We used in situ hybridization techniques to study the distribution of neurones synthesizing somatostatin mRNA and neuropeptide Y mRNA in the hilar region of the hippocampal formation of patients with temporal lobe epilepsy. In the dentate gyrus, somatostatin mRNA- and neuropeptide Y mRNA-synthesizing neurones were found to be exclusively located within the hilar region. Unlike animal models, no ectopic expression of either peptide was found in principal cells. The numbers of hilar interneurones expressing somatostatin mRNA and neuropeptide Y mRNA were compared with the degree of hilar cell loss determined by immunohistochemistry against neuronal nuclear antigen. The numbers of somatostatin and neuropeptide Y mRNA-synthesizing neurones varied considerably between patients, but both were found to be highly correlated to the total number of neuronal nuclear antigen-immunoreactive hilar neurones. These results suggest that loss of somatostatin and neuropeptide Y interneurones occurs in proportion to overall hilar cell loss, and therefore the hypothesis of a selective loss of these interneurones in temporal lobe epilepsy seems unlikely.
GAD = glutamate decarboxylase
HIPP = hilar–perforant path (cells)
ISH = in situ hybridization
NeuN = neuronal nuclear antigen
TLE = temporal lobe epilepsy
A disruption in the balance between excitation and inhibition has frequently been advanced as an explanation for chronic epilepsy over the last century (e.g. Gowers, 1881). Several studies have been carried out over the last few decades to determine levels of excitatory and inhibitory neurotransmitters in the brains of epileptic patients, but studies at this level have not provided any clear-cut results (Glass and Dragunow, 1995). Morphological studies on tissue resected from patients undergoing surgery for the treatment of temporal lobe epilepsy (TLE) have, however, provided a number of useful insights into the structural changes associated with this condition, which might affect the balance between excitatory and inhibitory circuitry.
Tissue from patients diagnosed as having TLE often bears a characteristic neuropathological scar, in which a profound loss of hippocampal pyramidal cells is observed as well as a reduction in local circuit neurones, many of which are assumed to be inhibitory in nature (Magerison and Corsellis, 1966; Dam, 1980; Babb and Brown, 1987; Meldrum, 1997). This is often accompanied by distorted connectivity of excitatory neurones, such as the sprouting of granule cell axons (mossy fibres) into the inner molecular layer of the dentate gyrus (Sutula et al., 1989) and the dispersion of the remaining granule cells, resulting in widening of the stratum granulosum (Houser, 1990). The degree of mossy fibre sprouting and dispersion of dentate granule cells correlates with the degree of hilar cell loss, suggesting a link between these parameters (Lurton et al., 1997; El Bahh et al., 1999).
It has been hypothesized that a precipitating factor in triggering such reorganization in the dentate gyrus may be the loss of hilar interneurones, particularly hilar mossy cells and somatostatin- and neuropeptide Y-expressing interneurones (Sloviter, 1987; Robbins et al., 1991). Several experimental studies have suggested that hilar interneurones are differentially vulnerable to experimentally induced seizures. For instance, chronic stimulation of the perforant path input to the dentate gyrus in rats resulted in a graded response in which hilar interneurones were damaged before pyramidal cells (Sloviter, 1983). Enhancing GABAergic function by vigabatrin treatment has been shown to reduce the degree of loss of principal cells in this model (Ylinen et al., 1991), also suggesting a precipitatory role for reduced inhibition in the generation of hippocampal damage. In the chronic perforant path stimulation model, a marked decrease in immunoreactivity for somatostatin and neuropeptide Y interneurones correlates with a reduction in functional inhibition (Sloviter, 1987, 1991). The great majority of neuropeptide Y and somatostatin-immunoreactive interneurones co-localize GABA (Freund and Buszaki, 1996), suggesting that their loss could be directly responsible for this functional disinhibition. Decreases in the numbers of neurones that are immunoreactive for somatostatin and neuropeptide Y have also been reported in many other animal models of limbic epilepsy, such as animals treated with kainic acid (Sperk et al., 1992) and animals with tetanus toxin-induced seizures (Mitchell et al., 1995), and in other pathologies such as those following global forebrain ischaemia (Johansen et al., 1987) and traumatic brain injury (Lowenstein et al., 1992). These results raise the possibility that hilar interneurones could be more vulnerable than principal cells to `overstimulation' such as that which might occur during seizures and that the loss of these could precede other structural changes in the hippocampus.
