Surviving CA1 pyramidal cells receive intact perisomatic inhibitory in
http://www.100md.com
《大脑学杂志》
1 Institute of Experimental Medicine, Hungarian Academy of Science,2 Neurosurgical Department, MaV Hospital Budapest,3 National Institute of Neurosurgery, H-1577
4 National Institute of Psychiatry and Neurology, Epilepsy Center, Budapest, Hungary
Summary
Temporal lobe epilepsy (TLE) is known to be linked to an impaired balance of excitation and inhibition. Whether inhibition is decreased or preserved in the human epileptic hippocampus, beside the excess excitation, is still a debated question. In the present study, quantitative light and electron microscopy has been performed to analyse the distribution, morphology and input–output connections of parvalbumin (PV)-immunopositive interneurons, together with the entire perisomatic input of pyramidal cells, in the human control and epileptic CA1 region. Based on the degree of cell loss, the patients with therapy-resistant TLE formed four pathological groups. In the non-sclerotic CA1 region of TLE patients, where large numbers of pyramidal cells are preserved, the number of PV-immunopositive cell bodies decreased, whereas axon terminal staining, and the distribution of their postsynaptic targets was not altered. The synaptic coverage of CA1 pyramidal cell axon initial segments (AISs) remained unchanged in the epileptic tissue. The somatic inhibitory input is also preserved; it has been decreased only in the cases with patchy pyramidal cell loss in the CA1 region (control, 0.637; epileptic with mild cell loss, 0.642; epileptic with patchy cell loss, 0.424 μm synaptic length/100 μm soma perimeter). The strongly sclerotic epileptic CA1 region, where pyramidal cells can hardly be seen, contains a very small number of PV-immunopositive elements. Our results suggest that perisomatic inhibitory input is preserved in the epileptic CA1 region as long as pyramidal cells are present. Basket and axo-axonic cells survive in epilepsy if their original targets are present, although many of them lose their PV content or PV immunoreactivity. An efficient perisomatic inhibition is likely to take part in the generation of abnormal synchrony in the non-sclerotic epileptic CA1 region, and thus participate in the maintenance of epileptic seizures driven, for example, by hyperactive afferent input.
Key Words: perisomatic inhibition; GABA; basket cell; chandelier cell; selective vulnerability
Abbreviations: AIS = axon initial segment; CA1–3 = regions of the cornu Ammonis; PV = parvalbumin; TLE = temporal lobe epilepsy
Introduction
Epilepsy is thought to be related to a changed balance between excitation and inhibition (Avoli, 1983; Mody et al., 1992). In epilepsy, the CA1 region was shown to receive an enhanced excitatory input from the CA3 region in the rat hippocampus (Buzsaki et al., 1989; Rafiq et al., 1993; Bragin et al., 1997), from the dentate gyrus and/or CA2 region in the human hippocampus (Wittner et al., 2002) and from the sprouting CA1 pyramidal cell axons in rats (Esclapez et al., 1999; Lehmann et al., 2001) and humans (Lehmann et al., 2000). It is still not clear whether inhibition is decreased or preserved in the human epileptic CA1 region. There is evidence for interneuronal cell death (de Lanerolle et al., 1988, 1989; Magloczky et al., 2000), as well as for the preservation of glutamic acid dehydrogenase (GAD)-positive cells (Babb et al., 1989). Since CA1 pyramidal cells provide the cortical output connections of the hippocampus (Ramon y Cajal, 1909–1911; Lorente de No, 1934; Maclean, 1992) and are known to be the most vulnerable in epilepsy (Sommer, 1880; Corsellis, 1955; Falconer, 1968), alterations in the inhibitory network in the CA1 region may be crucial in the pathophysiology of the epileptic temporal lobe.
The calcium-binding protein parvalbumin (PV) is known to be present exclusively in perisomatic inhibitory interneurons in the hippocampus of rodents (Kosaka et al., 1987; Katsumaru et al., 1988b; Holm et al., 1990; Nitsch et al., 1990), monkeys (Seress et al., 1991; Ribak et al., 1993) and humans (Braak et al., 1991; Seress et al., 1993). Two perisomatic inhibitory cell types, basket and chandelier or axo-axonic cells, were demonstrated to be PV-immunopositive in all examined species, including humans (Seress et al., 1993; for a review see Freund and Buzsaki, 1996) and to control the output of principal cells (Miles et al., 1996). The survival of PV-immunopositive interneurons in epilepsy, as well as their role in the pathophysiological mechanisms of seizures and cell death are still controversial questions. Some authors have described the preservation (Sloviter et al., 1991; Arellano et al., 2003) and others the vulnerability (Zhu et al., 1997; Wittner et al., 2001) of these cells in the human epileptic hippocampus, and a decrease in the number of PV-immunopositive interneurons has been found in the epileptic neocortex (DeFelipe et al., 1993; Marco et al., 1997; DeFelipe, 1999). PV immunoreactivity is very sensitive to epileptic and ischaemic conditions and might disappear from neurons that are otherwise preserved (Johansen et al., 1990; Tortosa and Ferrer, 1993; Bazzett et al., 1994; Scotti et al., 1997a, b). Since PV immunostaining probably reveals only part of the originally PV-immunopositive cells, which in turn represent only one of the at least two basket cell types (the other being the cholecystokinin-containing cells; for review see Freund, 2003), perisomatic inhibition cannot be fully characterized by the analysis of PV-immunopositive interneurons. Controversial results have been found in different rat models of epilepsy and human studies concerning the perisomatic inhibitory input of hippocampal principal cells. In the dentate gyrus, loss of axo-axonic and axo-somatic inhibitory inputs of granule cells was found in the kindling (Sayin et al., 2003) and pilocarpine models (Kobayashi and Buckmaster, 2003), but the preservation of perisomatic inhibitory input (Wittner et al., 2001) and an increased rather than decreased recurrent inhibition of granule cells (Isokawa-Akesson et al., 1989) has been shown in the dentate gyrus of epileptic patients. Perisomatic GABAergic input of CA1 pyramidal cells remained unchanged in the kainate model (Morin et al., 1999), whereas in the pilocarpine model dendritic but not somatic inhibition is decreased (Cossart et al., 2001), together with a slight decrease also in axo-axonic input (Dinocourt et al., 2003). These contradictory results also show that an extensive comparison and correlation with data obtained from human epileptic tissue is needed to validate predictions from animal models.
No detailed electron microscopic analyses have been done to date investigating the perisomatic inhibitory input of CA1 pyramidal cells in the human hippocampus. Here we analysed the distribution, morphology and input–output characteristics of PV-immunopositive interneurons, as well as the inhibitory synaptic input of CA1 pyramidal cell somata and axon initial segments (AISs) of human control subjects and epileptic patients.
Material and methods
PV-containing elements were examined in the CA1 region of 32 temporal lobe samples surgically removed from epileptic patients, and in seven control hippocampi. Patients with intractable TLE were operated on in the National Institute of Neurosurgery and in the MaV Hospital in Budapest; standard anterior temporal lobectomies were performed (Spencer and Spencer, 1985). The anterior third of the temporal lobe was removed together with the temporomedial structures. Control brain samples were obtained from a 37-year-old woman who died by accident, from 47- and 48-year-old women and 47-, 51- and 56-year-old men who had a cardiac arrest, and from a 53-year-old man who died by suffocation. None of the control subjects had a record of any neurological disorder. Brains were removed 2–4 h after death; the dissection was performed in the Institute of Pathology and Forensic Medicine, Semmelweis University, Budapest. In all cases, the patient's consent was obtained according to the Declaration of Helsinki, and the ethical regulations of the Hungarian Ministry of Health were followed.
After surgical removal, the epileptic tissue was immediately dissected into 2 mm thick blocks, and immersed in a fixative containing 4% paraformaldehyde, 0.1% glutaraldehyde and 0.2% picric acid in 0.1 M phosphate buffer (PB, pH 7.4), as described earlier (Magloczky et al., 1997). The fixative was changed every hour to a fresh solution during constant agitation for 6 h, then the blocks were post-fixed in the same fixative overnight. Four of the control hippocampi were subjected to the same procedure. Three of the control brains (C2, C10 and C11) were removed from the skull 2 h after death and both internal carotid and vertebral arteries were cannulated; the brain was perfused with physiological saline (2 l in 30 min) followed by a fixative solution containing 4% paraformaldehyde and 0.2% picric acid in 0.1 M PB (5 l in 3.5 h). The hippocampus was removed after perfusion and cut into 2 mm thick blocks, which were post-fixed in the same fixative solution overnight.
Immunocytochemistry
Vibratome sections (60 μm thick) were cut from the blocks and, following washing in PB, they were immersed in 30% saccharose for 1–2 days, then frozen three times over liquid nitrogen. Sections were processed for immunostaining against PV, as follows. Sections were transferred to Tris-buffered saline (TBS, pH 7.4), then endogenous peroxidase was blocked by 1% H2O2 in TBS for 10 min. TBS was used for all the washes (3 x 10 min between each sample) and dilution of the antisera. Non-specific immunostaining was blocked by 5% milk powder and 2% bovine serum albumin. A monoclonal mouse antibody against PV (1 : 5000; Sigma, St Louis, MO) was used for 2 days at 4°C. The specificity of the antibody has been thoroughly tested by the manufacturer. For visualization of immunopositive elements, biotinylated anti-mouse immunoglobulin G (IgG) (1 : 300, Vector) was applied as secondary antiserum followed by avidin-biotinylated horseradish peroxidase complex (ABC; 1 : 300, Vector). The immunoperoxidase reaction was developed by 3,3'-diaminobenzidine tetrahydrochloride (DAB; Sigma) dissolved in Tris buffer (TB, pH 7.6) as a chromogen. Sections were then osmicated (1% OsO4 in PB, 40 min) and dehydrated in ethanol (1% uranyl acetate was added at the 70% ethanol stage for 40 min) and mounted in Durcupan (ACM, Fluka). The control hippocampi were processed in the same way.
Controls of the method included the incubation of sections in the same way, but the primary antibody was replaced by normal mouse serum. No specific immunostaining could be detected under these conditions. Non-specific staining visualized large numbers of lipofuscin granules in the cells, which were therefore considered as background.
After light microscopic examination, areas of interest were re-embedded and sectioned for electron microscopy. Ultrathin serial sections were collected on Formvar-coated single slot grids, stained with lead citrate, and examined with a Hitachi 7100 electron microscope.
Quantitative analysis
Cell counting was performed in three control (C6, C10 and C11) and in six epileptic (P16, P22, P31, P40, P75 and P79) brains. Each PV-immunopositive cell together with the outlines of the CA1 region were drawn by camera lucida from two to five sections per patient. The area of the CA1 region was measured using the analySIS program, and the number of PV-immunopositive cells per unit area (mm2) was determined in the same area. The numbers of cells derived from different sections of the same subject were summed and divided by the sum of the areas of the same sections. The averages of the different pathological groups (control, mild, patchy and sclerotic) were determined by calculating the average of data from all patients in each group.
The target elements of PV-immunopositive axon terminals were examined in four control brains (C2, C6, C10 and C11) and in four epileptic patients (P15, P31, P40 and P54). Based on light microscopic examination, sections parallel to the plane of the CA1 pyramidal cell apical dendrites were chosen for the electron microscopic analysis. The entire width of CA1 stratum pyramidale including the zone of AISs located at the border of stratum oriens was re-embedded from PV-immunostained sections; the size of the blocks was similar. Serial sections were made from the blocks, and PV-containing terminals were analysed in every tenth section in order, following the rules of systematic random sampling, to avoid sampling of the same axon terminals. Photographs were taken from every PV-immunopositive synaptic terminal in each section, and the distribution of the postsynaptic target elements of the PV-immunopositive terminals was determined. In all electron microscopic analyses, the means of the different pathological groups (control versus epileptic, mild and patchy groups) were determined by calculating the average of the data of every patient belonging to that group.
The synaptic inputs of CA1 pyramidal cell AISs and somata were analysed in three control hippocampi (C2, C6 and C10) and in five epileptic patients (AIS, P15, P16, P40, P54 and P79; cell body, P15, P31, P40, P54 and P79). The sampling of the AISs of pyramidal cells was made as described above; each AIS observed in the stratum pyramidale and the bordering zone of the stratum oriens was digitized from every tenth section using a MegaView II camera. About 40–80 AISs were analysed in every subject. The border zone of the stratum oriens was included in the examined area to analyse the whole length of the pyramidal cell AISs. The AISs were obliquely cut, and the average perimeter was similar in all control and epileptic subjects. This varied between 7.38 and 13.64 μm, and no statistical differences were found between the groups of AISs of different subjects [Kruskal–Wallis ANOVA (analysis of variance), P > 0.05]. For the analysis of synaptic input of CA1 pyramidal cell bodies, serial sections were made from two blocks from each subject containing the entire width of the stratum pyramidale. About 20 pyramidal cells were digitized from each block (40 cells per subject). The perimeters of the pyramidal cell AISs and somata and the length of the synapses they receive were measured by the analySIS program. The synaptic coverage (μm synaptic length/100 μm of AIS or soma perimeter) of every AIS and soma has been recorded. Statistical analysis (Kruskal–Wallis ANOVA and Mann–Whitney U test) were performed using the Statistica 6.0 program. The number of synapses (terminals forming simple and perforated synapses) per 100 μm perimeter was calculated in every patient by dividing the total number of synapses by the total perimeter recorded. Asymmetric somatic synapses were only occasionally found, and were not included in the calculation.
