Neurochem Res (2015) 40:410–419 DOI 10.1007/s11064-014-1494-9
ORIGINAL PAPER
The Subcellular Localization of GABA Transporters and Its Implication for Seizure Management Karsten K. Madsen • Gert H. Hansen • E. Michael Danielsen • Arne Schousboe
Received: 28 September 2014 / Revised: 27 November 2014 / Accepted: 1 December 2014 / Published online: 18 December 2014 Ó Springer Science+Business Media New York 2014
Abstract The ability to modulate the synaptic GABA levels has been demonstrated by using the clinically effective and selective GAT1 inhibitor tiagabine [(R)-N-[4,4-bis(3methyl-2-thienyl)-3-butenyl]nipecotic acid]. N-[4,4-bis(3methyl-2-thienyl)-3-butenyl]-3-hydroxy-4-(methylamino)4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol (EF1502) which not only inhibits GAT1 like tiagabine but also BGT1 has been shown to modulate extrasynaptic GABA levels. The simultaneous inhibition of synaptic and extrasynaptic GABA transporters using tiagabine and EF1502, respectively has been demonstrated to exert a synergistic anticonvulsant effect in several seizure models in mice. The pharmacological profile of these and similar compounds has been thoroughly investigated in in vitro systems, comparing the GAT subtype selectivity with the ability to inhibit GABA uptake in primary cultures of neurons and astrocytes. However, an exact explanation has not yet been found. In the present study, the ability of GATs to form homo and/or heterodimers was investigated as well as to which membrane Special Issue: In Honor of Michael Norenberg. K. K. Madsen (&) A. Schousboe Department of Drug Design and Pharmacology, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen Ø, Denmark e-mail: [email protected] A. Schousboe e-mail: [email protected] G. H. Hansen E. M. Danielsen Department of Cellular and Molecular Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen Ø, Denmark e-mail: [email protected] E. M. Danielsen e-mail: [email protected]
123
micro environment the GATs reside. To investigate dimerization of GATs, fusion proteins of GATs tagged with either yellow fluorescent protein or cerulean fluorescent protein were made and fluorescence resonance energy transfer (FRET) was measured. It was found that GATs form both homo- and hetero-dimers in N2A and HEK-293 cells. Microdomain localization of GATs as investigated by detergent resistant membrane fractions after treatment of tissue with Brij-98 or Triton X-100 revealed that BGT1 and GAT1 mostly localize to non-membrane rafts independent of the detergent used. However, GAT3 localizes to membrane rafts when using Brij-98. Taken together, these results suggest that the observed hetero dimerization of GATs in the FRET study is unlikely to have functional implications since the GATs are located to very different cellular compartments and cell types. Keywords GABA transporters Epilepsy FRET Dimerization Localization Raft
Introduction Modulation of GABA neurotransmission accomplished via regulation of synaptic and extrasynaptic GABA levels by inhibition of GABA transporters (GATs) or GABA-transaminase has proven useful in management of seizure events [1, 2]. Pharmacological intervention using GAT inhibitors has been successfully carried out using the GAT1 selective inhibitor tiagabine [(R)-N-[4,4-bis(3-methyl-2thienyl)-3-butenyl]nipecotic acid] which has been established as a clinically effective drug for the treatment of focal seizures in humans hereby providing a proof of concept that inhibition of GATs constitutes an interesting target in the development of anticonvulsant drugs [3].
Neurochem Res (2015) 40:410–419
The physiological role of GATs is to facilitate the removal of GABA from the extracellular space by transport into presynaptic neurons and surrounding astrocytes. GABA taken up from the synaptic cleft into the presynaptic neuron serves to replenish the GABA transmitter pool and thus to recycle the neurotransmitter, whereas GABA entering the astrocytes is completely metabolized and lost from the transmitter pool (for references, see Schousboe et al. [4]). Interestingly, a study by Bender and Norenberg [5] has provided evidence that astrocytic GABA transporters may play a role in the adverse effects of ammonia in brain function as seen during hyperammonemia associated with hepatic encephalopathy (HE). Ammonia was shown to increase release of GABA from astrocytes and at the same time the uptake capacity was decreased. This could possibly contribute to the dysfunction of GABAergic neurotransmission associated with HE (for references, see Schousboe et al. [6]). Pharmacological inhibition of GATs serves to increase the extracellular concentration of GABA and thereby the duration of its action at synaptic and extrasynaptic GABA receptors, thereby increasing the inhibitory tonus on phasic and tonic GABA receptors, respectively (for references, see Schousboe et al. [7]). Four GATs have been cloned from human, rat, and mouse tissue resulting in a rather confusing nomenclature and that proposed by the HUGO Gene Nomenclature Committee will be used throughout, i.e. GAT1 (SLC6A1), BGT1 (SLC6A12), GAT2 (SLC6A13), and GAT3 (SLC6A11). Many attempts to deduce the cellular localization of GATs have been undertaken but unfortunately a comprehensive immunohistochemical study utilizing knockout control animals remains to be fully undertaken. However, a general consensus regarding GAT localization is as follows: GAT1 is predominantly expressed on neurons around the synapse and to a minor extent on distal astrocytic processes throughout the mature brain [8, 9]. GAT3 is predominantly expressed on distal astrocytic processes which are in direct contact with GABAergic neurons. GAT3 is highly expressed in retina, olfactory bulb, brainstem, diencephalon but shows low levels of expression in hippocampus and cortex. In this regard GAT3 displays a much more restricted localization than GAT1 [10]. GAT2 is found in the leptomeninges and the neonatal brain and this transporter subtype is not believed to have a major effect on termination of GABAergic neurotransmission [11, 12]. BGT1 has been found in the hippocampus and cortex. In contrast to GAT1, this transporter is not located close to GABAergic synapses but resides mainly in extrasynaptic regions [13, 14]. The expression of BGT1 in the brain is, however, controversial as it has been reported to be extremely low [15]. In polarized MDCK cells GAT1 and GAT3 are found exclusively on the apical surface whereas GAT2 and BGT1 are found on the basolateral surface [9, 16, 17].
411
The pharmacological properties of GAT inhibitors have been extensively studied both in vitro and in vivo (for references, see Schousboe et al. [7]) and the following will provide a brief account of studies relevant to the understanding of the roles of GAT1, BGT1 and GAT3. Thus, the selective GAT1 inhibitor tiagabine, and EF1502 [N-[4,4bis(3-methyl-2-thienyl)-3-butenyl]-3-hydroxy-4-(methylamino)-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol] a compound which displays selectivity towards GAT1 and BGT1 have increased our understanding about the modulatory potential of GAT inhibitors on the synaptic and extrasynaptic GABA levels in the brain. With the discovery of EF1502, the first compound to display selectivity towards BGT1, it was anticipated that a detailed pharmacological investigation of BGT1 as a target could be undertaken. Using the isobologram paradigm a synergistic anticonvulsant effect was observed when combining EF1502 with tiagabine or another GAT1 selective drug such as LU-32-176B [N-[4,4-bis(4-fluorophenyl)-butyl]-3-hydroxy-4-amino-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol]. However, when combining the two GAT1 selective compounds tiagabine and LU-32-176B only an additive anticonvulsant effect was observed. Of importance in this study was that only the anticonvulsant activity was synergistic but not the toxicity of tiagabine [2]. Later it was shown using the same experimental approach that when combining the GAT2/3 selective inhibitor SNAP-5114 with EF1502 a synergistic anticonvulsant effect was observed, whereas the combination of SNAP-5114 and tiagabine resulted in an additive effect [18]. The correlation between inhibitory potential of GAT inhibitors on neurons and astrocytes and their anticonvulsant activity compared to GAT subtype selectivity seems to further add to the complexity of GAT pharmacology. It has been established that inhibiting neuronal GAT could evoke seizure activity in mice due to depletion of neurotransmitter levels in presynaptic neurons. Therefore, it was hypothesized that selective inhibition of astrocytes would be advantageous (see, Schousboe et al. [7]). The present study was undertaken in an effort to further characterize GATs in their local environment and to investigate whether GATs could make hetero dimerization. The pharmacological properties of GAT inhibitors might be affected if GATs are able to form homo or heterodimers in different cellular compartments. To this end GAT fusion proteins with two variants of the green fluorescent protein were prepared and fluorescence resonance energy transfer (FRET) was measured to assess the potential of GAT homo- and hetero-dimerization. These studies were undertaken in HEK-293 cells and N2a cells since transfection into neurons and astrocytes resulted in cultures which changed morphology or died upon transfection. It was also investigated if the different GATs resided in the same environment in the cell membrane in primary cultures
123
412
of astrocytes and neurons, adult mouse brain tissue, and immortalized cell cultures. Moreover, it was investigated if the transporters reside in detergent resistant membrane fractions upon treatment of the tissue with Brij-98 or Triton X-100 followed by ultracentrifugation overnight and subsequent detection using immunoblotting. Results obtained by these technologies were anticipated to help elucidate some of the pharmacological enigmas outlined above.
Experimental Procedures Materials Adult, new born, and E15 NMRI mice were obtained from Taconic (Ry, Denmark). Plastic materials were purchased from NUNC (Roskilde, Denmark). Reagents were of the highest grade and usually obtained from Sigma-Aldrich (St. Louis, MO, USA). Penicillin was purchased from Leo Pharma A/S (Ballerup, Denmark). Blasticidin was purchased from InvivoGen (San Diego, USA). Fetal calf serum was obtained from Invitrogen (Carlsbad, USA, lot nos. 216628 and 72118) and fetal bovine serum was from Lonza (Verviers, Belgium). Transfection reagent jetPEITM was from Polyplus-transfections SA (Illkirch, France). Complete mini protease inhibitor was from Roche Diagnostics (Mannheim, Germany) ECF substrate and StormTM scanner was from GE healthcare (Chalfont St Giles, United Kingdom). Primary Culture of Cerebral Cortical Neurons Primary cultures of cerebral cortical neurons were prepared essentially as described by Hertz et al. [19]. After dissection of the cerebral cortex the tissue was exposed to a mild trypsinization (0.2 mg/mL trypsin, 10 min at 37 °C) followed by mechanical trituration in a DNase solution (75 I.U./mL) containing soya bean trypsin inhibitor (0.44 mg/mL). The cells were suspended in a slightly modified Dulbecco’s medium containing: 17.1 mM KCl, 27.9 mM G-glucose, 0.2 mM L-glutamine, 0.05 I.U. insulin, 7.3 lM pABA, 50,000 I.U. penicillin, and 10 % fetal calf serum, and counted to ensure a cell density of 3.33 mill./mL before they were cultured in 80 cm2 plastic flasks precoated with poly-D-lysine (Mw [ 300.000 g/mol) overnight at room temperature. To prevent astrocytic proliferation cytosine arabinoside was added after 32–48 h to a final concentration of 20 lM in the culture medium. The cells were maintained at 37 °C in a 95/5 % (v/v) mixture of atmospheric air and CO2 with a humidified atmosphere for 7 days at which point the culture consist of a confluent monolayer of 95 % GABAergic neurons [19].
123
Neurochem Res (2015) 40:410–419
Primary Culture of Cerebral Cortical Astrocytes Cortical astrocytes were cultured essentially as described by Hertz et al. [20]. Using newborn NMRI mice not older than 24 h, and the rather harsh condition under which they are obtained, the culture is devoid of meningeal cells, debris, blood vessels, and ramified neurons, leaving a culture highly enriched in astroglial cells (95 %) with no neurons remaining. Furthermore, because of the extensive biochemical and biophysical similarities between this preparation of astrocytes and that of its in vivo counterparts makes this technique ideal to study astrocytes in vitro [20, 21]. The neopallium was isolated from newborn NMRI mice and passed through a Nitex nylon sieve (80 lm pore size) into a slightly modified Dulbecco’s modified Eagle’s medium (DMEM) containing 2.5 mM L-glutamine, 6 mM D-glucose, 100,000 I.U. penicillin and cultured in 80 cm2 plastic flasks at a density of 0.13 cortices/mL. The cells were maintained at 37 °C in a 95/5 % (v/v) mixture of atmospheric air and CO2 with a humidified atmosphere and cultured for 3 weeks. Twice a week the medium was changed. The first week the medium contained 20 % fetal calf serum and the following week it was lowered to 15 %. The third week the serum concentration was lowered to 10 % and the final week it was lowered to 10 % and 0.25 mM dibutyryl cyclic AMP (dBcAMP) was added to the medium. dBcAMP causes the astrocytes to differentiate morphologically with extension of cell processes and biochemically, which facilitates the further reduction of phagocytic contamination [21].