A reduction in somatostatin and neuropeptide Y-immunoreactive neurones has also been suggested to be present in human patients with TLE (de Lanerolle et al., 1989; Robbins et al., 1991; Mathern et al., 1995). De Lanerolle and colleagues first reported almost total loss of immunoreactivity for somatostatin and neuropeptide Y interneurones, confined almost exclusively to hilar interneurones, in cryptogenic TLE, compared with patients experiencing tumour-related seizures (de Lanerolle et al., 1989). Mathern and colleagues quantified this loss of immunoreactive cells as a percentage of the total number of hilar neurones, determined from cresyl violet-stained sections, and concluded that the number of somatostatin but not of neuropeptide Y neurones was selectively reduced (Mathern et al., 1995). Because both peptides co-localize glutamate decarboxylase (GAD), these results are difficult to reconcile with the finding that GAD-immunoreactive inhibitory interneurones are not lost relative to other hilar interneurones in humans with TLE (Babb et al., 1989; Meldrum et al., 1997). Selective loss of somatostatin and neuropeptide Y interneurones, with respect to overall hilar cell loss, therefore remains to be clearly established in tissue from humans with epilepsy.
Both de Lanerolle and colleagues and Mathern and colleagues have remarked on the high degree of plasticity in the staining patterns of these neuropeptides (de Lanerolle et al., 1989; Mathern et al., 1995) and similar observations can be made regarding experimentally induced seizures in animal models (e.g. Mitchell et al., 1995). It is well known that experimentally induced seizures can cause release of these peptides (Vezzani et al., 1996), and therefore in situ hybridization studies provide a more sensitive method for quantifying numbers of neuropeptide-synthesizing neurones, because release would not be expected to reduce expression patterns.
In the present study, we used radiolabelled oligonucleotide probes to detect neurones synthesizing somatostatin and neuropeptide Y messenger RNA (mRNA) in 20 patients with TLE, and compared the density of somatostatin- and neuropeptide Y-expressing neurones with the total numbers of hilar neurones determined by neuronal nuclear antigen (NeuN) immunocytochemistry. The objectives of this study were to determine more accurately the distribution of somatostatin and neuropeptide Y neurones in tissue from patients with TLE, and more specifically the relationship between somatostatin and neuropeptide Y neurone densities with respect to total numbers of hilar neurones.
Twenty patients were included in the study, all of whom had a history of recurrent seizures and were diagnosed as having TLE. The age of onset of epilepsy varied from 2 to 20 years (mean 10.4 years) and the age at surgery varied from 11 to 36 years (mean 26.8 years). In 12 cases, records showed evidence of an initial precipitating injury and in the other cases none was recorded (Table 1).
Human hippocampal tissue was obtained during surgery at the Centre Hospitalier Régional de Bordeaux (Professor A. Rougier). The hippocampus was removed en bloc and processed immediately for in situ hybridization. The extent of the resection was not standardized, but all tissue was cut into approximately 5–10 mm blocks and fixed by immersion in 1% p-formadehyde in 0.1 M phosphate buffer, pH 7.2, at 4°C for 24 h under gentle agitation. In all cases, the interval between removal and fixation was <1 h. Blocks were then left in the same phosphate buffer with 15% sucrose for a further 24 h at 4°C, frozen and stored at –80°C until required, as described previously (Brana et al., 1995). Serial 10 μm frozen sections were cut using a cryostat, collected on APES (3-aminopropyltriethoxysilane)-coated slides and stored at –80°C until used. Sections were either stained directly with toluidine blue or processed for immunocytochemistry or in situ hybridization.
