Proc. Nati. Acad. Sci. USA Vol. 83, pp. 9241-9244, December 1986 Neurobiology The permeability of y-aminobutyric acid-gated chloride channels is described by the binding of a "cage" , t-butylbicyclophosphoro[35S]thionate (anions/picrotoxinin) H. HAVOUNDJIAN*t, S. M. PAUL*, AND P. SKOLNICK*§ *Laboratory of Bioorganic Chemistry, National Institute of Arthritis, Diabetes, and Digestive and Kidney Diseases; tHoward Hughes Medical Institute; and *Clinical Neuroscience Branch, National Institute of Mental Health, National Institutes of Health, Bethesda, MD 20892 Communicated by J. E. Rall, August 7, 1986

ABSTRACT The "cage" convulsant t-butylbicydophos- supramolecular complex (12-15). However, neither the pre- phoro[35S]thionate ([-"SJTBPS) binds with high affinity to sites cise location of [35S]TBPS binding sites nor the relationship at or near a y-aminobutyric acid (GABA)-gated chloride between the binding characteristics of this radioligand to the channel according to current hypothesis. We now report that functional state of the chloride channel are known. the potencies of a series of anions in enhancing [3 S]TBPS We now report that under a defined set of conditions the binding correlated highly with their relative permeabilities effects of anions on the apparent affinity (Kd) of [35S]TBPS through GABA-gated chloride channels. Furthermore, statis- highly correlates with the presumed binding of these anions tically significant correlations are obtained between the appar- to a selectivity site in the chloride-channel; we conclude that ent affinity (Kd) of [35S]TBPS estimated in the presence of these [35S]TBPS binding is related to the permeability of GABA- anions and their relative permeabilities through GABA-gated gated chloride channels. These findings strongly suggest that chloride channels. The latter relationships obtained whether [35S]TBPS binds directly to GAfA-gated chloride channels. the Kd of [35S]TBPS as estimated in rat cerebral cortex was Thus, [35S]TBPS may be used to estimate the permeability correlated with the relative permeabilities of these anions in characteristics of GABA-gated chloride channels and should either frog dorsal root gangion cells or primary cultures of prove valuable for both in vitro and in vivo studies of the mouse spinal cord neurons. These findings strongly suggest biophysical properties of this channel. that [35S]TBPS binds to GABA-gated chloride channels and that the apparent affinity of this radioligand is directly related METHODS to the permeability of these channels. Thus, radioreceptor techniques using [35S]TBPS may provide a simple means of Adult male Sprague-Dawley rats (Taconic Farms, German- describing permeability characteristics of GABA-gated chlo- town, NY) were subjected to an ambient temperature swim ride channels. for 10 min and immediately killed by decapitation. Swim stress increases both the maximal number of [35S]TBPS Many psychoactive drugs such as the (1-3) binding sites and the apparent affinity ofthis radioligand (16). and (3-5) affect neuronal excitability by aug- This procedure increased the sensitivity of the radioreceptor menting y-aminobutyric acid (GABA)-gated chloride current assay sufficiently to analyze the effects of anions under through the plasma membrane. Biochemical studies have identical incubation conditions. Cerebral cortex (pooled demonstrated that these effects are mediated through an frontal, parietal, and temporal) was removed, weighed, and oligomeric protein complex referred to as the benzodiaze- homogenized in 50 volumes of ice-cold 50 mM Tris-HCl pine/GABA receptor-chloride ionophore complex ("supra- buffer, pH 7.4, using a Brinkmann Polytron (setting 6-7 for molecular complex") (6, 7). The availability of selective, 15 sec). The homogenates were centrifuged at 20,000 x g for high-affinity ligands for this receptor complex has permitted 20 min at 40C, and the resulting pellets were resuspended in a detailed examination of the physical and functional rela- 50 volumes of 50 mM Tris citrate buffer, pH 7.4/100 mM tionships between the and GABA recogni- NaCl. The tissue was resuspended and recentrifuged four tion sites (8, 9). In contrast, neurochemical characterization more times in the Tris citrate/NaCl buffer. The resulting of the GABA-gated, benzodiazepine receptor-coupled chlo- pellets were washed a final time in 50 mM Tris citrate and ride channel has been hampered by the lack of a suitable were resuspended in 50 volumes of Tris citrate buffer. high-affinity ligand. [3H]a-, a derivative [35S]TBPS binding was determined in a total volume of 1 ml of the "cage" convulsant picrotoxinin and the first radio- consisting of 400 A.l of tissue suspension (0.3-0.4 pug of ligand used to characterize sites associated with the GABA- protein)/100 1ul of [35S]TBPS (60-100 Ci/mmol; 1 Ci = 37 gated chloride channel (10, 11), proved unsatisfactory be- GBq; New England Nuclear/buffer or salt solutions to final cause ofits low affinity (1-2 ,AM) and high level ofnonspecific volume. Reactions were initiated by addition of the binding (10, 11). radioligand and terminated after 90 min at 250C by rapid Recently, t-butylbicyclophosphoro[35S]thionate ([35S]- filtration and washing (two 5-ml aliquots of50 mM Tris citrate TBPS), a cage convulsant structurally related to picrotoxinin, buffer, pH 7.4, through Whatman GF/B filter strips using a was reported to bind to sites associated with the GABA-gated Brandel M-24R filtering manifold (Brandel Instruments, chloride channel with high affinity (Kd =50 nM) (12). Because Gaithersburg, MD). Scatchard plots were constructed adding ofthe high affinity and low nonspecific binding of [35S]TBPS, various amounts of TBPS (New England Nuclear) to a fixed this compound has been used to characterize these binding concentration (-5 nM) of radioligand to achieve the desired sites and their relationship to other components of the final concentration. Nonspecific binding was defined using 10

