Neuropathology and Applied Neurobiology 1992,18,145-157
Ultrastructure of neurons containing somatostatin in the dentate hilus of the rat hippocampus after cerebral ischaemia, and a note on their commissural connections F. F. JOHANSEN*, T. SPIRENSEN?, N. TPINDER?, J. Z I M M E R t A N D N. H. DIEMER* PharmaBiotec Research Centre, *Cerebral Ischaemia Research Group, Institute of Neuropathology, University of Copenhagen and ?Institute of Neurobiology, University of Aarhus. Denmark
JOHANSEN F. F., S0RENSEN T., T0NDER N., ZIMMER J. & DIEMER N. H.(1992) Neuropathology and Applied Neurobiology 18,145-1 57 Ultrastructureof neurons containing somatostatin in the dentate hilus of the rat hippocampusafter cerebral ischaemia, and a note on their commissural connections
In a light microscopical study, we previously showed that more than 80% of somatostatin (SS) immunoreactive(4) neurons in the hilus of the dorsal part of the rat dentate gyrus are lost 4 days after ischaemia. In order to verify that the loss of SS immunostainingis due to an actual loss of the SS-i neurons and not merely a loss in expression of S S immunoreactivity, we have now performed an ultrastuctural study of these neurons before and 40 h after 20 min of global cerebral ischaemia in adult rats. The normal SS-i neurons were multipolar and fusiform in shape. The SS-i product was associated with the endoplasmic reticulum and occasionally the Golgi apparatus. The cell nuclei had indentations of the nucleolemma and contained intranuclear rods. After ischaemia, many SS-i neurons in the dentate hilus showed increased electron density of both the cell nucleus and the cytoplasm. In addition the cytoplasm was heavily vacuolated with the SS-i associated with some of these vacuoles. Other SS-i neurons had, in addition to the vacuoles a more homogeneous, and abnormal electron lucent nucleus and cytoplasm. These ultrastructural changes correspond to previously reported irreversible, ischaemic cell changes of neurons. Based on this we conclude that the S S immunoreactivity in the dentate hilus of the dorsal hippocampus is lost after ischaemia because of neuronal necrosis. As a minor part of this study, we examined whether the ischaemia-susceptible SS-i neurons in dentate hilus had commissural'axonal projections. This was done utilizing double fluorescencemicroscopy of retrograde axonal transport of the fluorescent dye, Fluoro-Gold, and the observation that vulnerable SS-i neurons display homogeneously dispersed immunostaining 40 h after ischaemia. Fluoro-Gold was injected unilaterally into the dorsal dentate gyrus 5 days prior to ischaemia. Then, 40 h after ischaemia, sections were stained for SS immunofluorescence, and examined, in the dentate hilus contralateral to the injection, for neuronal co-localization of both events. Cell counts revealed double-labelling of 13% of all neurons which displayed one of the events. This observation suggests that at least some of the ischaemia-susceptible SS-i neurons in dentate hilus do project commissurally. The pathophysiological significanceof ischaemic loss of commissurally projecting SS-i neurons in dentate hilus remains to be determined. Correspondence to: Dr F. F. Johansen, Institute of Neuropathology, Frederik V's vej 11, DK-2100, Copenhagen 0, Denmark.
146
F. F. Johansen et al.
Keywords: electron microscopy, immunocytochemistry, somatostatin, ischaemic cell changes, retrograde axonal tracing, hippocampus, rat
INTRODUCTION We have previously reported the disappearance of somatostatin ( S S ) immunoreactive (4) neurons in the dentate hilus (CA4) of the dosal hippocampus from the 2nd to the 4th day after transient cerebral ischaemia in the rat (Johansen, Zimmer & Diemer, 1987). In the neuronal cell bodies, the SS-i is normally arranged as whorled bands or clusters of immunopositive particles when observed in the light microscope (LM). During the first 24-48 h after ischaemia, the immunoreactivity is increased and becomes homogeneously dispersed in the cell bodies (Johansen et al., 1987). Thereafter about 80% of the immunostained cells in dentate hilus disappear. This loss of immunocytochemical reactivity coincides with an actual loss of neurons in the dentate hilus (Johansen, Lin & Schousboe, 1989;Johansen & O’Hare, 1989;Ternder et al., 1990). Since we, however, in other studies have found that the levels of immunocytochemical staining for glutamic acid decarboxylase (GAD), neuropeptide Y, and parvalbumin, respectively, can increase, disappear permanently, or decrease transiently in hippocampal neurons that otherwise survive an ischaemic insult (Johansen et al., 1989, 1990; Johansen & O’Hare, 1989), we wanted to determine whether the hilar SS-i neurons actually died or merely lost their S S immunoreactivity after ischaemia. For this purpose we decided to examine the hilar SS-i neurons 40 h after ischaemia in the electron microscope (EM), and correlate the observed changes with the ischaemic cell changes described by others (Brown & Brierley, 1972). Several authors have described commissural and ipsilateral hilodenate projections from dentate hilar neurons, and some of these cells have been identified as SSi, where at least the ipsilateral projection goes to the outer part of the dentate molecular layer (Zimmer, Laurberg & Sunde, 1983; Leranth & Frotscher, 1987; Bakst et al., 1986; Leranth, Malcolm & Frotscher, 1990). The main commissural and associational projection terminates in the inner part of the dentate molecular layer and is unlikely to be somatostatinergic (Laurberg & Serrensen, 1981; Swanson, Sawchenko & Cowan, 1981; Bakst et al., 1986; Leranth et al., 1990). Since it is possible that associational and commissural projections from SS-i neurons play a role in the inhibition of the granule cells (Sloviter, 1987),a loss of these projections may influence the postischaemic pathophysiology. Therefore, in order to determine whether the ischaemia-susceptible SS-i cells in the dentate hilus do project commissuraily, we combined the use of retrograde axonal transport of the fluorescencedye, Fluoro-Gold, with immunofluorescent staining for S S at 40 h after ischaemia. We examined the co-localization of Fluoro-Gold and post-ischaemic S S immunofluorescence of the homogeneously distributed type in the hilar neurons (Johansen et al., 1987). MATERIALS A N D METHODS
Induction of ischaernia All experiments were conducted on adult male Wistar rats, weighing 350400 g. Twenty minutes of cerebral ischaemia was induced by means of four-vessel occlusion (hlsinelli & Brierley, 1979) with minor modifications as described in detail by Johansen et al. (1989) and
Ultrastructure of somatostatin neurons after ischaemia
147
Johansen & O'Hare (1989). Briefly, the vertebral arteries were electrocauterized under methohexital anaesthesia (50 mg/kg body weight). The rats were then left without food, but with free access to water, until the following day, when the carotid arteries were gently exposed during brief halothane anaesthesia. Anaesthesia was discontinued for 2 min, before both carotid arteries were ligated for 20 min. After the occlusion period, the rats were allowed to survive for 40 h. Tracing of commissural projection Five days prior to ischaemia, three rats received stereotaxic injections of 1 pl of a 4% solution of Fluoro-Gold (Fluorochrome-Inc., Englewood, Colorado, USA) dissolved in Ringer's solution, into the dentate hilus of the dorsal hippocampus on one side using the following coordinates; 2.6 mm behind the bregma, 2.0 mm lateral to the midline and 3.5 mm below the cortical surface (Pellegrino, Pellegrino & Cushman, 1979).This procedure results in a widespread Fluoro-Gold labelling of dentate hilar neurons along the longitudinal septotemporal axis on the injected and the contralateral side corresponding to the ipsilateral and commissural projections from these cells (Laurberg Kc Serrensen, 1981). Forty hours after ischaemia hippocampal tissues were processed for immunofluorescencestaining for SS.