Probe characteristics and in situ hybridization
Synthetic radiolabelled oligonucleotide probes were used to detect somatostatin and neuropeptide Y mRNAs in the human tissue and have been described and characterized by Brana and colleagues (Brana et al., 1995). Human somatostatin mRNA-synthesizing neurones were identified using a synthetic 48-mer oligonucleotide probe recognizing human and rat preprosomatostatin (Funckes et al., 1983) consisting of the sequence 5′-CCA GAA GAA GTT CTT GCA GCC AGC TTT GCG TTC CCG GGG TGC CAT-3′. Human neuropeptide mRNA-synthesizing neurones were similarly identified using a 36-mer-specific synthetic radiolabelled oligonucleotide probe recognizing human neuropeptide Y (Minth et al., 1984), consisting of the sequence 5′-CGA GTA GTA TCT GGC CAT (GTC CTC CGC TGG TGC GTC-3′. Both probes were synthesized commercially (Oswell DNA Services, Southampton, UK) and radiolabelled in our laboratory by tailing with [35S]dATP (NEN; >1000 Ci/mmol). Adjacent sections were hybridized as described previously (Brana et al., 1995).
Slides were exposed in contact with X-Omat film (Kodak) for 1 week (somatostatin) or 2 weeks (neuropeptide Y) and then coated with Ilford K-5 emulsion. Emulsion-dipped sections were left in the dark for up to 12 weeks, developed, lightly counterstained with toluidine blue and mounted. Controls for specificity of the in situ hybridization (ISH) signal included the use of hybridization buffer without radioactive probe and the addition of unlabelled oligonucleotides to the hybridization buffer as described by Le Moine and colleagues (Le Moine et al., 1990).
Determination of hilar neurones by NeuN immunocytochemistry
Hilar neurones were identified using a mouse monoclonal antibody directed against neuronal nuclei (NeuN, Chemicon, USA). Sections were defrosted, washed in TBS (Tris-buffered saline), pH 7.4, incubated with 5% NHS (normal horse serum) in TBS/0.1% Triton X-100 for 1 h at room temperature and then incubated overnight at 4°C with the primary antibody at a dilution of 1 : 200 (2.5 μg/ml protein) in (TBS/0.1% Triton X-100)/1% NHS. Immunoreactivity was revealed with an immunoperoxidase procedure using biotinylated anti-mouse IgG (1 : 200; Amersham, Amersham, UK) and avidin–biotin peroxidase complex (ABC kit: Dako, Glostrup, Denmark). Peroxidase activity was revealed with 3,3′-diaminobenzidine (Sigma, Poole, UK) and sections were then dehydrated, mounted and observed under a light microscope. To validate NeuN staining as a reliable method for detecting cell loss, we compared results from this study with quantified hilar cell counts performed in a previous study (Lurton et al., 1997) on toluidine blue-stained sections from the same patients.
The hilar region was identified on all NeuN-stained sections and all the emulsion-dipped sections that had been processed for ISH. A coverslip was placed on top of each slide between the upper and lower blades of the stratum granulosum and secured in place with nail varnish. The area bounded by granule cells up to the line delimited by the coverslip was defined as the hilar region. All cell counts were performed on this area and the surface area of the hilar region on each slide was determined using a digital image analysis system (Open Lab; Improvision, Coventry, UK). The total numbers of NeuN-immunoreactive cells and of cells positive for somatostatin and neuropeptide Y mRNA in the hilar region were counted under light microscopy (×40 objective) on two non-contiguous sections by at least three investigators. We elected not to use the physical dissector to evaluate ISH data because it is difficult to obtain a reliable signal on adjacent sections and we used the same method to estimate the numbers of NeuN-immunoreactive cells in order to standardize our methods of estimating cell density.
In the case of somatostatin ISH, two separate ISH runs were performed to verify the reproducibility of the ISH signal and, in this case, because no significant differences were found between the two runs (r = 0.93 P < 0.0001), the data were pooled. All individual results from ISH for somatostatin and neuropeptide Y and from NeuN immunocytochemistry were corrected for the area of the hilus on each slide and the data were expressed as the number of neurones per square millimetre. For each section, we determined the numbers of cells positive for NeuN, somatostatin mRNA and neuropeptide Y mRNA per square millimetre and then pooled the data to obtain a value for each patient. Data were analysed by regression analysis with a statistical analysis software package (Prism-GraphPad Software, San Diego, USA).