The publication costs of this article were defrayed in part by page charge Abbreviations: TBPS, t-butylbicyclophosphorothionate; GABA, Y- payment. This article' must therefore be hereby marked "advertisement" aminobutyric acid. in accordance with 18 U.S.C. §1734 solely to indicate this fact. §To whom reprint requests should be addressed. 9241 Downloaded by guest on September 27, 2021 9242 Neurobiology: Havoundjian et al. Proc. Natl. Acad. Sci. USA 83 (1986) jLM picrotoxinin (Sigma) and was usually <20% of total 600 binding under standard conditions. All anions were used as the sodium salts. Protein was determined using the Miller (17) modification of the Lowry et al. technique (18). 0 CH3COO-

RESULTS While it has been established that anions (e.g., Br- and CP-) 400[ are required for optimum binding of [355]TBPS (12), the precise relationship between the physicochemical properties E of these ions and the binding of [35S]TBPS is not known. Thus, [35S]TBPS binding was studied in the presence of each ui0l of a series of eight monovalent anions that showed a wide range of permeabilities through chloride channels (19). In the first series of experiments, the concentration of [35S]TBPS was held constant at a concentration well below *cl- the Kd (estimated in the presence of 200 mM Cl-), and the anion concentrations were varied between 15-1600 mM (Fig. 1). Perchlorate and thiocyanate were the most potent ions in stimulating [35S]TBPS binding with EC50 values (defined as kI- the ion concentration increasing [35S]TBPS binding to 50% of 1.0 2.0 3.0 maximum) of 18 and 25 mM, respectively. The least potent anion in this series was acetate, with an EC50 of =550 mM. Px/PCI- The limited solubility of NaF prevented an estimation of the FIG. 2. Relationship between anion permeability in frog dorsal potency of fluoride, but this ion was among the least potent. root ganglion and [35S]TBPS binding. The EC50 effective in increas- Significant differences were also observed in the efficacies of ing [35S]TBPS binding as estimated from Fig. 1 of a series of anions the ions in increasing [35S]TBPS binding, with the was compared with the permeability of anions, relative to chloride, most, and perchlorate the least efficacious. A biphasic effect (Px/PcI-), in the bullfrog spinal ganglion. The limited solubility of was observed with thiocyanate, perchlorate, and iodide: low NaF prevented an accurate estimation of the EC50 of fluoride. concentrations of these anions stimulated, and high concen- Permeability data were taken from ref. 19; r = 0.90, P < 0.01. trations inhibited [35S]TBPS binding (Fig. 1). Because the movements of anions through chloride chan- apparent affinity (Kd) of [35S]TBPS. For example, in the nels appear to be controlled, at least in part, by anion presence of anions at 25 mM (Fig. 3), the Kd of [3 S]TBPS in interactions with a binding site in the channel (20-22), the the presence ofthe most permeable ion, thiocyanate (Kd =35 permeabilities, relative to chloride, for a number of anions nM), and the least permeable ion, acetate Kd "170 nM), measured in the bullfrog dorsal root ganglion (19) were varied almost five-fold. No consistent effects on the number compared with their potencies in stimulating [35S]TBPS of [35S]TBPS binding sites were observed with any ofthe ions binding. A plot of ion permeabilities, relative to chloride, as tested except thiocyanate, which typically increased the Bm. a function oftheir potencies in stimulating [35S]TBPS binding 20-30% relative to the values obtained with the other ions shows these parameters to be highly correlated (r = 0.90; P (Fig. 3). < 0.01) (Fig. 2). The Kd of [35S]TBPS determined in the presence of fixed In the second series of experiments the concentration of concentrations of anions (25 mM) and their permeabilities, [35S]TBPS was varied from 5 to 320 nM and the anion relative to chloride, in bullfrog dorsal root ganglion correlated concentrations were held constant at 10, 25, 40, or 50 mM. In highly (r = 0.96; P < 0.001) (Fig. 4). Moreover, similar this concentration range the anions all increased, [35S]TBPS correlations were obtained between the Kd of [35S]TBPS and binding (Fig. 1). Scatchard plots of this data demonstrated the permeabilities of these anions, relative to chloride, at 10, that at a fixed concentration of anion, the differences in 40, and 50 mM in bullfrog ganglion, and at 25, 40, and 50 mM potencies were almost exclusively due to an effect on the in cultured mouse spinal neurons (22) (Table 1).

