12 September 2010. Despite recent setbacks on the clinical front (see ARF related news story), the hunt for small molecules that can cleanly tweak γ-secretase to slow Alzheimer disease seems to be alive and well. In the September 9 Neuron, researchers led by Steven Wagner, University of California, San Diego, report on a new class of γ-secretase modulators that reduces β amyloid deposition in an AD mouse model without affecting Notch or other γ-secretase substrates that mediate critical functions. Originally identified at Neurogenetics, Inc., a San Diego-based firm that became TorreyPines Therapeutics, Inc., which has since folded, the compound penetrates the brain, reduces levels of toxic Aβ more efficiently than do other γ-secretase modulators, and was safe over a six-month treatment in mice. “Overall, I thought this was a very nice drug discovery paper,” said Michael Wolfe of Brigham and Women’s Hospital in Boston. “This is an important new class of molecules that is definitely worth following up with more preclinical studies.”
Early efforts to target γ-secretase for AD focused on inhibiting the enzyme, which cuts a C-terminal piece of amyloid precursor protein (APP) to release the Aβ peptides that can glom together to form the hallmark AD plaques. However, because γ-secretase has numerous substrates besides APP, most notably Notch, the wholesale blockade approach carries with it a slew of potentially harmful side effects, which some say might have felled Phase 3 trials of Eli Lilly’s γ-secretase inhibitor compound last month (see ARF related news story). To avoid such problems, Wagner and coauthor Rudolph Tanzi, an AD geneticist at Massachusetts General Hospital, Boston, tossed around ways to find molecules that could temper γ-secretase more precisely. The compounds in the current report came out of a high-throughput screen first conceived in 1999 by the two researchers, who, together with coauthor Bill Comer, founded Neurogenetics, Inc., on a strategy rooted in early-onset AD genetics. “We felt that one of the obvious approaches besides inhibiting γ-secretase was to simply try to medicinally reverse the biochemical phenotype of mice with familial AD gene mutations, which was an increased Aβ42/Aβ40 ratio,” Wagner told ARF. He presented some of this work at the 2005 AD/PD Conference in Sorrento, Italy (see ARF related conference story).
Screening 80,000 compounds in a Chinese hamster ovary (CHO) cell line overexpressing mutant APP, first author Maria Kounnas and colleagues looked for molecules that could lower the Aβ42/Aβ40 ratio by curbing Aβ42 production. Candidates were excluded if they also affected generation of Notch intracellular domain (NICD), a cytoplasmic peptide with key roles in cell development and differentiation. One compound met both criteria. Fine-tuning its structure, the researchers designed and synthesized some 1,200 molecules, and eventually came up with several compounds with potency and pharmacokinetics worthy of in vivo testing in AD mice.
Tested in mutant APP-overexpressing SH-SY5Y human neuroblastoma cells, the “winner” (a diarylaminothiazole called “Compound 4”) had a five nanomolar IC50 (i.e., concentration of compound needed to slash Aβ42 production twofold), which falls squarely in the low nanomolar range that piques the interest of drug developers. This is comparable to some of the most potent γ-secretase inhibitors (e.g., the sulfonamide BMS-299897) and 1,000 to 10,000 times more potent than tarenflurbil and other non-steroidal anti-inflammatory (NSAID)-like γ-secretase modulators measured in comparable cell-based assays.
The Aβ effects of Compound 4 were different, too. Unlike the sulfonamide, which lowers total Aβ (Anderson et al., 2005), and NSAIDs, which lower Aβ42 and increase Aβ38, the diarylaminothiazole reduces both Aβ42 and 40 while raising Aβ37 and 38. The molecule does not affect total Aβ peptide levels, as confirmed using mass spectrometry of anti-Aβ immunoprecipitates of mutant APP-expressing CHO cells.
Importantly, when tested in human embryonic kidney HEK293 cells expressing APP and Notch, the new γ-secretase modulator did not affect NICD formation, even at concentrations more than 1,000 times higher than the IC50 for Aβ42 inhibition.
To define molecular targets, the researchers ran detergent-solubilized APP-CHO cell lysates over an agarose matrix coated with the compound. This affinity strategy captured Pen-2 (a protein in the γ-secretase complex) and, to a lesser extent, N-terminal and C-terminal fragments of presenilin-1. Another component of the γ-secretase complex, nicastrin, did not bind. Nor did APP. This shows the compound has “a measure of specificity,” Wolfe said, noting the data would be strengthened by several additional experiments. “What you’d like to see is competition with a free inhibitor that prevents [the Pen-2 binding]. And you’d like to see that an inactive, structurally related compound doesn’t do this,” he said. Furthermore, to show the binding site is distinct from those of other γ-secretase modulators, “you’d want to show that NSAIDs don’t compete for it,” Wolfe said. The authors noted that competition experiments “were not possible due to the core aqueous solubility” of Compound 4 and others in the diarylaminothiazole series.
