One of the surprises at the AD/PD 2005 conference held last week in Sorrento, Italy, lay in just how many groups presented new experimental therapy approaches in various stages of preclinical development. This welcome news accompanies the arrival of a constant stream of new mouse strains that model aspects of Alzheimer disease. Even with their limitations, these models will enable investigators to test their favorite fledgling therapy in a battery of systems that together will be more physiologically meaningful than the cell-free or cell-based assay that generated the original idea or compound. Many of those treatment approaches have in common that they try to reduce production of the Aβ peptide, considered a culprit in AD pathogenesis, by choking the γ-secretase complex, and some news on that effort is summarized here. For presentations focused on β- and α-secretase, see companion Sorrento story.
γ-Secretase Modulators: The New Wave?
Steve Wagner at the Californian biotechnology company TorreyPines Therapeutics introduced a series of new compounds. In the past few years, it has become clear that certain non-steroidal anti-inflammatory drugs (NSAIDs) lower levels of Aβ40/42 in a way that has little do to with their ostensible targets, COX1/2 (see ARF Live Discussion). These NSAIDs subtly alter the γ-secretase complex that produces Aβ in such a way that the enzyme spits out less of the pathogenic peptide while continuing apace with its necessary job of cutting a growing list of other, physiological substrate proteins (see ARF related news story). Once this idea of so-called allosteric modulation of the γ-secretase enzyme complex was in the air, the hunt was on for compounds that pull off this trick more potently than do NSAIDs. That is because most NSAIDs are quite toxic at the doses required for meaningful Aβ42 lowering. (In recent months, troubles about cardiovascular side effects have led to market withdrawal, more restrictive labeling, and lower dosing recommendations for some NSAIDS for arthritis treatment. This raises doubts about whether they'll ever be approved for AD prevention, as prevention requires a greater safety margin than does the treatment of chronic pain, see ARF related news story, see ARF related news story).
Wagner's team appears to have made a catch with a series of small molecules called NGX 83232 and another compound called NGX96256. In Sorrento, Wagner explained his team's efforts to screen for chemicals that lower Aβ42 production while weeding out those that also lower production of NICD. This is the intracellular tail of the γ-secretase's best studied other substrate, Notch, and it is also the protein that caused the downfall of first-generation γ-secretase inhibitors. These mechanism-based inhibitors went into clinical trials, including a phase 2 trial by Bristol Myers Squibb, where they proved too toxic (see more in May section below.) From Wagner's screen, however, one compound emerged that looked promising, and medicinal chemists fashioned 650 derivatives of it until they found some that did the job in the single-digit or sub-nanomolar concentration ranges drug developers like to use.
In addition to sparing Notch, the NGX 83232 compounds also leave unchanged levels of AICD, the cytoplasmic tail of APP. This peptide is widely thought to enter the nucleus for signaling, and recently turned out to regulate the production of none other than its parent APP in a positive, auto-regulatory feedback loop (von Rotz et al., 2004, and confirmed independently in Sorrento with a poster presentation by Sébastien Hébert in Bart de Strooper's laboratory.)
The NGX compounds did not change the total amount of Aβ produced, but they did shift the relative amounts of its various forms. There was less of the pathogenic Aβ 42 and 40 and more of its shorter forms having 37 and 38 amino acids. Little is known about these smaller, less fibrillogenic versions of Aβ, but scientists presume them to be relatively harmless.
Wagner reported that the compounds showed this effect in human neuroblastoma cells overexpressing APP and in mixed brain cultured from Tg2576 mice. Largely the same result appeared in plasma and brain extracts from Tg257 mice fed the compounds for three days. Wagner noted that the compounds did not affect the normal cleavage of ε-cadherin and ErbB4, which are two of the other physiological substrates of γ-secretase. They also did not disturb ε cleavage of APP, which is thought to precede the γ-cleavage that then releases Aβ. In response to a question, Wagner noted that his group tried NSAIDs but was unable to find a clear effect without toxicity.
(Editor's note added on March 21, 2005: This week's biotech newsletter BioCentury reported that the German/British biotechnology company Cellzome has signed an agreement with the pharmaceutical company Johnson and Johnson to develop its own small-molecule γ-secretase modulator.)
γ-Secretase Inhibitors: Down But Not Out
γ-secretase has been a favorite drug target throughout the 1990s, even before researchers identified presenilin as its catalytic component and later fingered three other members of the complex. Yet when the list of the enzyme's other substrates grew and notch-related side effects appeared, its luster dimmed and it became de rigeur among academic scientists to pronounce it dead as a drug target. In Sorrento, Patrick May of Eli Lilly and Company in Indianapolis, a veteran in the search of γ-secretase inhibitors, contended otherwise, saying newer compounds are still worth pursuing.
May recounted the history of the long and arduous search for such inhibitors, saying the work started with what by today's standards are unsophisticated tools and little supporting knowledge about just how complex this enzyme and the flux of Aα through the body is. "We were working in the dark in those days," he told the audience. Early successes with an inhibitor called DAPT made the goal look attainable, especially since the compounds appeared to have delayed pharmacodynamic effects on Aβ levels and amyloid deposition after the drug had left the mice's body (Dovey et al., 2001).
Next, Lilly scientists made the LY series of inhibitors, which are largely unpublished except for LY411575. Merck and Pfizer have studied such compounds, as well (see Best et al., 2005, Lanz, 2004), and some now act at picomolar doses without apparent cytotoxicity, May said. Yet it is also known that at least LY411575 at higher doses ravages the immune system and intestine of mice (Wong et al, 2004) The side effects have raised the question of whether one can achieve a clinically significant reduction in Aβ while otherwise preserving essential function of regulated intramembrane cleavage. This issue of the therapeutic window can only be answered in the clinic, May said.
Toward that end, May talked about the compound LY 450139, which inhibits γ-secretase less potently but also has a milder effect on Notch cleavage. "We want to gingerly go into the clinic with this second generation," May said. Long-term administration to beagle dogs went well, and the compound reduced CSF Aβ levels. LY450139 is now in preparation for use in clinical biomarker studies, May said. A randomized phase 1 trial of 70 AD patients taking the drug for six weeks has helped the company approximate an apparently safe dose, but otherwise was inconclusive. That is largely because lumbar CSF levels of Aβ did not drop as markedly as was hoped, and an observed Aβ drop in plasma levels is difficult to interpret at this point. A large efficacy trial, therefore, is not imminent, May concluded, as the company may have to try out a slightly higher dose first. Other companies developing g-secretase inhibitors include Bristol-Myers Squibb, Novartis, Amgen, Sanofi-Aventis, but they have not made data public.—Gabrielle Strobel.