100r DISCUSSION coCU There is good evidence that the benzodiazepine/GABA 80- receptor-chloride ionophore complex (supramolecular com- 0E plex) mediates the anxiolytic, anticonvulsant, sedative-hyp- 60 notic, and muscle relaxant actions of many clinically useful compounds (8, 9). The introduction of the cage convulsant cn [35S]TBPS (12) has made it possible to study the picrotoxinin 40 F binding site, which is distinct from recognition sites for both cn ct GABA and benzodiazepines on the supramolecular complex 20 F (12-15). Electrophysiological studies demonstrating that , which contains equal amounts of picrotoxinin and the inactive cage compound picrocrotin, noncompeti- 0 10 100 1000 tively inhibits GABA-gated chloride currents suggest that Anion, mM cage (e.g., picrotoxinin, TBPS) act at sites located at or near the chloride channel (23), although this FIG. 1. Effect of anions of [35S]TBPS binding to rat cerebral conclusion has been disputed (24). Moreover, the relation- cortex: *, NaCl; o, NaBr; A, NaNO3; *, NaI; r, NaSCN; *, NaC104; v, NaF; and v, NaC2H302. The concentration of [35S]TBPS in these ship between the characteristics ofradioligand binding to this experiments was 5 nM. This experiment is representative of three site and the functional state of the chloride channel is not experiments that had similar results. known. Downloaded by guest on September 27, 2021 Neurobiology: Havoundjian et al. Proc. Natl. Acad. Sci. USA 83 (1986) 9243

a)

~0

0.008

0 100 200 300 400 500 600 700 Bound FIG. 3. Scatchard analysis of [35S]TBPS binding in rat cortical membranes: effects of anions. In this representative experiment, anion concentration was 25 mM. Similar findings were obtained when tissues were assayed in the presence of 10, 40, or 50 mM concentrations ofanions (see Table 1). The concentration of [35S]TBPS in these experiments ranged from 5-325 nM. TBPS was added to [35S]TBPS (5 nM) to achieve the appropriate ligand concentration. Each assay contained 0.4 mg of protein. o, NaCl (Kd, 86 nM; Bn.,, 526 fmol per assay); A, NaBr (Kd, 49 nM; B., 523 fmol per assay); o, NaF (Kd, 149 nM; B,,,, 409 fmol per assay); *, Nal (Kd, 44 nM; B.,~, 443 fmol per assay); v, NaSCN (Kd, 37 nM; Bmgs, 669 fmol per assay); v, NaCl04 (Kd, 62 nM; B,,,, 475 fmol per assay); A, NaC2H302 (Kd, 170 nM; Bm,,, 443 fmol per assay); *, NaNO3 (Kd, 42 nM; Bmx,, 470 fmol per assay). Consequently, we investigated the effects of a series of The equilibrium binding properties of [35S]TBPS were also anions with different permeabilities through chloride chan- examined in the presence of fixed concentrations of anions nels on [35S]TBPS binding. It had been shown that certain (10-50 mM) that enhanced [35S]TBPS binding. Scatchard anions (e.g., bromide, chloride, thiocyanate) (12) significant- plots of the data in the presence of 25 mM anions (Fig. 3) ly increase [35S]TBPS binding. However, a systematic study demonstrated an effect primarily on the apparent affinity (Kd) attempting to quantitatively relate the effects ofthese anions of [35S]TBPS, rather than the number of binding sites (Bmax). on [35S]TBPS binding and their physicochemical properties However, small (-25%) but consistent changes in the Bmax of with respect to chloride channels has not been reported. [35S]TBPS were observed in the presence of thiocyanate All the anions used in this study increased [35S]TBPS (10-50 mM), the most permeant species tested (Fig. 3 and binding (Fig. 1), although their potencies varied widely. A data not shown). This increase could be due to an "unmask- statistically significant correlation (r = 0.90; P < 0.01) was ing" of sites, or the conversion ofa population oflow-affinity observed between the potencies of seven of the anions tested sites to a higher-affinity state now amenable to detection by (the potency of fluoride could not be accurately estimated the radioreceptor technique. due to