Alzheimer disease therapy: Can the amyloid cascade be halted? (original) (raw)

Secretase inhibitors. Research clarifying the metabolic pathways that regulate Aβ production has revealed that the secretases that produce the Aβ may be good therapeutic targets since inhibition of either β- or γ-secretase decreases Aβ production. More progress has been made in developing γ-secretase inhibitors, because high-throughput screens carried out in the pharmaceutical industry have identified numerous γ-secretase inhibitors. Multiple classes of potent γ-secretase inhibitors have now been described, and several of these have been shown to target both PS1 and PS2 (refs. 35–38; reviewed in ref. 7). At least one γ-secretase inhibitor is in clinical trials. Moreover, treatment of mice with a γ-secretase inhibitor reduces Aβ levels in the brain and attenuates Aβ deposition (39). However, despite these advances, numerous concerns over the use of γ-secretase inhibitors as AD therapeutics remain. These concerns center on target-mediated toxicity caused by interference with γ-secretase–mediated Notch signaling (40, 41); inhibition of signaling mediated by newly recognized γ-secretase substrates (such as the epidermal growth factor receptor ErbB4) or unrecognized substrates (42); or accumulation of potentially neurotoxic APP CTFβ, which invariably occurs when γ-secretase is inhibited (43, 44).

Although the development of β-secretase inhibitors has lagged behind the development of γ-secretase inhibitors, many believe that β-secretase is likely to be a better therapeutic target. β-Secretase (BACE1, for β-site APP-cleaving enzyme) knockout mice produce no Aβ, yet they have no obvious pathological phenotype (45, 46). Significantly, the crystal structure of BACE1 has been solved (47, 48). Such structural information will surely speed the drug discovery efforts, currently underway, to develop potent nonpeptidic BACE1 inhibitors. Although the knockout studies partially allayed fears that BACE1 inhibition might be problematic due to inhibition of cleavage of non-APP substrates, concerns remain regarding target-mediated toxicity. Moreover, the crystal structure of BACE1 reveals a wide, active-site gorge that may be difficult to target with small-molecule inhibitors (47, 48).

Very recently, several Food and Drug Administration–approved (FDA-approved) NSAIDs, including ibuprofen, sulindac, and indomethacin, have been shown to be selective Aβ42-lowering agents (49). Moreover, long-term treatment of APP transgenic mice with ibuprofen attenuates Aβ deposition (50). Although the mechanisms by which these NSAIDs lower Aβ42 have not been established, the effect is independent of cyclooxygenase inhibition, which is the primary anti-inflammatory target of these compounds (49). These substances do not change the total level of Aβ produced but, rather, shift cleavage from Aβ42 to a shorter 38–amino acid Aβ peptide (Aβ38). This finding suggests that they are interacting with γ-secretase. Although the contribution of Aβ38 to AD pathology is not known, it is generally accepted that Aβ42 is the pathogenic Aβ species (9). Therefore, investigators believe that lowering Aβ42 levels is a therapeutic strategy worthy of further investigation. The implications of these findings, with respect to the therapeutic potential of anti-inflammatory agents, will be discussed shortly.

Cholesterol-altering drugs. Epidemiologic data and data from model systems indicate that cholesterol-altering drugs may have an impact on the development of AD, and that this effect could be attributed to effects on Aβ accumulation. Retrospective studies on β-hydroxy-β-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitors (statins) show a large reduction in the risk for developing AD in individuals taking these drugs (51, 52), whereas individuals with elevated cholesterol are at higher risk for the development of AD (53–56). In culture and animal model systems, statins and other cholesterol-lowering agents decrease Aβ levels and Aβ deposition (57–59), whereas high-cholesterol diets in APP transgenic mice increase Aβ deposition (60). In addition, inhibitors of acetyl coenzyme A:cholesterol acyltransferase (ACAT), the enzyme that converts free cholesterol to cholesterol esters, also appear to decrease Aβ production (61).

Cholesterol’s role in Aβ metabolism appears to be quite complex and is the subject of recent reviews (62, 63). Cholesterol-modulating drugs could influence Aβ deposition by (a) directly influencing Aβ production through alterations in secretase activity, (b) directly altering Aβ deposition, or (c) indirectly influencing Aβ deposition by altering levels of factors such as apoE. Alternatively, it is possible that the beneficial effect of cholesterol-lowering drugs on AD is related not to effects on Aβ, but rather to the fact that a CNS ischemic event can convert preclinical AD to clinically diagnosable dementia (64). It is worth noting that in a prospective population-based study, high systolic blood pressure was associated with a higher relative risk for AD than elevated serum cholesterol levels were (56). Nevertheless, regardless of the mechanism, treatment with statins or other cholesterol-altering agents may have a significant clinical benefit in the prevention of AD.

The complex interaction of cholesterol with Aβ indicates that there are many potential ways to alter Aβ metabolism. Other examples of the complex effects of drugs on Aβ metabolism include the action of the PI3K inhibitor wortmannin (65). Wortmannin inhibits Aβ production, both in cells and in vivo, apparently by altering APP trafficking. Although such drugs do not selectively target Aβ, if these compounds are relatively nontoxic (which is not the case for wortmannin), they are reasonable candidates for anti-Aβ therapy.

