Interactions of benzodiazepines and β-carbolines with a histidine residue of the benzodiazepine receptor

Neurochem. Int. Vol. 15, No. 2, pp. 145-152, 1989 Printed in Great Britain. All rights reserved

0197-0186/89 $3.00+ 0.00 Copyright © 1989 MaxwellPergamonMacmillanplc

INTERACTIONS OF BENZODIAZEPINES A N D fl-CARBOLINES WITH A HISTIDINE RESIDUE OF THE BENZODIAZEPINE RECEPTOR BERTRANDLAMBOLEZ*, CLAUDE-MARGUERITEDESCHAMPSand JEAN ROSSIER Laboratoire de Physiologie Nerveuse, CNRS, 91198 Gif sur Yvette Cedex, France (Received 28 December 1988; accepted 20 February 1989)

Abstract--The involvement of a histidine residue in benzodiazepine binding is suggested by the decrease of Ro 15-1788 and flunitrazepam binding when pH is changed from 7.5 to 5.5. Diethyl pyrocarbonate inactivates the benzodiazepine binding site by modifying a histidine residue, as shown by its faster action at pH 7.5 than at pH 5.5. After a partial diethyl pyrocarbonate inactivation, the remaining Ro 15-1788binding sites still have the same sensitivity to acid pH. No difference in the pH sensitivity of the Ro 15-1788 binding is observed between the cerebellum and the cerebral cortex. It is thus proposed that the histidine residue is present at every benzodiazepine binding site, be it either subtype I or II. The pure antagonists Ro 15-1788 and propyl-fl-carboline-3-carboxylateprotect, respectively, 36.2 and 1% of the benzodiazepine binding sites from diethyl pyrocarbonate inactivation, the full agonists flunitrazepam and diazepam provide, respectively, a 98.5 and a 46.5% protection and the full inverse agonists methyl-6,7-dimethoxy-fl-carboline-3-carboxylateand methyl-fl-carboline-3-carboxylateprovide, respectively, a 60 and an I 1% protection. Data suggest that the histidine residue is absent from the propyl-fl-carboline-3-carboxylatebinding site, and that the ability of the compounds to protect the Bdz binding sites is not related to their agonist or inverse agonist potencies. The protections observed may be due to allosteric interactions between the Bdz and fl-carboline binding sites and the histidine residue.

complex (Maksay and Ticku, 1984). Since this compound is known to react preferentially with histidine residues (Miles, 1977), the presence of this amino acid in the benzodiazepine receptor has been suggested. Our recent observation that the binding of the Bdz antagonist Ro 15-1788 decreased at acid pH (Lambolez and Rossier, 1987), has given further support to this hypothesis that a histidine residue may be located near to or at the Bdz binding site. The present study has investigated whether the amino acid residue modified by DEP is indeed a histidine and also whether this residue is present at every Bdz receptor of subtypes I or II (Sieghart, 1985). The Bdz agonists, flurazepam (Maksay and Ticku, 1984) and flunitrazepam (Fnz), protect the Bdz binding site from DEP inactivation, whereas the antagonist propyl-fl-carboline-3-carboxylate (flCCP) does not (Lambolez and Rossier, 1987). In order to further understand the mechanism of protection against DEP *Author to whom correspondence should be addressed. modification, protection experiments were performed Abbreviations: Bdz, benzodiazepines: DEP, diethyl pyro- with various other ligands for the Bdz receptor. carbonate; Fnz, flunitrazepam; flCCP, propyl-flThe present study shows that the protonation, or carboline-3-carboxylate; Cz, diazepam; flCCM, methyl-fl-carboline-3-carboxylate;DMCM, methyl-6,7- the selective modification by DEP, of a histidine dimethoxy-fl-carboline-3-carboxylate. residue present at every Bdz receptor of either sub145 Benzodiazepines (Bdz) and fl-carbolines are two chemical families of molecules binding to the Bdz binding site located on the Bdz-GABA A receptorchloride channel complex. Bdz and fl-carbolines can be divided into three classes, according to their activity: (1) agonists, enhancing GABAergic transmission; (2) inverse agonists diminishing GABAergic transmission; and (3) antagonists blocking the effects of both agonists and inverse agonists by competitive inhibition, but devoid of intrinsic activity (Haefely et al., 1985). A pretreatment by diethyl pyrocarbonate (DEP) blocks the binding of Bdz (Burch and Ticku, 1981; Sherman-Gold and Dudai, 1981; Burch et al., 1983) and fl-carbolines (Burch et al., 1983; Maksay and Ticku, 1984) without modifying the other binding sites of the Bdz-GABA A receptor-chloride channel