the limited solubility of NaF) and their permeabilities Since the primary effect of anions appeared to be on the relative to chloride, in the bullfrog dorsal root ganglion (19) apparent affinity of [35S]TBPS, the relationship between the (Fig. 2). Three anions-iodide, perchlorate, and thiocyan- Kd of this ligand and ion permeability was examined. Highly ate-exerted a biphasic effect on [35S]TBPS binding, with significant correlations were observed between the relative enhancement observed at low, and inhibition at higher permeabilities of these ions determined by voltage clamp concentrations. A similar biphasic effect of iodide and techniques in the bullfrog dorsal root ganglion (19) and the Kd thiocyanate on [35S]TBPS binding has been reported (12). of [35S]TBPS (Fig. 4, Table 1). Furthermore, good correla- Such inhibition could be nonspecific, possibly related to the tions were also obtained between the Kd of [35S]TBPS and the chaotropic properties of these ions. At very high concentra- permeabilities of halides (determined by single-channel re- tions (> 2M) chloride and bromide also inhibited [35S]TBPS cording) (22) in cultured mouse spinal cord neurons. The (data not shown), which suggests that this inhibitory action finding that the selectivity ofanions for GABA-gated chloride may be common to all anions if sufficiently high concentra- channels differs dramatically from their relative hydrated tions could be achieved. The differences observed in the sizes (19) supports the presence ofspecific anion binding sites maximum increase in radioligand binding produced by the in the channel (19, 22). The direct relationship between the various anions (Fig. 1) could be related to differences in the apparent affinity of [35S]TBPS and relative ion permeabilities summation oftheir stimulatory effects that are related to their through GABA-gated chloride channels strongly suggests permeabilities (Fig. 2) and their nonspecific, inhibitory ef- that cage convulsants like picrotoxinin and TBPS act directly fects (Fig. 1). at this anion binding site in the chloride channel. Downloaded by guest on September 27, 2021 9244 Neurobiology: Havoundjian et al. Proc. Natl. Acad. Sci. USA 83 (1986) pretreatment of brain membranes with phospholipase A2 produces a concentration-dependent reduction in the number of [35S]TBPS binding sites (25) that is paralleled by a reduction in -stimulated 36C1- efflux from synap- 0 CH3COO- toneurosomes (26). These findings suggest the number of [35S]TBPS binding sites may be directly proportional to the number of GABA-gated chloride channels that are in an "open" conformation. Electrophysiological studies will be necessary to confirm or refute this hypothesis. The use of simple neurochemical techniques to study the radioligand binding properties of cage convulsants like [35S]TBPS may provide important insights into both the functional relationship of the GABA-gated chloride channel 100 \ to other components of the supramolecular complex and the physiological roles of these sites. 00- The authors thank Dr. Charles Edwards, National Institute of Diabetes, Digestive, and Kidney Diseases, for helpful discussion and XCl04- making his manuscript available to us prior to publication (19), and Mr. Michael Jackson for his excellent technical assistance. H.H. was 50 - Br- a Howard Hughes Medical Research Scholar during these studies. NO *0 SCNI 1. Haefely, W., Kyburz, E., Gerecke, M. & Mohler, H. (1985) in Advances in Drug Research, ed. Testa, B. (Academic, Lon- don), Vol. 14, pp. 166-322. 2. Study, R. & Barker, J. (1983) J. Am. Med. Assbr. 247, 2147-2151. 0 1.0 2.0 3.0 3. Study, R. & Barker, J. (1981) Proc. Natl. Acad. Sci. USA 78, 7180-7184. PX/ PCI- 4. Nicoll, R. & Wojtowicz, J. (1980) Brain Res. 191, 225-237. 