快猫短视频

AS HE PROBED his patients鈥� exposed brains during epilepsy surgery in the
1930s, Montreal neurologist Wilder Penfield discovered to his amazement that he
could bring back long-dormant memories. Just one jolt from a tiny electrode on
the outer surface of the brain鈥檚 temporal lobes, and distant memories played
back in vibrant detail. One patient even hummed an orchestral score to the
astounded surgical team, decades after he鈥檇 sat through its performance. It鈥檚 a
dramatic testimony to the longevity of memory.

Hidden away in each of us is a permanent record of our past. The smell of a
school canteen can conjure up long-forgotten images of childhood; vivid replays
of past events can flash by at times of intense fear. Even if it鈥檚 sometimes
hard to recall experiences, they are permanently inscribed somewhere amid the
billions of neurons in your brain.

How that happens is the sort of question that bothers neurobiologists. Most
argue that long-term memories are literally built into the brain as it creates
and strengthens connections between neighbouring neurons. These connections, or
synapses, are thought to join neurons up into intricate networks that can
recreate sequences of brain activity days, weeks or even years later.

But a handful of researchers has recently suggested a new and radical idea.
Perhaps long-lasting memories are inscribed not into synapse patterns but into
our brain鈥檚 DNA. Perhaps, say the researchers, we create gene-like codes in
which we permanently record the blueprints of our memories.

It鈥檚 not that hard to hang onto a memory for a short time鈥攚here you
last saw your keys, say, or where you parked your car. To strengthen the network
temporarily, all you apparently need is a more plentiful supply of messenger
chemicals poised at key synapses, ready to transmit the prepared signal. Trouble
is, this effect doesn鈥檛 last long. For a more permanent memory, your brain needs
something substantially more stable.

To lock a memory in place, most researchers say you need to shore up its
synaptic connections. One way is to build more synapses; another is to make
existing ones bigger and more stable. Either way requires cellular building
blocks in the form of proteins. Indeed, injecting chemicals that block protein
manufacture in the brain can disrupt the formation of long-term
memories鈥攂ut not short-term ones.

It sounds plausible enough. But building permanent, stable connections is no
simple process. Nearly all of the brain鈥檚 molecules, including those that form
the neural connections, are replaced every week or two. How long-lasting
memories can be stored with such distinctly impermanent media has confounded
theorists for years.

The problem, says neurobiologist Sandra Pe帽a de Ortiz of the
University of Puerto Rico in San Juan, is that molecular turnover would
eventually degrade these structural proteins. They鈥檇 become hopelessly fuzzy,
like photocopies of photocopies. No less a luminary than Francis
Crick鈥攃o-discoverer of the chemical structure of DNA鈥攆irst pointed
this out, back in 1984.

Electron microscopes have also shown how far from stable neurons are. Their
outstretched branches move from day to day. 鈥淚maging studies show strikingly
dynamic synapses,鈥� says neurobiologist Seth Grant from Edinburgh University.

Memories of memories

True, you could keep rebuilding the same structures, but how would you know
what to rebuild and where? For memories to be stable through time, Pe帽a
believes, what鈥檚 needed is some sort of archived blueprint for each acquired
experience. That way, the brain would have a reliable set of instructions for
how to rebuild each synapse.

Nature鈥檚 favourite blueprint, of course, is DNA. And as luck would have it,
DNA doesn鈥檛 undergo the turnover that other molecules go through. It is quite
stable on its own, and also enjoys the protection of a specialised repair
machinery if anything happens to go awry.

鈥淲e believe that permanent memories are stored in altered genes,鈥� says
Pe帽a. In the same way that DNA provides a blueprint for the proteins that
make up living cells, Pe帽a and her colleague Yuri Arshavsky from the
University of California鈥檚 Institute for Nonlinear Science in La Jolla believe
that new 鈥渕emory molecules鈥� are born when altered gene sequences are translated
into proteins.

They鈥檙e not suggesting that we make a new protein in the split second it
takes to recall a memory. 鈥淲hat I call molecules of memory,鈥� explains
Pe帽a, 鈥渁re used to stabilise networks of memory neurons.鈥� When a specific
network is stabilised, so is a particular memory.

