doi: 10.1053/ejpn.2000.0450 available online at http://www.idealibrary.com on II1|~1" European Journal of Paediatric Neurology 2001; 5(Suppl. A): 135-142
ORIGINAL
ARTICLE
Changes in GABAergic neuron distribution in situ and in neuron cultures in ovine (OCL6) Batten disease MANFRED J OSWALD, GRAHAM W KAY, DAVID N PALMER Animal and Food Sciences Division, Lincoln University, Canterbury, New Zealand
The neuronal ceroid lipofuscinoses (NCLs) are a group of inherited human and animal diseases characterized by progressive brain atrophy. A form in sheep is syntenic to the human CLN6 disease. Cell type specific neurodegeneration in these sheep was indicated by the distribution of GABAergic interneurons in coronal sections of normal and CLN6 affected sheep brains. A reduction of parvalbumin immunoreactive neurons in NCL cerebral cortex was the most striking feature. This was most pronounced in parietal cortex where very few positive cells remained. Calretinin immunoreactive somata in infragranular layers of the neocortex were also reduced while the number of calbindin positive cells was similar in affected and normal brains. There were fewer GAD immunoreactive neurons in the deeper layers of all NCL cortical areas examined. The parietal lobe was relatively more affected than frontal or temporal lobes while the cerebellum and the basal ganglia showed no signs of selective neuron loss. Since horizontally extending basket cells are mainly labelled by parvalbumin, the loss of these interneurons in the neocortex may render pyramidal neurons more excitable and compromise their co-ordinated output. In vitro, cultures of control and affected neurons from 60 to 70-day-old fetal brain hemispheres were examined for the presence of GABAergic and glutamatergic neurons. Different neurons developed distinct immunoreactivity to glutamate or GABA but the overall distribution was similar in normal and affected cultures. This culture system may provide a useful model to compare GABAergic cell function of normal and NCL affected neurons. Keywords: Batten disease.Sheep. Neocortex. Inhibitory interneurons. Parvalbumin. Neuron culture.
Introduction Naturally occurring animal models offer an invaluable tool for studying the pathophysiology of inherited diseases. Most studied in the neuronal ceroid lipofuscinoses (NCLs, Batten disease) is the form in a flock of New Zealand South Hampshire sheep that are bred and maintained as a model of the human diseases. The pathological features of this ovine NCL form (OCL) have been well described in the original diagnosis 1,2 and subsequent investigations.3-7 Clinical symptoms and the ultrastructure of the storage bodies in cerebral neurons, a mixture of curvilinear and fingerprint profiles, are similar to the human CLN6 variant late infantile form of NCL. s,9 Linkage studies have placed the genetic lesion at sheep chromosome
7q13-15, which is syntenic with the h u m a n CLN6 location on chromosome 15q 21-23. l° Severe brain atrophy and blindness are characteristic of the NCLs. In the ovine model atrophy is pronounced in the retina and the neocortex, affecting certain layers more than others. 6,7 The remaining structures of the central nervous system are relatively spared. Previous studies on neuron types in several forms of NCL suggested that these diseases affect y-aminobutyric acid (GABA)ergic cells but not cholinergic cells. 7,11,12 Lysosomal storage of protein is also characteristic of the NCLs. Although the accumulation of subunit c of mitochondrial ATP synthase occurs in most cell types including neurons in most forms of NCL 13-14 only certain classes of neurons appear to be affected and degenerate.
Correspondence: D N Palmer, Animal and Food Sciences Division, Lincoln University, PO Box 84, Canterbury, New Zealand. e-mail:
[email protected] 10900-3798/01/05/A135+8 $35.00
© 2001 European Paediatric Neurology Society
Original article: M J Oswald et al.
136
Little is known about the pre-clinical neuropathology of any form of Batten disease. A unique advantage of animal models is that the development of pathology can be studied from a presymptomatic state. Primary medium to long term fetal sheep neuron cultures have been established for the first time, 15 and the hallmark storage of subunit c has been reported in affected cells. 16 The purpose of this investigation is to describe the in situ progression of the disease at a cellular level, and to establish a neuron culture system to study the cell biology of the affected neuron types. We report on the disease specific loss of a subgroup of GABAergic neurons and evidence that sheep neurons in culture acquire specific functional features.
