γ-Secretase as a drug target for familial Alzheimer's disease: the road less traveled

Alzheimer's disease is a devastating neurodegenerative disorder that involves progressive loss of memory and cognitive function, affecting more than 35 million people worldwide, primarily those over 65 years of age [1]. Despite intense efforts over three decades, major gaps in our understanding of the molecular basis of the disease remain and no demonstrated disease-modifying medications are available. The complexity of Alzheimer's disease has presented serious challenges to the elucidation of pathogenic mechanisms and the discovery and development of effective therapeutics. A simpler path to understanding the disease biology and discovering new drugs is presented by early-onset hereditary forms of Alzheimer's disease. This familial Alzheimer's disease (FAD) is associated with dominantly inherited mutations that cause disease onset before age 60, some even before age 30 [2]. Despite its midlife onset and completely hereditary nature, FAD is quite similar in pathology, presentation and progression to the common sporadic Alzheimer's disease of the elderly.

FAD mutations are found in three genes, encoding APP and PSEN1 and PSEN2 [3]. APP is a single-pass membrane protein that undergoes successive proteolysis, first by β-secretase and then by γ-secretase, to produce the amyloid β-peptide (Aβ) that deposits in the brain in Alzheimer's disease. The presenilins are multipass membrane proteins that comprise the catalytic component of the γ-secretase protease complex. Thus, FAD mutations are found in the substrate and in the enzyme that produces Aβ. Alteration of APP proteolysis by γ-secretase is closely associated with FAD pathogenesis and pharmacological correction of this aberrant proteolysis would be expected to delay or prevent disease onset.

γ-Secretase processing of the APP substrate, however, is itself complex. This proteolysis occurs within the transmembrane domain of APP and the active site of γ-secretase is likewise embedded in cellular membranes [4]. That is, hydrolysis takes place within the hydrophobic environment of the lipid bilayer. The transmembrane substrate must move laterally into the internal, water-containing, active site of presenilin, with the water entering through pores exposed to the aqueous environment. Moreover, the γ-secretase complex cleaves the APP transmembrane domain processively, first releasing the APP intracellular domain and producing Aβ peptides of 48 or 49 residues in length (Aβ48 and Aβ49) that are then trimmed down, generally in intervals of three amino acids, to shorter secreted forms [5]. The two pathways to Aβ peptide production are Aβ49→Aβ46→Aβ43→Aβ40 and Aβ48→Aβ45→Aβ42→Aβ38.

Until relatively recently, the focus of Alzheimer's research has been almost exclusively on only two of these peptides: Aβ40 and Aβ42. The former is the major secreted form of Aβ and the latter is much more prone to aggregation and is the principal component of the cerebral plaques characteristic of the disease. FAD mutations can change the production or properties of Aβ and most have been reported to increase the Aβ42/Aβ40 ratio, thereby seeding aggregation. As a consequence, drug discovery efforts were focused on γ-secretase inhibitors (GSIs) to generally lower Aβ peptide production and γ-secretase modulators (GSMs) to specifically lower Aβ42 levels.

GSIs failed in human trials for the treatment of Alzheimer's disease, causing serious adverse events associated with blocking proteolysis of another γ-secretase substrate, the Notch1 receptor [6,7]. γ-Secretase processing of Notch1 is part of an evolutionarily conserved signaling pathway essential for cell differentiation. More concerning, GSIs caused cognitive worsening in clinical trial patients with Alzheimer's disease, findings that raised concerns about the amyloid hypothesis of Alzheimer's pathogenesis. GSMs appear to be safe [8]; however, efficacy in slowing disease progression has yet to be demonstrated for any GSM or for any clinical candidate targeting Aβ42 or amyloid plaques [9].

Although representing <1% of all Alzheimer's cases, the monogenic early-onset FAD cases present an opportunity to elucidate disease biology and explore new approaches to targeting γ-secretase for drug discovery. In contrast to the complexity of sporadic late-onset Alzheimer's disease, FAD is clearly caused by dominant missense mutations in the substrate and enzyme that produce Aβ. The challenge is to investigate the effects of the mutations on the complex proteolysis of γ-secretase on its APP substrate and then develop assays to identify small molecules that correct the aberrant processing. A similar strategy has already been proven successful for another genetic disorder involving mutations in a membrane protein; cystic fibrosis is caused by mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) that reduce the proper functioning of this chloride channel [10]. Small molecules that stabilize or otherwise rescue the function of mutant CFTR are approved drugs to treat cystic fibrosis [11].

A recent study quantified the effects of 14 FAD mutations in the APP transmembrane domain on all the proteolytic processing steps carried out by γ-secretase [12]. The analytical method employed liquid chromatography coupled with tandem mass spectrometry to measure the levels of the coproducts of each processing step. This method was quite similar to that utilized for the first demonstration of processive proteolysis of the APP substrate by γ-secretase [5]. Not all FAD mutations elevated the Aβ42/Aβ40 ratio compared with that generated from the wild-type APP substrate; however, all elevated Aβ variants of 45 residues and longer due to deficiencies in the first and/or second trimming steps. These longer Aβ peptides contain most of the APP transmembrane domain, are membrane-anchored and not secreted and the elevation of these peptides by all of the 14 tested APP FAD mutations raises questions about possible pathogenic roles of these understudied Aβ peptides, which we call ‘dark amyloid’. To date, a full analysis of the effects of any presenilin FAD mutation on all γ-secretase processing steps has not been conducted. However, the effects of 138 FAD mutations in PSEN1 on the Aβ42/Aβ40 ratio similarly found that many did not elevate this ratio, which has been long held as critical to triggering the disease [13].

Whatever the identity of the pathogenic entities and neurotoxic pathways, correcting the dysfunctional FAD proteolytic processing to yield a profile of Aβ peptides similar to that seen with wild-type γ-secretase and APP substrate is expected to be therapeutic. The ability of a single molecule to correct processive proteolysis for all FAD mutations seems unlikely. Nevertheless, demonstrating the ability to delay disease onset or slow disease progression for one or more FAD mutations would provide proof of principle and justify the search for other small molecules to correct defective proteolysis for other FAD mutations. Indeed, the first effective therapeutic for cystic fibrosis was approved for only a single mutation in CFTR, albeit one that occurs with much higher frequency than other disease-causing CFTR mutations [14].

With tens of millions afflicted by sporadic, late-onset Alzheimer's disease worldwide, the temptation to focus on drug discovery for this most common form of dementia is strong. Nevertheless, three decades of intense efforts have not yielded a single effective disease-modifying therapy. While FAD afflicts fewer patients and can be considered an orphan disease, its monogenic nature, affecting the enzyme and substrate that produce Aβ, makes it a much more tractable problem. Moreover, as FAD strikes in midlife, each successfully treated patient could potentially look forward to decades of happy and productive life. Finally, what is learned in the study of FAD is likely to inform us, at least in part, about mechanisms and targets for late-onset Alzheimer's disease. The example of drug discovery for cystic fibrosis provides a roadmap to drug discovery for FAD. Perhaps it is time we took this road less traveled.