Brain Research, 297 (1984) 181-186 18 l Elsevier

BRE 20108

Diazepam enhances the action but not the binding of the GABA analog, THIP

JOHN H. SKERRITT and ROBERT L. MACDONALD

Department of Neurology, Universi O' ~/" Michigan Medical Center, Ann Arbor, M148109 ( U. S. A. i (Accepted November 29th, 1983)

Key words: GABA receptors -- -- anticonvulsant -- GABA agonists -- spinal cord cullurcs

GABA (4-aminobutyric acid) and its bicyclic analog THIP (4,5,6,7-tetrahydroisoxazolo-[4,5-c]-pyridin-3-ol) produced membrane hyperpolarization and increased chloride ion conductance of mouse spinal cord neurons in cell culture. Above I nM diazepam en- hanced the actions of both GABA and THIP with similar potency and efficacy. Diazepam has been shown to enhance the binding of [~H]GABA to rat brain membranes over similar concentration ranges, with the EC~0 values for enhancement of [;H]GABA binding and increase in membrane conductance being similar. In contrast, binding of [~H]THIP has been shown to be unaltered by diazepam under a variety of conditions. The possible reasons for such a discrepancy between these electrophysiological and neurochemical re- sults with THIP are discussed.

The inhibitory neurotransmitter, GABA, has a aptic rather than extrasynaptic localization z.~. widespread distribution in the central nervous sys- Diazepam enhanced the effects of synaptically re- tem, and multiple cellular actions of GABA, likely to leased or exogenous GABA in a wmety of central arise from effects on GABA subclasses, nervous system preparations 9, including cultured have been identified ll. To develop GABA agonists mouse and chick spinal cord neurons '',iT. Diazepam with selective actions, certain GABA analogs of re- also has been shown to enhance [-~H1GABA binding stricted conformation have been designed 12-H. One to rat brain membranes ~,j~,"-2,2:'. In neurochemical key compound is THIP, a bicyclic GABA analog, studies, however, anomalous interactions of diaze- which has potential therapeutic use since it had anti- pare and THIP have been documented. Unlike convulsant and analgesic activity after oral adminis- GABA. THIP failed to enhance [3H]diazepam bind- tration to rodents ~0`~4,j~. ing at 4 °C and instead antagonized GABA-induced Like GABA, TH1P produced -sensitive enhancement of binding4,I.L While inhibition of firing of cat spinal interneurons and dis- diazepam stimulated the binding of [~H]GABA and placed [~H]GABA binding to rat brain membranes [3H] to rat brain membranes, [3H]THIP with moderate potency 7.1<~5. However, important binding was unaltered by diazepam under a variety of differences between THIP and GABA have been incubation conditions 2~. demonstrated. Although THIP was less potent than Using mouse spinal cord neurons in primary disso- GABA in depolarizing isolated spinal fibers of the ciated cell culture, we have investigated the postsyn- rat, the compound was 20 times stronger than GABA aptic actions of GABA and THIP and the effects of in depression of spinal synaptic activity -~. GABA diazepam upon GABA and THIP responses. We re- usually depolarized apical dendrites of hippocampal port that both amino acids hyperpolarized neurons CA1 pyramidal cells, while THIP has been shown to by increasing chloride ion conductance, the re- elicit hyperpolarizing responses exclusively I. Thus, it sponses were -sensitive and the responses has been suggested that TH1P may selectively acti- to both amino acids were enhanced with similar po- vate a subclass of GABA receptors, possibly of syn- tency and to a similar extent by diazepam. We dem-

Correspondence: R. L. Macdonald, University of Michigan, Neuroscience Laboratory Building, 1103 East Huron, Ann Arbor, MI 48109, U.S.A.

I)006-8993/84/$03.IlI) © 1984 Elsevier Science Publishers 13, V. 182

A1. KAc B 1. GABA

Pre Picro Post ;i !