Therapies targeting Aβ aggregation or removal. Because Aβ aggregation appears essential for the initiation of the AD pathogenic cascade, it may also be possible to prevent AD by altering Aβ aggregation or removing aggregates that are already formed. A number of research groups are currently exploring the development of direct Aβ aggregation inhibitors (66). While some encouraging results have been reported in animal models (67, 68), these compounds are peptide-like and unlikely to make good drugs. An alternative strategy for altering Aβ aggregation was reported recently. In APP transgenic mice treated with clioquinol (an antibiotic and bioavailable Cu/Zn chelator), marked reduction in Aβ deposition occurred after several months of treatment (69). Zinc and other divalent cations appear necessary for Aβ aggregation (70). Thus, metal chelation may have some therapeutic benefit in the treatment of AD, either by preventing Aβ aggregation or by disrupting preformed aggregates. Clioquinol is a reasonably well tolerated drug in humans and is currently in a phase II clinical trial for AD.

One of the most surprising developments in anti-Aβ therapy is Aβ immunization. Direct immunization with aggregated Aβ42 was originally shown to attenuate Aβ deposition significantly in APP transgenic mice (71). Aβ immunization now appears to be effective in reducing amyloid deposition in multiple mouse models when mice are immunized, either actively with Aβ, or passively with intact anti-Aβ antibodies (72–76). However, it appears that there are some limits to the ability of immunization to clear existing plaques. Immunization of mice with large initial amyloid loads does not have a significant impact on amyloid deposition (76). Whether this lack of clearance can be attributed to an inherent limitation of the immunization approach or to the lack of production of sufficient amounts of anti-Aβ to clear large amounts of Aβ is unknown. In the latter case, one would postulate that simply increasing the amount of anti-Aβ would cause more Aβ to be cleared. Significantly, several groups have shown that, even in the apparent absence of any effect on Aβ load in the brain, Aβ immunization can ameliorate a cognitive deficit in reference memory and working spatial memory in APP transgenic mice (74, 77). This suggests that, even in the absence of Aβ reduction, immunization may have some therapeutic effect. However, given that the relationship between memory deficits observed in these mice and those in humans with AD is unknown, the significance of this behavioral correction in mice is unclear.

Of interest are experiments showing that the local application of anti-Aβ to the brain can result in rapid clearance and resolution of the plaques, along with a robust microglial infiltration and activation (78). Based on these observations, it may be possible to rapidly clear existing Aβ deposits, at least in mouse models, given sufficient local concentrations of anti-Aβ in the brain. While questions regarding mechanisms abound, it is thought that antibodies to Aβ do one or more of the following: (a) enhance clearance of Aβ, (b) disrupt Aβ fibrils, (c) prevent Aβ fibril formation, and/or (d) block the toxic effects of Aβ aggregates.

Although the initial phase I trial of Aβ42 immunization in humans was well tolerated, the discontinuation of the phase II trial due to meningio-encephalitic presentation in about 5% of the study group represents a severe setback for direct immunization strategies. Unfortunately, due to the paucity of information on the nature of the side effects, all hypotheses regarding the nature of the postvaccination syndrome remain highly speculative (79). Very recent data now show that one patient with the postvaccination syndrome did have modest anti-Aβ titers and high levels of anti-Aβ antibodies in the cerebrospinal fluid (CSF) (80, 81). This patient did respond to steroid-induced immunosuppression, but the anti-Aβ titers remained unchanged after recovery. Although it has been suggested that the high CSF anti-Aβ titers caused disease in this individual, it is equally likely that the high CSF titers were the result of the meningio-encephalitic presentation, perhaps due to a T cell response against Aβ or APP.

Studies of Aβ metabolism reveal a number of potential therapeutic strategies that may alter Aβ accumulation in the AD brain. Agents targeting Aβ-induced cascades are also being evaluated; however, it is much more difficult to determine the potential efficacy of these, since the APP mouse models do not demonstrate all of the pathological features apparent in the AD brain. Moreover, because of the lack of clarity regarding how Aβ leads to neuronal dysfunction and death, most therapeutic modalities targeting downstream effects of Aβ are not necessarily specific to AD. Thus, agents such as antioxidants, neurotrophic factors, apoptosis inhibitors, and other neuroprotective agents may all be of benefit in the treatment of AD. They are also likely to be of general utility in other neurodegenerative conditions.

One intriguing modality currently being considered for AD treatment is the use of NSAIDs. Multiple epidemiologic studies support a role for the use of NSAIDs in preventing the development of AD. Based on the known mechanism of action of NSAIDs and the evidence for an Aβ-induced inflammatory cascade in the AD brain, it is proposed that the anti-inflammatory property of these drugs is responsible for their apparent benefit to patients with AD. The recent data demonstrating that some NSAIDs can selectively lower Aβ42 raise the possibility that this mechanism, rather than the anti-inflammatory property of these compounds, confers protection (49). Alternatively, it may be the anti-inflammatory property, or a combination of the anti-inflammatory and Aβ42-lowering properties, that confers protection. Significantly, several NSAIDs are currently being tested for efficacy in either treating or preventing AD. However, only one of the current trials is using an NSAID, ibuprofen, that potentially lowers Aβ42.