146

BERTRAND LAMBOLEZet al.

type I or II, blocks the binding o f Bdz. The properties o f the ligand-receptor interactions that influence the ability o f the ligands to protect the Bdz receptor from D E P inactivation are discussed.

EXPERIMENTAL PROCEDURES

[N-methyl-3H]Ro 15-1788 [2.87 TBq/mmol (78 Ci/mmol)] and [N-methyl-3H]Fnz [2.4 TBq/mmol (65 Ci/mmol)] were from CEA Saclay, France. Clonazepam, Fnz, Ro 15-1788, Ro 15-3505 and diazepam (Dz) were kindly provided by Dr W. Haefely and Dr P. Schoch, F. Hoffmann La Roche and Co. Ltd, Basel, Switzerland, Methyl-fl-carboline-3carboxylate (flCCM), flCCP and methyl-6,7-dimethoxy-flcarboline-3-carboxylate (DMCM) were gifts from Dr R. H. Dodd, ICSN CNRS, Gif sur Yvette. Absolute DEP from Fluka, stored at 4uC, was diluted 34-fold in ethanol just before use. All buffers, incubations and washes were at 4~'C unless otherwise stated, Tris-phosphate buffers used throughout the pH studies were prepared by adjusting the pH of a 50 mM Tris base solution with H 3PO4. p H studies Membrane preparation and 3H ligand binding were performed as previously described (Dodd et al., 1985). Briefly, male Wistar rats (200-250 g) were decapitated, the cerebral cortex and the cerebellum dissected, homogenized (Polytron + Potter-Elvehjem) in 50 mM Tris-HCl pH 7.4 (8.33 ml/g fresh tissue) and centrifuged (3 min at 400 g). The supernatant was recentrifuged for 20 min at 20,000 g and the pellet was resuspended in Tris-HC1 (6 ml/g original tissue) and recentrifuged (20 min at 20,000g). The resulting pellet was again suspended in the same buffer, frozen and stored at -3ff'C. Protein was estimated by the method of Lowry et al. (1951). For binding experiments, the membranes, of cerebral cortex unless otherwise stated, were diluted (75 100#g protein/ml) in the indicated buffer at the required pH (the pH of the incubation medium remained unchanged after adding membranes), and incubated for I h with the relevant ligands in a final volume of 1 ml. Non-specific binding was determined at each pH in the presence of 1 #M clonazepam except in saturation binding experiments where clonazepam was replaced by 1 # M of the unlabelled ligand used for saturation. Incubations were stopped by the addition of 3 ml of ice-cold incubation buffer to each tube and filtered under vacuum through Whatman GF/B glass fiber filters. The tube was rinsed once and the filter three times with 3ml of ice cold buffer. Bound radioactivity retained on the filters was counted in 10ml of Aquasol scintillation solution in a LKB Wallac 1215 Rackbeta 2 counter. Binding parameters were obtained from the saturation binding studies by Eadie-Hoffstee analysis (Zivin and Waud, 1982). DEP modifications Modification by DEP and protection experiments were performed as described in Lambolez and Rossier (1987) with minor modifications. Membranes were diluted 10-fold in sodium phosphate buffer 20mM containing 0.2 M NaCI, at pH 6 unless otherwise stated and centrifuged (I0 min at 29,000g). The pellet was resuspended in the same buffer at 1 2 mg protein/ml. For protection experiments, the ligand at a final concentration of 1 #M was preincubated with the membranes for 5 min at 25°C before addition of DEP. DEP

was added at a final concentration of 0.4 mM and after a 9 rain incubation at 25"C. The reaction was stopped by a 16-fold dilution into ice-cold 50mM Tris-HCl, pH7.5, followed by a 10 min centrifugation at 29,000g. The pellet was then resuspended in 35 ml of Tris HCI buffer and centrifuged again. This was repeated five times to remove the remaining DEP and protective ligands. The final resuspension was made in 18 ml Tris HCI, and the treated membranes were tested for [3H]Ro 15-1788 binding in saturation experiments as described. Control membranes were treated in an identical manner but without DEP.