Materials and methods Affected and normal sheep were maintained under normal N e w Zealand pastoral conditions. Affected sheep were diagnosed by brain biopsy. 17 Brains were fixed by whole body perfusion with 4% paraformaldehyde, performed under general anaesthetic. The brains were removed, fixed for a further 48 h in 4% paraformaldehyde, then equilibrated with 25% sucrose, frozen rapidly with dry ice and stored at - 8 0 ° C until cryosectioned. Four sets of 50/~m thick coronal sections were taken; across the frontal lobe; to expose the basal ganglia and surrounding cortex; through the parietal and temporal cortices including the hippocampus; and across the cerebellum. Cryosections were stained by the Nissl method or an immunohistochemical procedure. TM Antibodies used were monoclonal antibodies to parvalbumin (SWant, Bellinzona, Switzerland) 1:5000, or to calbindin (SWant) 1:2000; rabbit antiserum to calretinin (SWant) 1:5000, or to glutamate decarboxylase (GAD; Chemicon, Temecula, CA) 1:5000. All antibodies were diluted in 1% goat serum in phosphate buffered saline (PBS), p H 7.4, also containing 0.2% Triton X-100 and 0.04% Thimerosal (Sigma, St Louis, MO). Cryosections were pre-incubated in 1% H202 in 50% methanol for 20 minutes at 20°C and then in primary antibody for 3 days at 4°C. Immunostains were developed Using biotinylated goat anti-mouse IgG (Sigma) 1:500, or donkey anti-rabbit IgG (Amersham, Arlington Heights, IL) 1:1000, followed by ExtrAvidin peroxidase conjugate (Sigma) 1:500, or Streptavidin biotinylated peroxidase complex (Amersham) 1:1000, respectively, and then 0.05% diaminobenzidine with 0.01% hydrogen peroxide
in 0.1 M phosphate buffer, p H 7.4 for 20 min at 20°C.
Cell cultures Obligate affected fetuses were obtained by transfer of obligate affected embryos into synchronized normal surrogate ewes. is Cerebral neurons from 60 to 70-day-old normal and affected fetuses were prepared and cultured in DMEM/F12 containing B27 (Gibco BRL, Gaithersburg, MD) and 20 mM KC1 as described, is except that the glucose concentration was raised to 0.6% and Fungizone (Gibco BRL) was included at 1%. Cells were plated at a density of 170 000 cells/cm 2 and cultured in 200 #1 drops of medium, held under silicone fluid (Dow Coming 50CS), on German glass coverslips (Bellco, Vineland, NJ) that had been washed overnight with 1M KOH in ethanol, and were pre-treated in 15 # g / m l poly-L-lysine for 16 hours followed b y 10% fetal calf serum for 2 hours. Coverslips were removed at weekly intervals over 5 weeks, washed with PBS, p H 7.4, and fixed for 40 rain with 4% paraformaldehyde (PFA) or 2% PFA and 5% glutaraldehyde. For double immunofluorescent labelling fixed cultured neurons were incubated for 16 hours at 4°C in a mixture of monoclonal antibody to GABA (Sigma) 1:2000 and rabbit antiserum to glutamate (Sigma) 1:2000, followed by 3 hours at 20°C in a mixture of Alexa Fluor 488 goat anti-mouse IgG and Alexa Fluor 594 goat anti-rabbit IgG (Molecular Probes, Eugene, OR), each at 1:1000. To positively identify neurons, PFA fixed cultures were immunostained with a monoclonal antibody to fl-III-tubulin (Promega, Madison, WI) 1:5000, followed by Alexa Fluor 488 goat anti-mouse IgG as for GABA above. Fluorescent images were obtained on an Aristoplan epifluorescent microscope (Leica, Wetzlar, Germany) equipped with a CHROMA 8300 Filter series and a cooled digital CCD camera (Photometrix).
Results =
Nissl stain Atrophy of the cerebral cortex and thinning of the cortical layers were apparent in all areas of a 2year-old NCL sheep brain compared with agematched controls. The underlying white matter was proportionally reduced. This atrophy was
Original article: GABAergic neuron distribution in ovine NCL
137
w.
i~ ~
~
"3 :
,
::
.%
Fig. 1.
Coronal sections of parietal cortex (A-F) the hippocampal CA1 subfield (G, H).
(Cont. p. 138)
1~
Original article: M J Oswald et al.
I.
i, C!
