CELL AND MOLECULAR MECHANISMS OF LEARNING 
Eric R. Kandel, M.D., Senior Investigator 
Molecular Mechanisms of Memory in Aplysia 
Dynamics of cAMP and protein kinase A (PKA ) 
subunits in Aplysia sensory neurons. Binyamin 
Hochner, Bong-Kiun Kaang, and Eric Kandel, in col- 
laboration with Brian Bacskai, Martyn Mahaut-Smith, 
Stephen Adams, and Roger Tsien (HHMI, University 
of California, San Diego), injected cyclic AMP- 
dependent protein kinase, labeled with fluorescein 
and rhodamine on the catalytic and regulatory sub- 
units, respectively, into Aplysia sensory neurons ei- 
ther in culture or in intact cell clusters. Confocal 
fluorescence microscopy revealed that bath appli- 
cation of serotonin (5-HT) produced striking gra- 
dients of cAMP — high in the processes, low in the 
central bodies of the neurons. Perinuclear increases 
in cAMP slowly caused translocation of the freed 
catalytic subunits into the nucleus to an extent lin- 
early related to the percentage dissociation of the 
kinase. The difl'usional processes may act as a filter 
to ensure that only repeated or particularly strong 
synaptic stimulation of the cAMP cascade moves suf- 
ficient catalytic subunits into the nucleus to phos- 
phorylate transcription factors and activate gene 
expression. 
Transcriptional activation. One hallmark of 
most types of long-term memory is the requirement 
for new protein synthesis. In long-term sensitization 
of the gill-withdrawal reflex in Aplysia, this require- 
ment can be studied on the cellular level in the 
monosynaptic connections between the sensory and 
motor neurons of this reflex. Here, long-term but 
not short-term facilitation requires new protein syn- 
thesis and is reflected in an altered level of expres- 
sion of specific proteins regulated through the 
cAMP second messenger pathway. 
Based on gene transfer into individual sensory neu- 
rons of Aplysia in the intact nervous system, Kaang, 
Kandel, and Seth Grant found that 5-HT induces 
transcriptional activation of a reporter gene driven 
by the cAMP response element (CRE) , that this in- 
duction requires CRE-binding proteins (CREBs), 
and that it is blocked by competing CRE oligonucle- 
otides. The induction by 5-HT does not occur fol- 
lowing a single pulse of 5-HT but becomes progres- 
sively more effective following two or more pulses. 
Moreover, expression of GAL4-CREB fusion genes 
shows that 5-HT induction requires phosphory- 
lation of CREB on Ser''^ by PKA but not by the cal- 
cium/calmodulin-dependent kinase. These data 
provide supporting evidence for CREB-modulated 
transcriptional activation with long-term facilitation. 
Molecular Mechanisms of Long-Term 
Potentiation 
Genetic analysis reveals that the fyn tyrosine ki- 
nase gene is necessary for LTP and learning in 
mice. Tyrosine kinase inhibitors block long-term 
potentiation (LTP) in the CAl region of the hippo- 
campus. To identify specific tyrosine kinases in- 
volved in LTP, Grant, Thomas O'Dell, Kevin Karl, 
Paul Stein, Phillipe Soriano (HHMI, Baylor College 
of Medicine), and Kandel screened mice with muta- 
tions engineered in either of four cytoplasmic tyro- 
sine kinase genes: fyn, src, yes, and abl. Although 
these four kinases are coexpressed in the hippocam- 
pus, only the fyn mutant mice failed to show normal 
LTP. With low-intensity tetanic stimulation, slices 
from fyn mutant mice showed essentially no LTP. 
However, a modest amount of LTP could be induced 
using higher-intensity (75% of the maximum excit- 
atory postsynaptic potential [EPSP]) tetanic stimula- 
tion (133.2 ± 9 3% of control), but this potentia- 
tion was smaller than that observed in slices from 
control animals (168.5 ± 11.6% of control). By 
contrast, synaptic transmission appeared normal in 
slices from fyn mutant mice, and paired-pulse facili- 
tation was not different from that observed in con- 
trol animals. 
The impairment of LTP appears to correlate with 
impaired spatial learning in the Morris water maze, 
suggesting a functional link between LTP and spatial 
memory. In addition to its importance in LTP, the 
fyn gene is also necessary for the normal develop- 
ment of the pyramidal cell layer of the hippocam- 
pus, since the layer in the CA3 region shows struc- 
tural abnormalities. Together these data suggest that 
the fyn tyrosine kinase is important for the induc- 
tion of LTP and implicate a new biochemical path- 
way contributing to synaptic plasticity. 
Nitric oxide produces long-term enhancement 
of synaptic transmission in the CAl region of the 
hippocampus by an activity-dependent mecha- 
nism. There is evidence that the membrane- 
permeant molecule nitric oxide may act as a retro- 
grade message during LTP in the hippocampus. A 
difficulty with the retrograde message idea, how- 
ever, has been that lateral spread of a diffusible mes- 
sage could lead to potentiation of transmission at 
inactive presynaptic terminals, which would violate 
the observed pathway specificity of LTP. A possible 
solution to this problem would be for the effects of 
the message to be restricted to recently active pre- 
synaptic fibers. Scott Small, Kandel, and Robert 
416 
