A pair of papers proposes a new reason that learning and memory fade in Alzheimer’s disease (AD). Two independent groups found that astrocytes in the hippocampus of mouse models produce and release high levels of the inhibitory neurotransmitter GABA, dampening neuronal plasticity and causing cognitive deficits. Unusually high levels of GABA also fill hippocampal astrocytes in patients with AD, according to the studies. C. Justin Lee, Korea Institute of Science and Technology (KIST) in Seoul, was senior author on one paper in the June 29 Nature Medicine. The other, from the lab of Gong Chen, Pennsylvania State University, University Park, came out in the June 13 Nature Communications. “This GABA release heavily inhibits synaptic transmission and basically shuts down whole circuits,” Lee told Alzforum. “The hippocampus is rendered incapable of affecting memory formation or recall,” he said.

Glaring Inhibition. At left, a reactive astrocyte makes as much GABA (red) as a GABAergic neuron (arrow). At right, the astrocyte sits near an Aβ plaque (blue). See text below. Image courtesy of Jo et al., Nature Medicine.

GABA is the major inhibitory neurotransmitter in the brain. It can cause persistent, or tonic, inhibition of excitatory neurons when it acts on receptors outside the synapse, or acute inhibition when it acts at the synapse. Lee’s group previously reported that astrocytes in the cerebellum produce GABA and release it through bestrophin 1 (Best1), a calcium-activated channel that can pump GABA from the cell to tonically inhibit neurons (see Lee et al., 2010). They found that hippocampal astrocytes, on the other hand, contained no GABA, but still expressed Best 1 (see Yoon et al., 2011). Lee and colleagues wondered if hippocampal astrocytes might produce and release GABA through this channel under certain circumstances, such as in Alzheimer's disease.

To find out, co-first authors Seonmi Jo and Oleg Yarishkin examined hippocampal slices from APP/PS1 mice. They stained for plaques using Thioflavin-S and for reactive astrocytes using antibodies to glial fibrillary acidic protein, a marker of astrocyte activation. As others have described, they found reactive astrocytes clustered around plaques starting around four months of age, especially in the dentate gyrus (see for example Serrano-Pozo et al., 2013). Microdialysis revealed that in this brain region, the concentration of GABA in the interstitial fluid rose. Electrophysiological measures indicated that tonic inhibition of granule cells from the dentate gyrus increased accordingly. In brain samples from 8-month–old APP/PS1 mice, astrocytes in the dentate gyrus stained for five times more GABA than control tissue, as much as nearby GABAergic interneurons (see image above). The GABA increase also showed up in 5xFAD mice. An anion channel blocker called natriuretic peptide B (NPPB) halted GABA’s release, suggesting that GABA exited astrocytes via Best1 channels, which are sensitive to the compound.

What caused the excess GABA? The authors pointed out that the neurotransmitter can be produced by the breakdown of putrescine. They found this polyamine abundant near plaques. They also found more monoamine oxidase-B (Maob), which is reportedly elevated in the brains of AD patients (see Saura et al., 1994), in hippocampal astrocytes of the APP/PS1 mice, compared with control animals. Maob catalyzes one of the key steps in conversion of putrescine to GABA. Bathing isolated astrocytes for two hours in an irreversible inhibitor of Maob, called selegiline, reduced GABA release.

As previously reported, long-term potentiation (LTP), a form of synaptic plasticity crucial for learning and memory, was impaired in these APP/PS1 mice, and they forgot a context-based fear sooner than controls. However, after a week of selegiline treatment, LTP and behavioral deficits returned to normal. Selegiline effects wore off after two weeks of treatment, possibly because the inhibitor is irreversible and causes permanent damage. A reversible Maob inhibitor called safinamide worked longer, suggesting this type of treatment could be more effective. 

Chen and colleagues found something similar—that reactive astrocytes in the dentate gyrus of 5xFAD mice produce high amounts of GABA, which causes tonic inhibition in dentate granule cells. First author Zheng Wu and colleagues saw that this excess GABA impaired LTP in hippocampal slices.

However, these researchers proposed a different mechanism for increased GABA. They posit that glutamate builds up in astrocytes near plaques and glutamic acid decarboxylase (GAD) converts it to the inhibitory neurotransmitter. According to their results, reactive astrocytes of mice and humans have more of the GAD67 isoform than quiescent astrocytes do. They further diverge with Lee's group, proposing that the neurotransmitter exits astrocytes through the astrocyte-specific GABA transporter GAT3/4. They found more of this transporter in astrocytes of 5xFAD mice and in brain tissue from AD patients than in corresponding healthy control tissue. Shutting down GAT3/4 with an inhibitor called SNAP-5114 reduced tonic inhibition on dentate granule cells from 5xFAD mice. Lee pointed out that his group saw no difference in GAD67 between mouse models and controls, nor did GABA transporter inhibitors prevent tonic inhibition in the cerebellum in his experiments.