HEK Cells Stably Expressing mGAT1–4 Four cultures of stably transfected Human Embryonic Kidney (HEK)-293 cells expressing mGAT1–4 as described [22] were used. Shortly cDNA of the four GATs were inserted into the expression vector pIRES also encoding cDNA for blasticidin S deaminase. Each of the four vectors was transfected into its own culture of HEK-293 cells and blasticidin-S at a concentration of 20 lg/mL was used to force selection of stably transfected cells. The cells were cultured in DMEM containing 5 mL penicillin/streptomycin 9 100, 5 lg/mL blasticidin-S, and 6 % fetal bovine serum. Cells were split when they reached 80 % confluence by loosening the cells in 1X Trypsin/EDTA solution in 1XPBS. The cells were maintained at 37 °C in a 95/5 % (v/v) mixture of atmospheric air and CO2 with a humidified atmosphere. Cells were used for experiments when reaching 80 % confluency.
Neurochem Res (2015) 40:410–419
Adult Mouse Cortex The cortex of adult NMRI mice was dissected on ice and flash frozen in 2-methyl-butane (cooled on dry ice). Before membrane raft detergent analysis the brains were thawed and homogenized. Membrane Raft Isolation The detergent resistant membrane fraction analysis was preformed essentially as described by Dalskov et al. [23]. Cells and tissue after being harvested or homogenized were resuspended in 1 mL HEPES buffer (25 mM HEPES-HCl, 150 mM NaCl, pH 7.1 containing 1 tablet Complete mini protease inhibitor and extracted for 10 min with either 1 % Triton X-100 on ice or 1 % Brij 98 at 37 °C. The extracts were then mixed in equal volumes of 80 % (w/v) sucrose in the same buffer in centrifuge tubes (Beckman, Fullerton, USA; # 344060) and a continuous sucrose gradient of 15–40 % was carefully poured on top and finally the extracts were centrifuged at 100,0009g for 18 h at 5 °C in a Beckman ultracentrifuge (Optima LE 80K) using a Sw40Ti rotor head. Following the ultracentrifugation the gradient was harvested from the bottom of the tube in 1 mL fractions resulting in 12 fractions. Fraction 1–3 is considered to contain non-membrane raft proteins and fraction 5–12 is considered to contain membrane raft proteins. In each of the fractions 1:1 volume of ice cold acetone was added for 30 min on ice to precipitate the proteins which was pelleted at 20,0009g. The pellet was dissolved in 100 lL 19 sample buffer and used for western blotting. All membrane raft extractions were performed on at least two separate tissue preparations. The amount of protein used for each cell type varied. Two bottles of T80 flasks of confluent astrocytes and three T80 bottles of confluent neurons were subjected to the above procdure and finally the proteins were dissolved in 80 lL of 19 sample buffer. Three cortices of adult NMRI mice were harvested as described above, homogenized and split into two pools, each pool treated with Triton X-100 or Brij 98. The proteins were dissolved in 133 lL 19 sample buffer.
413
Millipore, Billerica, USA), BGT-1 1:1,000 and GAT-3 1:1,000 (BGT11-A and GAT31-A, respectively, AlphaDiagnostic, San Antonio, USA), Flotillin-1 1:250 (610820, Nordic Biosite, Ta¨by, Sweden), and Na?, K?-ATPase A-subunit 1:500 (MA3-928, Affinity Bio reagents, Golden, USA) diluted in 1.5 % non-fat skin milk. Protein bands were detected by an AP conjugated antibody 1:7,500 (S3721 and S3731, Promega, Madison, USA). Immunoblots were developed using ECF substrate according to the protocol provided by the manufacturer and finally visualized on a StormTM scanner. Construction of GAT Fusion Proteins C-terminal fusion proteins of the four mouse GATs were constructed by insertion of venus and cerulean, which were generously provided by, Steven S. Vogal (NIH). The constructs were made in the pIRES vector also containing a selection marker for Blasticidin-S. Furthermore, N-terminal tagged GAT1 with yellow fluorescent protein (YFP) has been made which was further tagged with cerulean in the C-terminus to give a double tagged construct. General cloning techniques have been employed and the constructs were verified by sequence analysis by (MWG, Ebersberg, Germany). The fusion proteins were connected by a linker region which is shown in Table 1. Positive FRET Controls A positive FRET pair was generously donated by Professor Kathryn M. Partin. Specifically, AMPA receptors tagged with cerulean fluorescent protein (CFP) and YFP (R1i15CFP and R1i46YFP) were obtained showing a FRET efficiency of 18.5 ± 2.9 % obtained by acceptor photobleaching [24]. Construction of N2a Cells Stably Expressing GAT Fusion Proteins with Cerulean The GATcerulean fusion protein described above was transiently transfected into N2a cells using JetPEI in Table 1 Linker region in the GAT fusion proteins C-terminal of GATs
Linker region
N-terminus of cerulean/venus
GAT1
SKEAYI
GPVAT
MW
BGT1
WEKETH(-L)
FPVAT
MW
GAT2
ELESNC
GPVAT
MW
GAT3
TEKETHF
GPVAT
MW
Western Blotting Precipitated proteins from each fraction were loaded onto 10 % SDS-PAGE gels in equal volumes. 20 lL of neurons and astrocytes and 5 lL of adult mouse cortex were loaded onto each lane, along with a molecular weight marker and separated at 100 V and subsequently transferred overnight at 25 V to an Immobilon-P PVDF membrane. Membranes were blocked in 3 % non-fat skim milk and blotted for 2 h in primary antibody against GAT-1 1:1,000 (AB1570 W;
The table shows the amino acid sequence of the C-terminal of GATs, the amino acid sequence of the linker region bridging the GAT and fluorescent protein, and the amino acid sequence of the N-terminal of the fluorescent proteins. In BGT1 the last amino acid L has been deleted
123
414
25 cm2 plastic flasks following the transfection protocol provided by the manufacturer. On the second day after transfection blasticidin-S were added at a concentration of 30 lg/mL to promote selection of stably transfected cells. Once the culture only consisted of stably transfected cells in was maintained in DMEM containing 5 mL penicillin/ streptomycin x100, 5 lg/mL blasticidin-S, and 6 % fetal bovine serum. Transient Transfection of N2a Cells for FRET Studies The stable N2a cell expressing GATcerulean or N2a cells were cultured with and without Blasticidin-S, respectively and were split when they reached 80 % confluence by loosening the cells in 1X Trypsin/EDTA solution in 1XPBS. N2a or stably transfected cells being transfected for FRET measurements were seeded at a confluency of 35 % on poly-D-lysine (mW [ 300,000 g/mol) precoated glass coverslips the day before transfection. 12 or 35 mm round coverslips (#1.5 Menzel-Gla¨ser, Menzel Gmbh, Braunschweig, Germany) were used. Transient transfections using JetPEI according to the manufactures descriptions were done 24 h prior to FRET measurements. Twelve mm coverslips were washed in ddH2O and fixed in 4 % paraformaldehyde for 10 min at 4 °C before being mounted on a glass slide using Dako Faramount mounting medium (Dako Inc., Carpinteria, USA) and used for sensitized emission FRET measurements. Verification of Biological Activity of Fusion Proteins Kinetic analysis of the fusion proteins was carried out as described [25] to determine Km and Vmax for GABA. Briefly, 24 well plates were seeded with N2a cells and transiently transfected as described above. 24 h after transfection the kinetic determination was carried out. The incubations were carried out at 37 °C in phosphate buffered saline containing 6 mM D-glucose and a constant 3 [H]GABA concentration. The GABA concentration was varied over the range of 1–400 lM. Cells were incubated for 3 min and subsequently washed in cold PBS before they were dissolved in 0.4 M KOH. Non-specific binding was done on ice. Radioactivity was measured in the presence of Ecoscint A (National Diagnostic, Atlanta, USA) on a Tri-Carb 2900TR counter (Perkin Elmer, Wellesley, USA). Protein concentration was determined using Pierce micro BCA protein kit (Thermo Scientific, Rockford, USA). The statistical analysis of Km and Vmax was done in Sigmaplot 9.0 (Systat Software Inc, San Jose, USA) using the Michaelis–Menten equation as fitting parameter.
123
Neurochem Res (2015) 40:410–419
Confocal Image Acquisition FRET measurements were performed on a Leica SP2 confocal microscope equipped with an argon laser (Leica, Wetzlar, Germany) upgraded with a 405 nm pulsed diode laser controlled by a Picoquant PDL-B laser driver (Picoquant, Berlin, Germany) and time-correlated single-photon counting detector DCC-100 (Becker and Hickl, Berlin, Germany). A 63X 1.2Na water HCX PL APO Lbd.bl objective was used for imaging. Sensitized Emission FRET Analysis Sensitized emission FRET was performed as described by Chen and Periasamy [26] using their PFRET-Analysis plugin for ImageJ (freeware, NIH, USA). N2a cells stably expressing the GAT-cerulean constructs or transiently expressing the positive control R1i15CFP were used as donor reference cells [donor alone (D)]. N2a cells transiently expressing GAT-venus constructs and R1i46YFP were used as acceptor reference cells [acceptor alone (A)]. The FRET samples [donor and acceptor (DA)] were prepared by transiently transfecting the stable expressing GAT-cerulean N2a cells with the GAT-venus constructs or in the case of the positive control a double transient transfection. The measured FRET efficiencies of the GAT dimers were compared to the background value (areas devoid of cells) and a Student’s t test was performed. Sensitized emission FRET was imaged on the above mentioned system using an Argon-laser exciting donor and acceptor at 458 and 514 nm, respectively. Donor and acceptor emission were gathered in the range of 462–510 and 520–660 nm, respectively. Each image was taken with a zoom factor of 5 with both frame averaging and line averaging set to 2. Before imaging laser power and PMT gain were adjusted to avoid to low uncorrected sensitized FRET signal and oversaturated donor signal in the acceptor and donor emission channels, respectively. All images were of 512 9 512 pixels at 8 bit color depth.
