Quantitative Estimation of the Ratio of GABA-Immunoreactive Cells in Neocortical Grafts

Somatosensory anlage from 17-18 day old rat embryos were transplanted in place of the removed barrel cortex in adult rats. Six to eight months after transplantation, the grafts were either completely separated by glial scar or partly separated and partly confluent with the host neocortex. Each was studied histologically and immunostained for GABA. It was found that in partly confluent grafts the neuronal density was similar or even higher than in the host cortex, while the cell number in the separate grafts was much lower than in the nearby host cortex. The number of GABA-positive cells, however, was in all grafts significantly lower (2.9% on average) than in the normal cortex (11.8% on average).The decline in GABA-stained nerve cells was highest in separated grafts, but was somewhat less marked in transplants partly confluent with the host tissue. The possible role of partial or total deafferentation as well as the relative vulnerability of the transplanted tissue by temporary hypoxia and other metabolic disturbances are discussed as the probable factors in selective decline of GABA-ergic cells in the transplanted somatosensory cortex.

embryos were transplanted in place of the removed barrel cortex in adult rats. Six to eight months after transplantation, the grafts were either completely separated by glial scar or partly separated and partly confluent with the host neocortex. Each was studied histologically and immunostained for GABA. It was found that in partly confluent grafts the neuronal density was similar or even higher than in the host cortex, while the cell number in the separate grafts was much lower than in the nearby host cortex. The number of GABA-positive cells, however, was in all grafts significantly lower (2.9% on average) than in the normal cortex (11.8% on average). The decline in GABAstained nerve cells was highest in separated grafts, but was somewhat less marked in transplants partly confluent with the host tissue. The possible role of partial or total deafferentation as well as the relative vulnerability of the transplanted tissue by temporary hypoxia and other metabolic disturbances are discussed as the probable factors in selective decline of GABA-ergic cells in the transplanted somatosensory cortex.

INTRODUCTION
It has been shown previously /2,3/ that neocortical grafts surviving within the barrel field of the adult rat can be subdivided into two groups. The first group became confluent with the host brain and was only partly separated from the host tissue by glial scar. The second group was eompletely separated from the host brain by an uninterrupted glial capsule.
These two groups of neocortieal grafts also exhibited different electrical activities. The pattern of background activity of the neurons in the first (confluent) group proved to be similar to that of the intact cortical neurons. These neurons responded both to electrical stimulation of the host brain and to sensory stimulation of the animal. In the second group (separated grafts), the background activity consisted of hypersynchronous discharges of many neurons., Spontaneous epileptiform discharges were detectable only in neurons which did not respond to stimulation in the host brain. It was assumed that the appearance of epileptic activity depended on dysfunction of the GABA-ergic system in the graft. The aim of this study was, therefore, to determine the numerical density of GABA-immunoreactive neurons in both separated and confluent cortical grafts. VOLUME 2, NO. 3-4, 1991

Grafting
Wistar rat embryos on the 17-18th gestation day were used as donors. The embryos were decapitated and small (about 2 x 2 ram) pieces of the brain were dissected from the anlage of the somatosensory area. Adult rats of the same strain served as recipients. Under pentobarbital anesthesia (40 mg/kg i.p.) a 5 x 6 mm piece was removed from the fight side of the skull, and the barrel field was removed by sucking through the opening of the dura. The depth of the cavity was 1 1.5 ram, its width was about 2 x 4 ram. One solid graft was placed under the incised dura in each recipient. Finally the bone was replaced over the opening of the skull and the skin sutured.
Six to eight months after transplantation the host rats were perfused through the heart (ascending aorta) by 300 ml 2.5% paraformaldehyde and 1% glutaraldehyde (in 0.1 M phosphate buffer, pH 7.2-7.4). A piece of the animal's braincontaining the graft and the surrounding neocortex-was removed and cut by tissue chopper in slices of about 500 600/m thick. The slices were stored overnight at 4"C in the same fixative, washed several times-in phosphate buffer and postosmicated in 1% OsO (in 0.1 M phosphate buffer, pH 7.2-7.4) and embedded in Epon.

GABA-immunostaining
The selection of the proper blocks was made by using toluidine blue stained semithin sections. Neuronal nuclei (test objects) were outlined at 800 times magnification, and their area as well as the whole investigated area measured. The numerical densities were calculated using the formula of Weibel and Gomez /23/ for ellipsoids: (N^) where Na is the number of nuclear profiles per unit area, V v is the volume (or area) fraction, K was chosen as 1.05 and/3 = 1.39, which is a function of the axial ratios of the ellipsoidal nuclear cross-sections studied.
The grafts (and the host cortices)were analyzed using a Hipad digitising tablet attached to an IBM-PC. In parallel with the GABAimmunostained semithin sections, the GABAimmunoreaetive neurons (neuronal nuclei) were outlined and their numerical density calculated similarly to the toluidine blue stained semithin sections. The incidence of GABA-positive neurons was calculated from the two numerical density data.

