Image Restoration Microscope (Applied Precision). Five fields per culture plate were photographed at 40× magnification. Illumination parameters were kept constant.