Results Validation of Biological Activity of GAT Fusion Proteins In order to assess the function of the GAT fusion proteins relative to the non-tagged versions, kinetic measurements were carried out to determine Km and Vmax values and the membrane localization was confirmed by confocal microscopy. As seen in Table 2 Km for the cerulean tagged
Neurochem Res (2015) 40:410–419
415
GAT1–3 was found to be closely related to the expected values around 12–19 lM and for BGT1 the Km of 49 lM was also close to the expected value. Confocal microscopy confirmed expression of the fusion proteins at the surface of the cells (results not shown). Sensitized Emission FRET Following the verification and validation of the basic pharmacological property of the GAT fusion proteins sensitized FRET studies were performed as described above using N2a cells as a model system. N2a cells were chosen since they grow without cell to cell contact allowing for better distinction of single cells when performing imaging studies. Hek-293 cells were also used for comparison generating comparable results (data not shown). To validate the sensitized emission FRET technique used in this study a positive FRET pair was generously donated by Professor Kathryn M. Partin. Specifically, AMPA receptors tagged with CFP and YFP (R1i15CFP and R1i46YFP) were obtained showing a FRET efficiency of 18.5 ± 2.9 % obtained by acceptor photobleaching [24]. As can be seen from Table 3 a FRET efficiency of 21.8 % was obtained when performing sensitized emission FRET, confirming FRET between the AMPA receptors and validating the imaging technique used in this study. To evaluate whether the obtained FRET signal was a consequence of random collision events a FRET experiment was initiated comparing cells with the same amount of donor (GAT1cerulean) with two different levels of acceptor (GAT1venus high and low). In the event of collision FRET the FRET efficiency would increase with higher amounts of acceptor added. As can be seen from Table 3 the combination of GAT1cerulean with GAT1venus high and GAT1venus low resulted in a FRET efficiency of 20.3 and 19.2 %, respectively suggesting that the FRET signal found is not due to random collision events. The FRET efficiency of the heterodimers consisting of GAT1 combined with GAT1, BGT1, GAT2, and GAT3 was estimated to be between 16.2 and 20.4 % showing a comparable efficiency regardless of the pair examined. The background FRET efficiency determined in areas devoid of cells was also calculated showing values of 7.0 % in the Table 2 Km and Vmax values for GABA on GAT fusion proteins
Km (lM) Vmax (nmol/ min mg)
GAT1cer
BGT1cer
GAT2cer
GAT3cer
15.3 ± 10.0
49.1 ± 14.3
18.5 ± 3.8
12.0 ± 5.8
7.8 ± 1.2
8.2 ± 0.7
5.5 ± 0.2
5.3 ± 0.6
The table shows Km and Vmax values ±SD for GABA of the GAT fusion proteins. The kinetic analysis was performed on N2a cells stably expressing the GATcerulean fusion proteins
Table 3 FRET efficiency of GAT homo/heterodimers FRET pair
sFRET efficiency
GAT1cerulean ? GAT1venus high
20.3 ± 10.3 (18)***
GAT1cerulean ? GAT1venus low
19.2 ± 6.2 (32)***
GAT1cerulean ? BGT1venus
16.2 ± 7.6 (25)***
GAT1cerulean ? GAT2venus
20.4 ± 6.4 (25)***
GAT1cerulean ? GAT3venus
16.9 ± 4.8 (24)***
Background R1i15CFP ? r1i46YFP Background
7.0 ± 3.4 (38) 21.8 ± 10.4 (12)*** 6.0 ± 1.0 (10)
The table shows a summary of the FRET efficiencies ± SD (number of analyzed cells) in N2a cells processed using the algorithm described by Chen and Periasamy [25]. *** P \ 0.0001 using a Student’s t test comparing to background. GAT1cerulean was analyzed with two different amounts of GAT1venus present to evaluate whether the FRET signal measured could be attributed to random collision events. Background sFRET efficiency was calculated in areas devoid of cells but within the same images used for the FRET pair measurements
GAT experiment and 6.0 % in the positive control experiment. Membrane Microdomain Localization of GATs The specificity of the commercial anti-GAT antibodies available were tested for cross reactivity using the stably transfected HEK-293 cells expressing the four different mouse GATs. The antibody towards GAT1 and BGT1 displayed no cross reactivity towards the other GAT subtypes and were considered selective. The GAT3 antibody showed a marginal cross reactivity towards GAT1 (not quantified). As can be seen in Fig. 1 GAT1 was detected in neuronal cultures and in adult mouse cortex but not in astroglial cell cultures. BGT1 was only observed in adult mouse cortex whereas GAT3 was observed in both primary cultures of neurons and astrocytes and in adult mouse cortex. Furthermore, GAT1 and BGT1 were primarily localized to non-raft membrane microdomains as can be seen by the intense bands in fractions 1–3 with only a slight intensity in fractions 5–12 regardless of the detergent used to isolate the detergent resistant membrane fractions in the adult mouse cortex. However, GAT1 does seem to be slightly associated to membrane rafts in primary neuronal cultures when using both detergents. In contrast GAT3 displayed quite the opposite localization when comparing the two detergents. When isolated with Triton-X 100 GAT3 localized to non-raft domains in neurons and adult mouse cortex showing intense bands in fraction 1–3. In astrocytes on the other hand, there might be a marginal trend to raft localization when using Triton-X 100. However, when isolated with Brij-98 GAT3 is seen to localize
123
416
Neurochem Res (2015) 40:410–419
Fig. 1 Membrane raft association of GAT1, BGT1, and GAT3. Membrane rafts were obtained in neurons (N), astrocytes (A), and mouse cortex (MC) treated with either 1 % Triton X-100 (T) or 1 % Brij 98 (B) (NT, NB, AT, AB, MCT, and MCB). GATs were detected at 70 kDa except for GAT1 in neurons which were found at 55 kDa.
The proteins from fraction 1–12 (bottom to top) were run on a 10 % SDS-PAGE. GAT1 and BGT1 are mostly associated to fraction 1–3 pointing towards a non-raft localization. GAT3 shows a non-raft distribution when treated with Triton X-100 (fraction 1–3) and a membrane raft association when treated with Brij 98 (fraction 5–12)
to raft domains as shown by the intense bands in fractions 5–12 in all three cell types.
leucine zipper might not be directly involved in the dimerization. While the present study and that performed by Scholze et al. [28] point towards the possibility of GATs forming dimers, the question is, whether such dimers are formed in an intact nervous system and not just in a heterologous expression system. Due to the fragile nature of primary neuronal cultures it was not possible to transfect these cells without significantly changing the morphology and killing the majority of the cell population and hence, a definite answer to this question cannot be provided. To shed more light on this issue the localization of GATs to membrane microdomains was performed using the detergent resistant membrane technique. The established membrane raft markers Flotillin-1 and Na?/K?ATPase [23, 29] were included in the study due to their differential membrane raft association depending on the detergent used to solubilize the cells (see Fig. 2). Flotillin-1 was found to be slightly more enriched in the raft fraction in astrocytes and cortex independent of the detergent used, whereas, it partitioned more uniformly in raft and non-raft fractions of neurons. This partitioning of Flotillin-1 has previously been reported by Hill et al. [29]. Na?/K?-ATPase as reported by Dalskov et al. [23] partitions exclusively in non-raft and raft fractions when Triton X-100 and Brij 98 is used, respectively, in all cell types tested. This shows that the established membrane raft associated proteins in this study were found in the expected fractions. This study showed, that in primary cultures of neurons and astrocytes and in mouse cortex GAT1 and BGT1 were found mostly in non-raft fractions, whereas, GAT3 was found to be membrane raft associated when Brij 98 was used as detergent but not when Triton X-100 was used. This indicates that GAT3 and Na?/K?-ATPase are found
Membrane Microdomain Localization of Established Proteins As part of the validation of the raft isolation two positive controls were chosen, i.e. Flotillin-1 and the Na?/K?ATPase. As can be seen in Fig. 2 Flotillin-1 localized to raft domains regardless of the detergent used whereas the Na?/K?-ATPase displayed a differential localization to non-raft and raft microdomains depending on whether it was isolated with Triton-X 100 or Brij-98, respectively.
Discussion The finding that the Km values found for the fluorescence tagged GAT fusion proteins resembled those reported for the native GATs expressed in HEK cells (see Bolvig et al. [27]) shows that the fusion proteins are likely to exhibit physiological properties in the plasma membrane of the cell lines used for expression, a notion supported by the high Vmax values that were also similar to those previously reported [27]. The results obtained from the FRET studies clearly show that GAT1 can form both homo- and heterodimers in the artificial environment of the N2a cell as well as HEK cells (data not shown). Studies by Scholze et al. [28] using the FRET technique also showed GAT1– GAT1 homodimers and in this study they suggested that the dimerization was due to a leucine zipper located in transmembrane domain 2 consisting of 4 leucine molecules separated by 6 amino acids. This leucine zipper can also be found in the mouse GAT1 but this repeat is not present in BGT1, GAT2, and GAT3 (see Table 4) suggesting that a
123
Neurochem Res (2015) 40:410–419
417
Fig. 2 Membrane raft association of the established membrane raft markers Flotillin-1 and Na?, K?-ATPase were detected at *45 and 110 kDa, respectively. Membrane rafts were obtained in neurons (N), astrocytes (A), and mouse cortex (MC) treated with either 1 % Triton X-100 (T) or 1 % Brij 98 (B) (NT, NB, AT, AB, MCT, and MCB). The proteins from fraction 1–12 (bottom to top) were run on a 10 %
SDS-PAGE. Flotillin-1 is almost evenly distributed in all the cell types exposed to both detergents with preference for membrane raft association in astrocytes and mouse cortex (fraction 5–12). Treatment with Triton X-100 revealed Na?, K?-ATPase as a soluble membrane component (non-raft, fraction 1–3). However, treatment with Brij 98 localizes Na?, K?-ATPase to membrane raft domains (fraction 5–12)
in the same membrane raft population. As reported in the literature, different detergents isolate different membrane raft populations [30]. The serotonin transporter (SERT) has been shown to associate to membrane rafts and non-raft fractions when Brij 58 and Triton X-100 were used as detergents [31]. The membrane raft association of SERT resembles that of GAT3 since both were found to associate to membrane rafts when a weaker detergent as Brij 58 or 98 was used but were solubilized in the presence of Triton X-100. The present study has demonstrated that GATs can form homo- and hetero-oligomers in N2a cells (and HEK-293 cells), but whether all combination possibilities are likely to have functional implications is questionable. Considering the membrane environment and localization studies mentioned above [7–14, 16, 17] it was reported that GAT1 and GAT3 were found on the apical surface of polarized MDCK cells and have been found on distal astrocytic processes in close proximity to GABAergic neurons, suggesting the possibility of interaction. However, in the present study GAT3 and GAT1 were found to be associated with different membrane microdomains since GAT3 partitioned into Brij 98 solubilized raft microdomains whereas GAT1 and BGT1 were not found in raft microdomains in the adult mouse cortex. BGT1 was found to be localized to distinct sites away from GABAergic synapses and to the basolateral surface of polarized MDCK cells [9, 16, 17]. Based on these findings it seems quite unlikely that hetero dimerization of GATs could be of major importance explaining the pharmacology observed for these transporters [1, 2]. The question that comes to mind is what is the physiological significance of GAT dimers? Unfortunately this study was unable to shed light on this issue. However, the raft association of GAT3 and the non-raft
association of GAT1 and BGT1 found in cortex from adult mice suggest that the micromembrane domain association of GATs may be relevant for their in vivo pharmacological profile. In order to obtain a synergistic anticonvulsant effect BGT1 needed to be inhibited using EF1502 combined with either a GAT1 selective compound like tiagabine or a GAT3 selective compound like SNAP-5114 [1, 18]. The selective inhibition of GAT1 and GAT3 only offered an additive anticonvulsant activity [18]. The localization of the GATs may appear to offer a reasonable explanation. Both GAT1 and GAT3 are localized synaptically or perisynaptically and inhibition of these two transporters in combination would only elevate the GABA levels around the synapse resulting in an activation of phasically activated GABA receptors. When inhibiting BGT1 which is located extrasynaptically GABA levels are elevated at sites containing tonically activated GABA receptors. With the combined inhibition of GATs in the synapse and extrasynaptically, two sets of GABA receptors are activated which results in the synergistic effect observed when combining EF1502 and tiagabine [2]. The fact that tiagabine and EF1502 indeed exhibit different mechanistic pharmacological differences was further established in a study where the two compounds were combined with the selective extrasynaptic GABA-A receptor agonist gaboxadol [4,5,6,7tetrahydroisoxazolo[5,4-c]pyridin-3-ol] [1]. Gaboxadol has a higher efficacy than GABA on the alpha-4 containing GABA-A receptors although being less potent [32–34]. As expected, inhibition of GAT1 using tiagabine was not able to elevate the GABA levels at extrasynaptic GABA-A receptors. However, inhibiting the extrasynaptically located BGT1 using EF1502 elevated the GABA levels significantly and caused a competition between the elevated
123
418
Neurochem Res (2015) 40:410–419
Table 4 Comparison of the amino acid sequence of GATs at the site of the GAT1 leucine zipper GAT1
R
F
P
Y
L
C
G
K
N
G
G
G
A
F
L
I
P
Y
F
L
BGT1
R
F
P
Y
L
C
Y
K
N
G
G
G
A
F
F
I
P
Y
F
I
GAT2
R
F
P
Y
L
C
Y
K
N
G
G
G
A
F
F
I
P
Y
L
I
GAT3
R
F
P
Y
L
C
Y
K
N
G
G
G
A
F
L
I
P
Y
V
V
GAT1
T
L
I
F
A
G
V
P
L
F
L
L
E
C
S
L
G
Q
Y
BGT1
F
F
F
S
C
G
I
P
V
F
F
L
E
V
A
L
G
Q
Y
GAT2
F
L
F
T
C
G
I
P
V
F
F
L
E
T
A
L
G
Q
Y
GAT3
F
F
I
C
C
G
I
P
V
F
F
L
E
T
A
L
G
Q
F
The table shows the amino acid sequence of the four mouse GATs from residues 80–118. The italics marking shows the leucine heptad repeat of GAT1 and the comparison to the other GATs. As it can be seen from the comparison only GAT1 contains this leucine zipper
GABA and gaboxadol at the extrasynaptic GABA-A receptors [1]. This underlines the above mentioned theory that in order to achieve a synergistic anticonvulsant activity GATs localized synaptically and extrasynaptically need to be inhibited simultaneously [1]. A recent study by Vogensen et al. [35] has investigated a new selective BGT1 inhibitor [4,4-Bis(3-methylthien-2yl)but-3-enyl](2-carboxycyclohex-2-enyl)methylammonium Chloride, compound 17b). An isobologram study was undertaken to investigate whether tiagabine and compound 17b would interact synergistically as one would expect based on the hypothesis that if BGT1 is inhibited by compound 17b and GAT1 is inhibited by tiagabine then both the extrasynaptic and synaptic GABA levels, respectively would rise and a synergistic anticonvulsant effect would be observed. However, the study only showed an additive effect. So it seems that the explanation of the synergistic effect observed between EF1502 and tiagabine might not be related to inhibition of BGT1 and GAT1, respectively. However, we do know from the Gaboxadol experiments described above, that tiagabine elevates synaptic GABA levels and EF1502 elevates extrasynaptic GABA levels. To investigate this more closely compound 17b should be tested in combination with Gaboxadol. This experiment would allow a conclusion as to whether or not selective inhibition of BGT1 can elevate the extrasynaptic GABA levels significantly to displace Gaboxadol. Until this experiment has been undertaken it is not possible with certainty to ascribe the synergistic effect between EF1502 and tiagabine to an inhibition of BGT1. Acknowledgments We would like to thank Ph.D. Steven S. Vogel, National Institute of Alcohol Abuse and Alcoholism, Bethesda, Maryland (20892), for providing cDNA of Cerulean and Venus and Professor Kathryn M. Partin, Colorado State University, Fort Collins (80523-1617), for providing the positive FRET constructs of AMPA receptors tagged with CFP and YFP (R1i15CFP and R1i46YFP). The work has been financially supported by the Carlsberg Foundation (2009_01_0501).
