The goal of BACE1 inhibition is to stop the production of toxic Aβ. However, a new study reports that at very high doses, the inhibitors may harm the very neurons they are meant to protect. Initial findings were presented by senior author Jochen Herms of Ludwig-Maximilians-Universität, Munich, at a conference last year (see Dec 2013 conference story). In press in Biological Psychiatry, the full study now reports that BACE inhibitors quash synaptic plasticity and cause memory loss in normal adult mice. The findings suggest that such inhibitors could pose a threat to neurons if used at high enough doses, but also offer reassurance that low doses could block Aβ production without affecting memory. The paper is available online.

BACE1 performs the rate-limiting step of processing amyloid precursor protein (APP) into amyloidogenic Aβ species, making the enzyme a prime therapeutic target. However, developing inhibitors to block the enzyme has not been easy. The enzyme’s sizable catalytic site acts on myriad substrates, and researchers walk a fine line when seeking to block the enzyme without unleashing undue side effects. Measuring neuronal reactions to the inhibitors, particularly reactions that might cause gradual cognitive decline, is a tall order in the context of clinical trials, Herms said.

First author Severin Filser and colleagues decided to monitor potential effects on synaptic plasticity in mice. The researchers treated normal adult mice with two different BACE inhibitors—SCH1682496 from Merck, or LY2811376 from Eli Lilly—twice daily for 16 days. Herms said the Merck inhibitor is likely similar to MK-8931, which currently is being tested in clinical trials. The Lilly compound was abandoned in a Phase 1 trial due to a retinal pigmentation defect in rats (see Mar 2011 conference story). The German researchers confirmed that the inhibitors reduced Aβ40 in the plasma and cortex of mice in a dose-dependent manner. Neither inhibitor caused weight loss or overt health problems.

To monitor the inhibitors’ effect on dendritic spine growth—a structural measure of synaptic plasticity—the researchers fitted cranial windows onto mice expressing green fluorescent protein (GFP) and monitored the spines on cortical neurons via two-photon imaging. Mice treated with a high dose (hereafter 100 mg/kg) of Merck’s inhibitor formed half as many spines as untreated mice by the last day of treatment; mice treated with the same dose of Lilly’s inhibitor produced about a third fewer spines. Spine growth returned to normal after removal of the inhibitors, and was unaffected by a low dose (hereafter 30 mg/kg) of either inhibitor. Adult BACE1 knockout mice had normal dendritic spine growth that was unaffected by the inhibitors, suggesting that the inhibitors’ effect on spines was specific to BACE1, and that mice deprived of BACE1 throughout life develop compensatory mechanisms.

The researchers then looked for electrophysiological effects of BACE1 inhibition in cortical neurons in brain slices. The high dose of both inhibitors abolished more than half of the spontaneous excitatory postsynaptic currents (sEPSCs). The inhibitors also reduced the frequency and amplitude of miniature EPSCs (mEPSCs), which are triggered by the release of small amounts of neurotransmitters that don’t require action potentials. The authors speculated that this might correlate directly with the reduction in dendritic spines. In BACE1 knockout mice, both sEPSCs and mEPSCs were unaffected by the inhibitors, once again suggesting that the BACE inhibitor was specific. However, the knockouts had a higher frequency of mEPSCs than wild-type mice. This hints that the dynamics of presynaptic vesicle release is altered in BACE1 knockouts, but more research is needed to understand how, Herms said.

The researchers measured long-term potentiation (LTP)—the prolonged period of heightened excitability following stimulation that promotes learning and memory. In hippocampal slices, they found that CA1 neurons from mice treated with high doses of either inhibitor had striking defects, losing their sensitivity to stimulation. Low doses of the inhibitors did not cause these electrophysiological defects. Neither high nor low doses of the inhibitors affected LTP in BACE1 knockout mice; however, neurons in these mice had lower initial responses to stimulation than did wild-type mice.

The effects of the inhibitors on synaptic plasticity played out in the form of mild memory deficits as well. While the treated mice performed normally on measures of anxiety, their performance on the Y-maze test of spatial memory indicated they may have problems with working memory, the authors report.

The study points to the importance of selecting the correct therapeutic dosage in trials, said Filser. “A low dose was able to significantly lower the amount of Aβ, but didn’t cause any of the synaptic deficits we see with the higher dose,” he said. Other researchers weighed in with similar comments (see below). Participants in Merck’s Phase 2/3 trial are receiving daily doses of 12, 40, or 60 mg of MK-8931. However, a direct comparison between mouse and human doses is not practical, Filser said, because the human brain retains BACE1 inhibitors longer than the mouse brain does. He believes researchers should keep these synaptic defects in mind when choosing the lowest effective dose in humans. The human equivalent of the high dose used in these mouse studies would be would be about 8 mg/kg, or roughly 600 mg per day, according to FDA guidance.

