9 March 2009. Two things you can find on any Alzheimerologist’s wish list are better drug targets and better diagnostics. Both were on people’s minds at the Keystone Symposium, Neurodegenerative Diseases: New Molecular Mechanisms held 17-22 February at Keystone, Colorado. The drug targets might yield better medicines, and the diagnostics will help identify people in the prodromal phase of the disease, when said medicines may work best. As David Holtzman, Washington University School of Medicine, St. Louis, Missouri, said in his talk, most of the clinical trials going on for AD right now are conducted after Aβ accumulation is close to or has reached its maximum, tau accumulation is already substantial, and many neurons have been lost. “We need to be diagnosing AD much earlier,” he said.
Holtzman reviewed some of the current biomarker data and also showed how a new pulse-chase method of monitoring changes in CSF Aβ can help monitor the efficacy of γ-secretase inhibitors. In his talk, Bart De Strooper, of the University of Leuven and Flaams Instituut voor Biotechnologie in Leuven, Belgium, revealed some protein partners that might be exploited for γ-secretase inhibition, particularly of its proteolytic processing of amyloid precursor protein (APP).
Holtzman’s lab studies antecedent biomarkers, particularly cerebrospinal fluid (CSF) markers such as Aβ and tau. Though CSF concentrations of these markers overlap considerably between controls and patients, it is now well established that AD patients have reduced CSF Aβ42 and elevated levels of CSF tau and p-tau. “The idea is that Aβ42 is going down in the CSF because aggregates form [in the brain] and shift the equilibrium toward plaques,” said Holtzman. To see if this is true, his lab has correlated CSF levels with brain amyloid load, judged by binding of the PET agent Pittsburgh compound B (PIB). That work showed that nearly every research participant with a high plaque load also had low CSF Aβ42, and almost everyone with high CSF Aβ42 had no plaque load (see ARF related news story and also Li et al., 2007). But the WashU researchers also follow a third group of people with low CSF Aβ42 and no or few plaques. Holtzman suggested that these people may only have diffuse plaques, which PIB does not bind well, not fibrillary Aβ. In other words, the CSF Aβ drops before they become PIB positive.
Holtzman said that biomarker data could be used to follow formation or clearance of aggregates as part of monitoring disease progression and/or treatment. One potential way to do that is with microdialysis. Holtzman’s group has developed this technique in mice to measure Aβ changes in the brain’s interstitial fluid (ISF) in response to treatments such as a γ-secretase inhibition (see ARF related news story). More recently, the researchers have developed a microdialysis probe for screening drugs in vivo. The probe can deliver compounds as well as measure solutes in the ISF, and Holtzman said the technique lends itself to moderate throughput (about 500 compounds a year). As a control, picrotoxin, which increases synaptic activity, increases ISF Aβ with this system, in keeping with the idea that synaptic activity promotes release of the peptide (see ARF related news story).
Aβ levels can and have been measured in human CSF as well, though there is a small snag. “Aβ levels are quite variable in a normal individual’s CSF even during a single 24-hour period,” said Holtzman. That variation may be linked to synaptic activity, he suggested. One way around that variability may be to simply look at fractional Aβ synthesis and clearance, rather than total Aβ, which reflects synthesis, clearance, and stability. To do this, Randy Bateman, working originally with Holtzman, has adopted pulse-chase using C13-labelled leucine. The amino acid is injected, readily gets into the CNS, and can be identified in Aβ peptides following sampling of the CSF. The researchers have tried this technique in both animal models and in humans.
Bateman, Holtzman, and colleagues have used the pulse-chase method to measure the efficacy of a γ-secretase inhibitor in humans. While measures of CSF Aβ did show an effect of the highest dose of the inhibitor, there was lot of noise in the data, reflecting normally fluctuating Aβ levels. In contrast, the pulse-chase data showed a drop in newly synthesized Aβ at every dose of the inhibitor tested. The results are highly significant with only five patients per group, Holtzman said. The drug had no effect on clearance of Aβ, which might be expected for a γ-secretase inhibitor. Holtzman said the data are currently in press in Annals of Neurology.
γ-secretase is a major drug target for pharma and some academic labs studying AD. De Strooper offered up some novel γ-secretase-related targets for reducing Aβ load. One is the G protein-coupled receptor GPR3, which shuttles γ-secretase to the cell surface, where it can contribute to amyloidogenic processing of APP. GPR3 makes a good drug candidate because it appears to have little to do with processing of Notch, another γ-secretase substrate, De Strooper said. That work was recently published (see ARF related news story). Another γ-secretase-linked target is Aph1B, a component of the presenilin complex. Aph1 comes in three homologs—A, B, and C, and De Strooper and colleagues have shown that Aph1A mouse knockouts look very much like Notch knockouts, while Aph1B/C knockouts appear almost normal (see ARF related news story). That finding suggests that Aph1A may confer Notch specificity on γ-secretase, which raises the question of what the other homologs are doing. Could they confer APP specificity on the protease? That idea is borne out by De Strooper’s work on Aph1 B/C homologs. In Keystone, he showed that knocking out Aph1B/C partially rescues an AD phenotype. Offspring of Aph1B/C-negative mice bred with AD mice performed better in the Morris water maze than the AD controls, and the crosses were also protected against Aβ plaque deposition. The finding suggests that γ-secretase with Aph1B is a major contributor to the amyloid deposition in the brain, De Strooper said. In support of this, De Strooper showed data from fluorescent lifetime imaging (FLIM) experiments of presenilin reconstituted in γ-secretase knockout fibroblasts. FLIM reveals how closely connected the different components of the γ-secretase complex are, and showed that while presenilin was in an open conformation with Aph1A, it was in a closed conformation with Aph1B. De Strooper concluded that Aph1B contributes to Aβ processing and might be a good target for a selective γ-secretase inhibitor.—Tom Fagan.