EM immunocytochemistry Eight adults rats subjected to ischaemia and six normal rats were perfused transcardially, during pentobarbitone anaesthesia (50 mg/kg body weight), with 4% paraformaldehyde, 0.05% glutaraldehyde and 0.2% saturated picric acid in 0.1 M phosphate buffer (PB), pH 7.4. The brains were gently removed from the skulls and stored at 4°C for 2 h in 4% paraformaldehyde and 0.2% picric acid in PB. The hippocampal region was then exposed, and the dorsal half cut perpendicular to the longitudinal axis in 50 pm thick sections on an Oxford vibratome in PB. To enhance antibody penetration, the sections were soaked in a 15% sucrose solution in PB, frozen in liquid nitrogen, and thawed at 4°C for 2 h. Sections were thereafter rinsed in TBS (Tris buffered saline) and processed for immunostaining (Sternberger, 1979). First the sections were pre-incubated in 1% swine serum (X901, Dakopatts) in TBS for 30 min and then incubated with the primary SS antibody (1 :2400, A566, Dakopatts) in TBS overnight at 4"C, rinsed in TBS and incubated with the secondary swine-anti-rabbit antibody ( ~ 3 0 2196, , Dakopatts) in 0.25% BSA (bovine serum albumin) in TBS for 30min at room temperature, rinsed in TBS and incubated in PAP (1:75,Z113, Dakopatts) in 0.25% BSA in TBS for 30 min, rinsed in TBS, then exposed to diaminobenizidine (DAB), finally rinsed once in TBS and twice in PB. The primary antibody was diluted in TBS with 0.25% BSA and 0.01% Triton XlOO which did not alter the preservation of the ultrastructure of the tissue. After the immunostaining and post-fixation in 1 % OsO, in PB for 1 h, the 50 pm vibratome sections were dehydrated and flat-embedded in Epon between two pieces of plastic foil. Selected sections with SS-i cells were then re-embedded on top of empty Epon blocks. From these, semithin 3 pm thick sections were cut and mounted on glass slides and every third section was stained with toluidine blue. Semithin sections containing SS-i cells were identified by LM, re-embedded on top of new Epon blocks, trimmed and cut in thin sections (50-60 nm). The thin sections were mounted on slot grids coated with formvar film (three grids) and on hexagonal 400 mesh slim bar grids (SPI supplies, 2240c) (two grids) and examined in a Jeol-100s electron microscope. From each rat, 10 semithin sections were selected as described, and from each of these sections, 2 4 SS-i cells were analysed by EM. Exclusion of the primary SS antibody was used as a negative control.
148
F. F. Johansen et al.
Table 1. Co-localization of somatostatin and Fluoro-Gold in ischaemic hilar neurons. Differential cell counts from the dentate hilus of the dorsal fascia dentata in rats, which 5 days prior to ischaemia had Fluoro-Gold injected into the contralateral hilus, and then 40 h after ischaemia were killed and processed for combined Fluoro-Gold and SS-FITC immunofluorescence. Counts of neurons with SS-FITC immunofluorescence (neurons containing somatostatin) refers selectively to ischaemically damaged neurons with homogeneously dispersed SS-FITC immunofluorescence
Numbers Percentage of total Mean kSD per hilus section
Animals
Sections
Somatostatin
FIuoroGold
Co-localization
Total
3
54
93 1 59%
843 54%
207 13%
IS67
17.4k 3.4
15.7+ 1.5
3.9k2.5
100%
Other vibratome sections were processed for electron microscopy without further immunocytochemicaltreatment. Somatostatin immunofluorescencestaining and fluorescence microscopy The three rats used for combined retrograde tracing and immunofluorescence for SS were perfused transcardially with 4% paraformaldehyde in PB. After removal from the skull, the brains were stored for 4 h in 4% paraformaldehyde in PB at 4°C before they were cut on a vibratome in 50 pm thick coronal sections. These were pre-incubated in 1% normal swine serum (X901, Dakopatts) in PB for 30 min and incubated with 1:2400 SS antibody in PB (Dakopatts, A566) overnight at 4"C, rinsed in PB and finally incubated for 1 h with 2.5% swine anti-rabbit immunoglobulin conjugated with fluorescein-isothiocyanate isomer 1 (SWAR-FITC F205, Dakopatts). Before LM examination, sections were mounted and coverslipped in Entellan@. For the analysis and photography, we used a Zeiss fluorescence microscope with the BP 450490, LP 520 and FT 570 filters for Fluoro-Gold, and LP 397 and FT 395 filters for FITC. Cell counts were performed at 400 times magnification in the dentate hilus of the dorsal dentate gyrus contralateral to the Fluoro-Gold injection. All neurons which showed Fluoro-Gold fluorescence, or homogeneously dispersed SS-FITC fluorescence, or both of these, were counted (Table 1). The dentate hilus was defined and demarcated as previously described (Johansen et al., 1989;Johansen & O'Hare, 1989). RESULTS Somatostatin immunoreactive neurons in the normal dentate hilus Many SS-i neurons were found scattered throughout the entire dentate hilus in both the 50 pm thick vibratome sections and the 3 pm thick semithin sections cut from these. The immunocytochemical staining was in the form of whorled bands and particles within the cell body, stretching into the proximal dendrites. When examined by EM, there were often 2-5 SS-i cell bodies present in the same ultrathin section. The SS-i cells all showed well-developed rough endoplasmic reticulum and Golgi apparatus. The immunostaining was localized to discrete areas of the rough endoplasmic reticulum (Figure 1) and, in a few cases, the inner sac of the Golgi apparatus. The nucleus was large and dominated by euchromatin with a prominent nucleolus and many nuclei contained an intranuclear rod (Figure 1). The nucleolemma had an irregular contour and sometimes deep infoldings,. but these never extended more than half
Ultrastructure of somatostatin neurons after ischaemia
149
Figure 1. SS-i neurons in the dentate hilus of control rat. The framed area outlined in a is shown enlarged in a neighbouring section in b. Immunostaining is confined to parts of the rough endoplasmic reticulum. The nucleus contains an intranuclear rod (arrowhead). (*) freezing artefact. Bar 5 pm (a), 1 pm (b).
way through the nucleus. The dendrites of the SS-i neurons appeared spineless and received asymmetrical synaptic contacts. Further details regarding the structure of the pre-synaptic terminals and the dendrites, were difficult to define, due to deterioration of the tissue caused by the immunocytochemical processing.
Somatostatin immunoreactive neurons in the dentate hilus after ischaemia Forty hours after the ischaemia there were many SS-i neurons present in the dentate hilus. By light microscopy, the majority of the neurons did, however, differ from the neurons in the normal rat. Some neurons had an almost normal, but irregularly distributed and abnormally intensified SS-i staining (Figure 2), while others displayed a very intense and homogeneous staining of the cell bodies extending into the dendrites (Figure 2). In semithin sections, some
150
F. F.Johansen et al.
Figure 2. Vibratome section from dentate hilus 40 h after ischaemia. Examples of the homogeneous (thick arrows) and increased (thin arrow) SS-immunoreactivity of cell bodies are shown. Bar 20 pm. Figure 3. SS-i neurons in a semithin section from the dentate hilus 40 h after ischaemia. The intracellular immunostaining appears in clumps, often associated with clear vesicles. g, granule cell layer. Bar 20 pm.
neurons in the dentate hilus had vacuoles in the cytoplasm and an increased toluidine blue staining of both the cytoplasm and the nucleus. In some vacuoles, SS-i staining could be discerned at the circumference of the vacuoles (Figure 3). By electron microscopy, the cytoplasm showed pronounced vacuolation (Figures 4 and 5), and the appearances corresponding to those described by Brown and Brierley (1972). In agreement with this, the vacuoles were found to consist primarily of dilatations of the rough endoplasmic reticulum. The SS-i staining was located in association with some of the vacuoles (Figures 4 and 5). Sometimes double-walled membranous material appeared in vacuoles without any immunostaining, indicating that they were swollen mitochrondria (Brown & Brierley, 1972; Brierley, 1984). Other mitochondria displayed intact cristae and had normal dimensions, but showed increased electron density of the matrix. In some of the vacuolated SS-i cells, the cytoplasm was dark (Figure 4). The increased density of the cytoplasm and the vacuolation also extended into the dendrites. The cell nuclei were dark and shrunken (Figure 4) with a homogeneous nucleoplasm in which nuclear inclusion bodies occasionally were recognizable. In other neurons with vacuoles, the cytoplasm was electron lucent and the nucleus swollen (Figure 5). Occasionally, cells with disintegrating cytoplasm were associated with reactive, swollen astrocytes.
Ultrastructure of somatostatin neurons a f e r ischaemia
151
Figure 4. Electron micrographs of SS-i neuron in dentate hilus 40 h after ischaemia. Framed area in (a) is enlarged in (b). The electron dense cytoplasm is filled with vacuoles and the immunoreaction product is associated with some of these vacuoles (arrows). T h e nucleus (N)is dark, shrunken and irregular in shape. Bar 3 pm (a), 1 pm @).
In general, the post-ischaemic ultrastructural appearances of the SS-i neurons can be summarized as follows: group A , neurons with electron dense cytoplasm, vacuoles with or without SS-i and a few swollen mitochondria; group B, neurons with the same changes as in A, and in addition a shrunken and incrusted nucleus (Figure 4); and group C, neurons with vacuoles with or without SS-i, and abnormally electron lucent cytoplasm and nucleus (Figure 5). Neurons that typically showed intensified and homogeneously dispersed SS-i staining by LM (Figure 2), showed group A or group B changes by EM. Forty hours after ischaemia, the majority of the dentate hilar SS-i neurons had ultrastructural characteristics corresponding to group B and group C.