The distributions of neurones synthesizing neuropeptide Y mRNA and somatostatin mRNA could be identified clearly on emulsion-dipped sections and in both cases there was a clear intrahilar distribution within the dentate gyrus itself. Strongly positive neurones were also seen in other hippocampal regions, particularly in the CA1 and subicular areas and these again showed clear evidence of a distribution consistent with that reported for interneurones (data not shown), similar to that seen with immunohistochemistry (Chan-Palay, 1987; Mathern et al., 1995).
The distributions of neuropeptide Y mRNA- and somatostatin mRNA-positive neurones were very similar to each other, being particularly prevalent within the hilar region itself (Fig. 1), but with the occasional cells located in the subgranular zone. This distribution would be consistent with labelling of mainly cells associated with the hilar–perforant path (HIPP cells) (Han et al., 1993) and a few basket cells (Freund and Buszaki, 1996).
In situ hybridization histochemistry for neuropeptide Y mRNA and somatostatin mRNA in the hilar region of tissue from patients with TLE. Detection of 35S-labelled oligonucleotide probes for neuropeptide Y and somatostatin on emulsion-dipped preparations counterstained with toluidine blue. (A) Distribution of neuropeptide Y mRNA-synthesizing neurones within the dentate hilus. Scale bar = 500 μm. (B) Higher magnification (scale bar = 100 μm) showing the clear intrahilar distribution of neuropeptide Y mRNA-synthesizing neurones. (C) Higher magnification still (scale bar = 50 μm), showing labelling pattern for somatostatin mRNA-positive hilar cells.
In some animal models, in addition to interneurones, granule cells have been shown to express neuropeptide Y mRNA after experimentally induced seizures (e.g. Gruber et al., 1994). This expression pattern is not apparent in naïve animals and, because neuropeptide Y has potent anticonvulsant actions in vitro, this could represent an endogenous neuroprotective/anticonvulsant mechanism (Vezzani et al., 1999). Previous studies using immunocytochemistry have revealed no labelling of human granule cells with antisera against neuropeptide Y (Mathern et al., 1995; Chan-Palay et al., 1987), and therefore it was of interest to determine whether, in our case, using the more sensitive technique of ISH, we were able to detect any expression in granule cells. In all the sections tested to date we have found no evidence for the expression of either neuropeptide Y mRNA or somatostatin mRNA in granule cells by ISH.
Hilar neurones were identified by immunocytochemistry using an antibody (NeuN) that specifically recognizes nuclei of neurones (Fig. 2). This method offers greater accuracy than methods used previously, which were based primarily on Nissl staining, because there is less possibility of accidentally including non-neuronal cells. Both principal neurones (i.e. granule cells) and interneurones were labelled with the NeuN antibody (Fig. 2B). In the case of deep hilar neurones these could be identified clearly (Fig. 2A); however, as with conventional Nissl staining, we had to rely on morphological criteria to distinguish basket cells lying in the subgranular zone from granule cells (Fig. 2B). Dispersion of granule cells was also clearly evident from NeuN staining in many cases, with a similar appearance to that reported by Lurton and colleagues using ISH for dynorphin (Lurton et al., 1997). As the present paper is the first report, to our knowledge, that has specifically addressed the issue of determining numbers of hilar neurones by staining with an antibody against neuronal nuclei, we compared our results with those published in a previous study in which hilar cell numbers were determined by more classical histochemical methods. As expected, NeuN-positive hilar neurones were highly correlated with toluidine blue-stained hilar cells (r = 0.69, P < 0.001), confirming that counting NeuN-stained cells reflects the total number of hilar neurones determined on Nissl-stained tissue (Fig. 3).
Visualization of hilar interneurones and principal neurones on 10 μm cryostat sections in the dentate gyrus of human hippocampi by immunohistochemistry using an antibody that specifically detects a neuronal nuclear antigen (NeuN). (A) Low-power (scale bar = 100 μm) photomicrograph of typical staining pattern of hilar neurones. (B) Higher magnification (scale bar = 50 μM) showing the boundary between granule cells and hilar neurones. Note the clear labelling of neuronal cell bodies compared with equivalent toluidine blue-stained section (Fig. 1B).
