4 April 2005. In terms of a treatment for Alzheimer disease, Aβ seems to be where it’s at these days. Some of the most promising strategies focus on preventing its production and oligomerization, or trying to enhance its clearance from the brain. At the Molecular Mechanisms of Neurodegeneration meeting in Dublin, all three possibilities were explored.
Grant Krafft, chairman and CSO of Acumen Pharmaceuticals, described his company’s strategy for tackling AD by targeting ADDLs, or Aβ-derived diffusible ligands. ADDLs—some would call them Aβ oligomers—are the real bad guys in AD, suggested Krafft, not misfolded proteins, fibrils, plaques, or even cell death. Misfolding is a misnomer, he contends, because at subnanomolar concentrations, thermodynamics favor the formation of Aβ assemblies. Though fibrils and plaques may not be good, they are not the root cause of AD either. And, while in late stages of the disease there is considerable loss of neurons, the real damage is done before cell death, and is caused by a breakdown in signal transduction, Krafft proposed. “How else could you have neurons packed with hyperphosphorylated tau?” he asked.
For all these reasons, Acumen, a relatively new kid on the biotech block, has focused on dealing with ADDL toxicity. Central to their strategy is data showing that ADDLs bind to synapses, particularly the synaptic cell surface receptor postsynaptic density 95 (PSD-95). This binding results in upregulation of intracellular signaling molecules including Arc and Rac (see Lacor et al., 2004): Arc is an immediate-early gene which has been linked to dysfunctional learning, while Rac activation has been linked to compromised long-term potentiation (LTP). Krafft reported that Acumen is using synaptosomes in high-throughput screens to find receptor antagonists that might prevent the toxic effect of these ADDLs.
Acumen is also developing molecules (in addition to antibodies) that interfere with ADDL assembly. Though interrupting protein-protein interactions presents quite a challenge to chemists (see ARF related Dublin story), in the case of Aβ42 it may be a little easier, suggested Krafft. The theory is that the two C-terminal amino acids, which are not present in the less “sticky” Aβ40, allow the formation of a β hairpin that then serves as the core of the ADDL structure. By targeting that hairpin it should be possible to develop drugs that prevent oligomerization of Aβ42. Acumen has obtained some lead compounds from high-throughput screens.
Krafft also described his collaboration with Bill Klein and chemist Chad Mirkin, both at Northwestern University, to develop antibodies that can be used to detect ADDLs in the CSF (see ARF related news story), and his ongoing cooperation with Merck & Co. to develop humanized antibodies that will selectively bind to ADDLs rather than Aβ monomers or fibrils.
Of course, given that Dublin serves as headquarters of Elan, the first company to bring an Alzheimer disease vaccine to clinical trial, it is probably fitting that at least some presentations would address the subject of AD and immunization. Elan’s vaccine program suffered a major setback when patients in a phase II clinical trial developed encephalitis. That disappointment has since been tempered by some encouraging results from those trials (see ARF immunotherapy update from the Sorrento AD/PD meeting). For example, postmortem examination of three patients who were injected with Elan’s AN-1792 has shown that the vaccine seems to dramatically reduce plaque burden. In Dublin, James Nicoll, a diagnostic pathologist at the University of Southampton General Hospital, England, showed that in a fourth patient there is evidence that plaques, though still present at death, were undergoing active removal (see Nicoll’s Sorrento presentation).
These postmortem results, plus tantalizing hints that the immunized patients do better cognitively and functionally (see ARF related news story), have spurred companies to develop second-generation passive and active immunotherapies that can circumvent the inflammatory response. Menelas Pangalos of Wyeth Pharmaceuticals, a partner in Elan’s vaccine program, reviewed some of the data coming out of Elan’s phase II trial (for a summary see coverage of Dale Schenk’s Sorrento presentation), but also hinted at ways to determine who can benefit from such therapy. Using transcriptional profiling, Pangalos and colleagues have been able to retrospectively “predict” which patients in the AN-1792 trial best respond to the vaccine. The profiling approach may also be useful for identifying unwanted side effects. Pangalos and colleagues have found that the transcriptional relationship between two specific genes is a good indicator of which patients in the Elan trial developed encephalitis.