Dark neurons A small number of neurons in some of the semithin sections from both normal and ischaemic rats had an appearance similar to dark neurons (Cammermeyer, 1961; Stensaas, Edwards & Stensaas, 1972; Friedrich & Mugnaini, 1981). They displayed an intensely stained nucleus and cytoplasm, but without cytoplasmic vacuolation. Most of these cells were confined to the transition between the dentate granule cell layer and the hilus, in particular along the infrapyramidal part of the granule cell layer. By EM, some dark neurons showed electron dense cytoplasm with the Golgi apparatus standing out as a pale area, whereas other dark neurons in addition displayed a narrow gap between the nucleus and the cytoplasm (Cammermeyer, 1961; Stensaas er al., 1972).
152
F. F. Johansen et al.
Figure 5. Electron micrographs of a SSi neuron with a proximal dendrite in the dentate hilus 40 h after ischaemia. Framed area in a is enlarged in b. The electron lucent cytoplasm is filled with vacuoles which often are confluent and extend into the dendrite. Immunostaining is associated with the vacuoles (arrowheads). The nucleoplasm contains an intranuclear rod (arrow). Bar 5 pm (a), 1 pn (b).
Colocalization of retrogradely transported Fluoro-Gold and somatostatin immunofluorescence in dentate hilar neurons after ischaemia
In the dorsal fascia dentata contralateral to the Fluoro-Gold injection, the hilus contained an average of 17.4 k.3.4 neurons per section with homogeneously dispersed SS-immunofluorescence and 15.7f 1.5 neurons per section with Fluoro-Gold fluorescence. The total cell counts are listed in Table 1. Colocalization of Fluoro-Gold and homogeneously dispersed SSimmunofluorescence was on average found in 3.9 f2.5 neurons per section (Table 1). In the section from the dentate hilus of a rat 40 h after ischaemia shown in Figure 6a, at least five
Ultrastructure of somatostatin neurons after ischaemia
153
Figure 6. Section of dentate hilus from rat with Fluoro-Gold injected in the contralateral dentate gyrus 5 days prior to ischaemia and killed 40 h after ischaemia, as observed through Fluoro-Gold filter a and the FITC-filter b. In a, at least five neurons are retrogradely labelled displaying white Fluoro-Gold fluorescence (arrowheads). These neurons have commissural projections. In b four neurons are labelled with SS-FITC immunofluorescence (arrowheads), which in the microscope appears as intense green fluorescence. Non-specific and auto-fluorescence appears golden brown through the FITC filter. The arrows in a and b point to a neuron containing both fluorocromes and homogeneously dispersed SS-FITC immunoreactivity. Bar 40 pm.
retrogradely labelled, Fluoro-Gold fluorescent and hence commissurally projecting, neurons (arrowheads) were seen. In the same section photographed through the FITC-filter (Figure 6b) four neurons displayed S S immunofluorescence (arrowheads). One of these neurons (marked by arrow) displayed both homogeneously dispersed S S immunofluorescence and Fluoro-Gold fluorescence.
DISCUSSION Methodological considerations and dark neurons The immunostaining was performed largely in accordance with previous light microscopical procedures (Sternberger, 1979; Zimmer & Sunde, 1984; Johansen et al., 1987), after which the reacted tissue was embedded in Epon and processed for EM analysis according to routine procedures. We did not block the staining for non-specific endogenous peroxidase, but the absence of peroxidase staining in the sections incubated without the primary SS antibody clearly demonstrated that the presence of endogenous peroxidase did not present a problem. The absence of peroxidase staining of dark neurons also demonstrated that non-specific immunostaining could not be induced by this reaction.
154
F. F. Johansen et al.
The dark neurons were most likely induced by the dissection of the tissue and the vibratomesectioning after only 2 h of post-fixation, just as it might have been worsened by the subsequent freezing of the sections in liquid nitrogen (Cammermeyer, 1961; Stensaas et al., 1972; Friedrich & Mugnaini, 1981). The presence of the dark neurons, however, did not cause any problem with regard to the interpretation of the SS-i cells, either in the normal or in the ischaemic hippocampus. Nevertheless, we could not exclude the possibility that dark neuron reaction might also be induced in neurons with ischaemic damage. One example might be the occasional neurons which displayed both ischaemia-related microvacuolation and cell shrinkage. Somatostatin immunoreactive neurons in the normal and ischaernic dentate hilus
The major finding of our study was that hilar neurons in the dorsal fascia dentata, which expressed SS-i at both the LM and EM level 40 h after ischaemia, also displayed ischaemic cell changes at the ultrastructural level which are not compatible with survival of the neurons (Brown & Brierley, 1972; Brierley, 1984). Our post-ischaemic ultrastructural observations of SS-i neurons in group A correspond to neurons with ‘simple ischaemic cell changes’ (Brierley, 1984),in group B to neurons with ‘ischaemiccell change with incrustations’ (Brierley, 1984), and in group C to neurons with ‘homogenizing cell changes’ (Brierley, 1984). In hilar neurons of the normal rat, the SS-i was primarily associated with the endoplasmic reticulum. The increased and most often homogeneous distribution of SS-i seen 40 h after ischaemia by LM, and corresponding to group A and B neurons at the ultrastructural level, apparently resulted from dispersed SS-i now located in discrete areas of the dilated endoplasmic reticulum. These SS-i neurons (group A and B) in our EM study were likely to precede the ischaemic cell changes known as acidophilia/eosinophilia, because the latter have been relatsd to the changes seen in group C in our EM study (Brierley, 1984). All the SS-i neurons in the ischaemic animals analysed by EM showed ultrastructural changes of irreversible ‘ischaemic cell damage’ (Brierley, 1984). Both in time and appearance, the changes in the SS-i neurons could accordingly be related to the development of ischaemic cell damage. Four days after ischaemia, approximately 80% of the SS-i cells in the dentate hilus were absent (Johansen et al., 1987). This absence correlated with the post-ischaemic neuronal damage in the hilar area both with regard to chronology and subregional distribution (Johansen et al., 1989; Johansen & O’Hare, 1989).Thus, there is evidence that the majority of the dentate hilar SS-i cells die rapidly after ischaemia in contrast to the delayed pyramidal cell death previously described in the hippocampus (Kirino, Tamura & Sano, 1984). The population of hilar SS-i neurons comprises at least two different cell types (Milner & Bacon, 1989). From their ultrastructure the SS-i neurons examined in the present study resembled the sparsely spined fusiform hilar neurons (Ribak & Seress, 1988; Milner & Bacon, 1989). The reason that we did not identify spines on the SS-i hilar cells might be because the tissue preparation for immunostaining blurred the necessary details. In control and ischaemic animals, we observed that the SS-i primarily was associated with the endoplasmic reticulum, but in some of the hilar neurons, it was also associated with the Golgi apparatus. This intracellular distribution corresponded with both of the two types of hilar SS-i neurons described by Milner and Bacon (1989). Our observations, however, did not add any evidence as to whether these two cell types differ in their susceptibility to ischaemia. The dentate hilar SS-i neurons show some morphological features in common with neurons that contain y-arninobutyric acid (GABA), i.e. intranuclear rods, nuclear infoldings, and
Ultrastructure of somatostatin neurons after ischaemia
155
symmetric SS-i synapses (Leranth et af., 1990). Somogyi et a f . (1984) showed, immunocytochemically,the coexistence of GAD and SS in some hippocampal neurons. Sloviter and Nilaver (1987), on the other hand, suggested that GABA and SS are located in two separate neuronal populations in the rat dentate hilus. They based this suggestion upon morphological characteristics and divergence of the subregional distribution of hilar neurons immunoreactive for GABA and SS, respectively. Johansen et af. (1989) found increased levels of GAD in dentate hilar neurons 4 days after ischaemia, when the majority of hilar SS-i cells had disappeared (Johansen et af., 1987). This supports the view that two separate cell populations exist, maybe with some minor overlap (Somogyi et af., 1984; Kosaka, Wu & Benoit, 1988). To prove that a small number of SS-i neurons in dentate hilus that are vulnerable to ischaemia also contain GABA, it will be necessary to perform a proper immunocytochemical double-labelling study to demonstrate ischaemic cell changes in double-labelled neurons by EM. Somatostatin immunoreactive neurons with commissural projections in the ischaemic dentate hilus Commissural projections from hilar neurons have been demonstrated which have ultrastructural characteristics corresponding to the present description (Seroogy, Seress & Ribak, 1983). Subsequently, the same cell type, defined by its ultrastructure, was also found to be SS-i (Leranth & Frotscher, 1987).This correlated with earlier light microscopical observations with combined SS-immunofluorescence and retrograde axonal tracing (Zimmer et al., 1983). In accordance with these previous studies, we now find that an average of 13% of the hilar neurons in the dorsal fascia dentata which displayed homogeneously dispersed ischaemia-induced SSimmunofluorescence and/or contained retrogradely transported Fluoro-Gold, in fact contained both fluorescence markers. Accordingly, at least some of the SS-i neurons susceptible to ischaemia must have commissural projections. Regarding the actual number of cells, several factors could lead to an underestimate of the number of ischaemia-susceptible SS-i neurons with contralateral projections. The unilateral injection of Fluoro-Gold resulted in a dense and extensive labelling of hilar neurons in the contralateral dentate gyrus. Because we analysed sections from a dorsal level corresponding to the septo-temporal level of the contralateral injection, it is likely that the injection would label the majority of commissurally-projecting neurons present here. However, at the post-ischaemic survival time used in our study, only a part of the vulnerable hilar SS-i neurons displayed homogeneously dispersed SS-immunofluorescence.Complex cumulative observations would, therefore, have to be performed to give the total number. The electrophysiological significance of the commissural and associational projections from dentate hilar SS-i neurons on dentate granule cell inhibition should be awaited before pathophysiological consequences are drawn from an ischaemic loss of these hilar commissural and associational projections. Thus, it has been demonstrated that granule cell inhibition in the rat is enhanced after ischaemia (Chang, Steward & Kassel, 1990) which argues against the idea (Johansen et al. 1987) that loss of SS-i neurons should lead to disinhibition in the dentate gyrus. CONCLUSION We conclude that the loss of neuronal SS-i in the rat dentate hilus observed within the first 2 days after ischaemia, is primarily due to rapid neuronal death of the SS-i hilar neurons. Some of the SS-i neurons in the hilus of the septodorsal fascia dentata which are susceptible to ischaemia have, moreover, been shown to project commissurally.
