Amyloid-β (Aβ) precursor protein (APP) has something of a mixed reputation. Mutations in the protein, or in its scissor enzyme presenilin, cause early onset familial Alzheimer disease (eFAD). On the other hand, learning and memory are also compromised by lack of APP, suggesting the protein and its processing may support neural activity. In the April 30 PNAS online, a collaboration of scientists led by Karen Ashe at the University of Minnesota, Minneapolis, reports that mice overexpressing normal APP enjoy enhanced learning and memory; however, those benefits depend both on β-site cleavage and on the APP intracellular domain (AICD). The findings may rekindle debate on the physiologic role of AICD, and also raise a cautionary flag on efforts to target BACE1 for treatment of Alzheimer disease.

The authors uncovered the role of AICD by ablating BACE1 in transgenic mice overexpressing wild-type human APP. These TgAPP mice produce about sixfold more APP than controls. They have no Aβ pathology; they do, however, outperform controls in spatial learning and memory tests (see Westermann et al., 2002). To investigate if those superior faculties relate to APP processing, first author Huifang Ma and colleagues compared the performance of TgAPP mice missing none, one, or both copies of the BACE1 gene.

The scientists first confirmed the earlier finding that BACE+/+ TgAPP mice perform better than non-transgenics in the Morris water maze. This seems to be due to altered neural activity, because hippocampal slices from the animals exhibited enhanced synaptic plasticity when tested for two specific forms of long-term potentiation (LTP). Primed LTP (P-LTP), elicited by a weak single tetanic train of pulses and followed by four strong tetanic trains, was about 30 percent higher in TgAPP tissue than in control slices. Primed post-tetanic potentiation (P-PTP) is a form of short-term synaptic plasticity, according to the paper. It was enhanced by about 20 percent. By contrast, homozygous BACE1 knockout mice performed like wild-type in the water maze test, and their hippocampal tissue showed normal P-LTP and P-PTP. Knocking out a single copy of BACE was also sufficient to abolish both the spatial learning improvement and the enhanced synaptic plasticity, according to the study.

What is the role of BACE1 in enhanced learning and memory? BACE1 cleavage of APP yields a soluble N-terminal APP fragment, sAPPβ, and exposes the C-terminal end to γ-secretase cleavage, which yields Aβ and AICD. (The competing α-secretase pathway yields sAPPα instead, and prevents formation of Aβ.) The authors found that Aβ, sAPPα, and AICD are all elevated in TgAPP compared to non-transgenic strains, yet a single BACE1 knockout, which is sufficient to cut the memory and LTP enhancements, only affected AICD, which dropped to undetectable levels in either single- or double-BACE1 knockouts.

Curiously, the researchers found no change in Aβ levels in single BACE1 knockouts, despite complete loss of AICD. The authors note that this is not entirely without precedent, since mutation or inhibition of γ-secretase can have different effects on AICD and Aβ levels. But since γ-secretase is unaltered in this case, an alternative explanation is warranted. “Potential differences in kinetics of AICD and Aβ degradation may result in a more significant drop in AICD levels than in Aβ levels when one BACE1 gene is ablated,” write the authors. They also suggest that the concentration, rather than the activity of BACE1, may modulate AICD-generating ε cleavage of γ-secretase.

These results suggest that BACE1-derived AICD may be important for the enhanced learning and memory in mice overexpressing APP, but it is unclear how this may happen. AICD may be much more than a byproduct of APP processing. The intracellular domain can form complexes with other proteins, including Fe65 and Tip60, which may activate transcription (see related ARF live discussion). Moreover, a simple point mutation in AICD that abolishes a caspase site appears sufficient to protect mice against mutant human APP (see ARF related news story). The authors acknowledge that because TgAPP mice overexpress APP, the effects they observed may not reflect the physiological activity of APP. But, just in case, they offer a recommendation for drug developers. “To ensure that experimental therapies do not prevent BACE1-mediated facilitation of memory by APP, preclinical studies of experimental β-secretase inhibitors should be done not only in animal models of AD, but also in natural animals to evaluate their effects on normal cognitive function,” suggest the authors.—Tom Fagan