RESULTS p H studies

[3H]Fnz binding to rat cortical membranes is dependent on the p H o f the incubation medium (Fig. 1). The binding was maximal at p H 7.5, was 33% lower at pH 5.5 and 64% lower at p H 10.5. Since Fnz has no p K a in this pH range ( M a u p a s and Fleury, 1982), these variations can be attributed to modifications o f the binding site. Increasing the ionic strength o f the binding buffer by addition o f 0.45 M NaCI [which is known to diminish electrostatic interactions between surface residues o f the proteins (Thomas et al., 1985)], did not change the decrease o f the binding between pH 7.5 and 5.5. In contrast, the decrease of the binding betwen pH 7.5 and 10.5 was greatly reduced in 0.45 M NaC1 (27% lower than m a x i m u m binding instead o f 64% lower at low ionic strength). Binding constants o f [3H]Fnz at pH 7.5 and 5.5 were obtained through saturation experiments (Fig. 2). The Bmax (number o f sites) was 1.25 p m o l / m g

0.05 M buffer • 100

<-~

0.5 M buffer ~,

,~

~ -...,,

z~ .

~ 50,

\

, .~>

N pH 6

7

8

9

10

Fig. I. Specific binding of [3H]Fnz (1.4 nM) to membranes of rat cerebral cortex at various pHs in 50raM Tris phosphate buffer ( 0 ) , or in 50 mM Tris phosphate buffer containing 0.45 M NaCI (~). The specific binding is expressed as a percentage of the specific binding observed at pH 7.5 (0.45 pmol/mg protein). The results are the mean of 2 experiments each performed in triplicate. The non-specific binding was determined in the presence of I p M clonazepam and was less than 7% of the total binding. Buffer of the same pH was used to rinse the filters.

A histidine residue at the benzodiazepine receptor

pH7.5/~

' ~ i

147 • 0MIN O 4 MIN

pH 5.5 •

•4

, •

~

• 7MIN ~ 15 MIN

~

1

lb

tuml (m~

Ca

0

• 0

o

.~

.'2

protein (_+0.05 SEM) at p H 7 . 5 and 0 . 9 p m o l / m g protein (_+0.04 SEM) at pH 5.5 which is 28% lower than at pH 7.5, in good agreement with the 33% lower binding of [3H]Fnz = 1.4 nM. N o effect on the KD (affinity) was observed. Figure 3 illustrates the pH sensitivity of [3H]Ro 15-1788 binding to membranes of cerebral cortex and cerebellum• Qualitatively identical results were obtained in both regions: the binding was maximal at pH 7.5, and decreased markedly at both acid (pH 5.5) and basic (pH 10.5) pH. As with Fnz, the increase in

•

/-~"/~X,,. o

Cerebral Cortex

c~,~

~ 5c

<2

6

7

8

9

10

pH

Fig. 3. Specific binding of 0.8 nM of [3H]Ro 15-1788 to membranes of rat cerebral cortex ( 0 ) and cerebellum (O) at various pHs in 50mM Tris-phosphate buffer. The specific binding is expressed as a percentage of the specific binding observed at pH 7.5. The results are the mean of 2 experiments in triplicate. The non-specific binding was determined in the presence of 1/aM clonazepam and was less than 3% of the total binding. Buffer of the same pH was used to rinse the filters.