156
F. F. Johansen et al.
ACKNOWLEDGEMENTS The technical help of P. Kjaxgaard Merller and Lisbeth Thatt Jensen is gratefully acknowledged. The study was supported by the Danish Medical Research Council, the Danish State Biotechnology Programme 1987-1990, Aarhus University Research Foundation, the NOVO Foundation, the Carlsberg Research Foundation and the Lundbeck Foundation.
REFERENCES Bakst I., Avendano C., Morrison J.H. & Amaral D.G. (1986) An experimental analysis of the origins of somatostatinlike immunoreactivity in the dentate gyrus of the rat. Journal of Neuroscience ( U S A ) 6,1452-1462 Brierley J.B. (1984) Cerebral hypoxia. In Greenfield’s Neuropathology, Eds. J.H. Adams, J.A.N. Corsellis & L.W. Duchen, pp. 43-85. Edward Arnold, London Brown A.W. & Brierley J.B. (1972) Anoxic-ischaemic cell change in the rat brain light microscopic and fine structural observations. Journal of the Neurological Sciences 16,59-84 Cammermeyer J. (1961) The importance of avoiding ‘dark’ neurons in experimental neuropathology. Acfa Neuropathologica (Berl) 1,245-270 Chang H.S., Steward 0.& Kassel N.F. (1989) Decreases in excitatory transmission and increases in recurrent inhibition in the rat dentate gyrus after transient cerebral ischaemia. Brain Research 505,220-224 Friedrich Jr V.L. & Mugnaini E. (1981) Electron microscopy. Preparation of neural tissues for electron microscopy. In Neuroanatomical Tract-Tracing Method, Eds L. Heimer & M.J. Robards, pp. 345-375. Plenum Press, New York Johansen F.F., Lin C-T., Schousboe A. & Wu J-Y. (1989) Immunocytochemical investigation of L-glutamic acid decarboxylase in the rat hippocampal formation: the influence of transient cerebral ischaemia. The Journal of Comparative Neurology 281,40-53 Johansen F.F. & O’Hare M.M.T. (1989) Loss of neuropeptide Y immunoreactivity in the rat hippocampus following transient cerebral ischaemia. Journal of Neurosurgical Anesthesiology 1,339-345 Johansen F.F., Tsnder N., Zimmer J., BaimbridgeK.G. & Diemer N.H. (1990) Short-term changes of parvalbumin and calbindin immunoreactivity in the rat hippocampus following cerebral ischaemia. Neuroscience Letters 120, 171-1 74 Johansen F.F., Zimmer J. & Diemer N.H. (1987) Early loss of somatostatin neurons in dentate hilus after cerebral ischaemia in the rat precedes CA-I pyramidal cell loss. Acra Neuropathologica (Berl) 73,110-1 14 Kinno T., Tamura A. & Sano K. (1984) Delayed neuronal death in the rat hippocampus following transient forebrain ischaemia. Acta Neuropafhologica (Berl) 64,139-147 Kosaka T., Wu J-Y. & Benoit R. (1988) GABAergic neurons containing somatostatin-like immunoreactivity in the rat hippocampus and dentate gyrus. Experimental Brain Research 71,388-398 Laurberg S. & Ssrensen K.E. (1981) Associational and commissural collaterals of neurons in the hippocampal formation (hilus fasciae dentatae and subfield CA3). Brain Research 212,287-300 Leranth C. & Frotscher M. (1987) Cholinergic innervation of hippocampal GAD and somatostatin-immunoreactive commissural neurons. The Journal of Comparative Neurology 261,33-47 Leranth C., Malcolm A.J. & Frotscher M. (1990) Merent and efferent synaptic connections of somatostatinhimunoreactive neurons in the rat fascia dentata. The Journal of Comparative Neurology 295,111-122 Milner T.A. &Bacon C.E. (1989) Ultrastructural localization of somatostatin-like immunoreactivity in the rat dentate gyms. The Journal of Comparative Neurology 290,544-560 Pellegrho L.J., Pellegrino AS. & Cushman A.J. (1979) A SfereotaxicAtlas of the Raf Brain. Plenum Press, New York Pulsinelli W.A. & Brierley J.B. (1979) A new model of bilateral hemispheric ischaemia in the unanesthetized rat. Stroke 10,267-272 Rib& C.E. & Seress L. (1988) A Golgi-electron microscopic study of fusiform neurons in the hilar region of the dentate gyrus. The Journalof Comparative Neurology 271,67-78 Seroogy K.B., Seress L. & Ribak C.E. (1983) Ultrastructure ofcommissural neurons of the hilar region in the hippocampal dentate gym.Experimental Neurology 82,594408 Sloviter R.S. (1987) Decreased hippocampal inhibition and a selective loss of interneurons in experimental epilepsy. Science U5,73-76
Ultrastructure of somatostatin neurons after ischaemia
157
Sloviter R.S. & Nilaver G. (1987) Immunocytochemical localization of GABA-, cholecystokinin-, vasoactive intestinal polypeptide-, and somatostatin-like immunoreactivity in the area dentata and hippocampus of the rat. The Journal of Comparative Neurology 256,4240 Somogyi P., Hodgson A.J., Smithy A.D., Nunzi M.G., Gorio A. & Wu J-Y. (1984) Different populations ofGABAergic neurons in the visual cortex and hippocampus of cat contain somatostatin or cholecystokinin-immunoreactive material. Journal of Neuroscience 4,2590-2603 Stensaas S.S., Edwards C.Q. & Stensaas L.J. (1972) An experimental study of hypercromatic nerve cells in the cerebral cortex. Experimental Neurology 36,472487 Sternberger L.A. (1979) Immunocytochemisfry(2nd edition). John Wiley, New York Swanson L.W., Sawchenko P.E. & Cowan W.M. (1981) Evidence for collateral projections by neurons in Ammon’s horn, the dentate gyrus, and the subiculum: a multiple retrograde labelling study in the rat. Journal of Neuroscience 1,548-559 Tender N., Johansen F.F., Frederickson C.J., Zimmer J. & Diemer N.H. (1990) Possible role of zinc in the selective degeneration of dentate hilar neurons after cerebral ischaemia in the adult rat. Neuroscience Letters 109,247-252 Zimmer J., Laurberg S. & Sunde N. (1983) Neuroanatomical aspects ofnormal and transplanted tissue. In Neurobiology of the Hippocampus, Ed. W. Seifert, pp. 39-64. Academic Press, London Zimmer J. & Sunde N. (1984) Neuropeptides and astroglia in intracerebral hippocampal transplants: an immunohistochemical study in the rat. The Journal of Comparative Neurology 227,33 1-347
Received 3 May 1991 Accepted 30 August 1991
Ultrastructure of neurons containing somatostatin in the dentate hilus of the rat hippocampus after cerebral ischaemia, and a note on their commissural connections F. F. JOHANSEN*, T. SPIRENSEN?, N. TPINDER?, J. Z I M M E R t A N D N. H. DIEMER* PharmaBiotec Research Centre, *Cerebral Ischaemia Research Group, Institute of Neuropathology, University of Copenhagen and ?Institute of Neurobiology, University of Aarhus. Denmark
JOHANSEN F. F., S0RENSEN T., T0NDER N., ZIMMER J. & DIEMER N. H.(1992) Neuropathology and Applied Neurobiology 18,145-1 57 Ultrastructureof neurons containing somatostatin in the dentate hilus of the rat hippocampusafter cerebral ischaemia, and a note on their commissural connections