But these are no run-of-the-mill structural proteins. Memory molecules, she
believes, are novel proteins created from a unique blueprint that she thinks
could be formed by neurons rearranging their DNA in response to each new
experience. Each protein鈥檚 novel structure would allow it to snap into a unique
position at the synapse, she surmises, helping make memory traces stable without
screwing up other synaptic structures. 鈥淐hanges in synaptic connections wouldn鈥檛
remain intact for long,鈥� says Pe帽a, 鈥渂ut gene rearrangements could be
kept throughout the neuron鈥檚 lifetime.鈥�

Arshavsky has a slightly more ambitious proposal. He wonders whether the
memory molecules might store information themselves鈥攍iterally molecular
representations of memories. 鈥淚n contrast with the widespread concept that brain
functions are realised mainly on the network level,鈥� he says, 鈥淚 believe that
individual neurons play an important role.鈥�

Either way, this is a radical suggestion. We鈥檙e used to thinking of our
genetic code as something fixed at the very beginning of our lives, not
something that gets rewritten on a daily basis. After all, it gives us each our
individuality. What would happen if every cell in the brain was allowed to
tamper with that code? Colleagues are intrigued. 鈥淢aybe there is something
outrageous taking place,鈥� says Harvard geneticist Gary Rathbun. 鈥淥ur
conventional views can get in our way.鈥� Still, he is quick to point out, 鈥渢he
story is not complete鈥�.

鈥淭here are certainly considerable limitations to the existing [synaptic]
model,鈥� Grant admits. But he feels the DNA hypothesis is a long shot (see
Brains within brains). Proteins are important, he says, but you don鈥檛 need
novel genes to make them.

So what makes Pe帽a and Arshavsky even consider such an off-the-wall
idea as rewriting our own genetic blueprint? There鈥檚 only circumstantial
evidence so far, but enough to convince them to keep looking. DNA certainly has
the capacity to act as a stable blueprint for memory molecules, they argue. Once
neurons are fully grown they don鈥檛 divide again, so there鈥檚 no danger of
disrupting the DNA during cell division, and no problem if you disrupt genes
that are needed to make new cells. In any case, there鈥檚 plenty of DNA to spare
for archiving our memories. As much as 97 per cent of our DNA has no obvious
function.

But how could you go about storing a new memory? According to Pe帽a and
Arshavsky it would require some kind of genetic rearrangement鈥攁
reshuffling of the A鈥檚, T鈥檚, G鈥檚 and C鈥檚 that make up our genomic alphabet. This
might not be so outlandish a notion as it seems. Rearranging DNA may not happen
in most of our cells, but it is exactly how our immune system remembers
infectious pathogens, point out geneticists Axel Dietrich and Willem Been from
the University of Amsterdam.

鈥淲e know of three 鈥榤emory鈥� systems in nature,鈥� says Dietrich. There is an
evolutionary memory of how to build an organism, a cognitive memory of events we
experience, and an immune memory of past infections. 鈥淭wo are based on DNA,鈥�
says Dietrich. 鈥淲e can expect nature to be efficient enough to use the same
tools for the third as well.鈥�

Immune memories come in the form of recognition proteins called antigen
receptors. These receptors are uniquely shaped to lock onto specific pathogens,
and they sit on the surface of the B lymphocyte cells which patrol the
bloodstream.

A specialised toolkit of enzymes gets to work on three groups of lymphocyte
genes, called V(D)J. The enzymes snip the groups apart, rearrange them, and glue
them back together to create a template for a protein of exactly the right shape
to match a particular invader. By shuffling together genes from three genetic
鈥渄ecks鈥�, our immune system has a tremendously varied range of protein 鈥渉ands鈥� at
its disposal. The same mechanism could create an almost infinite variety of
memory proteins, argue the researchers.

Intriguingly, there are two gene complexes in the brain, called cadherins and
protocadherins, which have a similarly variable structure to the V(D)J complex.
Both complexes play a role in adhesion between cells鈥攁nd are known to be
important in the early stages of forming synaptic connections between neurons.
鈥淸They] have caused a flurry of excitement,鈥� notes Rathbun. 鈥淭hey seem to be
screaming out at the world, 鈥業 rearrange!'鈥� No one has yet shown that they
undergo genetic recombination, he explains. Still, the genes do produce slightly
different RNA products鈥攖he intermediate step between genes and
proteins鈥攊n different regions of the brain. Which is exactly what you鈥檇
expect if you wanted to build novel proteins.

There might be other clues too. If this is how memories are stored, the brain
should come equipped with a similar enzymatic toolkit as the immune system.
Indeed, animals as diverse as fish, salamanders and mice have the same enzymes
in their brains as in immune cells to snip apart V(D)J genes, as well as another
group of enzymes known to glue rearranged gene complexes back together.
Genetically engineered mice lacking these enzymes die before birth, and suffer
massive loss of neurons as well as immune cells. It鈥檚 perhaps a hint that
neurons and immune cells have something in common.

