Valerenic Acid Potentiates and Inhibits GABAA Receptors: Molecular Mechanism and Subunit Speciﬁcity

ARTICLE IN PRESS

+ MODEL

Neuropharmacology xx (2007) 1e10 www.elsevier.com/locate/neuropharm

Valerenic acid potentiates and inhibits GABAA receptors: Molecular mechanism and subunit speciﬁcity

S. Khom a, I. Baburin a, E. Timin a, A. Hohaus a, G. Trauner b, B. Kopp b, S. Hering a,*

a Department of Pharmacology and Toxicology, University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria b Department of Pharmacognosy, University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria Received 8 December 2006; received in revised form 11 April 2007; accepted 30 April 2007

Abstract

Valerian is a commonly used herbal medicinal product for the treatment of anxiety and insomnia. Here we report the stimulation of chloride currents through GABAA receptors (IGABA) by valerenic acid (VA), a constituent of Valerian. To analyse the molecular basis of VA action, we expressed GABAA receptors with 13 different subunit compositions in Xenopus oocytes and measured IGABA using the two-microelectrode voltage-clamp technique. We report a subtype-dependent stimulation of IGABA by VA. Only channels incorporating b2 or b3 subunits were stimulated by VA. Replacing b2/3 by b1 drastically reduced the sensitivity of the resulting GABAA channels. The stimulatory effect of VA on a1b2 receptors was substantially reduced by the point mutation b2N265S (known to inhibit loreclezole action). Mutating the corresponding residue of b1 (b1S290N) induced VA sensitivity in a1b1S290N comparable to a1b2 receptors. Modulation of IGABA was not significantly dependent on incorporation of a1, a2, a3 or a5 subunits. VA displayed a significantly lower efficiency on channels incorporating a4 subunits. IGABA modulation by VA was not g subunit dependent and not inhibited by flumazenil (1 mM). VA shifted the GABA concentrationeeffect curve towards lower GABA concentrations and elicited substantial currents through GABAA channels at 30 mM. At higher concentrations (100 mM), VA and acetoxy-VA inhibit IGABA. A possible open channel block mechanism is discussed. In summary, VA was identified as a subunit specific allosteric modulator of GABAA receptors that is likely to interact with the loreclezole binding pocket. Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: GABAA-receptors; 2-Microelectrode voltage-clamp technique; Xenopus oocyte; Subunit speciﬁc modulation; Valerenic acid

1. Introduction (Hevers and Lu¨ddens, 1998; Sieghart, 1995; Boileau et al., 2002). Both binding sites for GABA and for benzodiazepines g-Aminobutyric acid (GABA) mediates fast synaptic inhi- (BZDs) are assumed to be located at subunit interfaces (for re- bition by interaction with the GABA type A (GABAA) recep- view see Galzi and Changeux, 1995; Sigel and Buhr, 1997). tor. GABAA receptors are assembled from individual subunits Mutational studies suggest that the binding pocket for BZDs forming a pentameric structure. Nineteen isoforms of mamma- is located at the interface between a and g subunits, whereas lian GABAA receptor subunits have been cloned: a1e6, b1e3, binding of GABA is believed to occur at the interfaces g1e3, d, 3, p, r1e3 and q (Barnard et al., 1998; Simon et al., between a and b subunits (Sigel, 2002; Ernst et al., 2003; 2004). Ernst et al., 2005). The subunit composition determines the GABA sensitivity GABAA channels are modulated by numerous structurally and the pharmacological properties of the GABAA receptor distinct substances including clinically important drugs as BZDs, barbiturates and various general anaesthetics (see Sieghart, 1995 for review), but also by several compounds * Corresponding author. Tel.: þ43 1 4277 55301; fax: þ43 1 4277 9553. of plant origin, including ﬂavonoids, e.g. methyl-apigenin E-mail address: [email protected] (S. Hering). (Marder et al., 2003) or wogonin (from Scutellaria baicalensis,

Please cite this article in press as: Khom, S. et al., Valerenic acid potentiates and inhibits GABAA receptors: Molecular mechanism and subunit speciﬁcity, Neuropharmacology (2007), doi:10.1016/j.neuropharm.2007.04.018 ARTICLE IN PRESS

+ MODEL 2 S. Khom et al. / Neuropharmacology xx (2007) 1e10

Hui et al., 2002; Huen et al., 2003), monoterpenes, e.g. bor- Microelectrodes were ﬁlled with 2 M KCl and had resistances between 1 neol (Granger et al., 2005) and thymol (Garcia et al., 2006) and 3 MU. or polyacetylenes, e.g. MS-1, MS-2 and MS-4 (from Cussonia zimmermannii, Baur et al., 2005). In the present study, we an- 2.3. Perfusion system alysed the molecular mechanism of action of VA and the two structurally closely related compounds acetoxy-VA and hy- GABA and VAwere applied by means of a newly developed fast perfusion droxy-VA from Valerian root on GABA channels. Previous system (Baburin et al., 2006; see also Khom et al., 2006). Drug or control A solutions were applied by means of a TECAN Miniprep 60 permitting automa- in vitro studies on a neonatal rat brainstem preparation suggest tion of the experiments. To elicit IGABA the chamber was perfused with 120 ml that the Valerian effects may be mediated through VA action of GABA-containing solution at volume rate between 300 and 1000 ml/s. The on GABAA receptors (Yuan et al., 2004). The molecular mech- IGABA rise time ranged between 100 and 250 ms (see Khom et al., 2006). Care anism of action of VA remained, however, unknown. In order was taken to account for possible slow recovery from increasing levels of de- to clarify the molecular basis of VA action on GABA recep- sensitization in the presence of high GABA or VA concentrations. The dura- A tion of washout periods was therefore extended from 1.5 min (1e20 mM tors, we expressed 13 different subunit combinations in GABA, <10 mM VA) to 30 min (100 mM GABA, 10 mM VA), respectively. Xenopus oocytes and analysed the modulation of the corre- Oocytes with maximal current amplitudes >3 mA were discarded to exclude sponding receptors by VA. voltage-clamp errors.

