Probing the size of a hydrophobic binding pocket within the allosteric site of muscarinic acetylcholine M2-receptors

Life

ELSEVIER

Sciences,Vol. 66, No.

18, pp. 1675-1682,200O Copyright 0 2000 Elsevier Science Inc. Printed in the USA. All rights reserved 0024-3205/00/Esee front matter

PI1 SOO24-3205(00)00490-Z

PROBING THE SIZE OF A HYDROPHOBIC BINDING POCKET WITHIN THE ALLOSTERIC SITE OF MUSCARINIC ACETYLCHOLINE M,-RECEPTORS Wiebke Bender, Markus Staudt, Christian Tr%nkle*, Klaus Mohr* and Uhike Holzgrabe$ Department of Pharmaceutical Chemistry, Institute of Pharmacy, University of Wtirzburg, Am Hubland, 97074 Wiirzburg *Department of Pharmacology and Toxicology, Institute of Pharmacy, University of Bonn, An der Immenburg 4,53 121 Bonn, Germany

(Received in final form November 30, 1999)

Summary Hexane-bisammonium-type compounds containing lateral phthalimide moieties are known to have a rather high affmity for the allosteric site of muscarinic M, receptors. In order to get more insight into the contribution of the lateral substituents for alloster binding affinity, a series of compounds with unilaterally varying imide substituents were synthesized and tested for their ability to retard allosterically the dissociation of [3H]N-methylscopolamine from the receptor protein (control t,,Z = 2 min; 3 mM MgHCO, , 50 mM Tris, pH 7.3, 37 “C). Among the test compounds, the naphthalimide containing agent (half maximum effect at ECSo,diss= 60 nM) revealed the highest potency. Apparently, its affinity for the allosteric site in NMS-occupied receptors is 20fold higher compared with the phthalimide containing parent compound W 84. Analysis of quantitative structure-activity relationships yielded a parabolic correlation between the volume of the lateral substituents and the allosteric potency. The maximal volume was determined to be approximately 600 A’ suggesting that the allosteric binding site contains a binding pocket of a defined size for the imide moiety. Key Words: alkane bisammonium compounds, allosteric modulation, M, acetylcholine receptors, quantitative structure-activity relationships

The allosteric modulation of ligand binding to G protein coupled receptors is an alternative mechanism of influencing receptor function offering novel therapeutic perspectives. The formation of ternary complexes consisting of the receptor protein, the ligand bound to the orthosteric site and the allosteric agent can result in an altered dissociation of the orthosteric ligand. Allosteric effects have been studied extensively in muscarinic acetylcholine receptors, especially the M, subtype. Drugs from various pharmacological groups were found to alter the dissociation of antagonists, such as N-methylscopolamine (NMS; 1,2,3,4,5). Interestingly the most potent compounds (6) are structurally almost symmetrical, e.g. alcuronium (7), tubocurarine (8), WDuo3 (9,10), Chin3/6 (ll), and dimethyl-W84 (12). Consequently, the derived pharmacophore model established by means of molecular modeling is characterized by J To whom correspondance

should be addressed.

Vol. 66, No. 18,200O

Alkane Bisammonium Allosteric Modulators

1676

symmetry: according to the current view two aromatic areas and two positively charged nitrogens in a distance of about 10 A arranged in a sandwich-like conformation are found to be essential for high allosteric potency (13,14). Variations of the spacers between the pharmacophoric elements and of substituents on the lateral aromatic rings strongly influenced the potency and, thus, supported the model of the sandwich-like conformation (1 ,10,15,16). However, the allosteric binding site on the acetylcholine receptor is unlikely to be symmetrical. Therefore, the purpose of this study was to synthesize and pharmacologically evaluate unilaterally modified compounds which are derived from the parent structure W84. The new derivatives carry a phthalimido substituent on one end and differently substituted cyclic imides on the other end of a hexane-bisammonium spacer. It will be shown that from the analysis of quantitative structure-activity relationships (QSAR) a more precise model of the pharmacophore can be derived.

