Memories lost and found

For nearly a decade, neuro­scientist Li-Huei Tsai and her colleagues have been studying senile mice. In a lab at MIT her team has genetically fast-forwarded the mice into a condition much like dementia: They have problems making new memories and retrieving old ones. The mice forget how to navigate water mazes they had mastered; they don’t recognize signs of imminent danger they had once responded to fearfully.

Last year, Tsai’s group found a way to reverse the process. When given a drug known to strengthen nerve cell connections in the brain, the mice not only gained back the ability to learn new tasks, but also remembered many forgotten behaviors.

On the opposite coast, researchers are using a similar drug to rewire long-held memories in mice facing another kind of mental challenge: drug addiction. Neurobiologist Marcelo Wood of the University of California, Irvine coaxes cocaine-seeking mice to view the sights and sounds they’ve learned to associate with getting cocaine. He then creates a new, harmless memory around those cues. After a single treatment, mice placed near their drug den forget their cravings.

Though Tsai and Wood use different drugs in their studies, both draw on research showing that the ability to learn and remember can be influenced by subtle changes to DNA — changes that affect how genes turn on and off without altering the underlying genetic information. Such epigenetic modifications, it turns out, might have a profound impact on long-term memory.

Exploring these methods has opened a growing field of research, called neuroepigenetics, aimed at finding ways to boost memory in humans. Results so far offer the prospect of new types of medication to improve memory and even restore long-forgotten information in disorders such as Alzheimer’s disease, Huntington’s disease or other types of dementia. Someday drugs might also treat other memory impairments, including the fogginess that plagues many people as they age, and developmental deficits such as autism. The findings also suggest potential new strategies to treat drug addiction in people.

Memory lane

Memory provides a link between the present and past, and it creates a foundation for learning throughout life. Without your memory, you couldn’t read this sentence or find your way home at night. Recall helps animals, from the lowly sea slug to mice to humans, navigate through life. But only in recent decades have scientists begun to unravel the mysteries of how memories are created and stored for the long haul.

For years, memory research was largely confined to studies of animals and to a few people whose memory had dramatically unraveled. One of the most famous human cases was a man named Henry Molaison, known as H.M. in scientific studies. Scientists learned from H.M. that the hippocampus is an essential part of the brain for making and retaining memories.

Newer studies focus on the part of the memory process that involves strengthening links between nerve cells. For this job, cells have to make proteins.

Several mechanisms can turn protein production on and off in brain cells. One employs various enzymes that change how genes, segments of DNA, are bundled together. DNA is tightly intertwined with proteins known as histones, assembled in a complex called chromatin. Through a process called acetylation — the attaching of a little molecule called an acetyl group — some enzymes relax chromatin. This allows it to open so machinery can access the genetic blueprint for a protein. Other enzymes clamp down on chromatin, blocking genes from being activated when they’re not needed.

One enzyme family — called “histone deacetylases,” or HDACs — helps keep DNA and histones tightly bound by keeping acetyl groups off. In the late 1990s, researchers developed HDAC inhibitors as chemotherapy agents against cancer. Working in a way that sounds like a double-negative, the chemo agents repress the DNA-inhibiting action of the HDACs, resulting in free-flowing gene activity. These anticancer agents proved helpful for treating some tumors in people.

When J. David Sweatt, a neurobiologist at the University of Alabama at Birmingham, gave the agents to lab animals, he found that the drugs turned absent-minded subjects into attentive ones. Intrigued by these findings, scientists tested the idea that histone acetylation might come into play when new memories are formed.

In 2004, Sweatt’s group showed that HDAC inhibitors helped boost memory in rats learning to navigate in an unfamiliar situation. The rats’ improved memory was accompanied by changes in nerve cells, or neurons, in the hippocampus. In 2007, Tsai’s group gave HDAC inhibitors to mice with memory problems and found that they were able to recall things they had forgotten — in this case, dangers associated with certain environments.

