- Email: [email protected]
GABAB receptor activation protects GABAA receptor from cyclic AMP-dependent down-regulation in rat cerebellar granule cells
GABAB receptor modulation
Pergamon PII: S0306-4522(99)00257-2
Neuroscience Vol. 93, No. 3, pp. 1077–1082,1077 1999 Copyright q 1999 IBRO. Published by Elsevier Science Ltd Printed in Great Britain. All rights reserved 0306-4522/99 $20.00+0.00
GABAB RECEPTOR ACTIVATION PROTECTS GABAA RECEPTOR FROM CYCLIC AMP-DEPENDENT DOWN-REGULATION IN RAT CEREBELLAR GRANULE CELLS ` ,* A. CUPELLO† and M. ROBELLO*‡ B. BARILA *I.N.F.M., Dipartimento di Fisica dell’Universita` di Genova, Genova, Italy †Centro di Neurofisiologia Cerebrale, C.N.R., Genova, Italy
Abstract—Interaction between GABAA and GABAB receptors was studied in rat cerebellar granule cells in culture, by the wholecell patch-clamp approach. Our data show that the GABAB agonist (2)baclofen is not able, per se, to significantly change the muscimol-activated chloride current. However, (2)baclofen dose-dependently prevents the reduction of GABAA receptor function by forskolin, an activator of adenylate cyclase. The effect of baclofen is mediated by a pertussis toxin-sensitive G protein. In fact, in cells treated with pertussis toxin, baclofen and forskolin, the toxin is able to block baclofen action, allowing forskolin to act fully. The protective effect by GABAB receptor activation under these circumstances is most probably related to the prevention of cyclic AMP increases after forskolin treatment. In fact, in these neurons cyclic AMP and protein kinase A activation result in a downregulation of GABAA receptor function. On the whole, the data indicate the presence of complex modulation of GABAA receptors by GABAB receptor types in cerebellum granule cells. q 1999 IBRO. Published by Elsevier Science Ltd. Key words: GABAA modulation, patch-clamp, GABAB receptors, G protein.
The main inhibitory neurotransmitter in the mammalian brain, GABA, acts via two main receptor families: GABAA and GABAB. The former family consists of GABA-activated chloride channels, which are pentameric structures made up of five protein subunits around a central chloride pore. 24,32 GABAB receptors were discovered and characterized as (2)baclofen-sensitive and bicuculline-insensitive a little less than 20 years ago. 5,17 Functionally, they cause postsynaptic inhibition activating a K 1 conductance and presynaptic inhibition of neurotransmitter release probably blocking Ca 21 channels. 3,4,35 These receptors belong to the family of the G protein-associated receptors and two forms of them have been recently cloned. 20 The distribution in the brain of GABAA and GABAB receptors has been studied and they show similar concentrations in some regions, but they differ in the overall rank order of their concentration. 6 A clear and well-known example of co-localization of these two receptor types in the same neuron is in the coupling of fast (GABAA-mediated) and slow (GABAB-mediated) inhibitory postsynaptic potentials (IPSPs) described in several neuronal types. 7,10,15,23,26 These two phases of postsynaptic hyperpolarizing effects involve two different ionic types, Cl 2 and K 1, respectively, and display a limited level of interaction. However, an indirect interaction is the one by which GABAB presynaptic autoreceptors block GABAA (and GABAB) postsynaptic action, inhibiting GABA release. 8,9,14,27,28 It is possible that the G protein mechanisms set into motion by GABAB receptors may alter second messenger systems in a postsynaptic target neuron so to modify GABAA receptor function. This may happen via, for
example, phosphorylation/dephosphorylation events. GABAA receptors of cerebellar granule cells have been shown by our lab to be down-regulated by second messenger activated serine/threonine kinases such as protein kinase A (PKA) and G. 29,30 The possibility of this direct interaction between GABAA and GABAB receptors has been evaluated in the present work in rat cerebellum granule cells in culture. EXPERIMENTAL PROCEDURES
Cell culture Granule cells were prepared from cerebella of eight-day-old Wistar rats following the procedure of Levi et al., 22 as previously described. 30 Briefly, the minced tissue was first suspended in trypsin (0.25 mg/ml, Type III Sigma) for 15 min at 378C in a shaking water bath, and then in deoxyribonuclease and trypsin inhibitor. Finally, it was dispersed by gently drawing it into a fire-polished Pasteur pipette. Cells were resuspended in basal Eagle’s medium with Earle’s salts supplemented with 10% fetal calf serum (Gibco Bio-Cult, U.K.), 25 mM KCl, 2 mM glutamine and 100 mg/ml gentamicine and plated on poly-l-lysinecoated glass coverslips placed in 20 mm plastic dishes at a density of 1x10 6 per dish and kept at 378C in a humidified 95% air–5% CO2 atmosphere. Experiments were performed between days 5 and 12 after plating. Solutions The standard external solution consisted of (in mM): 135 NaCl, 5.4 KCl, 1.8 CaCl2, 1 MgCl2, 5 HEPES, 10 glucose. The pH was adjusted to 7.4 using NaOH. The pipette filling solution contained (in mM): 142 KCl, 10 HEPES, 2 EGTA, 4 MgCl2, 3 ATP. The pH was adjusted to 7.3 with a Trizma base. Electrophysiology Membrane currents were measured with the standard whole-cell patch-clamp technique by an EPC-7 (List-medical). Patch electrodes were manufactured from borosilicate glass capillaries (Type 1406129 Hilgenberg, Malsfield, Germany) with a programmable Sachs and Flaming puller (model PC-84) with a resistance of 5–10 MV when filled with the internal standard solution. The access resistance was not different among different experimental groups and it did not change significantly during the period of data acquisition. Cell responses were
‡To whom correspondence should be addressed. Abbreviations: EGTA, ethyleneglycolbis(aminoethylether)tetra-acetate; HEPES, N-2-hydroxyethylpiperazine-N 0 -2-ethanesulphonic acid; IPSP, inhibitory postsynaptic potential; PKA, protein kinase A; PTX, pertussis toxin. 1077
