Maggie Chiang

Boosting the Brain’s Brakes to Beat Memory Loss

Overactivity in the hippocampus, likely tied to lack of inhibition, underlies some age-related cognitive decline. Targeting those circuits shows early promise in slowing memory loss.

As we get older, our memory begins to fail in predictable ways. We’re more likely to confuse similar memories, for example, forgetting whether it was little Tim or Samantha who threw the turkey leg across the room at Thanksgiving, and whether that happened the same year the dog snatched the sweet potato pie. In experiments in the lab, old people can easily remember very different objects but have a harder time remembering similar ones. “One of the hallmarks of age-related cognitive decline is loss of memory precision,” says Amar Sahay, a neuroscientist at Harvard Medical School and an investigator with the Simons Collaboration on Plasticity and the Aging Brain (SCPAB).

Research from Sahay’s group and others has shown that this loss of precision stems from a flaw in the hippocampus, part of the brain that is essential for both navigation and ‘episodic memories,’ or memories of personal experience. Study after study has shown that as we get older, cells in part of the hippocampus become hyperactive. The overactive circuit can no longer create distinct representations specific to new experiences, impeding the formation of new memories. “This deficiency in memory precision can be thought of as a result of increased interference between similar representations,” Sahay says.

The good news is that these circuits are not beyond repair. Scientists are now deciphering the molecular, cellular and circuit-level changes that contribute to hyperactivity and memory issues. The resulting map of age-related differences is highlighting new routes for rejuvenation. “This is an invaluable opportunity,” Sahay says. “By identifying the substrates for this widely recognized phenomenon, we might be able to rectify or reverse those cellular changes.” Researchers have already shown early success in reducing both hyperactivity and memory impairment, including with low doses of an existing epilepsy drug.

A spectrum of aging

Much like humans, animals show remarkable variability in how they age. Michela Gallagher, a neuroscientist at Johns Hopkins University in Baltimore, has spent the last two decades digging into these differences using a genetically diverse strain of outbred rats. Some of these rats show serious memory impairments with age, while others maintain their memory almost as well as their younger counterparts. Examining the molecular and cellular sources of these differences has provided crucial insight to the aging brain.

About 15 years ago, researchers discovered that the old rats with memory problems showed specific deficits in the hippocampus. Neurons in part of the hippocampus called CA3 had abnormally high firing rates compared with their younger counterparts. Rats with more age-related hyperactivity in the hippocampus had worse memory function, and old animals with little or no memory issues show no or less hyperactivity.

The CA3 region of the hippocampus, along with the dentate gyrus, plays an essential role in ‘pattern separation,’ diminishing overlap in how the hippocampus represents similar memories. For example, when an animal explores new surroundings, place cells in the hippocampus, which encode an animal’s location, generate novel place fields. That helps the brain form new spatial maps specific to each environment — and helps us learn the route to a new favorite restaurant or the location of the coffee stand at a new conference venue. “It’s a good setting for looking at plasticity that is relevant for memory,” Gallagher says.

Old people have a harder time remembering if they've previously seen one of a pair of highly similar objects (right) than one of a pair of less similar objects (left.) Yassa et al. 2011

The researchers found that this process is impaired in old rats — their CA3 neurons failed to fire more in novel surroundings, and they no longer adapted their place fields to new settings. In short, the older animals could no longer adequately process new information. “From a computational perspective, hyperactivity in the hippocampus reduces the capacity to discriminate among similar memories,” says Michael Yassa, a neuroscientist at the University of California, Irvine. Without adequate separation, new memories can override previously formed representations.

Since that discovery, other groups have demonstrated similar patterns across species in both brain activity and behavior. Aging rodents, primates and humans all show hyperactivity in the CA3 cells of the hippocampus and difficulty distinguishing similar memories. For example, using brain imaging, Yassa and collaborators have shown that old people have more activity in the dentate gyrus and CA3 than younger people, and those with the worst memory impairments have the most hyperactivity.

