Selective inhibitory synapse loss in chronic cortical slabs: a morphological basis for epileptic susceptibility.

a for epileptic susceptibility. Electron microscopic examination of pyramidal neurones at the edges of chronic slabs of cerebral cortex in the cat revealed a selective loss of inhibitory (symmetric axosomatic) synapses compared with pyramidal neurones in the centers of the slabs. It appears likely that the neurons at the edges, which retain excitatory input (asymmetric axodcndritic synapses) in the ncuropil, but totally lack the somatic inhibitory input, act as the focus for the prolonged seizure activity which occurs in chronic cortical slabs. RIBAK, ct 1982. Selective inhibitory synapse loss in chronic cortical slabs: morphological for epileptic susceptibility.


Introduction
edges the nerarones were devoid of all axoso~maatic Chronically denervated cerebral cortex exhibits prolonged epileptiform activity after electrical or ciaernical stinlulation (Grafstein and Sastry 1957;Reiffenstein 1979). 'The suggestion that a generalized loss 08' recurrent collakerals to inhibitory neurones was responsible for the prtplonged seiza~res (MrnjeviC r7t trl. 197CBb was recentIy shown to be incorrect (Weiffenstsin 1979). Although inhibition and excitation in the nriddle of ciaronic slabs of cortex is normal (KrnjeviC et al. 1970a, 197BBb), there is a decrease in total y-amiraobutyric acid (GABA) content of chronic cortical shbs (Reiffenstein and Neal 19'74;Kx~yarna and Jasper 1977), which suggested that there is a localized deficit of inhibitory transmission, possibly at the edgcs of the slabs. A deficit of' this sort has recently been demonstrated at the edges of alumina gel epileptic lesions (Ribak c7t a/. 19'79). '%'his report is a qualitative description of an electron microscopic examination of synapses at the edges of chronic cortical slabs. methods CIiros~ic cortical slabs were prepared in suprasylvian gyri 08' four cat5 according to the method of Burns (Krn.jcvic ot ( I / . 19900; Reiffenstein 19791. At 9-868 weeks, when thc epileptic susceptibility is frilly developecf (Slrarpiess and Halpcrn 148621, the cats werc again anesthetized with pentobarbital and the brains fixed by Bntracarc1Bac perfusion of aldehydes (Kibak er trl. 1C379j. Blocks of' tissue wsrc cut from both contralateral suprasylvian gyrus and chror1ic slabs. 'B'hesc werc osrnicated, en bloc, staineci with ur:mayl acetate, dehydrated, and ernbeddect in cpon resin. Se~nithin. I-ylm sections were takcn from each block to identify cell layers and the edges of each slab (or the homotopic area in the comtralaterril cortex). 'Thin sections for clectron ~~~i c r o s c o p y were obtained from layers V and VE. to compare slab edges, the middle of the slabs, and normal cortex.

Results
FPyranaida% neurones with their Bong apical dendrites could easily be identified in the contralaterai cortex and throughorat the chronic slaks. Tn contralateral cortex axosomatic symmetric synapses (inhibitory) were regularly present on pyramidal cells, and in the adjacent neuropil there were numerous ternainals rnaking synapses (mostly asymrnctric, which is thc type corrc-Bated with excitatory function B with dendrites and dcndritic spines (Fig. I ). The number of these asymnnetric synapses is undoubtedly diminished, since input fibres from ahc opposite cortex (which form purely excitatory. asymnnetric synapses: for a review see Colc~nnier I98 I have been severed by the undercut. In spite of the lesion, however, ahcre is no obvious lack of asymmetric synapses, and astrogliia are relatively infrequent. B n contrast, astroglia were ntarncrous klsroughout the chronically isolated slabs (Figs. 2 -8), particu8arly near the edges ( i .e., the isolation cut). Within I00 prn of the synapscs (Figs. 3-61 and the somata of the neuronex were almost completely surrounded by layers of astroglia. The ternainals ~h i c h were present in this pan? of the slab appeared to be nnaking asynairaetric synapse\ mith dendrite5 or derldritic spines (Figs. 2,4,and 6 ) . Neurtsales in the middle of the slabs had approximately nc~rnaal nunalbers of i~~hibitory axososnatic synapses, while neurones 1 mm from the edges had intermediate nunrhers of axosornatic synapse\, and also T Y B O~~ terminals which were apposed to thcsc \onsiata ccmparcd with neurones near the edgcs (Figs. 7 and 8). Fewer astrocytes were pracnt in the center than at the edge\, and in contrast the nea~ronal sornata in thc center were 1101 completely surrouirded by them. Axc~dendritic synapses in the neraropil \+ere pre4ent aII aca-o\s the isolated slabs, even within 100 p n froit1 the edge\ of the slabs (Figs. 2-81.