B. Barila` et al.
1078
filtered at 3 kHz. Capacitance transient neutralization and series resistance compensantion were optimized. Both stimulation and data acquisition were performed with a Labmaster D/A A/D converter driven by a pCLAMP software (Axon Instruments, Burlingame, CA). Analysis was performed with pCLAMP and SIGMA PLOT (Jandel Scientific, Erkrath, Germany) software. Data are given as mean^S.E.M. (cell number). All experiments were performed at room temperature. Drug application procedure The holding potential was set to 280mV in all experiments reported as this was the most suitable condition for recording the total chloride current. Run-down phenomena of chloride currents were prevented by the presence of ATP in the internal solution. Chloride currents were elicited by 10 26 M muscimol perfusion (flow speed 4–6 ml/min) at 2 min intervals. In order to verify a possible interaction between the two receptors, first of all only GABAB receptor was activated with the specific agonist (2)baclofen and then, in succession, using muscimol agonist, GABAA receptor function was tested. Stock solutions of baclofen (10 mM) and forskolin (50 mM) were diluted at the desired final concentration when added to the culture medium bathing the cells. The drug treatments were performed with cells incubated at 378C for at least 30 min; afterwards the culture medium was substituted with external solution and finally the cell patched was superfused with 10 26M muscimol. In some experiments the culture was treated with pertussis toxin (500 ng/ml) in the culture medium for 16–20 h. All chemicals were purchased from Sigma Chemical Co., St Louis, MO.
RESULTS
Muscimol acting on GABAA receptors of the granule cells in culture activates chloride currents, which under our registration conditions are inward (with Cl 2 going out of the cells). These currents reach a peak current (IP) followed by a development of steady-state level (ISS) over continued application of the agonist (Fig. 1A). The concentration–effect curve for muscimol is reported in Fig. 1B. The experimental points could be fitted by the Hill equation: Y A·
Xn Bn 1 X n
1
where Y is the peak muscimol activated current and X is the muscimol concentration. The parameters A and B are, respectively, the maximum current and the concentration of muscimol corresponding to A/2. The best fit gave A2,320 pA, B1.6 mM and n0.8. In these experiments we have applied drugs, forskolin, (2)baclofen or both and evaluated the muscimol activated Cl 2 current change over time. Chloride current referred to time 0 was the one recorded in the presence of the drugs before washout. Under these conditions, an increase in muscimol-activated current upon washout of the drugs (baclofen, forskolin or both) means that the drug treatment was inhibiting GABAA receptors. When cells treated with 200 mM baclofen added to the culture medium for at least 30 min were tested with 1 mM muscimol immediately after the treatment and onward, the amplitude of the peak of the current activated by muscimol did not significantly change. In fact, the peak changed only by 7^5% (six cells) after 8 min washing with external solution (Fig. 1A, C). However, forskolin (50 mM) treatment resulted in a down-regulation of GABAA receptor activity (Fig. 2A, C). This effect was mainly for the rapidly desensitizing (peak) component. In fact, the ratio ISS/IP became 0.18^0.02 (n6)
Fig. 1. Effect of baclofen on the granule cells chloride current. (A) Currents activated by 1 mM muscimol in one cell incubated with 200 mM baclofen for at least 30 min. Traces were clamped at 280 mV, after 0 (left) and 8 (right) min of washing with external solution. (B) Dose–response curve. Semilogarithmic plot of the peak current amplitude as a function of muscimol concentration. Each cell was voltage clamped at 280 mV. Each point is the average of 3420 cells. Theoretical fitting is obtained using the Hill equation (1). The best fit gave A2,320 pA, B1.6 mM and n0.8. (C) Normalized peak current activated by 1 mM muscimol at 280 mV as a function of the washing time in cells treated as described above (A). The data are the average of at least eight values.
after the washout of forskolin, whereas immediately after the forskolin treatment was 0.28^0.03 (n6).
GABAB receptor modulation
1079
Fig. 3. Dose dependence of the forskolin effect. Each cell was treated with the same dose of forskolin and different concentrations, C, of baclofen. Points represent mean and the number of cells tested is indicated in the brackets. Experimental points can be fitted using equation (2). The best fit gave A0.98, B2.5 mM and n0.4.
In this case, X is the baclofen concentration and Y was computed using the following expression: " !# Iin 12 Ifin " !#bf 3 Iin 12 Ifin f
Fig. 2. Combined effect of baclofen and forskolin. Currents activated by 1 mM muscimol in granule cells clamped at 280 mV, where the whole-cell configuration was reached after 0, 4, 8 min of washing with external standard solution. (A) In one cell preincubated with 50 mM forskolin (B) in one cell preincubated with 200 mM baclofen and 50 mM forskolin for at least 30 min. (C) Normalized peak current activated by 1 mM muscimol as a function of washing time. (W) in cells preincubated with 50 mM forskolin and (X) in cells preincubated with 200 mM baclofen and 50 mM forskolin. The data are the average of at least seven values.
In this case, the effect of 50 mM forskolin was completely inhibited by 200 mM baclofen such that muscimol-activated current remained constant over time of washing (Fig. 2B, C). These data must be compared with those observed in cells treated only with 50 mM forskolin, where an increase by 59^3% (eight cells) was found after 6 min washing with external solution (Fig. 2A, C). Also the change in ISS/IP ratio was prevented by baclofen. Different concentrations of (2)baclofen were used to test its action on inhibition of GABAA receptor activity induced by forskolin. The experimental points could be fitted using the equation: Xn Y A· 1 2 n : 2 B 1 Xn
where the initial peak current (Iin) refers to the current in the presence of the drugs and the final peak current (Ifin) after their washing out; in this expression the numerator refers to cells treated with baclofen and forskolin while the denominator to cells treated only with forskolin. The parameters A and B are, respectively, the maximum effect of forskolin and the concentration of baclofen corresponding to A/2. The best fit gave A0.98, B2.5 mM and n0.4 (Fig. 3). After incubation with pertussis toxin (PTX) for 16 h and then with 200 mM baclofen and 50 mM forskolin for at least 30 min, the chloride current increased by 43^5% (11 cells) after 6 min washing with external solution (Fig. 4A, B). This experiment showed that PTX blocked the effect of baclofen leaving forskolin free to act. In fact, similar results were obtained with cells pretreated with PTX and then only with forskolin. In these conditions, in nine cells there was an increase of 60^6% of muscimol response following a 6-min wash with the external solution (Fig. 4C, D). These observations indicate that PTX per se does not influence the forskolin effect whereas it prevents the baclofen effect by blocking a G protein. DISCUSSION
In a previous report of our own together with another lab, 2 it was shown that GABAB receptor activation by (2)baclofen, their prototypical agonist, 17 prevented the forskolin induced increase in cAMP in rat cerebellum granule cells. The biphasic concentration–response curve for the antagonism by CGP52432 suggested that more than one subtype of GABAB receptors was involved. In the present experiments we went further, evaluating the