RESULTS
Seven rats with surviving grafts were used in this experiment (Table I). The exact position of the grafts within the cortex (i.e., continuous or contiguous with the host tissue) was determined in serial semithin sections. Measurements on the graft-host border revealed that the grafts exhibited different grades of confluence with the host brain. Two grafts (No. 130, 231) were found to be completely separated from the host brain by a continuous glial scar (Fig. 1C). Five other grafts were confluent with the host brain in one or more regions (Fig. 1A,B; Fig. 2A). The length of the confluent and separated 'borders" were normalized to 1.0 and their ratio varied from 0.1 to 0.8.
In GABA-immunoreacted semithin sections some GABA-positive neurons were seen in the separated grafts, while in the host cortex and in the "confluent" grafts the frequency of immunopositive cells was high (Fig. 2B,C). Immunopositive punctuate structuresmost probably corresponding to the cross-sections of A. The transplant (G) is separated in most places from the host cortex (H) by glial scar, but is continuous .with the host cortex between arrows. The larger blood vessels are indicated by arrowheads. Insert: schematic drawing of the topographical localization of the graft in the host somatosensory cortex. B. The continuous part indicated by arrows in Fig 1A is 3A). The apical dendrites of grafted cortical pyramidal cells were also surrounded by small dark dots (GABA-positive terminals), particularly in the confluent grafts (Fig. 3B).  A. GABA-positive perikarya appear somewhat larger than GABA-immunonegative somata. B. GABA-immunopositive (arrow)as well as immunonegative (arrowhead) cell bodies and apical dendrite are covered by GABA-immunopositive axon terminals.
GABA-positive cells showed considerable individual variation. In the confluent grafts the GABA cell density varied between 0.98 and 2.08 x 10 3 neurons/mm3; in the separated grafts, however, the GABA cell numerical density ranged between 0.36 and 0.55 x 103 neurons/mm 3 only.

DISCUSSION
The morphometric and quantitative estimation of the neurons in the neocortical grafts as compared to the host somatosensory neocortex reveal some remarkable differences.
The mean neuronal nuclear diameter, usually paralleling the size of the neuronal soma, was slightly, but non-significantly, larger in the grafts than in the host somatosensory neocortex. This difference, however, became significant when populations of the GABA-positive cells were compared. The increased size of GABA-positive cells in the graft might be regarded as an indication of delay (or of some disturbance) in their development/16/.
In the host somatosensory cortex of the analyzed samples the average numerical density was about 41 x 103 neurons/mm .T his is about thirty percent less than the value found by Gabbott and Stewart/9/in the visual cortex of the same species (60 x 103 neurons/mm3). Since, however, the N v values for the visual cortex are significantly higher in other mammals /4,7,11,17,22/ as compared to other cortical cytoarchitectonic areas, our data apparently represent the true values in the somatosensory cortex of the rat.
It was found that the average numerical density of neurons in the grafts was practically the same as in the host cortex, although a detailed comparison with normal cortex was difficult, since the organotypic layered 241 arrangement was usually absent in the grafts. It is well known that the numerical density of the cortical neurons is lowest in the Ist layer, high in the Vth and VIth layers, and maximal in the Ilnd-IVth layers/8,9/. It appears that neuronal numerical density in the graft depends on the histological characteristics of the embryonic tissue, which may also explain the scattering of N v in the various grafts. However, as a general trend, it was recognized that complete separation of the transplant resulted in a more severe cell loss as compared to the partly confluent grafts. In more integrated transplants, the cell density was, in fact, even higher than in the host cortex.
Nevertheless, irrespective of the cellular density in the grafts, the absolute number of GABA-cells is much lower in the transplants than in the host cortex. The mean numerical density of GABA-immunoreactive neurons in the host cortex was 4.98 x 103 neurons/ram (11.8% of all neurons). This value agrees with that of Gabbott and Stewart/9/, who reported that about 10% of cortical neurons in the rat visual cortex are non-pyramidal (local inhibitory nerve cells). The  neurons are generated only later. In the present study, embryonic material was taken for grafting on the 17-18th day and, therefore, it was likely that it contained a nearly complete number of GABA-ergic neurons. Available exper.imental evidence suggests that GABA-ergic neurons have prominent functional plasticity, and that expression of GABA-markers depends strongly upon the degree of their normal sensory afferentation. A rapid reduction in the intensity of immunostaining for GAD was observed 3 days after peripheral deafferentation in the barrel cortex of adult animals /24/. A decrease of the number of GABAimmunostained neurons in the dominance columns of the primary visual cortex of monkeys was also elicited by simply occluding the corresponding eye /10/. At the same time, functional sensory deprivation resulted in disturbances of the cortical inhibitory processes. Neurons in the cortical barrels deprived of sensory input by destroying the appropriate vibrissae exhibit higher spontaneous activity and reactivity to stimulation of the neighboring vibrissae; their on-off responses are more prolonged, because they are not curtailed by inhibitory postsynaptic potentials /19/. Interestingly, the functional change was not paralleled by any changes of the general cytoarchitectonics of the deprived barrels.
Thus, as one of the possible explanations, it may be suggested that the observed reduction in the number of GABA-immunoreactive neurons in the grafts would be caused by disturbance or by absence of sensory afferentation of the grafted tissue. This assumption seems to be supported by the finding of an extremely low number of GAD-positive neurons in intraocular neocortical grafts completely deprived of extrinsic nervous connections/5/.
Another possible explanation for the decline of GABA-positive neuronal density in the grafted cortical tissue may be the relative vulnerability of GABA-ergic neurons to temporary hypoxia /1,18/ or other metabolic damage, which is practically unavoidable during the transplantational procedures. It is also possible that the disturbed or non-existent specific afferentation and the temporary hypoxia would jointly contribute to the decline in the number of GABA cells in transplanted cortical tissue. In either case, it appears that low levels of GABA markers correlate with reduced efficiency of intracortical inhibition.
Abnormal epileptiform activity in the isolated cortical and hippocampal grafts were explained in another work by formation of recurrent "autosynapses" between the intergraft neurons, deprived of normal afferent connections and efferent targets 242 A. BRAGIN, .TAKACS,O . VINOGRADOVA & .HAMORI /25/. The present findings indicate that the low number of GABA-ergic elements might directly contribute also to the observed abnormal epileptic activity.