Linear regression analysis of somatostatin mRNA- and neuropeptide-Y mRNA-synthesizing neurones per square millimetre in the dentate hilus as a function of total number of hilar interneurones determined by NeuN immunohistochemistry. Dotted lines represent 95% confidence limits; the solid line is the linear regression line. (A) Relationship between somatostatin mRNA-expressing cells in the hilus and NeuN-immunoreactive hilar interneurones. (B) Relationship between neuropeptide Y mRNA-expressing cells in the hilus and NeuN-immunoreactive hilar interneurones. (C) Relationship between neuropeptide Y mRNAand somatostatin mRNA-synthesizing hilar interneurones. To validate the NeuN method for determining hilar interneurones, we compared results for NeuN staining with previously reported values for the total numbers of CA4 neurones determined by toluidine blue staining (CA4 total) on the same material, which we have reported in a previous publication (Lurton et al., 1997) (see table). (D) Relationship between hilar neurones determined by NeuN immunocytochemistry compared with total numbers of hilar neurones determined by Nissl staining.
Regression analysis was also performed to determine the relationship between somatostatin mRNA- and neuropeptide Y mRNA-synthesizing hilar neurones and the total number of hilar neurones determined by NeuN immunohistochemistry. There was a high degree of correlation between somatostatin mRNA- and neuropeptide Y mRNA-positive cells (r = 0.82, P < 0.0001) (Fig. 3). This very high degree of correlation, which can be seen clearly in Fig. 3C, could be explained partially by the fact that neuropeptide Y and somatostatin may, in fact, be expressed to a large degree in the same cellular population (Kohler et al., 1987; Freund and Buszaki 1996). Unfortunately, within the constraints of these experiments it was not possible to carry out double-labelling studies to confirm the degree of co-localization, since we had to rely on radioactively labelled probes to achieve sufficient sensitivity.
The major finding of our study is that the number of cells positive for somatostatin mRNA and the number positive for neuropeptide Y mRNA were correlated with the total number of hilar neurones determined either by NeuN histochemistry (somatostatin mRNA, r = 0.779, P < 0.0001; neuropeptide Y mRNA, r = 0.51, P < 0.05) or toluidine blue staining (somatostatin mRNA, r = 0.851, P < 0.0001; neuropeptide Y mRNA, r = 0.814, P < 0.0001) (Fig. 3).
The main finding of the present study is that both somatostatin mRNA and neuropeptide Y mRNA-synthesizing neurones in the hilus of patients with TLE correlate strongly with the total numbers of hilar neurones, determined either by NeuN immunohistochemistry or by toluidine blue staining. This suggests that, as with numbers of GAD-positive inhibitory interneurones (Babb et al., 1989), the loss of somatostatin and neuropeptide Y interneurones is proportional to overall hilar cell loss. Our results indicate that, while neuropeptide Y mRNA- and somatostatin mRNA-positive neurones are clearly reduced in some patients with TLE compared with others, the hypothesis of a selective loss of these interneurones with respect to overall hilar cell loss seems unlikely from our data.
These results contrast with some earlier findings relying on immunocytochemistry for determining the densities of somatostatin- and neuropeptide Y-positive hilar neurones. De Lanerolle and colleagues originally reported a highly significant reduction in both hilar somatostatin-immunoreactive interneurones in 29 out of 35 patients with cryptogenic TLE (de Lanerolle et al., 1989). Similarly, Mathern and colleagues also reported a significant reduction in numbers of somatostatin-immunoreactive neurones in patients with hippocampal sclerosis compared with patients without sclerosis and post-mortem tissue from non-epileptic patients (Mathern et al., 1995). Both of these studies remarked on the impressive degree of plasticity seen in terminal fibre staining, suggesting that levels of the peptide may be altered and that sprouting of terminal fibres may have occurred. A similar degree of apparent reactive plasticity was also reported in both of these studies with respect to neuropeptide Y immunoreactivity.
