Following a recent report linking Fragile X mental retardation protein (FMRP) with translation of amyloid-β precursor protein (see ARF related live discussion), new evidence suggests an additional connection between Alzheimer disease and Fragile X syndrome, the most commonly inherited form of mental retardation. Lowering the activity of p21-activated kinase (PAK), which is decreased in the brains of Alzheimer patients (see ARF related news story), reverses the synaptic and behavioral deficits imposed by the Fragile X gene mutation in mice, according to a paper in the June 25 PNAS online. The finding suggests that inhibiting PAK might be a valuable therapeutic strategy for treating this neurodevelopmental disease, which afflicts—to varying degrees of severity—1 in 4,000 males and 1 in 6,000 females, according to the National Fragile X Foundation.

That cross-breeding the mouse model of Fragile X syndrome (FXS) with transgenic mice with low PAK function would create a seemingly normal mouse may seem too simple. So thought senior author Susumu Tonegawa, at the Massachusetts Institute of Technology, when Mansuo Hayashi, the study’s first author, suggested the idea after she noticed remarkable dendritic spine characteristics in PAK mutants.

“Mansuo looked at the spines of the PAK mutants and found abnormalities that were opposite of what was found in the Fragile X knockouts,” said Tonegawa. In contrast to the spindly, immature spines observed in the brains of the FXS mouse model and in patients with the syndrome, spines in the PAK mutants were stocky. “Mansuo thought it was interesting if you could cross the PAK knockout with the Fragile X mutant, then maybe the effects would cancel out each other,” Tonegawa said.

After cross-breeding the mutants, the researchers quantified and analyzed Golgi-impregnated spines in pyramidal neurons in the temporal cortex. “To my surprise, but maybe not to hers, the mouse with the mutation in both genes had spines that were very normal,” Tonegawa said. “Not too many or too few. Not too thin or too fat. They looked like normal spines in a normal mouse.”

While basal electrophysiological activity did not differ between FXS and PAK/FXS double mutants, there were differences in responses to electrical stimulation. In the double mutant, PAK inhibition reversed deficits in long-term potentiation observed in the FXS-only mutant. In open field and trace conditioning tasks, the researchers also found that the double mutation reversed behavioral abnormalities, including hyperactivity, stereotypy, and attention deficits. The results are the first demonstration of both a cellular and a behavioral reversal of abnormalities in the FXS mouse model.

To test if protein-protein interaction might explain these findings, Hayashi and colleagues performed an immunoprecipitation study, which indicated that PAK and the Fragile X protein existed in a protein complex in vivo. When the researchers purified the proteins and then mixed them in vitro, they bound together.

Because reduction of PAK activity does not occur until the third postnatal week, the study also indicates that PAK inhibition can reverse FXS syndrome after it appears. “Extrapolate this to children 2 or 3 or 4 years old after they are diagnosed, then you can apply this drug therapy and you can reverse it,” Tonegawa said.

Chemical compounds that inhibit PAK exist, although they were not made for that specific purpose. For instance, the pharmaceutical company Cephalon has a PAK inhibitor in development. “It’s not a strong PAK inhibitor and it was developed for a different kinase,” Hayashi said. Evidence from clinical trials suggests that it crosses the blood-brain barrier, she added.

The relationship between PAK signaling and Alzheimer disease remains unclear, although the pathway seems to be impaired in Alzheimer patients. However, the interplay between the enzyme pathway and synapses in both Alzheimer’s and FXS appears more established (see comment from Greg Cole below). “In general, synaptic structure seems to be a common target,” said Hayashi, who is now identifying new drug targets for Alzheimer’s at Merck Research Laboratories in Boston. “Synapse loss seems to precede cognitive loss in AD and FXS.”—Molly McElroy

Molly McElroy is a freelance writer based in Melbourne, Florida.


  1. The apparent reversal of some of the phenotypic characteristics of Fragile X syndrome by inhibition of enzymatic activity on a signaling pathway, by Dr. Tonegawa’s group, is very exciting. Of course, not all phenotypes are addressed in this paper, and some behavioral reversal effects appear to be at best partial. That said, this appears to be an elegant demonstration that at least some aspects of the neuromorphological and behavioral phenotypes associated with Fragile X syndrome involve the signaling pathway whereby glutamate triggers ERK phosphorylation leading to protein synthesis. While application of this knowledge at the clinical level remains only a distant possibility, this very clear demonstration that downregulation of a signaling pathway can rescue broad aspects of the Fragile X phenotype indicates that an array of seemingly disparate characteristics may share a common dependence upon the activation of well-known enzymatic signaling systems.

  2. Sex and Drugs and Rac and Rho. How Can You Handle a Sick PAK?
    Sex-linked mental retardation genes are probably the best understood genetic causes of cognitive deficits. As proximate causes of cognitive dysfunction from birth, they may provide clues to any final common pathways involved in more tangled pathogenic cascades, for example, late-onset dementias. FMRP KO mice are a model for a common genetic cause of autism and mental retardation, Fragile X syndrome, a disease with greater-than-normal cortical spine density and elongated spines. This paper from Hayashi et al. in the Tonegawa laboratory shows that LTP and behavioral deficits (locomotor activity, stereotypy, anxiety, and trace fear conditioning), and spine defects in FMRP KO mice can be ameliorated by crossing in a dominant negative PAK transgene that limits PAK activity by 41 percent in WT mice. They also provide evidence for an interaction between PAK1 and FMRP using Co-IP and a GST PAK1 pull down assay. The authors suggest that PAK inhibitors might be developed as drugs to be used to treat the disorder.

