Download Learning from a fly`s memory

news and views
as this will be essential in using models to
make global extrapolations of the effects of
increasing atmospheric CO2 on the carbon
cycle as a whole.
■
Eric A. Davidson and Adam I. Hirsch are at The
Woods Hole Research Center, PO Box 296, Woods
Hole, Massachusetts 02543, USA.
e-mail: [email protected]
1. Schimel, D. et al. in Climate Change 1995: The Science of
Climate Change (eds Houghton, J. T. et al.) 65–131 (Cambridge
Univ. Press, 1996).
2. Oren, R. et al. Nature 411, 469–472 (2001).
3. Schlesinger, W. H. & Lichter, J. Nature 411, 466–469 (2001).
4. Idso, S. & Kimball, B. Glob. Biogeochem. Cycles 7, 537–555
(1993).
5. Trumbore, S. E. Ecol. Appl. 10, 399–411 (2000).
6. Andrews, J. & Schlesinger, W. Glob. Biogeochem. Cycles 15,
149–162 (2001).
7. DeLucia, E. et al. Science 284, 1177–1179 (1999).
8. Nadelhoffer, K. J. et al. Nature 398, 145–147 (1999).
9. Aber, J. et al. BioScience 48, 921–934 (1998).
10. Caspersen, J. P. et al. Science 290, 1148–1151 (2000).
11. Davidson, E. et al. Nature 408, 789–790 (2000).
12. Schiffman, P. & Johnson, W. Can. J. For. Res. 19, 69–78 (1989).
Neurobiology
Learning from a fly’s memory
Randolf Menzel and Uli Müller
Two facets of learning are the formation and the retrieval of memories.
Genetic manipulation of the fruitfly’s brain allows them to be dissociated,
and may lead to a better understanding of memory.
S
NATURE | VOL 411 | 24 MAY 2001 | www.nature.com
olfactory learning)4. Several memory phases,
involving several sets of molecular machinery, are involved in acquiring and storing this
information5.
Dubnau et al.3 now use state-of-the-art
molecular genetic tools to tease apart the
storage and retrieval of olfactory memories
in Drosophila. Their target is a gene called
shibire. This gene encodes a motor protein,
dynamin, needed for the release of neurotransmitters from neurons at synapses
(presynaptic function). Fruitflies in which
a temperature-sensitive mutant shibire gene
is expressed move normally at 20 7C, but
become paralysed at 30 7C — the temperature at which the mutant gene is switched on.
A few minutes after being shifted from the
Odour
Acquisition
Foot
shock
Signal to
modulate
output
synapses?
Transient temperaturesensitive blockade of
presynaptic function
(output)
Retrieval
eminal work done over 100 years ago1,2
established the basics of memory and
learning. But, then as now, studies of
learning were hampered by the fact that it is
difficult (if not impossible) to distinguish
between the formation and the retrieval of
memories. Yet it will be important to make
this distinction if we are to unravel fully the
molecular and neural processes underlying
the sequential stages of memory. On page
476 of this issue3, Dubnau and colleagues
describe how they used fruitflies to tackle the
problem.
During learning, memories are formed
and stored in several stages, and can be
retrieved at any moment when circumstances demand it. But these stages are hard
to separate in laboratory studies, because
memories are accessible only through
retrieval. In a typical experiment, an animal
might learn to associate one stimulus (such
as an odour) with an electric shock. The animal would then be subjected to some sort
of interfering experimental procedure, and
tested some time later to see whether it can
remember the association. For example,
inhibitors of protein synthesis might be
injected into the brain, to find out whether
learning requires the synthesis of new proteins. If the interfering procedure were used
some time after the association was learnt,
and well before the ‘retention’ test, then
changes in retention would be interpreted
as revealing aspects of memory formation
and storage, rather than retrieval. But these
techniques have flaws; for example, proteinsynthesis inhibitors might alter the memories that are stored.
Fruitflies (Drosophila melanogaster) are
often used when studying complex research
problems, and provide an ideal test case here.
They too can learn to associate an odour
with a shock, and remember the unpleasant
association for a few days (this is known as
Output
(conditional odour avoidance)
© 2001 Macmillan Magazines Ltd
restrictive temperature (30 7C) to the permissive one, the flies recover (as the mutant
gene is switched off).