1080
B. Barila` et al.
Fig. 4. Effect of PTX. Each cell was incubated with PTX for 12–16 h and then treated at least for 30 min either with 200 mM baclofen and 50 mM forskolin (A, B) or only with 50 mM forskolin (C, D). (A) Currents activated by 1 mM muscimol in one cell after 0, 4, 8 min perfusing with external standard solution. (B) Normalized peak current activated by 1 mM muscimol at 280 mV as a function of washing time. The data are the average of at least seven values. (C) Currents activated in one cell by 1 mM muscimol after 0, 4, 8 min perfusing with external standard solution. (D) Normalized peak current activated by 1 mM muscimol at 280 mV as a function of washing time. The data are the average of at least nine values.
functional consequences of the block of cAMP increases in terms of GABAA receptor activity, which has been shown to be down-regulated by PKA. 30 GABAB receptors have been demonstrated to have several functions. A classical division of their effect is in their negative action on presynaptic Ca 21 channels and their activation of postsynaptic K 1 channels. 3 This last effect results in time delayed late IPSPs in the hippocampus which are not sensitive to block by bicuculline. 1,13,26 These K 1-mediated late IPSPs involve the intermediation of a G protein. 3,19 In addition, GABAB receptors, at postsynaptic sites in the case of cerebellar granule cells interfere with the activity of adenylate cyclase via PTX-sensitive Gi proteins. 37 A well-known apparent paradox is the fact that GABAB receptor activation results in a decrease in forskolin-activated accumulation of cAMP in rat brain slices, whereas it causes a larger increase in cAMP under the activation of b adrenergic receptors. 11,16,18,21 These apparently contradictory results appear to find their explanation in the different pattern of regulation for the various isoforms of adenylate cyclase (I–VIII), which may be present in different cells and subcellular compartments. 25,34 In fact, type II adenylate cyclase is activated by GSa picomolar concentration, which is liberated by b adrenergic receptor stimulation, but can be further activated by the bg complex. 12,33 However, the effect of bg requires higher (nanomolar) concentrations, which can be reached under the concurrent stimulation of Gi proteins via GABAB receptors. 34 Referring to cerebellum, the conditional stimulatory effect on adenylate cyclase, via GABAB receptor activation does not appear to be present. 11 In addition, rat cerebellar granule cells in culture did not show stimulation of cAMP synthesis by isoproterenol, a b-adrenergic agonist. 37 No increase of cAMP was also caused by prostaglandin E1, serotonin and
vasoactive intestinal protein. Thus, in granule cells the conditional activation of adenylate cyclase II by GABAB receptors may not be present. In contrast, we did not find in cerebellar granules cells in culture any change in cAMP levels by (2)baclofen, whereas baclofen could block the forskolininduced increase in cAMP. 2 This last result had already been described by another group. 37 We report here that in granule cells forskolin down-regulates GABAA receptors. This result confirms our previous data 30 that activation of PKA, via cAMP increases, impairs GABAA receptor function. Simultaneous application of (2)baclofen erases the forskolin effect with an EC50 of 2.5 mM. This result is in accord with the prevention of cAMP level increase by GABAB receptor activation in granule cells in culture. 2 Evidently, the PKA activation via cAMP increase affects mainly the rapidly desensitizing component as it was appreciated by the change in the ISS/IP ratio. This effect too was prevented by baclofen. On the other hand, a preferential down-regulation of the rapidly desensitizing component of GABA-activated chloride current by protein kinase G was already shown by us. 29 These results may suggest that GABA or muscimol activate two different GABAA receptor populations differentially sensitive to down-regulation by serine/threonine kinases. The (2)baclofen prevention of forskolin-induced increases in cAMP seems to suggest that this diterpene activates a type I adenylyl cyclase in these cells. This enzyme isoform is in fact inhibited by both GIa and bg. 34 This conclusion is strengthened by the fact that the baclofen effect on GABAA receptors is prevented by PTX. Cerebellar granule cells indeed express mRNA for type I (and type II) adenylate cyclase. 25 From a functional point of view, the protective effect of
GABAB receptor modulation
GABAB on GABAA receptors may play a role under those physiological conditions in which type I adenylate cyclase is activated. One of such conditions, according to the paradigms proposed by Taussig and Gilmann, 34 is an increase of intracellular Ca 21. In fact, type I adenylyl cyclase is one of those directly activated by nanomolar concentrations of Ca 21/ calmodulin. In a previous report, we have shown that Ca 21 influx into granule cells via N-methyl-d-aspartate (NMDA)-activated channels down-regulates GABAA receptor function. 31 Although that effect was accounted for by nitric oxide synthase and calcineurin activation, increases in cAMP via adenylate cyclase activation may represent another pathway
1081
of inhibition of GABAA receptors. Under this circumstance GABAB receptor activity may partially rescue their ionotropic counterparts. However, it should be made clear that GABAB/GABAA receptor interaction modality is most probably different in different neuronal populations. In fact, Xi et al. 36 have reported that in bullfrog dorsal root ganglion cells GABAB receptor activation depresses GABAA receptor function, most probably via cAMP increases. This is not surprising in view of the wide range of possible situations in terms of GABAB receptors responsive G proteins, adenylate cyclase types (see Discussion above) and, of course, of GABAA receptors isoforms. 24,32
REFERENCES
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.