Dependence of PV immunostaining on fixation and post-mortem delay
In a preliminary experiment, we examined 12 control brains with different post-mortem delays and age in both genders (Wittner et al., 2002). Briefly, the progression of age influences the quality and quantity of immunostaining, i.e. most of the antisera showed weaker staining and a lower number of cells in aged subjects. Therefore, the subjects older than 60 years have been excluded from the detailed analysis. For electron microscopic studies, subjects were chosen with a post-mortem delay of <4 h, because longer post-mortem delays (6–24 h) resulted in poor ultrastructural preservation, and a decrease in immunoreactivity. The electron microscopic analysis of these tissues revealed acceptable ultrastructural preservation even in the immersion-fixed control (C6), although it was inferior to the perfused tissue. The preservation of the 2–3 h post-mortem perfused controls (C2, C10 and C11) was comparable to the immediately fixed epileptic samples and to the perfused animal tissues. The PV staining in the control subjects included in the study was similar to each other and to previously described human or monkey PV immunostaining (Braak et al., 1991; Seress et al., 1991, 1993; Sloviter et al., 1991; Ribak et al., 1993). Therefore, we concluded that the differences found between control and epileptic tissues in the present study are likely to be caused by epilepsy.
Pathological classification of the tissue samples
All patients examined in the present study had therapy-resistant epilepsy of temporal lobe origin. The seizure focus was identified by multimodal studies including video-EEG monitoring, single photon emission computed tomography (SPECT) and/or PET. Only patients without gross temporal lobe damage or tumour were included in the study. The patients had different degrees of hippocampal atrophy and/or sclerosis. Similarly to recent results (de Lanerolle et al., 2003), we established four groups of epileptic patients based on the principal cell loss and interneuronal changes examined at the light microscopic level as follows (Table 1). (i) Type 1 (mild): similar to control, no considerable cell loss in the CA1 region, pyramidal cells are present, layers are visible, their borders are clearly identified. There is a slight loss in certain interneuron types, mostly in the hilus and the stratum oriens of the CA1 region. (ii) Type 2 (patchy): pyramidal cell loss in patches in the CA1 pyramidal cell layer, but these segments of the CA1 region are not atrophic. Interneuron loss is more pronounced. (iii) Type 3 (sclerotic): the CA1 region is shrunken, atrophic, >90% cell loss, occasionally scattered pyramidal cells remained in the CA1 region, separation of the layers is impossible. Only the stratum lacunosum-moleculare is present in the CA1 region as a distinct layer; the other strata could not be separated from each other due to the lack of pyramidal cells and shrinkage of the tissue. This remaining part should contain the layers determined in the control as strata oriens, pyramidale and radiatum. Mossy fibre sprouting and considerable changes in the distribution and morphology of interneurons can be observed in the samples of this group. (iv) Type 4 (gliotic): the whole hippocampus is shrunken, atrophic, not only the CA1 region, the loss of all cell types can be observed, including the resistant neurons (granule cells, calbindin-positive interneurons). The number of samples belonging to this group was small and therefore they were not included in this study.
Results
The distribution and morphological features of PV-immunopositive cells were examined at light and electron microscopic levels in the CA1 region derived from human controls and TLE patients. The number and distribution of cells proved that the antigenicity was retained for PV in the post-mortem controls included in this study (see Material and methods: considerations of the quality of post-mortem tissue). Samples taken from four control brains (three of them were perfused, see Material and methods) as well as from epileptic patients with mild and strong sclerosis were analysed further in the electron microscope to obtain data about the distribution of PV-positive axon terminals, their postsynaptic targets, the synaptic input of PV-positive interneuron cell bodies and dendrites, as well as the perisomatic synaptic input of CA1 pyramidal cells.
Light microscopy of PV-positive interneurons
In describing different regions of the hippocampal formation, we used the nomenclature of Lorente de No (1934) and Rosene and Van Hoesen (1987) with the proposed modifications of Seress concerning the CA3c region (Seress, 1988) (see also Fig. 1).
In the human control CA1 region, PV-immunopositive interneurons were located in the strata oriens and pyramidale (Fig. 2A), as was reported in previous publications (Braak et al., 1991; Sloviter et al., 1991; Seress et al., 1993). Large multipolar cells can be seen in the stratum pyramidale; their long dendrites run radially through the region and terminate in the stratum lacunosum-moleculare. Horizontal PV-immunopositive interneurons are situated in the stratum oriens; their dendrites arborize mostly in this layer (Fig. 2A). As it was described in the rat hippocampus (Fukuda and Kosaka, 2000), we often found long dendritic segments originating from different cells, attached to each other. Such dendritic juxtapositions were seen most frequently among radially running dendrites in the strata radiatum and lacunosum-moleculare (Fig. 2A, arrowheads) and the horizontal dendrites of the stratum oriens. Dense PV-immunopositive terminal staining was observed in the stratum pyramidale (Fig. 2A, insert). The PV-immunopositive interneurons in the human control CA1 region are known to belong to the populations of basket and axo-axonic cells (Braak et al., 1991; Seress et al., 1993; Sloviter et al., 1991), and terminate within the stratum pyramidale.
In the non-sclerotic epileptic CA1 region with mild cell loss (type 1), the distribution and number of PV-immunopositive interneurons did not change considerably; only a slight decrease was observed in the number of the horizontal stratum oriens cells (Fig. 2B, Table 2). Interestingly, the dendrites of PV-immunopositive interneurons were shorter than in the control tissue. Dense axonal staining of a density similar to, if not more than, that of controls was observed in the stratum pyramidale (Fig 2B, insert); basket-like formations (small arrow) and vertically oriented bouton rows characteristic of chandelier cells can also be distinguished (double arrows). In the CA1 region with patchy cell loss (type 2), a considerable decrease was found in the number of PV-immunopositive cells (Fig. 2C, Table 2), with the preservation of a dense axonal staining (Fig. 2C, insert). The termination pattern of PV-immunopositive cells did not change; the axonal cloud was confined to the stratum pyramidale. It was distributed homogeneously in the pyramidal cell layer, and also covered the patches with no pyramidal cells visible at the light microscopic level (Fig. 2C, insert). In the sclerotic epileptic tissue (type 3), the number of PV-immunopositive elements was dramatically decreased (Fig. 2D, Table 2). PV-immunopositive cells and axons can hardly be found; only a few PV-immunopositive dendrites were present in the sclerotic CA1 region (Fig. 2D, insert). The border between the sclerotic CA1 region and the prosubiculum was clearly visible; this latter was not sclerotic, a large number of pyramidal cells were present and the PV staining was similar to the control (Fig. 1). Since very small numbers of PV-immunopositive elements were found in the sclerotic CA1 region, samples from only the non-sclerotic epileptic hippocampi (with mild or patchy cell loss) were chosen for further quantitative analyses performed at the electron microscopic level. Cell counting confirmed the loss of PV-immunopositive cells in the epileptic CA1 region (Table 2). Three controls and two patients from each pathological group were chosen for the analysis. The area of the CA1 region was measured and the number of PV-immunopositive cells per unit area was determined (see Material and methods). We found a decrease of PV-immunopositive cells that correlated well with the degree of hippocampal damage in general (Table 2), i.e. the number of PV-immunopositive neurons was similar to control in those hippocampi where no considerable pyramidal cell loss was found in the CA1 region (type 1), a severe decrease in the number of PV-immunopositive somata accompanied by the preservation of axonal staining in cases with patchy cell loss (type 2), and severe cell loss with no axonal staining for PV in the sclerotic hippocampi (type 3).
Electron microscopy of PV-positive elements
Cell bodies and dendrites
Five PV-immunopositive cells from the CA1 region of human controls C2, C6 and C11 were analysed at the electron microscopic level (Fig. 3A). Every examined cell body was located in the stratum pyramidale. PV-immunopositive interneuron cell bodies received a large number of asymmetrical (excitatory) synapses, and axon terminals forming symmetrical (inhibitory) synapses were rarely seen (Fig. 3A1–3). Ten to 12 synaptic contacts per section were observed to terminate on the cell body of PV-immunopositive interneurons, usually one or two of them giving symmetrical synapses. PV-stained dendrites were covered by a large number of asymmetrical synapses. Zonula adhaerentia were frequently observed between PV-immunopositive dendrites (Fig. 4A).
Four PV-immunopositive interneurons derived from epileptic tissue (P15, P40, P54 and P79) were analysed in the electron microscope (Fig. 3B). Only PV-immunopositive cells from the non-sclerotic CA1 region were analysed, since PV-immunopositive cell bodies can hardly be found in the sclerotic tissue. In the non-sclerotic epileptic CA1 region, the synaptic input of PV-immunopositive cell bodies and dendrites was similar to the control, a large number of terminals forming asymmetrical synapses and a few others giving symmetrical synapses terminated on the PV-immunopositive elements (Fig. 3B1–2). A small number of PV-immunopositive dendrites were also analysed in the sclerotic CA1 region (P21, P22, type 3, sclerotic tissue). These dendrites were also covered by asymmetrical synapses, and by glial elements (Fig. 4B). In the epileptic CA1 region, zonula adhaerentia were frequently found between PV-immunopositive dendrites, but PV-immunopositive and -negative dendrites also formed zonula adhaerentia with each other (Fig. 4B).
Axon terminals and postsynaptic targets
The postsynaptic target distribution of PV-immunopositive boutons was determined in the CA1 region of four control (C2, A6, C10 and C11, n = 52–55) and four epileptic samples (P40 and P54, mild; and P15 and P31, patchy; Fig. 5, n = 52–107). In the CA1 region of patients with strong sclerosis, PV-immunopositive terminals were too sparse to study; in most areas they were absent. In the sample from P15, we distinguished patches containing (P15PC+) and lacking (P15PC–) pyramidal cells. Blocks were re-embedded from both regions, and the target distribution of PV-immunopositive terminals determined separately. Since the results deriving from the two separate blocks were similar, the data were pooled. PV-immunopositive terminals gave exclusively symmetrical synapses mostly to pyramidal cells in both control and epileptic tissue: they terminated on AISs (control, 14.5%; epileptic, 17.3%), pyramidal cell bodies (control, 14.8%; epileptic, 11.1%; Fig. 4D), proximal dendrites (control, 18.2%; epileptic, 23.1%), distal dendrites of pyramidal cells (control, 11.6%; epileptic, 6.4%) and spines (control, 17.6%; epileptic, 10.9%). They rarely established synaptic contacts on interneuron dendrites (control, 2.3%; epileptic, 1.9%), and even less frequently on PV-immunopositive interneuron dendrites. We also distinguished a category of unidentified dendrites with small diameter (control, 21.0%; epileptic, 29.4%). There was a larger variance within the groups of control and epileptic samples than between the control and the epileptic groups. Some of the targets were degenerating profiles in the epileptic samples from P15, P31 and P54 (P15, 54.2%; in P15PC+, 11.7%; in P15PC–, 92.2%; P31, 81.5%; P54, 60.7%; Fig. 4C).
Perisomatic inhibition of pyramidal cells
Cell bodies. Somatic inhibitory input of CA1 pyramidal cells was analysed in three control (C2, C6 and C10) and in five epileptic (P40, P54 and P79, mild; and P15 and P31, patchy) samples (Table 3, Fig. 6A). From each subject, we analysed on average 40 pyramidal cells from different areas (two separate blocks, 20 cells each). Synaptic coverage (μm synaptic length/100 μm soma perimeter) and the number of synapses/100 μm soma perimeter were determined (see Materials and methods). Every symmetric synapse, irrespective of its PV content, was recorded and measured. Asymmetric synapses were only occasionally found and were not included in the analysis. In the control tissue, 16–29% of the terminals were PV immunopositive, giving symmetric synapses to the pyramidal cell bodies. This ratio was higher in certain epileptic cases (45–48%, in P79, mild; P15 and P31, patchy), and did not change in others (27%, in P40; P54, mild).
In both control and epileptic tissue, some small differences were found in the synaptic coverage and the number of terminals giving synapses to pyramidal cell somata in the different blocks derived from the same subject. After the data were pooled by subjects, very similar results were found within the different groups (control, mild and patchy epileptic, Table 3, Fig. 6A). In the epileptic group with mild cell loss the synaptic input of pyramidal cells was very similar to the control (synaptic coverage: 0.637 ± 0.043 (mean ± SEM) versus 0.642 ± 0.048 in control and epileptic with mild cell loss, respectively).
In the epileptic tissue with patchy cell loss (P15 and P31), we took samples from light microscopically identified patches that were rich or poor in (contained hardly any) pyramidal cells. At the electron microscopic level, we found that in ‘poor’ patches, pyramidal cells were still present, but the majority of them showed signs of degeneration (100% in P15; 81% in P31). Interestingly, all examined parameters (synaptic coverage, number of synapses/100 μm soma perimeter, ratio of PV-immunopositive terminals forming synapses on the pyramidal cell somata and synaptic active zone length) were similar in the blocks derived from patches containing degenerating and healthy pyramidal cells. Therefore, the results were pooled. Thus, in the epileptic tissue with patchy cell loss, the synaptic input was significantly decreased (synaptic coverage 0.424 ± 0.045, P < 0.05) compared with the control, or with the mildly epileptic samples (Table 3, Fig. 6A). In the samples containing a considerable number of degenerating pyramidal cells (P15, P31 and P54), we determined the synaptic input of degenerating and non-degenerating pyramidal cells. The synaptic coverage of the degenerating cells was slightly increased in all cases, but this difference was not higher than the variance between different areas or subjects.