Usefully, the gene can also be targeted to
particular parts of the brain, and Dubnau
et al.3 use fruitflies in which the mutant gene
is expressed specifically in a region known
as the mushroom body. This area in the fly
brain has been implicated in short-term
olfactory learning6, as confirmed in genetargeting experiments7,8. The mushroom
body is a central, multisensory processing
structure, consisting of thousands of tightly
packed neurons. These neurons receive
information from nerve cells outside the
mushroom body, and transmit signals to
brain regions that are involved in controlling
movements (Fig. 1). As mentioned above,
the shibire mutant affects the transmission
of signals, rather than the receipt of information. So, if this mutant is targeted to the
mushroom body, it will affect only output
(that is, neurotransmission) from the mushroom body, and not input. This makes it possible to identify the stages of learning that
require neurotransmitter release from the
mushroom body.
Dubnau et al.3 first shifted their mutant
flies to the restrictive temperature, so impeding neurotransmission from the mushroom
body, during memory acquisition (when
they were learning to associate the odour
with the shock) or during memory storage
(the five minutes after training ended). Thirty
minutes after training, the authors tested the
flies for their recall of the association at the
permissive temperature, when neurotransmission was normal. The flies showed normal retention of the memory. But when they
were trained at the permissive temperature
Figure 1 Teasing apart the different stages of
learning in fruitflies. The mushroom body,
pictured here, is part of the fruitfly brain, and
is required for flies to learn to associate an
odour with an electric shock in laboratory
experiments. This process involves inputs to
the mushroom body (neurons that carry the
information of the odour and shock), and
outputs (signals to the parts of the brain that
coordinate motor processes, enabling the flies to
respond to the odour). Dubnau et al.3 have used
a temperature-sensitive mutant shibire gene to
block output from mushroom-body neurons at
different times — either while the flies are
learning the odour–shock association, or while
they are being tested for their recall of the
association. They remember normally when
output is blocked during learning, indicating
that the output is not needed for memory
storage. But the flies forget the association
when the output is blocked during recall. This
shows that the output is needed for memory
retrieval, not acquisition.
433
news and views
Randolf Menzel and Uli Müller are at the Institut
434
für Biologie–Neurobiologie, Freie Universität Berlin,
Königin-Luise-Strasse 28-30, D-14195 Berlin,
Germany.
e-mail: [email protected]
1. Ebbinghaus, H. Memory: A Contribution to Experimental
Psychology (Dover, New York, 1885; reprinted in 1964).
2. Müller, G. E. & Pilzecker, A. Z. Psychol. 1, 1–288
(1900).
3. Dubnau, J., Grady, L., Kitamoto, T. & Tully, T. Nature 411,
476–480 (2001).
4. Tully, T. & Quinn, W. G. J. Comp. Physiol. Psychol. 156, 263–277
(1985).
5. Dubnau, J. & Tully, T. Annu. Rev. Neurosci. 21, 407–444 (1998).
6. Heisenberg, M. Learning Memory 5, 1–10 (1998).
7. Connolly, J. B. et al. Science 275, 2104–2107 (1996).
8. Zars, T., Wolf, R., Davis, R. & Heisenberg, M. Learning Memory
7, 18–31 (2000).
9. Davis, R. L. Neuron 11, 1–14 (1993).
10. Martin, K. C. et al. Cell 91, 927–938 (1997).
11. Squire, L. R. Memory and Brain (Oxford Univ. Press, New
York, 1987).
Condensed-matter physics
Why vortices matter
Peter Gammel
In a magnetic field, a superconductor is threaded by swirling whirlpools
of electric current. Understanding these magnetic vortices is important
because they control the flow of current through the superconductor.
he defining property of a superconductor is the ability to carry an electrical
current without resistance when cooled
below a transition temperature, typically
a few degrees above absolute zero. But
below this temperature, most of the superconductor’s useful properties are governed
by its response to external magnetic fields.
Under certain conditions magnetic field
lines can permeate some superconductors,
leading to the creation of magnetic vortices.
Vortices have non-superconducting cores,
which carry the magnetic field lines, surrounded by swirling ‘supercurrents’. The
behaviour of these magnetic vortices determines the physical properties of superconductors, including the maximum electrical
current the superconductor can support.
The movement of vortices is particularly
damaging because it creates electrical resistance, thereby destroying the superconducting state.