•
=
....
ii~iiiii=ii: il :11!2 iiii~i =::
i == ;=;=i!~i ~=::
i
~i ~ ~ ~i
i
:[:
: [
:i,
:=
=
IC~ •
Fig. 1. (Continued)Frontal cortex (I-L) from normal (A, C, E and G, I, K) and affected (B, D, F, H, J, L) sheep brains. They are stained by the Nissl method (A, B), and by immunohistochemistry for parvalbumin (C, D, G, H) and calretinin (E, F), calbindin (I, J), and GAD (K, L). Scale bars represent 100#m (H, L) and 20#m (insert in D). Sets A-H and I-K are each at the same magnification.
severe in the parietal lobe where the thickness of the cortical plate was reduced by about 50% (Figs 1A, B). Cortical layers are poorly defined in normal sheep brain and could only be distinguished in some areas of normal neocortex. They were barely evident in Nissl sections of normal parietal cortex but in NCL parietal cortex layers 3-5 were discernable (Fig. 1B). This may be a consequence of the overall compression of the cortex.
GABAergic cell-type distribution The most striking feature of an investigation of GABAergic subtypes in normal and affected brains was an obvious reduction in parvalbumin positive neurons in NCL cerebral cortex, compared with normal. This was most pronounced in parietal cortex where very few positive cells remained, and in some areas none were found (Figs 1C, D). Parvalbumin immunoreactive cell loss was also clearly evident in all other cortical areas examined. Few of the remaining cells looked healthy and many had distorted and often rounded somata (Fig. 1D). In normal brains, parvalbumin
immunostaining revealed a characteristic laminar signature in neocortex of frontal and parietal lobes. Numerous parvalbumin positive cells and extensions were present through layers 2-6, with the highest numbers in layers 4 and 5 (Fig. 1C). Most neurons were multipolar of varying size. Pericellular baskets were outlined by parvalbumin positive axons in layers 3-5. Less dramatic differences were evident in the calretinin staining pattern. The infragranular layers, layers 5 and 6, of all NCL cortical areas contained fewer calretinin positive cells than normal, and a horizontal staining pattern in layer 1 was apparent. This was most marked in the temporal lobe (Figs 1E, F). Calretinin positive neurons were most concentrated in layers 2 and 5 of normal neocortex. Staining intensity and numbers were highest in layer 5 of temporal (Fig. 1E) and frontal cortex. Very few neurons were stained in either normal or affected parietal cortex. In the affected hippocampus parvalbumin but not calretinin or calbindin immunoreactive cells were depleted while Nissl staining did not reveal any changes. Parvalbumin positive cells were restricted to the dentate gyrus and the stratum
Original article: GABAergic neuron distribution in ovine NCL
A
1:39
'
/LGlutamate •
A
r
~J
e
-ml
B
Fig. 2. Neurons from a 65-day-old affected fetus after (A) 7 and (B) 21 days in culture, stained by immunofluorescent labelling of fl-lll-tubulin. Scale bar represents 20/~m.
oriens of the CA1 hippocampal subfield in normal brains, but were not seen in NCL brains (Figs 1G, H). No difference in GABAergic and Nissl staining patterns or intensities were found in the basal ganglia or the cerebellum. Both affected and normal sections stained weakly for calbindin and glutamate decarboxylase (GAD). In contrast to parvalbumin and calretinin, calbindin immunoreactive cell loss was not notable in the affected brain sections (Figs 1I, J). However, anomalous cellular staining of layer 1 was seen in affected neocortex (Fig. 1J). In normal frontal cortex, calbindin immunoreactive neurons were mainly localized in layers 2-3 and 5-6. The difference in the GAD staining pattern suggests that fewer GAD reactive neurons are present in affected neocortex compared with normal but the weak staining intensity made this difficult to evaluate (Figs 1K, L). There were also apparently fewer GAD positive neurons in the deeper layers of all NCL cortical areas examined.
Fig. 3. A 28-day culture of neurons from a 65-day-old affected fetus double immunostained for both glutamate and GABA. These images are of the same field, the two antibodies being distinguished by different fluorescent emissions (see Materials and methods). Neurons immuno-
reactive for glutamate only (solid arrowhead), for GABA only (arrow), and for glutamate and GABA (open arrowhead) are marked. Scale bar represents 20 #m.
Cultures Neurons dissociated from 60 to 70-day-old fetuses started to grow extensions within a day of plating and the basic neuronal cell shapes were established within 3 days. Neuronal somata and the neuritic arbor continued to develop and mature beyond 7 days in culture and formed an extensive network around astrocytes also present in the culture. Immunostaining for the neuronal tubulin ]%III isoform at 7 and 21 days in culture detected a variety of bipolar, multipolar and pyramidal neurons (Fig. 2). The development of affected and normal cultures was similar. Tubulin staining showed neurite beading at frequent intervals in affected cultures (Fig. 2). This was less obvious in normal cultures.