  1. Yet another fascinating report from the Ashe laboratory is published in PNAS. Here, Karen Ashe and colleagues demonstrate that BACE1 cleavage of APP is needed for the enhancement of memory and the activity-dependent synaptic plasticity. A mouse model is used that overexpresses wild-type human APP at six times the endogenous level. These mice do not produce Aβ oligomers, or plaques, and exhibit enhanced spatial memory. Biochemically, these mice have more Aβ, APPsα, and AICD when compared to non-transgenic controls. So, which of the APP fragments is responsible for the enhanced memory? Ma et al. provide evidence that the enhanced synaptic plasticity is diminished by the lack of one or two copies of BACE1. Ablation of BACE1 leads to a parallel decrease in AICD but not to changes in APP or other APP fragments including Aβ and APPsα. The major novel conclusions of the paper are as follows:

    1. AICD plays an important role in synaptic plasticity and, thus, memory.

    2. The beneficial effects are not caused by Aβ or APPsα, as previously reported.

    3. We need to exercise caution when designing BACE1 inhibitors to reduce Aβ load in AD patients.

    These are exciting findings that could be open to additional interpretations. For example, BACE1 also cleaves neuregulin, and that has been shown to affect synaptic plasticity and myelination (Harrison and Law, 2006; Willem et al. 2006; Hu et al., 2006). In addition, the increased levels of APPsα in the WT APP mice suggest that the αCTF fragment levels are increased. These are the substrates for γ-secretase to produce AICD by cleaving at the ε site. So, why then would lack of BACE1 affect the γ-secretase cleavage of αCTF fragments? As always, an intriguing article opens many more stimulating questions.


    . Neuregulin 1 and schizophrenia: genetics, gene expression, and neurobiology. Biol Psychiatry. 2006 Jul 15;60(2):132-40. PubMed.

    . Control of peripheral nerve myelination by the beta-secretase BACE1. Science. 2006 Oct 27;314(5799):664-6. PubMed.

    . Bace1 modulates myelination in the central and peripheral nervous system. Nat Neurosci. 2006 Dec;9(12):1520-5. Epub 2006 Nov 12 PubMed.

  2. This intriguing study from Karen Ashe’s lab shows that mice overproducing wild-type human APP (TgAPP) display enhanced spatial memory in the Morris water maze. To determine which of the proteolytic products of APP plays a role in this process, the TgAPP mice were crossed with BACE1 knockout mice. TgAPP mice bearing a single targeted allele of BACE1 no longer displayed enhanced memory. The only APP fragment that was reduced in the TgAPP/BACE1 heterozygote mice when compared to TgAPP controls was AICD. These data clearly implicate BACE1 activity in APP-dependent enhanced memory and suggest that AICD plays a role in memory, thus providing the first physiological evidence for AICD function in vivo.

    Electrophysiology studies were also carried out to determine whether synaptic plasticity in the Schaffer collateral pathway of the hippocampus correlated with the respective behaviors of the mice in this spatial memory task. Primed long-term potentiation (P-LTP) was enhanced in TgAPP mice, but LTP was unchanged relative to non-transgenic controls. This suggests that the synaptic plasticity affected by APP overproduction was dependent on prior synaptic activity. These changes in primed LTP were no longer observed in mice bearing a single targeted allele of BACE1. The implication from these data is that AICD plays a role in this form of synaptic plasticity. We are left wondering whether AICD levels change in response to synaptic activity. Specifically, are AICD levels affected differently in primed LTP versus LTP, and does AICD modulate transcriptional events associated with memory? These are only a couple of the questions that arise from this fascinating report.

    View all comments by Suzanne Guenette
  3. I wonder why there is no mention of the neuropathology, including neuritic plaques and neurofibrillary tangles, that develops in most, but not all (see Argellati et al., 2006) individuals afflicted with Down syndrome. BACE1 and BACE2 levels have been reported to be elevated in the brains of Down's patients (Sun et al., 2006; Motonaga et al., 2002). Duplication of the APP locus has been found in families that lack clinical features of DS, but develop EOAD (see Cabrejo et al., 2006). This is an interesting paper, but it again reveals that the value of mice as models to study human disease is limited.