THIP 2001~M A2. KCI B2. lsec

Pre PIcro Post

GABA 8pM THIP 200pM Fig. 1. GABA and THIP responses recorded from spinal cord neurons in cell culture. A: miniperfusion (2 s) with 5/~M GABA or 2(10 .uM THIP increased membrane conductance and evoked hyperpolarizing responses during recordings with 4 M KAc-filled intraeellu- lar micropipettes (A 0, and evoked depolarizing responses with 3 M KCl-filled intracellular micropipettes (A2). Resting membrane potentials were --57 mV and --70 mV, respectively. B: picrotoxin (100 F~M) reversibly blocked the effects of iontophoretically applied GABA (8 nA, 400 ms, B 0 and THIP (45 hA, 400 ms, B2) during recordings with KCl-filled micropipettes. (Membrane hyperpolarized to --90 inV.) onstrate that enhancement of GABA responses and recording microplpette: data were recorded on a [3H]GABA binding are similar in dose-dependency 6-channel Gould polygraph. and discuss the possible reasons for difference in di- GABA (0.5 M, pH 3.2) or THIP (0.2 M. pH 3.2~ azepam action on THIP responses and [3H]THIP were applied iontophoretically using 400 ms duration binding. rectangular current pulses ( +0.5--60 hA) at 4-s inter- Cell cultures were prepared from spinal cords with vals. GABA or THIP were also applied at known attached dorsal root ganglia from 12-14-day-old concentrations by miniperfusion using large (2-10 mouse embryos as described earlier 21. Cultures were ~tm) tip diameter micropipettes (0.8 psi ejection pres- maintained for 5-8 weeks prior to electrophysiolog- sure, 2 s duration). ical experiments. Spinal cord neurons were pen- Diazepam or its vehicle (0.01% or less dimethyl- etrated under visual guidance on the modified stage sulfoxide) were applied by miniperfusion (0.2 psi of an inverted phase-contrast microscope, with high ejection pressure, 30 s duration) from a micropipette impedance (25-40 Mf2) glass micropipettes filled positioned 15-100~m from the soma of the cell under with 4 M potassium acetate (KAc) or 3 M potassium study. Such application of medium containing only chloride (KC1). Impalements were accepted only if 0.01% vehicle did not alter responses to iontophoret- membrane potential was greater than 45 mV and ac- ically applied GABA (99.7 +__ 0.7% control. 6 cells) tion potentials could be evoked. or THIP (98.6 -- 2.8% control. 6 cellsj. During as- Cultures were bathed in a phosphate-buffered sa- sessment of diazepam action upon amino acid re- line containing elevated magnesium ions to suppress sponses, cells were impaled with 3 M KCl-containing spontaneous activity (composition in mM: NaCI micropipettes, and GABA or THIP were applied 143.4, KCI 4.2, CaCI 2 0.9, MgCI 2 10.0 and glucose iontophoretically using constant current pulses. 5.6 in 9.5 mM sodium phosphate buffer at pH 7.35- GABA and TH1P responses of 6-9 mV amplitude 7.40) and maintained at 33-34 °C for electrophysio- were evoked following membrane hyperpolarization logical experiments. Using a conventional bridge cir- to --80 to --90 mV and used for assessment of diaze- cuit, simultaneous current injection and membrane pare actions Data were accepted only if the amino potential measurement were possible with a single acid responses returned to control level within 5 rain 183 following removal of the diazepam-containing pi- 35 pette from the vicinity of the cell under study. In most experiments diazepam effects on GABA and THIP 30 / E / responses were studied on the same cell, with 3 cycles v T T/ w 25 /! t ' of diazepam application possible. +,,', Z / When KAc recording micropipettes were used, ap- II T ,,'i GABA },, THIP plication of GABA or THIP to spinal cord neurons w T //. rapidly and reversibly produced membrane hyperpo- 15 ~ /// • larization and an increase in membrane conductance ) , i" 10 (Fig. 1Aj). Use of KCI-containing recording micro- JJ /¢ pipettes allowed chloride ions to enter the cell, chan- i 3 ging the equilibrium potential for chloride ions from about --65 mV to--20 mVl% and both GABA and 2 5 10 20 s'0,00 260 s60 TH1P responses became depolarizing (Fig. IA;). At (/JM) similar amplitudes, responses to THIP were of longer Fig. 2. Concentration-dependency of GABA and THIP re- duration than those to GABA, applied either ionto- sponses. Depolarizing responses were obtained following im- palement of cells with 3 M KCI-containing micropipettes anti phoretically or by miniperfusion for similar durations membrane hyperpolarizatJon to --80 to --90 inV. Responses (Fig. IA). In 21 cells for which GABA and TH1P re- at each concentration are averages of responses on 4-8 cells. sponses of approximately equal size (about 8 mV) Bars indicate _,+ standard error of the mean (S.E.M.). (Portions of the GABA data are from ref. 19.) were obtained following 4{){) ms iontophoretic appli- cations (KCl-filled recording micropipettes), the mean duration at half-maximal amplitude of GABA sponses were studied using miniperfusion and re- responses was 0.47 _+ 0.02 s and of THIP responses cording with 3 M KCl-containing micropipettes was {).66 _+ 0.02 s (I' < 0.001, 2-tailed Student's t- (Fig. 2), Threshold for GABA responses was ap- test). It has been suggested by otherst that the great- proximately 1-2/,M and for THIP 5 !~M, A 10 mV er duration of THIP action may be due to the absence depolarization was produced by approximately 4 ItM of active uptake mechanisms for this amino acid in GABA and 16 !~M THIP, a 20 mV depolarization neuronal tissue. Indeed, TH1P does not inhibit was produced by about 8 uM GABA and 60 ,M 13H]GABA uptake by minislices of rat cerebral cor- THIP and a 30 mV depolarization was produced by text5, and [~H]TH1P is not itself accumulated by such 19/,M GABA and 350!~M THIP. Thus, over the con- slices (Skerritt, unpublished observations). Howev- centration range studied, GABA was 2.5-25 times er, the GABA uptake inhibitor, nipecotic acid (200 more potent than THIP. TH1P has been shown to be /zM). failed to increase the duration of responses to 3.9-fold less potent than GABA as an inhibitor of GABA recorded from spinal cord neurons in cell cul- [~HJGABA binding to rat brain membranes ¢,. THIP ture (unpublished observations). The longer dura- has also been shown to be several-fold less potent tion of THIP responses could also arise from a great- than GABA in depolarization of ventral roots of neo- er mean chloride channel open time for THIP than natal rat spinal cord-" and depolarization of apical (}ABA, However, noise analysis studies have actual- dendrites of CA 1 hippocampal pyramidal cells~. The Ix.' revealed a shorter mean chloride channel open dose-response relationship for THIP was thus less time for THIP than for GABA ~. Thus, the basis for steep than that for GABA (Fig. 2), but the slopes of the greater duration of THIP responses is uncertain. log-log plots of both GABA and THIP dose response- Both GABA and THIP responses were virtually curves are greater than 1.0, suggesting that both abolished in a reversible fashion by the GABA an- GABA and TH1P bind cooperatively to receptors. tagonist, picrotoxin (lt)I),M) (Fig. 1B). The actions Diazepam (1-100{/ riM) enhanced the effects of of GABA and THIP have been shown also to be re- iontophoretically applied GABA (Fig. 3. filled cir- xersed by the GABA receptor antagonist, bicucul- cles) and TH1P (Fig. 3, open circles), while !J. I nM line. in the rat spinal cordl4. diazepam was ineffective• Maximal enhancement for The dose-dependencies of GABA and THIP re- both GABA and THIP occurred at 1{)() nM and half- 184

