Dale Bredesen at the Buck Institute, Novato, California, noted that the paper provides further evidence that APP functions as a single-pass transmembrane G protein-coupled receptor, and also lays out a detailed signaling mechanism. “I think this opens up a whole new way to evaluate the effects mediated by APP, and also provides an additional suggestion that APP signaling could be quite important in normal plasticity,” he told ARF. Bredesen was not involved in the work.

In contrast, in the December 14, 2011, European Molecular Biology Organization (EMBO) Molecular Medicine, researchers led by Luciano D’Adamio at the Albert Einstein College of Medicine, Bronx, New York, reported that inhibiting BACE1, but not γ-secretase, rescues synaptic plasticity and memory in a mouse model of dementia. Together, these studies underscore how APP fragments may appear to have varied effects. They also imply that researchers should look beyond Aβ for the causes of AD pathology and for therapeutic targets.

Membrane-bound APP can be cut on the extracellular side of the cell membrane by either α- or β-secretase, creating a soluble shed portion (sAPPα or sAPPβ) and a membrane-bound fragment with an intracellular tail (α-CTF or β-CTF). The β-CTF fragment is then snipped by γ-secretase to generate Aβ and an APP intracellular domain (AICD) that may travel to the nucleus. It is not yet clear what all these fragments do, although there are clues. For example, numerous studies have shown that APP plays a physiological role in modulating neurite outgrowth (see, e.g., ARF related news story; ARF news story; Small et al., 1999; and Young-Pearse et al., 2008), but the exact mechanisms are unknown. Since synapses form on dendritic spines, neurite growth is intimately linked to synaptic health, making this pathway particularly pertinent to AD.

Parent and colleagues focused on intracellular signaling by membrane-bound APP fragments. First author Carole Deyts designed a fusion protein, mAICD, by joining the intracellular portion of human APP to membrane-bound motifs. Deyts and colleagues transfected various cell cultures (including neuroblastoma cells, primary mouse cortical neurons, and immortalized hippocampal neurons) with the construct, and found that transfected cells grew longer and more complex neurites. The findings contrast with previous studies that highlighted axon and tau toxicity by BACE1-cleaved APP products, including N-terminal (see, e.g., ARF related news story) and C-terminal fragments (see ARF related news story and ARF news story).

Looking for a mechanism, the authors saw higher levels of cyclic AMP (cAMP) and activated protein kinase A (PKA) in transfected cells. Membrane-bound adenylate cyclase produces cAMP, which is known to activate PKA. Inhibiting either adenylate cyclase or PKA in these cultures abolished the neurite outgrowth, the authors report, indicating that this pathway is crucial for mediating mAICD’s effects. PKA modifies numerous downstream proteins. It activates cAMP-response element binding protein-1 (CREB), a key protein for learning and memory, and inactivates the kinase GSK-3β. Researchers have associated active GSK-3β with neurite retraction (see Hur and Zhou, 2010), and it also phosphorylates tau. The authors confirmed that CREB activity increased, and GSK-3β activity decreased, in cells overexpressing mAICD.

Deyts and colleagues next explored how mAICD might activate adenylate cyclase. They showed by co-immunoprecipitation that mAICD binds the G protein GαS, which is known to stimulate cAMP signaling. The authors identified a G protein-binding site in the intracellular portion of APP, and showed that mutating one amino acid in this motif prevented mAICD from interacting with GαS, and also blocked neurite growth. Expression of a dominant negative GαS protein likewise inhibited mAICD-induced neurite extension. The results imply that membrane-bound APP fragments act as G protein-coupled receptors to promote neurite growth, suggest the authors.

To show that physiological membrane-bound APP fragments have the same effect as mAICD, the authors treated neuroblastoma cultures with a γ-secretase inhibitor, which causes APP-CTF fragments to build up. Neurite length increased in treated cells, and again the effect could be blocked by inhibiting adenylate cyclase, indicating the same pathway was at work as in mAICD-laced cells. Finally, in conjunction with γ-secretase inhibition, the authors blocked either β or α cleavage to allow either α-CTF or β-CTF to accumulate, but found that neurites grew equally well in either condition, suggesting both fragments can stimulate outgrowth.

