Recent failures of clinical trials targeting amyloid-β in mild to moderate Alzheimer’s disease (AD) have led to claims that the interventions are too little too late, and that irreversible damage had occurred in the brain. A March 1 article in Nature suggests that at least some of the damage is epigenetic, and offers hope that it can eventually be repaired. Aβ restricts expression of genes related to learning and memory—a blockade that persists even if Aβ is removed, suggest researchers led by Li-Huei Tsai, Massachusetts Institute of Technology, Cambridge. Tsai and colleagues found that a histone deacetylase that silences a host of memory-related genes is activated in the brains of both mouse models of neurodegenerative disease and in AD patients. Knocking down the enzyme reversed gene repression in the mice and restored their memory capacity. The results could have implications for drug development. "We think that this blockade of gene expression involved in memory formation plays a very important role in the cognitive aspect of Alzheimer's disease," Tsai told ARF.

Histones—which act like spools to wrap and compact DNA in the cell—keep genes relatively quiet until acetyl groups are added to the proteins and the DNA unwinds. Histone deacetylases (HDACs)—18 types in all—yank off those gene-freeing acetyl groups to quiet gene expression again. If HDACs are too active, histone acetylation may run amok and important genes stay repressed.

Tsai's lab has probed the role of HDACs in AD for some years. Previously, the team found that general HDAC inhibitors, which lead to increased histone acetylation, restored lost long-term memories and enhanced learning in a mouse model of AD (ARF related news story on Fischer et al., 2007). The lab then showed that a particular class I HDAC called HDAC2 was important in learning and memory. Overexpression in mice quashed synaptic density and memory formation, while underexpression did the opposite (see ARF related news story on Guan et al., 2009). It was unclear, however, how HDAC2 exerted its effects.

To probe HDAC2's action, first author Johannes Gräff and colleagues looked at expression in the CK-p25 mouse developed in Tsai’s lab. This transgenic model of neurodegenerative disease expresses p25, a truncated form of p35 that activates cyclin-dependent kinase 5 (Cdk5). The mice develop Aβ and tau pathology, and reduced synaptic density and neuron loss (ARF related news story). Comparing this model to wild-type mice, Gräff and colleagues found higher levels of HDAC2 in hippocampal area CA1 and in the prefrontal cortex, while related class I HDACs (HDAC1 and HDAC3, also involved in memory formation) remained normal. A different, 5xFAD mouse model showed similar elevation in HDAC2. Chromatin immunoprecipitation revealed enriched HDAC2 binding to genes involved in learning and memory, including genes that are tuned down in humans with AD. These included immediate-early genes such as Homer1 and Arc, and genes related to synaptic plasticity, such as certain glutamate receptor subunits and synaptophysin. Histones associated with those genes sported fewer acetyl groups than usual and expression of those genes dropped off. HDAC2 seemed to be silencing genes involved in neuroplasticity.

Would knocking down HDAC2 restore gene expression? To find out, the team injected HDAC2-specific short hairpin RNAs (shRNA) into the hippocampal CA1 areas of CK-p25 mice. They injected, as controls, scrambled, non-specific shRNAs into CK-p25 and wild-type mice. The shRNA squashed HDAC2 levels and boosted mRNA expression of the target genes involved in learning and memory. Synaptic plasticity and abundance of dendrites, as visualized by MAP2 immunoreactivity, exceeded that of control CK-p25 mice and nearly matched levels of wild-type mouse controls. Knocking down HDAC2 did not boost the number of surviving neurons in CK-p25 mice. However, compared to controls, the animals’ associative memory in fear-conditioning tasks returned to normal and the mice also found refuge on the hidden platform in the Morris water maze more quickly. Treated mice spent as much time as wild-type mice in the correct quadrant in probe trials the next day. "The good news is that, even though there's an epigenetic blockade, this is potentially reversible," said Tsai.

To find Aβ's link to a surge in HDAC2 expression, the researchers exposed cultured hippocampal neurons to synthetic peptide. HDAC2 mRNA levels rose. What drove the increased expression? In the promoter region of HDAC2, Gräff and colleagues found a recognition element for glucocorticoid receptor 1 (GR1), a transcriptional activator when phosphorylated during stress. They wondered if stress brought on by Aβ might activate GR1. Sure enough, after application of Aβ42 to cultured wild-type hippocampal neurons, GR1 displayed more phosphate and bound to the HDAC2 promoter, enhancing HDAC2 transcription. Phosphorylated GR1 also decorated HDAC2 promoters in CK-p25 hippocampi, where the researchers saw an increase in HDAC2 transcription. This provides a path between neurotoxic Aβ and HDAC2 expression, said the authors.