2. Methods 2.4. Analysing concentration e response curves

2.1. Chemicals Stimulation of chloride currents by modulators of the GABAA receptor was measured at a GABA concentration eliciting between 5 and 10% of the

Valerenic acid was obtained from Extrasynthese, France, Lyon, acetoxy- maximal current amplitude (EC5e10). The EC5e10 was determined at the be- valerenic acid from LGC Promochem, Wesel, Germany and hydroxy-valerenic ginning of each experiment. acid was isolated from Valerian root. The structural formulae are given in Enhancement of the chloride current was defined as (I(GABAþComp)/ Fig. 5. Stock solutions (100 mM) were prepared in DMSO (dimethyl IGABA) 1, where I(GABAþComp) is the current response in the presence of sulfoxide, Sigma, Austria). Because of low solubility in ND96, VA and a given compound and IGABA is the control GABA current. To measure the two derivatives (Fig. 5) were only used up to a concentration of the sensitivity of the GABAA receptor for a given compound, it was applied 300 mM. Equal amounts of DMSO were present in control and VA-containing for an equilibration period of 1 min before applying GABA (EC5e10). Con- solutions. The maximum DMSO concentration in the bath (0.3%) did not centrationeresponse curves were generated and the data were fitted by non- affect IGABA. linear regression analysis using Origin Software (OriginLab Corporation, nH USA). Data were fitted to the equation 1=ð1 þðEC50=½CompÞ Þ, where nH is the Hill coefficient. Each data point represents the mean S.E. 2.2. Expression and functional characterization of GABAA receptors from at least four oocytes and 2 oocyte batches. Statistical significance was calculated using paired Student t-test with a confidence interval of p < 0.05. Preparation of stage VeVI oocytes from Xenopus laevis, synthesis of cap- ped off run-off poly(Aþ) cRNA transcripts from linearized cDNA templates (pCMV vector) was performed as previously described (Khom et al., 2006). 3. Results Briefly, female Xenopus laevis (NASCO, USA) were anaesthetised by expos- ing them for 15 min to a 0.2% solution of MS-222 (methane sulfonate salt of 3-aminobenzoic acid ethyl ester; Sandoz) before surgically removing parts of 3.1. Potentiation of IGABA by VA through a1b2, a1b2g1 the ovaries. Follicle membranes from isolated oocytes were enzymatically and a1b2g2S channels digested with 2 mg/ml collagenase (Type 1 A, Sigma). One day after isolation, e the oocytes were injected with about 10 50 nl of DEPC-treated water (diethyl Functional effects of VA were investigated on recombinant pyrocarbonate, Sigma, Germany) containing the different cRNAs at a concentration of approximately 300e3000 pg/nl/subunit. The amount of cRNA was GABAA receptors expressed in Xenopus laevis oocytes. Mod- determined by means of a NanoDrop ND-1000 (Kisker-Biotech, Steinfurt, ulation of IGABA by VA was first studied on GABAA channels Germany). composed of either a1b2 or a1b2g1/2S subunits. As shown in Mutation b1S290N was introduced by the ‘‘gene SOEing’’ technique Fig. 1A, B, VA exhibited a positive allosteric modulatory ef- (Horton et al., 1989). This involved synthesizing mutagenic oligonucleotides fect at concentrations 1 mM by enhancing IGABA. The effect to introduce the desired mutation, and a silent restriction site was used to e screen for the mutation. The mutant cDNA was verified by sequencing. was dose-dependent and the averaged concentration response

To ensure expression of the g subunit in the case of a1b2g1, a1b2g2S, curve shows that maximum stimulation of a1b2g2S receptors a2b2g2S and a2b2g1 receptors, cRNAs were mixed in a ratio of 1:1:10 except occurred at w100 mM(Fig. 1B). The maximal potentiation a b g (ratio 3:1:5). For receptors comprising only a and b subunits (a b , 4 2 2S 1 2 of IGABA (efficiency) through a1b2g1 (235.6 46.4%, n ¼ 8) a2b2, a1b3, a1b2N265S (cDNA gift of E. Sigel), a2b2, a3b2 and a5b2), the and a1b2g2S channels (400.0 77.6%, n ¼ 12) was, however, cRNAs were mixed in a ratio 1:1. cRNAs for of a1b1 and a1b1S290N channels not significantly higher than in channels composed of a1 and were injected in a ratio 3:1 to avoid formation of b1 homooligomeric GABAA receptors (Boileau et al., 2002; Krishek et al, 1996). Oocytes were stored b2 subunits (245.7 53.3%, n ¼ 17) (see Fig. 1A, B, Tables at 18 C in ND96 solution (Methfessel et al., 1986). Electrophysiological 1 and 2). At concentrations 30 mM we observed a direct ac- experiments were done using the two-microelectrode voltage-clamp method tivation of GABAA channels by VA. This finding is illustrated at a holding potential of 70 mV making use of a TURBO TEC 01C in the inset of Fig. 1A (see Fig. 6 for detailed analysis of chan- amplifier (npi electronic, Tamm, Germany) and an Axon Digidata 1322A interface (Molecular Devices, Sunnyvale, CA). Data acquisition was carried nel activation by VA). At concentrations 100 mM current en- out by using pCLAMP v.9.2. The bath solution contained 90 mM NaCl, hancement by VA was less pronounced or even inhibition of $ 1 mM KCl, 1 mM MgCl2 6H2O, 1 mM CaCl2 and 5 mM HEPES (pH 7.4). control IGABA was evident.