Axon initial segments. Synaptic input of CA1 pyramidal cell AISs was analysed in three control (C2, C6 and C10) and five epileptic hippocampi (P40, P54 and P79, mild; P15 and P16, patchy; for sampling see Material and methods; Table 4, Fig. 6A). Synaptic coverage (μm synaptic length/100 μm AIS perimeter) and number of synapses/100 μm AIS perimeter were determined. Every axon terminal giving symmetric synapses to AISs was included in the study, irrespective of their PV content. About half of the terminals (45–46%) forming synaptic contact on pyramidal cell AISs were PV immunopositive in the control tissue, and this did not change considerably in the epileptic tissue (36–50%).
The synaptic coverage of AISs did not change significantly in the epileptic samples (Kruskal–Wallis ANOVA and Mann–Whitney U tests, P > 0.05; Table 4, Fig. 6B). We separated the degenerating AISs in samples P15 and P54, and analysed their synaptic input. Degenerating AISs were identified on the basis of the following ultrastructural criteria: they showed signs of degeneration (homogeneous and electrondense cytoplasm, degenerating organelles) and at the same time the characteristic undercoating of the plasma membrane and microtubule fascicles was preserved. The synaptic coverage of degenerating AISs from both samples was decreased [synaptic coverage in P15, 3.60 ± 0.65 (mean ± SEM) versus 2.20 ± 0.53 of non-degenerating and degenerating AIS, respectively; synaptic coverage in P54, 3.98 ± 0.94 versus 3.69 ± 0.49 of non-degenerating and degenerating AIS, respectively]. However, the sample is too small (16 degenerating AISs from P15, and eight non-degenerating AISs from P54) to evaluate the significance of this result.
We found an increase of active zone length of synapses terminating on pyramidal cell somata and AISs (Tables 3 and 4), as it was reported earlier in an animal model of epilepsy (Nusser et al., 1998) and human epileptic tissue (Wittner et al., 2002).
Discussion
The number of PV-immunopositive cells is decreased in the epileptic CA1 region
Animal models and human studies have provided evidence for both the vulnerability (Best et al., 1993; Magloczky and Freund, 1995; Buckmaster and Dudek, 1997; Zhu et al., 1997) and the preservation (Sloviter et al., 1991) of PV-immunopositive interneurons in epilepsy. In the present study, we observed a decrease in the number of PV-immunopositive interneurons in the human epileptic CA1 region, and this reduction was correlated with the degree of sclerotic cell loss. In the non-sclerotic epileptic CA1 region with mild cell loss (type 1), a slight decrease was observed in the number of PV-immunopositive interneurons (Fig. 2B), which was more pronounced in the CA1 region with patchy pyramidal cell loss (type 2). This phenomenon was observed in conjunction with the preservation of the dense terminal staining in the stratum pyramidale (Fig. 2C, Table 2). The distribution of postsynaptic target elements of PV-positive axon terminals remained unchanged (Fig. 5) and the perisomatic inhibitory input of CA1 pyramidal cells was preserved in the non-sclerotic epileptic CA1 region (Tables 3 and 4, Fig. 6). A reduction of PV immunoreactivity without basket cell loss was found in animal models of ischaemia and epilepsy (Johansen et al., 1990; Tortosa and Ferrer, 1993; Bazzett et al., 1994; Magloczky and Freund, 1995; Scotti et al., 1997a), as well as in the human epileptic dentate gyrus (Wittner et al., 2001). We found that the loss of PV immunostaining from cell bodies and dendrites was accompanied by the preservation of their terminals as well as the inhibitory input of pyramidal cells, suggesting that here again we are dealing with the decrease of immunoreactivity rather than the death of the cells. In animal models of epilepsy, slower synthesis of PV or a change in conformation due to calcium overloading was suggested as a possible cause for the reduced PV immunostaining (Johansen et al., 1990; Scotti et al., 1997a). This might also operate in the human hippocampus.
In contrast to the non-sclerotic epileptic hippocampus, in the sclerotic CA1 region PV-immunopositive interneurons are lost, the number of PV-immunopositive elements has dramatically decreased and usually small numbers of dendrites are visible in the region (Fig. 2D). The nomenclature of the human hippocampal formation is still controversial, including the determination of the border between the CA1 region and subiculum and the existence of the prosubiculum (Rosene and Van Hoesen, 1987; Amaral and Insausti, 1990; Duvernoy, 1998). This might explain the contradictory observations concerning the preservation (Arellano et al., 2003) or disappearance (present work) of PV-immunopositive elements in the sclerotic CA1 region.
PV-immunopositive interneurons receive the highest number of synapses among the different interneuron types in the rat hippocampus, and they receive the smallest proportion of GABAergic input (Gulyas et al., 1999). No difference was observed in the synaptic input of PV-immunopositive cell bodies in the epileptic CA1 region compared with the control (Fig. 3), but the sprouting of mossy fibres in the dentate gyrus (Sutula et al., 1989; Houser et al., 1990; Represa et al., 1990), as well as the sprouting of a calbindin-positive excitatory pathway in the CA1 region (Wittner et al., 2002) is likely to increase the density of excitatory input to PV-immunopositive dendrites further (Blasco-Ibanez et al., 2000; Magloczky et al., 2000). Direct evidence for enhanced excitation in the CA1 region was shown only in animal models (Buzsaki et al., 1989; Rafiq et al., 1993; Bragin et al., 1997). The overexcitation of these interneurons might play a role in the decrease of PV below the level of immunocytochemical detection (Bazzett et al., 1994), and/or in the excitotoxic death of these cells, as was shown earlier for CA1 pyramidal cells (Sano and Kirino, 1990; Meldrum, 1991, 1993). PV-immunopositive cell bodies located in the CA1 region receive a much larger excitatory input than those in the dentate gyrus (10–12 versus 4–5 synapses/cell body perimeter in one section, in the CA1 region or the dentate gyrus, respectively; Wittner et al., 2001). It is possible that these two mechanisms together, i.e. loss of original targets and overexcitation of the cells, play a role in the disappearance of PV-immunopositive interneurons in the sclerotic CA1 region.
Perisomatic inhibition is preserved if pyramidal cells are present
Basket cells and axo-axonic (or chandelier) cells are responsible for the perisomatic input of principal cells in the hippocampus of all examined species, including rat and monkey (Somogyi et al., 1983; Lewis and Lund, 1990; Seress and Frotscher, 1991; Freund and Buzsaki, 1996; Halasy et al., 1996; Martinez et al., 1996) and humans as well (Seress et al., 1993; Braak et al., 1991). The majority of these cells are PV immunopositive, but cholecystokinin-containing basket cells (Lotstra and Vanderhaeghen, 1987) and calbindin-positive chandelier cells (Wittner et al., 2002) have also been described in the human hippocampus.
Since the PV content of perisomatic inhibitory cells is sensitive to epileptic and ischaemic conditions (see above), some surviving PV-immunopositive cells may not be visible with PV immunostaining. Therefore, we examined the perisomatic inhibitory synapses on CA1 pyramidal cells irrespective of their PV immunoreactivity.
Analysis of synaptic coverage of pyramidal cell somata and AISs showed that perisomatic inhibitory input is preserved in the epileptic CA1 region, except in the samples belonging to the pathological group with patchy cell loss, where a slight, but significant decrease was observed (Table 3, Fig. 6). Furthermore, the distribution of postsynaptic targets of PV-positive terminals in the epileptic tissue was similar to that of the control (Fig. 5). These results suggest that perisomatic inhibitory cells may survive in epilepsy, if their targets are present in the region. The lack of profound changes in perisomatic inhibitory input in the CA1 region suggests that other factors are likely to account for the selective vulnerability of pyramidal cells.
In the rodent hippocampus, electrical coupling plays an important role in the synchronization of PV-immunopositive interneurons (Deans et al., 2001; Meyer et al., 2002). Gap junction-coupled ensembles of PV-immunopositive cells (Katsumaru et al., 1988a; Fukuda and Kosaka, 2000) synchronize the firing pattern of pyramidal cells and generate gamma oscillations in the neocortex and hippocampus (Penttonen et al., 1998; Galarreta and Hestrin, 1999; Gibson et al., 1999; Tamas et al., 2000). Zonula adhaerentia have been observed between PV-immunopositive dendrites in the human control (Seress et al., 1993) and epileptic hippocampus (Fig. 4A and B), as well as in the vicinity of gap junctions in rats (Kosaka and Hama, 1985), and they are thought to ensure the mechanical connection of dendrites. Our results suggest that similar coupling mechanisms occur between PV-immunopositive interneurons in human tissue, which in turn probably lead to a more effective synchronization of pyramidal cells that may participate in the maintenance of epileptic seizures.
Conclusions
The balance of excitation and inhibition in the human CA1 region is considerably altered during epileptic changes. An excess excitation is provided by the sprouted CA1 pyramidal cell axons (Lehmann et al., 2000), as well as by the sprouting of afferents from the dentate gyrus and/or the CA2 region (Wittner et al., 2002), and/or from extrahippocampal pathways (Lehmann et al., 2000). Certain populations of dendritic inhibitory cells survive in epilepsy, and participate in the intense synaptic reorganization (Wittner et al., 2002), while others are lost (de Lanerolle et al., 1988, 1989; Magloczky et al., 2000). Perisomatic inhibitory input is preserved in the non-sclerotic epileptic CA1 region, where pyramidal cells survive. The preserved perisomatic inhibition together with an excess excitation and an impaired dendritic inhibition might lead to an abnormal synchrony in output regions of the hippocampus, which still contain large enough numbers of projection neurons. This might have an important role in the generation and spread of epileptic seizures. In this group of patients, the preservation and possible increase in the efficacy of the perisomatic inhibitory network may be causally related to hypersynchrony during seizures. This may partly explain the difficulties in therapies that rely on the modulation of GABAergic transmission, that does not distinguish between perisomatic (controlling synchrony) and dendritic (controlling input plasticity) inhibition.
Acknowledgements
We wish to thank Drs M. Palkovits, P. Sotonyi and Zs. Borostyanki (Semmelweis University, Budapest) for providing control human tissue, and Ms K. Lengyel, E. Simon and Mr Gy. Goda for excellent technical assistance. This study was supported by grants from the NIH (MH 54671), the Howard Hughes Medical Institute (55000307) and OTKA (T032251) Hungary.
References
Amaral DG, Insausti R. Hippocampal formation. In: Paxinos G, editor. The human nervous system. San Diego: Academic Press; 1990. p. 711–55.
Arellano JI, Munoz A, Ballesteros-Yanez I, Sola RG, DeFelipe J. Histopathology and reorganization of chandelier cells in the human epileptic sclerotic hippocampus. Brain 2004; 127(pt 1): 45–64.
Avoli M. Is epilepsy a disorder of inhibition or excitation Prog Clin Biol Res 1983; 124: 23–37.
Babb TL, Pretorius JK, Kupfer WR, Crandall PH. Glutamate decarboxylase-immunoreactive neurons are preserved in human epileptic hippocampus. J Neurosci 1989; 9: 2562–74.
Bazzett TJ, Becker JB, Falik RC, Albin RL. Chronic intrastriatal quinolinic acid produces reversible changes in perikaryal calbindin and parvalbumin immunoreactivity. Neuroscience 1994; 60: 837–41.
Best N, Mitchell J, Baimbridge KG, Wheal HV. Changes in parvalbumin-immunoreactive neurons in the rat hippocampus following a kainic acid lesion. Neurosci Lett 1993; 155: 1–6.
Blasco-Ibanez JM, Martinez-Guijarro FJ, Freund TF. Recurrent mossy fibers preferentially innervate parvalbumin-immunoreactive interneurons in the granule cell layer of the rat dentate gyrus. Neuroreport 2000; 11: 3219–25.
Braak E, Strotkamp B, Braak H. Parvalbumin-immunoreactive structures in the hippocampus of the human adult. Cell Tissue Res 1991; 264: 33–48.
Bragin A, Csicsvari J, Penttonen M, Buzsaki G. Epileptic afterdischarge in the hippocampal–entorhinal system: current source density and unit studies. Neuroscience 1997; 76: 1187–203.
Buckmaster PS, Dudek FE. Neuron loss, granule cell axon reorganization, and functional changes in the dentate gyrus of epileptic kainate-treated rats. J Comp Neurol 1997; 385: 385–404.
Buzsaki G, Ponomareff GL, Bayardo F, Ruiz R, Gage FH. Neuronal activity in the subcortically denervated hippocampus: a chronic model for epilepsy. Neuroscience 1989; 28: 527–38.
Corsellis JAN. The incidence of Ammon's horn sclerosis. Brain 1955; 80: 193–208.
Cossart R, Dinocourt C, Hirsch JC, Merchan-Perez A, De Felipe J, Ben-Ari Y, et al. Dendritic but not somatic GABAergic inhibition is decreased in experimental epilepsy. Nat Neurosci 2001; 4: 52–62.
Deans MR, Gibson JR, Sellitto C, Connors BW, Paul DL. Synchronous activity of inhibitory networks in neocortex requires electrical synapses containing connexin36. Neuron 2001; 31: 477–85.
DeFelipe J. Chandelier cells and epilepsy. Brain 1999; 122: 1807–22.