Pre Dz Post

GABA

F- 80 I nM ...... _...... Z W 6.vL._ 21ec 60 UJ 0 100nM ...... - ...... :.~ .... Z < 40 "1- THIP Z W @ 20

^ 0

i i I i,, i 0.1 1 10 100 1000 DIAZEPAM (nM) Fig. 3. Concentration-dependency of diazepam effects on GABA and THIP responses and on [3H]GABA and [3H]THIPbinding (binding data from ref. 23). 6-15 cells were tested at each diazepam concentration, bars indicate +_ S.E.M. Filled circles and squares represent GABA response and binding, respectively, empty circles and squares represent THIP response and binding, respectivelY. Data shown at right were obtained on one cell (MP = --82 mV) for studies of the effects of 1 nM diazepam on GABA and THIP re- sponses (applied by iontophoresis, KC1 recording) and another cell (MP = --80 mV) for 100 nM diazepam.

maximal enhancement at 2.6 +__ 0.5 nM (GABA) and azepam and 50% maximal enhancement at 4.6 nM. 3.6 + 0.7 nM (THIP) (THIP ECs0 not significantly While peak enhancement of GABA responses occur- different from GABA ECs0). The concentration-de- red at lower concentrations than that of [3H]GABA pendencies of diazepam enhancement of GABA and binding, it should be noted that the binding assays il- THIP responses were strikingly similar, and at no di- lustrated were incubated at 25 °C and the electro- azepam concentration tested was there a statistically physiological experiments at 33-34 °C: mcreases in significant difference between the degree of en- the incubation temperature have been shown to de- hancement of GABA and of THIP responses (Stu- crease the ECs0 of concentration-response curve for dent's t-test). However, since we did not determine enhancement of [3H]GABA binding by diazepam2X whether or not diazepam had comparable effects on The similarities between the concentration-response GABA and THIP dose-response curves, we cannot relationship for diazepam enhancement of [3H]GABA be certain that the mechanism of enhancement was binding and enhancement of GABA action would the same. suggest that the neurochemical results obtained with Diazepam (1-1000 nM) also enhanced the binding diazepam enhancement of [3H]GABA binding are of [3H]GABA to fresh, washed synaptosomal mem- closely related to the effects of diazepam observed in branes from rat brain z2,z3' but in contrast, did not en- various electrophysiological assays. A good correla- hance [3H]THIP binding under a variety of incuba- tion has also been shown between the degree of dis- tion conditions24. The concentration-dependency of placement of benzodiazepine binding in cerebral col- enhancement of [3H]GABA binding was quite similar tical cultures, and the concentration-dependency of to that of enhancement of GABA responses reported enhancement of GABA actions by diazepam z6. The in this study with significant enhancement of 1 nM di- disparity between diazepam effects. THIP action and 1~5 binding raises the question of whether [-~H]THIP and perpolarizing actions of a high TttlP concentration [3H]GABA are binding to the same sites, (5012M) by 1/~M diazepam in cultured cerebral corti- Binding studies with {3H]THIP suggest that the li- cal neurons. If a THIP receptor with a dissociation gand labels sites with properties similar to pH]GABA constant were to exist with a k'j of 2tl-50 uM, such a binding sites. [3H]THIP binding was inhibited by a site would not be detectable bv conventional recep-. wide range of