+ MODEL S. Khom et al. / Neuropharmacology xx (2007) 1e10 3

Fig. 1. (A) Typical IGABA recordings illustrating concentration-dependent modulation by VA of GABA elicited chloride currents through a1b2g2S-containing receptors. Direct activation of GABAA receptors during preincubation with 30 mM VA is shown in an inset. Concentrationeeffect curves for action of VA on (B): a1b2 (C), a1b2g1 (:) and a1b2g2S (-); (C): a2b2 (C), a2b2g1 (:) and a2b2g2S (-); and (D): a3b2 (C), a4b2g2S (:) and a5b2 (-) receptors using a GABA EC5e10 (EC50 and nH values are given in Table 1). Data points represent means S.E. from at least four oocytes from 2 batches. IGABA at 100 and 300 mM(a1b2) and at 300 mM(a1b2g2S and a1b2g1) (grey symbols), IGABA at 100 and 300 mM(a2b2) and at 300 mM(a2b2g2S) (grey symbols) and IGABA at 100 mM and 300 mM (a3b2, a4b2g2S and a5b2) (grey symbols) were excluded from the ﬁt.

3.2. Potentiation of IGABA by VA through a2b2, a3b2, VA concentrations (see Fig. 8). Maximal stimulation of IGABA a4b2g2S, a5b2, a2b2g1 and a2b2g2S channels through GABAA receptors composed of a3b2 and a5b2 subunits was comparable to a2b2 or a1b2 receptors (see Table 1 A possible a subunit specificity was analysed by substitut- for efficiencies and EC50 values, compare Fig. 1BeD and ing the a1 by a2, a3, a4 and a5 subunits and subsequent anal- Table 2). Interestingly, a4b2g2S receptors displayed a signifi- ysis of the modulation of IGABA through the corresponding cantly lower sensitivity for VA with a maximal enhancement GABAA channels. A comparison of Fig. 1B, C reveals of IGABA of 68.9 14.1%. a very similar modulation of GABAA receptors incorporating a2 instead of a1 subunits. As illustrated in Fig. 1B, C we 3.3. Modulation of IGABA by VA at different GABA observed a slightly higher efficiency of VA in g2S-containing concentrations receptors (see Table 1 for fitted efficiencies and EC50s). A reliable estimation of the EC50 values and maximal IGABA In order to gain insight into the mechanism of IGABA mod- stimulation was complicated by an apparent IGABA inhibition ulation by VA, we compared GABA concentrationeresponse at concentrations 100 mM that was consistently observed curves in the presence and absence of VA on a1b3 channels. for all subunit compositions (except for a2b2g1 receptors). VAwas applied in a concentration of 30 mM to minimise a po- We will show later that the bell-shaped concentrationeeffect tential inhibition of IGABA (see Fig. 1BeD). At this concentra- curves may reflect a low-affinity open channel block at high tion VA shifted the concentrationeeffect curve towards lower

+ MODEL 4 S. Khom et al. / Neuropharmacology xx (2007) 1e10

Table 1 also suggest that the binding sites of VA and the BZD like Potency and efﬁciency of Valerenic acid for GABAA-receptors of different diazepam are independent of each other (Fig. 3C, D). subunit compositions

Subunit EC50 Maximum Hill Number of combination (mM) stimulation coefﬁcient experiments 3.5. b subunit dependence of GABAA receptor

