Aging and apolipoprotein E4 (ApoE4) can increase the risk of cognitive impairment and neurodegenerative disorders, including Alzheimer's disease (AD), and patients with type 2 diabetes mellitus are highly susceptible to cognitive dysfunction. Recent research has indicated that metformin, a prescribed drug for type 2 diabetes, may affect cognitive function; however, findings regarding its efficacy are largely controversial. The current study reported that a 5‐mo metformin administration (300 mg/kg/d) starting at 13 mo old improved the spatial memory of ApoE3‐target replacement (TR) mice, not ApoE4‐TR mice. It found that in aged ApoE3‐TR mice, metformin treatment, at a molecular level, inhibited AMPK activity, increased insulin signaling, and activated mammalian target of rapamycin signaling, resulting in an enhanced expression of postsynaptic proteins; but the response of the neuronal AMPK activity and insulin signaling to metformin was blunt in aged ApoE4‐TR mice. Meanwhile, metformin treatment also increased the phosphorylation of tau in both ApoE3‐TR and ApoE4‐TR mice, implying that metformin may have side effects in human. These findings suggest that metformin can improve the cognitive performance of aged mice in an APOE genotype–dependent manner, which provides empirical insights into the clinical value of metformin for ApoE4‐ and age‐related AD prevention and treatment.—Zhang, J., Lin, Y., Dai, X., Fang, W., Wu, X., Chen, X. Metformin treatment improves the spatial memory of aged mice in an APOE genotype–dependent manner. FASEB J. 33, 7748–7757 (2019). http://www.fasebj.org

Background Nuclear acetyl-CoA pools govern histone acetylation that controls synaptic plasticity and contributes to cognitive deterioration in patients with Alzheimer’s disease (AD). Nuclear acetyl-CoA pools are generated partially from local acetate that is metabolized by acetyl-CoA synthetase 2 (ACSS2). However, the underlying mechanism of histone acetylation dysregulation in AD remains poorly understood. Methods We detected ACSS2 expression and histone acetylation levels in the brains of AD patients and 5 × FAD mice. When we altered ACSS2 expression by injecting adeno-associated virus into the dorsal hippocampus of 5 × FAD mice and replenished ACSS2 substrate (acetate), we observed changes in cognitive function by Morris water maze. We next performed RNA-seq, ChIP-qPCR, and electrophysiology to study molecular mechanism underlying ACSS2-mediated spatial learning and memory in 5 × FAD mice. Results We reported that ACSS2 expression and histone acetylation (H3K9, H4K12) were reduced in the hippocampus and prefrontal cortex of 5 × FAD mice. Reduced ACSS2 levels were also observed in the temporal cortex of AD patients. 5 × FAD mice exhibited a low enrichment of acetylated histones on the promoters of NMDARs and AMPARs, together with impaired basal and activity-dependent synaptic plasticity, all of which were rescued by ACSS2 upregulation. Moreover, acetate replenishment enhanced ac-H3K9 and ac-H4K12 in 5 × FAD mice, leading to an increase of NMDARs and AMPARs and a restoration of synaptic plasticity and cognitive function in an ACSS2-dependent manner. Conclusion ACSS2 is a key molecular switch of cognitive impairment and that targeting ACSS2 or acetate administration may serve as a novel therapeutic strategy for the treatment of intermediate or advanced AD. Graphical Abstract Nuclear acetyl-CoA pools are generated partly from local acetate that is metabolized by acetyl-CoA synthetase 2 (ACSS2). Model depicts that ACSS2 expression is downregulated in the brains of 5×FAD model mice and AD patients. Of note, ACSS2 downregulation mediates a reduction in ionotropic glutamate receptor expression through histone acetylation, which exacerbates synaptic plasticity impairment in AD. These deficits can be rescued by ACSS2 upregulation or acetate supplementation (GTA, an FDA-approved food additive), which may serve as a promising therapeutic strategy for AD treatment.