DeFelipe J, Garcia Sola R, Marco P, del Rio MR, Pulido P, Ramon y Cajal S. Selective changes in the microorganization of the human epileptogenic neocortex revealed by parvalbumin immunoreactivity. Cereb Cortex 1993; 3: 39–48.
de Lanerolle NC, Sloviter RS, Kim JH, Robbins RJ, Spencer DD. Evidence for hippocampal interneuron loss in human temporal lobe epilepsy. Epilepsia 1988; 29: 674.
de Lanerolle NC, Kim JH, Robbins RJ, Spencer DD. Hippocampal interneuron loss and plasticity in human temporal lobe epilepsy. Brain Res 1989; 495: 387–95.
de Lanerolle NC, Kim JH, Williamson A, Spencer SS, Zaveri HP, Eid T, et al. A retrospective analysis of hippocampal pathology in human temporal lobe epilepsy: evidence for distinctive patient subcategories. Epilepsia 2003; 44: 677–87.
Dinocourt C, Petanjek Z, Freund TF, Ben-Ari Y, Esclapez M. Loss of interneurons innervating pyramidal cell dendrites and axon initial segments in the CA1 region of the hippocampus following pilocarpine-induced seizures. J Comp Neurol 2003; 459: 407–25.
Duvernoy HM. The human hippocampus. Berlin: Springer-Verlag; 1998.
Esclapez M, Hirsch JC, Ben-Ari Y, Bernard C. Newly formed excitatory pathways provide a substrate for hyperexcitability in experimental temporal lobe epilepsy. J Comp Neurol 1999; 408: 449–60.
Falconer MA. The significance of mesial temporal sclerosis (Ammon's horn sclerosis) in epilepsy. Guys Hosp Rep 1968; 117: 1–12.
Freund TF. Interneuron diversity series: rhythm and mood in perisomatic inhibition. Trends Neurosci 2003; 26: 489–95.
Freund TF, Buzsaki G. Interneurons of the hippocampus. Hippocampus 1996; 6: 347–470.
Fukuda T, Kosaka T. Gap junctions linking the dendritic network of GABAergic interneurons in the hippocampus. J Neurosci 2000; 20: 1519–28.
Galarreta M, Hestrin S. A network of fast-spiking cells in the neocortex connected by electrical synapses. Nature 1999; 402: 72–5.
Gibson JR, Beierlein M, Connors BW. Two networks of electrically coupled inhibitory neurons in neocortex. Nature 1999; 402: 75–9.
Gulyas AI, Megias M, Emri Z, Freund TF. Total number and ratio of excitatory and inhibitory synapses converging onto single interneurons of different types in the CA1 area of the rat hippocampus. J Neurosci 1999; 19: 10082–97.
Halasy K, Buhl EH, Lorinczi Z, Tamas G, Somogyi P. Synaptic target selectivity and input of GABAergic basket and bistratified interneurons in the CA1 area of the rat hippocampus. Hippocampus 1996; 6: 306–29.
Holm IE, Geneser FA, Zimmer J, Baimbridge KG. Immunocytochemical demonstration of the calcium-binding proteins calbindin-D 28k and parvalbumin in the subiculum, hippocampus and dentate area of the domestic pig. Prog Brain Res 1990; 83: 85–97.
Houser CR, Miyashiro JE, Swartz BE, Walsh GO, Rich JR, Delgado-Escueta AV. Altered patterns of dynorphin immunoreactivity suggest mossy fiber reorganization in human hippocampal epilepsy. J Neurosci 1990; 10: 267–82.
Isokawa-Akesson M, Wilson CL, Babb TL. Inhibition in synchronously firing human hippocampal neurons. Epilepsy Res 1989; 3: 236–47.
Johansen FF, Tonder N, Zimmer J, Baimbridge KG, Diemer NH. Short-term changes of parvalbumin and calbindin immunoreactivity in the rat hippocampus following cerebral ischemia. Neurosci Lett 1990; 120: 171–4.
Katsumaru H, Kosaka T, Heizmann CW, Hama K. Gap junctions on GABAergic neurons containing the calcium-binding protein parvalbumin in the rat hippocampus (CA1 region). Exp Brain Res 1988a; 72: 363–70.
Katsumaru H, Kosaka T, Heizmann CW, Hama K. Immunocytochemical study of GABAergic neurons containing the calcium-binding protein parvalbumin in the rat hippocampus. Exp Brain Res 1988b; 72: 347–62.
Kobayashi M, Buckmaster PS. Reduced inhibition of dentate granule cells in a model of temporal lobe epilepsy. J Neurosci 2003; 23: 2440–52.
Kosaka T, Hama K. Gap junctions between non-pyramidal cell dendrites in the rat hippocampus (CA1 and CA3 regions): a combined Golgi–electron microscopy study. J Comp Neurol 1985; 231: 150–61.
Kosaka T, Katsumaru H, Hama K, Wu JY, Heizmann CW. GABAergic neurons containing the Ca2+-binding protein parvalbumin in the rat hippocampus and dentate gyrus. Brain Res 1987; 419: 119–30.
Lehmann TN, Gabriel S, Kovacs R, Eilers A, Kivi A, Schulze K, et al. Alterations of neuronal connectivity in area CA1 of hippocampal slices from temporal lobe epilepsy patients and from pilocarpine-treated epileptic rats. Epilepsia 2000; 41: S190–4.
Lehmann TN, Gabriel S, Eilers A, Njunting M, Kovacs R, Schulze K, et al. Fluorescent tracer in pilocarpine-treated rats shows widespread aberrant hippocampal neuronal connectivity. Eur J Neurosci 2001; 14: 83–95.
Lewis DA, Lund JS. Heterogeneity of chandelier neurons in monkey neocortex: corticotropin-releasing factor- and parvalbumin-immunoreactive populations. J Comp Neurol 1990; 293: 599–615.
Lorente de No R. Studies on the structure of the cerebral cortex—II. Continuation of the study of the ammonic system. J Psychol Neurol 1934; 46: 113–77.
Lotstra F, Vanderhaeghen JJ. Distribution of immunoreactive cholecystokinin in the human hippocampus. Peptides 1987; 8: 911–20.
Maclean P. The limbic system concept. In: Trimble M, Bolwig T, editors. The temporal lobes and the limbic system. Wrighston Biomedical; 1992. p. 1–265.
Magloczky Z, Freund TF. Delayed cell death in the contralateral hippocampus following kainate injection into the CA3 subfield. Neuroscience 1995; 66: 847–60.
Magloczky Z, Halasz P, Vajda J, Czirjak S, Freund TF. Loss of calbindin-D28K immunoreactivity from dentate granule cells in human temporal lobe epilepsy. Neuroscience 1997; 76: 377–85.
Magloczky Z, Wittner L, Borhegyi Z, Halasz P, Vajda J, Czirjak S, et al. Changes in the distribution and connectivity of interneurons in the epileptic human dentate gyrus. Neuroscience 2000; 96: 7–25.
Marco P, Sola RG, Cayal SRY, DeFelipe J. Loss of inhibitory synapses on the soma and axon intial segment of pyramidal cells in human epileptic peritumoral neocortex: implications for epilepsy. Brain Res Bull 1997; 44: 47–66.
Martinez A, Lubke J, Del Rio JA, Soriano E, Frotscher M. Regional variability and postsynaptic targets of chandelier cells in the hippocampal formation of the rat. J Comp Neurol 1996; 376: 28–44.
Meldrum B. Excitotoxicity and epileptic brain damage. Epilepsy Res 1991; 10: 55–61.
Meldrum BS. Excitotoxicity and selective neuronal loss in epilepsy. Brain Pathol 1993; 3: 405–12.
Meyer AH, Katona I, Blatow M, Rozov A, Monyer H. In vivo labeling of parvalbumin-positive interneurons and analysis of electrical coupling in identified neurons. J Neurosci 2002; 22: 7055–64.
Miles R, Toth K, Gulyas AI, Hajos N, Freund TF. Differences between somatic and dendritic inhibition in the hippocampus. Neuron 1996; 16: 815–23.
Mody I, Otis TS, Staley KJ, Kohr G. The balance between excitation and inhibition in dentate granule cells and its role in epilepsy. Epilepsy Res Suppl 1992; 9: 331–9.
Morin F, Beaulieu C, Lacaille JC. Alterations of perisomatic GABA synapses on hippocampal CA1 inhibitory interneurons and pyramidal cells in the kainate model of epilepsy. Neuroscience 1999; 93: 457–67.
Nitsch R, Bergmann I, Kuppers K, Mueller G, Frotscher M. Late appearance of parvalbumin-immunoreactivity in the development of GABAergic neurons in the rat hippocampus. Neurosci Lett 1990; 118: 147–50.
Nusser Z, Hajos N, Somogyi P, Mody I. Increased number of synaptic GABA(A) receptors underlies potentiation at hippocampal inhibitory synapses. Nature 1998; 395: 172–7.
Penttonen M, Kamondi A, Acsady L, Buzsaki G. Gamma frequency oscillation in the hippocampus of the rat: intracellular analysis in vivo. Eur J Neurosci 1998; 10: 718–28.
Rafiq A, DeLorenzo RJ, Coulter DA. Generation and propagation of epileptiform discharges in a combined entorhinal cortex/hippocampal slice. J Neurophysiol 1993; 70: 1962–74.
Ramon y Cajal S. Histologie du systeme nerveux de l'homme et des vertebres. Vol. I, II. Paris: Maloine; 1909–1911.
Represa A, Tremblay E, Ben-Ari Y. Sprouting of mossy fibers in the hippocampus of epileptic human and rat. Adv Exp Med Biol 1990; 268: 419–24.
Ribak CE, Seress L, Leranth C. Electron microscopic immunocytochemical study of the distribution of parvalbumin-containing neurons and axon terminals in the primate dentate gyrus and Ammon's horn. J Comp Neurol 1993; 327: 298–321.
Rosene DL, Van Hoesen GW. The hippocampal formation of the primate brain: a review of some comparative aspects of cytoarchitecture and connections. In: Jones EG, Peters A, editors. Cerebral cortex. Vol. 6. New York: Plenum Press; 1987. p. 345–456.
Sano K, Kirino T. Ammon's horn sclerosis: its pathogenesis and clinical significance. Tohoku J Exp Med 1990; 161 Suppl: 273–95.
Sayin U, Osting S, Hagen J, Rutecki P, Sutula T. Spontaneous seizures and loss of axo-axonic and axo-somatic inhibition induced by repeated brief seizures in kindled rats. J Neurosci 2003; 23: 2759–68.
Scotti A, Bollag O, Kalt G, Nitsch C. Loss of perikaryal parvalbumin immunoreactivity from surviving GABAergic neurons in the CA1 field of epileptic gerbils. Hippocampus 1997a; 7: 524–35.
Scotti AL, Kalt G, Bollag O, Nitsch C. Parvalbumin disappears from GABAergic CA1 neurons of the gerbil hippocampus with seizure onset while its presence persists in the perforant path. Brain Res 1997b; 760: 109–17.
Seress L. Interspecies comparison of the hippocampal formation shows increased emphasis on the regio superior in the Ammon's horn of the human brain. J Hirnforsch 1988; 29: 335–40.
Seress L, Frotscher M. Basket cells in the monkey fascia dentata: a Golgi/electron microscopic study. J Neurocytol 1991; 20: 915–28.
Seress L, Gulyas AI, Freund TF. Parvalbumin- and calbindin D28k-immunoreactive neurons in the hippocampal formation of the macaque monkey. J Comp Neurol 1991; 313: 162–77.
Seress L, Gulyas AI, Ferrer I, Tunon T, Soriano E, Freund TF. Distribution, morphological features, and synaptic connections of parvalbumin- and calbindin D28k-immunoreactive neurons in the human hippocampal formation. J Comp Neurol 1993; 337: 208–30.
Sloviter RS, Sollas AL, Barbaro NM, Laxer KD. Calcium-binding protein (calbindin-D28K) and parvalbumin immunocytochemistry in the normal and epileptic human hippocampus. J Comp Neurol 1991; 308: 381–96.
Sommer W. Erkrankung des Ammonshornes als Aetiologisches Moment der Epilepsie. Arch Psychiat Nervkrankh 1880; 10: 631–75.
Somogyi P, Nunzi MG, Gorio A, Smith AD. A new type of specific interneuron in the monkey hippocampus forming synapses exclusively with the axon initial segments of pyramidal cells. Brain Res 1983; 259: 137–42.
Spencer DD, Spencer SS. Surgery for epilepsy. Neurol Clin 1985; 3: 313–30.
Sutula TP, Cascino G, Cavazos JE, Ramirez L. Mossy fiber synaptic reorganization in the epileptic human temporal lobe. Ann Neurol 1989; 26: 321–30.
Tamas G, Buhl EH, Lrincz A, Somogyi P. Proximally targeted GABAergic synapses and gap junctions synchronize cortical interneurons. Nat Neurosci 2000; 3: 366–71.
Tortosa A, Ferrer I. Parvalbumin immunoreactivity in the hippocampus of the gerbil after transient forebrain ischaemia: a qualitative and quantitative sequential study. Neuroscience 1993; 55: 33–43.
Wittner L, Magloczky Z, Borhegyi Z, Halasz P, Toth S, Erss L, et al. Preservation of perisomatic inhibitory input of granule cells in the epileptic human dentate gyrus. Neuroscience 2001; 108: 587–600.
Wittner L, Erss L, Szabo Z, Toth S, Czirjak S, Halasz P, et al. Synaptic reorganization of calbindin-positive neurons in the human hippocampal CA1 region in temporal lobe epilepsy. Neuroscience 2002; 115: 961–78.