of IGABA (nH) (n) potentiation by VA (EC5e10) (%) a1b2 5.2 2.4 245.7 53.3 1.5 0.4 17 We observed a strong dependence of the modulatory effect a1b3 16.6 3.8 487.3 78.9 1.7 0.3 12 of VA on the b subunit composition of the receptor. Fig. 4 il- a1b2g1 18.0 7.1 235.6 46.4 1.3 0.2 8 lustrates the effect of VA on IGABA through GABAA channels a b g 13.6 4.1 400.0 77.6 1.5 0.3 12 1 2 2S with different b subunit isoforms (channels without g sub- a2b2 7.2 2.6 187.4 36.8 1.8 0.6 10 a2b2g1 16.8 6.4 280.5 36.1 1.0 0.2 9 units). The figure illustrates that only channels incorporating a2b2g2S 11.3 1.6 421.4 26.3 1.5 0.1 8 b2 or b3 subunits were stimulated by VA at concentrations a3b2 13.2 10.5 188.5 90.4 1.4 0.5 5 between 1 and 30 mM. The highest efficiency of VA was ob- a5b2 7.8 6.1 174.2 57.0 1.7 1.1 11 served for a1b3 channels. Replacing b2/3 by b1 drastically re- a4b2g2S 2.5 2.0 68.9 14.1 1.1 0.4 5 duced the sensitivity of the resulting GABAA channels. As a1b1S290N 10.4 3.8 371.3 69.6 1.5 0.3 5 shown in Fig. 4A, the stimulation of IGABA through a1b1 receptors was almost absent. Interestingly, the point mutation GABA concentrations and also slightly increased the maximal b2N265S (known to reduce stimulatory effects by loreclezole, GABA response (Fig. 2). Wingrove et al., 1994) almost completely abolished the stimulatory effect of VA. The corresponding mutation in b1 (b1S290N) induced sensitivity for VA. These findings suggest 3.4. VA does not interact with the BZD site an involvement of this residue in either binding of VA or trans- duction of the effect. Corresponding representative IGABA in Our data demonstrate that VA has a strong stimulatory effect absence and presence of 10 mM VA are shown in Fig. 4B. on GABA channels not containing g subunits. A trend to- The structures of VA, acetoxy-VA, hydroxy-VA and lorecle- A zole are illustrated in Fig. 5. wards a slightly higher efficiency of VA on a1/2b2 receptors containing a g2S subunit (Fig. 1BeD) prompted us, however, to investigate a possible interaction of VA with the BZD bind- 3.6. Direct activation of GABAA receptors by VA ing pocket. To do this, we stimulated IGABA through a1b2g2 receptors with 5 mM VA either in the absence or presence of As shown in Fig. 1A, VA activated currents through GA 1 mM of the BZD antagonist flumazenil. Flumazenil did not af- BAA receptors (IVA). Fig. 6A illustrates VA-evoked currents fect the IGABA potentiation significantly (77.4 2.3%, n ¼ 3 through channels composed of a1 and b2 subunits. A compar- control vs. 77.3 8.6%, n ¼ 3) (Fig. 3A, B). Moreover, addi- ison with the kinetics of an IGABA at low GABA concentra- tive effects of VA and 300 nM diazepam on a1b2g2S receptors tions revealed a significantly slower activation of IVA

Table 2

Comparison of efﬁciencies (upper-right) and potencies (lower-left) for GABAA receptors of different subunit compositions

Imax α β α β α β γ α β γ α β α β γ α β γ α β α β α β γ α β 1 2 1 3 1 2 1 1 2 2S 2 2 2 2 1 2 2 2S 3 2 5 2 4 2 2S 1 1S290Ν EC50

α1β2 ***

α1β3 * ******

α1β2γ1 ** α β γ 1 2 2S ***

α2β2 * ***

α2β2γ1 ** α β γ 2 2 2S * ** *

α3β2

α5β2 * α β γ 4 2 2S ** * * * * α β 1 1S290Ν (*) Indicates statistically significant (p < 0.05) differences.

+ MODEL S. Khom et al. / Neuropharmacology xx (2007) 1e10 5

(Fig. 6A). Compared to a mean current rise time (t10e90%)of IGABA at EC10 (180 45 ms, n ¼ 7) we observed a significantly slower activation of IVA with mean t10e90% of 680 50 ms (n ¼ 5) at VA concentrations between 30 and 300 mM. Interestingly, not only the activation of GABAA channels but also the deactivation rate of IVA were remarkably slower compared to IGABA. The mean time constants of IVA deactivation of 2083 232 ms (n ¼ 5at30e300 mM VA) were about 10 times larger compared to IGABA decay upon washout (200 10 ms, GABA EC10, n ¼ 7, Fig. 6A). With higher VA concentrations, a ‘‘wash-off’’ current (WOC) was observed during rapid perfusion of the oocytes with control solution (see Fig. 6A, C). This finding suggests rapid unbinding of VA from a (low-affinity) binding site located in the open channel pore (compare with Thompson et al., 1996; Akk and Steinbach, 2000; Feng et al., 2004). The amplitude of the ‘‘wash-off’’ current increased with increasing VA concentration from 30 to 300 mM(Fig. 6). Fig. 2. GABA concentrationeeffect curve of a1b3 GABAA receptors (control, -) and in the presence of 30 mMVA(C). Corresponding EC50-values were 5.3 0.6 mM(nH ¼ 1.1 0.1, n ¼ 4) for control and 1.7 mM 0.7 mM(nH ¼ 3.7. Evidence for open channel block by VA and 0.7 0.2, n ¼ 4) in the presence of VA, respectively. The increase in the max- acetoxy-VA imal IGABA may result from a direct activation of GABAA receptors (see Fig. 6). Data points represent means S.E. from at least four oocytes from Our data (grey symbols in Figs. 1AeC, 3A) an open chan- 2 batches. nel block mechanism for VA. This hypothesis is supported by the observed ‘‘wash-off’’ currents that increased at high VA concentrations (Fig. 6). Interestingly, acetoxy-VA induced no

Fig. 3. (A) Stimulation of IGABA by 5 mM VA is not inhibited by flumazenil. The left bar shows the positive allosteric modulation of the GABA (EC5e10)-induced chloride current by 5 mM VA through a1b2g2S receptors. The right bar illustrates that flumazenil (1 mM) does not antagonize VA induced enhancement of IGABA. (B) Typical IGABA through a1b2g2S receptors in the absence and presence of the indicated concentrations of VA or VA and flumazenil, respectively. (C) Additive effects of VA 10 mM and diazepam 300 nM. The left bar illustrates the enhancement of IGABA by 10 mM VA, the bar in the middle by 300 nM diazepam, whereas the right bar illustrates the enhancement of IGABA under coapplication of VA 10 mM and diazepam 300 nM through receptors composed of a1, b2 and g2S subunits. (D) Representative currents through a1b2g2S channels for the additive effects of VA and diazepam.