In a light microscopical study, we previously showed that more than 80% of somatostatin (SS) immunoreactive(4) neurons in the hilus of the dorsal part of the rat dentate gyrus are lost 4 days after ischaemia. In order to verify that the loss of SS immunostainingis due to an actual loss of the SS-i neurons and not merely a loss in expression of S S immunoreactivity, we have now performed an ultrastuctural study of these neurons before and 40 h after 20 min of global cerebral ischaemia in adult rats. The normal SS-i neurons were multipolar and fusiform in shape. The SS-i product was associated with the endoplasmic reticulum and occasionally the Golgi apparatus. The cell nuclei had indentations of the nucleolemma and contained intranuclear rods. After ischaemia, many SS-i neurons in the dentate hilus showed increased electron density of both the cell nucleus and the cytoplasm. In addition the cytoplasm was heavily vacuolated with the SS-i associated with some of these vacuoles. Other SS-i neurons had, in addition to the vacuoles a more homogeneous, and abnormal electron lucent nucleus and cytoplasm. These ultrastructural changes correspond to previously reported irreversible, ischaemic cell changes of neurons. Based on this we conclude that the S S immunoreactivity in the dentate hilus of the dorsal hippocampus is lost after ischaemia because of neuronal necrosis. As a minor part of this study, we examined whether the ischaemia-susceptible SS-i neurons in dentate hilus had commissural'axonal projections. This was done utilizing double fluorescencemicroscopy of retrograde axonal transport of the fluorescent dye, Fluoro-Gold, and the observation that vulnerable SS-i neurons display homogeneously dispersed immunostaining 40 h after ischaemia. Fluoro-Gold was injected unilaterally into the dorsal dentate gyrus 5 days prior to ischaemia. Then, 40 h after ischaemia, sections were stained for SS immunofluorescence, and examined, in the dentate hilus contralateral to the injection, for neuronal co-localization of both events. Cell counts revealed double-labelling of 13% of all neurons which displayed one of the events. This observation suggests that at least some of the ischaemia-susceptible SS-i neurons in dentate hilus do project commissurally. The pathophysiological significanceof ischaemic loss of commissurally projecting SS-i neurons in dentate hilus remains to be determined. Correspondence to: Dr F. F. Johansen, Institute of Neuropathology, Frederik V's vej 11, DK-2100, Copenhagen 0, Denmark.
146
F. F. Johansen et al.
Keywords: electron microscopy, immunocytochemistry, somatostatin, ischaemic cell changes, retrograde axonal tracing, hippocampus, rat
INTRODUCTION We have previously reported the disappearance of somatostatin ( S S ) immunoreactive (4) neurons in the dentate hilus (CA4) of the dosal hippocampus from the 2nd to the 4th day after transient cerebral ischaemia in the rat (Johansen, Zimmer & Diemer, 1987). In the neuronal cell bodies, the SS-i is normally arranged as whorled bands or clusters of immunopositive particles when observed in the light microscope (LM). During the first 24-48 h after ischaemia, the immunoreactivity is increased and becomes homogeneously dispersed in the cell bodies (Johansen et al., 1987). Thereafter about 80% of the immunostained cells in dentate hilus disappear. This loss of immunocytochemical reactivity coincides with an actual loss of neurons in the dentate hilus (Johansen, Lin & Schousboe, 1989;Johansen & O’Hare, 1989;Ternder et al., 1990). Since we, however, in other studies have found that the levels of immunocytochemical staining for glutamic acid decarboxylase (GAD), neuropeptide Y, and parvalbumin, respectively, can increase, disappear permanently, or decrease transiently in hippocampal neurons that otherwise survive an ischaemic insult (Johansen et al., 1989, 1990; Johansen & O’Hare, 1989), we wanted to determine whether the hilar SS-i neurons actually died or merely lost their S S immunoreactivity after ischaemia. For this purpose we decided to examine the hilar SS-i neurons 40 h after ischaemia in the electron microscope (EM), and correlate the observed changes with the ischaemic cell changes described by others (Brown & Brierley, 1972). Several authors have described commissural and ipsilateral hilodenate projections from dentate hilar neurons, and some of these cells have been identified as SSi, where at least the ipsilateral projection goes to the outer part of the dentate molecular layer (Zimmer, Laurberg & Sunde, 1983; Leranth & Frotscher, 1987; Bakst et al., 1986; Leranth, Malcolm & Frotscher, 1990). The main commissural and associational projection terminates in the inner part of the dentate molecular layer and is unlikely to be somatostatinergic (Laurberg & Serrensen, 1981; Swanson, Sawchenko & Cowan, 1981; Bakst et al., 1986; Leranth et al., 1990). Since it is possible that associational and commissural projections from SS-i neurons play a role in the inhibition of the granule cells (Sloviter, 1987),a loss of these projections may influence the postischaemic pathophysiology. Therefore, in order to determine whether the ischaemia-susceptible SS-i cells in the dentate hilus do project commissuraily, we combined the use of retrograde axonal transport of the fluorescencedye, Fluoro-Gold, with immunofluorescent staining for S S at 40 h after ischaemia. We examined the co-localization of Fluoro-Gold and post-ischaemic S S immunofluorescence of the homogeneously distributed type in the hilar neurons (Johansen et al., 1987). MATERIALS A N D METHODS
Induction of ischaernia All experiments were conducted on adult male Wistar rats, weighing 350400 g. Twenty minutes of cerebral ischaemia was induced by means of four-vessel occlusion (hlsinelli & Brierley, 1979) with minor modifications as described in detail by Johansen et al. (1989) and
Ultrastructure of somatostatin neurons after ischaemia
147
Johansen & O'Hare (1989). Briefly, the vertebral arteries were electrocauterized under methohexital anaesthesia (50 mg/kg body weight). The rats were then left without food, but with free access to water, until the following day, when the carotid arteries were gently exposed during brief halothane anaesthesia. Anaesthesia was discontinued for 2 min, before both carotid arteries were ligated for 20 min. After the occlusion period, the rats were allowed to survive for 40 h. Tracing of commissural projection Five days prior to ischaemia, three rats received stereotaxic injections of 1 pl of a 4% solution of Fluoro-Gold (Fluorochrome-Inc., Englewood, Colorado, USA) dissolved in Ringer's solution, into the dentate hilus of the dorsal hippocampus on one side using the following coordinates; 2.6 mm behind the bregma, 2.0 mm lateral to the midline and 3.5 mm below the cortical surface (Pellegrino, Pellegrino & Cushman, 1979).This procedure results in a widespread Fluoro-Gold labelling of dentate hilar neurons along the longitudinal septotemporal axis on the injected and the contralateral side corresponding to the ipsilateral and commissural projections from these cells (Laurberg Kc Serrensen, 1981). Forty hours after ischaemia hippocampal tissues were processed for immunofluorescencestaining for SS.
