Quantification of synapses formed with apical dendrites of Golgi-impregnated pyramidal cells: variability in thalamocortical inputs, but consistency in the ratios of asymmetrical to symmetrical synapses.

Five pyramidal cells from the posteromedial barrel subfield of mouse SmI cortex were labeled by Golgi impregnation and then gold-toned and de-impregnated (Fairen, Peters & Saldanha, 1977). Subsequently, 40 to 70 pm-long segments of their apical dendrites occurring in layer IV were graphically reconstructed from serial thin sections to determine the distribution of their synapses. Thalamocortical synapses onto these dendritic segments were identified by lesion-induced degeneration. The synaptic pattern of the pyramidal cell apical dendrites was consistent with previous reports in that most synapses occurred on spines and were asymmetrical and the smaller number of shaft synapses were primarily symmetrical. Some axospinous synapses were formed by degenerating thalamocortical axon terminals. The proportion of thalamocortical synapses onto reconstructed dendritic segments was different for different neurons. For example, thalamocortical axon terminals formed 15% of the synapses involving the spines of the reconstructed segment from a medium superficial layer V pyramidal cell and 10% of the synapses onto portions of the segment from a large layer VI pyramidal cell. In contrast, reconstructed dendritic segments of three other layer VI pyramidal cells formed no more than one thalamocortical synapse. An analysis of the distribution of synapses onto reconstructed dendritic segments revealed that the segments of 3 medium and large pyramidal cells had a ratio of about 12.5 asymmetrical synapses per symmetrical synapse, whereas the segments of 2 small pyramidal cells had ratios of only 6.5 asymmetrical synapses per symmetrical synapse. That these ratios fall into 2 distinct groups suggests that the relative number of asymmetrical and symmetrical synapses is stereotyped within populations of neurons. BECAUSE of their large size, pyramidal cells were among the first neuronal types studied by the pioneering neurohistologists of the 19th century (see historical review by BRAZIER, 1978). Subsequent examination of Golgi impregnations of the cerebral cortex has shown pyramidal cells to be the primary source of axons leaving the cortex (e.g. RAM~N Y CAJAL, 1911; L~RENTE DE No, 1938). Thus, to a large extent, the study of information processing by the cerebral cortex must involve the study of synapses made with pyramidal cells. Electron microscopy has revealed that pyramidal cells form symmetrical synapses on their cell bodies, dendrites, and axon initial segments, and asymmetrical synapses primarily on their dendritic spines (PETERS & KAISERMAN-ABRAMOF, 1969; JONES & POWELL, 1970a; LEVAY, 1973; PARNAVELAS, ULLIVAN, LIEBERMAN & WEBSTER, 1977; SOMOGYI, 1977, 1978, 1979; CHRISTENSEN & EBNER, 1978; SOMOGYI, HODGS~N & SMITH, 1979). Two sources of symmetrical synapses onto pyramidal cells are non-spiny and sparsely-spined stellate cells (PETERS & FAIII~~N, 1978; PETERS & PROSKAUER, 1980) and chandelier cells (SOMOGYI, 1977, 1979; SOMOGYI et al., 1979; PETERS, 1980). Using lesion-induced degeneration to label corAbbreviation: PMBSF, posteromedial barrel subfield. tical afferents, it has been demonstrated that some of the asymmetrical synapses onto pyramidal cells arise from the thalamus (STRICK & STERLING, 1974; WHITE, 1978; SOMOGYI, 1978; SOMOCYI et al., 1979; PETERS, PROSKAUER, FELDMAN & KIMERER, 1979; HER~CH & WHITE, 1980; HORNUNG & GAREY, 1980; HENDRY & JONES, 1980), whereas others arise from the contralateral hemisphere (P. B. CIPOLLONI & A. PETERS, personal communication). Missing from these studies is any information on the proportion of synapses contributed by thalamic or callosal sources to pyramidal cells. This study is intended in part to rectify this deficiency by determining the proportions of thalamocortical synapses impingeing on portions of the apical dendrites of different types of pyramidal cells. In this way, an indication of the potential significance and variability of thalamocortical input to pyramidal cells can be obtained. Furthermore, the commonly accepted notion of the differential effects of asymmetrical versus symmetrical synapses strongly suggests that the ratio of these synaptic types is an important determinant of the contribution of dendrites to the excitability of their parent neurons. For these reasons, extensive portions of the apical dendrites of Golgi-impregnated pyramidal cells have been reconstructed from serial thin sections and the numbers of