In Rainbows End, by Vernor Vinge, a 2006 science-fiction novel set in the near future, modern medicine brings a talented Chinese-American poet, Robert Gu, back from end-stage Alzheimer’s disease. Before treatment, Gu is bedridden and can neither talk nor remember his children. After the therapy, his memory returns, although he develops a different set of talents. Flowers for Algernon, the 1959 short story by Daniel Keyes, entertains a related fantasy in which a futuristic treatment transforms Charlie, a mentally retarded man, into a genius.

Though fanciful, both these works echo research hinting that certain chemical treatments can reinvigorate the ability to learn and remember even in the face of brain damage or innate mental deficits. The studies—so far done in mice and sea slugs—indicate that the key to such cognitive improvements lies in epigenetics, the study of changes in DNA that do not affect the genetic code. Instead these chemical changes influence gene expression—that is, how actively the gene is used to make protein. Such alterations, it turns out, can have a profound impact on long-term memory. A drug compound, or even an environmental manipulation, that acts as a kind of volume knob for gene expression could someday help treat memory disorders and facilitate learning.

Gene expression is, after all, critical to memory formation. As a person learns and a memory takes shape, ebbs and flows in the activity of neurons incite the synthesis of new proteins, which help to cement or create connections between nerve cells. In this process, genes are first transcribed into RNA, which is then translated into protein.

Gene expression is strictly regulated. In chromosomes inside cells, DNA wraps around proteins called histones that serve as packaging material. In places where this packaging is looser, the underlying genes are accessible to the proteins that transcribe them, whereas tightly packaged DNA cannot be transcribed [see “The New Genetics of Mental Illness,” by Edmund S. Higgins]. Certain chemical changes to DNA or histones can loosen or tighten this chromosome structure and thereby enable or thwart the expression of memory genes.

Recently biologists have found that loosening part of a chromosome using drugs, or environments that provide more intellectual stimulation, can improve learning and memory in cognitively impaired animals. If such effects can be extended to humans, future therapies for memory disorders might work by altering DNA packaging in specific ways.

Rescuing Recollections
In the past few years several scientific teams have revealed that making a memory requires enzymes called histone acetyltransferases (HATs). HATs attach chemical units called acetyl groups to histones, thereby opening up DNA and facilitating gene expression. These enzymes counteract the activities of histone deacetylases, or HDACs, which remove acetyl groups from histones and condense DNA.

One 2004 study, for example, points to the importance of HATs in a mouse’s ability to remember objects and locations. Neuroscientist Mark Mayford of the Scripps Research Institute in La Jolla, Calif., and his colleagues engineered mice with an abnormal gene for a HAT called CREB binding protein. The inserted gene produced CREB binding protein devoid of all HAT activity, eliminating its capacity to stick acetyl groups onto histones near important memory genes. (They engineered the defect so that it appeared only in adulthood and did not affect development.)

These mice displayed distinct memory deficits—they had difficulty recognizing familiar objects and recalling the path to a hidden platform in a water maze—suggesting that normal memory requires the capacity to attach a sufficient number of acetyl groups to histones. And to prove that the memory impairment resulted from a lack of HAT activity, the researchers showed they could abolish the cognitive deficit by compensating for the molecular one. Gene-altered mice performed normally on the object memory test after receiving a chemical that inhibits HDACs, the enzymes that remove acetyl groups, and therefore boosts the number of acetyl groups bound to histones.

But could such a drug recover memories in other situations? Certain clinical phenomena show that memory loss is not always permanent. When patients emerge from anesthesia after receiving electric shock treatment for depression, their memory returns in stages. At first they remember nothing; then childhood memories emerge, and, within minutes, memory lane takes patients to the present, indicating that recollections can indeed reappear after they might seem to have vanished.

Animal experiments now indicate that retrieving lost memories might even be possible after severe neuronal damage—and that epigenetic mechanisms are central to this recovery. Neuroscientist Li-Huei Tsai of the Massachusetts Institute of Technology and the Howard Hughes Medical Institute and her colleagues genetically engineered a group of mice to develop an Alzheimer’s-like dementia when the scientists gave them the antibiotic doxycycline: the antibiotic flipped the genetically programmed dementia switch to the “on” position at the desired time.

While the mice were still cognitively healthy, the scientists taught them to associate an electric shock with a particular chamber so that the mice froze in fear whenever they were in the chamber. When the researchers administered doxycycline to some of the mice, however, those rodents suffered brain damage and memory loss and forgot their fear, frequently failing to freeze in the chamber. In contrast, mice that did not get the antibiotic froze as much as they ever had.