140
Immunostaining neurons for the major excitatory and inhibitory neurotransmitters glutamate and GABA, respectively, showed differential expression of them (Fig. 3). Both glutamate and GABA staining was seen in most neurons at 7 days in culture but neuronal staining became more distinct over the following weeks. At 28 days in culture neurons expressing either glutamate or GABA alone, and neurons expressing both of these neurotransmitters to varying levels were identified (Fig. 3). More glutamatergic than GABAergic neurons were identified. This cell type distribution was similar in normal and affected cultures.
Discussion General neuron staining revealed that the profound loss of cortical mass observed before in affected sheep 4,5 resulted in part from the compression of the cortical plate into discernible layers (Figs 1A, B). There was also a marked loss of particular cell types. The loss of parvalbumin immunoreactive neurons was most striking and severe in parietal lobe cortex (Figs 1C, D). Loss of calretinin immunoreactivity in the infragranular layers was also marked but comparatively less pronounced (Figs 1E, F). The combined effect of parvalbumin and calretinin cell loss changed the GAD staining profile. There were fewer GAD positive neurons in the deeper layers of all NCL cortical areas examined. In following the sequential neuropathological changes in the NCL sheep model, Jolly et al. s found that severity and progression of the atrophy affected first and most the parietal > occipital and frontal > temporal lobes. The extent to which the neocortical interneurons immunoreactive for parvalbumin and calretinin were missing in different lobes of NCL brain also suggests that the parietal lobe is relatively more affected than frontal or temporal lobes. The laminar GABAergic cell type distribution in normal sheep brain is comparable with that in other mammals. In rat primary somatosensory cortex GAD and parvalbumin staining is most prominent in layer 4.19-21 The overall staining patterns of calretinin and calbindin are similar and often bilaminar. Immunoreactive neurons d i s t r i b u t e d t o varying levels through supragranular, layers 1-3, and infragranular layers, the relative distribution depending on the cortical area examined.22-24 Williams et al. 2s first noted a loss of type II inhibitory synapses on perikarya and axon hillocks
Original article: M J Oswald et al. of large projection neurons relatively early in human late infantile NCL. A loss of GAD- and parvalbumin-reactive synapses with a basket profile around pyramidal cells of the neocortex or Purkinje cells of the cerebellum was apparent in canine NCL neocortex. 12 In the rand mouse model of NCL, interneurons of cerebral cortex and the hippocampus were found to accumulate storage bodies early in the disease and a loss of GABAergic neurons was seen at an advanced stage. 11 Parvalbumin immunoreactivity in particular was decreased in rand mouse hippocampus. GABAergic neuron loss was also seen when stained for somatostatin and calbindin, but it was less apparent when stained for GAD. The pattern of GABAergic neuron loss seen in affected sheep was different again. However, despite the heterogeneity amongst NCL types, there is accumulating evidence that GABAergic cells, and those expressing parvalbumin in particular are specifically affected in these NCL forms. Interneurons are typically inhibitory in nature, the majority utilizing GABA as their neurotransmitter. 26 A loss of the parvalbumin immunoreactive basket cells in neocortex could cause a lesion in neuronal circuits ultimately affecting the pyramidal output cells in these forms of NCL. Basket and chandelier cells are typically labelled by parvalbumin. 21 Chandelier cells innervate the axon initial segment of their target cells. 27 Basket cells mainly contact pyramidal cells but also provide a strong input to each other and innervate vertically aligned interneurons immunoreactive for vasoactive intestinal polypeptide (VIP). 21 It is thought that the combined action of basket cells and VIP interneurons on pyramidal cells is responsible for the generation of long-range synchronous gamma oscillations. 21,2sA loss of parvalbumin interneurons would render pyramidal neurons more excitable and compromise their co-ordinated output. A study of the relative pathological involvement of individual components within a functional circuit is required to investigate this hypothesis. The differences in the distribution of certain cell types in control and affected brain highlight the desirability of developing neuron cultures that contain cells representative of the range of cells affected in situ. Previously we reported on the establishment of primary sheep neuron cultures derived from the cerebral hemispheres of midgestation, 60 to 90-day-old normal and NCL affected fetuses. 15,16 Under these culture conditions, bipolar and a few tripolar MAP-2 positive neurons developed, both normal and affected cells were viable, and no obvious morphological and viability differences were apparent, is Tubulin
Original article: GABAergic neuron distribution in ovine NCL staining of the cultures described here s h o w e d that the majority of neurons are m o r e complex, m a n y being multipolar (Fig. 2). The stages of neuronal d e v e l o p m e n t in vitro were comparable with those described for rat h i p p o c a m p a l and cerebral cortical neurons.29, 30 Glutamate a n d GABA are two major neurotransmitters f o u n d in neocortex having an excitatory and inhibitory synaptic effect respectively. 26,31 Both of these neurotransmitters were expressed in distinct partially overlapping cell populations of neurons cultured from a 65-day-old sheep fetus (Fig. 3). Both GABA and glutamate are implicated with neurotrophic functions d u r i n g d e v e l o p m e n t 32. This m a y explain the m o r e general staining of all neurons b y these factors seen in the initial stages of the culture. The emergence of distinct cell populations reactive for either GABA or glutamate indicates that these cells utilize these factors as neurotransmitters. The similarities of relative number, m o r p h o l o g y , time course of d e v e l o p m e n t and biochemistry of cultured GABAergic n e u r o n s c o m p a r e d with those in situ 3°~3 suggest that the applied c u l ~ r e system provides a useful m o d e l to compare GABAergic cell function of normal and NCL affected n e u r o n s at the cellular level. This range of cell types m a y be extended b y the culture of neurons from specific regions of the relatively large fetal sheep brains. Specific cell types m a y be selected this way.