Parent told ARF she is interested in the therapeutic possibilities of targeting G proteins in AD, pointing out that many such drugs are available from cancer research (see Thathiah and De Strooper, 2011). She is currently investigating whether stimulating signaling through membrane-bound APP fragments can help restore memory in animal models of AD. “People have been paying a lot of attention to Aβ production, and thinking APP is bad for you. I think APP is actually very good [for you] if it works well,” Parent suggested.

Intriguingly, APP was previously shown by a Japanese group to bind to a different G protein, Go (see Nishimoto et al., 1993). In opposition to GαS, Go signaling decreases cAMP levels and can lead to cell death. Nishimoto and colleagues showed that three AD-associated mutations in APP all increased Go-mediated cell death. Bredesen noted that the new paper complements Nishimoto’s work. “There has been the suggestion for years that APP can mediate both neurite retraction and neurite extension. We’ve called it a molecular switch,” he said. These papers add to the evidence that APP can have either beneficial or harmful effects, depending on which G proteins are present. The next step will be to show which ligands activate the positive and negative effects of APP, Bredesen added.

In the second paper, D’Adamio and colleagues provide more evidence for the harmful effects of signaling by BACE1 cleavage products, using a mouse model of familial Danish dementia (FDDKI). Danish dementia is caused by loss of the Bri2 protein, which normally binds to APP and prevents its cleavage. To make the mouse, the researchers inserted the Danish mutation into the endogenous Bri2 mouse gene. D’Adamio noted that, unlike many AD mouse models, this animal does not overexpress APP; instead, it builds up APP cleavage products. Joint first authors Robert Tamayev and Shuji Matsuda found they were able to rescue synaptic and memory deficits in FDDKI mice by inhibiting BACE1, but not by inhibiting γ-secretase. This implies that a BACE1 cleavage fragment (β-CTF or secreted sAPPβ) is the toxic entity, rather than Aβ.

The data also hint that inhibiting γ-secretase could be harmful, as that would cause BACE1 cleavage products to build up. Intriguingly, some previous papers have suggested that AD-causing presenilin mutations might be loss-of-function mutations, fitting with the idea that lack of γ-secretase could lead to memory problems (see, e.g., Shen and Kelleher, 2007; ARF related news story on Qiang et al., 2011). A clinical trial of the γ-secretase inhibitor semagacestat was halted because patients on the drug did worse than those on placebo (see ARF related news story and ARF news story).

The authors developed a short peptide, called modulator of β-cleavage of APP (MoBA), that binds to APP and prevents BACE1 cleavage. MoBA does not prevent α-secretase cleavage, nor does it bind to β-CTF, the authors report. MoBA had comparable effects to BACE1 inhibition in the FDDKI mice, restoring memory and synaptic plasticity. This peptide might be a therapeutic alternative to BACE1 inhibitors, D’Adamio suggested, as it binds only to APP and does not affect other BACE1 substrates. In ongoing work, preliminary data indicate that MoBA is as effective in restoring synaptic plasticity in APP/PS1 mice as it is in the FDDKI mice, D’Adamio added.—Madolyn Bowman Rogers.

References:
Deyts C, Vetrivel KS, Das S, Shepherd YM, Dupré DJ, Thinakaran G, Parent AT. Novel GαS-protein signaling associated with membrane-tethered amyloid precursor protein intracellular domain. J Neurosci. 2012 Feb 1;32(5):1714-29. Abstract

Tamayev R, Matsuda S, Arancio O, D’Adamio L. β- but not γ-secretase proteolysis of APP causes synaptic and memory deficits in a mouse model of dementia. EMBO Mol Med. 2011 Dec 14. Abstract