16 January 2005. Reduced activity of p21-activated kinase (PAK) may cause the synaptic defects that leave Alzheimer patients with poor memory and cognitive skills. That is the conclusion of a paper appearing in January 14 Nature Neuroscience online. Greg Cole, from the University of California at Los Angeles, and colleagues also show that PAK activity in hippocampal neurons is attenuated by oligomers of amyloid-β (Aβ), thus providing a link between the fibrillogenic peptide and synaptic dysfunction.
There is overwhelming evidence that Alzheimer disease (AD) pathology begins with synaptic failure (for a review, see Coleman et al., 2004). In this respect, AD is similar to some developmental diseases, such as Down syndrome, and to some mental retardation syndromes. While it is unclear what molecular sequence of events leads to synaptic loss in AD, it is clear that X-linked, nonspecific mental retardation is caused by a simple missense mutation in the gene for one of the PAK isoforms, PAK3 (see ARF related news story). This prompted Cole and colleagues to examine postmortem samples of AD brain tissue for PAK defects.
Lead author Lixia Zhao and coworkers found that both major isoforms of the kinase, PAK1 and PAK3, are reduced by about 65 and 38 percent, respectively, in the hippocampus of patients who had had moderate AD. When compared to normal brain tissue, levels of PAK3 in the temporal cortex were significantly lower, and levels of active, phosphorylated PAK (all three isoforms) were lower still. These losses seem related to the formation of some kind of aggregate, because Zhao and colleagues spotted abnormally dense and focal deposits of phosphorylated PAD in AD hippocampus tissue.
The findings beg two major questions: What causes the loss of PAK activity/aggregation and what significance does this loss have? The latter may be easier to address because it is well known that PAK can regulate actin filament growth through its ability to induce phosphorylation of cofilin, a protein that otherwise binds to F-actin and promotes disassembly of actin filaments. Actin, of course, is essential for remodeling of the cytoskeleton and the growth and maintenance of dendritic spines and synapses.
In fact, the authors showed that cofilin prevents the spine protein drebrin (developmentally regulated brain protein) from binding to actin, suggesting that PAK, cofilin, and drebrin may be inextricably linked. The authors also noted other correlates among these spine proteins. In samples from AD brains, they found a reciprocal relationship between cofilin and drebrin, such that when the level of one is up, then the other is down. In cells from diseased hippocampus, increasingly punctate cofilin labeling accompanied loss of PAK. The latter is reminiscent of the so-called Hirano bodies that are found in the hippocampus and cortex of AD patients (see ARF related news story). These contain cofilin, actin rods, and other actin-binding proteins.
Many of these observations are purely correlative. The authors used an animal model of PAK deficiency to tie them all together, showing that inhibiting PAK leads to a loss of actin-bound drebrin, punctate labeling of cofilin, and even loss of memory in 11-month-old mice. It would be interesting to see whether synapses are lost, whether synaptic function is compromised in these animals, or how a cross with APP/PS-transgenics would fare. Previous studies have shown, for example, that reducing PAK activity by up to 40 percent leads to spine defects and synaptic plasticity in the cortex, but no observable synaptic defects in the hippocampus (see Hayashi et al., 2004).
What leads to loss of PAK in the first place? The authors suggest that it might be Aβ because in the brains of 22-month-old Tg2576 transgenic mice, they found PAK aggregates that resemble those seen in human tissue. (These animals, harboring human Aβ precursor protein (AβPP) with the Swedish mutations, accumulate human Aβ as they age.) Zhao and colleagues also found that these transgenic mice have reduced levels of both phosphorylated PAK and drebrin, and they appear to have Hirano body-like aggregates of cofilin. The authors also found that levels of phosphorylated PAK and drebrin were restored if the animals were treated by passive immunization with an Aβ antibody to reduce Aβ in the brain. All of these findings suggest a role for Aβ in PAK dysfunction, and this was supported by in-vitro experiments on primary hippocampal neurons. The addition by the authors of Aβ oligomers to these cultures caused about a 40 percent reduction in PAK activity within 15 minutes. During the next two days in culture, the neurons lost nearly 80 percent of their drebrin.
This isn’t the first time PAK has been implicated in AD. The kinase is phosphorylated by Cdk5 (see ARF related news story), which has also been proposed to play a role in disease pathology. Rachael Neve and colleagues reported in 2003 that interaction between the C-terminal of AβPP and PAK3 may send neurons into an apoptotic spiral (see ARF related news story). The kinase may, therefore, pack a double punch, hitting synapses hard and contributing to cell death.—Tom Fagan.
Zhao L, Ma Q-L, Calon F, Harris-White ME, Yang F, Lim GP, Morihara T, Ubeda OJ, Ambegaokar S, Hausen JE, Weisbart RH, Teter B, Frautschy SA, Cole GM. Role of p21-activated kinase pathway defects in the cognitive deficits of Alzheimer disease. Nature Neurosci. Advanced online publication. Jan. 15, 2006. Abstract