GABA analogs with a similar order of tor binding methodologies that use a centrifugation potencies to their inhibition of [3H]GABA binding, or filtration means of separately bound and unbound and [-~H]THIP, like [-~H]GABA, labeled both higher radioligand due to the high dissociation rate expected and lower affinity sites in brain preparations 724. for a site of such low affinity. Whether such a site However, the maximal binding capacity of the would be a third binding site for [~tf]fHIP, or is de- [3H]THIP sites was considerably lower than that of rived from the lower (331 nM) affinit~ sile when [3HIGABA, suggesting that [3H]TH1P labeled only a binding is assayed under physiological conditions is subclass of GABA receptors. In addition, the affini- unknown. The functions of the subclasses of GABA ties of [-~H]THIP for its higher and lower affinity binding sites labeled by [-q-t]TttlP and unaffected bx binding sites were greater than those of [3H]GABA, is unclear: possib b these sites may although the potency of THIP on spinal neurons in be involved in another known action of IHIP. such as cell culture was found to be 2.5-25-fold lower than analgesia >. that of GABA. Under standard [3H]GABA binding The finding that diazepam enhances Ftt lP actions conditions (Tris-citrate buffer, 2 °C), the equilibrium may be of possible clinical application, While both dissociation constant (K,t) of low affinity [-~H]GABA THIP and diazepam have potent anticonvulsant ef- binding to washed synaptosomal rat brain mem- fects upon acute administration to mice, tolerance branes was 822 nM >, near the threshold for GABA develops to the anti-epileptic activity of diazepam ~, responses observed on spinal cord neurons in cell cul- and certain side effects (trelnor, sedation, ataxia and ture (1-2 I2M). However, the K d value for pH]THIP dizziness) may prevent moderate doses of TItIP be- was 331 nM, while the threshold for THIP responses ing used in the clinical setting 2', Since the en- was some 15-fold higher (51~M). hancement of THIP effects by diazcpam could lead Several explanations arc possible for the discrep- to a synergistic rather than purely additive anticon- ancy between the neurochemical and electrophysio- vulsant effect of the two compounds, clinical testing logical results obtained with THIP. Under a w~riety of low dose combinations of -[HIP and henzodiaze- of conditions diazepam did not enhance the binding pine may prove worthwhile. of [~H]THIP to the lower affinity site found in brain membranes. It is possible that a THIP receptor of We thank Professor Graham Johnston, Dr. Pml even lower affinity exists. The threshold for THIP ac- Krogsgaard-Larsen and Dr. Mary Ann Werz for tions in spinal cord neurons in cell culture was 5 jiM, their advice, J.H.S. is a recipient of an award from with larger depolarizations (10 mV) requiring ap- the Rotary Foundation of Rotary International and proximately 16 uM THIP. Similarly, for en- the research was supported by NSF (}rant 019722 hancement of benzodiazepine binding (at 30 °C), and NIH Grants NS (I04(/8 and NS 19692 to R.I,.M. threshold effects occurred at 10 I~M THIP while the We wish to acknowledge the gifts of diazepam (Hoff- ECs0 value was approximately 40 I~M 2s. Other work- man-La Roche, Nutley, N J) and THIP (Lundbeck, ers > have demonstrated the enhancement of the hy- Copenhagen, Denmark ).

