12 Jun 2009

To Douglas Adams fans, it’s the answer to life, the universe, and everything. To Alzheimer disease researchers, the numeral 42 holds similarly high esteem, denoting the amino-acid length of one of the field’s top biomarkers. But absent Deep Thought’s superior computing powers, calculating the risk of impending AD based on levels of cerebrospinal fluid (CSF) Aβ42 has been tricky because the peptide forms insoluble aggregates that muddle estimates of its production rate in the central nervous system (CNS)—especially in the pivotal years preceding the onset of dementia. In a report published online yesterday in EMBO Molecular Medicine, a team of Japanese and German researchers introduces APL1β28 as a potential surrogate marker for CNS Aβ42 production. This Aβ-like peptide gets cleaved by the same enzymes that unleash Aβ42, but it does not form plaques and appears in the CSF at elevated levels in sporadic and familial AD patients.

Brain accumulation of pathological Aβ forms, especially Aβ42, is a key feature of AD. As such, unusually high ratios of Aβ42, relative to total Aβ, should serve as a red flag for future development of dementia. However, during the years leading up to detectable memory loss, this prediction hits a snag. Within that critical window, which many see as the best time for initiating potential treatments, CSF Aβ42 levels do not rise, as would be predicted if one simply assumes that increased production is a harbinger of disease. Quite the opposite; Aβ42 CSF levels drop during the preclinical phase of AD, presumably because these peptides form the insoluble deposits that clog AD brains. However, if the enzymes that unleash Aβ42 from amyloid precursor protein (APP) were also to cleave other substrates in a similar fashion, and if those latter cleavage products did not aggregate in the brain, they might offer a more straightforward way to measure Aβ42 production. This is what researchers led by Masatoshi Takeda and first author Kanta Yanagida, Osaka University Graduate School of Medicine, Japan, have shown in the new study.

The researchers discovered APL1β28 in their hunt for in vivo Aβ-like peptides. They looked for peptides derived from APLP1 (βAPP like protein1/2) by probing human CSF samples with antibodies against its juxtamembrane domain. They found that like Aβ42 and its cousins, APL1β species can be produced by sequential cutting by β- and γ-secretase enzymes. A key difference is that APL1β does not appear to be amyloidogenic. It did not aggregate when incubated in vitro under the same conditions used to drive Aβ fibril formation, and APL1β plaques were not found in immunohistochemical analyses of AD temporal lobe tissues.

When Takeda and colleagues analyzed human CSF, they detected APL1β peptides at concentrations comparable to Aβ, and levels of the APL1β28 species correlated with Aβ42 production. Furthermore, the researchers found that the ratio of APL1β28 to total APL1β was higher in familial and sporadic AD patients, relative to healthy patients, even at the mild cognitive impairment (MCI) stage preceding bona fide AD. “If further studies confirm that the ratio of APL1β28 to total APLP1β species reflects Aβ42 production relative to total Aβ species, this ratio may be an excellent endophenotype to identify new genetic risk factors for AD as has been done recently with Aβ42 and tau (Kauwe et al., 2008 and ARF related news story),” writes David Holtzman, Washington University, St. Louis, Missouri, in a commentary accompanying the new study. Holtzman and colleagues combined stable isotope labeling with CSF sampling to measure CNS production and clearance of Aβ peptides in people in real time (Bateman et al., 2006 and ARF related news story), and recently used this method to measure the effect of an experimental AD drug on Aβ production (Bateman et al., 2009 and ARF related news story).—Esther Landhuis