γ-secretase modulators: hopes and setbacks for the future of Alzheimer’s treatment

Alzheimer’s disease (AD) is the major cause of dementia and a debilitating illness that affects a large proportion of the aged population, yet no effective treatment exists that could halt or reverse its progression. Therapeutic strategies are currently being developed that are aimed at controlling the levels of Aβ amyloid in the brain. Indeed, compelling genetic and experimental evidence indicate that accumulation of Aβ amyloid peptide is directly linked to the etiology of the disease [1]. Aβ was identified as the major component of the amyloid plaques first observed over a century ago by the German pathologist Alois Alzheimer, and that are regarded as the hallmark of the AD brain pathology. Aβ is a polypeptide fragment that derives from the amyloid precursor protein (APP) by a sequential cleavage process involving two proteolytic enzymes, known as β- and γ-secretase. Aβ peptide can undergo a concentration-dependent oligomerization and self-aggregation, which results in the formation of cytotoxic protofibrils that trigger a cascade of events leading to neurodegeneration. Therefore, both β- and γ-secretase represent rational targets for drug research. γ-secretase is an attractive target, since it carries out the final and decisive step in Aβ production. Indeed, γ-cleavage generates heterogenous Aβ peptides, with alternative C-terminal ends. Aβ₄₀ is the principal form produced by normal cells (90%) with lesser amounts of Aβ₄₂(<10%). However, Aβ₄₂ is believed to be the culprit of AD pathogenesis since it is more insoluble and has a greater propensity to aggregate and forms cytotoxic fibrils. Therefore, the quest for γ-secretase inhibitors, particularly for those that specifically inhibit the production of Aβ₄₂, has attracted considerable efforts from the academia and pharmaceutical industry [2]. This has led to a novel class of drugs termed γ-secretase modulators (GSMs).

γ-secretase is a membrane-associated proteolytic machinery consisting of four subunits: presenilin (either PS1 or PS2), nicastrin, anterior-pharynx 1 (APH-1) and presenilin-enhancer 2 (PEN-2) [3]. The catalytic site of the enzyme is embedded in the membrane and consists of two aspartate residues located in transmembrane domains 6 and 7 of presenilin [4]. Since proteolysis takes place within the membrane, the identification of the enzyme and the search for specific inhibitors has been particularly challenging. The architecture of the γ-secretase active site and the enzyme mechanism remain elusive, although the reconstitution of its proteolytic activity by recombinant gene expression has allowed cryo-electron microscopic studies that provide the first information on the complex structure [5]. Accessibility scans are also improving our knowledge of the spatial arrangement of the presenilin nine transmembrane domains [6]. The γ-secretase complex cleaves the transmembrane domain of APP at multiple sites, with a first ε-cut that allows cytosolic release of APP intracellular domain (AICD) and its translocation to the nucleus for signaling; at least two further cuts lead to the production of Aβ peptides. The search for γ-secretase inhibitors (GMI) through the design of substrate analogues, and screening of libraries of aspartyl protease inhibitors and other classes of drugs has provided highly potent molecules with subnanomolar activity. However, a major hurdle in the development of GMI has been the finding that the same, or a similar γ-secretase activity is implicated in processing of a subset of type I receptors, including Notch [7] (which is involved in embryonic development and cell differentiation) to release transcription-active fragments that play a role in signaling and, therefore, has raised concerns about the therapeutic value of γ-secretase inhibition for AD treatment. Animal trials of the first generation of GMI have shown considerable detrimental side effects, including gastrointestinal toxicity and interference with B- and T-lymphocyte maturation owing to inhibition of Notch processing and signaling.

The finding that some NSAIDs behaved as GSMs opened a new promising avenue. NSAIDs, such as aspirin and ibuprofen, were first used in the clinic for their potential protective effect against inflammation of the brain that is commonly associated with AD. A report that ibuprofen could reduce plaque deposition and brain inflammation in a mouse AD model [8] sparked excitement and initiated laboratory studies to understand its molecular mechanism of action. NSAID effects were thought to result from COX inhibition. Surprisingly, some selected NSAIDs, specifically, ibuprofen, indomethacin and sulindac sulfide, were found to preferentially decrease Aβ₄₂ production from cultured cells with parallel increased production of shorter, less aggregating Aβ₃₈ peptide and without significant perturbation of Notch cleavage [9]. These inhibitors demonstrated noncompetitive inhibition kinetics, consistent with inhibition of γ-secretase activity by an allosteric mechanism and binding to a site distinct to that of the transition-state analogues [10]. Fluorescence energy transfer studies demonstrated that ibuprofen and naproxen affected the proximity of APP and PS1, and altered the relative position of the two PS1 active site aspartates [11]. R-flurbiprofen (tarenflurbil or flurizan), a compound related to ibuprofen but devoid of COX-2 inhibition offered a very attractive therapeutic potential [12]. Chronic administration of this drug to an APP transgenic mouse model was able to rescue learning deficits [13].