Acknowledgements We w o u l d like to acknowledge Willeke vanRoon, H e n r y Waldvogel and Professor Richard Faull (Department of A n a t o m y , Auckland University) for their generous help with the immunostaining methodology, along with Nigel Jay's assistance in reproductive physiology. Financial s u p p o r t was received from the C a n t e r b u r y Medical Research Fund (DNP and MJO), the N e w Zealand Neurological F o u n d a t i o n (DNP and GWK), and the Children Brain Disease Foundation.
3
4
5
6 7 8
9
10
11
12
13
14
15 16
References 1 Jolly RD, Janmaat A, West DM, Morrison I. Ovine ceroid-lipofuscinosis: A model of Batten's disease. Neuropathol Appl Neurobiol 1980; 6: 195-209. 2 Jolly RD, Janmaat A, Graydon RJ, Clemett RS. Ceroidlipofuscinosis: The ovine model. In: Armstrong D, Koppang N, Rider JA (eds) Ceroid-Lipofuscinosis
17 18
141
(Batten's Disease). Amsterdam: Elsevier Biomedical Press, 1982: 219-228. Graydon RJ, Jolly RD. Ceroid-lipofuscinosis (Batten's disease). Sequential electrophysiologic and pathologic changes in the retina of the ovine model. Invest Ophthalmol Visual Sci 1984; 25: 294-301. Mayhew IG, Jolly RD, Pickett BT, Slack PM. Ceroidlipofuscinosis (Batten's disease): Pathogenesis of blindness in the ovine model. Neuropathol Appl Neurobiol 1985; 11: 273-290. Jolly R D, Shimada A, Dopfmer I e t al. Ceroid lipofuscinosis (Batten's disease): Pathogenesis and sequential neuropathological changes in the ovine model. Neuropathol Appl Neurobiol 1989; 15: 371-383. Jolly RD, Palmer DN. The neuronal ceroid-lipofuscinoses (Batten disease): Comparative aspects. Neuropathol Appl Neurobiol 1995; 21: 50-60. Walkley SU, March PA, Schroeder CE et al. Pathogenesis of brain dysfunction in Batten disease. A m J Med Genet 1995; 57: 196--203. Sharp JD, Wheeler RB, Lake BD et al. Loci for classical and variant late infantile neuronal ceroid lipofuscinosis map to chromosomes 11p15 and 15q21-23. Hum Molec Genet 1997; 6: 591-595. Williams RE, Lake BD, Elleder M, Sharp JD. CLN6 variant late infantile/early juvenile NCL. In: Goebel HH, Mole SE, Lake B (eds) The Neuronal Ceroid Lipofuscinoses (Batten Disease). Amsterdam: IOS Press; 1999: 102-13. Broom MF, Zhou C, Broom JE, et al. Ovine neuronal ceroid lipofuscinosis: a large animal model syntenic with the human neuronal ceroid lipofuscinosis variant CLN6. J Med Genet 1998; 35: 717-721. Cooper JD, Messer A, Feng AK et al. Apparent loss and hypertrophy of interneurons in a mouse model of neuronal ceroid lipofuscinosis: Evidence for partial response to insulin-like growth factor-1 treatment. J Neurosci 1999; 19: 2556--67. March PA, Wurzelmann S, Walkley SU. Morphological alterations in neocortical and cerebellar GABAergic neurons in a canine model of juvenile Batten disease. Am J Med Genet 1995; 57: 204-212. Palmer DN, Fearnley IM, Walker JE et al. Mitochondrial ATP synthase subunit c storage in the ceroidlipofuscinoses (Batten disease). Am J Med Genet 1992; 42: 561-567. Palmer DN, Jolly RD, van Mil HC et al. Different patterns of hydrophobic protein storage in different forms of neuronal ceroid lipofuscinosis (NCL, Batten