John Libbey Eurotext

Effects of valproate, vigabatrin and tiagabine on GABA uptake into human astrocytes cultured from fœtal and adult brain tissue Volume 1, numéro 3, Septembre 1999

Astrocytes account for more than half of the glial cell population in the brain. One of their prime physiological roles is in the regulation of extracellular concentrations of a host of endogenous compounds including the inhibitory neurotransmitter gamma-aminobutyric acid (GABA) [1, 2]. Astrocytes are involved in the active transport of GABA out of the synapse and in its mitochondrial metabolism. Given their important role in GABA homeostasis, astrocyte cultures have been extensively employed in neurochemical and neuropharmacological studies of GABA [3-7] and represent a suitable medium for the investigation of the mechanisms of action of antiepileptic drugs (AEDs).

The antiepileptic agents sodium valproate (VPA), vigabatrin (VGB) and tiagabine (TGB) have all been proposed to exert their effects, at least in part, by an action on the GABAergic system. VPA is an established AED whose precise mechanism of action remains to be elucidated. In addition to recognised effects on neuronal sodium channels, it has also been reported to increase GABA synthesis [8], inhibit GABA metabolism [9] and block GABA uptake [10]. VGB and TGB are newer AEDs, specifically designed to act on the GABAergic system. VGB has been reported to irreversibly inhibit the metabolism of GABA by an action on the enzyme GABA-transaminase [11] and also to block glial GABA uptake [12]. TGB has been proposed to selectively block GABA uptake alone [13]. Thus, it would appear that, to a greater or lesser degree, blockade of GABA uptake from the synaptic cleft may contribute to the clinical activity of all three agents.

Much of this information regarding the mechanism of action of AEDs has been gleaned from studies conducted in tissues or isolated cells derived from experimental animals. There is little or no evidence to suggest whether these observations are reciprocated in human tissues and/or cells, despite the obvious implications of clinical relevance. Given their supposed common mechanism of action, and with a ready supply of human brain tissue, we have performed a preliminary study of the effects of VPA, VGB and TGB on GABA uptake into primary cultures of human cerebral cortical astrocytes derived from both adult and foetal brain.

Materials and methods


Dulbecco's modified eagle medium (DMEM)-F12, foetal calf serum (FCS), fungizone®, and penicillin/streptomycin were purchased from GIBCO BRL (Paisley, UK). All other chemicals (reagent/cell culture grade as appropriate), including VPA (2-propylpentanoic acid), were obtained from Sigma Chemical Company (Poole, UK). Radiolabelled GABA (gamma-[14C(U)]aminobutyric acid) was obtained from NEN Research Products (Stevenage, UK). VGB (D,L-4-aminohex-5-enoic acid) and TGB [(R-)-(-)-1-[4,4-Bis(3-methyl-2-thienyl)-3-butenyl]-3-piperidine-carboxylic acid, hydrochloride] were gifts from Hoechst Marion Roussel (Uxbridge, UK) and Novo Nordisk A/S (Bagsvaerd, Denmark), respectively. In the absence of convincing evidence to confirm therapeutic brain drug levels in man, the respective drug concentrations employed in this study were selected to reflect those employed in previous in vitro investigations and brain concentrations known to be effective in experimental seizure models.

Human brain tissue

Cerebral cortex from adult brain (n = 4) was obtained from the temporal lobe of patients undergoing surgery for intractable epilepsy and processed immediately following resection. Anatomically matched foetal tissue (n = 4) was obtained from the temporal lobes of spontaneously aborted foetuses (16-24 weeks gestation). Foetal samples were pathologically normal and could be successfully cultured up to 36 hours post-mortem. Approval to work with adult surgical and foetal post-mortem tissue was granted by the medical ethics committees of the Southern General Hospital NHS Trust, Glasgow and Yorkhill NHS Trust, Glasgow, respectively.