Zhu ZQ, Armstrong DL, Hamilton WJ, Grossman RG. Disproportionate loss of CA4 parvalbumin-immunoreactive interneurons in patients with Ammon's horn sclerosis. J Neuropathol Exp Neurol 1997; 56: 988–98.(L. Wittner, L. Erss, S. Czirjak, P. Hala)
4 National Institute of Psychiatry and Neurology, Epilepsy Center, Budapest, Hungary
Summary
Temporal lobe epilepsy (TLE) is known to be linked to an impaired balance of excitation and inhibition. Whether inhibition is decreased or preserved in the human epileptic hippocampus, beside the excess excitation, is still a debated question. In the present study, quantitative light and electron microscopy has been performed to analyse the distribution, morphology and input–output connections of parvalbumin (PV)-immunopositive interneurons, together with the entire perisomatic input of pyramidal cells, in the human control and epileptic CA1 region. Based on the degree of cell loss, the patients with therapy-resistant TLE formed four pathological groups. In the non-sclerotic CA1 region of TLE patients, where large numbers of pyramidal cells are preserved, the number of PV-immunopositive cell bodies decreased, whereas axon terminal staining, and the distribution of their postsynaptic targets was not altered. The synaptic coverage of CA1 pyramidal cell axon initial segments (AISs) remained unchanged in the epileptic tissue. The somatic inhibitory input is also preserved; it has been decreased only in the cases with patchy pyramidal cell loss in the CA1 region (control, 0.637; epileptic with mild cell loss, 0.642; epileptic with patchy cell loss, 0.424 μm synaptic length/100 μm soma perimeter). The strongly sclerotic epileptic CA1 region, where pyramidal cells can hardly be seen, contains a very small number of PV-immunopositive elements. Our results suggest that perisomatic inhibitory input is preserved in the epileptic CA1 region as long as pyramidal cells are present. Basket and axo-axonic cells survive in epilepsy if their original targets are present, although many of them lose their PV content or PV immunoreactivity. An efficient perisomatic inhibition is likely to take part in the generation of abnormal synchrony in the non-sclerotic epileptic CA1 region, and thus participate in the maintenance of epileptic seizures driven, for example, by hyperactive afferent input.
Key Words: perisomatic inhibition; GABA; basket cell; chandelier cell; selective vulnerability
Abbreviations: AIS = axon initial segment; CA1–3 = regions of the cornu Ammonis; PV = parvalbumin; TLE = temporal lobe epilepsy
Introduction
Epilepsy is thought to be related to a changed balance between excitation and inhibition (Avoli, 1983; Mody et al., 1992). In epilepsy, the CA1 region was shown to receive an enhanced excitatory input from the CA3 region in the rat hippocampus (Buzsaki et al., 1989; Rafiq et al., 1993; Bragin et al., 1997), from the dentate gyrus and/or CA2 region in the human hippocampus (Wittner et al., 2002) and from the sprouting CA1 pyramidal cell axons in rats (Esclapez et al., 1999; Lehmann et al., 2001) and humans (Lehmann et al., 2000). It is still not clear whether inhibition is decreased or preserved in the human epileptic CA1 region. There is evidence for interneuronal cell death (de Lanerolle et al., 1988, 1989; Magloczky et al., 2000), as well as for the preservation of glutamic acid dehydrogenase (GAD)-positive cells (Babb et al., 1989). Since CA1 pyramidal cells provide the cortical output connections of the hippocampus (Ramon y Cajal, 1909–1911; Lorente de No, 1934; Maclean, 1992) and are known to be the most vulnerable in epilepsy (Sommer, 1880; Corsellis, 1955; Falconer, 1968), alterations in the inhibitory network in the CA1 region may be crucial in the pathophysiology of the epileptic temporal lobe.
The calcium-binding protein parvalbumin (PV) is known to be present exclusively in perisomatic inhibitory interneurons in the hippocampus of rodents (Kosaka et al., 1987; Katsumaru et al., 1988b; Holm et al., 1990; Nitsch et al., 1990), monkeys (Seress et al., 1991; Ribak et al., 1993) and humans (Braak et al., 1991; Seress et al., 1993). Two perisomatic inhibitory cell types, basket and chandelier or axo-axonic cells, were demonstrated to be PV-immunopositive in all examined species, including humans (Seress et al., 1993; for a review see Freund and Buzsaki, 1996) and to control the output of principal cells (Miles et al., 1996). The survival of PV-immunopositive interneurons in epilepsy, as well as their role in the pathophysiological mechanisms of seizures and cell death are still controversial questions. Some authors have described the preservation (Sloviter et al., 1991; Arellano et al., 2003) and others the vulnerability (Zhu et al., 1997; Wittner et al., 2001) of these cells in the human epileptic hippocampus, and a decrease in the number of PV-immunopositive interneurons has been found in the epileptic neocortex (DeFelipe et al., 1993; Marco et al., 1997; DeFelipe, 1999). PV immunoreactivity is very sensitive to epileptic and ischaemic conditions and might disappear from neurons that are otherwise preserved (Johansen et al., 1990; Tortosa and Ferrer, 1993; Bazzett et al., 1994; Scotti et al., 1997a, b). Since PV immunostaining probably reveals only part of the originally PV-immunopositive cells, which in turn represent only one of the at least two basket cell types (the other being the cholecystokinin-containing cells; for review see Freund, 2003), perisomatic inhibition cannot be fully characterized by the analysis of PV-immunopositive interneurons. Controversial results have been found in different rat models of epilepsy and human studies concerning the perisomatic inhibitory input of hippocampal principal cells. In the dentate gyrus, loss of axo-axonic and axo-somatic inhibitory inputs of granule cells was found in the kindling (Sayin et al., 2003) and pilocarpine models (Kobayashi and Buckmaster, 2003), but the preservation of perisomatic inhibitory input (Wittner et al., 2001) and an increased rather than decreased recurrent inhibition of granule cells (Isokawa-Akesson et al., 1989) has been shown in the dentate gyrus of epileptic patients. Perisomatic GABAergic input of CA1 pyramidal cells remained unchanged in the kainate model (Morin et al., 1999), whereas in the pilocarpine model dendritic but not somatic inhibition is decreased (Cossart et al., 2001), together with a slight decrease also in axo-axonic input (Dinocourt et al., 2003). These contradictory results also show that an extensive comparison and correlation with data obtained from human epileptic tissue is needed to validate predictions from animal models.
No detailed electron microscopic analyses have been done to date investigating the perisomatic inhibitory input of CA1 pyramidal cells in the human hippocampus. Here we analysed the distribution, morphology and input–output characteristics of PV-immunopositive interneurons, as well as the inhibitory synaptic input of CA1 pyramidal cell somata and axon initial segments (AISs) of human control subjects and epileptic patients.
Material and methods
PV-containing elements were examined in the CA1 region of 32 temporal lobe samples surgically removed from epileptic patients, and in seven control hippocampi. Patients with intractable TLE were operated on in the National Institute of Neurosurgery and in the MaV Hospital in Budapest; standard anterior temporal lobectomies were performed (Spencer and Spencer, 1985). The anterior third of the temporal lobe was removed together with the temporomedial structures. Control brain samples were obtained from a 37-year-old woman who died by accident, from 47- and 48-year-old women and 47-, 51- and 56-year-old men who had a cardiac arrest, and from a 53-year-old man who died by suffocation. None of the control subjects had a record of any neurological disorder. Brains were removed 2–4 h after death; the dissection was performed in the Institute of Pathology and Forensic Medicine, Semmelweis University, Budapest. In all cases, the patient's consent was obtained according to the Declaration of Helsinki, and the ethical regulations of the Hungarian Ministry of Health were followed.
After surgical removal, the epileptic tissue was immediately dissected into 2 mm thick blocks, and immersed in a fixative containing 4% paraformaldehyde, 0.1% glutaraldehyde and 0.2% picric acid in 0.1 M phosphate buffer (PB, pH 7.4), as described earlier (Magloczky et al., 1997). The fixative was changed every hour to a fresh solution during constant agitation for 6 h, then the blocks were post-fixed in the same fixative overnight. Four of the control hippocampi were subjected to the same procedure. Three of the control brains (C2, C10 and C11) were removed from the skull 2 h after death and both internal carotid and vertebral arteries were cannulated; the brain was perfused with physiological saline (2 l in 30 min) followed by a fixative solution containing 4% paraformaldehyde and 0.2% picric acid in 0.1 M PB (5 l in 3.5 h). The hippocampus was removed after perfusion and cut into 2 mm thick blocks, which were post-fixed in the same fixative solution overnight.
Immunocytochemistry
Vibratome sections (60 μm thick) were cut from the blocks and, following washing in PB, they were immersed in 30% saccharose for 1–2 days, then frozen three times over liquid nitrogen. Sections were processed for immunostaining against PV, as follows. Sections were transferred to Tris-buffered saline (TBS, pH 7.4), then endogenous peroxidase was blocked by 1% H2O2 in TBS for 10 min. TBS was used for all the washes (3 x 10 min between each sample) and dilution of the antisera. Non-specific immunostaining was blocked by 5% milk powder and 2% bovine serum albumin. A monoclonal mouse antibody against PV (1 : 5000; Sigma, St Louis, MO) was used for 2 days at 4°C. The specificity of the antibody has been thoroughly tested by the manufacturer. For visualization of immunopositive elements, biotinylated anti-mouse immunoglobulin G (IgG) (1 : 300, Vector) was applied as secondary antiserum followed by avidin-biotinylated horseradish peroxidase complex (ABC; 1 : 300, Vector). The immunoperoxidase reaction was developed by 3,3'-diaminobenzidine tetrahydrochloride (DAB; Sigma) dissolved in Tris buffer (TB, pH 7.6) as a chromogen. Sections were then osmicated (1% OsO4 in PB, 40 min) and dehydrated in ethanol (1% uranyl acetate was added at the 70% ethanol stage for 40 min) and mounted in Durcupan (ACM, Fluka). The control hippocampi were processed in the same way.
Controls of the method included the incubation of sections in the same way, but the primary antibody was replaced by normal mouse serum. No specific immunostaining could be detected under these conditions. Non-specific staining visualized large numbers of lipofuscin granules in the cells, which were therefore considered as background.
After light microscopic examination, areas of interest were re-embedded and sectioned for electron microscopy. Ultrathin serial sections were collected on Formvar-coated single slot grids, stained with lead citrate, and examined with a Hitachi 7100 electron microscope.
Quantitative analysis
Cell counting was performed in three control (C6, C10 and C11) and in six epileptic (P16, P22, P31, P40, P75 and P79) brains. Each PV-immunopositive cell together with the outlines of the CA1 region were drawn by camera lucida from two to five sections per patient. The area of the CA1 region was measured using the analySIS program, and the number of PV-immunopositive cells per unit area (mm2) was determined in the same area. The numbers of cells derived from different sections of the same subject were summed and divided by the sum of the areas of the same sections. The averages of the different pathological groups (control, mild, patchy and sclerotic) were determined by calculating the average of data from all patients in each group.
The target elements of PV-immunopositive axon terminals were examined in four control brains (C2, C6, C10 and C11) and in four epileptic patients (P15, P31, P40 and P54). Based on light microscopic examination, sections parallel to the plane of the CA1 pyramidal cell apical dendrites were chosen for the electron microscopic analysis. The entire width of CA1 stratum pyramidale including the zone of AISs located at the border of stratum oriens was re-embedded from PV-immunostained sections; the size of the blocks was similar. Serial sections were made from the blocks, and PV-containing terminals were analysed in every tenth section in order, following the rules of systematic random sampling, to avoid sampling of the same axon terminals. Photographs were taken from every PV-immunopositive synaptic terminal in each section, and the distribution of the postsynaptic target elements of the PV-immunopositive terminals was determined. In all electron microscopic analyses, the means of the different pathological groups (control versus epileptic, mild and patchy groups) were determined by calculating the average of the data of every patient belonging to that group.
The synaptic inputs of CA1 pyramidal cell AISs and somata were analysed in three control hippocampi (C2, C6 and C10) and in five epileptic patients (AIS, P15, P16, P40, P54 and P79; cell body, P15, P31, P40, P54 and P79). The sampling of the AISs of pyramidal cells was made as described above; each AIS observed in the stratum pyramidale and the bordering zone of the stratum oriens was digitized from every tenth section using a MegaView II camera. About 40–80 AISs were analysed in every subject. The border zone of the stratum oriens was included in the examined area to analyse the whole length of the pyramidal cell AISs. The AISs were obliquely cut, and the average perimeter was similar in all control and epileptic subjects. This varied between 7.38 and 13.64 μm, and no statistical differences were found between the groups of AISs of different subjects [Kruskal–Wallis ANOVA (analysis of variance), P > 0.05]. For the analysis of synaptic input of CA1 pyramidal cell bodies, serial sections were made from two blocks from each subject containing the entire width of the stratum pyramidale. About 20 pyramidal cells were digitized from each block (40 cells per subject). The perimeters of the pyramidal cell AISs and somata and the length of the synapses they receive were measured by the analySIS program. The synaptic coverage (μm synaptic length/100 μm of AIS or soma perimeter) of every AIS and soma has been recorded. Statistical analysis (Kruskal–Wallis ANOVA and Mann–Whitney U test) were performed using the Statistica 6.0 program. The number of synapses (terminals forming simple and perforated synapses) per 100 μm perimeter was calculated in every patient by dividing the total number of synapses by the total perimeter recorded. Asymmetric somatic synapses were only occasionally found, and were not included in the calculation.