In experimental studies, Vezzani and colleagues have demonstrated increased potassium-evoked release of both peptides in the hippocampus and entorhinal cortex after kainic acid-induced seizures in the rat (Vezzani et al., 1996). This indicates that seizures may affect the levels of peptides found in interneurones and this could affect the ability to detect the peptides by immunocytochemical methods. We have observed previously that blocking release mechanisms with tetanus toxin considerably enhances immunocytochemical staining for neuropeptide-Y in organotypic hippocampal slice cultures (Mitchell et al., 1996), further suggesting that release of the peptide does affect the ability to detect it using immunocytochemistry. The increased sensitivity of ISH methods over immunocytochemical detection methods is also clear with respect to expression of neuropeptide Y in granule cells, for instance where neuropeptide Y is readily detected in granule cells by ISH but only sporadically by immunocytochemistry (Sperk et al., 1992; Lurton et al., 1996).
Robbins and colleagues addressed the issue of the depletion of somatostatin immunoreactivity in human epileptic tissue by using an in situ approach on a limited number of patients (Robbins et al., 1991). While their results apparently confirmed a reduction in somatostatin mRNA-synthesizing neurones, the results were not correlated to the overall hilar cell loss and the precise localization of the somatostatin signal is difficult to assess as no data were presented at the cellular level.
In our study, we elected not to compare results from our epileptic patients with data from post-mortem tissue, but rather to compare results with overall cell loss within the same homogeneous population. One reason for this choice was the fact that mRNA levels are known to decay with post-mortem delay. Thus, signal strength, and hence the ability to detect mRNA for these peptides, is proportional to the delay before fixation and freezing of the tissue. It is virtually impossible to obtain sufficient control material from autopsy with equivalent post-mortem delay periods and with standardized preservation techniques. All of our samples were therefore collected and treated identically to optimize the tissue bank for in situ studies and therefore should represent a population in which postoperative delay is standardized.
Several experimental studies have reported the ectopic expression of neuropeptide Y in granule cells in a variety of animal models (Sperk et al., 1992; Chafetz et al., 1995; Lurton et al., 1997; Vezzani et al., 1999). The functional significance of such neo-expression in animal models is unclear, but it has been suggested that this may represent an endogenous damping mechanism to limit excitability in the face of excessive epileptogenic discharges. This hypothesis suggests that neuropeptide Y released from dense core vesicles under high-frequency stimulation may act to reduce glutamate release at mossy fibre synapses (for review, see Vezzani et al., 1999). A critical role for neuropeptide Y in controlling seizure activity is also suggested by the fact that mice lacking neuropeptide Y have a reduced seizure threshold to convulsants (Erikson et al., 1996). We found no evidence of any ectopic expression of neuropeptide Y in granule cells, which tends to suggest that this mechanism may not be operational in humans. It must be stressed, however, that our data only pertain to a specific time point within the epileptogenic process and are therefore of limited value in this respect.
Whereas it seems unlikely from our data that selective loss of somatostatin mRNA- and neuropeptide Y mRNA-synthesizing neurones occurs in TLE, there is nevertheless significant loss of these neurones in parallel with the loss of other hilar interneurones. In this respect, the loss of different types of hilar interneurones may have different functional effects. Our data do not indicate whether the loss of somatostatin and neuropeptide Y interneurones results in the loss of functional inhibition; however, one might expect this loss to have functional consequences because somatostatin and neuropeptide Y interneurones are known to co-localize GABA and are therefore likely to be inhibitory in nature. Neuropeptide Y neurones are believed to represent a slightly more heterogeneous population of interneurones than somatostatin interneurones, which are probably exclusively of the HIPP variety (Freund and Buszaki, 1996). The high degree of correlation between somatostatin and neuropeptide Y interneurone loss in our study suggests that it may be precisely those that co-localize both peptides that are affected in TLE. This would almost certainly represent HIPP cells (Han et al., 1993), which are known to have cell bodies in the hilar region, but whose axons are believed to project mainly to the outer molecular layer, innervating mainly granule cell dendrites (Han et al., 1993; Freund and Buszaki, 1996). Interestingly, the dendritic tree of these cells corresponds exclusively to the area occupied by mossy fibre collaterals in the hilus. These neurones are therefore unlikely to participate in commissurally or associationally evoked inhibition, because these inputs are predominantly to the inner molecular layer, but probably represent interneurones involved in feed-back inhibition. Loss of these interneurones could therefore be expected to result in a reduction in recurrent inhibition in the dentate gyrus and in this way could contribute towards hyperexcitability.