In an attempt to restore the memory in the brain-damaged mice, the investigators injected some of them daily for four weeks with a chemical that inhibits the acetyl-removing HDACs, a process that invigorates HATs and unwraps DNA from its protein packaging. In 2007 Tsai’s team reported that the epigenetic treatment restored the fear memory in the mice that received it and that no such memory reappeared in the mice that had been injected with an inert saline solution. Changing the packaging of DNA and reinvigorating gene expression somehow unmasked this simple fear memory—probably, the researchers speculate, by spawning new connections between healthy neurons rather than by repairing damaged ones.

The M.I.T. group also came up with a drug-free way to restore the obliterated fear memory: changing the rodents’ environment. Enriching the surroundings—giving the mice new toys to play with and running wheels that enabled them to exercise—similarly increased the number of acetyl groups on histones, apparently revving up the expression of memory genes just as the HDAC inhibitors did. Such a finding may help explain why scholars, who presumably live in an intellectually enriched world, are less susceptible to Alzheimer’s. A mentally stimulating job may be a form of environmental enrichment for humans, alleviating the effects of neurodegenerative processes in people by loosening chromosome structure.

Correcting Cognition
If medicine can revive memory after brain degeneration, could it also ameliorate inborn mental deficiencies such as those the fictional character Charlie displays in Flowers for Algernon? In a study published in 2004, biologist Angel Barco, then at the College of Physicians and Surgeons at Columbia University, and his colleagues tested this hypothesis in mice that had a genetic disorder resembling Rubinstein-Taybi syndrome, which in humans leads to mental retardation as well as skeletal abnormalities such as facial deformities and broad thumbs.

Underlying this syndrome is a mutation in the gene for CREB binding protein. A defect in one of a person’s two copies of the gene renders its protein nonfunctional; in such cases, cells generally produce only half the normal amount of protein. The resulting deficit in CREB binding protein activity seems to stymie the gene expression necessary for long-term memories to form, among its other effects. Similar to what Mayford’s group saw in their HAT-deficient adult mice, Barco’s team confirmed that mice born with a defective gene for CREB binding protein (and displaying classic Rubinstein-Taybi-like traits) have poor long-term memory. In their experiments, the mutant rodents had trouble recollecting having been shocked in a particular environment or after hearing a tone. They froze less often than normal mice did when they were exposed to the setting or sound that had been paired with the shock.

Mice with the CREB binding protein deficit displayed no such cognitive problems, however, if they received an HDAC inhibitor three hours before their training sessions with the shock, suggesting that the deficit can be reversed by loosening DNA’s protein packaging—even if this unraveling occurs belatedly, in adulthood. Such findings hint that the remodeling of this DNA wrapping might help improve cognition even in the face of ingrained developmental deficits, presumably by facilitating the expression of important memory genes. In Rubinstein-Taybi syndrome, such fixes may directly compensate for the low rates of acetylation that result from the lack of functional CREB binding protein.

Other molecules affecting DNA’s wrapping are also involved in memory and learning. The sea slug Aplysia, for instance, contains a pair of compounds called polyADP-ribose (PAR) and PAR polymerase (PARP), the enzyme that attaches PAR to DNA’s protein packaging. This enzyme facilitates transcription by stacking PAR molecules on histones as well as on various proteins involved in the reading of the DNA template.

To study the role of this enzyme in memory and learning, the late neurobiologist James H. Schwartz of Columbia University and his colleagues tempted Aplysia with a seaweed these creatures love and that the researchers had deviously encased in a cotton mesh, making the seaweed impossible for the slug to eat. The slugs learned that the seaweed was inedible and stopped trying to get it, eliciting the formation of a long-term memory, which required protein synthesis. But when the scientists treated some sea slugs with a compound that inhibits the PARP enzyme shortly before showing them the covered seaweed, the mollusks failed to remember that the food was inaccessible: the next day they still attempted to eat it. Thus, PARP seems to be an essential memory enzyme, suggesting that chemically enhancing its effects could be yet another avenue for bolstering memory in humans, who also bear a version of this protein.

Such work, along with the rodent studies, reveals the tremendous potential of epigenetic alterations to mold memories and, in the future, to reverse cognitive disorders as diverse as Alzheimer’s and mental retardation. A better understanding of the systems that modify the packaging of DNA may help us one day make science-fiction stories such as Rainbows End and Flowers for Algernon a reality.