Since then, several labs have begun studying how histone acetylation and deacetylation work to activate or shut down learning and memory. It’s now clear, Sweatt says, that the epigenetic mechanisms are key in controlling  gene activity that’s necessary for many different forms of long-term memory.

Ordinarily, the chromatin in brain cells responds to all kinds of activity and stimulation, Sweatt says. When experiencing an event — a child’s birthday party, a lively lecture or even a good book — some chromatin relaxes, some genes are turned on and the brain pumps out proteins that help store the memory.

But disruptions of epigenetic mechanisms can lead to gene silencing, changing a neuron’s behavior for months or even years. In some cases, genes needed for memory and learning can be permanently deactivated. Tsai says this appears to happen in the brains of people with Alzheimer’s.

Making connections

Though it’s hard to fully re-create Alzheimer’s disease in a rodent, Tsai and colleagues found a way to genetically induce similar conditions in mice. Their method causes a mouse to lose 30 to 40 percent of the brain cells in its hippocampus.

To test the effects of this loss, the researchers first trained the mice to perform various tasks with their hippocampi still intact, then waited four weeks for the memories to consolidate and become stored in various parts of the brain. The team then triggered brain cell loss and allowed several more weeks for changes in the hippocampus to unfold.

As their brains began to shrink and shrivel, the mice forgot how to do jobs they had previously mastered. In one task, mice had learned to associate a shock with moving in a certain cage and would freeze in place to avoid pain. After losing hippocampus cells, the same mice forgot to associate fear with the cage and kept walking. In another test, the mice learned to escape from a pool of murky water by finding a submerged platform. The brain-damaged mice couldn’t remember where the platform was.

When Tsai treated the mice with the HDAC inhibitors that had been designed to attack cancer, their memory improved. But there was a problem: Tsai wasn’t sure which enzymes the drugs were targeting in the brain. The HDAC family comprises 11 different enzymes, sequentially named HDAC1 through HDAC11. The cancer drugs contained a cocktail of the HDAC inhibitors targeting a mix of the enzymes.

So Tsai gave the brain-damaged mice a small molecule that targeted only one of the HDAC enzymes, HDAC2. Studies had shown that HDAC2 levels rise in the aging brain, even in healthy people. After treatment with HDAC2 inhibitors, the mice again went through their training paces.

Remarkably, brain-damaged mice given HDAC2 inhibitors performed nearly as well as healthy mice in the water experiment. When the mice were placed in the cage where they had once been shocked, their memories again kicked in: They froze just as often as healthy mice.

“That was a remarkable piece of information,” Tsai says. “After the small-molecule treatment, we found that the seemingly lost memory somehow was recovered.”

In the mice given HDAC2 inhibitors, the scientists found an increase in the number of connections between nerve cells in the hippocampus. Tsai believes that by activating the genes used in learning and memory, the brain rewires surviving neurons, helping them reconnect to cells that may have been damaged. “We think that the key to memory formation and memory retrieval, even in the Alzheimer’s brain, is the healthy connection between neurons,” she says.

Despite how dissimilar mice are to humans, Tsai’s data may also apply to people. “These results suggest that perhaps even in humans, when people start to show signs of dementia, there are still memory traces left somewhere in the brain,” she says. Her group also examined HDAC2 levels in autopsied human brain tissue and found that even people in the earliest stages of Alzheimer’s disease had elevated HDAC2 levels.

Tsai says the findings, published last year in Nature, suggest that gene regulation at the epigenetic level works as a sort of master switch, coordinating a variety of genes needed for learning and memory. If this switch is turned off, memories fade.

Molecular brake pad

When it comes to creating a memorable event, it’s not just what the brain does that matters. What it doesn’t do also affects long-term memory. Studies of how the brain selectively decides what to keep are illuminating how this process works.