Discussion
In the chronic slabs of cerebral cortex there is La total loss of synapses locrated on pyramidal cell bodies close to the ecfges c~f the slabs. Synapses normally prssenat at this site are always of the syanrnetrical type (Colonnier 1968, I98 1 ). 'b'er~~linals forming thcse synapses in salonkeys arad rats contain glutansic acid dccarboxylase (GAD), and these synapses arc tkaeretiarc correlated with the raacdiatiora of GABAergic inhibition in the cere-bra1 cc~rtex (Rilsak or c d . 1979). Ribak c)t trl. ( 1982) lriave recently extendcd these stutfies to show slaat virtually all symsmetric cortical synapses lare GAHAergic. In contrast to their total loss 611'somakic inhibitory input, these neulnnes at the edges of the slabs appear to havc nairatained an important excitatory input frona axodendritic synapses in the saeuropil. Moreover, the so~aaaita of these lseurcsncs near the edges also were conaplctely wr:tpped ir-a multiple layers of rctlctive astrocytcs. This appears very similar to the morphology of' alua-maina lesioras reported by Ribak 6.1 c d . (  FIG. 3. Low magnification EM of soma of a neuroanc less than 50 prn froan the edge of the slab (to the left). An accumulation of organelles, \uc& ac cisternae of G'solgl complex. occur at the apical end where the nucleus (N) shows an indentation. This soma is alrt~ost entirely surrounded by profiles of reactive astrocytes. This glial coverii?g is only interrupted by collagen fibres (CB and an axon terminal (arrow) that synapses with an adjacent dendrite (see Fig. 6). The boxed areas are shown at higher nlagnifications in Figc. 3-6. x I 1 000. F~G .
4. Higher ranagnification of area in upper right of Fig. 2. Fjilarnents sectioned either transver\cly or longikb~dinallj appear in profiles of reactive astrocytes ( ( a ) that are closely contacting the ncuronal soma. A slightly dicruptcd dendrite (~i ) appears to form a synapse (arrow) with rr \mall axon terminal. X 27 000. 'This soma has three axon terminals ( i ) close to it, and one of these ft,rans : a symmetric synapse (arrsw~he:aal). Nn~~sercaus astrocytes ( a ) are present on the neuropil. 'Two other terminals (;, and r2) form asynametric (arrcpws) and sy~nmetric (arrowhead) synapses with a dendrite. x 25 000. FIG. 8. lanother portion of thc same pyramitial cell soma as shown in Fig. 7 shows a tsrrninal ( t , ) forming a symmetric axosornatic synapse (arrowherads), with astocytic processes (tr) covering the remraiaming sonla! s u r f a x . In the acijacent neurcapil, a bouton en ycrsscrnf it2) forms an asymmetric synapse iarrcw) with a tlentlrite, and :antather terminal ( b3! fomls a similar type s y n a p x (arrow). X 22 500. or if there has been a loss of acpir~oars stellate cell bodies. The latter has been reported in some human lesions (Hraak and GobeI 1978). Nevertheless, it seems probable that the pyramidal cells at the edges are acting as the focus for the prolonged epileptiforrn discharge which is characteristic of chrc~alically denea-vatcci slabs of cerebral cortex. Other nearby pyrarniclal cells, with partial losses sf inhibitory synapses, are probably more easily recruited into the activity of the focus than [leurmes in thc middle of ttac slab, or in normal cortex. On the basis of the decrease in CAWA content (Reiffenstein and Neal 1974;Koyania and Jaspcr 1977) one w o~~l d predict that the fottri loss of inhibitory synapses throughaalt the slab is about 3eBCk. The absence of rnlajor losses of inhibitory or excitatory inputs to the neurones in the naiddlc of the slabs does at least provide an explanation why most other investigations have found an apparent normalcy oT most neurones in epileptic chronic cortical slabs be.g., Krnjevii e f (&I. 1970u, 1970b(&I. 1970u, : Reiffenstcin 1979. Qeeantitativc data for the distributions s f inhibitory and excitatory synapses in these preparations, and also in slabs of 1-2 weeks duration (whcn the slabs do not yet show pra~longed scizurcs (Sharpless and Halpern 1962)) arc currently being accumeelatcd.
'I'he authors thank Dr. Eugene Roberts for Facilitating this cc~EEaboration, and Lan-y Richrnan and Stephen Nuara fc~r techraicral assisttince. R.J. K. thanks Ilr. K . Kra?jeviC for his hospatality during a year of sabbatical leave. Thc work was supported by grants fromn the United States Parblic ifealth Service to C.E.W. (NS-156691, frc~m the Medical Research Council of Canada to Dr. Krnjevie, and a travel grant fro111 the Alberta Hcrttage Foundation bkjr Medical Research to R.J.R.