123
References 1. Madsen KK, Ebert B, Clausen RP, Krogsgaard-Larsen P, Schousboe A, White HS (2011) Selective GABA transporter inhibitors tiagabine and EF1502 exhibit mechanistic differences in their ability to modulate the ataxia and anticonvulsant action of the extrasynaptic GABA(A) receptor agonist gaboxadol. J Pharmacol Exp Ther 338:214–219 2. White HS, Watson WP, Hansen SL, Slough S, Perregaard J, Sarup A, Bolvig T, Petersen G, Larsson OM, Clausen RP, Frølund B, Falch E, Krogsgaard-Larsen P, Schousboe A (2005) First demonstration of a functional role for central nervous system betaine/{gamma}-aminobutyric acid transporter (mGAT2) based on synergistic anticonvulsant action among inhibitors of mGAT1 and mGAT2. J Pharmacol Exp Ther 312:866–874 3. Nielsen EB, Suzdak PD, Andersen KE, Knutsen LJ, Sonnewald U, Braestrup C (1991) Characterization of tiagabine (NO-328), a new potent and selective GABA uptake inhibitor. Eur J Pharmacol 196:257–266 4. Schousboe A, Bak LK, Waagepetersen HS (2013) Astrocytic control of biosynthesis and turnover of the neurotransmitters glutamate and GABA. Front Endocrinol (Lausanne) 4:102 5. Bender AS, Norenberg MD (2000) Effect of ammonia on GABA uptake and release in cultured astrocytes. Neurochem Int 36: 389–395 6. Schousboe A, Waagepetersen HS, Leke R, Bak LK (2014) Effects of hyperammonemia on brain energy metabolism: controversial findings in vivo and in vitro. Metab Brain Dis 29:913–917 7. Schousboe A, Madsen KK, Barker-Haliski ML, White HS (2014) The GABA synapse as a target for antiepileptic drugs: a historical overview focused on GABA transporters. Neurochem Res 39: 1980–1987 8. Conti F, Melone M, De Biasi S, Minelli A, Brecha NC, Ducati A (1998) Neuronal and glial localization of GAT-1, a high-affinity gamma-aminobutyric acid plasma membrane transporter, in human cerebral cortex: with a note on its distribution in monkey cortex. J Comp Neurol 396:51–63 9. Borden LA (1996) GABA transporter heterogeneity: pharmacology and cellular localization. Neurochem Int 29:335–356 10. Minelli A, DeBiasi S, Brecha NC, Zuccarello LV, Conti F (1996) GAT-3, a high-affinity GABA plasma membrane transporter, is localized to astrocytic processes, and it is not confined to the vicinity of GABAergic synapses in the cerebral cortex. J Neurosci 16:6255–6264
Neurochem Res (2015) 40:410–419 11. Conti F, Zuccarello LV, Barbaresi P, Minelli A, Brecha NC, Melone M (1999) Neuronal, glial, and epithelial localization of gamma-aminobutyric acid transporter 2, a high-affinity gammaaminobutyric acid plasma membrane transporter, in the cerebral cortex and neighboring structures. J Comp Neurol 409:482–494 12. Liu QR, Lopez-Corcuera B, Mandiyan S, Nelson H, Nelson N (1993) Molecular characterization of four pharmacologically distinct gamma-aminobutyric acid transporters in mouse brain [corrected]. J Biol Chem 268:2106–2112 13. Borden LA, Smith KE, Gustafson EL, Branchek TA, Weinshank RL (1995) Cloning and expression of a betaine/GABA transporter from human brain. J Neurochem 64:977–984 14. Zhu XM, Ong WY (2004) A light and electron microscopic study of betaine/GABA transporter distribution in the monkey cerebral neocortex and hippocampus. J Neurocytol 33:233–240 15. Zhou Y, Holmseth S, Hua R, Lehre AC, Olofsson AM, PobleteNaredo I, Kempson SA, Danbolt NC (2012) The betaine-GABA transporter (BGT1, slc6a12) is predominantly expressed in the liver and at lower levels in the kidneys and at the brain surface. Am J Physiol Renal Physiol 302:F316–F328 16. Pietrini G, Suh YJ, Edelmann L, Rudnick G, Caplan MJ (1994) The axonal gamma-aminobutyric acid transporter GAT-1 is sorted to the apical membranes of polarized epithelial cells. J Biol Chem 269:4668–4674 17. Ahn J, Mundigl O, Muth TR, Rudnick G, Caplan MJ (1996) Polarized expression of GABA transporters in Madin–Darby canine kidney cells and cultured hippocampal neurons. J Biol Chem 271:6917–6924 18. Madsen KK, Clausen RP, Larsson OM, Krogsgaard-Larsen P, Schousboe A, White HS (2009) Synaptic and extrasynaptic GABA transporters as targets for anti-epileptic drugs. J Neurochem 109(Suppl 1):139–144 19. Hertz E, Yu ACH, Hertz L, Juurlink BHJ, Schousboe A (1989) Preparation of primary cultures of mouse cortical neurons. In: Shahar A, de Vellis J, Vernadakis A, Haber B (eds) A dissection and tissue culture manual of the nervous system. Alan R. Liss Inc, New York, pp 183–186 20. Hertz L, Juurlink BHJ, Hertz E, Fosmark H, Schousboe A (1989) Preparation of primary cultures of mouse (rat) astrocytes. In: Shahar A, de Vellis J, Vernadakis A, Haber B (eds) A dissection and tissue culture manual of the nervous system. Alan R. Liss Inc, New York, pp 105–108 21. Hertz L, Juurlink BHJ, Szuchet S (1985) Cell Cultures. In: Lajtha A (ed) Handbook of Neurochemistry. Plenum Publishing Corporation, New York, pp 603–653 22. White HS, Sarup A, Bolvig T, Kristensen AS, Petersen G, Nelson N, Pickering DS, Larsson OM, Frølund B, Krogsgaard-Larsen P, Schousboe A (2002) Correlation between anticonvulsant activity and inhibitory action on glial gamma-aminobutyric acid uptake of
419
23.
24.
25.
26.
27.
28.
29.
30. 31.
32.
33. 34.
35.