The concept of GSMs was further extended when the antilipidemic agent fenofibrate and the COX-2-selective NSAID celecoxib (a drug then prescribed for the treatment of arthritis) were found to have a reverse effect to flurbiprofen and to promote the production of Aβ₄₂[14]. This suggested that pharmacological manipulation could be used to alter the cleavage precision of γ-secretase. To design more specific drugs, it was necessary to uncover the sites where NSAID bound to the γ-secretase complex. For this purpose, photoreactive derivatives of fenofibrate (Aβ₄₂ raising molecule) and flurbiprofen (Aβ₄₂ lowering molecule) were designed for crosslinking to the γ-secretase complex [15]. A similar technique had been used previously to demonstrate that the presenilins contained the catalytic site of γ-secretase. Unexpectedly, these NSAIDs could not be crosslinked to any of the four γ-secretase complex subunits. Instead, these could be crosslinked to APP C100 C-terminal fragment, the substrate of γ-secretase cleavage and direct precursor to Aβ. Suggesting that the drugs were interacting with the substrate rather than the enzyme, this result was puzzling but also appealing as it could open a new therapeutic window. Indeed, targeting individual substrates rather than the polyvalent γ-secretase could preclude many side effects. The fact that the juxta-membrane region of the ectodomain of APP confers some specificity for recognition by γ-secretase and cleavage at the γ-site further supports the potential validity of this approach [16]. However, whereas potent inhibitors can be tailored to fit into the defined catalytic core of an enzyme, the structure of a substrate near its cleavage site is usually flexible, thus achieving a tight interaction may be difficult. This problem would explain why NSAID analogues have so far proved to be weak inhibitors. Furthermore, substrate targeting molecules could alter the function of APP by impeding its molecular interactions [17].

The first clinical trial of a NSAID in AD patients was halted because of secondary effects caused by COX-2 inhibition that could not be tolerated in chronic treatment at the high dose prescribed [101]. Myriad has recently tested R-flurbiprofen (tarenflurbil), a second generation of NSAID devoid of COX-2 inhibitory activity, and reported positive results of the Phase II trials that indicate the drug is well-tolerated [18]. Unfortunately, the results of the Phase III trials that have just been disclosed at the International Conference on Alzheimer’s Disease in Chicago are disappointing, showing no significant amelioration of patients receiving the treatment compared with those given a placebo [102]. BBB penetrance appears to be the major problem of the drug, as it is almost totally recovered in the blood circulation.

Although the therapeutic potential of NSAIDs for AD treatment seems to be compromised at present, the concept of γ-secretase modulation remains the most valid and promising intervention to target this enzyme with minimal side effects. A recent study demonstrated that membrane lipid composition influences the precision of the γ-secretase cleavage [19]. Thus, some drugs that interfere with cellular membrane architecture may be valuable, and the example that sulindac sulfide decreases the production of Aβ from cultured cells by redistributing γ-secretase out of lipid rafts indicates that this is feasible [20]. Further investigation of endogenous γ-secretase may also reveal natural modifiers of its activity. In this regard, TMP21 interaction with presenilin modulates Aβ production but does not affect the preceding cleavage step that releases AICD [21]. In addition, the demonstration that γ-secretase complexes with alternative subunit compositions can produce different proportions of Aβ₄₂ and Aβ₄₀ amyloid peptides offers yet another new therapeutic lead [22]. In conclusion, although recent results of clinical trials have shattered some hopes that a γ-secretase modulator would soon become available to treat AD, some exciting fresh findings offer new perspectives to continue the search for GSMs and other modifiers of γ-secretase.

Financial & competing interests disclosure

G Evin is supported by the National Health and Medical Research Council of Australia (grant 400073). The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.