7

.,i

B/F

Fig. 2. Eadie-Hofstee analysis of [3H]Fnz binding to membranes of rat cerebral cortex at pH 7.5 (A) and 5.5 (A) in 50 mM Tris-phosphate buffer. The points are means of duplicates. Buffer of the same pH was used to incubate and to rinse the filters. The B,,~x were 0.9 ___0.04SEM and 1.25+0.03SEMpmol/mg protein and the KD were 4.45 nM + 0.4 SEM and 4.1 nM + 0.4 SEM at pH 5.5 and 7.5, respectively.

100"

15'

4 B/F

Fig. 4. Eadie-Hofstee analysis of [3H]Ro 15-1788 binding to membranes pretreated with 0.4 mM DEP at pH 6 at 25°C for 0min (O), 4min (O), 7min ( 0 ) and 15min (O). The reaction with DEP was stopped at the indicated time by adding ice-cold 50 mM Tris-HCl pH 7.5. For the time 0 min considered as control, DEP and Tris-HCl were added simultaneously to the membranes. [3H]Ro 15-1788 binding was tested after washing 3 times the membranes as described. Assays were carried out in triplicate. This experiment was repeated with similar results. Inset: decrease with time of the Bm~x of [3H]Ro 15-1788 calculated by EadieHofstee analysis. The Bmax is expressed as a percentage of the control (0 min) B~x which was 0.48 ___0.03 SEM pmol/mg protein.

ionic strength by addition of 0.45 M N a C l did not change the p H dependency of the binding to cortical membranes between pH 7.5 and 5.5, but abolished the pH effect between p H 7.5 and l0 (not shown).

Analysis of DEP action Incubating the membranes at pH 6 with 0.4 m M D E P at 25°C caused a time-dependent reduction o f [3H]Ro 15-1788 Bmax without changing the K o (Fig. 4). The loss of binding sites followed an exponential time course (inset). Figure 5 compares the effects of 0.4 m M D E P treatment at 25°C during 9 min, either at pH 5.5 or 7.5, on [3H]Ro 15-1788 binding. D E P at p H 7.5 reduces by 70% the BmaXof [3H]Ro 15-1788 (from its control value of 1.9 pmol/mg protein + 0.03 SEM to 0.58 pmol/mg protein _+ 0.03 SEM). D E P acts therefore faster at p H 7.5 than 5.5, where it only reduces the Bmax by 50% [from 1.73 pmol/mg protein ___0.03 SEM (control) to 0.84 pmol/mg protein _ 0.04 SEM]. Furthermore, at pH 7.5, D E P strongly increases the K D of [3H]Ro 15-1788 from 0.56nM___0.03 SEM (control) to 1.44 nM ___0.14 SEM (which represents a 157% increase). In contrast, D E P had minor effects on KD at p H 5.5 (a 22% increase from 0.68 nM + 0.03 SEM (control) to 0.83 nM_+0.12 SEM) as seen at pH 6 (see Fig. 4).

BERTRAND LAMBOLEZ~~ ul.

148

-8 i’

1 0 CONTROL pti 7.5 2+ CONTROL pH 5.5 30 OEP pH 5.5 4 0 OEP pH 7.5

I

I

0 CONTROL

,d

0 DEP TREATED

f

Fig. 5. Eadie-Hofstee analysis of [‘H]Ro 15-1788 binding to membranes incubated 9 min at 25°C: (0) at pH 7.5 (control), (+) at pH 5.5 (control), (0) with 0.4 mM DEP at pH 5.5 and (0) with 0.4 mM DEP at pH 7.5. The binding of [)H]Ro 15-1788 was performed after rinsing the membranes as described. The B,,, _+SEM (pmol/mg protein) and K,, k SEM (nM) were, respectively: (1) 1.9 k 0.03 and 0.56 +0.03; (2) 1.73 + 0.03 and 0.68 + 0.03; (3) 0.84 k 0.04 and 0.83 + 0.12: and (4) 0.58 + 0.03 and 1.44 +0.14. The experiment, in triphcate, was repeated twice with qualitatively similar results.