Benardo L. S. (1995) N-methyl-d-aspartate transmission modulates GABAB-mediated inhibition of rat hippocampal pyramidal neurons in vitro. Neuroscience 68, 637–643. Bonanno G., Fassio A., Robello M., Amico C., Marchetti C., Sale R. and Raiteri M. (1996) Pharmacology of GABAB receptors modulating cyclic AMP production and calcium currents in rat granule cells. Pharmac. Rev. Commun. 8, 99–102. Bormann J. (1988) Electrophysiology of GABAA and GABAB receptor subtypes. Trends Neurosci. 11, 112–116. Bowery N. G. (1993) GABAB receptor pharmacology. A. Rev. Pharmac. Toxicol. 33, 109–147. Bowery N. G., Hill D. R., Hudson A. L., Doble A., Middlemiss D. N., Shaw J. and Turnbull M. (1980) (2)Baclofen decreases neurotransmitter release in the mammalian CNS by an action at a novel GABA receptor. Nature 283, 92–94. Bowery N. G., Hudson A. L. and Price G. W. (1987) GABAA and GABAB receptor site distribution in the rat central nervous system. Neuroscience 20, 365–383. Crunelli V., Haby M., Jassik-Gerschenfeld D., Leresche N. and Pirchio M. (1988) Cl 2 and K 1 dependent inhibitory postsynaptic potentials evoked by interneurones of the rat lateral geniculate nucleus. J. Physiol. 399, 153–176. Davies C. H., Davies S. N. and Collingridge G. L. (1990) Paired-pulse depression of monosynaptic GABA mediated inhibitory postsynaptic responses in rat hippocampus. J. Physiol. 424, 513–531. Deisz R. A. and Prince D. A. (1989) Frequency dependent depression of inhibition in guinea pig neocortex in vitro by GABAB receptor feedback on GABA release. J. Physiol. 412, 513–542. Dutar P. and Nicoll R. A. (1988) A physiological role for GABAB receptors in the central nervous system. Nature 332, 156–158. Enna S. J. and Karbon E. W. (1984) GABAB receptors and transmitter-stimulated cAMP accumulation in rat brain. Neuropharmacology 23, 821–822. Federman A. D., Conklin B. R., Schrader K. A., Reed R. R. and Bourne H. R. (1992) Hormonal stimulation of adenylyl cyclase through Gi-protein bg subunits. Nature 356, 159–161. Ga¨hwiler B. H. and Brown D. A. (1985) GABAB-receptor-activated K 1 current in voltage-clamped CA3 pyramidal cells in hippocampal cultures. Proc. natn. Acad. Sci. U.S.A. 82, 1558–1562. Harrison N. L., Lange G. D. and Barker J. L. (1988) (2)Baclofen activates presynaptic GABAB receptors on GABAergic inhibitory neurons from embryonic rat hippocampus. Neurosci. Lett. 85, 105–109. Hasuo H. and Gallagher J. P. (1988) Comparison of antagonism by phaclofen of baclofen induced hyperpolarizations and synaptically mediated late hyperpolarizing potentials recorded intracellularly from rat dorsolateral septal neurons. Neurosci. Lett. 86, 77–81. Hill D. R. (1985) GABAB receptor modulation of adenylate cylase activity in rat brain slices. Br. J. Pharmac. 84, 249–257. Hill D. R. and Bowery N. G. (1981) 3H-baclofen and 3H-GABA bind to bicuculline-insensitive GABAB sites in rat brain. Nature 290, 149–152. Hill D. R. and Dolphin A. C. (1984) Modulation of adenylate cyclase activity by GABAB receptors. Neuropharmacology 23, 829–830. Hille B. (1992) Ionic Channels of Excitable Membranes. pp. 170–201. Sinauer, Sunderland, MA. Kaupmann K., Huggel K., Heid J., Flor P. J., Bischoff S., Mickel S. J., Mc Master G., Angst C., Bittiger H., Froestl W. and Bettler B. (1997) Expression cloning of GABAB receptors uncovers similarity to metabotropic glutamate receptors. Nature 386, 234–246. Knight A. R. and Bowery N. G. (1996) The pharmacology of adenylyl cyclase modulation by GABAB receptors in rat brain slices. Neuropharmacology 35, 703–712. Levi G., Aloisi F., Ciotti M. T. and Gallo V. (1984) Autoradiographic localization and depolarization induced release of amino acids in differentiating granule cell cultures. Brain Res. 290, 77–86. McCormick D. A. (1989) GABA as an inhibitory neurotransmitter in human cerebral cortex. J. Neurophysiol. 62, 1018–1027. McKernan R. M. and Whiting P. J. (1996) Which GABAA receptor subtypes really occur in the brain? Trends Neurosci. 19, 139–143. Mons N. and Cooper D. M. F. (1995) Adenylate cyclases: critical foci in neuronal signaling. Trends Neurosci. 18, 536–542. Newberry N. R. and Nicoll R. A. (1985) Comparison of the action of baclofen with GABA on rat hippocampal pyramidal cells in vitro. J. Physiol. 360, 161–185. Pearce R. A., Grunder S. D. and Faucher L. D. (1995) Different mechanisms for use-dependent depression of two GABAB mediated IPSCs in rat hippocampus. J. Physiol. 484, 425–435. Raiteri M. and Bonanno G. (1993) Multiple GABAB receptors. Trends pharmac. Sci. 14, 259–261. Robello M., Amico C., Bucossi G., Cupello A., Rapallino M. V. and Thellung S. (1996) Nitric oxide and GABAA receptor function in the rat cerebellar cortex and cerebellar granule cells. Neuroscience 74, 99–105. Robello M., Amico C. and Cupello A. (1993) Regulation of GABAA receptors in cerebellar granule cells in culture: differential involvement of kinase activities. Neuroscience 53, 131–138. Robello M., Amico C. and Cupello A. (1997) A dual mechanism for impairment of GABAA receptor activity by NMDA receptor activation in rat cerebellum granule cells. Eur. Biophys. J. 25, 181–187. Stephenson A. (1995) The GABAA receptors. Biochem. J. 310, 1–9. Taussig R., Tang W.-J., Hepler J. R. and Gilman A. G. (1994) Distinct patterns of bidirectional regulation of mammalian adenylyl cyclases. J. biol. Chem. 269, 6093–6100. Taussig R. and Gilman A. G. (1995) Mammalian membrane-bound adenylyl cyclases. J. biol. Chem. 270, 1–4. Thompson S. M. (1994) Modulation of inhibitory synaptic transmission in the hippocampus. Prog. Neurobiol. 42, 575–609.