EM immunocytochemistry Eight adults rats subjected to ischaemia and six normal rats were perfused transcardially, during pentobarbitone anaesthesia (50 mg/kg body weight), with 4% paraformaldehyde, 0.05% glutaraldehyde and 0.2% saturated picric acid in 0.1 M phosphate buffer (PB), pH 7.4. The brains were gently removed from the skulls and stored at 4°C for 2 h in 4% paraformaldehyde and 0.2% picric acid in PB. The hippocampal region was then exposed, and the dorsal half cut perpendicular to the longitudinal axis in 50 pm thick sections on an Oxford vibratome in PB. To enhance antibody penetration, the sections were soaked in a 15% sucrose solution in PB, frozen in liquid nitrogen, and thawed at 4°C for 2 h. Sections were thereafter rinsed in TBS (Tris buffered saline) and processed for immunostaining (Sternberger, 1979). First the sections were pre-incubated in 1% swine serum (X901, Dakopatts) in TBS for 30 min and then incubated with the primary SS antibody (1 :2400, A566, Dakopatts) in TBS overnight at 4"C, rinsed in TBS and incubated with the secondary swine-anti-rabbit antibody ( ~ 3 0 2196, , Dakopatts) in 0.25% BSA (bovine serum albumin) in TBS for 30min at room temperature, rinsed in TBS and incubated in PAP (1:75,Z113, Dakopatts) in 0.25% BSA in TBS for 30 min, rinsed in TBS, then exposed to diaminobenizidine (DAB), finally rinsed once in TBS and twice in PB. The primary antibody was diluted in TBS with 0.25% BSA and 0.01% Triton XlOO which did not alter the preservation of the ultrastructure of the tissue. After the immunostaining and post-fixation in 1 % OsO, in PB for 1 h, the 50 pm vibratome sections were dehydrated and flat-embedded in Epon between two pieces of plastic foil. Selected sections with SS-i cells were then re-embedded on top of empty Epon blocks. From these, semithin 3 pm thick sections were cut and mounted on glass slides and every third section was stained with toluidine blue. Semithin sections containing SS-i cells were identified by LM, re-embedded on top of new Epon blocks, trimmed and cut in thin sections (50-60 nm). The thin sections were mounted on slot grids coated with formvar film (three grids) and on hexagonal 400 mesh slim bar grids (SPI supplies, 2240c) (two grids) and examined in a Jeol-100s electron microscope. From each rat, 10 semithin sections were selected as described, and from each of these sections, 2 4 SS-i cells were analysed by EM. Exclusion of the primary SS antibody was used as a negative control.
148
F. F. Johansen et al.
Table 1. Co-localization of somatostatin and Fluoro-Gold in ischaemic hilar neurons. Differential cell counts from the dentate hilus of the dorsal fascia dentata in rats, which 5 days prior to ischaemia had Fluoro-Gold injected into the contralateral hilus, and then 40 h after ischaemia were killed and processed for combined Fluoro-Gold and SS-FITC immunofluorescence. Counts of neurons with SS-FITC immunofluorescence (neurons containing somatostatin) refers selectively to ischaemically damaged neurons with homogeneously dispersed SS-FITC immunofluorescence
Numbers Percentage of total Mean kSD per hilus section
Animals
Sections
Somatostatin
FIuoroGold
Co-localization
Total
3
54
93 1 59%
843 54%
207 13%
IS67
17.4k 3.4
15.7+ 1.5
3.9k2.5
100%
Other vibratome sections were processed for electron microscopy without further immunocytochemicaltreatment. Somatostatin immunofluorescencestaining and fluorescence microscopy The three rats used for combined retrograde tracing and immunofluorescence for SS were perfused transcardially with 4% paraformaldehyde in PB. After removal from the skull, the brains were stored for 4 h in 4% paraformaldehyde in PB at 4°C before they were cut on a vibratome in 50 pm thick coronal sections. These were pre-incubated in 1% normal swine serum (X901, Dakopatts) in PB for 30 min and incubated with 1:2400 SS antibody in PB (Dakopatts, A566) overnight at 4"C, rinsed in PB and finally incubated for 1 h with 2.5% swine anti-rabbit immunoglobulin conjugated with fluorescein-isothiocyanate isomer 1 (SWAR-FITC F205, Dakopatts). Before LM examination, sections were mounted and coverslipped in Entellan@. For the analysis and photography, we used a Zeiss fluorescence microscope with the BP 450490, LP 520 and FT 570 filters for Fluoro-Gold, and LP 397 and FT 395 filters for FITC. Cell counts were performed at 400 times magnification in the dentate hilus of the dorsal dentate gyrus contralateral to the Fluoro-Gold injection. All neurons which showed Fluoro-Gold fluorescence, or homogeneously dispersed SS-FITC fluorescence, or both of these, were counted (Table 1). The dentate hilus was defined and demarcated as previously described (Johansen et al., 1989;Johansen & O'Hare, 1989). RESULTS Somatostatin immunoreactive neurons in the normal dentate hilus Many SS-i neurons were found scattered throughout the entire dentate hilus in both the 50 pm thick vibratome sections and the 3 pm thick semithin sections cut from these. The immunocytochemical staining was in the form of whorled bands and particles within the cell body, stretching into the proximal dendrites. When examined by EM, there were often 2-5 SS-i cell bodies present in the same ultrathin section. The SS-i cells all showed well-developed rough endoplasmic reticulum and Golgi apparatus. The immunostaining was localized to discrete areas of the rough endoplasmic reticulum (Figure 1) and, in a few cases, the inner sac of the Golgi apparatus. The nucleus was large and dominated by euchromatin with a prominent nucleolus and many nuclei contained an intranuclear rod (Figure 1). The nucleolemma had an irregular contour and sometimes deep infoldings,. but these never extended more than half
Ultrastructure of somatostatin neurons after ischaemia
149
Figure 1. SS-i neurons in the dentate hilus of control rat. The framed area outlined in a is shown enlarged in a neighbouring section in b. Immunostaining is confined to parts of the rough endoplasmic reticulum. The nucleus contains an intranuclear rod (arrowhead). (*) freezing artefact. Bar 5 pm (a), 1 pm (b).
way through the nucleus. The dendrites of the SS-i neurons appeared spineless and received asymmetrical synaptic contacts. Further details regarding the structure of the pre-synaptic terminals and the dendrites, were difficult to define, due to deterioration of the tissue caused by the immunocytochemical processing.
Somatostatin immunoreactive neurons in the dentate hilus after ischaemia Forty hours after the ischaemia there were many SS-i neurons present in the dentate hilus. By light microscopy, the majority of the neurons did, however, differ from the neurons in the normal rat. Some neurons had an almost normal, but irregularly distributed and abnormally intensified SS-i staining (Figure 2), while others displayed a very intense and homogeneous staining of the cell bodies extending into the dendrites (Figure 2). In semithin sections, some
150
F. F.Johansen et al.
Figure 2. Vibratome section from dentate hilus 40 h after ischaemia. Examples of the homogeneous (thick arrows) and increased (thin arrow) SS-immunoreactivity of cell bodies are shown. Bar 20 pm. Figure 3. SS-i neurons in a semithin section from the dentate hilus 40 h after ischaemia. The intracellular immunostaining appears in clumps, often associated with clear vesicles. g, granule cell layer. Bar 20 pm.
neurons in the dentate hilus had vacuoles in the cytoplasm and an increased toluidine blue staining of both the cytoplasm and the nucleus. In some vacuoles, SS-i staining could be discerned at the circumference of the vacuoles (Figure 3). By electron microscopy, the cytoplasm showed pronounced vacuolation (Figures 4 and 5), and the appearances corresponding to those described by Brown and Brierley (1972). In agreement with this, the vacuoles were found to consist primarily of dilatations of the rough endoplasmic reticulum. The SS-i staining was located in association with some of the vacuoles (Figures 4 and 5). Sometimes double-walled membranous material appeared in vacuoles without any immunostaining, indicating that they were swollen mitochrondria (Brown & Brierley, 1972; Brierley, 1984). Other mitochondria displayed intact cristae and had normal dimensions, but showed increased electron density of the matrix. In some of the vacuolated SS-i cells, the cytoplasm was dark (Figure 4). The increased density of the cytoplasm and the vacuolation also extended into the dendrites. The cell nuclei were dark and shrunken (Figure 4) with a homogeneous nucleoplasm in which nuclear inclusion bodies occasionally were recognizable. In other neurons with vacuoles, the cytoplasm was electron lucent and the nucleus swollen (Figure 5). Occasionally, cells with disintegrating cytoplasm were associated with reactive, swollen astrocytes.
Ultrastructure of somatostatin neurons a f e r ischaemia
151
Figure 4. Electron micrographs of SS-i neuron in dentate hilus 40 h after ischaemia. Framed area in (a) is enlarged in (b). The electron dense cytoplasm is filled with vacuoles and the immunoreaction product is associated with some of these vacuoles (arrows). T h e nucleus (N)is dark, shrunken and irregular in shape. Bar 3 pm (a), 1 pm @).
In general, the post-ischaemic ultrastructural appearances of the SS-i neurons can be summarized as follows: group A , neurons with electron dense cytoplasm, vacuoles with or without SS-i and a few swollen mitochondria; group B, neurons with the same changes as in A, and in addition a shrunken and incrusted nucleus (Figure 4); and group C, neurons with vacuoles with or without SS-i, and abnormally electron lucent cytoplasm and nucleus (Figure 5). Neurons that typically showed intensified and homogeneously dispersed SS-i staining by LM (Figure 2), showed group A or group B changes by EM. Forty hours after ischaemia, the majority of the dentate hilar SS-i neurons had ultrastructural characteristics corresponding to group B and group C.