pH sensitivity

after DEP action

The binding of 0.73 nM of [3H]Ro 15-1788 is shown to remain sensitive to pH after DEP treatment (Fig. 6). After a 7min incubation of the membranes with 0.4 mM DEP at 25°C at pH 6 the binding of [3H]Ro 15-1788 binding at pH 7.5 was 60% of control. Between pH 7.5 and 5.5, the binding decreased by 44% for DEP treated membranes and by 43% for control. Protection

experiments

Incubating membranes at pH 6 with 0.4 mM DEP for 9 min at 25°C was found to block reproducibly half of the Bdz binding sites (52.4 + 4.3%, n = 20). This protocol was used to compare the ability of various ligands to protect the [3H]Ro 15- 1788 binding sites against DEP inactivation (Fig. 7). The K, observed after protection was unchanged or slightly increased (less than 30%) probably be-

Fig. 6. specific binding of 0.73 nM of [‘H]Ro 15-1788 at various pHs in 50 mM Tris-phosphate buffer to membranes incubated for 7 min at 25°C with 0.4 mM DEP (0) and to control membranes treated in an identical manner but without DEP (0). Membranes were rinsed 3 times before the binding of [‘H]Ro 15- 1788. Specific binding is expressed as a percentage of that observed at pH 7.5. Non-specific binding was determined in the presence of 1 p M clonazepam and was below 3% of total binding. The binding at pH 7.5 of [3H]Ro 15-1788 to DEP treated membranes was 40% lower than that observed with control membranes.

cause of incomplete removal of the protecting ligands (since similar increases were found with control membranes). I -Benzodiazepines. Almost no decrease in B,,, was observed when Fnz, a full agonist (Haefely et al., 1985), was present during DEP treatment: this ligand protected 98.5% of the sites. The full agonist Dz (Haefely et al., 1985), protected 46.4% of the sites. Ro 15-3505, a partial inverse agonist (Haefely et al., 1985) protected 44% of the sites and the antagonist Ro 15-1788 protected 36.2% of the sites. 2-,5-Carbolines. DMCM, the most potent inverse agonist (Haefely et al., 1985), gave a 60% protection of the Bdz binding sites. PCCM, a full inverse agonist (Haefely et al., 1985) protected 10% of the sites whereas the antagonist BCCP did not protect the Bdz binding site. At 1 PM, each ligand was at a concentration 3 log units above its K,, and was therefore saturating its binding sites before the addition of DEP in the

Fig. 7 (@ing page). Protection of [‘H]Ro 15-1788 binding sites against the inactivation by 0.4 mM DEP at 25°C during 9 min by various ligands. Each ligand (1 PM) was incubated with the membranes 5 min before the addition of DEP. B,,,,, and K,, were calculated from Eadie-Hofstee transformations of the data. The mean (+ SEM) Bm, decrease was 52.4% & 4.3 of control for the 20 experiments reported here. For each ligand a control was made with 1PM ligand without DEP to control the washes. n is the number of independent experiments. For each experiment, the number of sites blocked by DEP (which is the difference: B,,, control - B,,,,, after DEP) was calculated. The percentage of sites protected is the difference: (number of sites blocked by DEP) - (number of sites blocked by DEP + ligand) expressed as a percentage of the sites blocked by DEP.

Ligand

% of sites protected CH3

0

Fnz

98.5 + 4.46

(n=3)

O2N

F

CH3

Oz

(n =2) g

ct~~S9

48.7 44.1

e~ t~

B O

Ro 15-3505 (n=3)

44 + 2

oc..s

Ct

O

CH3 O

Ro 15-1788 (n=3)

36.2 + 4.35

O

CHa

DMCM (n=4)

60--* 7.5

H

O ,.o

O

13-CCM (n =2)

O/CH3

11.2 8.9

I

H 1.3 + 2.7

-CCP (n--3)

OCH2 H

149

CH2~CH3

150

BERTRAND LAMBOLEZeta[.

conditions used (Chiu et al., 1982). Thin layer chromatography and spectrophotometry were used to verify that DEP did not react with Fnz, Ro 15-1788, f l C C M or flCCP.

