Interpreting "imitative" responses in early infancy.

highly significant decrease in the number of GABAergic axon terminals at sites of seizure foci. The decreased number of GAD-positive terminals in our immunocytochemical preparations of epileptic monkey cortex could be explained by an alumina gelproduced loss of antigenicity of GAD molecules. However, this is unlikely since most of the alumina is located in macrophages with slight amounts in astrocytes (16). Furthermore, not all of the GAD-positive terminals are lost at seizure foci, and the staining of the remaining terminals indicates that the antigenicity of GAD is not affected by the alumina gel. In addition, a differential loss of antigenicity caused by differences in the diffusion of alumina from the application site is unlikely since, in monkeys with subarachnoid injections, the deep cortical layers display similar decreases of GAD-positive terminals to those observed in the superficial layers directly subjacent to the alumina gel. Therefore, the loss of immunocytochemically detectable GAD from cortical axon terminals indicates an actual loss of GAD molecules and this could be due either to a severe impairment of GAD synthesis, or to the degeneration of GABAergic somata or their axon terminals, or both. Although previous biochemical data have indicated decreased GAD activities at seizure foci (5), our results extend this finding and show a numerical decrease of GAD-containing axon terminals. Whether these terminals actually degenerate or merely lose immunocytochemically detectable GAD is not known. However, a degeneration of GAD-containing terminals is suggested by the results of ultrastructural studies that show a decreased number of presumed inhibitory, symmetric synaptic junctions with somata and dendritic shafts of cortical neurons at seizure foci (17). In either event, a functional loss of GABAergic cortical neurons would occur. Our experimental preparations indicate that the magnitude of this loss is significant and could be expected to reduce the inhibitory synaptic control of pyramidal neurons, thus leading to a hypersensitivity of these cells to normal excitatory synaptic inputs (4, 5). The reason for this loss of GABAergic terminals at seizure foci is unknown, but it is possible that aspinous stellate neurons may be highly susceptible to alterations induced by alumina treatments. Further support for a GABAergic involvement in epilepsy is derived from pharmacological studies that show that certain convulsant and anticonvulsant drugs act at GABAergic synapses in the central nervous system (18). Thus, our results in combination with those of the other studies cited in this report support a hypothesis that a loss of functional GABAergic neurons leads to focal epilepsy. CHARLES E. RIBAK* Division ofNeurosciences, City ofHope National Medical Center, Duarte, California 91010 A. BASIL HARRIS Department of Neurological Surgery, University of Washington, Seattle 98195 JAMES E. VAUGHN EUGENE ROBERTS Division ofNeurosciences, City of Hope National Medical Center