Primary astrocyte cultures

Astrocytes were isolated and maintained in culture using a modification of the method of O'Connor et al. [14]. Tissue samples (adult and foetal) were collected in 10 ml of isolation medium (DMEM-F12 supplemented with 100 I.U./ml penicillin, 100 µg/ml streptomycin, 20 µg/ml gentamycin, and 1% fungizone®) on ice. Unless otherwise stated, all procedures were conducted at 4° C. The tissue was cleaned of blood vessels, meninges and other debris, and minced into 1 mm3 pieces by two passes (at 90°) in a McIlwain tissue chopper (Mickle Laboratory Engineering Company Ltd, Gomshall, UK). The minced tissue was transferred to a sterile tube containing 10 ml of fresh isolation medium, mixed gently to wash, and centrifuged at 560 x g for 5 minutes. The supernatant was discarded and the pellet digested for 20 minutes at 37° C in 2 ml of DMEM-F12 containing 2.5 mg trypsin, 40 µg DNAse, 40 U papain and 0.4 mg collagenase (Solution 1). Thereafter, the digest was centrifuged at 560 x g for a further 5 minutes and the supernatant decanted into a sterile tube containing 10 ml of plating medium (DMEM-F12 supplemented with 100 I.U./ml penicillin, 100 µg/ml streptomycin, 20 µg/ml gentamycin, 1% fungizone®, 15% FCS (v/v) and 20 mM D-glucose). The remaining pellet was digested for a further 20 minutes at 37° C in 2 ml of Solution 1 and again centrifuged at 560 x g for 5 minutes. The resulting supernatant was decanted into a second sterile tube containing 10 ml of plating medium. The pellet was then digested for 20 minutes at 37° C in 2 ml of DMEM-F12 containing 1 mg trypsin, 40 µg DNAse, 400 µg ethylene diaminetetra-acetic acid, and 400 µg collagenase. The tissue was then centrifuged at 560 x g for 5 minutes, and the supernatant decanted into a third sterile tube containing 10 ml of plating medium. The remaining pellet underwent a final digestion for 5 minutes at 37° C in 1 ml of DMEM-F12 containing 40 µg DNAse, 0.5 mg soya bean trypsin inhibitor, and 1.6 mg bovine serum albumin. The digest was then gently triturated 15 times and centrifuged at 560 x g for 5 minutes. The supernatant from this digestion was pooled with those from the previous three digestions and centrifuged at 315 x g for 10 minutes. The supernatant was discarded, the pellet resuspended in isolation medium and the resulting cell suspension plated onto 24-well tissue culture dishes (Sarstedt, Leicester, UK) at a density of 80,000 cells/cm2 in a final volume of 500 µl per well. Cells were left for four hours to allow for attachment. Thereafter, the existing medium was aspirated and the cells re-fed with fresh plating medium. The cultures were maintained at 37° C in an environment of 95% air/5% CO2 with a humidity of >= 90%. The medium was changed twice weekly until the cells reached confluence (around day 21).

GABA uptake assay

GABA uptake activity was determined by a modification of the method of Leach and co-workers [12]. A standard balanced salt solution (BSS) was used throughout. Its composition was as follows: 136 mM NaCl, 5 mM KCl, 0.8 mM MgSO4, 2.6 mM NaHCO3, 0.4 mM KH2PO4, 0.34 mM Na2HPO4, 1.3 mM CaCl2, 5.6 mM D-glucose and 15 mM HEPES. The solution was adjusted to pH 7.4 with 1 N NaOH, and stored at 4° C for up to one week. BSS was warmed to 37° C prior to use. Cultures were removed from the incubator, and the existing medium aspirated. Individual wells were washed twice (2 x 200 µl) with BSS, and 200 µl of the appropriate drug solution was added to each well. Control groups received BSS alone. Cultures were returned to the incubator for one hour. After the incubation period, a further 100 µl of BSS (with appropriate control/drug treatment) containing 150 µM [14C]-GABA (specific activity = 1 mCi/mmol) were added to each well. The cultures were incubated at 37° C for a further five minutes. Thereafter, the bathing medium was aspirated and the cultures washed with 4 x 200 µl BSS. Cells were removed from the wells by scraping in 250 µl of ice cold 1 N NaOH. Aliquots were taken for protein determination by the BIORAD method [12], and liquid scintillation counting in 6 ml of Picofluor 40 scintillation fluid (Canberra Packard, Pangbourne, UK). Disintegrations per minute (dpm) were counted in a liquid scintillation counter (2000CA TRI-CARB, Canberra Packard, Pangbourne, UK). The radioactive content of individual wells was compared to known radioactive standards, and results were quantified by the relation of GABA uptake to the protein concentration and expressed as pmol/minute/mg protein.


Four adult and four foetal brain samples were employed in this study. When plated on 24-well culture dishes, each tissue sample yielded between 4 and 8 results for each control/drug treatment, with a study total of n = 24 for each treatment group in each tissue. Analysis was performed using MINITAB for Windows statistical package (version 10.1) on an Elonex PC-5120/1 microcomputer. Individual results were initially calculated as the percentage of mean control values. Group results were then expressed as mean percentages ± the standard error of the mean (SEM). Drug treated groups were compared to control using one-way analysis of variance with a Dunnett correction for multiple comparisons.