[ Alger, B. E. and Nicoll, R. A., Pharmacological evidence 3 Barker. J. L., MacDonald, J. F., Mathers, l) A., McBur- for two kinds of GABA receptor on rat hippocampal py- ney, R. N. and ()ertel, W., GABA receplor funcmms m ramidal cells studied in vilro, J, Physiol. (Lond. L 328 cultured mouse spinal neurons. In F. V. De Fcudis and P. (1982) 125-14l. Mandel (Eds.), Amino Acid .,Veurotransmttter~, Raven 2 Allan, R. D., Evans, R. H. and Johnston, G. A. R.. y-Ami- Press, New York, 1981,pp.281 293. nobuLvric acid agonists: an m vitro comparison between de- 4 Braestrup, C., Nielsen, M., Krogsgaard-l,arsen, P. and pression ot spinal synaptic activity and depolarization ol Falch, E.. Partial agonists for brain GABA/benzodiazepine spinal root fibres in the rat, Brit. J. Pharmacol.. 70 (1980) receptor complex, Nature (Lond.), 28(I (197¢)) ~31- 333. 609-615, 5 Browne, T. R. and Penry, J. K., Benzodiazcpines in the 186

treatment of epilepsy, Epilepsia, 14 (1973) 277-310. GABA receptor agonists. In D. M. Woodbury, J. K. Penry 6 Choi, D. W., Farb, D. H. and Fishbach, G. D., Chlordiaze- and C. E. Pippenger (Eds.), Antiepileptic Drugs, Raven poxide selectively augments GABA action in spinal cord Press, New York, 1982, pp. 839--858. cell cultures, Nature (Lond.), 269 (1977) 342-344. 17 Macdonald, R. L. and Barker, J. L., Benzodiazepines spe- 7 Falch, E. and Krogsgaard-Larsen, P., The binding of the cifically modulate GABA-mediated postsynaptic inhibition specific GABA agonist [3H]THIP to rat brain synaptic in cultured mammalian neurons, Nature (Lond.), 271 membranes, J. Neurochem., 38 (1982) 1123-1129. (19781 563-564. 8 Guidotti, A., Toffano, G, and Costa, E., An endogenous 18 Meiners, B. A. and Salama, A. I~, Enhancement of benzo- protein modulates the affinity of GABA and benzodiaze- diazepine and GABA binding by the novel anxiolytic, tra- pine receptors in rat brain, Nature (Lond.), 275 (19781 cazolate, Europ. J. Pharmacol,, 78 (1982) 315-322. 553-555. 19 Nowak. L M.. Young, A. B. and Macdonald. R. L.. 9 Haefely, W., Pieri, L., Polc, P. and Schaffner, R., General GABA and bicuculline actions on mouse spinal cord and pharmacology and neuropharmacology of benzodiazepine cortical neurons in cell culture. Brain Research. 244 (19821 derivatives, Handb. Exp. Pharmacol., 55 (II) (19811 155-164, 13-283. 20 Petersen. H. R., Jensen. I. and Dam. M. THIP: a single 10 Hill, R. C., Maurer, R., Buescher, H. H. and Roemer, R., blind controlled trial in patients with epilepsy. Acta neurol. Analgesic properties of the GABA-mimetic THIP, Europ. scand.. 67 (19831 114-117. J. Pharmacol., 69 (1981) 221-224. 21 Ransom. B. R.. Neale. E.. Henkart. M , Bullock, P. N. 11 Johnston, G. A. R., GABA receptors. In J. B. Lombardini and Nelson. P. G.. Mouse spinal cord cells in culture. 1. and A. D. Kenny (Eds.), The Role of Peptides and Amino Morphologic and intrinsic neuronal eleetrophysiological Acids as Neurotransmitters, Alan R. Liss, New York, 1981, properties. J. Neurophysiol.. 40 ( 19771 1132-1150. pp. 1-17. 22 Skerritt, J. H.. Chen Chow, S. and Johnston. G. A, R., Dif- 12 Johnston, G. A. R., Allan, R. D., Kennedy, S. M. E. and ferences in the interactions between GABA and benzodia- Twitchin, B., Systematic study of GABA analogues of re- zepine binding sites. Neurosei. Lett.. 33 (19821 17,3--178. stricted conformation. In P. Krogsgaard-Larsen, J. Scheel- 23 Skerritt. J. H.. Willow. M. and Johnston, G. A. R.. Diaze- Kruger and H. Kofod (Eds.), GABA-Neurotransmitters, pare enhancement of low affinity GABA binding to rat Munksgaard, Copenhagen, 1978, pp. 149-164. brain membranes, Neurosci. Lett., 29 (19821 63-66. 13 Karobath, M., Placheta, P., Lippitsch, M. and Krogsgaard- 24 Skerritt. J. H. and Johnston, G. A. R.. Diazepam stimu- Larsen, P., Is stimulation of benzodiazepine receptor bind- lates the binding of GABA and muscimol but not THIP to ing mediated by a novel GABA receptor? Nature (Lond.), rat brain membranes, Neurosci. Lett.. in press, 278 (1979) 748-749. 25 Supavilai. P. and Karobath, M.. The effect of temperature 14 Krogsgaard-Larsen, P., Johnston, G. A. R., Lodge, D. and and chloride ions on the stimulation of [3H]flunltrazepam Curtis, D. R., A new class of GABA agonist, Nature binding by the muscimol analogues THIP and piperidine-4- (Lond.), 268 (1977) 53-55. sulfonic acid. Neurosci. Lett.. 19 (19801 337-341. 15 Krogsgaard-Larsen, P. and Johnston, G. A. R., Structure- 26 White, W. F., Dichter, M. A. and Snodgrass, S. R.. Benzo- activity studies on the inhibition of GABA binding to rat diazepine binding and interactions with the GABA recep- brain membranes by muscimol and related compounds..1. tor complex in living culture of rat cerebral cortex, Brain Neurochem., 30 (1978) 1377-1382. Research, 215 (1981) 162-176. 16 Lloyd. K. G. and Morselli, P. L.. Potential anticonvulsants-