Dark neurons A small number of neurons in some of the semithin sections from both normal and ischaemic rats had an appearance similar to dark neurons (Cammermeyer, 1961; Stensaas, Edwards & Stensaas, 1972; Friedrich & Mugnaini, 1981). They displayed an intensely stained nucleus and cytoplasm, but without cytoplasmic vacuolation. Most of these cells were confined to the transition between the dentate granule cell layer and the hilus, in particular along the infrapyramidal part of the granule cell layer. By EM, some dark neurons showed electron dense cytoplasm with the Golgi apparatus standing out as a pale area, whereas other dark neurons in addition displayed a narrow gap between the nucleus and the cytoplasm (Cammermeyer, 1961; Stensaas er al., 1972).
152
F. F. Johansen et al.
Figure 5. Electron micrographs of a SSi neuron with a proximal dendrite in the dentate hilus 40 h after ischaemia. Framed area in a is enlarged in b. The electron lucent cytoplasm is filled with vacuoles which often are confluent and extend into the dendrite. Immunostaining is associated with the vacuoles (arrowheads). The nucleoplasm contains an intranuclear rod (arrow). Bar 5 pm (a), 1 pn (b).
Colocalization of retrogradely transported Fluoro-Gold and somatostatin immunofluorescence in dentate hilar neurons after ischaemia
In the dorsal fascia dentata contralateral to the Fluoro-Gold injection, the hilus contained an average of 17.4 k.3.4 neurons per section with homogeneously dispersed SS-immunofluorescence and 15.7f 1.5 neurons per section with Fluoro-Gold fluorescence. The total cell counts are listed in Table 1. Colocalization of Fluoro-Gold and homogeneously dispersed SSimmunofluorescence was on average found in 3.9 f2.5 neurons per section (Table 1). In the section from the dentate hilus of a rat 40 h after ischaemia shown in Figure 6a, at least five
Ultrastructure of somatostatin neurons after ischaemia
153
Figure 6. Section of dentate hilus from rat with Fluoro-Gold injected in the contralateral dentate gyrus 5 days prior to ischaemia and killed 40 h after ischaemia, as observed through Fluoro-Gold filter a and the FITC-filter b. In a, at least five neurons are retrogradely labelled displaying white Fluoro-Gold fluorescence (arrowheads). These neurons have commissural projections. In b four neurons are labelled with SS-FITC immunofluorescence (arrowheads), which in the microscope appears as intense green fluorescence. Non-specific and auto-fluorescence appears golden brown through the FITC filter. The arrows in a and b point to a neuron containing both fluorocromes and homogeneously dispersed SS-FITC immunoreactivity. Bar 40 pm.
retrogradely labelled, Fluoro-Gold fluorescent and hence commissurally projecting, neurons (arrowheads) were seen. In the same section photographed through the FITC-filter (Figure 6b) four neurons displayed S S immunofluorescence (arrowheads). One of these neurons (marked by arrow) displayed both homogeneously dispersed S S immunofluorescence and Fluoro-Gold fluorescence.
DISCUSSION Methodological considerations and dark neurons The immunostaining was performed largely in accordance with previous light microscopical procedures (Sternberger, 1979; Zimmer & Sunde, 1984; Johansen et al., 1987), after which the reacted tissue was embedded in Epon and processed for EM analysis according to routine procedures. We did not block the staining for non-specific endogenous peroxidase, but the absence of peroxidase staining in the sections incubated without the primary SS antibody clearly demonstrated that the presence of endogenous peroxidase did not present a problem. The absence of peroxidase staining of dark neurons also demonstrated that non-specific immunostaining could not be induced by this reaction.
154
F. F. Johansen et al.
The dark neurons were most likely induced by the dissection of the tissue and the vibratomesectioning after only 2 h of post-fixation, just as it might have been worsened by the subsequent freezing of the sections in liquid nitrogen (Cammermeyer, 1961; Stensaas et al., 1972; Friedrich & Mugnaini, 1981). The presence of the dark neurons, however, did not cause any problem with regard to the interpretation of the SS-i cells, either in the normal or in the ischaemic hippocampus. Nevertheless, we could not exclude the possibility that dark neuron reaction might also be induced in neurons with ischaemic damage. One example might be the occasional neurons which displayed both ischaemia-related microvacuolation and cell shrinkage. Somatostatin immunoreactive neurons in the normal and ischaernic dentate hilus
The major finding of our study was that hilar neurons in the dorsal fascia dentata, which expressed SS-i at both the LM and EM level 40 h after ischaemia, also displayed ischaemic cell changes at the ultrastructural level which are not compatible with survival of the neurons (Brown & Brierley, 1972; Brierley, 1984). Our post-ischaemic ultrastructural observations of SS-i neurons in group A correspond to neurons with ‘simple ischaemic cell changes’ (Brierley, 1984),in group B to neurons with ‘ischaemiccell change with incrustations’ (Brierley, 1984), and in group C to neurons with ‘homogenizing cell changes’ (Brierley, 1984). In hilar neurons of the normal rat, the SS-i was primarily associated with the endoplasmic reticulum. The increased and most often homogeneous distribution of SS-i seen 40 h after ischaemia by LM, and corresponding to group A and B neurons at the ultrastructural level, apparently resulted from dispersed SS-i now located in discrete areas of the dilated endoplasmic reticulum. These SS-i neurons (group A and B) in our EM study were likely to precede the ischaemic cell changes known as acidophilia/eosinophilia, because the latter have been relatsd to the changes seen in group C in our EM study (Brierley, 1984). All the SS-i neurons in the ischaemic animals analysed by EM showed ultrastructural changes of irreversible ‘ischaemic cell damage’ (Brierley, 1984). Both in time and appearance, the changes in the SS-i neurons could accordingly be related to the development of ischaemic cell damage. Four days after ischaemia, approximately 80% of the SS-i cells in the dentate hilus were absent (Johansen et al., 1987). This absence correlated with the post-ischaemic neuronal damage in the hilar area both with regard to chronology and subregional distribution (Johansen et al., 1989; Johansen & O’Hare, 1989).Thus, there is evidence that the majority of the dentate hilar SS-i cells die rapidly after ischaemia in contrast to the delayed pyramidal cell death previously described in the hippocampus (Kirino, Tamura & Sano, 1984). The population of hilar SS-i neurons comprises at least two different cell types (Milner & Bacon, 1989). From their ultrastructure the SS-i neurons examined in the present study resembled the sparsely spined fusiform hilar neurons (Ribak & Seress, 1988; Milner & Bacon, 1989). The reason that we did not identify spines on the SS-i hilar cells might be because the tissue preparation for immunostaining blurred the necessary details. In control and ischaemic animals, we observed that the SS-i primarily was associated with the endoplasmic reticulum, but in some of the hilar neurons, it was also associated with the Golgi apparatus. This intracellular distribution corresponded with both of the two types of hilar SS-i neurons described by Milner and Bacon (1989). Our observations, however, did not add any evidence as to whether these two cell types differ in their susceptibility to ischaemia. The dentate hilar SS-i neurons show some morphological features in common with neurons that contain y-arninobutyric acid (GABA), i.e. intranuclear rods, nuclear infoldings, and
Ultrastructure of somatostatin neurons after ischaemia
155
symmetric SS-i synapses (Leranth et af., 1990). Somogyi et a f . (1984) showed, immunocytochemically,the coexistence of GAD and SS in some hippocampal neurons. Sloviter and Nilaver (1987), on the other hand, suggested that GABA and SS are located in two separate neuronal populations in the rat dentate hilus. They based this suggestion upon morphological characteristics and divergence of the subregional distribution of hilar neurons immunoreactive for GABA and SS, respectively. Johansen et af. (1989) found increased levels of GAD in dentate hilar neurons 4 days after ischaemia, when the majority of hilar SS-i cells had disappeared (Johansen et af., 1987). This supports the view that two separate cell populations exist, maybe with some minor overlap (Somogyi et af., 1984; Kosaka, Wu & Benoit, 1988). To prove that a small number of SS-i neurons in dentate hilus that are vulnerable to ischaemia also contain GABA, it will be necessary to perform a proper immunocytochemical double-labelling study to demonstrate ischaemic cell changes in double-labelled neurons by EM. Somatostatin immunoreactive neurons with commissural projections in the ischaemic dentate hilus Commissural projections from hilar neurons have been demonstrated which have ultrastructural characteristics corresponding to the present description (Seroogy, Seress & Ribak, 1983). Subsequently, the same cell type, defined by its ultrastructure, was also found to be SS-i (Leranth & Frotscher, 1987).This correlated with earlier light microscopical observations with combined SS-immunofluorescence and retrograde axonal tracing (Zimmer et al., 1983). In accordance with these previous studies, we now find that an average of 13% of the hilar neurons in the dorsal fascia dentata which displayed homogeneously dispersed ischaemia-induced SSimmunofluorescence and/or contained retrogradely transported Fluoro-Gold, in fact contained both fluorescence markers. Accordingly, at least some of the SS-i neurons susceptible to ischaemia must have commissural projections. Regarding the actual number of cells, several factors could lead to an underestimate of the number of ischaemia-susceptible SS-i neurons with contralateral projections. The unilateral injection of Fluoro-Gold resulted in a dense and extensive labelling of hilar neurons in the contralateral dentate gyrus. Because we analysed sections from a dorsal level corresponding to the septo-temporal level of the contralateral injection, it is likely that the injection would label the majority of commissurally-projecting neurons present here. However, at the post-ischaemic survival time used in our study, only a part of the vulnerable hilar SS-i neurons displayed homogeneously dispersed SS-immunofluorescence.Complex cumulative observations would, therefore, have to be performed to give the total number. The electrophysiological significance of the commissural and associational projections from dentate hilar SS-i neurons on dentate granule cell inhibition should be awaited before pathophysiological consequences are drawn from an ischaemic loss of these hilar commissural and associational projections. Thus, it has been demonstrated that granule cell inhibition in the rat is enhanced after ischaemia (Chang, Steward & Kassel, 1990) which argues against the idea (Johansen et al. 1987) that loss of SS-i neurons should lead to disinhibition in the dentate gyrus. CONCLUSION We conclude that the loss of neuronal SS-i in the rat dentate hilus observed within the first 2 days after ischaemia, is primarily due to rapid neuronal death of the SS-i hilar neurons. Some of the SS-i neurons in the hilus of the septodorsal fascia dentata which are susceptible to ischaemia have, moreover, been shown to project commissurally.