the highly selective mouse gamma-aminobutyric acid transporter 1 inhibitor 3-hydroxy-4-amino-4,5,6,7-tetrahydro-1,2-benzisoxazole and its N-alkylated analogs. J Pharmacol Exp Ther 302:636–644 Dalskov SM, Immerdal L, Niels-Christiansen LL, Hansen GH, Schousboe A, Danielsen EM (2005) Lipid raft localization of GABA A receptor and Na?, K?-ATPase in discrete microdomain clusters in rat cerebellar granule cells. Neurochem Int 46:489–499 Bedoukian MA, Weeks AM, Partin KM (2006) Different domains of the AMPA receptor direct stargazin-mediated trafficking and stargazin-mediated modulation of kinetics. J Biol Chem 281:23908–23921 Larsson OM, Griffiths R, Allen IC, Schousboe A (1986) Mutual inhibition kinetic analysis of gamma-aminobutyric acid, taurine, and beta-alanine high-affinity transport into neurons and astrocytes: evidence for similarity between the taurine and beta-alanine carriers in both cell types. J Neurochem 47:426–432 Chen Y, Periasamy A (2006) Intensity range based quantitative FRET data analysis to localize protein molecules in live cell nuclei. J Fluoresc 16:95–104 Bolvig T, Larsson OM, Pickering DS, Nelson N, Falch E, Krogsgaard-Larsen P, Schousboe A (1999) Action of bicyclic isoxazole GABA analogues on GABA transporters and its relation to anticonvulsant activity. Eur J Pharmacol 375:367–374 Scholze P, Freissmuth M, Sitte HH (2002) Mutations within an intramembrane leucine heptad repeat disrupt oligomer formation of the rat GABA transporter 1. J Biol Chem 277:43682–43690 Hill WG, An B, Johnson JP (2002) Endogenously expressed epithelial sodium channel is present in lipid rafts in A6 cells. J Biol Chem 277:33541–33544 Mishra S, Joshi PG (2007) Lipid raft heterogeneity: an enigma. J Neurochem 103(Suppl 1):135–142 Mu¨ller HK, Wiborg O, Haase J (2006) Subcellular redistribution of the serotonin transporter by secretory carrier membrane protein 2. J Biol Chem 281:28901–28909 Storustovu SI, Ebert B (2006) Pharmacological characterization of agonists at delta-containing GABAA receptors: functional selectivity for extrasynaptic receptors is dependent on the absence of gamma2. J Pharmacol Exp Ther 316:1351–1359 Wafford KA, Ebert B (2006) Gaboxadol—a new awakening in sleep. Curr Opin Pharmacol 6:30–36 Cremers T, Ebert B (2007) Plasma and CNS concentrations of Gaboxadol in rats following subcutaneous administration. Eur J Pharmacol 562:47–52 Vogensen SB, Jorgensen L, Madsen KK, Borkar N, Wellendorph P, Skovgaard-Petersen J, Schousboe A, White HS, KrogsgaardLarsen P, Clausen RP (2013) Selective mGAT2 (BGT-1) GABA uptake inhibitors: design, synthesis, and pharmacological characterization. J Med Chem 56:2160–2164
123
ORIGINAL PAPER
The Subcellular Localization of GABA Transporters and Its Implication for Seizure Management Karsten K. Madsen • Gert H. Hansen • E. Michael Danielsen • Arne Schousboe
Received: 28 September 2014 / Revised: 27 November 2014 / Accepted: 1 December 2014 / Published online: 18 December 2014 Ó Springer Science+Business Media New York 2014
Abstract The ability to modulate the synaptic GABA levels has been demonstrated by using the clinically effective and selective GAT1 inhibitor tiagabine [(R)-N-[4,4-bis(3methyl-2-thienyl)-3-butenyl]nipecotic acid]. N-[4,4-bis(3methyl-2-thienyl)-3-butenyl]-3-hydroxy-4-(methylamino)4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol (EF1502) which not only inhibits GAT1 like tiagabine but also BGT1 has been shown to modulate extrasynaptic GABA levels. The simultaneous inhibition of synaptic and extrasynaptic GABA transporters using tiagabine and EF1502, respectively has been demonstrated to exert a synergistic anticonvulsant effect in several seizure models in mice. The pharmacological profile of these and similar compounds has been thoroughly investigated in in vitro systems, comparing the GAT subtype selectivity with the ability to inhibit GABA uptake in primary cultures of neurons and astrocytes. However, an exact explanation has not yet been found. In the present study, the ability of GATs to form homo and/or heterodimers was investigated as well as to which membrane Special Issue: In Honor of Michael Norenberg. K. K. Madsen (&) A. Schousboe Department of Drug Design and Pharmacology, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen Ø, Denmark e-mail: [email protected] A. Schousboe e-mail: [email protected] G. H. Hansen E. M. Danielsen Department of Cellular and Molecular Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen Ø, Denmark e-mail: [email protected] E. M. Danielsen e-mail: [email protected]
123
micro environment the GATs reside. To investigate dimerization of GATs, fusion proteins of GATs tagged with either yellow fluorescent protein or cerulean fluorescent protein were made and fluorescence resonance energy transfer (FRET) was measured. It was found that GATs form both homo- and hetero-dimers in N2A and HEK-293 cells. Microdomain localization of GATs as investigated by detergent resistant membrane fractions after treatment of tissue with Brij-98 or Triton X-100 revealed that BGT1 and GAT1 mostly localize to non-membrane rafts independent of the detergent used. However, GAT3 localizes to membrane rafts when using Brij-98. Taken together, these results suggest that the observed hetero dimerization of GATs in the FRET study is unlikely to have functional implications since the GATs are located to very different cellular compartments and cell types. Keywords GABA transporters Epilepsy FRET Dimerization Localization Raft
Introduction Modulation of GABA neurotransmission accomplished via regulation of synaptic and extrasynaptic GABA levels by inhibition of GABA transporters (GATs) or GABA-transaminase has proven useful in management of seizure events [1, 2]. Pharmacological intervention using GAT inhibitors has been successfully carried out using the GAT1 selective inhibitor tiagabine [(R)-N-[4,4-bis(3-methyl-2thienyl)-3-butenyl]nipecotic acid] which has been established as a clinically effective drug for the treatment of focal seizures in humans hereby providing a proof of concept that inhibition of GATs constitutes an interesting target in the development of anticonvulsant drugs [3].
Neurochem Res (2015) 40:410–419
The physiological role of GATs is to facilitate the removal of GABA from the extracellular space by transport into presynaptic neurons and surrounding astrocytes. GABA taken up from the synaptic cleft into the presynaptic neuron serves to replenish the GABA transmitter pool and thus to recycle the neurotransmitter, whereas GABA entering the astrocytes is completely metabolized and lost from the transmitter pool (for references, see Schousboe et al. [4]). Interestingly, a study by Bender and Norenberg [5] has provided evidence that astrocytic GABA transporters may play a role in the adverse effects of ammonia in brain function as seen during hyperammonemia associated with hepatic encephalopathy (HE). Ammonia was shown to increase release of GABA from astrocytes and at the same time the uptake capacity was decreased. This could possibly contribute to the dysfunction of GABAergic neurotransmission associated with HE (for references, see Schousboe et al. [6]). Pharmacological inhibition of GATs serves to increase the extracellular concentration of GABA and thereby the duration of its action at synaptic and extrasynaptic GABA receptors, thereby increasing the inhibitory tonus on phasic and tonic GABA receptors, respectively (for references, see Schousboe et al. [7]). Four GATs have been cloned from human, rat, and mouse tissue resulting in a rather confusing nomenclature and that proposed by the HUGO Gene Nomenclature Committee will be used throughout, i.e. GAT1 (SLC6A1), BGT1 (SLC6A12), GAT2 (SLC6A13), and GAT3 (SLC6A11). Many attempts to deduce the cellular localization of GATs have been undertaken but unfortunately a comprehensive immunohistochemical study utilizing knockout control animals remains to be fully undertaken. However, a general consensus regarding GAT localization is as follows: GAT1 is predominantly expressed on neurons around the synapse and to a minor extent on distal astrocytic processes throughout the mature brain [8, 9]. GAT3 is predominantly expressed on distal astrocytic processes which are in direct contact with GABAergic neurons. GAT3 is highly expressed in retina, olfactory bulb, brainstem, diencephalon but shows low levels of expression in hippocampus and cortex. In this regard GAT3 displays a much more restricted localization than GAT1 [10]. GAT2 is found in the leptomeninges and the neonatal brain and this transporter subtype is not believed to have a major effect on termination of GABAergic neurotransmission [11, 12]. BGT1 has been found in the hippocampus and cortex. In contrast to GAT1, this transporter is not located close to GABAergic synapses but resides mainly in extrasynaptic regions [13, 14]. The expression of BGT1 in the brain is, however, controversial as it has been reported to be extremely low [15]. In polarized MDCK cells GAT1 and GAT3 are found exclusively on the apical surface whereas GAT2 and BGT1 are found on the basolateral surface [9, 16, 17].
411
The pharmacological properties of GAT inhibitors have been extensively studied both in vitro and in vivo (for references, see Schousboe et al. [7]) and the following will provide a brief account of studies relevant to the understanding of the roles of GAT1, BGT1 and GAT3. Thus, the selective GAT1 inhibitor tiagabine, and EF1502 [N-[4,4bis(3-methyl-2-thienyl)-3-butenyl]-3-hydroxy-4-(methylamino)-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol] a compound which displays selectivity towards GAT1 and BGT1 have increased our understanding about the modulatory potential of GAT inhibitors on the synaptic and extrasynaptic GABA levels in the brain. With the discovery of EF1502, the first compound to display selectivity towards BGT1, it was anticipated that a detailed pharmacological investigation of BGT1 as a target could be undertaken. Using the isobologram paradigm a synergistic anticonvulsant effect was observed when combining EF1502 with tiagabine or another GAT1 selective drug such as LU-32-176B [N-[4,4-bis(4-fluorophenyl)-butyl]-3-hydroxy-4-amino-4,5,6,7-tetrahydrobenzo[d]isoxazol-3-ol]. However, when combining the two GAT1 selective compounds tiagabine and LU-32-176B only an additive anticonvulsant effect was observed. Of importance in this study was that only the anticonvulsant activity was synergistic but not the toxicity of tiagabine [2]. Later it was shown using the same experimental approach that when combining the GAT2/3 selective inhibitor SNAP-5114 with EF1502 a synergistic anticonvulsant effect was observed, whereas the combination of SNAP-5114 and tiagabine resulted in an additive effect [18]. The correlation between inhibitory potential of GAT inhibitors on neurons and astrocytes and their anticonvulsant activity compared to GAT subtype selectivity seems to further add to the complexity of GAT pharmacology. It has been established that inhibiting neuronal GAT could evoke seizure activity in mice due to depletion of neurotransmitter levels in presynaptic neurons. Therefore, it was hypothesized that selective inhibition of astrocytes would be advantageous (see, Schousboe et al. [7]). The present study was undertaken in an effort to further characterize GATs in their local environment and to investigate whether GATs could make hetero dimerization. The pharmacological properties of GAT inhibitors might be affected if GATs are able to form homo or heterodimers in different cellular compartments. To this end GAT fusion proteins with two variants of the green fluorescent protein were prepared and fluorescence resonance energy transfer (FRET) was measured to assess the potential of GAT homo- and hetero-dimerization. These studies were undertaken in HEK-293 cells and N2a cells since transfection into neurons and astrocytes resulted in cultures which changed morphology or died upon transfection. It was also investigated if the different GATs resided in the same environment in the cell membrane in primary cultures
123
412
of astrocytes and neurons, adult mouse brain tissue, and immortalized cell cultures. Moreover, it was investigated if the transporters reside in detergent resistant membrane fractions upon treatment of the tissue with Brij-98 or Triton X-100 followed by ultracentrifugation overnight and subsequent detection using immunoblotting. Results obtained by these technologies were anticipated to help elucidate some of the pharmacological enigmas outlined above.
Experimental Procedures Materials Adult, new born, and E15 NMRI mice were obtained from Taconic (Ry, Denmark). Plastic materials were purchased from NUNC (Roskilde, Denmark). Reagents were of the highest grade and usually obtained from Sigma-Aldrich (St. Louis, MO, USA). Penicillin was purchased from Leo Pharma A/S (Ballerup, Denmark). Blasticidin was purchased from InvivoGen (San Diego, USA). Fetal calf serum was obtained from Invitrogen (Carlsbad, USA, lot nos. 216628 and 72118) and fetal bovine serum was from Lonza (Verviers, Belgium). Transfection reagent jetPEITM was from Polyplus-transfections SA (Illkirch, France). Complete mini protease inhibitor was from Roche Diagnostics (Mannheim, Germany) ECF substrate and StormTM scanner was from GE healthcare (Chalfont St Giles, United Kingdom). Primary Culture of Cerebral Cortical Neurons Primary cultures of cerebral cortical neurons were prepared essentially as described by Hertz et al. [19]. After dissection of the cerebral cortex the tissue was exposed to a mild trypsinization (0.2 mg/mL trypsin, 10 min at 37 °C) followed by mechanical trituration in a DNase solution (75 I.U./mL) containing soya bean trypsin inhibitor (0.44 mg/mL). The cells were suspended in a slightly modified Dulbecco’s medium containing: 17.1 mM KCl, 27.9 mM G-glucose, 0.2 mM L-glutamine, 0.05 I.U. insulin, 7.3 lM pABA, 50,000 I.U. penicillin, and 10 % fetal calf serum, and counted to ensure a cell density of 3.33 mill./mL before they were cultured in 80 cm2 plastic flasks precoated with poly-D-lysine (Mw [ 300.000 g/mol) overnight at room temperature. To prevent astrocytic proliferation cytosine arabinoside was added after 32–48 h to a final concentration of 20 lM in the culture medium. The cells were maintained at 37 °C in a 95/5 % (v/v) mixture of atmospheric air and CO2 with a humidified atmosphere for 7 days at which point the culture consist of a confluent monolayer of 95 % GABAergic neurons [19].
123
Neurochem Res (2015) 40:410–419
Primary Culture of Cerebral Cortical Astrocytes Cortical astrocytes were cultured essentially as described by Hertz et al. [20]. Using newborn NMRI mice not older than 24 h, and the rather harsh condition under which they are obtained, the culture is devoid of meningeal cells, debris, blood vessels, and ramified neurons, leaving a culture highly enriched in astroglial cells (95 %) with no neurons remaining. Furthermore, because of the extensive biochemical and biophysical similarities between this preparation of astrocytes and that of its in vivo counterparts makes this technique ideal to study astrocytes in vitro [20, 21]. The neopallium was isolated from newborn NMRI mice and passed through a Nitex nylon sieve (80 lm pore size) into a slightly modified Dulbecco’s modified Eagle’s medium (DMEM) containing 2.5 mM L-glutamine, 6 mM D-glucose, 100,000 I.U. penicillin and cultured in 80 cm2 plastic flasks at a density of 0.13 cortices/mL. The cells were maintained at 37 °C in a 95/5 % (v/v) mixture of atmospheric air and CO2 with a humidified atmosphere and cultured for 3 weeks. Twice a week the medium was changed. The first week the medium contained 20 % fetal calf serum and the following week it was lowered to 15 %. The third week the serum concentration was lowered to 10 % and the final week it was lowered to 10 % and 0.25 mM dibutyryl cyclic AMP (dBcAMP) was added to the medium. dBcAMP causes the astrocytes to differentiate morphologically with extension of cell processes and biochemically, which facilitates the further reduction of phagocytic contamination [21].