1

wa4

I

R R 0 mm:

0 cN 1

0’

0

-

\ 4a

=

_ m

O/

I :

4d

-O

:

;

4g

@

Structural formula of the parent compound W84 and the newly synthesized hexanebisammonium derivatives 4a-i

The congruent syntheses (Fig. 2) started off on the one hand with the alkylation of dimethylaminopropylphthalimide with dibromohexane to give 1. In order to avoid the bisamination of the bisbromoalkane this reagent was used in a huge excess without any solvent. On the other hand, a corresponding 1,2-dicarbonic acid was condensed in acetic anhydride to give an anhydride 2 which can be easily converted to the dimethylaminopropylimide 3. From the alkylation of 3 with 2 the corresponding unilaterally varied hexane-bisammonium derivatives 4a-i were obtained.

Vol. 66, No. 18,200O

Alkane Bisammonium Allosteric Modulators

1677

Methods

General procedures of the synthesis: In the first step, N[3(N,N-dimethylamino)propyl]phthalimide was stirred in a @@fold excess of dibromohexane at room temperature for 3 days. The so obtained solid can be filtered, several times washed with petroleum ether and recrystallized from methanol to achieve 78% 1. In order to obtain the imides 3, equimolar amounts of the corresponding anhydrides 2 and 3(N,N-dimethylamino)propylamine were refluxed in toluene using a water separator. Evaporation of the solvent and purification by means of column chromatography (silica gel, methanolKH,Cl,=l:l) gave 3 in good yields. Refluxing equimolar amounts of the compounds 1 and 3 in acetonitrile for several hours resulted in white solids which can be filtered and washed with hot acetonitrile to obtain the pure bisammonium compounds 4. The spectroscopic data are in accordance with the structures assigned to them.

(H&M

1,6-dibromohexane

0

0 /

CH3

c,---.9MBr

LHg

BrQ

3

1

H,CCN

4

Fig. 2 Reaction pathway of the bisammonium

compounds 4a-i

Preparation of porcine cardiac membranes. The preparation of the homogenate was carried out as described previously in more detail (5,9,11,12,17,18). During the preparation procedure the temperature did not exceed 4°C. Porcine ventricular tissue (20g) was obtained from the local slaughterhouse and minced and rinsed with 0.32 M sucrose solution in order to remove adherent blood. Using the same solution the tissue was homogenized first with a commercial blender (Waringm, New Hartford, CT, USA, five 10 set bursts at setting “high”) and then with a motordriven Potter Elvehjem homogenizer. After discarding a fraction of crude material by centrihgation at 300 x g for 10 min, the resulting supematant was pelleted at 80000 x g for 40 min. The resulting pellet was resuspended in 50 mM Tris-HCl pH 7.4 (4 ml/g original tissue wet