+ MODEL 6 S. Khom et al. / Neuropharmacology xx (2007) 1e10

Fig. 4. (A) Concentration-dependent effect for VA on a1b1, a1b3, a1b2, a1b2N265S and a1b1S290N receptors using a GABA EC5e10. (B) Typical traces for modulation of chloride currents through a1b1, a1b3, a1b2, a1b2N265S and a1b1S290N channels by 10 mM valerenic acid at GABA EC5e10. Control currents (GABA, single bar) and corresponding currents elicited by coapplication of GABA and the indicated valerenic acid concentration (double bar) are shown. stimulation at concentrations up to 100 mM (and did not evoke currents when applied at this concentration alone, data not Fig. 6. (A) Representative currents illustrating direct activation of GABAA shown) but accelerated the current decay through a1b3 chan- receptors (a b ) by VA at the indicated concentrations in comparison to nels (Fig. 7A). We made use of these properties (no stimula- 1 2 a GABA-induced current at EC10. (B) Inhibition of IVA (300 mM) by bicucul- tion but acceleration of current decay) to study inhibition of line. The left bar illustrates the percentage of the VA induced current (300 mM)

in relation to the maximal IGABA at 1 mM. The right bar shows the effect of VA in the presence of 5 mM bicuculline. Asterisk indicates statistically signiﬁcant differences from zero ( p < 0.05). (C) Estimation of the relative ‘‘wash-off’’

current (WOC) in percentage of corresponding IVA at the indicated VA concentrations.

IGABA in more detail. The shape of IGABA in the presence of 100 and 300 mM acetoxy-VA revealed a concentration- dependent acceleration of the current decay (Fig. 7B) which would agree with an open channel block. Interestingly, only acetoxy- but not hydroxy-VA accelerated the IGABA decay (data not shown). Plotting the reciprocal of the time constants of the acetoxy-VA induced current decay against concentration enabled the estimation of an apparent IC50 value (w190 mM) for an apparent open channel block mechanism. A simulation of the bell-shaped concentrationeresponse curve of VA action on a1b3 channels assuming a low-afﬁnity open channel Fig. 5. (A) Structures of VA and the structurally related compounds hydroxy- block mechanism is shown in Supplementary Materials and acetoxy-VA. (B) For comparison the structure of loreclezole. (Fig. 8, details are in Supplementary Materials).

+ MODEL S. Khom et al. / Neuropharmacology xx (2007) 1e10 7

4. Discussion

In the present study we analysed the modulation of GABAA receptors by VA, acetoxy-VA and hydroxy-VA (all constituents of Valerian root). A principal ﬁnding of our study is that only VA (but neither acetoxy-VA nor hydroxy-VA, Figs. 1 and 5) acts as a positive allosteric modulator of GABAA receptors. Half-maximal stimulation occurred at concentrations between 2.5 2.1 and 18.0 7.1 mM depending on the subunit composition (see Table 1). The threshold of stimulation was around 1 mM and a maximal enhancement of IGABA was observed for concentrations between 30 and 100 mM(Fig. 1BeD). In line with data reported for other allosteric modulators, at high concentrations (30 mM) VA directly activated GABAA receptors (Fig. 6). Furthermore, our data suggest that VA blocks open GABAA channels (Figs. 7 and 8).

4.1. Subunit speciﬁcity of IGABA stimulation by VA

In order to obtain insight into the molecular basis of the VA action, we analysed the stimulation of GABAA channels composed of different a, b and g subunits. As illustrated in Fig. 1, coexpression of g2S reveals a trend towards enhanced IGABA stimulation (Fig. 1B, C). This effect was, however, only significant for a2b2g2S receptors (Table 2). Omitting the g1 subunit did not significantly affect stimulation of either a1b2 or a2b2 receptors (Tables 1 and 2). The stimulatory action of VA was not significantly affected by coexpression a2, a3 or a5 instead of a1. A lower efficiency of VA was observed for a4 incorporating receptors (a4b2g2S) (maximal enhancement of IGABA 68.9 14.1%), significantly different from all of the subunit compositions investigated except a3b2 and a5b2. This finding might help to identify determinants of VA efficiency in future studies (Fig. 1D, Tables 1 and 2). In contrast, variation of the b subunits strongly influenced the effect on IGABA stimulation by VA (Fig. 4). The highest efficiency was observed for a1b3 followed by a1b2 channels (Table 1). Replacing b2 by b1 almost completely abolished the stimulatory effect of VA. These data suggest a preferential action of VA on receptors containing b2 or b3 subunits. This finding was confirmed by the lack of VA action on a1b2N265S channels. Point mutation N265S in b2 abolished the modulatory action of VA (Fig. 4). Replacing the serine in the b1 subunit (b ) by the corresponding asparagine of b induced Fig. 7. (A) Concentration-dependent effects for acetoxy-VA (A), hydroxy-VA 1S290 2 VA sensitivity. This finding suggests that I modulation (-) on the enhancement of IGABA (EC5e10) through a1b3 channels. Broken GABA line illustrates the enhancement of IGABA by VA (taken from Fig. 4A). (B) may be mediated by VA interaction with the loreclezole site, Representative traces illustrating GABA-induced chloride currents through or alternatively, reflects an effect of b2N265S on the transduc- a1b3 receptors in the absence (control) and the presence of 30 mM, 100 mM tion pathway of VA action. Similar b subunit specificity of lor- and 300 mM acetoxy-VA. In the presence of 30, 100 and 300 mM acetoxy- eclezole action displaying strong stimulation of receptors VA I decayed with time constants of 10.9 0.4 s, 7.4 0.5 s and GABA containing either a b or a b subunit supports this notion 5.0 0.2 s. (C) Reciprocals of time constants of acetoxy-VA induced IGABA 2 3 decay plotted as function of the acetoxy-VA concentration. The regression (Wafford et al., 1994). line yields a y-intercept (rate constant of dissociation from the open channel, An interaction of VA with N289 at the carboxyl terminal 1 k ¼ 0.08 s ) and a slope (rate constant of association with the open channel, 1 1 end of the pore-forming M2 transmembrane domain could kþ ¼ 420 s M ) suggesting an IC50 for open channel block of w190 mM. explain enhancement of IGABA by lower VA concentrations (e.g. by a destabilizing the closed channel conformation) and an apparent ‘‘open channel’’ inhibition at higher concentrations.