Researchers are studying how these changes might contribute to dementia, but Gallagher emphasizes that this pattern occurs even in the absence of brain pathology. “This is a signature of normal aging that is seen in humans and is sensitive to the same interventions as in rats,” she says.

Resilience and inhibition

Of course, not all of Gallagher’s elderly rats failed their memory tests. What distinguished the resilient rats from vulnerable ones? It wasn’t simply that unimpaired old rats resisted the ravages of age. Instead, they seemed to compensate better. “The reason has to do with the balance of excitation and inhibition,” Gallagher says.

Inhibitory neurons, which dampen activity of downstream cells, are an essential component in shaping neural dynamics. The tight control that comes with inhibition is especially important for episodic memory, which requires circuitry capable of creating highly specific representations. Without it, similar memories, such as trying to find your car parked in a different spot in the same lot each day, might interfere with one another. “If you have too much noise in the brain, you lose specificity of encoding,” Gallagher says.

Animal data suggests that age-related hippocampal hyperactivity comes from loss of the inhibitory neurotransmitter GABA. “In general, GABAergic tone goes down in aging,” Yassa says. That can have an especially serious impact on the medial temporal lobe, which houses the hippocampus, because these networks are kept under tight inhibition, he says. The temporal lobe is rife with recurrent connections, which can lead to runaway excitation if not kept under close inhibitory control. “Network dynamics require tight excitatory-inhibitory balance,” Yassa says. “A noisy representation makes it more difficult for the brain to do its job.”

Gallagher and collaborators found that unimpaired aged rats maintained their memory capacity by recruiting more inhibition in the hippocampus than memory-impaired animals, correcting the excitatory-inhibitory balance.

Sahay and collaborators are taking a closer look at how inhibitory neurons shape the circuit, focusing on a class of inhibitory interneurons known as parvalbumin cells. These cells receive input from the dentate gyrus and make synapses onto CA3 and CA2 pyramidal cells, principal neurons in the hippocampus. Encoding of past experiences requires synchronized and precise activation of populations of principal neurons, Sahay says.

Processes of parvalbumin neurons (red) encircle the cell bodies of CA3 principal neurons (blue). Sahay lab

In a 2018 paper published in Nature Medicine, they showed that learning boosts inhibition in the circuit — when animals learn, parvalbumin cells get more excitatory inputs and dial up inhibition onto downstream neurons in CA3.

This inhibition has a number of important roles, including acting as a filter to decrease noise in the network. “Recruitment of GABAergic inhibition is critical for governing the temporal fidelity of spiking, for synchronizing the activity of principle neurons and for generating network oscillations, which we think is essential for storing, encoding and transfer of information from the hippocampus to cortex,” Sahay says. Aging seems to blunt learning’s impact on inhibition, causing memory issues; in aged animals, parvalbumin cells are not recruited to the same degree and fail to show experience-dependent plasticity.

In a recently posted preprint, the researchers showed more specifically how inhibition shapes memory: Boosting this type of inhibition in healthy animals helps generate more precise and stable neural representations, enhancing certain types of long-term memory. A similar strategy may be helpful for improving age-related memory issue.

Parvalbumin neurons aren’t the only inhibitory neurons whose function seems to suffer with age. “GABA inhibition comes in many flavors,” says Yassa. In primates, for example, deficits in somatostatin neurons — which are found in the same region as parvalbumin interneurons but are thought to play a different role in memory — have also been implicated in hyperactivity.

Scientists don’t yet know why inhibitory neurons seem to be particularly vulnerable to aging.

“What about this cell type makes them less efficient?” Sahay asks. “It could be changes within the neurons themselves or how principal neurons recruit them.” Inflammation could also play a role, through changes to surrounding microglia that may alter the circuit. (For more on microglia’s role in brain aging, see Are Similar Processes at Work in Both Development and Cognitive Decline?)

