All of us forget some things and remember others every day. “Being able to
cope in the real world means you have to filter things out,” explains Fernando
Nottebohm, who led the research at Rockefeller University, New York. Yet while
researchers have tried hard to understand how information is committed to
memory, they have devoted little effort, says Nottebohm, to finding out how long
memories last.

Nottebohm, Sek Jin Chew and David Vicario studied the responses of zebra
finches to sounds, including birdsong and human voices. Zebra finches use song
to distinguish members of their own species from others, and to distinguish
between relatives and strangers. The team recorded the firing patterns of
neurons in the birds’ brains as they listened. They varied the number of
repetitions of each song, and the intervals between repetitions.

When a bird hears a new song, neurons in one specialised region of its brain
fire intensely for a short period. If the bird hears the same song again soon
afterwards, the firing of the neurons is decreased— suggesting that the
bird remembers the song and has become used to it. But when the bird is tested
with the song again after a long period, it reacts as if the song were new: its
neurons fire as intensely as they did in the first experiment.

The amount of time it takes a bird to forget a song depends partly on how
often it has heard it repeated. For example, a song heard 50 times is forgotten
sooner than one heard 1000 times. The birds also retain information for longer
when it matters to them. For a given number of repeats, unimportant information,
such as a human voice, is more rapidly forgotten than the songs of zebra
finches, which are remembered for longer than other species’ songs.

Surprisingly, the duration of each memory always fell into one of several
discrete categories (Science, vol 274, p 1909). For example, the
memories with the shortest shelf life lasted between 6.5 and 7 hours,
medium-life memories lasted between 17.5 and 18.5 hours, and long-life memories
lasted for 47.5 hours.

Just how the memories are recorded is not clear. However, other researchers
have found that the reinforcement of memory in mammals can be blocked by
preventing the synthesis of all the proteins in nerve cells during critical
periods. The team found that this was true of the zebra finches: if they blocked
protein synthesis between 6.5 and 7 hours after training, a memory that would
otherwise have lasted for 18 hours was lost there and then. By contrast, if the
team blocked protein synthesis before or after this time, the memory was
retained as normal.

Whether human memories also have discrete shelf lives is not clear, but
Nottebohm is cautiously optimistic. “If the process we are dealing with is a
basic one, we hope that it will turn up everywhere,” he says. Tim Tully, who
studies the formation of memory in fruit flies at Cold Spring Harbor Laboratory
in New York, agrees.

BY training cells to grow in precise patterns, researchers have come one step
closer to growing complex tissues such as nerves or blood vessels in the
laboratory. Their ideas could also pave the way for living electronic
circuits.

Most cells will only grow attached to a surface, so George Whitesides and
colleagues at Harvard University generated a pattern of sticky and unsticky
surfaces for the cells they were studying. They hoped that the cells would only
grow on the sticky surfaces. Using a tiny rubber stamp cast from a silicon chip,
the researchers made a repeating pattern of plateaus and valleys in clear
plastic. A flat stamp spread the sticky material on the plateaus, and the
plastic was dipped in a repellent material to coat the valleys.

Cells encouraged to grow on this terrain unswervingly chose the plateaus,
even when the sticky areas were only a single cell wide (Proceedings of the
National Academy of Sciences, vol 93, p 10775). If the sticky and repellent
materials were swapped the cells chose the valleys instead.

The potential applications are numerous. “If you help cells to align like
this you can make something that begins to look like a nerve or a capillary,”
says Whitesides. And nerve cells trained to grow in complex patterns could be
linked to conventional electronics to make hybrid circuits. “Being able to
manipulate where cells go is very important,” says Jeffrey Hubbell of the
California Institute of Technology in Pasadena.

He believes this system could help overcome the body’s tendency to form scars
around implanted medical devices. Scarring could be reduced if the implant’s
surface texture encouraged blood vessels to grow towards it. “This system has
the potential to change the way we make medical devices,” he says.

Researchers led by Carl Johnson of Vanderbilt University in Nashville studied
Synechococcus, which belongs to a group of primitive photosynthetic
organisms called cyanobacteria. Because Synechococcus cells depend on
light for energy, says Johnson, “when the lights go out they need to shut things
down”. The researchers worked with a particularly fast-reproducing strain which
passes through several generations every day, before resting at night.

Many biologists have assumed that circadian clocks cannot pass their timing
information from one generation to another. This means that
Synechococcus’s cycle of activity should break down without the regular
cues of dusk and dawn. But when Johnson’s team put their colonies under constant
light, the daily cycle of cell division and rest continued (Proceedings of
the National Academy of Sciences, vol 93, p 10183). Clearly,
Synechococcus possesses a circadian clock that can be passed down the
generations.

“Conceptually, this bothers some people,” says Susan Golden of Texas A&M
University, who has made similar observations of the same strain, which have not
yet been published, working with researchers at Nagoya University in Japan. But
in organisms that reproduce by cell division, she says, there is no reason why a
sense of time cannot be inherited. Circadian clocks depend on proteins whose
concentrations inside a cell cycle up and down. Provided these proteins are
divided equally between daughter cells after division, a clock should keep
perfect time.