156
F. F. Johansen et al.
ACKNOWLEDGEMENTS The technical help of P. Kjaxgaard Merller and Lisbeth Thatt Jensen is gratefully acknowledged. The study was supported by the Danish Medical Research Council, the Danish State Biotechnology Programme 1987-1990, Aarhus University Research Foundation, the NOVO Foundation, the Carlsberg Research Foundation and the Lundbeck Foundation.
REFERENCES Bakst I., Avendano C., Morrison J.H. & Amaral D.G. (1986) An experimental analysis of the origins of somatostatinlike immunoreactivity in the dentate gyrus of the rat. Journal of Neuroscience ( U S A ) 6,1452-1462 Brierley J.B. (1984) Cerebral hypoxia. In Greenfield’s Neuropathology, Eds. J.H. Adams, J.A.N. Corsellis & L.W. Duchen, pp. 43-85. Edward Arnold, London Brown A.W. & Brierley J.B. (1972) Anoxic-ischaemic cell change in the rat brain light microscopic and fine structural observations. Journal of the Neurological Sciences 16,59-84 Cammermeyer J. (1961) The importance of avoiding ‘dark’ neurons in experimental neuropathology. Acfa Neuropathologica (Berl) 1,245-270 Chang H.S., Steward 0.& Kassel N.F. (1989) Decreases in excitatory transmission and increases in recurrent inhibition in the rat dentate gyrus after transient cerebral ischaemia. Brain Research 505,220-224 Friedrich Jr V.L. & Mugnaini E. (1981) Electron microscopy. Preparation of neural tissues for electron microscopy. In Neuroanatomical Tract-Tracing Method, Eds L. Heimer & M.J. Robards, pp. 345-375. Plenum Press, New York Johansen F.F., Lin C-T., Schousboe A. & Wu J-Y. (1989) Immunocytochemical investigation of L-glutamic acid decarboxylase in the rat hippocampal formation: the influence of transient cerebral ischaemia. The Journal of Comparative Neurology 281,40-53 Johansen F.F. & O’Hare M.M.T. (1989) Loss of neuropeptide Y immunoreactivity in the rat hippocampus following transient cerebral ischaemia. Journal of Neurosurgical Anesthesiology 1,339-345 Johansen F.F., Tsnder N., Zimmer J., BaimbridgeK.G. & Diemer N.H. (1990) Short-term changes of parvalbumin and calbindin immunoreactivity in the rat hippocampus following cerebral ischaemia. Neuroscience Letters 120, 171-1 74 Johansen F.F., Zimmer J. & Diemer N.H. (1987) Early loss of somatostatin neurons in dentate hilus after cerebral ischaemia in the rat precedes CA-I pyramidal cell loss. Acra Neuropathologica (Berl) 73,110-1 14 Kinno T., Tamura A. & Sano K. (1984) Delayed neuronal death in the rat hippocampus following transient forebrain ischaemia. Acta Neuropafhologica (Berl) 64,139-147 Kosaka T., Wu J-Y. & Benoit R. (1988) GABAergic neurons containing somatostatin-like immunoreactivity in the rat hippocampus and dentate gyrus. Experimental Brain Research 71,388-398 Laurberg S. & Ssrensen K.E. (1981) Associational and commissural collaterals of neurons in the hippocampal formation (hilus fasciae dentatae and subfield CA3). Brain Research 212,287-300 Leranth C. & Frotscher M. (1987) Cholinergic innervation of hippocampal GAD and somatostatin-immunoreactive commissural neurons. The Journal of Comparative Neurology 261,33-47 Leranth C., Malcolm A.J. & Frotscher M. (1990) Merent and efferent synaptic connections of somatostatinhimunoreactive neurons in the rat fascia dentata. The Journal of Comparative Neurology 295,111-122 Milner T.A. &Bacon C.E. (1989) Ultrastructural localization of somatostatin-like immunoreactivity in the rat dentate gyms. The Journal of Comparative Neurology 290,544-560 Pellegrho L.J., Pellegrino AS. & Cushman A.J. (1979) A SfereotaxicAtlas of the Raf Brain. Plenum Press, New York Pulsinelli W.A. & Brierley J.B. (1979) A new model of bilateral hemispheric ischaemia in the unanesthetized rat. Stroke 10,267-272 Rib& C.E. & Seress L. (1988) A Golgi-electron microscopic study of fusiform neurons in the hilar region of the dentate gyrus. The Journalof Comparative Neurology 271,67-78 Seroogy K.B., Seress L. & Ribak C.E. (1983) Ultrastructure ofcommissural neurons of the hilar region in the hippocampal dentate gym.Experimental Neurology 82,594408 Sloviter R.S. (1987) Decreased hippocampal inhibition and a selective loss of interneurons in experimental epilepsy. Science U5,73-76
Ultrastructure of somatostatin neurons after ischaemia
157
Sloviter R.S. & Nilaver G. (1987) Immunocytochemical localization of GABA-, cholecystokinin-, vasoactive intestinal polypeptide-, and somatostatin-like immunoreactivity in the area dentata and hippocampus of the rat. The Journal of Comparative Neurology 256,4240 Somogyi P., Hodgson A.J., Smithy A.D., Nunzi M.G., Gorio A. & Wu J-Y. (1984) Different populations ofGABAergic neurons in the visual cortex and hippocampus of cat contain somatostatin or cholecystokinin-immunoreactive material. Journal of Neuroscience 4,2590-2603 Stensaas S.S., Edwards C.Q. & Stensaas L.J. (1972) An experimental study of hypercromatic nerve cells in the cerebral cortex. Experimental Neurology 36,472487 Sternberger L.A. (1979) Immunocytochemisfry(2nd edition). John Wiley, New York Swanson L.W., Sawchenko P.E. & Cowan W.M. (1981) Evidence for collateral projections by neurons in Ammon’s horn, the dentate gyrus, and the subiculum: a multiple retrograde labelling study in the rat. Journal of Neuroscience 1,548-559 Tender N., Johansen F.F., Frederickson C.J., Zimmer J. & Diemer N.H. (1990) Possible role of zinc in the selective degeneration of dentate hilar neurons after cerebral ischaemia in the adult rat. Neuroscience Letters 109,247-252 Zimmer J., Laurberg S. & Sunde N. (1983) Neuroanatomical aspects ofnormal and transplanted tissue. In Neurobiology of the Hippocampus, Ed. W. Seifert, pp. 39-64. Academic Press, London Zimmer J. & Sunde N. (1984) Neuropeptides and astroglia in intracerebral hippocampal transplants: an immunohistochemical study in the rat. The Journal of Comparative Neurology 227,33 1-347
Received 3 May 1991 Accepted 30 August 1991