1082
B. Barila` et al.
36. Xi Z.-X., Yamada K., Tsurusaki M. and Akasu T. (1997) Baclofen reduces GABAA receptor responses in acutely dissociated neurons of bullfrog dorsal root ganglia. Synapse 26, 165–174. 37. Xu J. and Wojcik W. J. (1986) GABAB receptor-mediated inhibition of adenylate cyclase in cultured cerebellar granule cells: blockade by islet-activating protein. J. Pharmac. exp. Ther. 239, 568–573. (Accepted 19 April 1999)
Pergamon PII: S0306-4522(99)00257-2
Neuroscience Vol. 93, No. 3, pp. 1077–1082,1077 1999 Copyright q 1999 IBRO. Published by Elsevier Science Ltd Printed in Great Britain. All rights reserved 0306-4522/99 $20.00+0.00
GABAB RECEPTOR ACTIVATION PROTECTS GABAA RECEPTOR FROM CYCLIC AMP-DEPENDENT DOWN-REGULATION IN RAT CEREBELLAR GRANULE CELLS ` ,* A. CUPELLO† and M. ROBELLO*‡ B. BARILA *I.N.F.M., Dipartimento di Fisica dell’Universita` di Genova, Genova, Italy †Centro di Neurofisiologia Cerebrale, C.N.R., Genova, Italy
Abstract—Interaction between GABAA and GABAB receptors was studied in rat cerebellar granule cells in culture, by the wholecell patch-clamp approach. Our data show that the GABAB agonist (2)baclofen is not able, per se, to significantly change the muscimol-activated chloride current. However, (2)baclofen dose-dependently prevents the reduction of GABAA receptor function by forskolin, an activator of adenylate cyclase. The effect of baclofen is mediated by a pertussis toxin-sensitive G protein. In fact, in cells treated with pertussis toxin, baclofen and forskolin, the toxin is able to block baclofen action, allowing forskolin to act fully. The protective effect by GABAB receptor activation under these circumstances is most probably related to the prevention of cyclic AMP increases after forskolin treatment. In fact, in these neurons cyclic AMP and protein kinase A activation result in a downregulation of GABAA receptor function. On the whole, the data indicate the presence of complex modulation of GABAA receptors by GABAB receptor types in cerebellum granule cells. q 1999 IBRO. Published by Elsevier Science Ltd. Key words: GABAA modulation, patch-clamp, GABAB receptors, G protein.
The main inhibitory neurotransmitter in the mammalian brain, GABA, acts via two main receptor families: GABAA and GABAB. The former family consists of GABA-activated chloride channels, which are pentameric structures made up of five protein subunits around a central chloride pore. 24,32 GABAB receptors were discovered and characterized as (2)baclofen-sensitive and bicuculline-insensitive a little less than 20 years ago. 5,17 Functionally, they cause postsynaptic inhibition activating a K 1 conductance and presynaptic inhibition of neurotransmitter release probably blocking Ca 21 channels. 3,4,35 These receptors belong to the family of the G protein-associated receptors and two forms of them have been recently cloned. 20 The distribution in the brain of GABAA and GABAB receptors has been studied and they show similar concentrations in some regions, but they differ in the overall rank order of their concentration. 6 A clear and well-known example of co-localization of these two receptor types in the same neuron is in the coupling of fast (GABAA-mediated) and slow (GABAB-mediated) inhibitory postsynaptic potentials (IPSPs) described in several neuronal types. 7,10,15,23,26 These two phases of postsynaptic hyperpolarizing effects involve two different ionic types, Cl 2 and K 1, respectively, and display a limited level of interaction. However, an indirect interaction is the one by which GABAB presynaptic autoreceptors block GABAA (and GABAB) postsynaptic action, inhibiting GABA release. 8,9,14,27,28 It is possible that the G protein mechanisms set into motion by GABAB receptors may alter second messenger systems in a postsynaptic target neuron so to modify GABAA receptor function. This may happen via, for
example, phosphorylation/dephosphorylation events. GABAA receptors of cerebellar granule cells have been shown by our lab to be down-regulated by second messenger activated serine/threonine kinases such as protein kinase A (PKA) and G. 29,30 The possibility of this direct interaction between GABAA and GABAB receptors has been evaluated in the present work in rat cerebellum granule cells in culture. EXPERIMENTAL PROCEDURES
Cell culture Granule cells were prepared from cerebella of eight-day-old Wistar rats following the procedure of Levi et al., 22 as previously described. 30 Briefly, the minced tissue was first suspended in trypsin (0.25 mg/ml, Type III Sigma) for 15 min at 378C in a shaking water bath, and then in deoxyribonuclease and trypsin inhibitor. Finally, it was dispersed by gently drawing it into a fire-polished Pasteur pipette. Cells were resuspended in basal Eagle’s medium with Earle’s salts supplemented with 10% fetal calf serum (Gibco Bio-Cult, U.K.), 25 mM KCl, 2 mM glutamine and 100 mg/ml gentamicine and plated on poly-l-lysinecoated glass coverslips placed in 20 mm plastic dishes at a density of 1x10 6 per dish and kept at 378C in a humidified 95% air–5% CO2 atmosphere. Experiments were performed between days 5 and 12 after plating. Solutions The standard external solution consisted of (in mM): 135 NaCl, 5.4 KCl, 1.8 CaCl2, 1 MgCl2, 5 HEPES, 10 glucose. The pH was adjusted to 7.4 using NaOH. The pipette filling solution contained (in mM): 142 KCl, 10 HEPES, 2 EGTA, 4 MgCl2, 3 ATP. The pH was adjusted to 7.3 with a Trizma base. Electrophysiology Membrane currents were measured with the standard whole-cell patch-clamp technique by an EPC-7 (List-medical). Patch electrodes were manufactured from borosilicate glass capillaries (Type 1406129 Hilgenberg, Malsfield, Germany) with a programmable Sachs and Flaming puller (model PC-84) with a resistance of 5–10 MV when filled with the internal standard solution. The access resistance was not different among different experimental groups and it did not change significantly during the period of data acquisition. Cell responses were
‡To whom correspondence should be addressed. Abbreviations: EGTA, ethyleneglycolbis(aminoethylether)tetra-acetate; HEPES, N-2-hydroxyethylpiperazine-N 0 -2-ethanesulphonic acid; IPSP, inhibitory postsynaptic potential; PKA, protein kinase A; PTX, pertussis toxin. 1077
