I can already see the grim look many Americans will have as they chew on their Christmas roast tomorrow. They'll be thinking about yesterday's report that a cow in Washington state tested positive for mad cow disease. There's some comfort in knowing that so far it's just a single cow, and that American cattle are regularly screened for bovine spongiform encephalitis. The grimmest look this Christmas may be on the faces of McDonald's shareholders and cattle ranchers. A single Canadian cow that test positive wreaked havoc on the entire beef industry up north. But this Christmas also brings a fascinating discovery about the bizarre agents that cause disorders such as mad cow disease: they may actually record our memories.

The work comes from the lab of Eric Kandel, the Columbia University neuroscientist who won the 2000 Nobel Prize for medicine. Kandel got the prize for figuring out some of the molecular underpinnings of memories. Each neuron has one set of branches that send outgoing signals and another set that receives incoming ones. These signals can only jump from one neuron to the next if an outgoing branch nuzzles up to an incoming one, creating a junction called a synapse. Kandel studied how the neurons in a sea slug change as memories are laid down. (These are obviously not memories of the Proustian sort--just simple associations, such as the memory of a shock coming after the flash of a light.) He showed that new synapses are created and other ones grow stronger as memories form. Kandel also identified a number of the molecules that seem to be responsible for strengthening these connection. (His Nobel prize lecture makes for good reading.)

Kandel did not rest on his laurels, but immediately tackled some of the big questions about memory that he and other neuroscientists had yet to figure out. A neuron may have tens of thousands of synapses, but only a few of them may change as a memory forms. Yet the instructions to make proteins that cause this change come from a neuron's single bundle of DNA. If the nucleus gets a signal to form new synapse-strengthening proteins, how do the proteins go only to the right synapses. And, even more importantly, how do those synapses stay strong for decades, when proteins themselves live only a short period of time?

Kandel and his coworkers reasoned that a memory-forming synapse must get some sort of "synaptic mark" that tagged it for synapse-strengthening proteins. They then looked for molecules that might be responsible for the mark. As they report in the December 26 issue of Cell, they have discovered what may well be the synaptic mark in a compound called cytoplasmic polyadenylation element binding protein (CPEB for short). CPEB can be found in cells throughout the body, but they found a special form of it in the neurons of sea slugs, and then later found it in fruit flies and mammals. They found that CPEB is synthesized during the earliest stages of memory formation, and probably drives the production of molecules that physically lay down new synapses and tells them where to grow. Evidence suggests that the protein can do this by "waking up" dormant RNA molecules in the synapse. (RNA is the messenger molecule that carries copies of genetic information to the protein-building factories of the gene.)

To understand how CPEB could do all this, the researchers looked closely at its structure. That's when they had a shock: CPEB has much the same structure as the agent that causes mad cow disease.

Mad cow disease is infectious, but it's caused not by a virus or a bacterium. Instead, it's caused by a rogue protein called a prion. The normal version of the protein (called PrP) may do a number of jobs in the body, and seems particularly important in the brain. But sometimes a PrP gets a funny kink in it and folds into a new shape. This new prion then bumps into a normal PrP and forces the normal copy to take on its own strange shape. The prions clump together and force others to join them in Borg-like fashion. Mad cow disease can spread if cows eat feed that has been supplemented with other cows--in particular, if the feed contains prions. Humans eating those sick cows can take in the prions as well and get a fatal brain disease of their own called Creutzfeld-Jacob disease.

Prions were the object of scorn and skepticism for years, in part because they were so different as pathogens from viruses or bacteria. Prions had no genetic material, and yet they spread like genetically-based pathogens. Eventually the evidence became too much to ignore (and also won Stanely Prusiner of the University of California at San Francisco a Nobel of his own). But prions were revolutionary in another way that most people don't know about: they enjoy a unique kind of evolution.

In the early 1990s scientists realized that yeast contain prions. These aren't mutant PrPs, however, but two completely different proteins that just so happen to have the ability to change shape and force other proteins to clump with them. Unlike mad cow prions, yeast prions don't necessarily harm their hosts--in fact, they actually make yeasts thrive better than without them. And since yeasts are single-celled, they can pass down their prions to their offspring. (A prion in your brain won't get down to your sperm or eggs, so you can't infect your kids.)

In other words, a yeast can inherit prions from its parents, despite the fact that it has inherited no prion gene. This non-DNA based inheritance is a lot more like what Lamarck was talking about than Darwin.

Kandel and his Columbia team joined forces with an expert on prions in yeast, Susan Lindquist of MIT. Together, they inserted copies of the gene for the synaptic mark CPEB into yeast so that they could experiment on them and see whether they were in fact prions. They found that indeed, CPEB can exist in two different states. In one, the protein roams the cell alone. In the other, it forces other CPEB to change shape and form clumps with it. They also found that only when it takes on its prion form can CPEB bind to RNA.

The researchers propose a simple but elegant hypothesis for how prions can build memories. They suggest that certain signals entering a synapse can trigger CPEB to become a prion. As a prion, it can wake up sleeping RNA in the synapse, creating proteins for strengthening it. It also keeps grabbing other CPEB molecules and turning them into prions as well, so that even after the original prion has fallen apart, others continue to do the job. The neverending power of prions, in other words, is what keeps our memories alive.

In a commentary in the same issue of Cell, Robert Darnell of Rockefeller University says that if this work holds up to scrutiny (if it's replicated in neurons rather than yeast, for one thing), it will prove "nothing less than extraordinary." It would be extraordinary enough if memory proved to be based on prions, but the finding--along with the earlier work on yeast--raises the possibility that prions actually do a lot of important things in our bodies, and that we cannot understand them unless we are willing to let go of our vision of life as nothing but genes creating proteins. That may not make this Christmas's roast any tastier, but it should help revive the low reputation of prions.