5 November 2002. In a sparsely attended slide presentation yesterday at the Neuroscience meeting, Inez Vincent of the University of Washington, Seattle, presented a feat no one seems to have pulled off before: With an experimental small-molecule compound that inhibits the kinase cdk5, she almost completely reversed neurodegeneration in a mouse model that may be obscure, but whose neurofibrillary pathology is very similar to that seen in human AD.
Vincent studied a natural mouse model of Niemann-Pick Type C (NPC). This human disease was so named because of its lysosomal storage defect, which causes ballooning neurons filled with lipid, spheroid structures in axons, and neuronal loss. Trying to understand how this pathology comes about, Vincent studied phosphorylation of cytoskeletal proteins from human disease and the mouse model. She found that tau was hyperphosphorylated in both human and mouse, and correlated with increased cdk5 activity and conversion of p35 to p25. This started at one month of age in mice. P25 and cdk5 accumulated together with hyperphosphorylated cytoskeletal proteins in axon spheroids.
The kicker, however, lies in the treatment experiment. Vincent treated four- to six-week-old mice with the cdk5/p25 inhibitor roscovitine, originally developed by a French group. The mice received intraventricular infusion driven by an osmotic pump for two weeks. Not only did the treated mice lose less weight and improve their locomotor scores compared to controls, but their protein phosphorylation decreased by about 90 percent in immunoblot, ELISA, and immunohistochemistry assays. The axonal spheroids partly resolved and neuronal loss decreased markedly. The authors conclude that cdk5/p25 mediates Niemann-Pick C neuropathology, and that inhibitors of this enzyme may treat this rare disease. The obvious implication is whether this cdk5/p25 mechanism is at work in AD, which also features tau hyperphosphorylation and neuronal degeneration. This apparently has not been tested. The drug could be tried in tau transgenic models, as well as in combined tau-APP transgenic models, which feature a more severe, combined phenotype.
Proteins Behaving Badly: P25/CDK5 Consorts with Pathologic Substrates.
Today, J.C. Cruz in Li-Huei Tsai’s lab presented an extension of Tsai’s presentation on Sunday (see related news story), in which she had suggested that hyperphosphorylation of APP’s cytoplasmic domain by p25/cdk5 influences the sorting of APP away from α-cleavage in the cell membrane and instead toward BACE cleavage in endosomes (see related news story). To examine this question in vivo, Crux et al. generated inducible transgenic mice that express p25-GFP under the control of the CamKinase2 promoter, so that the transgene is expressed only in the forebrain. With a panel of different phospho-specific antibodies, Crux found that with the induced p25, cdk5 no longer phosphorylated its normal targets, for example, the postsynaptic protein PSD-95. Instead, cdk5 now hyperphosphorylated APP, tau, and neurofilament. The authors suggest that the two regulatory proteins-p35 and p25-direct cdk5 to different sets of targets, one normal, one pathological.
The idea connecting all these dots, speculates Tsai, would be that p25/cdk5 phosphorylates tau, leading to axonal transport blockages and later to tangle deposition, and also phosphorylates APP, leading to Aβ generation in endosomes. How is that evil p25 made? The enzyme calpain, which has been implicated in neuronal death, cleaves p35 to generate p25 (see related news story).
So what activates calpain? This is where the argument becomes circular, as Aβ 42 has been shown to activate. However, other toxic conditions do, as well, for example, oxidative stress and increased intracellular calcium. This ion has long been a suspect in AD, and recent studies indicate that presenilin-1 mutations, besides revving up Aβ generation, also disrupt the restorative flows of calcium between the inside and the outside of the cell in such a way that intracellular calcium rises to dangerous levels. Perhaps one way in which PS1 mutations might induce tau pathology is via this connection?
Does this simplified speculation hold water in humans? The p25-transgenic mice do exhibit neurodegeneration, and yet, does the overexpression of p25 make this an artificial, irrelevant finding? Tsai says that a mouse overexpressing an endogenous mouse protein may not be more artificial than a mouse overexpressing a human protein. Her original finding that p25 protein levels were elevated in the brains of AD patients (see news) did not address whether that varied between regions. Today, B.A. Samuels, and others in Tsai’s group report a follow-up study on another series of human autopsy samples from multiple brain regions. Working with Yong Shen of the Sun Health Research Institute in Sun City, Arizona, Samuels found that frontal cortex shows the strongest overproduction of p25 relative to p35.-Gabrielle Strobel.