DISCUSSION

The antagonist [3H]Ro 15-1788 was used throughout this study to investigate the Bdz recognition site since its binding is far less susceptible than those of agonists to allosteric modulation (Haefely et al., 1985), and thus allows the ligand recognition site interaction to be studied in isolation of the GABA receptor-chloride channel function. The pH sensitivities of [3H]Fnz (Fig. 1), [3H]Ro 15-1788 (Fig. 3; and Lambolez and Rossier, 1987) and [3H]Dz (Braestrup and Squires, 1977), show the same common features with an optimum at pH 7.5. They are therefore probably characteristic of the Bdz binding site itself and of its interaction with its ligands. The pH sensitivity is reversible (Lambolez and Rossier, 1987) and therefore not caused by receptor denaturation. Consequently, the pH sensitivity can be attributed to protonation or deprotonation of amino acid residues, in the area of the binding site or interacting with it. It should be noted, however, that the decrease of Fnz binding seen at acid pH is lower than that observed for Ro 15-1788 (Lambolez and Rossier, 1987) and Dz (Braestrup and Squires, 1977), this may be due to anomalies in the interactions of Fnz with the Bdz receptor, as discussed below. The pH sensitivity of fl-carbolines was not studied because these molecules have a pKa between 7 and 8 (Pr B. Fleury personal communication) which complicates the interpretation of pH induced modifications in fl-carboline binding. The principal conclusions of this study are the following: (a) The protonation or DEP modification of a histidine residue in the area of the binding site, or interacting with it, inactivates Bdz binding. This hypothesis is reinforced by three points. (1) The decrease in both Fnz (Fig. 1) and Ro 15-1788 (Fig. 3) bindings at acid pH is not shifted by increasing the ionic strength. This indicates that the pKa of the amino acid residue whose protonation decreased the binding of Bdz between pH 7.5 and 5.5 is close to its value in solution. Histidine is the only amino acid having a pKa in this pH range (Lehninger, 1982). It is therefore likely that the binding decrease of Bdz between

pH 7.5 and 5.5 is due to the protonation of a histidine residue, (2) We found that DEP acts faster at pH 7.5 than at pH 5.5 (Fig. 5) as found in (Maksay and Ticku, 1984), but in contrast with (Sherman-Gold and Dudai, 1983). This is expected from the reaction of DEP with a histidine residue [ p K a = 6 (Lehninger, 1982)], since DEP acts on unprotonated residues. At a pH at which DEP is selective for histidine residues (at pH 6 or below), the effect of this reagent on Ro 15-1788 binding was limited to a decrease in Bin,~. However, at pH 7.5, where it becomes non-selective (Miles, 1977), an effect on K D was also observed. (3) The sensitivity of Fnz and Ro 15-1788 bindings to basic pH is strongly dependent on ionic strength. This may result from a shift of the pKa of an amino acid residue, in or close to the binding site, towards the more basic value it has in solution. In this case, the pKa in solution would be above 10 since the decrease in Ro 15-1788 binding is above pH 10 in 0.5 M buffer. This rules out a direct involvement of a tyrosine residue [pKa = 9.5 (Lehninger, 1982)] in the Bdz binding. From the points (1), (2) and (3) we conclude that at pH 6, the DEP modification of a histidine residue, and not a tyrosine as suggested in Sherman-Gold and Dudai (1983), is responsible for the inactivation of the Bdz binding site. The protonation of this histidine has a similar effect. (b) The histidine residue is present at every Bdz receptor. It is possible that the 50% of the Ro 15-1788 binding sites remaining at pH 5.5 (Fig. 3) may correspond to the reported distribution of the putative Bdz receptor subtypes 1 and 2 in cortex (Sieghart, 1985). This possibility can be ruled out by the observation that in spite of the modification of 40% of the Bdz binding sites by DEP, the remaining sites conserve the same sensitivity to acidic pH (Fig. 6), indicating thay they also have a histidine residue. Furthermore, in cerebellum, where 90% of Bdz receptor 1 and 10% of Bdz receptor 2 are found (Sieghart, 1985), the pH sensitivity was the same as in cortex (Fig. 3). Therefore, if such Bdz receptor subtypes exist, their binding sites both have or interact with a histidine residue and share sufficient structural homology to have the same pH dependency. This is consistent with reports showing that DEP can also block 100% of the Bdz binding sites in cerebellum as it can in cortex (Burch et al., 1983). (c) Mechanisms underlying protection.