At UC Irvine, Wood studies a protein that causes chromatin to relax called the CREB-binding protein. In this state, genes needed for memory formation are easily accessed. But one member of the HDAC enzyme family — HDAC3 — counteracts this effect, clamping down on chromatin and turning genes off.

In 2011, Wood’s group deleted the HDAC3 gene in a small group of hippocampal neurons in mice. This genetic manipulation transformed ordinary events into unforgettable ones. Later, Wood’s group used a drug to selectively inhibit HDAC3 activity in the brain. The results were the same: New environments or experiences were immediately committed to the animal’s memory, and easily summoned days or weeks later.

Wood says HDAC3 and HDAC2 appear to serve as brake pads that are always engaged, working to slow down or stop the steady stream of information encoded into memory. This mechanism allows you to hold onto information if it’s important, but discard it if it’s not.

“Our long-term memory is very, very selective,” he says. “One of the most important things your brain does with respect to memory is to actively prevent you from encoding everything that you experience.”

So inhibiting HDAC3 doesn’t just simply enhance memory, Wood says. “This was fundamentally different. It was like you had released a molecular brake pad so that information that is being acquired enters the realm of long-term formation without any constraints.”

Wood’s group is now working on ways to harness this gating mechanism to manipulate certain kinds of memories, such as those associated with addictive drugs. Studies show that as drugs take over the brain’s reward system, they change the way neurons communicate with each other. Then the mere sight of the location where the owner of the brain indulged — or even sounds and odors associated with drug use — can trigger an intense desire for the drug. Such changes are long-lasting and persistent.

Wood’s group designed an experiment to try to override the changes produced by addictive drugs such as cocaine. First, mice were taught to associate a particular environment with the drug. Upon entering a chamber with checkered walls and scented bedding, the mice received a drug reward. After several visits, mice develop what’s called a cocaine-associated memory and a preference for the environment. Given a choice, the animal will spend time in the chamber. Such behavior is central to addiction and poses and obstacle to therapy for many individuals, Wood says.

The mice then go through a process called extinction learning where they no longer receive any drug in the chamber. After many trials, the animals replace drug-related memories with associations that have no drug reward. “In a sense they are re-writing their original cocaine-associated memory,” Wood says.

But mice given an HDAC3 inhibitor shortly after their first drug-free visit to the chamber re-write their memory much faster. After a single treatment with the HDAC3 inhibitors, the mice forgot the urge to indulge. This new memory is persistent. Days and weeks later, the scientists tried to initiate relapse-like behavior, but the mice continued to show no preference for the drug den, the researchers reported in the Feb. 12 Proceedings of the National Academy of Sciences.

The study suggests that boosting the activity of memory-related genes during a trial with no reward helps wipe out any drug associations with the cues, Wood says.

Though scientists have yet to figure out how manipulations of HDAC3 produce the long-lasting change, Wood says timing is the key. Giving mice HDAC3 inhibitors immediately after a drug-free visit to the chamber resulted in a long-lasting effect. But treatment delayed by several hours elicited no long-term change in drug-seeking behavior. The timely treatment came during a critical period in memory processing, called the consolidation phase, when certain genes must be coordinated and turned on to strengthen communication between neurons, Woods says.

Down the road

Scientists might be able to help addicts create new memories around cues associated with drugs. But there are many hurdles to overcome before HDAC inhibitors of any type make it to human trials. Tsai says drugs currently used in animal studies are too destructive for use in the human brain. And histone acetylation has many roles in the body; a viable drug would somehow have to target specific processes in the brain.

And not all HDAC inhibitors work in the brain to enhance memory. Scientists are still unraveling the effects that various HDAC enzymes have. André Fischer of the University of Göttingen in Germany says that inhibiting certain histone deacetylase enzymes, such as HDAC5, can make memory problems worse.

Some labs are now turning their attention to finding other targets, such as the genes that the HDACs regulate. So far, Wood’s group has identified at least one key gene, NR4a2, that must be turned on. This gene probably sets off another wave of genes, some of which may be involved directly in the brain’s reward system, he says.

