BACE inhibitors, currently in clinical trials for Alzheimer’s disease, can almost completely suppress the production of Aβ40 and Aβ42. In the January 23 Proceedings of the National Academy of Sciences, researchers led by William Netzer and Nobel laureate Paul Greengard at The Rockefeller University, New York City, propose an alternate way to slow BACE processing. The authors report data suggesting that, in cells, the cancer drug imatinib (trade name Gleevec) somehow shunts amyloid precursor protein away from BACE to be cleaved by lysosomal proteases instead. “Imatinib selectively inhibits BACE processing of APP, indicating that pharmacological exploitation of this mechanism could be useful for preventing AD,” Netzer told Alzforum. Commentators were skeptical of the selectivity claims. They noted the effect could be generic, ushering many proteins from the plasma membrane and endoplasmic reticulum to the lysosome.

The new paper contrasts earlier work from the same lab contending imatinib blocked γ-secretase activity. “This is an interesting initial observation pointing toward an indirect mode of action of imatinib, as we postulated previously,” Dirk Beher at Asceneuron, Lausanne, Switzerland, wrote to Alzforum (see Hussain et al., 2013). “Its relevance for the development of new therapeutics remains to be seen, and is subject to confirmation with more potent compounds that follow a defined structure-activity relationship.” Netzer and colleagues report that an inactive version of another kinase inhibitor has the same effect in mice, but it bears little structural resemblance to imatinib, noted Beher.

Better in Brain.

DV2-103, an inactive derivative of the kinase inhibitor PD173955, acts like Gleevec but stays in the brain longer. [Courtesy of Netzer et al., PNAS.]

In their earlier study, Greengard’s group had reported that imatinib lowered Aβ by suppressing γ-secretase cleavage (see Oct 2003 news). Researchers led by Michael Wolfe, then at Brigham and Women’s Hospital, Boston, attributed the inhibitory activity to an impurity in imatinib preparations (see Nov 2004 conference news; Fraering et al., 2005). Greengard and Netzer later proposed that imatinib blocked γ-secretase binding to a γ-secretase activating protein (GSAP), but other groups were unable to replicate this finding and GSAP was not mentioned in the new paper (see Sep 2010 newsJan 2014 news). 

In the new study, Greengard and Netzer re-examined imatinib’s effects and concluded that the drug impairs BACE processing of APP more than it does γ-secretase cleavage. The authors treated neuroblastoma cells expressing human APP with either 10 μM imatinib or 1 μM of γ-secretase inhibitor L-685,458. Neuroblastoma cells express GSAP. Both compounds lowered soluble Aβ40 levels roughly by half, but they produced distinct patterns of APP fragments, suggesting different mechanisms of action. γ-Secretase snips the C-terminal fragments left behind by β or α cleavage (β-CTFs and α-CTFs) and γ-secretase inhibitor, as expected, prevented this, allowing β-CTFs and α-CTFs to build. Imatinib, on the other hand, increased only α-CTFs but reduced levels of β-CTF, as well as the soluble APP fragment produced by BACE cleavage, sAPPβ. Together, these data implied that BACE was poorly cleaving APP. Meanwhile, two new C-terminal fragments of 10 and 16 kD appeared in cells treated with imatinib. The 16 kD fragment was the more abundant of the two, and mass spectrometry determined it to be about 141 amino acids long, longer than the fragments left behind by β or α cleavage.

Further tests indicated that the 10 and 16 kD fragments resulted from neither BACE nor α-secretase cleavage. Interestingly, a BACE inhibitor generated the same fragments in the neuroblastoma cultures, as did cells expressing human APP carrying the protective Icelandic A673T mutation, which suppresses BACE cleavage (see Jul 2012 news). All told, it seemed that protecting APP from BACE somehow creates these other cleavages.

How could that be? The drug did not affect cutting of two other BACE substrates, L1CAM and Sez6. Nor did it directly inhibit BACE in cell-free assays. The researchers say they do not know if these fragments might result from cleavage by η-secretase, which cuts about 90 amino acids upstream from the β-secretase site (see Aug 2015 news). Because imatinib accumulates in lysosomes, the authors wondered if it acted there. After they blocked lysosomal proteases by alkalization, imatinib no longer suppressed Aβ production nor induced the 10 and 16 kD fragments. Likely, imatinib pushes APP toward processing by various lysosomal proteases, and the 10 and 16 kD fragments are the products of lysosomal degradation, Netzer speculated. The authors are trying to identify the proteases involved and to determine exactly how imatinib might stimulate lysosomal degradation of APP.

Some commenters noted that more work needs to be done to demonstrate that the proposed lysosomal degradation acts specifically on APP and not any other BACE substrates. Because such a trafficking mechanism could be generic, more substrates should be tested, noted Jenny Gunnersen at the University of Melbourne, Australia, in an email to Alzforum (see full comment below). Michael Wolfe, now at the University of Kansas, Lawrence, agreed, suggesting that numerous substrates, especially neuregulin, should be tested side by side with APP. In the present paper, the authors investigated the cleavage of L1CAM and Sez6 in separate experiments from APP, and only reported effects on the C-terminal fragments, not all cleavage products, Wolfe pointed out. Since CTF levels are affected by cellular degradation as well as production, they might not reflect effects of BACE alone, said Wolfe.