So far, the best evidence for age-related issues with inhibition comes from the hippocampus. It’s an essential structure for memory, and often implicated in both aging and Alzheimer’s disease. But inhibition plays a vital role in other brain regions, and may be similarly vulnerable. “It’s very likely that alterations in inhibitory neurons might occur in other parts of the brain as well,” Sahay says. Other researchers in the SCPAB are studying loss of inhibition in the entorhinal cortex and how that affects grid cells, another type of neuron important for navigation. Some studies have found that parts of the entorhinal cortex are hypoactive, which correlates with hyperactivity in the hippocampus. 

Targeting hyperactivity

When researchers first observed hippocampal hyperactivity in aging brains, some speculated it was not a deficit but rather the brain’s attempt to compensate for a sluggish memory system. Indeed, this perspective was so prevalent 10 years ago that Yassa’s team had difficulty publishing their brain imaging results, which showed that hyperactivity correlated with the level of memory impairment. “At the time, most people thought this was a compensatory ramp-up in activity to deal with the demands of the aging brain,” Yassa says. Pharmaceutical companies began developing therapeutics to block parts of the GABAergic system, enhancing hyperactivity.

Growing evidence produced over the last 10 years supports the opposite theory: Hyperactive neurons are the problem, not the solution. Both Sahay’s and Gallagher’s teams have demonstrated that reducing overactivity in aged rats improves memory deficits. “Overactivity is part of the problem, not overcompensation in any beneficial way,” Gallagher says.

Gallagher’s team has shown that low doses of levetiracetam, an epilepsy drug, reduces hyperactivity in people with mild cognitive impairment, a precursor to Alzheimer’s disease, and improves performance on memory tests. “It was the first hard piece of evidence to show hyperactivity was problematic, that you could reverse it and improve performance,” says Yassa, who collaborated on the study.

The drug, sold under the brand name Keppra, inhibits release of glutamate, reducing excitation. “It replaces interneuron function by boosting the receptor the interneurons use, increasing the gain in the GABA-activated receptor,” Gallagher says. “It’s not creating a different information-processing pattern, just turning up what’s going on anyway.”

Gallagher has launched a company, AgeneBio, to develop drugs for dementia and other neurological conditions. The company is running a late-stage clinical trial of a specially formulated version of levetiracetam for mild cognitive impairment, with results expected late next year. Because overactivity starts a couple of decades before memory impairment, it’s a promising target for slowing cognitive decline, Gallagher says.

Sahay and collaborators have taken a different approach to reducing hippocampal hyperactivity: They countered age-related decline in inhibition by downregulating a protein called ABLIM3. ABLIM3 is expressed in the dentate gyrus, rather than the inhibitory neurons themselves, and stops granule cells from recruiting parvalbumin inhibitory neurons. Removing it boosts inhibition and improves memory.

Sahay’s team is now searching for other molecular factors that recruit parvalbumin inhibitory neurons, as well as molecules within inhibitory interneurons themselves that enhance downstream connections. “What is the molecular code that allows these neurons to respond to experience and then dial up inhibition?” Sahay asks. Deciphering that code will inform efforts to enhance it in the aging brain, in turn improving memory encoding and storage, and restoring network oscillations, synchrony between the hippocampus and cortex, and other important neural activity patterns, he says.

Inhibitory neurons are a useful target for intervention because of their potential for broad impact. Dampening hyperactivity restores synchrony, generating a large effect on network activity. Sahay likens inhibitory interneuron to conductors capable of regulating the activity of thousands of neurons. “It makes sense to regulate the conductor rather than the violinist,” he says. AgeneBio and others are developing drugs to directly enhance inhibition, such as targeting specific GABA receptors.

These efforts will likely have applications beyond aging. Issues with inhibition and synchronization have been implicated in a number of conditions, such as epilepsy, autism and schizophrenia. Indeed, Sahay points to a recent newsworthy development: Cronutt the epileptic sea lion, who was successfully treated with a transplant of inhibitory neurons to the hippocampus. As of October, Cronutt was one year seizure-free.

“Disorders that arise in early life and disorders that evolve over time appear to be influenced by optimal functioning of parvalbumin neurons,” says Sahay, who is also studying autism. “Many different sets of data are now motivating people to continue to study these inhibitory neurons.”



Recent Articles