The production of toxic Aβ peptides from amyloid precursor protein (APP) proceeds via the sequential cleavage of the extracellular domain by β-secretase, followed by an intramembrane cut by γ-secretase. However, if α-secretase finds APP first, it can lop off the β-secretase cleavage site, simultaneously blocking production of Aβ and releasing a neuroprotective soluble APP fragment. Activation of muscarinic acetylcholine receptors is known to steer APP down the non-amyloidogenic, α-secretase pathway, but it has been difficult to pin the effect on a specific receptor subtype: There are five different muscarinic receptors (M1-M5) and knowing which one is involved is critical for making selective drugs and minimizing cholinergic side effects. Allan Levey and colleagues at Emory University, Atlanta, Georgia, have now used knockout mice to show that the M1 receptor is responsible for regulating APP processing in vivo. Their work, published in the March 24 Journal of Neuroscience, offers support for the ongoing development of M1-selective agonists as modulators of amyloid load.
At the same time, a second paper raises the question of whether α-secretase processing of APP is as benign as assumed. Ratnesh Lal, University of California, San Diego, and Ruth Nussinov, National Cancer Institute, Frederick, Maryland, present evidence that the small peptide derived from α- then γ-secretase cleavage forms ion channels in cells, reminiscent of pore-forming channels proposed for Aβ peptides. At high concentrations the peptide (known as P3) is toxic to neuronal cells in culture, apparently by causing calcium leakage. That work appears in this week’s edition of PNAS online.
The muscarinic acetylcholine receptor has been a target for AD therapies for decades, based first on the loss of acetylcholine in the disease and the role of M1 receptors in memory and cognition, and then on the finding that the receptors control the processing of amyloid precursor protein to Aβ peptides. Attempts to tweak muscarinic receptors to modify APP processing pointed to M1 as the main regulator, and selective agonists of the M1-type receptor have been shown to reduce amyloid and tau pathology in a mouse model of AD (see ARF related news story on Caccamo et al., 2006). However, the compound used in that study (AF267B) was subsequently shown to also activate M3 receptors (Jones et al., 2008). In the new study, first author Albert Davis and coworkers use a mouse knockout to examine regulation of APP processing by M1. They show that treating cultured mouse neurons expressing human APP with the acetylcholine receptor agonist carbachol increased the production of α-secretase cleavage products, but no such effect occurred in cells from M1 knockout mice. Expressing M1 receptor in knockout cells restored carbachol-induced α-secretase processing.
To look at the effect in vivo, the researchers knocked out M1 in a mouse that overexpresses a mutated human APP. Without M1, the mice showed a three- to fivefold increase in Aβ content in the brain at 16 months of age, and a doubling in plaque number compared to M1-expressing animals. The researchers do not report whether cognition was negatively affected in the animals, either by M1 loss or enhanced amyloid accumulation.
The results suggest that an M1-specific agonist could be doubly beneficial in AD by making up for the loss of acetylcholine and by decreasing Aβ production. Selectivity is important on both counts: Non-specific muscarinic receptor agonists can produce side effects due to peripheral cholinergic effects including gastrointestinal disturbances, changes in blood pressure, and excessive sweating. In addition, Davis and coworkers show that treating neurons from M1 knockout mice with carbachol increased Aβ production, probably via stimulation of other non-M1 muscarinic receptor subtypes. Selective agonists have been hard to come by, although pharma is hard at work on the problem, with recent progress toward promising candidates that are both selective and possess the requisite pharmacokinetic properties for testing in animal models (see ARF related news story on Jones et al., 2008 and ARF related news story on Ma et al., 2009).
But is there be a twist to the story? The peptide products of sequential α- and γ-secretase processing (known as P3, spanning residues 17-40/42) are widely viewed as harmless. However, the Lal study suggests that P3 peptides share properties with their amyloidogenic relatives, including the ability to form channels in the cell membrane that allow calcium to leak into the cell. In fact, calling P3 “non-amyloidogenic” is a misnomer, Lal told ARF. The P3 peptide includes the transmembrane region of Aβ and can form amyloid fibrils (Pike et al., 1995). The new work, led by first authors Hyunbum Jang, Fernando Teran Arce, and Srinivasan Ramachandran, shows that the peptides form channel-like structures in vitro similar to those described for the Aβ peptides (Quist et al., 2005). When reconstituted into lipid bilayers, the peptides assemble into cation-selective channels, and in cells, they appear to mediate calcium influx. The peptides were toxic for neurons, but only at very high concentrations (100 μM). The physiological significance of the findings remains to be seen. While these peptides do appear as a component of amyloid plaques in AD and Down syndrome, it is not clear what concentrations they achieve and whether channel formation plays a role in amyloid toxicity in vitro. One recent report suggests the P3 peptides do not form synaptotoxic oligomers in the way that Aβ does (Dulin et al., 2008). Based on their results, however, Lal and colleagues caution that efforts to detoxify APP by steering processing toward these peptides may be a futile effort.—Pat McCaffrey