Activity-dependent synapse elimination can be demonstrated in a compartmental tissue culture system in which two populations of cholinergic neurons converge on and innervate a common target population of muscle cells. This synapse elimination can be blocked by the inhibitor of serine and thiol proteases, leupeptin, at concentrations of 10 nM and greater. The more specific blocker of serine proteases, aprotinin, fails to block elimination as does the thiol protease blocker cystatin. Hirudin is a highly specific inhibitor of the serine protease, thrombin and hirudin does block synapse elimination at 1 nM concentration (aprotinin does not block thrombin). Protease nexin I (PNI), a naturally occurring protease inhibitor (identical with glial derived nexin or GDN) also blocks thrombin and prevents activity- dependent synapse elimination. These data strongly implicate thrombin in the process of synapse elimination. Kinetic analysis of internal calcium, [Ca++]i,, shows that single action potentials produce a larger increase in [Ca++]i than do the individual action potentials of rapid (10 Hz) pulse trains. The peak [Ca++]i reached during repetitive bursts of action potentials, however, is much higher than with single action potentials. These kinetic data are useful in interpreting the mechanisms coupling electrical activity to gene expression. Retinal ganglion cells (RGC) in living chick embryos can be labelled with horseradish peroxidase. The growth cones of the axons of these RGC are visualized with this technique and exhibit dramatic changes at different places in the optic pathway (nerve, chiasm and tract). This preparation should allow evaluation of the effects of different manipulations of the visual system (electrical stimulation, protease inhibitors, cell adhesion molecule antibodies) upon growth cone morphology and axonal projection patterns.