5. Schulz, W. & MacDonald, R. (1981) Brain Res. 209, 77-188. FIG. 4. Relationship between anion permeability and apparent 6. Tallman, J., Paul, S., Skolnick, P. & Gallager, D. (1980) affinity of [35S]TBPS. Abcissa: PX/Pc1-, anion permeability relative Science 207, 274-281. to chloride (from ref. 19). Ordinate: Kd of [35S]TBPS in nM estimated 7. Skolnick, P. & Paul, S. (1982) Int. Rev. Neurobiol. 23, at anionic concentration of 25 mM. r - 0.96, P < 0.001. 103-140. 8. Squires, R., ed. (1986) GABA & Benzodiazepine Receptors The high correlations between the apparent affinity of (CRC, Boca Raton, FL), in press. [35S]TBPS and the relative permeability of anions through 9. Usdin, E., Skolnick, P., Tallman, J., Greenblatt, D. & Paul, GABA-gated chloride channels should make detection of S., eds. (1983) Pharmacology ofBenzodiazepines (MacMillan, subtle changes in the permeability characteristics of these London). to neurochemical For 10. Ticku, M., Ban, M. & Olsen, R. (1978) Mol. Pharmacol. 14, channels amenable simple analysis. 391-402. example, we have recently demonstrated (16) that acute 11. Leeb-Lundberg, F. & Olsen, R. (1980) in Psychopharma- stress rapidly increases the apparent affinity of [35S]TBPS in cology and Biochemistry ofNeurotransmitter Receptors, eds. rat cerebral cortex. The mechanism by which this effect Yamamura, H., Olsen, R. & Usdin, E. (Elsevier/North Hol- occurs is not understood, but the present findings suggest that land, New York), pp. 593-606. stress increases the permeability of GABA-gated chloride 12. Squires, R., Casida, J., Richardson, M. & Saederup, E. (1983) channels. Mol. Pharmacol. 23, 326-336. 13. Supavilai, P. & Karobath, M. (1983) Eur. J. Pharmacol. 91, Although the significance of the number of [35S]TBPS 145-146. binding sites is unclear, we have recently observed that 14. Supavilai, P. & Karobath, M. (1984) Mol. Pharmacol. 24, 1193-1200. Table 1. Relationship between the apparent affinity of 15. Ticku, M. & Ramanjaneyulu, R. (1984) Pharmacol. Biochem. [35S]TBPS and anion permeability of Behav. 21, 1573-1579. GABA-gated chloride channels 16. Havoundjian, H., Paul, S. & Skolnick, P. (1986) Brain Res. 375, 403-408. A-, Dorsal root ganglion Spinal neuron 17. Miller, G. (1959) Anal. Chem. 31, 964. mM r t P r t P 18. Lowry, O., Rosebrough, N., Farr, A. & Randall, R. (1951) J. 10 0.90 4.53 <0.01 0.60 1.04 <0.4 Biol. Chem. 193, 265-275. 25 0.96 8.43 <0.001 0.96 4.94 <0.05 19. Inomata, N., Oomura, Y., Akaike, N. & Edwards, C. (1986) 40 0.97 8.49 <0.001 0.94 4.06 <0.10 Neurosci. Res. 3, 371-383. <0.01 20. Edwards, C. (1982) Neuroscience 7, 1335-1365. 50 0.92 5.10 <0.01 0.99 9.34 21. Hille, B. (1984) Ionic Channels of Excitable Membranes, Scatchard plots were constructed from data of seven or eight ions (Sinauer, Sunderland, MA). at concentrations between 10 and 50 mM. The apparent affinities (Kd 22. Hamill, O., Bormann, J. & Sakmann, B. (1983) Nature (Lon- values) of [35S]TBPS estimated at a single ion concentration of 10-50 don) 305, 805-808. mM were correlated with the relative permeabilities of these ions at 23. Akaike, N., Hattori, K., Oomura, Y. & Carpenter, D. (1985) GABA-gated chloride channels in the frog dorsal root ganglion (19) Experientia 41, 70-71. and cultured fetal mouse spinal cord neuron (22). Only halide ions- 24. Shank, R., Pong, S., Freeman, A. & Graham, L. (1974) Brain fluoride, chloride, iodine, and bromide-were studied in the latter Res. 72, 71-78. report. r = correlation coefficient; t = Student's t value; P = 25. Havoundjian, H., Cohen, R., Paul, S. & Skolnick, P. (1986) J. probability of statistical significance. Data correlating the Kd of Neurochem. 46, 804-811. [35S]TBPS estimated in the presence of25 mM anions and the relative 26. Schwartz, R., Skolnick, P., Seale, T. & Paul, S. (1986) in permeabilities ofthese anions in the bullfrog dorsal root ganglion are GABAergic Transmission & Anxiety, eds. Biggio, G. & Costa, illustrated in Fig. 4. E. (Raven, New York), Vol. 41, pp. 33-50. Downloaded by guest on September 27, 2021