Adult astrocyte cultures

VPA (1,000 µM), VGB (100 µM) and TGB (200 nM) all significantly (p < 0.05) reduced GABA uptake into primary cultures of human adult astrocytes following a 1 hour exposure (figure 1).

Foetal astrocyte cultures

VPA (1,000 µM) and VGB (100 µM) significantly (p < 0.05) reduced GABA uptake into primary cultures of human foetal astrocytes following a 1 hour exposure. TGB (200 and 500 nM) was without effect (figure 2).


VPA significantly reduced the uptake of GABA into both adult and foetal cells. This effect was consistent with previous studies employing frog dorsal root ganglion cells [15] and primary astrocyte cultures derived from rodent brain [10, 16, 17], in which it was reported that VPA reduced the affinity, but not the capacity, of the transport protein for GABA. Other investigators [18, 19] have, however, reported a lack of effect of VPA on GABA uptake. It is possible that this inconsistency reflects the experimental medium used in each case, with positive results being reported from studies employing cell cultures and negative data from investigations of crude synaptosomal preparations. Thus, whether or not inhibition of GABA uptake is a clinically relevant, in vivo action of VPA remains to be determined.

Like VPA, VGB significantly reduced the uptake of GABA into human astrocytes derived from both adult and foetal tissue. This observation supports a previous study from our laboratory [12], employing primary cultures of rat cerebral cortical astrocytes. Other investigators [20] have, however, found VGB to exert only a very weak blockade of GABA uptake into mouse astrocytes, with an IC50 well above the clinical range. Species differences may account for the discrepancies observed between rat, mouse and human cells in culture, but again, like VPA, the clinical relevance of this potential mechanism of VGB action requires further investigation.

Of the three drugs studied, only TGB is a recognised GABA uptake inhibitor [13], an effect which has been demonstrated in rat astrocyte and neurone cultures and rat brain synaptosomes. Blockade of GABA uptake, specifically at the GAT-1 transporter [21], is the only known mechanism of TGB action and is thought to be responsible for its experimental anticonvulsant [22] and clinical antiepileptic activity [23]. In contrast to the effects of VPA and VGB, TGB failed to reduce GABA uptake into astrocytes derived from human foetal brain. The drug did, however, block uptake into cells derived from adult tissue. Although further investigations are clearly required, these results suggest that the blockade of GABA uptake observed with VPA and VGB manifests by means of a mechanism different from that observed with TGB.

The uptake of GABA in the mammalian nervous system is regulated by three specific transport systems named GAT-1, GAT-2 and GAT-3 [21]. A fourth, less specific transporter, BGT-1, also contributes to the removal of GABA from the extra-cellular space [24]. These transporters are located on the plasma membrane of both neurones and glial cells, exhibit high affinity for GABA and require Na+ and Cl- for transport [25]. Despite similar structural features and 50-70% amino acid sequence homology, the GAT transporters have distinct pharmacological profiles and are differentially distributed along the neuroaxis [26].

The GAT-1 GABA transporter is the recognised molecular target for TGB, whereas conclusions from our previous studies [12] loosely inferred an action of VGB at the GAT-3 transporter. The current observation that VGB (and VPA) differ in their actions from TGB might lend a little further support to this proposal. These data also suggest that the specific GABA transporters exhibit not only a distinct distribution in the nervous system, but also a differential development. Delayed development of the GAT-1 transporter could account for the failure of TGB to block GABA uptake in astrocytes derived from foetal (16-24 weeks gestation) tissue.

It must also be borne in mind that the "adult astrocytes" employed in this study are derived from tissue which is not necessarily normal, given that it has been resected from the region of an epileptic focus. This fact may have an additional bearing on any discrepancies between the results obtained from adult and foetal sources, and may influence the nature and extent of the pharmacological response in the respective cells.


This preliminary study, conducted in astrocytes derived from human brain tissue, suggests that inhibition of GABA uptake may contribute to the clinical activity of VPA, VGB and TGB. The failure of TGB to block uptake in foetally-derived cultures suggests the involvement of distinct mechanisms of blockade amongst the three drugs under investigation and may be indicative of a differential development of GABA transporters in mammalian brain. However, the failure of TGB to block GABA uptake in foetal-derived astrocytes does not imply that the drug would be of little value in the treatment of paediatric epileptic disorders. Further studies are required to substantiate these observations and to investigate the proposed ontogeny in the pharmacological response to TGB.


The authors would like to thank Dr Edward O'Connor (Department of Neurosurgery, Yale University School of Medicine, New Haven, CT, USA) for his invaluable help in establishing the cell culture method and Hoechst Marion Roussel and Novo Nordisk A/S for their kind gifts of VGB and TGB, respectively.