EVEN the most accident-prone motorist can glance at road signs without
veering off the highway. But experiments on monkeys have only now begun to
reveal how our brains avoid such catastrophes. Richard Andersen and his
colleagues at the California Institute of Technology in Pasadena have identified
a group of brain cells that enable us to keep heading in the same direction even
when our eyes are moving.

We can tell which way we are going because, as we travel forwards, the world
ahead of us appears to expand. To take a well-known image, as the Starship
Enterprise zooms through space, its crew sees a pattern of stars flying outwards
from a central point which defines the direction of movement.

But if a crew member’s eyes follow the course of one particular star moving
to the left, the entire field of view will appear to move to the right. When
this apparent movement is superimposed on the expanding star pattern, the point
from which the pattern is expanding will appear to move to the left (see
Figure). Without a mechanism to compensate for eye movements, the crew of the
Enterprise would instinctively change the direction of their craft whenever they
glanced sideways.

Until now, this mechanism has eluded neuroscientists. But in the latest issue
of Science (vol 273, p 1544), Andersen’s team provides the first clues
to how it works. The researchers placed rhesus monkeys in front of an expanding
pattern of dots like the pattern of stars seen from the bridge of the
Enterprise. They then trained each monkey to move its eyes so that it tracked a
single, large dot which was part of the expanding pattern and moving to the
left.

Andersen and his team recorded the electrical activity of nerve cells in part
of the brain’s visual cortex called the dorsomedial superior temporal area. This
contains cells that respond to moving patterns, including some that fire most
strongly when presented with an expanding scene that is indicative of forward
movement.

The researchers first identified brain cells that fired strongly in response
to the pattern of dots expanding from a central point in each monkey’s field of
view. When each animal’s eyes moved left to follow the large dot, some of these
neurons dramatically decreased their rate of firing—they evidently did not
register that the expanding pattern had been distorted as a result of eye
movement.

Other neurons, however, continued firing as before, having compensated for
the fact that eye movement had shifted the central point of the expanding
pattern to the left. Andersen argues that the brain uses these “shifting”
neurons to determine the true direction of movement.

Next, the researchers tried moving the screen—still with the same
expanding display—so that the large dot remained straight ahead of the
monkey, which was able to follow it without moving its eyes. In this case, both
sets of neurons changed their pattern of firing in the same way, responding as
if the direction of movement had shifted. This suggests that the key to the
compensation mechanism is a signal direct from our moving eyes which feeds back
to the shifting neurons, says Andersen.

Now that Andersen’s team has identified the neurons that keep us on a steady
course, the next problem is to work out how the brain receives the information
needed to compensate for eye movement. “There is evidence that there is
compensation in the right direction, and in some cases of sufficient magnitude
to fully compensate for eye movement,” says Bill Warren, a neuroscientist at
Brown University in Providence, Rhode Island. “But the detailed mechanism is
still unclear.”

More than 90 per cent of the DNA in the genome of most organisms does not code for any protein (see “Message in a genome?”, New ĐÓ°ÉÔ­´´, 12 August 1995, p 30). The highest concentration of genetic junk is at centromeres, which consist largely of a pattern of five or six letters of the genetic code repeated thousands of times over. During cell division, centromeres serve as the attachment points for the cellular cables that pull chromosomes apart after they have been duplicated.

Abby Dernburg, a PhD student at the University of California, San Francisco, and her supervisor John Sedat were investigating how isolated stretches of junk DNA positioned away from the centromeres could switch neighbouring genes off, preventing them from producing any protein. One theory was that these islands of junk DNA turn genes off by making them bind to the centromeres, where the normal cellular machinery of protein production does not seem to work.

To test this idea, Dernburg and Sedat studied a mutant fruit fly which has an island of junk DNA near a gene called brown. Fruit flies usually have red eyes, but if the brown gene is inactive, as in these mutants, the flies cannot make the red pigment and the eyes appear brown.

The geneticists “painted” different parts of one of the flies’ chromosomes different colours, using short pieces of DNA that bind to known genetic sequences and carry fluorescent molecular tags. They tagged the brown gene pink and the centromere of the same chromosome orange. In normal flies, the pink and orange regions remained far apart. But in the mutants, they were always remarkably close (Cell, vol 85, p 745).

Dernburg and Sadat argue that the “stickiness” that held the brown gene at the centromere in the mutant flies normally binds the two members of a pair of chromosomes together at their centromeres in the early stages of cell division, when the chromosomes are duplicated. “Ever since junk DNA was discovered in the 1920s people have wondered why cells bother to carry so much around,” says Dernburg. “We’ve shown that it is not just a parasite.”

Duplicating a chromosome, and attaching the cellular cables that subsequently drag one chromosome copy to each end of the dividing cell takes time. If the junk DNA at the centromere did not act like glue, say the researchers, the chromosome copies could drift away randomly, resulting in daughter cells containing too few chromosomes, or too many. “This would be a genetic disaster,” says Dernburg.

The finding that the brown gene in the mutant flies always stays close to the centromere provides “strong support for the idea that [junk DNA] is sticky”, agrees Gary Karpen of the Salk Institute for Biological Studies in La Jolla, California. He and Dernburg believe that the junk DNA does not actually stick to itself. The real glue, they say, consists of as yet unidentified proteins that bind to DNA sequences and then to each other.