A histidine residue at the benzodiazepine receptor (1) The histidine residue is probably not protected by direct interactions with the ligands. The protonation of the histidine residue causes a decrease in the Bmgxof RO 15-1788 and Fnz which is characteristic of a non-competitive inhibition (a KD decrease would be observed in case of a direct interaction between the ligands and the protonated histidine residue). In addition, all the ligands tested, except Fnz, do not or only partially prevent the inactivation of the Bdz binding site by DEP (Fig. 7). This rules out a direct interaction of these ligands with the histidine residue modified by DEP. (2) The protective potency may come from allostery of the Bdz receptor unrelated to agonist activity. The partial inverse agonist Ro 15-3505, the antagonist Ro 15-1788 and the full agonist Dz provided a similar degree of protection. The protective potencies do not seem therefore related to the in vitro efficacy of the Bdz ligands. Furthermore, Fnz and flurazepam, which are very similar to Dz in in vitro agonist potency (Karobath and Supavilai, 1985; Chan and Farb, 1985), provide complete protection (Fig. 7 and Maksay and Ticku, 1984). Similar conclusions can be drawn for fl-carbolines, where again no correlation exists between their in vitro efficacies and the protection of the histidine residue against DEP. Three observations indicate that Fnz interacts with the Bdz receptor differently from Dz and Ro 15-1788 (although its in vitro agonist properties are quite similar to those of the former). - - F n z binding was much less sensitive to acid pH than Ro 15-1788 (Figs 1 and 2 and; Lambolez and Rossier, 1987) or Dz (Braestrup and Squires, 1977) binding. - - T h e capacity of Fnz to protect the Bdz receptor against DEP modification was much higher than that of Dz or Ro 15-1788 (Fig. 7). - - T h e dissociation kinetics of Fnz show an important slow component, not present or very slight for Dz and Ro 15-1788 (Doble, 1982; Brown and Martin, 1984; Chiu and Rosenberg, 1982). These anomalous interactions of Fnz with the Bdz receptor might be explained by a strong ability of this ligand to induce conformational changes of the receptor (Chiu and Rosenberg, 1982), independent to those involved in the potentiation of GABAergic transmission. The discrepancy between the protective potency of flCCM and of D M C M may perhaps be due to the

151

same phenomenon: D M C M also shows polyphasic dissociation kinetics (Braestrup et al., 1983), suggesting that it may change the conformation of the receptor, when the dissociation of ethyl-fl-carboline3-carboxylate [another fl-carboline inverse agonist (Haefely et al., 1985)], proceeds monoexponentially (Martin and Doble, 1983). These conformational changes would be induced to different degrees by the binding of the other iigands tested, apart from flCCP devoid of this activity. We suggest in conclusion that reciprocal allosterical interactions between a histidine residue and the Bdz and fl-carboline binding sites may explain the inactivation of the binding sites by histidine modification and the protection of the histidine against DEP by some of the ligands tested. Acknowledgements--We would like particularly to thank

Dr Adam Doble for critical appraisal of the manuscript. B. Lambolez holds a Minist6re de la Recherche et de l'Enseignement Sup6rieur grant. This work was supported in part by the Fondation pour la Recherche M6dicale.