HEK Cells Stably Expressing mGAT1–4 Four cultures of stably transfected Human Embryonic Kidney (HEK)-293 cells expressing mGAT1–4 as described [22] were used. Shortly cDNA of the four GATs were inserted into the expression vector pIRES also encoding cDNA for blasticidin S deaminase. Each of the four vectors was transfected into its own culture of HEK-293 cells and blasticidin-S at a concentration of 20 lg/mL was used to force selection of stably transfected cells. The cells were cultured in DMEM containing 5 mL penicillin/streptomycin 9 100, 5 lg/mL blasticidin-S, and 6 % fetal bovine serum. Cells were split when they reached 80 % confluence by loosening the cells in 1X Trypsin/EDTA solution in 1XPBS. The cells were maintained at 37 °C in a 95/5 % (v/v) mixture of atmospheric air and CO2 with a humidified atmosphere. Cells were used for experiments when reaching 80 % confluency.
Neurochem Res (2015) 40:410–419
Adult Mouse Cortex The cortex of adult NMRI mice was dissected on ice and flash frozen in 2-methyl-butane (cooled on dry ice). Before membrane raft detergent analysis the brains were thawed and homogenized. Membrane Raft Isolation The detergent resistant membrane fraction analysis was preformed essentially as described by Dalskov et al. [23]. Cells and tissue after being harvested or homogenized were resuspended in 1 mL HEPES buffer (25 mM HEPES-HCl, 150 mM NaCl, pH 7.1 containing 1 tablet Complete mini protease inhibitor and extracted for 10 min with either 1 % Triton X-100 on ice or 1 % Brij 98 at 37 °C. The extracts were then mixed in equal volumes of 80 % (w/v) sucrose in the same buffer in centrifuge tubes (Beckman, Fullerton, USA; # 344060) and a continuous sucrose gradient of 15–40 % was carefully poured on top and finally the extracts were centrifuged at 100,0009g for 18 h at 5 °C in a Beckman ultracentrifuge (Optima LE 80K) using a Sw40Ti rotor head. Following the ultracentrifugation the gradient was harvested from the bottom of the tube in 1 mL fractions resulting in 12 fractions. Fraction 1–3 is considered to contain non-membrane raft proteins and fraction 5–12 is considered to contain membrane raft proteins. In each of the fractions 1:1 volume of ice cold acetone was added for 30 min on ice to precipitate the proteins which was pelleted at 20,0009g. The pellet was dissolved in 100 lL 19 sample buffer and used for western blotting. All membrane raft extractions were performed on at least two separate tissue preparations. The amount of protein used for each cell type varied. Two bottles of T80 flasks of confluent astrocytes and three T80 bottles of confluent neurons were subjected to the above procdure and finally the proteins were dissolved in 80 lL of 19 sample buffer. Three cortices of adult NMRI mice were harvested as described above, homogenized and split into two pools, each pool treated with Triton X-100 or Brij 98. The proteins were dissolved in 133 lL 19 sample buffer.
413
Millipore, Billerica, USA), BGT-1 1:1,000 and GAT-3 1:1,000 (BGT11-A and GAT31-A, respectively, AlphaDiagnostic, San Antonio, USA), Flotillin-1 1:250 (610820, Nordic Biosite, Ta¨by, Sweden), and Na?, K?-ATPase A-subunit 1:500 (MA3-928, Affinity Bio reagents, Golden, USA) diluted in 1.5 % non-fat skin milk. Protein bands were detected by an AP conjugated antibody 1:7,500 (S3721 and S3731, Promega, Madison, USA). Immunoblots were developed using ECF substrate according to the protocol provided by the manufacturer and finally visualized on a StormTM scanner. Construction of GAT Fusion Proteins C-terminal fusion proteins of the four mouse GATs were constructed by insertion of venus and cerulean, which were generously provided by, Steven S. Vogal (NIH). The constructs were made in the pIRES vector also containing a selection marker for Blasticidin-S. Furthermore, N-terminal tagged GAT1 with yellow fluorescent protein (YFP) has been made which was further tagged with cerulean in the C-terminus to give a double tagged construct. General cloning techniques have been employed and the constructs were verified by sequence analysis by (MWG, Ebersberg, Germany). The fusion proteins were connected by a linker region which is shown in Table 1. Positive FRET Controls A positive FRET pair was generously donated by Professor Kathryn M. Partin. Specifically, AMPA receptors tagged with cerulean fluorescent protein (CFP) and YFP (R1i15CFP and R1i46YFP) were obtained showing a FRET efficiency of 18.5 ± 2.9 % obtained by acceptor photobleaching [24]. Construction of N2a Cells Stably Expressing GAT Fusion Proteins with Cerulean The GATcerulean fusion protein described above was transiently transfected into N2a cells using JetPEI in Table 1 Linker region in the GAT fusion proteins C-terminal of GATs
Linker region
N-terminus of cerulean/venus
GAT1
SKEAYI
GPVAT
MW
BGT1
WEKETH(-L)
FPVAT
MW
GAT2
ELESNC
GPVAT
MW
GAT3
TEKETHF
GPVAT
MW
Western Blotting Precipitated proteins from each fraction were loaded onto 10 % SDS-PAGE gels in equal volumes. 20 lL of neurons and astrocytes and 5 lL of adult mouse cortex were loaded onto each lane, along with a molecular weight marker and separated at 100 V and subsequently transferred overnight at 25 V to an Immobilon-P PVDF membrane. Membranes were blocked in 3 % non-fat skim milk and blotted for 2 h in primary antibody against GAT-1 1:1,000 (AB1570 W;
The table shows the amino acid sequence of the C-terminal of GATs, the amino acid sequence of the linker region bridging the GAT and fluorescent protein, and the amino acid sequence of the N-terminal of the fluorescent proteins. In BGT1 the last amino acid L has been deleted
123
414
25 cm2 plastic flasks following the transfection protocol provided by the manufacturer. On the second day after transfection blasticidin-S were added at a concentration of 30 lg/mL to promote selection of stably transfected cells. Once the culture only consisted of stably transfected cells in was maintained in DMEM containing 5 mL penicillin/ streptomycin x100, 5 lg/mL blasticidin-S, and 6 % fetal bovine serum. Transient Transfection of N2a Cells for FRET Studies The stable N2a cell expressing GATcerulean or N2a cells were cultured with and without Blasticidin-S, respectively and were split when they reached 80 % confluence by loosening the cells in 1X Trypsin/EDTA solution in 1XPBS. N2a or stably transfected cells being transfected for FRET measurements were seeded at a confluency of 35 % on poly-D-lysine (mW [ 300,000 g/mol) precoated glass coverslips the day before transfection. 12 or 35 mm round coverslips (#1.5 Menzel-Gla¨ser, Menzel Gmbh, Braunschweig, Germany) were used. Transient transfections using JetPEI according to the manufactures descriptions were done 24 h prior to FRET measurements. Twelve mm coverslips were washed in ddH2O and fixed in 4 % paraformaldehyde for 10 min at 4 °C before being mounted on a glass slide using Dako Faramount mounting medium (Dako Inc., Carpinteria, USA) and used for sensitized emission FRET measurements. Verification of Biological Activity of Fusion Proteins Kinetic analysis of the fusion proteins was carried out as described [25] to determine Km and Vmax for GABA. Briefly, 24 well plates were seeded with N2a cells and transiently transfected as described above. 24 h after transfection the kinetic determination was carried out. The incubations were carried out at 37 °C in phosphate buffered saline containing 6 mM D-glucose and a constant 3 [H]GABA concentration. The GABA concentration was varied over the range of 1–400 lM. Cells were incubated for 3 min and subsequently washed in cold PBS before they were dissolved in 0.4 M KOH. Non-specific binding was done on ice. Radioactivity was measured in the presence of Ecoscint A (National Diagnostic, Atlanta, USA) on a Tri-Carb 2900TR counter (Perkin Elmer, Wellesley, USA). Protein concentration was determined using Pierce micro BCA protein kit (Thermo Scientific, Rockford, USA). The statistical analysis of Km and Vmax was done in Sigmaplot 9.0 (Systat Software Inc, San Jose, USA) using the Michaelis–Menten equation as fitting parameter.
123
Neurochem Res (2015) 40:410–419
Confocal Image Acquisition FRET measurements were performed on a Leica SP2 confocal microscope equipped with an argon laser (Leica, Wetzlar, Germany) upgraded with a 405 nm pulsed diode laser controlled by a Picoquant PDL-B laser driver (Picoquant, Berlin, Germany) and time-correlated single-photon counting detector DCC-100 (Becker and Hickl, Berlin, Germany). A 63X 1.2Na water HCX PL APO Lbd.bl objective was used for imaging. Sensitized Emission FRET Analysis Sensitized emission FRET was performed as described by Chen and Periasamy [26] using their PFRET-Analysis plugin for ImageJ (freeware, NIH, USA). N2a cells stably expressing the GAT-cerulean constructs or transiently expressing the positive control R1i15CFP were used as donor reference cells [donor alone (D)]. N2a cells transiently expressing GAT-venus constructs and R1i46YFP were used as acceptor reference cells [acceptor alone (A)]. The FRET samples [donor and acceptor (DA)] were prepared by transiently transfecting the stable expressing GAT-cerulean N2a cells with the GAT-venus constructs or in the case of the positive control a double transient transfection. The measured FRET efficiencies of the GAT dimers were compared to the background value (areas devoid of cells) and a Student’s t test was performed. Sensitized emission FRET was imaged on the above mentioned system using an Argon-laser exciting donor and acceptor at 458 and 514 nm, respectively. Donor and acceptor emission were gathered in the range of 462–510 and 520–660 nm, respectively. Each image was taken with a zoom factor of 5 with both frame averaging and line averaging set to 2. Before imaging laser power and PMT gain were adjusted to avoid to low uncorrected sensitized FRET signal and oversaturated donor signal in the acceptor and donor emission channels, respectively. All images were of 512 9 512 pixels at 8 bit color depth.