Wu and colleagues also detected differences in the receptors that bind GABA. These complexes comprise five subunits that mix and match to form different isoforms. Extrasynaptic receptors containing the α5 subunit proliferated on dentate granule cells from 5xFAD mice compared to control animals. Furthermore, an α5-specific inverse agonist L-655,708 blocked most of the GABAergic tonic inhibition in hippocampal slices from these mice. Both SNAP-5114 and L-655,708 reversed LTP deficits in hippocampal slices. Mice given these compounds also performed normally in a Y maze test of cognition. 

Are these data relevant to humans, or is this just a quirk of animal models? Lee said he expected that since people have larger astrocytes than mice do, and more per neuron, any effects they see in rodents may be amplified in those with AD. In AD patient brains from the brain bank at the Emory University School of Medicine in Atlanta, Chen’s group found that reactive astrocytes in the dentate gyrus had more GABA than did astrocytes in healthy control tissues. Likewise, Lee’s group examined 22 postmortem human brain samples—half from healthy donors and half with evidence of AD pathology—from the Boston University Alzheimer’s Disease Center. The AD brains had more GABA, GFAP, and Maob in astrocytes of the temporal cortex. Astrocytic GABA could be a good biomarker, diagnostic tool, and drug target for AD, suggested Chen’s group. What’s more, Lee believes the pathology may be common to other diseases that involve reactive astrocytes, such as Huntington’s and Parkinson’s diseases. Safinamide is currently in Phase 3 trials for PD. Meanwhile, Lee and colleagues are looking for similar compounds with fewer side effects. 

“These results add a new cellular dimension to findings showing aberrant GABAergic function in the dentate gyrus of multiple mouse models of AD and Down’s syndrome,” Jorge Palop, University of California, San Francisco, wrote to Alzforum in an email. “Targeting specifically tonic inhibition adds a novel and clever therapeutic angle to address the over-inhibition of dentate granule cells in AD and DS.”

Combined with previous studies, the two papers highlight the need for balance between excitatory and inhibitory signaling in the brain, said Yadong Huang, Gladstone Institute of Neurological Disease, San Francisco. Work from Huang's group suggests that too little inhibition causes memory deficits in ApoE4 knock-in mice (see Feb 2012 news story). It will be interesting to dissect out the true mechanism of GABA production and release from reactive astrocytes, as it will be important for development of a therapy to target it, he said. It will also be useful to know when in disease the elevated tonic inhibition occurs, to predict at which stage a drug might be effective, he said.—Gwyneth Dickey Zakaib


  1. To add to the discussion about mechanism, the undersigned and co-author Stanislaw Mitew would like to point out the findings of a recent study from our lab at the Wicking Centre (Mitew et al., 2013). In it, we reported no significant differences in the protein levels of GAD65/67 between wild-type and APPswe/PS1dE9 transgenics aged 12 months, but GAD activity (i.e., GABA synthesis) was significantly elevated in the cerebral cortex of the APP/PS1 mice. Furthermore, this increased activity was solely attributable to the glial fraction and not the synaptosome fractions after Percoll gradient separation of the crude synaptosome fraction from cortex. So clearly the GAD enzymes themselves are rendered more active in the presence of deposited Aβ. This would either amplify any increased-substrate effect of putrescine or glutamate accumulation, or explain higher GABA levels without requiring more substrate. The assay measures GABA production by tissue homogenates directly without differentiating between GAD65 and GAD67. 


    . Altered synapses and gliotransmission in Alzheimer's disease and AD model mice. Neurobiol Aging. 2013 Oct;34(10):2341-51. PubMed.

Make a Comment

To make a comment you must login or register.


Research Models Citations

  1. 5xFAD (B6SJL)

News Citations

  1. San Francisco: GABA Neurons Blamed for Memory Loss in ApoE Mice

Paper Citations

  1. . Channel-mediated tonic GABA release from glia. Science. 2010 Nov 5;330(6005):790-6. PubMed.
  2. . The amount of astrocytic GABA positively correlates with the degree of tonic inhibition in hippocampal CA1 and cerebellum. Mol Brain. 2011 Nov 22;4:42. PubMed.
  3. . Differential relationships of reactive astrocytes and microglia to fibrillar amyloid deposits in Alzheimer disease. J Neuropathol Exp Neurol. 2013 Jun;72(6):462-71. PubMed.
  4. . Increased monoamine oxidase B activity in plaque-associated astrocytes of Alzheimer brains revealed by quantitative enzyme radioautography. Neuroscience. 1994 Sep;62(1):15-30. PubMed.

External Citations

  1. APP/PS1
  2. Phase 3 trials

Further Reading

Primary Papers

  1. . GABA from reactive astrocytes impairs memory in mouse models of Alzheimer's disease. Nat Med. 2014 Aug;20(8):886-96. Epub 2014 Jun 29 PubMed.
  2. . Tonic inhibition in dentate gyrus impairs long-term potentiation and memory in an Alzhiemer's disease model. Nat Commun. 2014 Jun 13;5:4159. PubMed.