B. Barila` et al.
1078
filtered at 3 kHz. Capacitance transient neutralization and series resistance compensantion were optimized. Both stimulation and data acquisition were performed with a Labmaster D/A A/D converter driven by a pCLAMP software (Axon Instruments, Burlingame, CA). Analysis was performed with pCLAMP and SIGMA PLOT (Jandel Scientific, Erkrath, Germany) software. Data are given as mean^S.E.M. (cell number). All experiments were performed at room temperature. Drug application procedure The holding potential was set to 280mV in all experiments reported as this was the most suitable condition for recording the total chloride current. Run-down phenomena of chloride currents were prevented by the presence of ATP in the internal solution. Chloride currents were elicited by 10 26 M muscimol perfusion (flow speed 4–6 ml/min) at 2 min intervals. In order to verify a possible interaction between the two receptors, first of all only GABAB receptor was activated with the specific agonist (2)baclofen and then, in succession, using muscimol agonist, GABAA receptor function was tested. Stock solutions of baclofen (10 mM) and forskolin (50 mM) were diluted at the desired final concentration when added to the culture medium bathing the cells. The drug treatments were performed with cells incubated at 378C for at least 30 min; afterwards the culture medium was substituted with external solution and finally the cell patched was superfused with 10 26M muscimol. In some experiments the culture was treated with pertussis toxin (500 ng/ml) in the culture medium for 16–20 h. All chemicals were purchased from Sigma Chemical Co., St Louis, MO.
RESULTS
Muscimol acting on GABAA receptors of the granule cells in culture activates chloride currents, which under our registration conditions are inward (with Cl 2 going out of the cells). These currents reach a peak current (IP) followed by a development of steady-state level (ISS) over continued application of the agonist (Fig. 1A). The concentration–effect curve for muscimol is reported in Fig. 1B. The experimental points could be fitted by the Hill equation: Y A·
Xn Bn 1 X n
1
where Y is the peak muscimol activated current and X is the muscimol concentration. The parameters A and B are, respectively, the maximum current and the concentration of muscimol corresponding to A/2. The best fit gave A2,320 pA, B1.6 mM and n0.8. In these experiments we have applied drugs, forskolin, (2)baclofen or both and evaluated the muscimol activated Cl 2 current change over time. Chloride current referred to time 0 was the one recorded in the presence of the drugs before washout. Under these conditions, an increase in muscimol-activated current upon washout of the drugs (baclofen, forskolin or both) means that the drug treatment was inhibiting GABAA receptors. When cells treated with 200 mM baclofen added to the culture medium for at least 30 min were tested with 1 mM muscimol immediately after the treatment and onward, the amplitude of the peak of the current activated by muscimol did not significantly change. In fact, the peak changed only by 7^5% (six cells) after 8 min washing with external solution (Fig. 1A, C). However, forskolin (50 mM) treatment resulted in a down-regulation of GABAA receptor activity (Fig. 2A, C). This effect was mainly for the rapidly desensitizing (peak) component. In fact, the ratio ISS/IP became 0.18^0.02 (n6)
Fig. 1. Effect of baclofen on the granule cells chloride current. (A) Currents activated by 1 mM muscimol in one cell incubated with 200 mM baclofen for at least 30 min. Traces were clamped at 280 mV, after 0 (left) and 8 (right) min of washing with external solution. (B) Dose–response curve. Semilogarithmic plot of the peak current amplitude as a function of muscimol concentration. Each cell was voltage clamped at 280 mV. Each point is the average of 3420 cells. Theoretical fitting is obtained using the Hill equation (1). The best fit gave A2,320 pA, B1.6 mM and n0.8. (C) Normalized peak current activated by 1 mM muscimol at 280 mV as a function of the washing time in cells treated as described above (A). The data are the average of at least eight values.
after the washout of forskolin, whereas immediately after the forskolin treatment was 0.28^0.03 (n6).
GABAB receptor modulation
1079
Fig. 3. Dose dependence of the forskolin effect. Each cell was treated with the same dose of forskolin and different concentrations, C, of baclofen. Points represent mean and the number of cells tested is indicated in the brackets. Experimental points can be fitted using equation (2). The best fit gave A0.98, B2.5 mM and n0.4.
In this case, X is the baclofen concentration and Y was computed using the following expression: " !# Iin 12 Ifin " !#bf 3 Iin 12 Ifin f
Fig. 2. Combined effect of baclofen and forskolin. Currents activated by 1 mM muscimol in granule cells clamped at 280 mV, where the whole-cell configuration was reached after 0, 4, 8 min of washing with external standard solution. (A) In one cell preincubated with 50 mM forskolin (B) in one cell preincubated with 200 mM baclofen and 50 mM forskolin for at least 30 min. (C) Normalized peak current activated by 1 mM muscimol as a function of washing time. (W) in cells preincubated with 50 mM forskolin and (X) in cells preincubated with 200 mM baclofen and 50 mM forskolin. The data are the average of at least seven values.
In this case, the effect of 50 mM forskolin was completely inhibited by 200 mM baclofen such that muscimol-activated current remained constant over time of washing (Fig. 2B, C). These data must be compared with those observed in cells treated only with 50 mM forskolin, where an increase by 59^3% (eight cells) was found after 6 min washing with external solution (Fig. 2A, C). Also the change in ISS/IP ratio was prevented by baclofen. Different concentrations of (2)baclofen were used to test its action on inhibition of GABAA receptor activity induced by forskolin. The experimental points could be fitted using the equation: Xn Y A· 1 2 n : 2 B 1 Xn
where the initial peak current (Iin) refers to the current in the presence of the drugs and the final peak current (Ifin) after their washing out; in this expression the numerator refers to cells treated with baclofen and forskolin while the denominator to cells treated only with forskolin. The parameters A and B are, respectively, the maximum effect of forskolin and the concentration of baclofen corresponding to A/2. The best fit gave A0.98, B2.5 mM and n0.4 (Fig. 3). After incubation with pertussis toxin (PTX) for 16 h and then with 200 mM baclofen and 50 mM forskolin for at least 30 min, the chloride current increased by 43^5% (11 cells) after 6 min washing with external solution (Fig. 4A, B). This experiment showed that PTX blocked the effect of baclofen leaving forskolin free to act. In fact, similar results were obtained with cells pretreated with PTX and then only with forskolin. In these conditions, in nine cells there was an increase of 60^6% of muscimol response following a 6-min wash with the external solution (Fig. 4C, D). These observations indicate that PTX per se does not influence the forskolin effect whereas it prevents the baclofen effect by blocking a G protein. DISCUSSION
In a previous report of our own together with another lab, 2 it was shown that GABAB receptor activation by (2)baclofen, their prototypical agonist, 17 prevented the forskolin induced increase in cAMP in rat cerebellum granule cells. The biphasic concentration–response curve for the antagonism by CGP52432 suggested that more than one subtype of GABAB receptors was involved. In the present experiments we went further, evaluating the
