PROBABILISTIC FUNCTION OF CENTRAL SYNAPSES 
Charles F. Stevens, M.D., Ph.D., Investigator 
One fundamental characteristic of synaptic trans- 
mission revealed by studies of the neuromuscular 
junction is that neurotransmitter release is quantal 
and probabilistic: that is, the presynaptic terminal 
can release only integral multiples of a minimum 
quantity of transmitter, and the exact number of 
transmitter packets released varies in a random 
manner. This quantal release is viewed as reflecting 
the stochastically occurring exocitosis of neuro- 
transmitter-containing vesicles. The ability to iden- 
tify the probability of quantal release and the size 
of the individual quantum is crucial for the analysis 
of synaptic transmission. For example, when a syn- 
apse becomes stronger with use, is it the quantal 
size or the release probability (or both) that 
changes? The mechanism of synaptic strengthening 
cannot be elucidated without answering basic 
questions of this sort. 
Although synaptic transmission in the central 
nervous system has also been reported to be quan- 
tal, no satisfactory analysis of probabilistic release 
at a central synapse has been accomplished. The 
difficulty has been technical. As part of their pro- 
gram to understand the mechanisms underlying 
neuronal information processing, Dr. Stevens and 
his colleagues have developed methods for carrying 
out quantal analysis of synaptic transmission be- 
tween hippocampal neurons. The results have con- 
firmed that the same formalism originally devel- 
oped to describe release at the neuromuscular 
junction is adequate as well for these central syn- 
apses. A simplifying assumption made by all previ- 
ous investigators— that the quantal size at an indi- 
vidual synapse is constant— has been shown, 
however, to be invalid. Thus a successful applica- 
tion of the original formalism for quantal analysis is 
more complicated for these interneuronal synapses 
than it is for the neuromuscular junction. 
Two main problems have plagued quantal analy- 
sis in the central nervous system. First, synapses are 
widely distributed over the neuron's synaptic tree. 
Because of cable attenuation of events that occur at 
more remote dendritic locations, one cannot distin- 
guish between fluctuations in the size of synaptic 
currents due to the site of synaptic contact from 
those that are reflections of the probabilistic nature 
of quantal release. 
Second, the neuronal interconnections are com- 
plex. For a valid quantal analysis, only a single pre- 
synaptic neuron can be stimulated, so that the 
studied connection is restricted to just one syn- 
apse; but in the central nervous system, this is ex- 
tremely difficult. To activate only a single cell, intra- 
cellular stimulation is required, but a given pair of 
neurons are connected so infrequently that the 
chances of recording simultaneously from an ap- 
propriate pair are remote. 
The first requirement of quantal analysis is that 
individual quanta, miniature synaptic currents 
(minis), be identified. Isolated minis are seen in 
central neurons as spontaneously occurring small 
transient currents that continue to appear when 
evoked synaptic transmission is blocked (with 
tetrodotoxin, low calcium/high magnesium, or 
presence of cadmium ions, for example). These 
minis are variable in size and shape, however, and 
one cannot generally know what fraction of the 
variability is inherent and what fraction is the result 
of cable attenuation of events that occur far out the 
dendritic tree. 
Dr. Stevens and his colleagues have circumvented 
these problems by using cultures of rat hippocam- 
pal neurons and evoking mini release with locally 
applied hypertonic solution. By correlating the effi- 
cacy with which hypertonic solution produces 
minis with the presence of boutons (revealed by 
synapsin immunohistochemical localization), the 
laboratory was able to show that the hypertonic so- 
lution action is restricted to the immediate site of 
application. The fact that minis could be evoked at 
any specified location on the dendritic tree made 
two important classes of observations possible. 1) 
The laboratory was able to determine the size dis- 
tribution of minis that occurred adjacent to the re- 
cording site so that sizes were unaffected by cable 
filtering. 2) Precisely what effect cable filtering had 
on the size and shape of minis could be determined 
by causing them to occur at different but known lo- 
cations on the dendritic tree. The cable theory 
could be used to describe these effects quantita- 
tively. Because the shape of minis is dramatically al- 
tered by cable filtering, the site of origin of a mini 
can be identified from its shape. 
Evoked release is conveniently studied in culture, 
because a high fraction of the adjacent neurons are 
connected. Furthermore, synaptic contacts that one 
cell makes on another usually are all formed at one 
site of the dendritic tree, and if multiple contacts 
are made, this can be detected by direct visualiza- 
tion (by dye-filling both neurons that are being 
Continued 
543 