Results Validation of Biological Activity of GAT Fusion Proteins In order to assess the function of the GAT fusion proteins relative to the non-tagged versions, kinetic measurements were carried out to determine Km and Vmax values and the membrane localization was confirmed by confocal microscopy. As seen in Table 2 Km for the cerulean tagged
Neurochem Res (2015) 40:410–419
415
GAT1–3 was found to be closely related to the expected values around 12–19 lM and for BGT1 the Km of 49 lM was also close to the expected value. Confocal microscopy confirmed expression of the fusion proteins at the surface of the cells (results not shown). Sensitized Emission FRET Following the verification and validation of the basic pharmacological property of the GAT fusion proteins sensitized FRET studies were performed as described above using N2a cells as a model system. N2a cells were chosen since they grow without cell to cell contact allowing for better distinction of single cells when performing imaging studies. Hek-293 cells were also used for comparison generating comparable results (data not shown). To validate the sensitized emission FRET technique used in this study a positive FRET pair was generously donated by Professor Kathryn M. Partin. Specifically, AMPA receptors tagged with CFP and YFP (R1i15CFP and R1i46YFP) were obtained showing a FRET efficiency of 18.5 ± 2.9 % obtained by acceptor photobleaching [24]. As can be seen from Table 3 a FRET efficiency of 21.8 % was obtained when performing sensitized emission FRET, confirming FRET between the AMPA receptors and validating the imaging technique used in this study. To evaluate whether the obtained FRET signal was a consequence of random collision events a FRET experiment was initiated comparing cells with the same amount of donor (GAT1cerulean) with two different levels of acceptor (GAT1venus high and low). In the event of collision FRET the FRET efficiency would increase with higher amounts of acceptor added. As can be seen from Table 3 the combination of GAT1cerulean with GAT1venus high and GAT1venus low resulted in a FRET efficiency of 20.3 and 19.2 %, respectively suggesting that the FRET signal found is not due to random collision events. The FRET efficiency of the heterodimers consisting of GAT1 combined with GAT1, BGT1, GAT2, and GAT3 was estimated to be between 16.2 and 20.4 % showing a comparable efficiency regardless of the pair examined. The background FRET efficiency determined in areas devoid of cells was also calculated showing values of 7.0 % in the Table 2 Km and Vmax values for GABA on GAT fusion proteins
Km (lM) Vmax (nmol/ min mg)
GAT1cer
BGT1cer
GAT2cer
GAT3cer
15.3 ± 10.0
49.1 ± 14.3
18.5 ± 3.8
12.0 ± 5.8
7.8 ± 1.2
8.2 ± 0.7
5.5 ± 0.2
5.3 ± 0.6
The table shows Km and Vmax values ±SD for GABA of the GAT fusion proteins. The kinetic analysis was performed on N2a cells stably expressing the GATcerulean fusion proteins
Table 3 FRET efficiency of GAT homo/heterodimers FRET pair
sFRET efficiency
GAT1cerulean ? GAT1venus high
20.3 ± 10.3 (18)***
GAT1cerulean ? GAT1venus low
19.2 ± 6.2 (32)***
GAT1cerulean ? BGT1venus
16.2 ± 7.6 (25)***
GAT1cerulean ? GAT2venus
20.4 ± 6.4 (25)***
GAT1cerulean ? GAT3venus
16.9 ± 4.8 (24)***
Background R1i15CFP ? r1i46YFP Background
7.0 ± 3.4 (38) 21.8 ± 10.4 (12)*** 6.0 ± 1.0 (10)
The table shows a summary of the FRET efficiencies ± SD (number of analyzed cells) in N2a cells processed using the algorithm described by Chen and Periasamy [25]. *** P \ 0.0001 using a Student’s t test comparing to background. GAT1cerulean was analyzed with two different amounts of GAT1venus present to evaluate whether the FRET signal measured could be attributed to random collision events. Background sFRET efficiency was calculated in areas devoid of cells but within the same images used for the FRET pair measurements
GAT experiment and 6.0 % in the positive control experiment. Membrane Microdomain Localization of GATs The specificity of the commercial anti-GAT antibodies available were tested for cross reactivity using the stably transfected HEK-293 cells expressing the four different mouse GATs. The antibody towards GAT1 and BGT1 displayed no cross reactivity towards the other GAT subtypes and were considered selective. The GAT3 antibody showed a marginal cross reactivity towards GAT1 (not quantified). As can be seen in Fig. 1 GAT1 was detected in neuronal cultures and in adult mouse cortex but not in astroglial cell cultures. BGT1 was only observed in adult mouse cortex whereas GAT3 was observed in both primary cultures of neurons and astrocytes and in adult mouse cortex. Furthermore, GAT1 and BGT1 were primarily localized to non-raft membrane microdomains as can be seen by the intense bands in fractions 1–3 with only a slight intensity in fractions 5–12 regardless of the detergent used to isolate the detergent resistant membrane fractions in the adult mouse cortex. However, GAT1 does seem to be slightly associated to membrane rafts in primary neuronal cultures when using both detergents. In contrast GAT3 displayed quite the opposite localization when comparing the two detergents. When isolated with Triton-X 100 GAT3 localized to non-raft domains in neurons and adult mouse cortex showing intense bands in fraction 1–3. In astrocytes on the other hand, there might be a marginal trend to raft localization when using Triton-X 100. However, when isolated with Brij-98 GAT3 is seen to localize
123
416
Neurochem Res (2015) 40:410–419
Fig. 1 Membrane raft association of GAT1, BGT1, and GAT3. Membrane rafts were obtained in neurons (N), astrocytes (A), and mouse cortex (MC) treated with either 1 % Triton X-100 (T) or 1 % Brij 98 (B) (NT, NB, AT, AB, MCT, and MCB). GATs were detected at 70 kDa except for GAT1 in neurons which were found at 55 kDa.
The proteins from fraction 1–12 (bottom to top) were run on a 10 % SDS-PAGE. GAT1 and BGT1 are mostly associated to fraction 1–3 pointing towards a non-raft localization. GAT3 shows a non-raft distribution when treated with Triton X-100 (fraction 1–3) and a membrane raft association when treated with Brij 98 (fraction 5–12)
to raft domains as shown by the intense bands in fractions 5–12 in all three cell types.
leucine zipper might not be directly involved in the dimerization. While the present study and that performed by Scholze et al. [28] point towards the possibility of GATs forming dimers, the question is, whether such dimers are formed in an intact nervous system and not just in a heterologous expression system. Due to the fragile nature of primary neuronal cultures it was not possible to transfect these cells without significantly changing the morphology and killing the majority of the cell population and hence, a definite answer to this question cannot be provided. To shed more light on this issue the localization of GATs to membrane microdomains was performed using the detergent resistant membrane technique. The established membrane raft markers Flotillin-1 and Na?/K?ATPase [23, 29] were included in the study due to their differential membrane raft association depending on the detergent used to solubilize the cells (see Fig. 2). Flotillin-1 was found to be slightly more enriched in the raft fraction in astrocytes and cortex independent of the detergent used, whereas, it partitioned more uniformly in raft and non-raft fractions of neurons. This partitioning of Flotillin-1 has previously been reported by Hill et al. [29]. Na?/K?-ATPase as reported by Dalskov et al. [23] partitions exclusively in non-raft and raft fractions when Triton X-100 and Brij 98 is used, respectively, in all cell types tested. This shows that the established membrane raft associated proteins in this study were found in the expected fractions. This study showed, that in primary cultures of neurons and astrocytes and in mouse cortex GAT1 and BGT1 were found mostly in non-raft fractions, whereas, GAT3 was found to be membrane raft associated when Brij 98 was used as detergent but not when Triton X-100 was used. This indicates that GAT3 and Na?/K?-ATPase are found
Membrane Microdomain Localization of Established Proteins As part of the validation of the raft isolation two positive controls were chosen, i.e. Flotillin-1 and the Na?/K?ATPase. As can be seen in Fig. 2 Flotillin-1 localized to raft domains regardless of the detergent used whereas the Na?/K?-ATPase displayed a differential localization to non-raft and raft microdomains depending on whether it was isolated with Triton-X 100 or Brij-98, respectively.
Discussion The finding that the Km values found for the fluorescence tagged GAT fusion proteins resembled those reported for the native GATs expressed in HEK cells (see Bolvig et al. [27]) shows that the fusion proteins are likely to exhibit physiological properties in the plasma membrane of the cell lines used for expression, a notion supported by the high Vmax values that were also similar to those previously reported [27]. The results obtained from the FRET studies clearly show that GAT1 can form both homo- and heterodimers in the artificial environment of the N2a cell as well as HEK cells (data not shown). Studies by Scholze et al. [28] using the FRET technique also showed GAT1– GAT1 homodimers and in this study they suggested that the dimerization was due to a leucine zipper located in transmembrane domain 2 consisting of 4 leucine molecules separated by 6 amino acids. This leucine zipper can also be found in the mouse GAT1 but this repeat is not present in BGT1, GAT2, and GAT3 (see Table 4) suggesting that a
123
Neurochem Res (2015) 40:410–419
417
Fig. 2 Membrane raft association of the established membrane raft markers Flotillin-1 and Na?, K?-ATPase were detected at *45 and 110 kDa, respectively. Membrane rafts were obtained in neurons (N), astrocytes (A), and mouse cortex (MC) treated with either 1 % Triton X-100 (T) or 1 % Brij 98 (B) (NT, NB, AT, AB, MCT, and MCB). The proteins from fraction 1–12 (bottom to top) were run on a 10 %
SDS-PAGE. Flotillin-1 is almost evenly distributed in all the cell types exposed to both detergents with preference for membrane raft association in astrocytes and mouse cortex (fraction 5–12). Treatment with Triton X-100 revealed Na?, K?-ATPase as a soluble membrane component (non-raft, fraction 1–3). However, treatment with Brij 98 localizes Na?, K?-ATPase to membrane raft domains (fraction 5–12)
in the same membrane raft population. As reported in the literature, different detergents isolate different membrane raft populations [30]. The serotonin transporter (SERT) has been shown to associate to membrane rafts and non-raft fractions when Brij 58 and Triton X-100 were used as detergents [31]. The membrane raft association of SERT resembles that of GAT3 since both were found to associate to membrane rafts when a weaker detergent as Brij 58 or 98 was used but were solubilized in the presence of Triton X-100. The present study has demonstrated that GATs can form homo- and hetero-oligomers in N2a cells (and HEK-293 cells), but whether all combination possibilities are likely to have functional implications is questionable. Considering the membrane environment and localization studies mentioned above [7–14, 16, 17] it was reported that GAT1 and GAT3 were found on the apical surface of polarized MDCK cells and have been found on distal astrocytic processes in close proximity to GABAergic neurons, suggesting the possibility of interaction. However, in the present study GAT3 and GAT1 were found to be associated with different membrane microdomains since GAT3 partitioned into Brij 98 solubilized raft microdomains whereas GAT1 and BGT1 were not found in raft microdomains in the adult mouse cortex. BGT1 was found to be localized to distinct sites away from GABAergic synapses and to the basolateral surface of polarized MDCK cells [9, 16, 17]. Based on these findings it seems quite unlikely that hetero dimerization of GATs could be of major importance explaining the pharmacology observed for these transporters [1, 2]. The question that comes to mind is what is the physiological significance of GAT dimers? Unfortunately this study was unable to shed light on this issue. However, the raft association of GAT3 and the non-raft
association of GAT1 and BGT1 found in cortex from adult mice suggest that the micromembrane domain association of GATs may be relevant for their in vivo pharmacological profile. In order to obtain a synergistic anticonvulsant effect BGT1 needed to be inhibited using EF1502 combined with either a GAT1 selective compound like tiagabine or a GAT3 selective compound like SNAP-5114 [1, 18]. The selective inhibition of GAT1 and GAT3 only offered an additive anticonvulsant activity [18]. The localization of the GATs may appear to offer a reasonable explanation. Both GAT1 and GAT3 are localized synaptically or perisynaptically and inhibition of these two transporters in combination would only elevate the GABA levels around the synapse resulting in an activation of phasically activated GABA receptors. When inhibiting BGT1 which is located extrasynaptically GABA levels are elevated at sites containing tonically activated GABA receptors. With the combined inhibition of GATs in the synapse and extrasynaptically, two sets of GABA receptors are activated which results in the synergistic effect observed when combining EF1502 and tiagabine [2]. The fact that tiagabine and EF1502 indeed exhibit different mechanistic pharmacological differences was further established in a study where the two compounds were combined with the selective extrasynaptic GABA-A receptor agonist gaboxadol [4,5,6,7tetrahydroisoxazolo[5,4-c]pyridin-3-ol] [1]. Gaboxadol has a higher efficacy than GABA on the alpha-4 containing GABA-A receptors although being less potent [32–34]. As expected, inhibition of GAT1 using tiagabine was not able to elevate the GABA levels at extrasynaptic GABA-A receptors. However, inhibiting the extrasynaptically located BGT1 using EF1502 elevated the GABA levels significantly and caused a competition between the elevated
123
418
Neurochem Res (2015) 40:410–419
Table 4 Comparison of the amino acid sequence of GATs at the site of the GAT1 leucine zipper GAT1
R
F
P
Y
L
C
G
K
N
G
G
G
A
F
L
I
P
Y
F
L
BGT1
R
F
P
Y
L
C
Y
K
N
G
G
G
A
F
F
I
P
Y
F
I
GAT2
R
F
P
Y
L
C
Y
K
N
G
G
G
A
F
F
I
P
Y
L
I
GAT3
R
F
P
Y
L
C
Y
K
N
G
G
G
A
F
L
I
P
Y
V
V
GAT1
T
L
I
F
A
G
V
P
L
F
L
L
E
C
S
L
G
Q
Y
BGT1
F
F
F
S
C
G
I
P
V
F
F
L
E
V
A
L
G
Q
Y
GAT2
F
L
F
T
C
G
I
P
V
F
F
L
E
T
A
L
G
Q
Y
GAT3
F
F
I
C
C
G
I
P
V
F
F
L
E
T
A
L
G
Q
F
The table shows the amino acid sequence of the four mouse GATs from residues 80–118. The italics marking shows the leucine heptad repeat of GAT1 and the comparison to the other GATs. As it can be seen from the comparison only GAT1 contains this leucine zipper
GABA and gaboxadol at the extrasynaptic GABA-A receptors [1]. This underlines the above mentioned theory that in order to achieve a synergistic anticonvulsant activity GATs localized synaptically and extrasynaptically need to be inhibited simultaneously [1]. A recent study by Vogensen et al. [35] has investigated a new selective BGT1 inhibitor [4,4-Bis(3-methylthien-2yl)but-3-enyl](2-carboxycyclohex-2-enyl)methylammonium Chloride, compound 17b). An isobologram study was undertaken to investigate whether tiagabine and compound 17b would interact synergistically as one would expect based on the hypothesis that if BGT1 is inhibited by compound 17b and GAT1 is inhibited by tiagabine then both the extrasynaptic and synaptic GABA levels, respectively would rise and a synergistic anticonvulsant effect would be observed. However, the study only showed an additive effect. So it seems that the explanation of the synergistic effect observed between EF1502 and tiagabine might not be related to inhibition of BGT1 and GAT1, respectively. However, we do know from the Gaboxadol experiments described above, that tiagabine elevates synaptic GABA levels and EF1502 elevates extrasynaptic GABA levels. To investigate this more closely compound 17b should be tested in combination with Gaboxadol. This experiment would allow a conclusion as to whether or not selective inhibition of BGT1 can elevate the extrasynaptic GABA levels significantly to displace Gaboxadol. Until this experiment has been undertaken it is not possible with certainty to ascribe the synergistic effect between EF1502 and tiagabine to an inhibition of BGT1. Acknowledgments We would like to thank Ph.D. Steven S. Vogel, National Institute of Alcohol Abuse and Alcoholism, Bethesda, Maryland (20892), for providing cDNA of Cerulean and Venus and Professor Kathryn M. Partin, Colorado State University, Fort Collins (80523-1617), for providing the positive FRET constructs of AMPA receptors tagged with CFP and YFP (R1i15CFP and R1i46YFP). The work has been financially supported by the Carlsberg Foundation (2009_01_0501).