1678

Alkane Bisammonium Allosteric Modulators

weight). For storage of the homogenate nitrogen and kept at -80°C. Dissociation

USSU_VS. Retardation

in porcine heart ventricular

aliquots amounting

of the dissociation

homogenates

Vol. 66, No. 18,200O

to 1 ml were frozen in liquid

of L3H]NMS from muscarinic

M,-receptors

(3 mM MgHPO,, 50 mM Tris, pH 7.3, 37 “C) served

as a measure for the allosteric effect of the test compounds. [)H]NMS dissociation from [‘H]NMS receptor complexes formed in a 30 min preincubation period was induced by addition of 1 pM atropine alone, or in combination with the test compounds. Radioligand dissociation under control conditions (t,,? -2 min) and in the presence of the test compounds proceeded monophasically. Analysis of the corresponding data was done by non-linear regression applying a monoexponential decay function. It provided an estimate for the apparent dissociation rate constant k~, of the radioligand. All test compounds retarded [3H]NMS dissociation, “concentration-dependently”. Concentration-effect curves for the retarding action of the compounds on the dissociation were established by plotting the apparent rate constant of dissociation k_, versus the concentrations of the test compounds. The individual data for each compound were fitted applying a four parameter logistic function. The parameter “top” was set constant to a value 100%. It was then checked in successive steps, whether the goodness of fit improved upon setting the parameters “bottom” and “slope factor” nH as variables instead of “bottom” = 0% and nN = -1, respectively. This was not the case. The inflection points of the concentration effect curves thus obtained indicate the concentrations evoking a reduction of k_, to 50% (EC,,,) of the control value. The EC,,, values can be taken to reflect the equilibrium dissociation constant for alloster binding to the radioligand-occupied receptor. QUR analysis: The lateral varied N-methylimides were built up using the model builder in HyperChem 5.1 (Hypercube Inc. Waterloo, Ont. Canada 1997). Geometry optimization was achieved using the MM+ program implemented in HyperChem. By means of ChemPlus (Hypercube Inc.) the following physicochemical properties were calculated: the octanol/water partition coefficient (lo@), the steric parameters volume and surface area (grid) as well as polarizability and refractivity, a chameleon parameter comprising both steric bulk and polarizability. The allosteric potencies (EC,,) were transformed in the QSAR form log(l/EC,,). The QSARs were constructed by performing a multidimensional linear regression analysis using the BILIN software developed by Kubinyi (19). Results and Discussion The concentration-effect-curves for the allosteric stabilization of [3H]NMS-binding to the acetylcholine Mz-receptors are depicted in Fig. 3. For sake of comparison, the curve of W84 is included. Table 1 provides a list of the EC,,, values. With exception of pyridine compound 4a and the compound 4i lacking an aromatic moiety at one end of the molecule all other compounds showed a higher potency than the parent compound W84. All curves were shifted in a parallel fashion suggesting the same mode of binding as observed with W84. Qualitatively, it can be stated that the additional substituents on the phthalimide skeleton in 4b-e increased the potencies by factors of 4 to 10. Adding more aromatic rings to the imide moiety as in cpds 4f and 4g resulted in an even higher increase in potency (factor of 20). Whereas the lateral substituents in W84 and 4a-g are characterized by a flat arylimide moiety, the Diels-Alderadduct of anthracene and maleic imide in 4b occupies all three dimensions in space which led to a diminution of the potency in comparison to 4f and 4g. Omitting the aromatic area of one lateral substituent (4i) resulted in a 7fold decrease in potency in comparison with the parent

Vol. 66, No. 18,200O

Alkane Bisammonium Allosteric Modulators

1679

compound W84. Interestingly, the corresponding compound with two lateral succinimides was almost inactive (EC,, = 151 PM) (15). Thus, the bilateral replacement of a phthalimide with a succinimide evoked a stronger decrease in potency than the unilateral one. 100

1

+49 A& . 4f r4c o 4b n 4h 04d A w84

01 (mans

f S.EM,

Ok

~3-5)

I

I

I

I

-7.5

-7.0

-6.5

-6.0

test-compound

I

-5.5

I

-5.0

I

0

4i

-4.5

(log M)

Fig. 3 Concentration-effect-curves for the allosteric delay of [3H]NMS dissociation from porcine heart muscarinic M,-receptors by the test compounds. Ordinate: Apparent rate constant of [3H]NMS dissociation as a percentage of the control value in the absence of a test compound. Abscissa: log(concentration) of the test compounds. Experimental scatter is not shown when it does not exceed the symbols. Curves were fitted applying a four parameter logistic function. EC,,: concentration reducing k., to 50% of the control value, i. e., doubling the control t,,z of [‘H]NMS dissociation. Since the allosteric potencies of the compounds studied here cover more than two orders of magnitude the data set is suitable to establish a quantitative analysis of the structure-activity relationships. In Table 1 the log(l/EC,,) and the descriptors of the physicochemical properties calculated for the N-methylimides (for details see Methods) are depicted. In the first step, the QSAR of the arylimide compounds W84 and 4a-g were analyzed. In all cases, only poor or nonsignificant linear correlations were found. Probing a parabolic model for the regression analysis (20) with all descriptors, a rather good correlation was only detected between the volume of the lateral substitutent and the allosteric potency. log(liEC,,) n=S;

= -0.000089 (2 0.000036) vol’ + 0.108 (+ 0.041) ~01-25.85 s = 0.151; F =41.6; Q’ = 0.898; I2 =0.93;

