Finally, in the same neurons, treatment with Aβ42 oligomers led to a slight, albeit reproducible and significant, MK8776 increase in Tau phosphorylation on S262 in control AMPKα1+/+ but not in AMPKα1 null hippocampal neurons (Figures 6E and 6F), suggesting that AMPKα1 mediates the phosphorylation of Tau on S262 induced by Aβ42 oligomers in hippocampal neurons. Loss of synapses begins during the early stages of AD and progressively affects neuronal network activity, leading to cognitive dysfunction (Coleman and Yao, 2003; Palop and Mucke, 2010; Terry

et al., 1991). In vitro and in vivo studies have demonstrated that Aβ oligomers are contributing to early synapse loss (Hsia et al., 1999; Hsieh et al., 2006; Lacor et al., 2007; Mucke et al., 2000; Shankar et al., 2007), whereas recent studies support Tau as one of

the mediators of Aβ toxicity in dendrites (Ittner et al., 2010; Roberson et al., 2007, 2011). However, our understanding of the molecular mechanisms linking Aβ oligomers and Tau synaptotoxicity in dendritic spines remains incomplete. Here, we report that (1) AMPK is overactivated in hippocampal neurons upon application of Aβ42 oligomers, and this activation is dependent on CAMKK2; (2) CAMKK2 or AMPK activation is sufficient to induce dendritic spine loss in hippocampal neurons in vitro and in vivo; (3) Aβ-mediated activation of AMPK induces the phosphorylation of Tau on residue S262 in the microtubule-binding domain; check details and (4) inhibition of either CAMKK2 or AMPK catalytic activity, or expression of a nonphosphorylatable form of Tau (S262A), blocks Aβ42 oligomer-induced PDK4 synaptotoxicity in hippocampal neurons in vitro and in vivo. AMPK is an important homeostatic regulator and is activated by various forms of cellular and metabolic stresses (Mihaylova and Shaw, 2011; Shaw et al., 2004). Oxidative stress such as elevation of ROS can activate AMPK through a mechanism that is still

unclear (reviewed in Hardie, 2007). Because part of the neuronal toxicity induced by Aβ is thought to involve increased ROS production (Schon and Przedborski, 2011), future experiments should test if AMPK function during Aβ-mediated neurodegeneration requires the ability of ROS to activate AMPK. In the brain, AMPK activity is increased in response to metabolic stresses such as ischemia, hypoxia, or glucose deprivation (Culmsee et al., 2001; Gadalla et al., 2004; Kuramoto et al., 2007; McCullough et al., 2005) and is abnormally elevated in several human neurodegenerative disorders, including AD and other tauopathies, amyotrophic lateral sclerosis, and Huntington’s disease (Ju et al., 2011; Lim et al., 2012; Vingtdeux et al., 2011b). Whether activation of AMPK in these different pathological contexts has a neuroprotective or deleterious outcome in various neuronal subtypes remains controversial (Salminen et al., 2011).