disease). Neuropediatrics 1997; 28: 45-48. Kay GW, Hughes SM, Palmer DN. In vitro culture of neurons from sheep with Batten disease. Mol Genet Metabol 1999; 67: 83-88. Hughes SM, Kay GW, Jordan TW et al. Diseasespecific pathology in neurons cultured from sheep affected with ceroid lipofuscinosis. Mol Genet Metabol 1999; 66: 381-386. Dickson LR, Dopfmer I, Dalefield RR et al. A method of cerebro-cortical biopsy in lambs. N Z Vet J 1989; 37: 21-22. Waldvogel HJ, Kubota Y, Trevallyan SC et al. The morphological and chemical characteristics of striatal neurons immunoreactive for the alpha 1-subunit
Original article: M J Oswald et al.
142
19
20
21
22
23
24
25
of the GABA(A) receptor in the rat. Neuroscience 1997; 80: 775-792. Chmielowska J, Stewart MG, Bourne RC. ?-aminobutyric acid (GABA) immunoreactivity in mouse and rat first somatosensory (SI) cortex: Description and comparison. Brain Res 1988; 439: 155-168. Baimbridge KG, Celio MR, Rogers JH. Calciumbinding proteins in the nervous system. Trends Neurosci 1992; 15: 303-308. Staiger JF, Freund TF, Zilles K. Interneurons immunoreactive for vasoactive intestinal polypeptide (VIP) are extensively innervated by parvalbumincontaining boutons in rat primary somatosensory cortex. Eur J Neurosci 1997; 9: 2259-2268. del Rio MR, DeFelipe J. Colocalization of calbindin D-28k, calretinin, and GABA immunoreactivities in neurons of the human temporal cortex. J Comp Neurol 1996; 369: 472-482. Gabbott PLA, Jays PRL, Bacon SJ. Calretinin neurons in human medial prefrontal cortex (areas 24a,b,c, 32', and 25). J Comp Neurol 1997; 381: 389-410. Gonchar Y, Burkhalter A. Connectivity of GABAergic calretinin-immunoreactive neurons in rat primary visual cortex. Cereb Cortex 1999; 9: 683-696. Williams RS, Lott IT, Ferrante RJ, Caviness VS. The cellular pathology of neuronal ceroid-lipofuscinosis. Arch Neurol 1977; 34" 298-305.
26 Houser CR, Vaughn JE, Hendry SH et al. GABA neurons in the cerebral cortex. In: Jones EG, Peters A (eds) Cerebral Cortex, Vol 2. New York: Plenum; 1984: 63-89. 27 Somogyi P. Synaptic organization of GABAergic neurons and GABA(A) receptors in the lateral geniculate nucleus and visual cortex. In: Lam DKT, Gilbert CD (eds) Neuronal Mechanisms of Visual Perception. Texas: Portfolio; 1989: 35-62. 28 Traub RD, Whittington MA, Stanford IM, Jefferys JGR. A mechanism for generation of long-range synchronous fast oscillations in the cortex. Nature 1996; 383: 621-624. 29 Dotti CG, Sullivan CA, Banker GA. The establishment of polarity by hippocampal neurons in culture. J Neurosci 1988; 8: 1454-1468. 30 Stichel CC, M~iller HW. Dissociated cell culture of rat cerebral cortical neurons in serum-free, conditioned media: GABA-immunopositive neurons. Devel Brain Res 1991; 64: 145-154. 31 Fonnum F. Glutamate: a neurotransmitter in mammalian brain. J Neurochem 1984; 42: 1-11. 32 Leslie FM. Neurotransmitters as neurotrophic factors. In: Loughlin SE, Fallon JH (eds) Neurotrophic Factors. New York: Academic Press; 1993: 565-598. 33 Huettner JE, Baughman RW. Primary culture of identified neurons from the visual cortex of postnatal rats. J Neurosci 1986; 6: 3044-3060.