(k 12.2)(l) S-PRESS = 0.199

where n is the number of compounds, 12 is the square of the correlation coefficient, s the standard deviation, F the ratio of explained to unexplained variance, Q’ is the cross validated 12 by using the leave-one-out procedure (Q’ can adopt values between 1 and zero) and s-PRESS the standard deviation from the predictive residual sum of squares. The optimum volume was calculated to be 608A’. It is worth mentioning, that the most active compound 4g was found to lie above the curve (see Fig. 4) and the difference between the measured and found volume was slightly higher than one standard deviation. The explanation for this finding might be the fact that 4g contains a six-ring imide whereas all other imides are included in a cyclopentane ring. A correlation between the potency and the lipophilicity which was found to be significant in the cases of “saturated” phthalimides (16) was not significant in the series of compounds studied here.

Alkane Bisammonium Allosterlc Modulators

1680

Allosteric

Vol. 66, No. 18.2000

TABLE I Potency of Compounds studied and Physicochemical N-methylimide moieties

Parameters of the corresponding

Compound

log( 1/EC,,,)

log p

Volume

Surface area

W84 4a 4b 4c 4d 4e 4f 4g 4h 4i

5.8’) 5.63 6.49 6.5 I 0.3-I 6.94 6.83 7.20 0.40 5.01

I .46 0.55 I .03 3.00 1.74 2.50 2.47 2.47 7.84 -0.30

IA’] 496.87 484.03 549.13 68 I .02 515.32 578.00 624.05 604.22 747.84 383.73

[A?1 326.86 321.97 355.44 421.81 338.43 368.74 387.88 372.55 45 I .46 27 1.02

Polarizibility iA’1 16.69 15.98 18.42 24.03 16.52 20.54 23.96 23.96 29.24 10.70

Refractivity [A’1 43.51 40.98 48.55 62. I8 43.94 53.12 59.96 59.96 75.89 27.47

log(l/EC’,,,): reciprocal value of the inflection point of the curves, representing the log(liconccntration) at which the apparent rate constant of [‘H]NMS dissociation was reduced by 50%. LogF calculated octanoliwater partition coefficient. Refractivity: a means for the characterization of steric bulk.

75

67 65 log(

I/ E&) 61

59

57

570

620

670

720

Volume [ AT Fig. 4 (‘on-elation between allosteric potency of W84 and 4a-g and volume of the respective Nmcthylimide moiety. Ordinate log( I/EC,,,). Abscissa: volume [A’].

Vol. 66, No. 18,200O

Alkane Bisammonium Allosteric Modulators

1681

In the next step, we included in the QSAR analysis the compounds 4h and 4i, which do not have a aryl ring directly connected to the imide moiety. The correlation between the allosteric potency and the volume became weaker, which was caused by the higher structural diversity of compounds considered: log(l/EC,,) n= 10.>

= -0.000031 (k 0.000018) volZ + 0.0391 (2 0.021) vol - 5.733 (2 5.95) s = 0.30 F = 18.0; ? = 0.84; Q’ = 0.200; S-PRESS = 0.665

The optimal volume was found to be slightly higher (640 A3), but still in the same range. Again, the difference between volume measured and volume found of the compound 4g was higher than one standard deviation (factor 1.5). Both, linear and parabolic correlations between the potency and lipophilicity were still worse than the correlation with the volume. In particular, the lipophilicity was unable to explain the strong potency of the huge molecules 4g and 4h, which were, in comparison with the volume approach, almost outliers in the lipophilicity model. In sum, the QSAR analysis indicated that the variation of one lateral imide moiety strongly influenced the allosteric potency. Volume especially, determined the potency of the compounds. The optimal volume was found to be in a range of 600 to 650 A’. This indicates that the binding site for allosteric modulators of antagonist binding to the orthosteric site is located in a sort of binding pocket which has a well defined size. Compounds with “small” lateral phthalimides, e.g. W84 or 4d, were too far away from the walls of the pocket to properly interact with residues of the amino acids of the receptor and, thus, showed low potency. Compounds with bigger aromatic imide moieties, 4f and 4g, seemed to fit nicely into the pocket. It is tempting to speculate that the naphthaline rings in these compounds made face-to-face aromatic interactions with putative aromatic amino acids of the receptor protein causing the high potency. Compounds which were too big, i.e. 4h, did not properly fit into the cavity and exhibited a smaller potency. Since 4i is far too small and additionally lacks an aromatic ring it showed the lowest potency in this series. In conclusion, in this study we have found a new compound, i.e. the naphthalimide derivative of W84, which was 20fold more potent than the parent compound. In the future, the phthalimide ring contralateral to the favorable naphthyline ring connected to the imide has to be optimized. These studies are in progress. Acknowledgement Thanks are due to the Deutsche Forschungsgemeinschaft, DFG, for financial support, to Ilona Knoblauch, Iris Witten, and Frauke Moerschel for their skillful technical assistance and to Prof. Dr. Hugo Kubinyi (BASF, Ludwigshafen, FRG) for providing the BILIN program.