In spite of these biochemical caveats, a key strength of the paper is its demonstration of the molecule’s in vivo relevance, Wolfe said. The researchers fed the diarylaminothiazole to Tg2576 mice with their standard chow, starting at an age of eight months, when Aβ plaques are just starting to accumulate in this AD transgenic strain. “This was along the lines of a preventive study,” Wagner said. The mice fed on the supplemented food for eight months before being sacrificed for analysis at 15 months of age. At this point, the brains of control animals had oodles of neuritic plaques, whereas amyloid load was down two- to threefold in the treated group, as judged by silver-stained cortex and hippocampus sections. Furthermore, treated mice were free of gastrointestinal lesions that result from extended exposure to γ-secretase inhibitors. The authors did not formally test cognition or behavior. Wagner has moved to UCSD, where he and colleagues are working on a related series of γ-secretase modulators that appear more water soluble than the compounds in the Neuron paper. Further characterization of the related GSMs is ongoing at UCSD and at Tanzi’s lab at MGH.
The γ-secretase project got off the ground when Neurogenetics, Inc., was founded in 2000. Over the years, the company received some $15 million from Eisai Co. Ltd., the Tokyo-based pharmaceutical company that discovered and markets Aricept (donepezil). Meanwhile, Neurogenetics, Inc., went public through a reverse merger in 2005, at which point it was rechristened TorreyPines Therapeutics, Inc. The company licensed-in other projects, including an M1 agonist for AD (see ARF related news story) and a migraine drug.
After TorreyPines shut down its research operations in 2008, Wagner, Tanzi, and Comer co-founded Neurogenetic Pharmaceuticals the following year to continue working on the GSMs. Comer provided funds to license the γ-secretase technology from TorreyPines, and presently heads preclinical development of Compound 4. The private firm has scaled up production and launched additional safety studies of this molecule, in hopes of filing an Investigational New Drug application to begin clinical testing next year, Comer said.
While scientists await such trials with a mixture of anticipation and skepticism, some say the field will not advance without facing up to the bigger problem of translating findings from lab to clinic. “Ninety-nine percent of what we do in academia, or even pharma, is prevention studies in APP mouse models to support therapeutic studies in humans, but they’re completely different beasts,” said Todd Golde, University of Florida, Gainesville, noting that many promising candidates from prevention studies in animals go on to be tested in mild to moderate AD patients who already have extensive pathology. Wolfe agreed, and stressed that “amyloid-lowering compounds have to be tested earlier on. There’s good reason to believe there’s too much damage in patients who have clinically diagnosed AD. You really can’t expect Aβ-lowering agents to be very effective in these people.”
Moreover, recent work “suggests that the regulation of γ-secretase cleavage in terms of which substrate, and where, is quite complex and dependent on a whole host of factors,” Golde said. The September 2 Nature reported on a γ-secretase activating protein that specifically enhances binding of γ-secretase to APP, but not to Notch (He et al., 2010 and ARF related news story). Several years ago, Golde and colleagues reported an interesting twist on NSAID-like γ-secretase modulators, showing that they work by targeting APP (ARF related news story on Kukar et al., 2008). A study published last month in PNAS bolsters this substrate theory (ARF related news story on Richter et al., 2010).
These complexities make it all the more imperative to test γ-secretase modulators in people at milder disease stages, Golde urged. In this regard, there are hints that the field is moving in the right direction. Last year, Bristol-Myers Squibb began recruiting for a Phase 2 trial of its γ-secretase inhibitor BMS-708163 in prodromal AD.—Esther Landhuis.
Kounnas MZ, Danks AM, Cheng S, Tyree C, Ackerman E, Zhang X, Ahn K, Nguyen P, Comer D, Mao L, Yu C, Pleynet D, Digregorio PJ, Velicelebi G, Stauderman KA, Comer WT, Mobley WC, Li YM, Sisodia SS, Tanzi RE, Wagner SL. Modulation of Gamma-Secretase Reduces Beta-Amyloid Deposition in a Transgenic Mouse Model of Alzheimer’s Disease. Neuron. 2010 Sep 9;67:769-780. Abstract