Enzyme glue

Psychiatrist Georgy Bakalkin of the Karolinska Institute in Stockholm is
certainly intrigued. If DNA recombination did underlie long-term memory, he
points out, you鈥檇 expect to find much higher concentrations of these enzymes in
the brains of long-lived and socially complex creatures like us, who stash away
so much information in our long-term memory. At least one enzyme that could glue
DNA back together appears in the human brain at concentrations up to a hundred
times higher than in rodent brains. 鈥淣o other proteins,鈥� he notes, 鈥渄emonstrate
so profound a difference.鈥�

But not everyone is convinced this is evidence that new genes are created in
the brain. There could be other explanations as to why this enzyme toolkit is
present. Even if neuronal genes are rearranged, points out neurogeneticist
Jerold Chun of drugs company Merck and the University of California, San Diego,
this might mean moving genes around rather than creating new ones.
鈥淩earrangement could just involve putting a promoter gene in front of another
existing gene,鈥� he says鈥攔amping up production of whatever protein that
particular gene codes for.

Undeterred, Pe帽a and Arshavsky argue that there鈥檚 one final clue that
novel memory molecules might indeed exist鈥攖he brain鈥檚 mysterious 鈥渋mmune
privilege鈥� status. A physical barrier prevents immune cells from entering the
brain鈥攖hey鈥檙e too big to get through. This barrier deprives the brain of
protection against infection. Why would the brain take such a risk if there
wasn鈥檛 an extremely good reason? Other tissues with immune
privilege鈥攕perm, for example, and the fetus and placenta鈥攕hare one
thing in common. They all express novel proteins unfamiliar to their host鈥檚
immune cells. If the body鈥檚 immune cells came into contact with such tissues,
they would probably mistake the novel molecules for invading pathogens鈥攁nd
attack.

Pe帽a and Arshavsky believe that something very similar must be going
on in the brain. 鈥淲e think that new proteins are created in neurons with each
new experience,鈥� explains Pe帽a. If the immune system could get at these
new memory molecules, they would trigger autoimmune reactions which would
destroy the molecules鈥攁nd the brain鈥檚 capacity for lasting memories.

If the new theory is right, perhaps one day we will transplant or even clone
memories. We know that bone marrow donors can pass on their allergies and some
acquired immunities鈥攊mmune memories鈥攊n the form of immune B memory
cells. 鈥淎s we understand memory better, this may seem a less than crazy idea,鈥�
laughs Pe帽a. Of course this is all armchair speculation at the moment.
鈥淲e鈥檙e at a very early stage of this research,鈥� says Pe帽a. 鈥淣ot many labs
are working on this, but they will鈥攕oon.鈥�

Geneticists are well used to the idea that our individual complements of DNA
are crucial to who we become. Few, however, have seriously considered whether
our identity leaves a permanent mark on our genome in turn. That鈥檚 precisely
what the new DNA theory of memory proposes.

If it鈥檚 true, we will have to change the way we think about our minds and
bodies, memories and diseases. Perhaps you remember a miserable week with
chickenpox when you were little. Your immune system certainly would: its job is
to make sure it remembers how to deal with the virus when you meet it again. Two
kinds of remembering: are they written in the same ink?

There are still some big holes in our knowledge of how permanent stable
memories form, says Seth Grant of Edinburgh University. He thinks there鈥檚 more
to memory than simply passing signals across synapses to activate a
predetermined network of neurons鈥攁nd special proteins are very likely to
be part of it.

Dogma holds that synaptic strength is what stores memory, not the individual
neurons themselves. But in neurons that are known to be involved in long-term
memory circuits, Grant and his team have identified complex arrangements of
dozens of cellular proteins, dubbed 鈥淗ebbosomes鈥�. And many of the proteins are
definitely not part of the synapses.

Grant suspects that Hebbosomes might be miniature information-processing
centres. He thinks they could make decisions about memory storage and retrieval
right there in the cell. 鈥淭he complex has multiple inputs, internal pathways,
and outputs,鈥� Grant says鈥攚hich might just allow for complex processing
within the neuron. Hebbosomes, it seems could be the real brains of the
outfit.

Grant says that we know new proteins are built as new memories become fixed.
But to create the huge variety of memories we hold, there鈥檚 no need for each
protein to have a unique new structure, as other researchers are proposing (see
main text). Grant believes new proteins are added to Hebbosome complexes and
that the variety of memory comes from the many different structures you could
build from a standard set of components.

If he鈥檚 right, individual neurons play a richer role in the orchestration of
memory than is widely believed. They might well store information themselves.
鈥淟earning may involve integration and orchestration of many cellular changes,
not just synaptic ones,鈥� he says. 鈥淪ynapses may not be the locus of learning.鈥