References 1) S. LAZARENO, N.J.M. BIRDSALL, Mol. Pharmacol. 48 362-378 (1995). 2) U. HOLZGRABE, K. MOHR, Drug Discovery Today, 3 2 14-222 (1998) 3) S. TUCEK, J. PROSKA, TIPS, 16 205-212 (1995) 4) J. ELLIS, Drug Develop. Res. 40 193-204 (1997) 5) A. CHRISTOPOULOS, A. LANZAFAME, F. MITCHELSON, Clin. Exp. Pharmacol. Phys. 25-185-194 (1998) 6) C. TRANKLE, E. KOSTENIS, U. BURGMER, K. MOHR, J. Pharmacol. Exp. Ther., 279

1682

Alkane Bisammonium Allosteric Modulators

Vol. 66, No. l&2000

926-933 (1996) 7) J. PROSKA, S. TUCEK, Mol. Pharmacol., 45 709-717 (1998) 8) M. WAELBROECK, P. ROBBERECHT, P. DE-NEEF, J. CHRISTOPHE, J. Recept. Res., 8 787.808 (1988) 9) G. BEJEUHR, U. HOLZGRABE, K. MOHR, U. SiiRIG, A. v. PETERSENN, Pharm. Pharmacol. Lett., 2 loo-103 (1992) 10) E. KOSTENIS, H.M. BOTERO CID, U. HOLZGRABE, K. MOHR, Eur. J. Pharmacol., 314 385-392 (1996) 11) G. BEJEUHR, G. BLASCHKE, U. HOLZGRABE, K. MOHR, U. SiiRIG, G. TERFLOTH, J. Pharm. Pharmacol., 46 108-112 (1994) 12) C. T-E, E. MIES-KLOMFASS, M.H. BOTERO CID, U. HOLZGRABE, K. MOHR, Mol. Pharmacol., 54 139-145 (1998) 13) U. HOLZGRABE, M. WAGENER, J. GASTEIGER, J. Mol. Graphics, 14 185-193 (1996) 14) U. HOLZGRABE, A.J. HOPFINGER, J. Chem. Inf. Comput. Sci. 36 1018-1024 (1996) 15) T. NASSIF-MAKKI, C. TtiNKLE, D. ZLOTOS, G. BEJEUHR, A. CAMBARERI, C. PFLETSCHINGER, E. KOSTENIS, K. MOHR, U. HOLZGRABE; J. Med. Chem. 42 849858 (1999) 16) M. STAUDT, C. TtiNKLE, K. MOHR, U. HOLZGRABE, Life Sci. 62 423-429 (1998) 17) H. M. BOTERO CID, U. HOLZGRABE, E. KOSTEMS, K. MOHR, and C. T-E, J. Med. Chem., 37 1439-1445 (1994). 18) E. KOSTENIS, U. HOLZGRABE and K. MOHR, Eur. J. Med. Chem., 29 947-953 (1994). 19) H. KUBINYI, BASF AG, 67056 Ludwigshafen, Germany, 1997. 20) H. KUBINYI, QSAR: Hunsch Analysis and Related Approaches; VCH, Weinheim, 1993