+ MODEL 8 S. Khom et al. / Neuropharmacology xx (2007) 1e10

Fig. 8. (A) Simulated shift of the concentrationeresponse dependency making use of the experimental data (EC50s and nH taken from the concentrationeeffect curves shown in Fig. 2). The broken and straight lines illustrate the curves in the presence and absence of 30 mM VA, respectively. Direct activation of GABAA receptors by VA was not taken into account. (B) Simulation of concentration-dependent enhancement and inhibition of IGABA by VA (in %). Solid line represents a simulation of the positive allosteric modulation by VA (EC50 ¼ 15 mM, nH ¼ 1.8) and open channel block (IC50 ¼ 190 mM, nH ¼ 1.8, see Supplementary Mate- rials for model description). Broken line illustrates the simulated concentrationeeffect curve without open channel block. The approximated IC50 for apparent open channel block of GABAA channels corresponds to the estimated value of acetoxy-VA (Fig. 7C).

4.2. Slow activation and deactivation of IVA 4.3. Evidence for open channel block

At high concentrations, VA directly activated GABAA re- At high concentrations (30 mM) a ‘‘wash-off’’ current oc- ceptors. A direct activation of GABAA receptors was previ- curred upon washout of VA. A straightforward interpretation ously reported for a number of modulators including of this finding is that VA rapidly unbinds from its binding pentobarbital, etomidate, propofol or loreclezole (Schulz and site within the open channel (see also Rho et al., 1996; Macdonald, 1981; Thompson et al., 1996; Feng et al., 2004; Thompson et al., 1996; Dalziel et al., 1999; Akk and Stein- Lam and Reynolds, 1998; Sanna et al., 1996; Hong and Wang, bach, 2000; Krampfl et al., 2002; Feng et al., 2004). Such 2005). This VA induced current was relatively small and com- a scenario is supported by the concentration-dependent in- parable for a1b3 (6.2 2.0%) and a1b2 (10.7 1.5%, stimula- crease of this current suggesting a concentration-dependent tion of maximal IGABA) receptors. open channel block (Fig. 6). Further evidence for an open IVA developed much slower than IGABA and IVA was non- channel block mechanism comes from the accelerated IGABA desensitizing (Fig. 6). Furthermore, the current deactivation decay in the presence of high concentrations of the structurally time constants (tdeact.)weremuchslowerinIVA compared to related acetoxy-VA (Fig. 7B). In order to further explore this IGABA (tdeact(VA, 30e300 mM) ¼ 2083 232 ms, n ¼ 5). The possibility, we simulated the bell-shaped concentrationere- time of IVA deactivation thus significantly exceeds the character- sponse curve that was typically seen for VA action. The corre- istic time of solution exchange whereas the time required for sponding IC50 (w190 mM) value that was estimated assuming washout of 1 mM GABA (tdeact ¼ 200 10 ms, n ¼ 7) corre- that the accelerated IGABA decay reflects open channel block sponds to the time of solution exchange (Baburin et al., 2006). (Fig. 7C) nicely agrees with the simulation of the bell-shaped Such a slow deactivation usually accompanies enhanced desensi- concentrationeresponse curve assuming a low affinity open tization (e.g. Jones and Westbrook, 1995; Haas and Macdonald, channel block mechanism for VA (simulated IC50 ¼ 190 mM, 1999). This was obviously not the case for IVA (Fig. 6A). The see Supplementary Materials). These findings suggest that mechanism underlying slower kinetics of IVA is currently not VA and acetoxy-VA might act at the same site in the channel clear and warrants future studies. pore. The discussed open channel block mechanism remains, Direct activation of GABAA channels by VA may, however, however, hypothetical, because acetoxy-VA may also interact also be explained by an interaction of VA with the GABA with a different binding site than VA. binding site as IVA were efficiently inhibited by bicuculline Surprisingly, an enhanced IGABA decay was observed only (1 mM). We cannot exclude that conformational changes in- for acetoxy-VA and not for the structurally related hydroxy- duced by bicuculline indirectly affect VA action (see also VA. This finding suggests that substitution of the hydrogen Ueno et al., 1997). in position 1 of the hexahydro-indene ring by a hydroxyl- or Our data show that VA is an agonist with much lower acetoxy-group defines not only the modulatory action but efficiency than GABA. The maximal current induced by the also the affinity for open GABAA channels. highest VA concentration of 300 mM (higher concentrations The complex mechanism of VA action (including modula- were not testable due to limited solubility) did not exceed tion of IGABA, direct activation of GABAA channels and open 15% of the current induced by saturating (1 mM) GABA channel block) complicates a straightforward estimation of the concentrations. efficiencies and potencies of VA for the different subunit

