Computational Neurobiology of the Hippocampus 
EPSP-spike potentiation in two ways. First, we 
have looked for experimental conditions that 
would favor EPSP-spike potentiation over synap- 
tic potentiation. Second, we are undertaking di- 
rect measurements of intracellular free calcium 
in hippocampal neurons. The properties of the 
hippocampal network and its capacity for infor- 
mation storage depend on the properties of the 
synapses and dendrites. It is therefore important 
to determine these properties before models of 
the hippocampus are attempted. 
LTP can be induced at some synapses by simul- 
taneous postsynaptic depolarization and binding 
of glutamate to the NMDA (7V-methyl-D-aspartate) 
subtype of glutamate receptor. This is usually ac- 
complished by stimulating the synapses at a high 
frequency. Is it possible to induce LTP if the den- 
drite is instead depolarized by invasion from the 
cell body during an action potential? We have 
tested this hypothesis in the hippocampus by 
pairing antidromically elicited action potentials 
with low- frequency synaptic stimulation. By it- 
self, neither the antidromic spikes nor the synap- 
tic stimulation was sufficient to cause LTP. When 
these two stimuli were paired, however, there 
was potentiation of the population spike, al- 
though the EPSP did not seem to be increased 
consistently. This may be due to a specific poten- 
tiation of the EPSP-spike potentiation. 
The increase in the population spike upon re- 
peated pairing of synaptic and antidromic stimu- 
lation is reversibly blocked by the application of 
AP5, which is an antagonist of the NMDA recep- 
tor. This suggests that the depolarization caused 
by the action potential does invade the dendrites 
and is acting through the same mechanisms that 
occur when the cell body is depolarized by 
current through a recording electrode. It is inter- 
esting that Donald Hebb pointed out in 1949 that 
the firing of postsynaptic action potentials simul- 
taneously with presynaptic activity might be an 
essential condition for synaptic plasticity. This is 
the first suggestion that action potentials might in 
fact have a role to play in inducing LTP. Much 
more work needs to be done to confirm this 
interpretation. 
If calcium currents in the dendrites of hippo- 
campal pyramidal neurons are indeed responsi- 
ble for altering the coupling between synapses 
and the cell body, then it should be possible to 
verify this directly by measuring the intracellular 
calcium ion concentration. Calcium ions are 
known to be central to mechanisms of plasticity 
in the hippocampus and other regions of the 
brain, but little is known of the time course or 
spatial distribution of these changes. Calcium- 
sensitive fluorescent dyes are first injected into 
single pyramidal neurons in hippocampal slices. 
The changes in Ca^^ concentration in both the 
dendrites and soma of the neuron are then moni- 
tored with a confocal fluorescence microscope, 
which allows these changes to be detected with 
high spatial and temporal resolution. The Ca^"^ 
elevations seen in the dendrites were faster than 
those seen in the soma, but of shorter duration. 
This would be expected, as their volume is much 
smaller. We are continuing to develop these tech- 
niques and increase the sensitivity of our record- 
ings so that we may be able to correlate physiolog- 
ical changes with changes in the spatial 
distribution of intracellular calcium. 
The projections between areas of the hippo- 
campus have a complex three-dimensional geom- 
etry that is difficult to delineate. Anatomists have 
studied them by injecting tracer into one area and 
serially reconstructing the terminal arborization 
of filled neurons. In collaboration with David 
Amaral of the Salk Institute, we have examined 
such a data set from injections in area CA3 project- 
ing into CAl of the rat hippocampus. We then 
used a mathematical model of these projections 
to reconstruct a three-dimensional map of the 
pathway. 
By this technique we have uncovered an aspect 
of hippocampal structure previously unseen: a se- 
ries of stripe-like modulations along the longitu- 
dinal axis. We are now attempting to demonstrate 
these modulations in direct experiments. We in- 
tend to create a realistic model of neuronal net- 
works in the hippocampus from our recon- 
structed pathways. 
400 