Researchers wondered how the findings of weakened BACE cleavage fit with previous studies claiming that imatinib inhibits γ-secretase. To determine which mechanism predominated, the authors compared imatinib’s effects in neuroblastoma cells that express full-length APP versus those that express only the β-CTF fragment. In the latter, only inhibition of γ-secretase would affect Aβ40 levels. Imatinib suppressed Aβ40 four times more effectively in cells that expressed full-length APP than in those expressing β-CTF alone, indicating that most of its effect comes through influencing BACE processing. The finding belies previous results from the Greengard lab that imatinib slightly increased β-CTF levels in neuroblastoma cells and in the guinea pig brain (see Netzer et al., 2003). Netzer noted that those studies might have misidentified either α-CTF or the 16 kD APP fragment as β-CTF in autoradiographs, leading them to think β-CTF went up. His group now use western blots to more precisely identify fragments via antibodies, making the new results more rigorous, he said.

Imatinib never reached high concentrations in the brain because it is pumped out by p-glycoprotein and other efflux transporters. To study in vivo effects, the authors therefore used the related kinase inhibitor PD173955, which also lowers Aβ. They made a derivative, DV2-103, that lacks kinase activity (see image above). All three molecules similarly affected APP processing in cell culture, producing the 10 and 16 kD fragments. In vivo, however, DV2-103 was more potent than imatinib. The authors administered 15 mg/kg DV2-103 to 3xTg mice intraperitoneally and found it accumulated in brain. It reached concentrations of 10 μM or more after four hours and cut soluble Aβ40 levels by 40 percent.

In ongoing work, the authors are making additional derivatives of imatinib, looking for more potent compounds that stay in brain longer, Greengard wrote to Alzforum. Industry scientists prefer compounds that are active in the nM range in vivo, to minimize off-target effects. Many pharmaceutical scientists are awaiting results of ongoing BACE inhibitor studies before deciding whether to pursue further therapies targeting soluble Aβ, Beher added. Because current clinical BACE inhibitors can knock down production of the peptide almost to zero in the CSF, they offer a good proof of concept for whether curbing soluble Aβ can slow the progression of AD.—Madolyn Bowman Rogers


  1. While it would be important news to find a way to selectively inhibit β-secretase cleavage of APP without also blocking cleavage of other substrates of this protease, the new report from Netzer and colleagues falls short of making this claim. There are no side-by-side comparisons in the same experiments showing the differential effect on APP, and these experiments were carried out under different conditions, using stable transfection for APP and transient transfection for two other β-secretase substrates, L1CAM and Sez6. For L1CAM and Sez6, only their CTF levels were examined, not the soluble ectodomains, the other products of cleavage by β-secretase. Soluble ectodomain levels would provide a more reliable readout of effects on β-secretase cleavage of substrates, as levels of CTF are dependent on the degree of production by β-secretase and degradation by γ-secretase.

    These points are important, because if imatinib affects β-secretase activity more broadly, then the compound does not work like the protective Icelandic mutation, and there would be no obvious therapeutic advantage over direct β-secretase inhibitors. Moreover, Netzer et al. did not look at arguably the most important other substrate for β-secretase from an AD therapeutic perspective—neuregulin. Potential toxic effects of β-secretase inhibitors through blocking critical neuregulin signaling are of critical concern as these compounds advance through clinical trials. 

    Beyond the question of the validity of the new findings, I am concerned about the claim in this new study that it does not conflict with their original report on the effects of imatinib on Aβ production (Netzer et al., 2003). In that previous study, the authors claimed, and showed some evidence, that imatinib blocked cleavage of APP, but not Notch, by γ-secretase.  They showed some results where APP CTF-β was increased by imatinib, an effect consistent with inhibition of γ-secretase.  We and others were unable to reproduce these results with imatinib (e.g., Fraering et al., 2005). Later, this same lab reported that the relevant target of imatinib was a protein they dubbed γ-secretase interacting protein (GSAP), which selectively affected γ-secretase processing of APP over Notch (He G et al., 2010). This role of GSAP was never validated, although many labs—especially in biopharma companies—tried hard to do so, as it would have been a promising drug target (e.g., Hussain et al., 2013). 

    Now Netzer et al. say that imatinib is selectively inhibiting β-secretase cleavage of APP without affecting cleavage of other β-secretase substrates, with insufficient results to support this claim.  My worry is that many other labs will now spend precious time trying to reproduce and build upon this work, taking away from the pursuit of more promising avenues of investigation.


    . Gleevec inhibits beta-amyloid production but not Notch cleavage. Proc Natl Acad Sci U S A. 2003 Oct 14;100(21):12444-9. PubMed.

    . gamma-Secretase substrate selectivity can be modulated directly via interaction with a nucleotide-binding site. J Biol Chem. 2005 Dec 23;280(51):41987-96. PubMed.