"I think now we offer a novel explanation for the pathogenesis of Alzheimer's disease, and how the Aβ peptide can exert a long-term effect in the brain," said Tsai. "It can increase HDAC2 and cause a chronic blockade of expression of genes necessary for memory formation." Upon examination of postmortem human brains with AD, the team then found that HDAC2—but not HDAC1 or HDAC3—was present at higher levels in the CA1 and entorhinal areas, even in early stages of the disease. "This advocates for the development of HDAC2 inhibitors," added Tsai, who reported no involvement in companies working on HDAC inhibitors.

The implication that human AD deficits may be reversible "should create excitement in the patient community, especially given the fact that HDAC inhibitors are currently being explored in the clinic for a variety of conditions," said Randal Tibbetts, University of Wisconsin, Madison, in an e-mail to ARF. HDAC inhibitors are already used in cancer treatment, and companies including Envivo Pharmaceuticals are pursuing them as potential AD therapeutic targets. Since the authors only looked for certain genes that are suppressed by HDAC2, more research is needed to determine if others are affected as well, he added. Tsai said she and her team plan to do that next.

This provides an intriguing mechanism for one of Aβ's detrimental effects, said Jang-Ho Cha of Merck, North Wales, Pennsylvania. "One of the big gaps in Alzheimer's study is understanding what Aβ is doing," Cha said. "It is not clear that this is the only thing it is doing, but this pathway they are mapping out is really elegant." Addition and removal of acetyl groups are fairly transient forms of gene regulation, but like blood attracts sharks, HDACs attract enzymes that more permanently alter the accessibility of DNA's coils. For instance, some enzymes methylate DNA. Such enzymes might provide more selective drug targets than HDAC inhibitors, which may affect more than one HDAC, Cha pointed out.—Gwyneth Dickey Zakaib


  1. Overall, this is a nice paper that is comprehensive in its approach. The use of two different Alzheimer’s disease (AD) mouse models is a strength of the paper. Of course, none of these AD models fully reflects the human AD pathology, and it is always uncertain how these findings will translate to the human system. The gene expression studies demonstrating HDAC2-dependent repression of synaptic target genes in the AD mouse models was compelling. The gene expression changes will require additional study, as only a handful of genes were tested. Which of these genes is most critical to reversing the pathology is of interest, but will be hard to address. The fact that HDAC2 binds to many of the downregulated genes supports a causal role for HDAC2-mediated repression in the associated learning and memory defects. However, this is not an easy thing to prove; it is possible that HDAC2 knockdown reverses a different pathologic process, and that the gene expression changes are a secondary consequence. Of course, this does not diminish the fact that HDAC2 appears to be an interesting target. Although selective targeting of HDAC2 is certainly indicated from this study, it may be difficult given its high degree of similarity to HDAC1. In sum, the idea that crucial neuronal genes are epigenetically and reversibly downregulated in experimental AD is supported by the data. This would imply, as the authors suggest, that some of the deficits in human AD may be reversible, which should create excitement in the community, especially given the fact that HDAC inhibitors are currently being explored in the clinic for a variety of conditions.

    View all comments by Randal S. Tibbetts

Make a Comment

To make a comment you must login or register.


News Citations

  1. Memories—Forgotten, But Not Gone?
  2. It’s an HDAC2 Wrap— Memory-suppressing DNA Modifier Identified
  3. Tangles, Neurodegeneration, Plaques—p25 Does it All

Paper Citations

  1. . Recovery of learning and memory is associated with chromatin remodelling. Nature. 2007 May 10;447(7141):178-82. PubMed.
  2. . HDAC2 negatively regulates memory formation and synaptic plasticity. Nature. 2009 May 7;459(7243):55-60. PubMed.

External Citations

  1. CK-p25 mouse
  2. 5xFAD mouse model

Further Reading


  1. . Modulation of long-term memory for object recognition via HDAC inhibition. Proc Natl Acad Sci U S A. 2009 Jun 9;106(23):9447-52. PubMed.
  2. . HDAC inhibitor increases histone H3 acetylation and reduces microglia inflammatory response following traumatic brain injury in rats. Brain Res. 2008 Jun 11; PubMed.
  3. . HDAC activity is required for BDNF to increase quantal neurotransmitter release and dendritic spine density in CA1 pyramidal neurons. Hippocampus. 2011 Dec 7; PubMed.

Primary Papers

  1. . An epigenetic blockade of cognitive functions in the neurodegenerating brain. Nature. 2012 Mar 8;483(7388):222-6. PubMed.