123
References 1. Madsen KK, Ebert B, Clausen RP, Krogsgaard-Larsen P, Schousboe A, White HS (2011) Selective GABA transporter inhibitors tiagabine and EF1502 exhibit mechanistic differences in their ability to modulate the ataxia and anticonvulsant action of the extrasynaptic GABA(A) receptor agonist gaboxadol. J Pharmacol Exp Ther 338:214–219 2. White HS, Watson WP, Hansen SL, Slough S, Perregaard J, Sarup A, Bolvig T, Petersen G, Larsson OM, Clausen RP, Frølund B, Falch E, Krogsgaard-Larsen P, Schousboe A (2005) First demonstration of a functional role for central nervous system betaine/{gamma}-aminobutyric acid transporter (mGAT2) based on synergistic anticonvulsant action among inhibitors of mGAT1 and mGAT2. J Pharmacol Exp Ther 312:866–874 3. Nielsen EB, Suzdak PD, Andersen KE, Knutsen LJ, Sonnewald U, Braestrup C (1991) Characterization of tiagabine (NO-328), a new potent and selective GABA uptake inhibitor. Eur J Pharmacol 196:257–266 4. Schousboe A, Bak LK, Waagepetersen HS (2013) Astrocytic control of biosynthesis and turnover of the neurotransmitters glutamate and GABA. Front Endocrinol (Lausanne) 4:102 5. Bender AS, Norenberg MD (2000) Effect of ammonia on GABA uptake and release in cultured astrocytes. Neurochem Int 36: 389–395 6. Schousboe A, Waagepetersen HS, Leke R, Bak LK (2014) Effects of hyperammonemia on brain energy metabolism: controversial findings in vivo and in vitro. Metab Brain Dis 29:913–917 7. Schousboe A, Madsen KK, Barker-Haliski ML, White HS (2014) The GABA synapse as a target for antiepileptic drugs: a historical overview focused on GABA transporters. Neurochem Res 39: 1980–1987 8. Conti F, Melone M, De Biasi S, Minelli A, Brecha NC, Ducati A (1998) Neuronal and glial localization of GAT-1, a high-affinity gamma-aminobutyric acid plasma membrane transporter, in human cerebral cortex: with a note on its distribution in monkey cortex. J Comp Neurol 396:51–63 9. Borden LA (1996) GABA transporter heterogeneity: pharmacology and cellular localization. Neurochem Int 29:335–356 10. Minelli A, DeBiasi S, Brecha NC, Zuccarello LV, Conti F (1996) GAT-3, a high-affinity GABA plasma membrane transporter, is localized to astrocytic processes, and it is not confined to the vicinity of GABAergic synapses in the cerebral cortex. J Neurosci 16:6255–6264
Neurochem Res (2015) 40:410–419 11. Conti F, Zuccarello LV, Barbaresi P, Minelli A, Brecha NC, Melone M (1999) Neuronal, glial, and epithelial localization of gamma-aminobutyric acid transporter 2, a high-affinity gammaaminobutyric acid plasma membrane transporter, in the cerebral cortex and neighboring structures. J Comp Neurol 409:482–494 12. Liu QR, Lopez-Corcuera B, Mandiyan S, Nelson H, Nelson N (1993) Molecular characterization of four pharmacologically distinct gamma-aminobutyric acid transporters in mouse brain [corrected]. J Biol Chem 268:2106–2112 13. Borden LA, Smith KE, Gustafson EL, Branchek TA, Weinshank RL (1995) Cloning and expression of a betaine/GABA transporter from human brain. J Neurochem 64:977–984 14. Zhu XM, Ong WY (2004) A light and electron microscopic study of betaine/GABA transporter distribution in the monkey cerebral neocortex and hippocampus. J Neurocytol 33:233–240 15. Zhou Y, Holmseth S, Hua R, Lehre AC, Olofsson AM, PobleteNaredo I, Kempson SA, Danbolt NC (2012) The betaine-GABA transporter (BGT1, slc6a12) is predominantly expressed in the liver and at lower levels in the kidneys and at the brain surface. Am J Physiol Renal Physiol 302:F316–F328 16. Pietrini G, Suh YJ, Edelmann L, Rudnick G, Caplan MJ (1994) The axonal gamma-aminobutyric acid transporter GAT-1 is sorted to the apical membranes of polarized epithelial cells. J Biol Chem 269:4668–4674 17. Ahn J, Mundigl O, Muth TR, Rudnick G, Caplan MJ (1996) Polarized expression of GABA transporters in Madin–Darby canine kidney cells and cultured hippocampal neurons. J Biol Chem 271:6917–6924 18. Madsen KK, Clausen RP, Larsson OM, Krogsgaard-Larsen P, Schousboe A, White HS (2009) Synaptic and extrasynaptic GABA transporters as targets for anti-epileptic drugs. J Neurochem 109(Suppl 1):139–144 19. Hertz E, Yu ACH, Hertz L, Juurlink BHJ, Schousboe A (1989) Preparation of primary cultures of mouse cortical neurons. In: Shahar A, de Vellis J, Vernadakis A, Haber B (eds) A dissection and tissue culture manual of the nervous system. Alan R. Liss Inc, New York, pp 183–186 20. Hertz L, Juurlink BHJ, Hertz E, Fosmark H, Schousboe A (1989) Preparation of primary cultures of mouse (rat) astrocytes. In: Shahar A, de Vellis J, Vernadakis A, Haber B (eds) A dissection and tissue culture manual of the nervous system. Alan R. Liss Inc, New York, pp 105–108 21. Hertz L, Juurlink BHJ, Szuchet S (1985) Cell Cultures. In: Lajtha A (ed) Handbook of Neurochemistry. Plenum Publishing Corporation, New York, pp 603–653 22. White HS, Sarup A, Bolvig T, Kristensen AS, Petersen G, Nelson N, Pickering DS, Larsson OM, Frølund B, Krogsgaard-Larsen P, Schousboe A (2002) Correlation between anticonvulsant activity and inhibitory action on glial gamma-aminobutyric acid uptake of
419
23.
24.
25.
26.
27.
28.
29.
30. 31.
32.
33. 34.
35.
the highly selective mouse gamma-aminobutyric acid transporter 1 inhibitor 3-hydroxy-4-amino-4,5,6,7-tetrahydro-1,2-benzisoxazole and its N-alkylated analogs. J Pharmacol Exp Ther 302:636–644 Dalskov SM, Immerdal L, Niels-Christiansen LL, Hansen GH, Schousboe A, Danielsen EM (2005) Lipid raft localization of GABA A receptor and Na?, K?-ATPase in discrete microdomain clusters in rat cerebellar granule cells. Neurochem Int 46:489–499 Bedoukian MA, Weeks AM, Partin KM (2006) Different domains of the AMPA receptor direct stargazin-mediated trafficking and stargazin-mediated modulation of kinetics. J Biol Chem 281:23908–23921 Larsson OM, Griffiths R, Allen IC, Schousboe A (1986) Mutual inhibition kinetic analysis of gamma-aminobutyric acid, taurine, and beta-alanine high-affinity transport into neurons and astrocytes: evidence for similarity between the taurine and beta-alanine carriers in both cell types. J Neurochem 47:426–432 Chen Y, Periasamy A (2006) Intensity range based quantitative FRET data analysis to localize protein molecules in live cell nuclei. J Fluoresc 16:95–104 Bolvig T, Larsson OM, Pickering DS, Nelson N, Falch E, Krogsgaard-Larsen P, Schousboe A (1999) Action of bicyclic isoxazole GABA analogues on GABA transporters and its relation to anticonvulsant activity. Eur J Pharmacol 375:367–374 Scholze P, Freissmuth M, Sitte HH (2002) Mutations within an intramembrane leucine heptad repeat disrupt oligomer formation of the rat GABA transporter 1. J Biol Chem 277:43682–43690 Hill WG, An B, Johnson JP (2002) Endogenously expressed epithelial sodium channel is present in lipid rafts in A6 cells. J Biol Chem 277:33541–33544 Mishra S, Joshi PG (2007) Lipid raft heterogeneity: an enigma. J Neurochem 103(Suppl 1):135–142 Mu¨ller HK, Wiborg O, Haase J (2006) Subcellular redistribution of the serotonin transporter by secretory carrier membrane protein 2. J Biol Chem 281:28901–28909 Storustovu SI, Ebert B (2006) Pharmacological characterization of agonists at delta-containing GABAA receptors: functional selectivity for extrasynaptic receptors is dependent on the absence of gamma2. J Pharmacol Exp Ther 316:1351–1359 Wafford KA, Ebert B (2006) Gaboxadol—a new awakening in sleep. Curr Opin Pharmacol 6:30–36 Cremers T, Ebert B (2007) Plasma and CNS concentrations of Gaboxadol in rats following subcutaneous administration. Eur J Pharmacol 562:47–52 Vogensen SB, Jorgensen L, Madsen KK, Borkar N, Wellendorph P, Skovgaard-Petersen J, Schousboe A, White HS, KrogsgaardLarsen P, Clausen RP (2013) Selective mGAT2 (BGT-1) GABA uptake inhibitors: design, synthesis, and pharmacological characterization. J Med Chem 56:2160–2164
123