In 1986, a new class of high-temperature
superconductor was discovered with unprecedented transition temperatures above 77
K. Magnetic vortices usually form stable
and regular patterns, such as a triangular
lattice, but in the high-temperature superconductors they appear in exotic and frequently less stable patterns. Two papers by
Avraham et al.1 and Bouquet et al.2 (starting
on page 448 of this issue) examine the
response of vortices to changing temperatures and fields, and reveal some of their
unusual behaviour in the high-temperature
superconductors.
When a current flows in a high-temperature superconductor, it exerts a force on the
magnetic vortices. If the vortices move in
response to this force they dissipate energy
and produce a non-zero resistance. But the
vortices can also be fixed or ‘pinned’ to
defects in the material, so they don’t move.
Practical applications of superconductors
require zero resistance and therefore pinning. A solid hexagonal lattice was thought
T
© 2001 Macmillan Magazines Ltd
to be the only stable vortex state until it was
discovered that the vortex lattice can melt
into a ‘liquid’ phase3. In the liquid phase it
is more difficult to prevent vortex motion
(and, hence, electrical resistance), which
restricts applications of superconductivity
to the vortex solid phases.
Like the melting of an ordinary solid,
the melting of vortices is described as a
first-order phase transition because it is
accompanied by abrupt changes in physical
Liquid 1
Liquid 2
Magnetic field
and then tested at the restrictive temperature, memory retention was impaired.
So, memory acquisition and retrieval can
be dissociated by interfering with neurotransmission in the mushroom body. The
acquisition and early processing of memories are known to depend on a signalling
cascade in the mushroom body that involves
cyclic AMP9. This cascade seems to be
driven by simultaneous input from olfactory
interneurons and unknown modulatory
neurons that are activated by the shock. This
neural circuit, including input to the mushroom body and the mushroom body itself,
might be a site for associative learning and
memory storage. The sites of output from
the mushroom body, however, are required
for memory retrieval3 (at least for the early
stages of memory tested here).
These results lead to several broad conclusions, as well as some questions. First,
continuous neural activity at chemical
synapses — often considered to be the basis
of early memory phases — is not required in
the mushroom body. So it seems unlikely
that early memory formation requires any
feedback from the mushroom body to input
neurons, or to the mushroom body itself
(that is, if electrical synapses are also not
involved, a possibility not testable with the
methods used here).
Second, the parts of mushroom-body
neurons that are involved in memory acquisition and retrieval are clearly separate. But
there must be some sort of signal that coordinates these processes. The signal probably
includes the transport of molecules along
mushroom-body neurons, and other, more
rapid processes10. Dynamin is involved in
protein transport as well as neurotransmission, so the memory-retrieval defects seen in
the mutant flies might also reflect defects in
coordination signals.
Third, a common feature of memory is its
progression through different phases. The
period tested by Dubnau et al.3 may represent a phase during which the formation of
memories at mushroom bodies can be clearly dissociated from retrieval. But it is not
known whether this separation also exists in
later memory phases.
Finally, most memory traces in large
brains are distributed over much greater distances than in fruitflies, and involve several
areas of the brain11. In the tiny fly brain, the
memory trace of a rather simple, odourguided behaviour is instead distributed
between different sites in the same neuron,
and these sites are specialized for the formation and retrieval of acquired information.
Is this a design principle of a small brain, or
of a simple type of memory, or both? The
answers to these questions will help to show
whether there is a general difference in how
small and large brains cope with learning
and remembering.
■
Critical
point
Glass
Lattice
Melting
transition
Temperature
Figure 1 Phase diagram for magnetic vortices
in a superconductor as a function of magnetic
field strength and temperature. In the lattice
and glass states, the superconductor will have
zero electrical resistance. Motion of the
vortices in the liquid states will generally lead
to finite resistance. Taming the vortices is
crucial for practical applications of
superconductivity. Avraham et al.1 show that
the lattice–glass transition includes a region
of ‘inverse melting’ just below the critical
point. The critical point is where the solid,
liquid and glass phases become
indistinguishable. Bouquet et al.2 have
investigated the behaviour of vortices above
the critical point, and identify a second
transition between two distinct vortex
liquids. These studies highlight the unusual
properties of vortices in the high-temperature
superconductors.
NATURE | VOL 411 | 24 MAY 2001 | www.nature.com