1080
B. Barila` et al.
Fig. 4. Effect of PTX. Each cell was incubated with PTX for 12–16 h and then treated at least for 30 min either with 200 mM baclofen and 50 mM forskolin (A, B) or only with 50 mM forskolin (C, D). (A) Currents activated by 1 mM muscimol in one cell after 0, 4, 8 min perfusing with external standard solution. (B) Normalized peak current activated by 1 mM muscimol at 280 mV as a function of washing time. The data are the average of at least seven values. (C) Currents activated in one cell by 1 mM muscimol after 0, 4, 8 min perfusing with external standard solution. (D) Normalized peak current activated by 1 mM muscimol at 280 mV as a function of washing time. The data are the average of at least nine values.
functional consequences of the block of cAMP increases in terms of GABAA receptor activity, which has been shown to be down-regulated by PKA. 30 GABAB receptors have been demonstrated to have several functions. A classical division of their effect is in their negative action on presynaptic Ca 21 channels and their activation of postsynaptic K 1 channels. 3 This last effect results in time delayed late IPSPs in the hippocampus which are not sensitive to block by bicuculline. 1,13,26 These K 1-mediated late IPSPs involve the intermediation of a G protein. 3,19 In addition, GABAB receptors, at postsynaptic sites in the case of cerebellar granule cells interfere with the activity of adenylate cyclase via PTX-sensitive Gi proteins. 37 A well-known apparent paradox is the fact that GABAB receptor activation results in a decrease in forskolin-activated accumulation of cAMP in rat brain slices, whereas it causes a larger increase in cAMP under the activation of b adrenergic receptors. 11,16,18,21 These apparently contradictory results appear to find their explanation in the different pattern of regulation for the various isoforms of adenylate cyclase (I–VIII), which may be present in different cells and subcellular compartments. 25,34 In fact, type II adenylate cyclase is activated by GSa picomolar concentration, which is liberated by b adrenergic receptor stimulation, but can be further activated by the bg complex. 12,33 However, the effect of bg requires higher (nanomolar) concentrations, which can be reached under the concurrent stimulation of Gi proteins via GABAB receptors. 34 Referring to cerebellum, the conditional stimulatory effect on adenylate cyclase, via GABAB receptor activation does not appear to be present. 11 In addition, rat cerebellar granule cells in culture did not show stimulation of cAMP synthesis by isoproterenol, a b-adrenergic agonist. 37 No increase of cAMP was also caused by prostaglandin E1, serotonin and
vasoactive intestinal protein. Thus, in granule cells the conditional activation of adenylate cyclase II by GABAB receptors may not be present. In contrast, we did not find in cerebellar granules cells in culture any change in cAMP levels by (2)baclofen, whereas baclofen could block the forskolininduced increase in cAMP. 2 This last result had already been described by another group. 37 We report here that in granule cells forskolin down-regulates GABAA receptors. This result confirms our previous data 30 that activation of PKA, via cAMP increases, impairs GABAA receptor function. Simultaneous application of (2)baclofen erases the forskolin effect with an EC50 of 2.5 mM. This result is in accord with the prevention of cAMP level increase by GABAB receptor activation in granule cells in culture. 2 Evidently, the PKA activation via cAMP increase affects mainly the rapidly desensitizing component as it was appreciated by the change in the ISS/IP ratio. This effect too was prevented by baclofen. On the other hand, a preferential down-regulation of the rapidly desensitizing component of GABA-activated chloride current by protein kinase G was already shown by us. 29 These results may suggest that GABA or muscimol activate two different GABAA receptor populations differentially sensitive to down-regulation by serine/threonine kinases. The (2)baclofen prevention of forskolin-induced increases in cAMP seems to suggest that this diterpene activates a type I adenylyl cyclase in these cells. This enzyme isoform is in fact inhibited by both GIa and bg. 34 This conclusion is strengthened by the fact that the baclofen effect on GABAA receptors is prevented by PTX. Cerebellar granule cells indeed express mRNA for type I (and type II) adenylate cyclase. 25 From a functional point of view, the protective effect of
GABAB receptor modulation
GABAB on GABAA receptors may play a role under those physiological conditions in which type I adenylate cyclase is activated. One of such conditions, according to the paradigms proposed by Taussig and Gilmann, 34 is an increase of intracellular Ca 21. In fact, type I adenylyl cyclase is one of those directly activated by nanomolar concentrations of Ca 21/ calmodulin. In a previous report, we have shown that Ca 21 influx into granule cells via N-methyl-d-aspartate (NMDA)-activated channels down-regulates GABAA receptor function. 31 Although that effect was accounted for by nitric oxide synthase and calcineurin activation, increases in cAMP via adenylate cyclase activation may represent another pathway
1081
of inhibition of GABAA receptors. Under this circumstance GABAB receptor activity may partially rescue their ionotropic counterparts. However, it should be made clear that GABAB/GABAA receptor interaction modality is most probably different in different neuronal populations. In fact, Xi et al. 36 have reported that in bullfrog dorsal root ganglion cells GABAB receptor activation depresses GABAA receptor function, most probably via cAMP increases. This is not surprising in view of the wide range of possible situations in terms of GABAB receptors responsive G proteins, adenylate cyclase types (see Discussion above) and, of course, of GABAA receptors isoforms. 24,32
REFERENCES
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.