In the context of AD, where synapses already have taken a hit from Aβ pathology, losing additional synaptic plasticity may be particularly problematic, Herms said. On the flip side, he added that BACE1 activity ramps up in AD brains, especially in axons co-mingling with plaques. In this situation, perhaps BACE1 inhibition would bring the enzyme’s activity down to normal levels rather than wiping it out completely. This would theoretically cause fewer synaptic side effects—something that Herms’ lab is currently testing in AD mouse models.

Fred van Leuven of the Catholic University of Leuven in Belgium was not surprised by the results. “When you inhibit an enzyme with so many substrates, it is bound to have side effects,” he said. “But if the therapeutic activity is more beneficial than the negative effects, we should push ahead with clinical trials. My attitude is we should try everything we’ve got against this disease, but at the same time be very aware of potential side effects.”—Jessica Shugart


  1. This is an interesting paper, which fits well with earlier findings reported from Phil Wong’s lab in BACE1-deficient mice (Savonenko et al., 2008). After Carmen Birchmeier’s report last year (Cheret et al., 2013), this is now another story demonstrating that BACE inhibitors can have side effects in adult animals. This is a finding we need to consider for clinical trials.

    On the other hand, many of the findings reported in this new paper show up with the high, but not the low, concentration of inhibitors. Thus, with careful dosing of the inhibitor in patients, there may be a chance to avoid these side effects. Additionally, the phenotypes largely do not show up in BACE1-deficient mice compared to wild-type mice, which may be due to compensatory changes in the BACE1-deficient mice, as the authors speculate. Thus, it remains possible that similar compensatory (and side effect-preventing) changes would show up if the inhibitor was given not just for two weeks, but for several weeks or months—similar to the long-term dosing in an AD patient.

    For the future, it will be important to identify the substrate(s) that mediate the spine phenotype and to see whether this is mediated by neuregulin (see, e.g., Savonenko et al., 2008). Additionally, it would be helpful to confirm the phenotype in a second BACE1-deficient mouse model, as not all phenotypes appear to be shared by all BACE1-deficient mouse lines.


    . Alteration of BACE1-dependent NRG1/ErbB4 signaling and schizophrenia-like phenotypes in BACE1-null mice. Proc Natl Acad Sci U S A. 2008 Apr 8;105(14):5585-90. PubMed.

    . Bace1 and Neuregulin-1 cooperate to control formation and maintenance of muscle spindles. EMBO J. 2013 Jun 21; PubMed.

    View all comments by Stefan Lichtenthaler
  2. Over the last years a number of studies have addressed potential on-target liabilities of therapeutic BACE1 inhibition. Many of these studies were based on knockout data alone, or addressed developmental myelination. In contrast, the new study by Herms and colleagues investigates the effects of two drug-like BACE1 inhibitors in adult mice. It reports negative effects on synaptic plasticity. While these relatively subtle effects would not have been picked up in standard toxicology studies, the data suggesting that the BACE1 inhibitor SCH1682496 alters dendritic spine dynamics in vivo appears robust and more concerning than mouse behavior data. Interestingly, the effect is seen only at the higher dose, while Aβ is already substantially reduced at the lower dose.

    Clearly, these results deserve further follow up to see whether they are universally observed with all BACE1 inhibitors and whether they translate to other species and to humans. If that is the case, then dosing to a target level of plasma Aβ reduction would be preferable to “flooring” the signal.

    View all comments by Martin Citron

Make a Comment

To make a comment you must login or register.


News Citations

  1. Blocking BACE—Do Adult Mouse Phenotypes Predict Side Effects?
  2. Barcelona: Out of Left Field—Hit to The Eye Kills BACE Inhibitor

Alzpedia Citations

  1. BACE1

Therapeutics Citations

  1. Verubecestat

External Citations

  1. Phase 2/3 trial
  2. FDA guidance

Further Reading

No Available Further Reading

Primary Papers

  1. . Pharmacological inhibition of BACE1 impairs synaptic plasticity and cognitive functions. Biol Psychiatry. 2015 Apr 15;77(8):729-39. Epub 2014 Oct 29 PubMed.