Dependence of PV immunostaining on fixation and post-mortem delay
In a preliminary experiment, we examined 12 control brains with different post-mortem delays and age in both genders (Wittner et al., 2002). Briefly, the progression of age influences the quality and quantity of immunostaining, i.e. most of the antisera showed weaker staining and a lower number of cells in aged subjects. Therefore, the subjects older than 60 years have been excluded from the detailed analysis. For electron microscopic studies, subjects were chosen with a post-mortem delay of <4 h, because longer post-mortem delays (6–24 h) resulted in poor ultrastructural preservation, and a decrease in immunoreactivity. The electron microscopic analysis of these tissues revealed acceptable ultrastructural preservation even in the immersion-fixed control (C6), although it was inferior to the perfused tissue. The preservation of the 2–3 h post-mortem perfused controls (C2, C10 and C11) was comparable to the immediately fixed epileptic samples and to the perfused animal tissues. The PV staining in the control subjects included in the study was similar to each other and to previously described human or monkey PV immunostaining (Braak et al., 1991; Seress et al., 1991, 1993; Sloviter et al., 1991; Ribak et al., 1993). Therefore, we concluded that the differences found between control and epileptic tissues in the present study are likely to be caused by epilepsy.
Pathological classification of the tissue samples
All patients examined in the present study had therapy-resistant epilepsy of temporal lobe origin. The seizure focus was identified by multimodal studies including video-EEG monitoring, single photon emission computed tomography (SPECT) and/or PET. Only patients without gross temporal lobe damage or tumour were included in the study. The patients had different degrees of hippocampal atrophy and/or sclerosis. Similarly to recent results (de Lanerolle et al., 2003), we established four groups of epileptic patients based on the principal cell loss and interneuronal changes examined at the light microscopic level as follows (Table 1). (i) Type 1 (mild): similar to control, no considerable cell loss in the CA1 region, pyramidal cells are present, layers are visible, their borders are clearly identified. There is a slight loss in certain interneuron types, mostly in the hilus and the stratum oriens of the CA1 region. (ii) Type 2 (patchy): pyramidal cell loss in patches in the CA1 pyramidal cell layer, but these segments of the CA1 region are not atrophic. Interneuron loss is more pronounced. (iii) Type 3 (sclerotic): the CA1 region is shrunken, atrophic, >90% cell loss, occasionally scattered pyramidal cells remained in the CA1 region, separation of the layers is impossible. Only the stratum lacunosum-moleculare is present in the CA1 region as a distinct layer; the other strata could not be separated from each other due to the lack of pyramidal cells and shrinkage of the tissue. This remaining part should contain the layers determined in the control as strata oriens, pyramidale and radiatum. Mossy fibre sprouting and considerable changes in the distribution and morphology of interneurons can be observed in the samples of this group. (iv) Type 4 (gliotic): the whole hippocampus is shrunken, atrophic, not only the CA1 region, the loss of all cell types can be observed, including the resistant neurons (granule cells, calbindin-positive interneurons). The number of samples belonging to this group was small and therefore they were not included in this study.
Results
The distribution and morphological features of PV-immunopositive cells were examined at light and electron microscopic levels in the CA1 region derived from human controls and TLE patients. The number and distribution of cells proved that the antigenicity was retained for PV in the post-mortem controls included in this study (see Material and methods: considerations of the quality of post-mortem tissue). Samples taken from four control brains (three of them were perfused, see Material and methods) as well as from epileptic patients with mild and strong sclerosis were analysed further in the electron microscope to obtain data about the distribution of PV-positive axon terminals, their postsynaptic targets, the synaptic input of PV-positive interneuron cell bodies and dendrites, as well as the perisomatic synaptic input of CA1 pyramidal cells.
Light microscopy of PV-positive interneurons
In describing different regions of the hippocampal formation, we used the nomenclature of Lorente de No (1934) and Rosene and Van Hoesen (1987) with the proposed modifications of Seress concerning the CA3c region (Seress, 1988) (see also Fig. 1).
In the human control CA1 region, PV-immunopositive interneurons were located in the strata oriens and pyramidale (Fig. 2A), as was reported in previous publications (Braak et al., 1991; Sloviter et al., 1991; Seress et al., 1993). Large multipolar cells can be seen in the stratum pyramidale; their long dendrites run radially through the region and terminate in the stratum lacunosum-moleculare. Horizontal PV-immunopositive interneurons are situated in the stratum oriens; their dendrites arborize mostly in this layer (Fig. 2A). As it was described in the rat hippocampus (Fukuda and Kosaka, 2000), we often found long dendritic segments originating from different cells, attached to each other. Such dendritic juxtapositions were seen most frequently among radially running dendrites in the strata radiatum and lacunosum-moleculare (Fig. 2A, arrowheads) and the horizontal dendrites of the stratum oriens. Dense PV-immunopositive terminal staining was observed in the stratum pyramidale (Fig. 2A, insert). The PV-immunopositive interneurons in the human control CA1 region are known to belong to the populations of basket and axo-axonic cells (Braak et al., 1991; Seress et al., 1993; Sloviter et al., 1991), and terminate within the stratum pyramidale.
In the non-sclerotic epileptic CA1 region with mild cell loss (type 1), the distribution and number of PV-immunopositive interneurons did not change considerably; only a slight decrease was observed in the number of the horizontal stratum oriens cells (Fig. 2B, Table 2). Interestingly, the dendrites of PV-immunopositive interneurons were shorter than in the control tissue. Dense axonal staining of a density similar to, if not more than, that of controls was observed in the stratum pyramidale (Fig 2B, insert); basket-like formations (small arrow) and vertically oriented bouton rows characteristic of chandelier cells can also be distinguished (double arrows). In the CA1 region with patchy cell loss (type 2), a considerable decrease was found in the number of PV-immunopositive cells (Fig. 2C, Table 2), with the preservation of a dense axonal staining (Fig. 2C, insert). The termination pattern of PV-immunopositive cells did not change; the axonal cloud was confined to the stratum pyramidale. It was distributed homogeneously in the pyramidal cell layer, and also covered the patches with no pyramidal cells visible at the light microscopic level (Fig. 2C, insert). In the sclerotic epileptic tissue (type 3), the number of PV-immunopositive elements was dramatically decreased (Fig. 2D, Table 2). PV-immunopositive cells and axons can hardly be found; only a few PV-immunopositive dendrites were present in the sclerotic CA1 region (Fig. 2D, insert). The border between the sclerotic CA1 region and the prosubiculum was clearly visible; this latter was not sclerotic, a large number of pyramidal cells were present and the PV staining was similar to the control (Fig. 1). Since very small numbers of PV-immunopositive elements were found in the sclerotic CA1 region, samples from only the non-sclerotic epileptic hippocampi (with mild or patchy cell loss) were chosen for further quantitative analyses performed at the electron microscopic level. Cell counting confirmed the loss of PV-immunopositive cells in the epileptic CA1 region (Table 2). Three controls and two patients from each pathological group were chosen for the analysis. The area of the CA1 region was measured and the number of PV-immunopositive cells per unit area was determined (see Material and methods). We found a decrease of PV-immunopositive cells that correlated well with the degree of hippocampal damage in general (Table 2), i.e. the number of PV-immunopositive neurons was similar to control in those hippocampi where no considerable pyramidal cell loss was found in the CA1 region (type 1), a severe decrease in the number of PV-immunopositive somata accompanied by the preservation of axonal staining in cases with patchy cell loss (type 2), and severe cell loss with no axonal staining for PV in the sclerotic hippocampi (type 3).
Electron microscopy of PV-positive elements
Cell bodies and dendrites
Five PV-immunopositive cells from the CA1 region of human controls C2, C6 and C11 were analysed at the electron microscopic level (Fig. 3A). Every examined cell body was located in the stratum pyramidale. PV-immunopositive interneuron cell bodies received a large number of asymmetrical (excitatory) synapses, and axon terminals forming symmetrical (inhibitory) synapses were rarely seen (Fig. 3A1–3). Ten to 12 synaptic contacts per section were observed to terminate on the cell body of PV-immunopositive interneurons, usually one or two of them giving symmetrical synapses. PV-stained dendrites were covered by a large number of asymmetrical synapses. Zonula adhaerentia were frequently observed between PV-immunopositive dendrites (Fig. 4A).
Four PV-immunopositive interneurons derived from epileptic tissue (P15, P40, P54 and P79) were analysed in the electron microscope (Fig. 3B). Only PV-immunopositive cells from the non-sclerotic CA1 region were analysed, since PV-immunopositive cell bodies can hardly be found in the sclerotic tissue. In the non-sclerotic epileptic CA1 region, the synaptic input of PV-immunopositive cell bodies and dendrites was similar to the control, a large number of terminals forming asymmetrical synapses and a few others giving symmetrical synapses terminated on the PV-immunopositive elements (Fig. 3B1–2). A small number of PV-immunopositive dendrites were also analysed in the sclerotic CA1 region (P21, P22, type 3, sclerotic tissue). These dendrites were also covered by asymmetrical synapses, and by glial elements (Fig. 4B). In the epileptic CA1 region, zonula adhaerentia were frequently found between PV-immunopositive dendrites, but PV-immunopositive and -negative dendrites also formed zonula adhaerentia with each other (Fig. 4B).
Axon terminals and postsynaptic targets
The postsynaptic target distribution of PV-immunopositive boutons was determined in the CA1 region of four control (C2, A6, C10 and C11, n = 52–55) and four epileptic samples (P40 and P54, mild; and P15 and P31, patchy; Fig. 5, n = 52–107). In the CA1 region of patients with strong sclerosis, PV-immunopositive terminals were too sparse to study; in most areas they were absent. In the sample from P15, we distinguished patches containing (P15PC+) and lacking (P15PC–) pyramidal cells. Blocks were re-embedded from both regions, and the target distribution of PV-immunopositive terminals determined separately. Since the results deriving from the two separate blocks were similar, the data were pooled. PV-immunopositive terminals gave exclusively symmetrical synapses mostly to pyramidal cells in both control and epileptic tissue: they terminated on AISs (control, 14.5%; epileptic, 17.3%), pyramidal cell bodies (control, 14.8%; epileptic, 11.1%; Fig. 4D), proximal dendrites (control, 18.2%; epileptic, 23.1%), distal dendrites of pyramidal cells (control, 11.6%; epileptic, 6.4%) and spines (control, 17.6%; epileptic, 10.9%). They rarely established synaptic contacts on interneuron dendrites (control, 2.3%; epileptic, 1.9%), and even less frequently on PV-immunopositive interneuron dendrites. We also distinguished a category of unidentified dendrites with small diameter (control, 21.0%; epileptic, 29.4%). There was a larger variance within the groups of control and epileptic samples than between the control and the epileptic groups. Some of the targets were degenerating profiles in the epileptic samples from P15, P31 and P54 (P15, 54.2%; in P15PC+, 11.7%; in P15PC–, 92.2%; P31, 81.5%; P54, 60.7%; Fig. 4C).
Perisomatic inhibition of pyramidal cells
Cell bodies. Somatic inhibitory input of CA1 pyramidal cells was analysed in three control (C2, C6 and C10) and in five epileptic (P40, P54 and P79, mild; and P15 and P31, patchy) samples (Table 3, Fig. 6A). From each subject, we analysed on average 40 pyramidal cells from different areas (two separate blocks, 20 cells each). Synaptic coverage (μm synaptic length/100 μm soma perimeter) and the number of synapses/100 μm soma perimeter were determined (see Materials and methods). Every symmetric synapse, irrespective of its PV content, was recorded and measured. Asymmetric synapses were only occasionally found and were not included in the analysis. In the control tissue, 16–29% of the terminals were PV immunopositive, giving symmetric synapses to the pyramidal cell bodies. This ratio was higher in certain epileptic cases (45–48%, in P79, mild; P15 and P31, patchy), and did not change in others (27%, in P40; P54, mild).
In both control and epileptic tissue, some small differences were found in the synaptic coverage and the number of terminals giving synapses to pyramidal cell somata in the different blocks derived from the same subject. After the data were pooled by subjects, very similar results were found within the different groups (control, mild and patchy epileptic, Table 3, Fig. 6A). In the epileptic group with mild cell loss the synaptic input of pyramidal cells was very similar to the control (synaptic coverage: 0.637 ± 0.043 (mean ± SEM) versus 0.642 ± 0.048 in control and epileptic with mild cell loss, respectively).
In the epileptic tissue with patchy cell loss (P15 and P31), we took samples from light microscopically identified patches that were rich or poor in (contained hardly any) pyramidal cells. At the electron microscopic level, we found that in ‘poor’ patches, pyramidal cells were still present, but the majority of them showed signs of degeneration (100% in P15; 81% in P31). Interestingly, all examined parameters (synaptic coverage, number of synapses/100 μm soma perimeter, ratio of PV-immunopositive terminals forming synapses on the pyramidal cell somata and synaptic active zone length) were similar in the blocks derived from patches containing degenerating and healthy pyramidal cells. Therefore, the results were pooled. Thus, in the epileptic tissue with patchy cell loss, the synaptic input was significantly decreased (synaptic coverage 0.424 ± 0.045, P < 0.05) compared with the control, or with the mildly epileptic samples (Table 3, Fig. 6A). In the samples containing a considerable number of degenerating pyramidal cells (P15, P31 and P54), we determined the synaptic input of degenerating and non-degenerating pyramidal cells. The synaptic coverage of the degenerating cells was slightly increased in all cases, but this difference was not higher than the variance between different areas or subjects.