REFERENCES

Braestrup C. and Squires R. F. (1977) Specific benzodiazepine receptors in rat brain characterized by high affinity [3H]diazepam binding. Proc. natn. Acad. Sci. U.S.A. 74, 3805-3809. Braestrup C., Nielsen M. and Honore T. (1983) Binding of [3H]DMCM, a convulsive benzodiazepine ligand, to rat brain membranes: preliminary studies. J. Neurochem. 41, 454-465. Brown C. L. and Martin I. L. (1984) Kinetics of [3H]Ro 15-1788 binding to membrane-bound rat brain benzodiazepine receptors. J. Neurochem. 42, 918-923. Burch T. P. and Ticku M. K. (1981) Histidine modification with diethyl pyrocarbonate shows heterogeneity of benzodiazepine receptors. Proc. natn. Acad. Sci. U.S.A. 78, 3945-3949. Burch T. P., Thyagarajan R. and Ticku M. K. (1983) Group-selective reagent modification of the benzodiazepine-gamma-aminobutyric acid receptor-ionophore complex reveals that low-affinity gamma-aminobutyric acid receptors stimulate benzodiazepine binding. Molec. Pharmac. 23, 52-59. Chan C. Y. and Farb D. H. (1985) Modulation of neurotransmitter action: control of the gamma-aminobutyric acid response through the benzodiazepine receptor. J. Neurosci. 5, 2365-2373. Chiu T. H. and Rosenberg H. C. (1982) Comparisons of the kinetics of [3H]diazepam and [3H]flunitrazepam binding to cortical synaptosomal membranes. J. Neurochem. 39, 1716-1725. Chiu T. H., Dryden D. M. and Rosenberg H. C. (1982) Kinetics of [3H]flunitrazepam binding to membranebound benzodiazepine receptors. Molec. Pharmac. 21, 57-65. Doble A. (1982) GABA abolishes cooperativity between benzodiazepine receptor. Eur. J. Pharmac. 83, 313-316.

152

BERTRAND LAMBOLEZet al.

Dodd R. H., Ouannes C., Prado De Carvalho L., Valin A., Venault P., Chapouthier G., Rossier J. and Potier P. (1985) 3-Amino-fl-carboline derivatives and the benzodiazepine receptor. Synthesis of a selective antagonist of the sedative action of diazepam. J. reed. Chem. 28, 824-828. Haefely W., Kyburz E., Gerecke M. and Mohler H. (1985) Recent advances in the molecular pharmacology of benzodiazepine receptors and in the structure activity relationships of their agonists and antagonists. Adv. Drug. Res. 14, 165 322. Karobath M. and Supavilai P. (1985) Interaction of benzodiazepine receptor agonists and inverse agonists with the GABA benzodiazepine receptor complex. Pharmac. Biochem. Behav. 23, 671~74. Lambolez B. and Rossier J. (1987) Benzodiazepine agonists protect a histidine residue from modification by diethyl pyrocarbonate whereas propyl-fl-carboline does not. FEBS Lett. 219, 301-305. Lehninger L. A. (1982) Principles o f Biochemistry, 2nd edn. Worth, New York. Lowry O. H., Rosebrough N. J., Farr A. L. and Randall R. J. (1951) Protein measurement with the Folin phenol reagent. J. biol. Chem. 193, 265-275. Maksay G. and Ticku M. K. (1984) Characterization of gamma-aminobutyric acid-benzodiazepine receptor com-

plexes by protection against inactivation by group-specific reagents. J. Neuroehem. 42, 1715-1727. Martin I. L. and Doble A. (1983) The benzodiazepine receptor in rat brain and its interaction with ethyl-flcarboline-3-carboxylate. J. Neurochem. 40, 1613-1619. Maupas B. and Fleury M. B. (1982) Analysis of acid-base equilibriums of 1-4 benzodiazepines. Effect of mesomeric and steric effects. Analysis 10, 187-190. Miles E. W. (1977) Modification of histidyl residues in proteins by diethyl pyrocarbonate. Meth. Enzym. 47, 431-442. Sherman-Gold R. and Dudai Y. (1981) Involvement of tyrosyl residues in the binding of benzodiazepines to their brain receptors. FEBS Lett. 131, 313-316. Sherman-Gold R. and Dudai Y. (1983) Diethyl pyrocarbonate modification of benzodiazepine receptors from calf cerebral cortex. Neurochem. Res. 8, 259-267. Sieghart W. (1985) Benzodiazepine receptors: multiple receptors or multiple conformations. J. Neural Trans. 63, 191 208. Thomas P. G., Russel A. J. and Fersht A. R. (1985) Tailoring the pH dependence of enzyme catalysis using protein engineering. Nature 318, 375 376. Zivin J. A. and Waud D. R. (1982) How to analyse binding, enzyme and uptake data: the simplest case, a single phase. Life Sci. 30, 1407 1422.