Identifying molecules and targets that act “downstream” of memory genes could boost efforts to develop effective therapies. It might also help physicians tailor treatments to the memory deficits that occur in a long list of disorders, including Alzheimer’s, Parkinson’s, addiction, depression and neurodevelopmental disorders such as fragile X syndrome and Rett syndrome.

Tsai says additional targets will probably be discovered in the next few years. In the meantime, results from her lab mice suggest that even in cases where memories seem to be lost, there may be ways to regain the power of recall.

“It’s like a telephone line that gets broken so you lose communication,” she says. “But if that line can be repaired, there’s hope that long-term memory can be recovered.”

---

Making memories and getting them back

Billions of neurons create and retrieve memories by storing and then tapping into patterns of connections in the brain.

1. Input
A memory begins when information from the senses, such as a whiff of blueberry muffins or the sound of an ice cream truck’s jingle, arrives in the brain’s sensory cortex.

2. Processing
The frontal cortex can tap into the sensory information immediately for use as a short-term, or working, memory.

3. Encoding
The hippocampus and areas of the medial temporal lobes begin to encode this new information into a long-term memory by growing new neural connections and strengthening the brain’s existing circuitry.

4. Storing
Time and sleep help these new memories move to long-term storage regions throughout the brain. Facts, events, emotions and motor skills (such as riding a bike) take up permanent residence in brain regions involved in processing the original scent, sound or other sensory information.

5. Retrieving
When a memory is needed, or triggered by sensory information or emotions, the hippocampus and cortical brain regions help pull it out of long-term storage and relay it to the frontal cortex as working memory.

Sources: Lila Davachi/NYU, Carolee Winstein/USC

---

Amazing brain

Henry Molaison, age 60 in 1986, prepares for tests at the MIT Clinical Research Center. At this time he had been participating in brain studies for more than half his life. Credit: Jenni Ogden

Studies of how human memory works often focus on individuals whose memories have crumbled. Perhaps the most famous of all such people was Henry Molaison, known by researchers as H.M., an amnesiac who collaborated on hundreds of studies of memory for more than half a century until his death in 2008 at age 82.

H.M.’s memory largely disappeared in one day in 1953, when as a 27-year-old with epilepsy he underwent experimental brain surgery meant to relieve his debilitating seizures. During the operation, surgeons extracted two slivers of tissue, one from each side of the brain: the front half of the hippocampus along with nearby entorhinal, perirhinal, and parahippocampal cortices. Surgeons also removed most of H.M.’s amygdala, an almond-shaped structure that supports emotion. Together, these brain structures make up a region known as the medial temporal lobe.

The operation did curtail H.M.’s seizures. But he could no longer remember new words or experiences. He could, however, remember some of what he had learned before the operation. He recognized his parents and could recall childhood experiences or facts he had learned in school.

At the time of H.M.’s surgery, scientists were debating the nature of memory creation. Some thought it occurred as a single process in which information started off in the brain as a short-term trace and then, over time, consolidated and moved into a long-term memory bank. But studies of H.M. showed that by losing most of his hippocampus and the surrounding structures, he had no way to turn newly learned information into long-term memories. Researchers now realize that there are differences between short-term and long-term memory creation and that they involve separate processes. 

Studies of H.M. also showed that the brain processes different types of memories through different circuits. Notably, H.M. could learn new motor skills — such as drawing techniques — even without the parts of his brain that had been removed. And he could repeat his performance months or years later. This led scientists to make a distinction between declarative memory — recalling what you had for breakfast or dredging up historical facts — and non­declarative memory, which includes motor-skill learning, classical conditioning and perceptual learning.

When H.M. died, scientists took detailed MRI scans of his brain and preserved it for future study. Today, studies of his brain continue at the University of California, San Diego through a project at the Brain Observatory. —Susan Gaidos