Axon initial segments. Synaptic input of CA1 pyramidal cell AISs was analysed in three control (C2, C6 and C10) and five epileptic hippocampi (P40, P54 and P79, mild; P15 and P16, patchy; for sampling see Material and methods; Table 4, Fig. 6A). Synaptic coverage (μm synaptic length/100 μm AIS perimeter) and number of synapses/100 μm AIS perimeter were determined. Every axon terminal giving symmetric synapses to AISs was included in the study, irrespective of their PV content. About half of the terminals (45–46%) forming synaptic contact on pyramidal cell AISs were PV immunopositive in the control tissue, and this did not change considerably in the epileptic tissue (36–50%).
The synaptic coverage of AISs did not change significantly in the epileptic samples (Kruskal–Wallis ANOVA and Mann–Whitney U tests, P > 0.05; Table 4, Fig. 6B). We separated the degenerating AISs in samples P15 and P54, and analysed their synaptic input. Degenerating AISs were identified on the basis of the following ultrastructural criteria: they showed signs of degeneration (homogeneous and electrondense cytoplasm, degenerating organelles) and at the same time the characteristic undercoating of the plasma membrane and microtubule fascicles was preserved. The synaptic coverage of degenerating AISs from both samples was decreased [synaptic coverage in P15, 3.60 ± 0.65 (mean ± SEM) versus 2.20 ± 0.53 of non-degenerating and degenerating AIS, respectively; synaptic coverage in P54, 3.98 ± 0.94 versus 3.69 ± 0.49 of non-degenerating and degenerating AIS, respectively]. However, the sample is too small (16 degenerating AISs from P15, and eight non-degenerating AISs from P54) to evaluate the significance of this result.
We found an increase of active zone length of synapses terminating on pyramidal cell somata and AISs (Tables 3 and 4), as it was reported earlier in an animal model of epilepsy (Nusser et al., 1998) and human epileptic tissue (Wittner et al., 2002).
Discussion
The number of PV-immunopositive cells is decreased in the epileptic CA1 region
Animal models and human studies have provided evidence for both the vulnerability (Best et al., 1993; Magloczky and Freund, 1995; Buckmaster and Dudek, 1997; Zhu et al., 1997) and the preservation (Sloviter et al., 1991) of PV-immunopositive interneurons in epilepsy. In the present study, we observed a decrease in the number of PV-immunopositive interneurons in the human epileptic CA1 region, and this reduction was correlated with the degree of sclerotic cell loss. In the non-sclerotic epileptic CA1 region with mild cell loss (type 1), a slight decrease was observed in the number of PV-immunopositive interneurons (Fig. 2B), which was more pronounced in the CA1 region with patchy pyramidal cell loss (type 2). This phenomenon was observed in conjunction with the preservation of the dense terminal staining in the stratum pyramidale (Fig. 2C, Table 2). The distribution of postsynaptic target elements of PV-positive axon terminals remained unchanged (Fig. 5) and the perisomatic inhibitory input of CA1 pyramidal cells was preserved in the non-sclerotic epileptic CA1 region (Tables 3 and 4, Fig. 6). A reduction of PV immunoreactivity without basket cell loss was found in animal models of ischaemia and epilepsy (Johansen et al., 1990; Tortosa and Ferrer, 1993; Bazzett et al., 1994; Magloczky and Freund, 1995; Scotti et al., 1997a), as well as in the human epileptic dentate gyrus (Wittner et al., 2001). We found that the loss of PV immunostaining from cell bodies and dendrites was accompanied by the preservation of their terminals as well as the inhibitory input of pyramidal cells, suggesting that here again we are dealing with the decrease of immunoreactivity rather than the death of the cells. In animal models of epilepsy, slower synthesis of PV or a change in conformation due to calcium overloading was suggested as a possible cause for the reduced PV immunostaining (Johansen et al., 1990; Scotti et al., 1997a). This might also operate in the human hippocampus.
In contrast to the non-sclerotic epileptic hippocampus, in the sclerotic CA1 region PV-immunopositive interneurons are lost, the number of PV-immunopositive elements has dramatically decreased and usually small numbers of dendrites are visible in the region (Fig. 2D). The nomenclature of the human hippocampal formation is still controversial, including the determination of the border between the CA1 region and subiculum and the existence of the prosubiculum (Rosene and Van Hoesen, 1987; Amaral and Insausti, 1990; Duvernoy, 1998). This might explain the contradictory observations concerning the preservation (Arellano et al., 2003) or disappearance (present work) of PV-immunopositive elements in the sclerotic CA1 region.
PV-immunopositive interneurons receive the highest number of synapses among the different interneuron types in the rat hippocampus, and they receive the smallest proportion of GABAergic input (Gulyas et al., 1999). No difference was observed in the synaptic input of PV-immunopositive cell bodies in the epileptic CA1 region compared with the control (Fig. 3), but the sprouting of mossy fibres in the dentate gyrus (Sutula et al., 1989; Houser et al., 1990; Represa et al., 1990), as well as the sprouting of a calbindin-positive excitatory pathway in the CA1 region (Wittner et al., 2002) is likely to increase the density of excitatory input to PV-immunopositive dendrites further (Blasco-Ibanez et al., 2000; Magloczky et al., 2000). Direct evidence for enhanced excitation in the CA1 region was shown only in animal models (Buzsaki et al., 1989; Rafiq et al., 1993; Bragin et al., 1997). The overexcitation of these interneurons might play a role in the decrease of PV below the level of immunocytochemical detection (Bazzett et al., 1994), and/or in the excitotoxic death of these cells, as was shown earlier for CA1 pyramidal cells (Sano and Kirino, 1990; Meldrum, 1991, 1993). PV-immunopositive cell bodies located in the CA1 region receive a much larger excitatory input than those in the dentate gyrus (10–12 versus 4–5 synapses/cell body perimeter in one section, in the CA1 region or the dentate gyrus, respectively; Wittner et al., 2001). It is possible that these two mechanisms together, i.e. loss of original targets and overexcitation of the cells, play a role in the disappearance of PV-immunopositive interneurons in the sclerotic CA1 region.
Perisomatic inhibition is preserved if pyramidal cells are present
Basket cells and axo-axonic (or chandelier) cells are responsible for the perisomatic input of principal cells in the hippocampus of all examined species, including rat and monkey (Somogyi et al., 1983; Lewis and Lund, 1990; Seress and Frotscher, 1991; Freund and Buzsaki, 1996; Halasy et al., 1996; Martinez et al., 1996) and humans as well (Seress et al., 1993; Braak et al., 1991). The majority of these cells are PV immunopositive, but cholecystokinin-containing basket cells (Lotstra and Vanderhaeghen, 1987) and calbindin-positive chandelier cells (Wittner et al., 2002) have also been described in the human hippocampus.
Since the PV content of perisomatic inhibitory cells is sensitive to epileptic and ischaemic conditions (see above), some surviving PV-immunopositive cells may not be visible with PV immunostaining. Therefore, we examined the perisomatic inhibitory synapses on CA1 pyramidal cells irrespective of their PV immunoreactivity.
Analysis of synaptic coverage of pyramidal cell somata and AISs showed that perisomatic inhibitory input is preserved in the epileptic CA1 region, except in the samples belonging to the pathological group with patchy cell loss, where a slight, but significant decrease was observed (Table 3, Fig. 6). Furthermore, the distribution of postsynaptic targets of PV-positive terminals in the epileptic tissue was similar to that of the control (Fig. 5). These results suggest that perisomatic inhibitory cells may survive in epilepsy, if their targets are present in the region. The lack of profound changes in perisomatic inhibitory input in the CA1 region suggests that other factors are likely to account for the selective vulnerability of pyramidal cells.
In the rodent hippocampus, electrical coupling plays an important role in the synchronization of PV-immunopositive interneurons (Deans et al., 2001; Meyer et al., 2002). Gap junction-coupled ensembles of PV-immunopositive cells (Katsumaru et al., 1988a; Fukuda and Kosaka, 2000) synchronize the firing pattern of pyramidal cells and generate gamma oscillations in the neocortex and hippocampus (Penttonen et al., 1998; Galarreta and Hestrin, 1999; Gibson et al., 1999; Tamas et al., 2000). Zonula adhaerentia have been observed between PV-immunopositive dendrites in the human control (Seress et al., 1993) and epileptic hippocampus (Fig. 4A and B), as well as in the vicinity of gap junctions in rats (Kosaka and Hama, 1985), and they are thought to ensure the mechanical connection of dendrites. Our results suggest that similar coupling mechanisms occur between PV-immunopositive interneurons in human tissue, which in turn probably lead to a more effective synchronization of pyramidal cells that may participate in the maintenance of epileptic seizures.
Conclusions
The balance of excitation and inhibition in the human CA1 region is considerably altered during epileptic changes. An excess excitation is provided by the sprouted CA1 pyramidal cell axons (Lehmann et al., 2000), as well as by the sprouting of afferents from the dentate gyrus and/or the CA2 region (Wittner et al., 2002), and/or from extrahippocampal pathways (Lehmann et al., 2000). Certain populations of dendritic inhibitory cells survive in epilepsy, and participate in the intense synaptic reorganization (Wittner et al., 2002), while others are lost (de Lanerolle et al., 1988, 1989; Magloczky et al., 2000). Perisomatic inhibitory input is preserved in the non-sclerotic epileptic CA1 region, where pyramidal cells survive. The preserved perisomatic inhibition together with an excess excitation and an impaired dendritic inhibition might lead to an abnormal synchrony in output regions of the hippocampus, which still contain large enough numbers of projection neurons. This might have an important role in the generation and spread of epileptic seizures. In this group of patients, the preservation and possible increase in the efficacy of the perisomatic inhibitory network may be causally related to hypersynchrony during seizures. This may partly explain the difficulties in therapies that rely on the modulation of GABAergic transmission, that does not distinguish between perisomatic (controlling synchrony) and dendritic (controlling input plasticity) inhibition.
Acknowledgements
We wish to thank Drs M. Palkovits, P. Sotonyi and Zs. Borostyanki (Semmelweis University, Budapest) for providing control human tissue, and Ms K. Lengyel, E. Simon and Mr Gy. Goda for excellent technical assistance. This study was supported by grants from the NIH (MH 54671), the Howard Hughes Medical Institute (55000307) and OTKA (T032251) Hungary.
References
Amaral DG, Insausti R. Hippocampal formation. In: Paxinos G, editor. The human nervous system. San Diego: Academic Press; 1990. p. 711–55.
Arellano JI, Munoz A, Ballesteros-Yanez I, Sola RG, DeFelipe J. Histopathology and reorganization of chandelier cells in the human epileptic sclerotic hippocampus. Brain 2004; 127(pt 1): 45–64.
Avoli M. Is epilepsy a disorder of inhibition or excitation Prog Clin Biol Res 1983; 124: 23–37.
Babb TL, Pretorius JK, Kupfer WR, Crandall PH. Glutamate decarboxylase-immunoreactive neurons are preserved in human epileptic hippocampus. J Neurosci 1989; 9: 2562–74.
Bazzett TJ, Becker JB, Falik RC, Albin RL. Chronic intrastriatal quinolinic acid produces reversible changes in perikaryal calbindin and parvalbumin immunoreactivity. Neuroscience 1994; 60: 837–41.
Best N, Mitchell J, Baimbridge KG, Wheal HV. Changes in parvalbumin-immunoreactive neurons in the rat hippocampus following a kainic acid lesion. Neurosci Lett 1993; 155: 1–6.
Blasco-Ibanez JM, Martinez-Guijarro FJ, Freund TF. Recurrent mossy fibers preferentially innervate parvalbumin-immunoreactive interneurons in the granule cell layer of the rat dentate gyrus. Neuroreport 2000; 11: 3219–25.
Braak E, Strotkamp B, Braak H. Parvalbumin-immunoreactive structures in the hippocampus of the human adult. Cell Tissue Res 1991; 264: 33–48.
Bragin A, Csicsvari J, Penttonen M, Buzsaki G. Epileptic afterdischarge in the hippocampal–entorhinal system: current source density and unit studies. Neuroscience 1997; 76: 1187–203.
Buckmaster PS, Dudek FE. Neuron loss, granule cell axon reorganization, and functional changes in the dentate gyrus of epileptic kainate-treated rats. J Comp Neurol 1997; 385: 385–404.
Buzsaki G, Ponomareff GL, Bayardo F, Ruiz R, Gage FH. Neuronal activity in the subcortically denervated hippocampus: a chronic model for epilepsy. Neuroscience 1989; 28: 527–38.
Corsellis JAN. The incidence of Ammon's horn sclerosis. Brain 1955; 80: 193–208.
Cossart R, Dinocourt C, Hirsch JC, Merchan-Perez A, De Felipe J, Ben-Ari Y, et al. Dendritic but not somatic GABAergic inhibition is decreased in experimental epilepsy. Nat Neurosci 2001; 4: 52–62.
Deans MR, Gibson JR, Sellitto C, Connors BW, Paul DL. Synchronous activity of inhibitory networks in neocortex requires electrical synapses containing connexin36. Neuron 2001; 31: 477–85.
DeFelipe J. Chandelier cells and epilepsy. Brain 1999; 122: 1807–22.
DeFelipe J, Garcia Sola R, Marco P, del Rio MR, Pulido P, Ramon y Cajal S. Selective changes in the microorganization of the human epileptogenic neocortex revealed by parvalbumin immunoreactivity. Cereb Cortex 1993; 3: 39–48.
de Lanerolle NC, Sloviter RS, Kim JH, Robbins RJ, Spencer DD. Evidence for hippocampal interneuron loss in human temporal lobe epilepsy. Epilepsia 1988; 29: 674.
de Lanerolle NC, Kim JH, Robbins RJ, Spencer DD. Hippocampal interneuron loss and plasticity in human temporal lobe epilepsy. Brain Res 1989; 495: 387–95.
de Lanerolle NC, Kim JH, Williamson A, Spencer SS, Zaveri HP, Eid T, et al. A retrospective analysis of hippocampal pathology in human temporal lobe epilepsy: evidence for distinctive patient subcategories. Epilepsia 2003; 44: 677–87.