    This Tonegawa group paper follows a similarly elegant Neuron paper in which the same scientific team characterized synaptic, LTP, and cognitive deficits in the dnPAK transgenic mice used in this FMRP KO paper. The paper definitively adds FMRP to the list of at least eight X-linked mental retardation genes implicated in Rho GTPase family regulation of actin dynamics in spines (Ramakers, 2002). Since spines are structurally dependent on actin assembly, their formation, motility, maturation, and maintenance require fine regulation of actin assembly and disassembly that is essential for synaptogenesis and synaptic plasticity underlying memory.

    The authors have shown that FMRP is a direct PAK1 interacting protein but do not yet know how FMRP fits into the complex signaling network. PAKs are part of a larger family with six PAKs. Group I PAKs ( PAKs 1, 2, and 3) are Rac1/Cdc42 effectors that respond by kinase activation. Group II PAKS (4, 5, and 6) are not activated by Rac1/Cdc42 and are subject to a different system of regulation. (Zhao and Manser, 2004). The Group I PAKS that are direct downstream effectors of Rac1 are embedded in a complex network of Rho GTPase and other targets of extracellular signaling cascades including Ephrins, Plexins, trophic factors, and NMDA receptors.

    Loss-of-function missense mutations in PAK3 cause severe mental retardation. Consistent with this, we have previously suggested that PAK defects in AD include focal ectopic hyperactivation but a general loss of cytosolic PAKs and soluble PAK activity (Zhao et al., 2006). This may contribute to spine deficits characterized by the loss of the dendritic spine, actin-binding protein, drebrin. Given the importance of at least PAK3 in normal cognitive function, it might seem counter-intuitive to treat a mental retardation syndrome with a PAK inhibitor. In our subsequent studies, we have found that Aβ oligomers cause an initial rapid aberrant PAK activation and translocation followed by cytosolic deficits in the chronic steady state and in vivo. In the face of PAK dysregulation, we would probably not want to either activate or inhibit to treat AD but restore more balanced regulation.

    Too much or too little activity in the network regulating spine actin dynamics and synaptic plasticity can promote cognitive deficits. Thus, AD neuroscientists are very familiar with the essential role of NMDA receptor activation in learning and memory, but also with the excesses of aberrant activation (excitotoxicity) and the successful use of the NMDA modulator memantine to treat AD when pure NMDA inhibitors are toxic. As with γ-secretase, the take-home lesson may be that a modulator aimed at shifting a pathogenic imbalance is likely to be more viable in the clinic.

    More generally, defects in these MR/synaptic plasticity pathways can either decrease or increase activity to cause imbalances which result in cognitive deficits. These can occur with excessive LTP or even too many spines as in the case of Fragile X. As Ramakers suggests in his 2002 review: “However, given the tight coupling between many components of the Rho signaling network, any deletion or mutation of crucial components is likely to shift the balance in the network to a suboptimal state, locking actin in a certain configuration (i.e., polymerized or depolymerized, contracted or relaxed). Consequently, neurons might be less responsive to environmental cues, giving rise to suboptimal neuronal connectivity and/or plasticity. In this scenario, it is not important whether a mutation stimulates or inhibits Rho signaling or which Rho GTPase is affected.” The same may be said for ectopic activation or deficits via oligomer-induced Rac/CdC42 activation and downstream PAK pathway effects. However, whether Rac and Rho are druggable remains to be seen.

    The fine balance in the Rho signaling network may not be easy to manipulate in a useful way and is certain to involve tight therapeutic windows. The 41 percent inhibition with dnPAK used in the Hayashi et al. study may offer some insight into a window where a dnPAK can inhibit an MR gene and induce cognitive deficits in WT mice and yet is therapeutic in the Fragile X model. This paper, a successful genetic effort to treat one synaptic plasticity defect with another, offers an interesting, albeit perhaps tangential insight into possibilities for AD therapeutics. It’s only Rac and Rho, but I like it.


    . Rho proteins, mental retardation and the cellular basis of cognition. Trends Neurosci. 2002 Apr;25(4):191-9. PubMed.

    . PAK and other Rho-associated kinases--effectors with surprisingly diverse mechanisms of regulation. Biochem J. 2005 Mar 1;386(Pt 2):201-14. PubMed.

    . Role of p21-activated kinase pathway defects in the cognitive deficits of Alzheimer disease. Nat Neurosci. 2006 Feb;9(2):234-42. PubMed.

Make a Comment

To make a comment you must login or register.


Webinar Citations

  1. Alzheimer Disease—The Fragile X Syndrome Connection

News Citations

  1. AD Pathology—Loss of Kinase Sends Synapses PAKing

Further Reading

Primary Papers

  1. . Inhibition of p21-activated kinase rescues symptoms of fragile X syndrome in mice. Proc Natl Acad Sci U S A. 2007 Jul 3;104(27):11489-94. PubMed.