+ MODEL S. Khom et al. / Neuropharmacology xx (2007) 1e10 9 compositions. This is illustrated in Figs. 1BeD and 4 where Ernst, M., Bruckner, S., Boresch, S., Sieghart, W., 2005. Comparative the modulatory action is apparently superimposed by inhibi- models of GABAA receptor extracellular and transmembrane domains: tory action. important insights in pharmacology and function. Mol. Pharmacol. 68, 1291e1300. Taken together we have described a subunit specific mod- Feng, H.J., Bianchi, M.T., Macdonald, R.L., 2004. Pentobarbital differentially ulation of GABAA receptors by VA (b3>b2>>b1-containing modulates alpha1beta3delta and alpha1beta3gamma2L GABAA receptor receptors, Fig. 4). Positive allosteric modulation is caused currents. Mol. Pharmacol. 66, 988e1003. by a VA induced increase in the GABA sensitivity (Fig. 2). Galzi, J.L., Changeux, J.P., 1995. Neuronal nicotinic receptors: molecular e The threshold concentration (1 mM) of this modulatory action organization and regulations. Neuropharmacology 34, 563 582. Garcia, D.A., Bujons, J., Vale, C., Sunol, C., 2006. Allosteric positive interac- is in the range of the estimated plasma concentration of VA tion of thymol with the GABAA receptor in primary cultures of mouse cor- (Anderson et al., 2005). At high concentrations, VA activates tical neurons. Neuropharmacology 50, 25e35. GABAA channels directly (Fig. 6) and also blocks the chan- Granger, R.E., Campbell, E.L., Johnston, G.A., 2005. (þ)- And ()-borneol: nel. The clinical evidence for Valerian effects is, however, efficacious positive modulators of GABA action at human recombinant still controversial (see Sampson, 2005 and De Smet, 2002 alpha1beta2gamma2L GABA(A) receptors. Biochem. Pharmacol. 69, 1101e1111. for review). Our study opens the perspective that the proposed Haas, K.F., Macdonald, R.L., 1999. GABAA receptor subunit gamma2 and delta sedative, hypnotic and anxiolytic effects suggested for Vale- subtypes confer unique kinetic properties on recombinant GABAA receptor e rian may be caused by interaction of VA with GABAA currents in mouse fibroblasts. J. Physiol. 514 (Pt 1), 27 45. channels. Hevers, W., Lu¨ddens, H., 1998. The diversity of GABAA receptors. Pharma- cological and electrophysiological properties of GABAA channel subtypes. Mol. Neurobiol. 18, 35e86. Acknowledgements Hong, Z., Wang, D.S., 2005. Potentiation, activation and blockade of GABAA receptors by etomidate in the rat sacral dorsal commissural neurons. Neu- We wish to thank Prof. W. Sieghart for helpful comments roscience 132, 1045e1053. on the manuscript, Dr. B. Klier, Vestenbergsgreuth, Germany Horton, R.M., Hunt, H.D., Ho, S.N., Pullen, J.K., Pease, L.R., 1989. Engineer- ing hybrid genes without the use of restriction enzymes: gene splicing by for samples of acetoxy-valerenic acid and hydroxy-valerenic overlap extension. Gene 77, 61e68. acid and Prof. Dr. A. Brattstro¨m (Zeller AG) for initiating Huen, M.S., Hui, K.M., Leung, J.W., Sigel, E., Baur, R., Wong, J.T., the research project on Valerian. Xue, H., 2003. Naturally occurring 20-hydroxyl-substituted flavonoids as high-affinity benzodiazepine site ligands. Biochem. Pharmacol. 66, 2397e2407. Hui, K.M., Huen, M.S., Wang, H.Y., Zheng, H., Sigel, E., Baur, R., Ren, H., Appendix A. Supplementary data Li, Z.W., Wong, J.T., Xue, H., 2002. Anxiolytic effect of wogonin, a benzodiazepine receptor ligand isolated from Scutellaria baicalensis Georgi. Supplementary data associated with this article can be found, Biochem. Pharmacol. 64, 1415e1424. in the online version, at doi:10.1016/j.neuropharm.2007.04.018. Jones, M.V., Westbrook, G.L., 1995. Desensitized states prolong GABAA channel responses to brief agonist pulses. Neuron 15, 181e191. Khom, S., Baburin, I., Timin, E.N., Hohaus, A., Sieghart, W., Hering, S., 2006. References Pharmacological properties of GABAA receptors containing gamma1 subunits. Mol. Pharmacol. 69, 640e649. Akk, G., Steinbach, J.H., 2000. Activation and block of recombinant GA- Krampfl, K., Wolfes, H., Dengler, R., Bufler, J., 2002. Kinetic analysis of the BA(A) receptors by pentobarbitone: a single-channel study. Br. J. Pharma- agonistic and blocking properties of pentobarbital on recombinant rat al- col. 130, 249e258. pha(1)beta(2)gamma(2S) GABA(A) receptor channels. Eur. J. Pharmacol. Anderson, G.D., Elmer, G.W., Kantor, E.D., Templeton, I.E., Vitiello, M.V., 435, 1e8. 2005. Pharmacokinetics of valerenic acid after administration of valerian Krishek, B.J., Moss, S.J., Smart, T.G., 1996. Homomeric beta 1 gamma-ami- in healthy subjects. Phytother. Res. 19, 801e803. nobutyric acid A receptor-ion channels: evaluation of pharmacological and Baburin, I., Beyl, S., Hering, S., 2006. Automated fast perfusion of Xenopus physiological properties. Mol. Pharmacol. 49, 494e504. oocytes for drug screening. Pflugers Arch. 453, 117e123. Lam, D.W., Reynolds, J.N., 1998. Modulatory and direct effects of propofol on Barnard, E.A., Skolnick, P., Olsen, R.W., Mohler, H., Sieghart, W., Biggio, G., recombinant GABAA receptors expressed in Xenopus oocytes: influence of Braestrup, C., Bateson, A.N., Langer, S.Z., 1998. International Union of alpha- and gamma2-subunits. Brain Res. 784, 179e187. Pharmacology. XV. Subtypes of gamma-aminobutyric acidA receptors: Marder, M., Viola, H., Wasowski, C., Fernandez, S., Medina, J.H., classification on the basis of subunit structure and receptor function. Phar- Paladini, A.C., 2003. 6-Methylapigenin and hesperidin: new valeriana macol. Rev. 50, 291e313. flavonoids with activity on the CNS. Pharmacol. Biochem. Behav. 75, Baur, R., Simmen, U., Senn, M., Sequin, U., Sigel, E., 2005. Novel plant sub- 537e545. stances acting as beta subunit isoform-selective positive allosteric modula- Methfessel, C., Witzemann, V., Takahashi, T., Mishina, M., Numa, S., tors of GABAA receptors. Mol. Pharmacol. 68, 787e792. Sakmann, B., 1986. Patch clamp measurements on Xenopus laevis oocytes: Boileau, A.J., Baur, R., Sharkey, L.M., Sigel, E., Czajkowski, C., 2002. The currents through endogenous channels and implanted acetylcholine recep- relative amount of cRNA coding for gamma2 subunits affects stimulation tor and sodium channels. Pflugers Arch. 407, 577e588. by benzodiazepines in GABA(A) receptors expressed in Xenopus oocytes. Rho, J.M., Donevan, S.D., Rogawski, M.A., 1996. Direct activation of Neuropharmacology 43, 695e700. GABAA receptors by barbiturates in cultured rat hippocampal neurons. Dalziel, J.E., Cox, G.B., Gage, P.W., Birnir, B., 1999. Mutant human alpha(1)- J. Physiol. 497 (Pt 2), 509e522. beta(1) (T262Q) GABA(A) receptors are directly activated but not modu- Sampson, W., 2005. Studying herbal remedies. N. Engl. J. Med. 353, 337e lated by pentobarbital. Eur. J. Pharmacol. 385, 283e286. 339. De Smet, P.A., 2002. Herbal remedies. N. Engl. J. Med. 347, 2046e2056. Sanna, E., Murgia, A., Casula, A., Usala, M., Maciocco, E., Tuligi, G., Ernst, M., Brauchart, D., Boresch, S., Sieghart, W., 2003. Comparative mod- Biggio, G., 1996. Direct activation of GABAA receptors by loreclezole, eling of GABA(A) receptors: limits, insights, future developments. Neuro- an anticonvulsant drug with selectivity for the beta-subunit. Neuropharma- science 119, 933e943. cology 35, 1753e1760.