Electrophysiological techniques can be used to determine the functional characteristics of epileptic tissue and it may be possible in the long run to determine whether excitatory or inhibitory neurotransmission is functionally altered in humans with TLE. So far, however, the interpretation of the results of electrophysiological studies has met with difficulties similar to those encountered in anatomical studies, particularly with respect to the standardization of tissue preparation and the lack of control material (Avoli and Williamson, 1996).
Brana C, Charron G, Aubert I, Carles D, Martin-Negrier ML, Trouette H, et al. Ontogeny of the striatal neurons expressing neuropeptide genes in the human fetus and neonate. J Comp Neurol1995; 360: 488–505.
Chan-Palay V. Somatostatin immunoreactive neurons in the human hippocampus and cortex shown by immunogold/silver intensification on vibratome sections: coexistence with neuropeptide Y neurons, and effects in Alzheimer-type dementia. J Comp Neurol1987; 260: 201–23.
Chafetz RS, Nahm WK, Noebels JL. Aberrant expression of neuropeptide Y in hippocampal mossy fibers in the absence of local cell injury following the onset of spike-wave synchronization. Brain Res Mol Brain Res1995; 31: 111–21.
El Bahh B, Lespinet V, Lurton D, Coussemaq M, Le Gal La Salle G, Rougier A. Correlation between granule cell dispersion, mossy fibre sprouting and hippocampal cell loss in temporal lobe epilepsy. Epilepsia1999; 40: 1393–401.
Han ZS, Buhl EH, Lorinczi Z, Somogyi P. A high degree of spatial selectivity in the axonal and dendritic domains of physiologically identified local-circuit neurons in the dentate gyrus of the rat hippocampus. Eur J Neurosci1993; 5: 395–410.
Kohler C, Eriksson LG, Davies S, Chan-Palay V. Co-localization of neuropeptide tyrosine and somatostatin immunoreactivity in neurons of individual subfields of the rat hippocampal region. Neurosci Lett1987; 78: 1–6.
Lowenstein DH, Thomas MJ, Smith DH, McIntosh TK. Selective vulnerability of dentate hilar neurons following traumatic brain injury: a potential mechanistic link between head trauma and disorders of the hippocampus. J Neurosci1992; 12: 4846–53.
Lurton D, Coussemacq M, Barrow P, Sundstrom LE, Rougier A. Widespread ectopic neuropeptide-Y immunoreactivity in contralateral mossy fibres after a unilateral intrahippocampal kainic acid injection in the rat. Neurosci Lett1996; 213: 181–4.
Mathern GW, Babb TL, Pretorius JK, Leite JP. Reactive synaptogenesis and neuron densities for neuropeptide Y, somatostatin, and glutamate decarboxylase immunoreactivity in the epileptogenic human fascia dentata. J Neurosci1995; 15: 3990–4004.
Sloviter RS. `Epileptic' brain damage in rats induced by sustained electrical stimulation of the perforant path. I. Acute electrophysiological and light microscopic studies. Brain Res Bull1983; 10: 675–97.
Sloviter RS. Permanently altered hippocampal structure, excitability, and inhibition after experimental status epilepticus in the rat: the `dormant basket cell' hypothesis and its possible relevance to temporal lobe epilepsy. Hippocampus1991; 1: 41–66.
Sperk G, Marksteiner J, Gruber B, Bellmann R, Mahata M, Ortler M. Functional changes in neuropeptide Yand somatostatin-containing neurons induced by limbic seizures in the rat. Neuroscience1992; 50: 831–46.
Vezzani A, Monhemius R, Tutka P, Milani R, Samanin R. Functional activation of somatostatin- and neuropeptide Y-containing neurons in the entorhinal cortex of chronically epileptic rats. Neuroscience1996; 75: 551–7.
Ylinen AM, Miettinen R, Pitkanen A, Gulyas AI, Freund TF, Riekkinen PJ. Enhanced GABAergic inhibition preserves hippocampal structure and function in a model of epilepsy. Proc Natl Acad Sci USA1991; 88: 7650–3.