    . Gamma-secretase activating protein is a therapeutic target for Alzheimer's disease. Nature. 2010 Sep 2;467(7311):95-8. PubMed.

    . The Role of γ-Secretase Activating Protein (GSAP) and Imatinib in the Regulation of γ-Secretase Activity and Amyloid-β Generation. J Biol Chem. 2013 Jan 25;288(4):2521-31. PubMed.

  2. The paper by Netzer and colleagues reports the results of their very interesting study describing the ability of imatinib mesylate (Gleevec) to lower Aβ levels. The authors show that Gleevec and inhibitors of BACE enzymatic activity likely function via different mechanisms because Gleevec and BACE inhibitor IV were found to have synergistic effects in lowering the levels of Aβ secreted from cells. The observation that Gleevec treatment resulted in production of the same proteolytic C-terminal fragments of APP as those generated from mutant APP bearing the protective mutation A673T suggests that Gleevec decreases the BACE-mediated processing of APP without interfering with its enzymatic activity per se. One of the major findings of this paper is that the observed effects were specific to APP. Notably, Gleevec treatment was permissive for BACE1 cleavage of the two other substrates tested, L1 and Sez6.

    These results are particularly promising since a major concern regarding the use of BACE inhibitors in the clinic is the potential for side effects arising from chronically blocking BACE cleavage of its numerous brain substrates. While the proposal that Gleevec sequesters APP away from BACE in the lysosomal compartment is plausible, it may be generic (i.e., favoring trafficking of proteins from the plasma membrane and endoplasmic reticulum to lysosomes). It will be important to test more BACE substrates before concluding that the Gleevec effect is absolutely specific for APP

    Another major advance described in this paper is the discovery of a compound related to Gleevec (DV2-103) that lacks the tyrosine kinase inhibitor properties but retains the ability to lower Aβ production by indirectly affecting APP processing by BACE. Unlike Gleevec, however, DV2-103 is able to cross the blood-brain barrier, accumulate in the brain, and significantly lower brain Aβ levels in a mouse model of Alzheimer’s disease. As it lacks tyrosine kinase inhibitor activity, DV2-103 might be predicted to cause fewer side effects than Gleevec. If DV2-103 has similar BACE substrate-sparing properties to Gleevec, its improved central nervous system availability makes it an extremely attractive proposition for further characterization as an Alzheimer’s disease therapeutic.

  3. This is an interesting paper. I am looking forward to seeing more mechanistic characterization in the future, in particular using primary neurons and endogenous APP, but also regarding the mechanistic basis of the differential effect on different BACE substrates (SEZ6 and L1 versus APP). Additionally, it would be great if the authors could identify similar compounds, but with much lower IC50 values.

  4. This is a very nice extended study concerning the effect of Gleevec on lowering Aβ generation. The authors provide convincing evidence that Gleevec inhibits APP-CTFβ production through a kinase-independent manner and that this requires an acidic environment. However, it is not fully revealed how Gleevec would inhibit APP-CTFβ production, but not the other BACE1 substrates such as Sez6 and L1CAM, since the mechanism for this inhibition is through trafficking to lysosomes. Membrane proteins such as APP and neuregulin-1 have similar cellular location patterns, and would be expected to be similarly inhibited. Animal studies may provide more insight into the mechanism and potential clinical use.

  5. Please excuse my tardiness in posting, but I just noticed a comment made by Michael Wolfe, concerning experimental conditions in our paper, Netzer et al. Dr. Wolfe's statement that N2a cells expressing a stable transfection of APP 695 were compared to transiently transfected N2a cells expressing the APP-A673T AD protective mutation is incorrect. The cells were transiently transfected for both APP 695 or APP-A673T, as noted in the Figure 2C legend and in the text p.1390, right side column lines 5-6, as well as in Methods.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Gleevec for Alzheimer's?
  2. San Diego: γ-Secretase Takes Scientists on a Wild Ride
  3. There’s a GSAP for That: Novel APP Partner a New Therapeutic Target?
  4. GSAP Revisited: Does It Really Play a Role in Processing Aβ?
  5. Protective APP Mutation Found—Supports Amyloid Hypothesis
  6. Enter Aη: Alternative APP Cleavage Creates Synaptotoxic Peptide

Research Models Citations

  1. 3xTg

Paper Citations

  1. . The Role of γ-Secretase Activating Protein (GSAP) and Imatinib in the Regulation of γ-Secretase Activity and Amyloid-β Generation. J Biol Chem. 2013 Jan 25;288(4):2521-31. PubMed.
  2. . gamma-Secretase substrate selectivity can be modulated directly via interaction with a nucleotide-binding site. J Biol Chem. 2005 Dec 23;280(51):41987-96. PubMed.
  3. . Gleevec inhibits beta-amyloid production but not Notch cleavage. Proc Natl Acad Sci U S A. 2003 Oct 14;100(21):12444-9. PubMed.

Further Reading

Primary Papers

  1. . Gleevec shifts APP processing from a β-cleavage to a nonamyloidogenic cleavage. Proc Natl Acad Sci U S A. 2017 Feb 7;114(6):1389-1394. Epub 2017 Jan 23 PubMed.