+ MODEL 10 S. Khom et al. / Neuropharmacology xx (2007) 1e10

Schulz, D.W., Macdonald, R.L., 1981. Barbiturate enhancement of GABA- Ueno, S., Bracamontes, J., Zorumski, C., Weiss, D.S., Steinbach, J.H., 1997. mediated inhibition and activation of chloride ion conductance: correlation Bicuculline and gabazine are allosteric inhibitors of channel opening of with anticonvulsant and anesthetic actions. Brain Res. 209, 177e188. the GABAA receptor. J. Neurosci. 17, 625e634. Sieghart, W., 1995. Structure and pharmacology of gamma-aminobutyric Wafford, K.A., Bain, C.J., Quirk, K., McKernan, R.M., Wingrove, P.B., acidA receptor subtypes. Pharmacol. Rev. 47, 181e234. Whiting, P.J., Kemp, J.A., 1994. A novel allosteric modulatory site on Sigel, E., 2002. Mapping of the benzodiazepine recognition site on GABA(A) the GABAA receptor beta subunit. Neuron 12, 775e782. receptors. Curr. Top. Med. Chem. 2, 833e839. Wingrove, P.B., Wafford, K.A., Bain, C., Whiting, P.J., 1994. Sigel, E., Buhr, A., 1997. The benzodiazepine binding site of GABAA recep- The modulatory action of loreclezole at the gamma-aminobutyric tors. Trends Pharmacol. Sci. 18, 425e429. acid type A receptor is determined by a single amino acid in the Simon, J., Wakimoto, H., Fujita, N., Lalande, M., Barnard, E.A., 2004. Anal- beta 2 and beta 3 subunit. Proc. Natl. Acad. Sci. U.S.A. 91, 4569e ysis of the set of GABA(A) receptor genes in the human genome. J. Biol. 4573. Chem. 279, 41422e41435. Yuan, C.S., Mehendale, S., Xiao, Y., Aung, H.H., Xie, J.T., Ang-Lee, M.K., Thompson, S.A., Whiting, P.J., Wafford, K.A., 1996. Barbiturate interactions 2004. The gamma-aminobutyric acidergic effects of valerian and valer- at the human GABAA receptor: dependence on receptor subunit combina- enic acid on rat brainstem neuronal activity. Anesth. Analg. 98, tion. Br. J. Pharmacol. 117, 521e527. 353e358.