Other therapeutic hopes are pinned on inhibiting β- and γ-secretases, the two proteases that sequentially cleave AβPP (see also related Sorrento news on γ-secretase and β-secretase). These strategies are not without controversy, however. One of the concerns about targeting γ-secretase is that not only does it cleave AβPP, but it also cleaves other transmembrane proteins including Notch, a major signaling molecule and a potential player in learning and memory (see ARF related news story and ARF news story). Some benzodiazepine derivatives, for example, inhibit γ-secretase but also cause goblet cell hyperplasia in the gut lining of the rat. This is thought to be caused by alteration of stem cell fate. The question on everyone’s mind, therefore, is whether a γ-secretase inhibitor can be developed that can prevent cleavage of AβPP but allow Notch cleavage to proceed as normal.
Mark Shearman reported that Merck Sharp & Dohme now have patents on three generic classes of γ-secretase inhibitor. These include compounds with IC50s in the low nanomolar range and which can reduce soluble Aβ in the brain of transgenic mice by 90 percent after a single dose. Though these inhibitors affect Notch and AβPP processing equally, he suggested that the “therapeutic window” between desirable and undesirable effects may be wide enough so that γ-secretase inhibitors can be used safely.
Shearman and colleagues have tested this in several ways, including molecular profiling. He reported how they used chip analysis of gene expression in the rat ileum to associate a specific gene expression profile to goblet cell hyperplasia. Then, when they tested benzodiazepine derivatives and their γ-secretase inhibitors on rat ileum, they found that the benzodiazepines give a huge hyperplasia profile, while one of their drug candidates, called compound F, showed very little signal.
In vivo data back up the profiling experiment. Shearman and colleagues have given the compound to transgenic mice daily for three months and seen no toxicity or Notch phenotype, even though the plaque load and area occupied by plaques were reduced by 60 and 40 percent, respectively. “So it is possible, with γ-secretase inhibitors, to obtain the efficacy that you want without the side effects based on Notch signaling,” said Shearman. He did caution, however, that rodents appear among the animals least sensitive to Notch-related effects.
Pangalos also reviewed the Wyeth γ-secretase program. The company has developed compounds that can inhibit the protease at IC50 values around 15 nM. In cell-based assays, these inhibitors prevent the release of Aβ42 and lead to an increase in levels of βCTFs produced by β-secretase.
In animals, a single dose (30mg/Kg) of one of the inhibitors, called GSI-1, causes a time-dependent inhibition of Aβ production, reducing synthesis by about 70 percent, reported Pangalos. Daily dosing at as little as 2.5mg/Kg can have a similar effect. As for side effects, even at the high doses there appeared to be little impact on thymocytes or the GI tract, he reported. As for behavior, when Pangalos and colleagues tested the compounds in transgenic mice (Tg2576), they found that the compounds can almost completely reverse defective fear responses.
Because γ-secretase shows a degree of promiscuity, β-secretases, or BACEs, have been considered by some to be more wholesome targets. There are two BACEs in the mammalian genome and we already know that one of them, BACE1, seems dispensable in mice (see ARF related news story). So what about BACE2?
Those bent on developing BACE inhibitors will be delighted to hear that BACE2 knockout mice are also viable, fertile, and seem no different from wild-type. In Dublin, Martin Citron, whose lab at Amgen Inc. discovered BACE1 in 1999 (see ARF related news story and Alzforum interview), reported his results with BACE2 knockouts and reported that BACE1/2 double knockouts are indistinguishable from wild-type animals, too.
The news may not come as a big surprise to many because BACE2 was always considered the minor β-secretase—its expression level is very low in neurons and overexpression of the protein actually leads to more α-secretase activity. The problem now will be to find a BACE inhibitor that can be turned into a useful drug. Citron reviewed the crystallographic data that shows the BACE1 active site as a very large binding pocket with as many as eight subsites. Some of the first-generation BACE inhibitors were bulky hexapeptides that could contact most of these subsites, but unfortunately these were not cell permeable, he said. Now, there are third-generation compounds available that are smaller, penetrate the cell, and can inhibit the protease while only binding to four of the subsites. “But as we still don’t know all the substrates for BACE, the question of toxicity is still one that needs careful attention,” he suggested.—Tom Fagan.