    AICD signaling to the nucleus, although frequently compared to NICD nuclear signaling, is still a matter of debate. This is largely due to the number of conflicting reports throughout the literature, including on AICD stabilizing proteins and the expression of putative AICD-targeted genes. In keeping with this, adaptor proteins (e.g., Fe65) can apparently activate transcription without the need of AICD (see Hebert et al., 2006; Cao and Sudhof, 2004).

    So what is it about AICD? Could it primarily function as a scaffolding protein or recruitment factor for APP-binding proteins? Kinases? Phosphatases? Cytoskeleton? Proteins that have such an incredibly short half-life are typically the most critical to the health of the cell. As such, they contribute to the control of signal transduction pathways, cell-cycle control, apoptosis, antigen processing, differentiation, and surface receptor desensitization, to name just a few. Maybe AICD is a nuclear signaling molecule with dual functionality, and we just haven't looked hard enough or in the right spot. Maybe AICD is one of many protein by-products involved in nuclear signaling within neurons, and is rapidly destroyed in order to maintain a mitotic block through gene repression or expression. A second role would still require rapid degradation in proliferating cells to allow for growth. Either way, it would make it an elusive molecule to detect. In a scenario where AICD generation from mAPP is no longer tightly regulated by α- or β-secretase activities (e.g., in FAD mutants), the genes would become deregulated and the neurons begin to slowly cycle into an abyss.

    I imagine this scenario: deregulated secretase activities such that a cell cycle checkpoint is breached into G1/S phase, expression of cell cycle proteins in the affected neurons increases, recruitment of microglia and increased ROS's, Aβ levels slowly increase, the neuron reaches G2 where it needs to prepare for division and so microtubules and MAPs increase, G2 continues very slowly, tau phosphorylation and NFTs increase, but the neuron is genetically programmed not to be in this predicament. It cannot reach M phase. Cell cycle markers accumulate and trigger apoptotic pathways. And that's it. The neuron has no choice but to die, leaving behind a dark hole full of protein aggregates, where a memory once lived.

    I recommend this paper.


    . Evidence against the overexpression of APP in Down syndrome. IUBMB Life. 2006 Feb;58(2):103-6. PubMed.

    . Increased BACE1 maturation contributes to the pathogenesis of Alzheimer's disease in Down syndrome. FASEB J. 2006 Jul;20(9):1361-8. PubMed.

    . Elevated expression of beta-site amyloid precursor protein cleaving enzyme 2 in brains of patients with Down syndrome. Neurosci Lett. 2002 Jun 21;326(1):64-6. PubMed.

    . Phenotype associated with APP duplication in five families. Brain. 2006 Nov;129(Pt 11):2966-76. PubMed.

    . Regulated intramembrane proteolysis of amyloid precursor protein and regulation of expression of putative target genes. EMBO Rep. 2006 Jul;7(7):739-45. PubMed.

    . Dissection of amyloid-beta precursor protein-dependent transcriptional transactivation. J Biol Chem. 2004 Jun 4;279(23):24601-11. PubMed.

Make a Comment

To make a comment you must login or register.


Webinar Citations

  1. The Functional Roles of APP Cytoplasmic Domain—Conflicts and Consensus

News Citations

  1. Paper Alert: Pathology Reversed by Abolishing APP Caspase Site

Paper Citations

  1. . The relationship between Abeta and memory in the Tg2576 mouse model of Alzheimer's disease. J Neurosci. 2002 Mar 1;22(5):1858-67. PubMed.

Further Reading

Primary Papers

  1. . Involvement of beta-site APP cleaving enzyme 1 (BACE1) in amyloid precursor protein-mediated enhancement of memory and activity-dependent synaptic plasticity. Proc Natl Acad Sci U S A. 2007 May 8;104(19):8167-72. PubMed.