Benardo L. S. (1995) N-methyl-d-aspartate transmission modulates GABAB-mediated inhibition of rat hippocampal pyramidal neurons in vitro. Neuroscience 68, 637–643. Bonanno G., Fassio A., Robello M., Amico C., Marchetti C., Sale R. and Raiteri M. (1996) Pharmacology of GABAB receptors modulating cyclic AMP production and calcium currents in rat granule cells. Pharmac. Rev. Commun. 8, 99–102. Bormann J. (1988) Electrophysiology of GABAA and GABAB receptor subtypes. Trends Neurosci. 11, 112–116. Bowery N. G. (1993) GABAB receptor pharmacology. A. Rev. Pharmac. Toxicol. 33, 109–147. Bowery N. G., Hill D. R., Hudson A. L., Doble A., Middlemiss D. N., Shaw J. and Turnbull M. (1980) (2)Baclofen decreases neurotransmitter release in the mammalian CNS by an action at a novel GABA receptor. Nature 283, 92–94. Bowery N. G., Hudson A. L. and Price G. W. (1987) GABAA and GABAB receptor site distribution in the rat central nervous system. Neuroscience 20, 365–383. Crunelli V., Haby M., Jassik-Gerschenfeld D., Leresche N. and Pirchio M. (1988) Cl 2 and K 1 dependent inhibitory postsynaptic potentials evoked by interneurones of the rat lateral geniculate nucleus. J. Physiol. 399, 153–176. Davies C. H., Davies S. N. and Collingridge G. L. (1990) Paired-pulse depression of monosynaptic GABA mediated inhibitory postsynaptic responses in rat hippocampus. J. Physiol. 424, 513–531. Deisz R. A. and Prince D. A. (1989) Frequency dependent depression of inhibition in guinea pig neocortex in vitro by GABAB receptor feedback on GABA release. J. Physiol. 412, 513–542. Dutar P. and Nicoll R. A. (1988) A physiological role for GABAB receptors in the central nervous system. Nature 332, 156–158. Enna S. J. and Karbon E. W. (1984) GABAB receptors and transmitter-stimulated cAMP accumulation in rat brain. Neuropharmacology 23, 821–822. Federman A. D., Conklin B. R., Schrader K. A., Reed R. R. and Bourne H. R. (1992) Hormonal stimulation of adenylyl cyclase through Gi-protein bg subunits. Nature 356, 159–161. Ga¨hwiler B. H. and Brown D. A. (1985) GABAB-receptor-activated K 1 current in voltage-clamped CA3 pyramidal cells in hippocampal cultures. Proc. natn. Acad. Sci. U.S.A. 82, 1558–1562. Harrison N. L., Lange G. D. and Barker J. L. (1988) (2)Baclofen activates presynaptic GABAB receptors on GABAergic inhibitory neurons from embryonic rat hippocampus. Neurosci. Lett. 85, 105–109. Hasuo H. and Gallagher J. P. (1988) Comparison of antagonism by phaclofen of baclofen induced hyperpolarizations and synaptically mediated late hyperpolarizing potentials recorded intracellularly from rat dorsolateral septal neurons. Neurosci. Lett. 86, 77–81. Hill D. R. (1985) GABAB receptor modulation of adenylate cylase activity in rat brain slices. Br. J. Pharmac. 84, 249–257. Hill D. R. and Bowery N. G. (1981) 3H-baclofen and 3H-GABA bind to bicuculline-insensitive GABAB sites in rat brain. Nature 290, 149–152. Hill D. R. and Dolphin A. C. (1984) Modulation of adenylate cyclase activity by GABAB receptors. Neuropharmacology 23, 829–830. Hille B. (1992) Ionic Channels of Excitable Membranes. pp. 170–201. Sinauer, Sunderland, MA. Kaupmann K., Huggel K., Heid J., Flor P. J., Bischoff S., Mickel S. J., Mc Master G., Angst C., Bittiger H., Froestl W. and Bettler B. (1997) Expression cloning of GABAB receptors uncovers similarity to metabotropic glutamate receptors. Nature 386, 234–246. Knight A. R. and Bowery N. G. (1996) The pharmacology of adenylyl cyclase modulation by GABAB receptors in rat brain slices. Neuropharmacology 35, 703–712. Levi G., Aloisi F., Ciotti M. T. and Gallo V. (1984) Autoradiographic localization and depolarization induced release of amino acids in differentiating granule cell cultures. Brain Res. 290, 77–86. McCormick D. A. (1989) GABA as an inhibitory neurotransmitter in human cerebral cortex. J. Neurophysiol. 62, 1018–1027. McKernan R. M. and Whiting P. J. (1996) Which GABAA receptor subtypes really occur in the brain? Trends Neurosci. 19, 139–143. Mons N. and Cooper D. M. F. (1995) Adenylate cyclases: critical foci in neuronal signaling. Trends Neurosci. 18, 536–542. Newberry N. R. and Nicoll R. A. (1985) Comparison of the action of baclofen with GABA on rat hippocampal pyramidal cells in vitro. J. Physiol. 360, 161–185. Pearce R. A., Grunder S. D. and Faucher L. D. (1995) Different mechanisms for use-dependent depression of two GABAB mediated IPSCs in rat hippocampus. J. Physiol. 484, 425–435. Raiteri M. and Bonanno G. (1993) Multiple GABAB receptors. Trends pharmac. Sci. 14, 259–261. Robello M., Amico C., Bucossi G., Cupello A., Rapallino M. V. and Thellung S. (1996) Nitric oxide and GABAA receptor function in the rat cerebellar cortex and cerebellar granule cells. Neuroscience 74, 99–105. Robello M., Amico C. and Cupello A. (1993) Regulation of GABAA receptors in cerebellar granule cells in culture: differential involvement of kinase activities. Neuroscience 53, 131–138. Robello M., Amico C. and Cupello A. (1997) A dual mechanism for impairment of GABAA receptor activity by NMDA receptor activation in rat cerebellum granule cells. Eur. Biophys. J. 25, 181–187. Stephenson A. (1995) The GABAA receptors. Biochem. J. 310, 1–9. Taussig R., Tang W.-J., Hepler J. R. and Gilman A. G. (1994) Distinct patterns of bidirectional regulation of mammalian adenylyl cyclases. J. biol. Chem. 269, 6093–6100. Taussig R. and Gilman A. G. (1995) Mammalian membrane-bound adenylyl cyclases. J. biol. Chem. 270, 1–4. Thompson S. M. (1994) Modulation of inhibitory synaptic transmission in the hippocampus. Prog. Neurobiol. 42, 575–609.
1082
B. Barila` et al.
36. Xi Z.-X., Yamada K., Tsurusaki M. and Akasu T. (1997) Baclofen reduces GABAA receptor responses in acutely dissociated neurons of bullfrog dorsal root ganglia. Synapse 26, 165–174. 37. Xu J. and Wojcik W. J. (1986) GABAB receptor-mediated inhibition of adenylate cyclase in cultured cerebellar granule cells: blockade by islet-activating protein. J. Pharmac. exp. Ther. 239, 568–573. (Accepted 19 April 1999)