For at least a decade, BACE1 has been thought to play a role in AD. BACE1 levels rise in sporadic AD patients (see Yang et al., 2003) and under various types of cellular stress (see Wen et al., 2004; Zhang et al., 2007; Tamagno et al., 2002). Overexpression of the enzyme also worsens learning and memory in transgenic mice that make human APP (see Cole and Vassar, 2007). Many assumed the impairment was due to a boost in Aβ production, but BACE1 cuts many substrates besides the Aβ precursor protein (APP) (see ARF related news story). Might the enzyme have Aβ-independent effects as well?

Researchers led by Eliezer Masliah at the University of California, San Diego, in La Jolla, observed striking memory impairments in mice that overexpressed human BACE1, but not human APP (see Rockenstein et al., 2005). Since mouse APP fragments are not terribly toxic, the team thought BACE1 impaired learning and memory through another pathway. Masliah teamed up with Xu to look for one.

First author Yaomin Chen and colleagues worked backwards, starting with the genetic underpinnings of learning and memory. They knew that cAMP response element binding protein (CREB) phosphorylation was crucial for these cognitive abilities, so they tested to see if BACE1 suppressed it. Sure enough, BACE1 overexpression in mouse N2a neuroblastoma cells and rat primary cortical neurons reduced phosphorylated CREB. BACE1 also quashed protein kinase A (PKA), which phosphorylates CREB, and cyclic adenosine monophosphate (cAMP), which activates PKA. The researchers suggest BACE exerts its effects at the top of that chain by interacting with adenylate cyclase, a 12-transmembrane-domain protein. They found that the cyclase co-immunoprecipitated with BACE1 in N2a cell and wild-type mouse brain lysates.

In-vivo experiments supported these results. In BACE1 transgenic mice, the cAMP/PKA/CREB pathway faltered in the brain compared to wild-type controls. But in BACE1 knockout mice (see ARF related news story on Luo et al., 2001), the cAMP/PKA/CREB pathway revved up, suggesting that a lack of the enzyme lifts the pathway’s brakes. Further, BACE1 transgenic mice explored an open-field arena more than did controls, suggesting impaired recall in the BACE1-overexpressing animals.

Is an Aβ boost at the root of this disruption? In engineered mouse embryonic fibroblast cells that produce no Aβ, human BACE1 overexpression still depleted phosphorylated CREB levels. Based on this, the authors claim that Aβ is likely not involved. Neither is enzymatic activity, it seems. Chen mutated one of two catalytic aspartic acids on BACE1, rendering it inactive, yet in N2a cells it reduced cAMP levels, PKA activity, and CREB phosphorylation just as well as wild-type BACE1. These reductions occurred even when a commercial BACE1 inhibitor was applied to the N2a cells, Xu said. The findings suggest the protease does not cleave adenylate cyclase. Xu noted that most of the protease's substrates contain only one transmembrane domain.

However, if aspartic acid mutants formed dimers, they might reconstitute catalytic activity (see Schmechel et al., 2004), said Stefan Lichtenthaler, German Center for Neurodegenerative Diseases, Munich. In that case, BACE1 may not necessarily cleave adenylate cyclase, but another substrate that acts on the cAMP/PKA/CREB pathway, he told Alzforum. However, it is also conceivable that BACE1 has a non-proteolytic function, he said. “The [CREB] pathway changes are impressive and quite robust,” added Lichtenthaler. “The mechanism is not yet fully clear, but this is an excellent basis for future follow-up.” What else could BACE1 be doing to adenylate cyclase if not cleaving it? It may be binding and changing the conformational shape, suggested Xu, though he agreed it will be hard to investigate the exact mechanism of action on a protein such as adenylate cyclase, which makes 12 passes through the cell membrane. Eric Parker, Merck Research Laboratories, Kenilworth, New Jersey, did caution that even though adenylate cyclase co-immunoprecipitated with BACE1, ‘”it is quite easy to get artifactual protein-protein interactions in these experiments." While adenylate cyclase's role is slightly tenuous, the findings are interesting, he said, and warrant rigorous replication.

The results may have implications for AD therapy. “The study suggests BACE inhibition may not be as complete as we expect,” said Xu. An inhibitor might fix the Aβ problem, but miss the cAMP/PKA/CREB pathway entirely, he said. It is possible that BACE inhibitors could sufficiently change the conformation of the protein to interfere with an adenylate cyclase interaction, suggested Parker. However, even if inhibitors only modulated the Aβ pathway, they should still hit the major function of BACE1 in AD, he told Alzforum. "BACE inhibitors’ ability to inhibit Aβ production will almost certainly be the key way these drugs work,” he said.

Nevertheless, by illuminating another mechanism of BACE1 action, this new work may help drug developers. It is important to understand all the possible routes by which an inhibitor will affect a cell, especially when its target has so many substrates, Xu said. Knowing the various pathways takes some of the guesswork out of a drug’s potential side effects.—Gwyneth Dickey Zakaib.

Reference:
Chen Y, Huang X, Zhang Y, Rockenstein E, Bu G, Golde TE, Masliah E, Xu H. Alzheimer’s β-Secretase (BACE1) Regulates the cAMP/PKA/CREB Pathway Independently of β-Amyloid. J Neurosci 2012 Aug 15; 32(33):11390-11395. Abstract