Dinocourt C, Petanjek Z, Freund TF, Ben-Ari Y, Esclapez M. Loss of interneurons innervating pyramidal cell dendrites and axon initial segments in the CA1 region of the hippocampus following pilocarpine-induced seizures. J Comp Neurol 2003; 459: 407–25.
Duvernoy HM. The human hippocampus. Berlin: Springer-Verlag; 1998.
Esclapez M, Hirsch JC, Ben-Ari Y, Bernard C. Newly formed excitatory pathways provide a substrate for hyperexcitability in experimental temporal lobe epilepsy. J Comp Neurol 1999; 408: 449–60.
Falconer MA. The significance of mesial temporal sclerosis (Ammon's horn sclerosis) in epilepsy. Guys Hosp Rep 1968; 117: 1–12.
Freund TF. Interneuron diversity series: rhythm and mood in perisomatic inhibition. Trends Neurosci 2003; 26: 489–95.
Freund TF, Buzsaki G. Interneurons of the hippocampus. Hippocampus 1996; 6: 347–470.
Fukuda T, Kosaka T. Gap junctions linking the dendritic network of GABAergic interneurons in the hippocampus. J Neurosci 2000; 20: 1519–28.
Galarreta M, Hestrin S. A network of fast-spiking cells in the neocortex connected by electrical synapses. Nature 1999; 402: 72–5.
Gibson JR, Beierlein M, Connors BW. Two networks of electrically coupled inhibitory neurons in neocortex. Nature 1999; 402: 75–9.
Gulyas AI, Megias M, Emri Z, Freund TF. Total number and ratio of excitatory and inhibitory synapses converging onto single interneurons of different types in the CA1 area of the rat hippocampus. J Neurosci 1999; 19: 10082–97.
Halasy K, Buhl EH, Lorinczi Z, Tamas G, Somogyi P. Synaptic target selectivity and input of GABAergic basket and bistratified interneurons in the CA1 area of the rat hippocampus. Hippocampus 1996; 6: 306–29.
Holm IE, Geneser FA, Zimmer J, Baimbridge KG. Immunocytochemical demonstration of the calcium-binding proteins calbindin-D 28k and parvalbumin in the subiculum, hippocampus and dentate area of the domestic pig. Prog Brain Res 1990; 83: 85–97.
Houser CR, Miyashiro JE, Swartz BE, Walsh GO, Rich JR, Delgado-Escueta AV. Altered patterns of dynorphin immunoreactivity suggest mossy fiber reorganization in human hippocampal epilepsy. J Neurosci 1990; 10: 267–82.
Isokawa-Akesson M, Wilson CL, Babb TL. Inhibition in synchronously firing human hippocampal neurons. Epilepsy Res 1989; 3: 236–47.
Johansen FF, Tonder N, Zimmer J, Baimbridge KG, Diemer NH. Short-term changes of parvalbumin and calbindin immunoreactivity in the rat hippocampus following cerebral ischemia. Neurosci Lett 1990; 120: 171–4.
Katsumaru H, Kosaka T, Heizmann CW, Hama K. Gap junctions on GABAergic neurons containing the calcium-binding protein parvalbumin in the rat hippocampus (CA1 region). Exp Brain Res 1988a; 72: 363–70.
Katsumaru H, Kosaka T, Heizmann CW, Hama K. Immunocytochemical study of GABAergic neurons containing the calcium-binding protein parvalbumin in the rat hippocampus. Exp Brain Res 1988b; 72: 347–62.
Kobayashi M, Buckmaster PS. Reduced inhibition of dentate granule cells in a model of temporal lobe epilepsy. J Neurosci 2003; 23: 2440–52.
Kosaka T, Hama K. Gap junctions between non-pyramidal cell dendrites in the rat hippocampus (CA1 and CA3 regions): a combined Golgi–electron microscopy study. J Comp Neurol 1985; 231: 150–61.
Kosaka T, Katsumaru H, Hama K, Wu JY, Heizmann CW. GABAergic neurons containing the Ca2+-binding protein parvalbumin in the rat hippocampus and dentate gyrus. Brain Res 1987; 419: 119–30.
Lehmann TN, Gabriel S, Kovacs R, Eilers A, Kivi A, Schulze K, et al. Alterations of neuronal connectivity in area CA1 of hippocampal slices from temporal lobe epilepsy patients and from pilocarpine-treated epileptic rats. Epilepsia 2000; 41: S190–4.
Lehmann TN, Gabriel S, Eilers A, Njunting M, Kovacs R, Schulze K, et al. Fluorescent tracer in pilocarpine-treated rats shows widespread aberrant hippocampal neuronal connectivity. Eur J Neurosci 2001; 14: 83–95.
Lewis DA, Lund JS. Heterogeneity of chandelier neurons in monkey neocortex: corticotropin-releasing factor- and parvalbumin-immunoreactive populations. J Comp Neurol 1990; 293: 599–615.
Lorente de No R. Studies on the structure of the cerebral cortex—II. Continuation of the study of the ammonic system. J Psychol Neurol 1934; 46: 113–77.
Lotstra F, Vanderhaeghen JJ. Distribution of immunoreactive cholecystokinin in the human hippocampus. Peptides 1987; 8: 911–20.
Maclean P. The limbic system concept. In: Trimble M, Bolwig T, editors. The temporal lobes and the limbic system. Wrighston Biomedical; 1992. p. 1–265.
Magloczky Z, Freund TF. Delayed cell death in the contralateral hippocampus following kainate injection into the CA3 subfield. Neuroscience 1995; 66: 847–60.
Magloczky Z, Halasz P, Vajda J, Czirjak S, Freund TF. Loss of calbindin-D28K immunoreactivity from dentate granule cells in human temporal lobe epilepsy. Neuroscience 1997; 76: 377–85.
Magloczky Z, Wittner L, Borhegyi Z, Halasz P, Vajda J, Czirjak S, et al. Changes in the distribution and connectivity of interneurons in the epileptic human dentate gyrus. Neuroscience 2000; 96: 7–25.
Marco P, Sola RG, Cayal SRY, DeFelipe J. Loss of inhibitory synapses on the soma and axon intial segment of pyramidal cells in human epileptic peritumoral neocortex: implications for epilepsy. Brain Res Bull 1997; 44: 47–66.
Martinez A, Lubke J, Del Rio JA, Soriano E, Frotscher M. Regional variability and postsynaptic targets of chandelier cells in the hippocampal formation of the rat. J Comp Neurol 1996; 376: 28–44.
Meldrum B. Excitotoxicity and epileptic brain damage. Epilepsy Res 1991; 10: 55–61.
Meldrum BS. Excitotoxicity and selective neuronal loss in epilepsy. Brain Pathol 1993; 3: 405–12.
Meyer AH, Katona I, Blatow M, Rozov A, Monyer H. In vivo labeling of parvalbumin-positive interneurons and analysis of electrical coupling in identified neurons. J Neurosci 2002; 22: 7055–64.
Miles R, Toth K, Gulyas AI, Hajos N, Freund TF. Differences between somatic and dendritic inhibition in the hippocampus. Neuron 1996; 16: 815–23.
Mody I, Otis TS, Staley KJ, Kohr G. The balance between excitation and inhibition in dentate granule cells and its role in epilepsy. Epilepsy Res Suppl 1992; 9: 331–9.
Morin F, Beaulieu C, Lacaille JC. Alterations of perisomatic GABA synapses on hippocampal CA1 inhibitory interneurons and pyramidal cells in the kainate model of epilepsy. Neuroscience 1999; 93: 457–67.
Nitsch R, Bergmann I, Kuppers K, Mueller G, Frotscher M. Late appearance of parvalbumin-immunoreactivity in the development of GABAergic neurons in the rat hippocampus. Neurosci Lett 1990; 118: 147–50.
Nusser Z, Hajos N, Somogyi P, Mody I. Increased number of synaptic GABA(A) receptors underlies potentiation at hippocampal inhibitory synapses. Nature 1998; 395: 172–7.
Penttonen M, Kamondi A, Acsady L, Buzsaki G. Gamma frequency oscillation in the hippocampus of the rat: intracellular analysis in vivo. Eur J Neurosci 1998; 10: 718–28.
Rafiq A, DeLorenzo RJ, Coulter DA. Generation and propagation of epileptiform discharges in a combined entorhinal cortex/hippocampal slice. J Neurophysiol 1993; 70: 1962–74.
Ramon y Cajal S. Histologie du systeme nerveux de l'homme et des vertebres. Vol. I, II. Paris: Maloine; 1909–1911.
Represa A, Tremblay E, Ben-Ari Y. Sprouting of mossy fibers in the hippocampus of epileptic human and rat. Adv Exp Med Biol 1990; 268: 419–24.
Ribak CE, Seress L, Leranth C. Electron microscopic immunocytochemical study of the distribution of parvalbumin-containing neurons and axon terminals in the primate dentate gyrus and Ammon's horn. J Comp Neurol 1993; 327: 298–321.
Rosene DL, Van Hoesen GW. The hippocampal formation of the primate brain: a review of some comparative aspects of cytoarchitecture and connections. In: Jones EG, Peters A, editors. Cerebral cortex. Vol. 6. New York: Plenum Press; 1987. p. 345–456.
Sano K, Kirino T. Ammon's horn sclerosis: its pathogenesis and clinical significance. Tohoku J Exp Med 1990; 161 Suppl: 273–95.
Sayin U, Osting S, Hagen J, Rutecki P, Sutula T. Spontaneous seizures and loss of axo-axonic and axo-somatic inhibition induced by repeated brief seizures in kindled rats. J Neurosci 2003; 23: 2759–68.
Scotti A, Bollag O, Kalt G, Nitsch C. Loss of perikaryal parvalbumin immunoreactivity from surviving GABAergic neurons in the CA1 field of epileptic gerbils. Hippocampus 1997a; 7: 524–35.
Scotti AL, Kalt G, Bollag O, Nitsch C. Parvalbumin disappears from GABAergic CA1 neurons of the gerbil hippocampus with seizure onset while its presence persists in the perforant path. Brain Res 1997b; 760: 109–17.
Seress L. Interspecies comparison of the hippocampal formation shows increased emphasis on the regio superior in the Ammon's horn of the human brain. J Hirnforsch 1988; 29: 335–40.
Seress L, Frotscher M. Basket cells in the monkey fascia dentata: a Golgi/electron microscopic study. J Neurocytol 1991; 20: 915–28.
Seress L, Gulyas AI, Freund TF. Parvalbumin- and calbindin D28k-immunoreactive neurons in the hippocampal formation of the macaque monkey. J Comp Neurol 1991; 313: 162–77.
Seress L, Gulyas AI, Ferrer I, Tunon T, Soriano E, Freund TF. Distribution, morphological features, and synaptic connections of parvalbumin- and calbindin D28k-immunoreactive neurons in the human hippocampal formation. J Comp Neurol 1993; 337: 208–30.
Sloviter RS, Sollas AL, Barbaro NM, Laxer KD. Calcium-binding protein (calbindin-D28K) and parvalbumin immunocytochemistry in the normal and epileptic human hippocampus. J Comp Neurol 1991; 308: 381–96.
Sommer W. Erkrankung des Ammonshornes als Aetiologisches Moment der Epilepsie. Arch Psychiat Nervkrankh 1880; 10: 631–75.
Somogyi P, Nunzi MG, Gorio A, Smith AD. A new type of specific interneuron in the monkey hippocampus forming synapses exclusively with the axon initial segments of pyramidal cells. Brain Res 1983; 259: 137–42.
Spencer DD, Spencer SS. Surgery for epilepsy. Neurol Clin 1985; 3: 313–30.
Sutula TP, Cascino G, Cavazos JE, Ramirez L. Mossy fiber synaptic reorganization in the epileptic human temporal lobe. Ann Neurol 1989; 26: 321–30.
Tamas G, Buhl EH, Lrincz A, Somogyi P. Proximally targeted GABAergic synapses and gap junctions synchronize cortical interneurons. Nat Neurosci 2000; 3: 366–71.
Tortosa A, Ferrer I. Parvalbumin immunoreactivity in the hippocampus of the gerbil after transient forebrain ischaemia: a qualitative and quantitative sequential study. Neuroscience 1993; 55: 33–43.
Wittner L, Magloczky Z, Borhegyi Z, Halasz P, Toth S, Erss L, et al. Preservation of perisomatic inhibitory input of granule cells in the epileptic human dentate gyrus. Neuroscience 2001; 108: 587–600.
Wittner L, Erss L, Szabo Z, Toth S, Czirjak S, Halasz P, et al. Synaptic reorganization of calbindin-positive neurons in the human hippocampal CA1 region in temporal lobe epilepsy. Neuroscience 2002; 115: 961–78.
Zhu ZQ, Armstrong DL, Hamilton WJ, Grossman RG. Disproportionate loss of CA4 parvalbumin-immunoreactive interneurons in patients with Ammon's horn sclerosis. J Neuropathol Exp Neurol 1997; 56: 988–98.(L. Wittner, L. Erss, S. Czirjak, P. Hala)