Chem. Res. Toxicol. 1992,5, 366-375
366
Quantitative Structute-Metabolism Relationship Analyses of MAO-Mediated Toxication of l-Methyl-4-phenyl-1,2,3,6-tetrahydropyridine and Analogues Cosimo Altomare,tJ Pierre-Alain Carrupt,t Patrick Gaillard,t Nabil El Tayar,? Bernard Testa,*??and Angelo CarottiJ Institut de Chimie Th&rapeutique,Ecole de Pharmacie, Universit6 de Lausanne, B.E.P., CH-1015 Lausanne, Switzerland, and Dipartimento Farmaco-chimico, Universitd di Bari, Trav. 200 Re David 4,I-70125Bari, Italy Received September 26, 1991 The l-octanol/water partition coefficients of a number of toxic and nontoxic analogues of
l-methy1-4phenyl-l,2,3,&tetrahydropyridine (MPTP)were determined using centrifugal partition chromatography (CPC), a novel and effective technique for measuring lipophilicity, and found to be highly correlated with values calculated by a fragmental method. Some conformational properties of these compounds were also assessed by molecular mechanics calculations and ‘H-NMR spectroscopy. A quantitative structure-metabolism relationship (QSMR) study of M P T P and analogues based on literature data was undertaken in order to determine the key features eliciting MAO-A and MAO-B reactivity and selectivity and influencing toxication. Multiple regression analysis (MRA) and comparative molecular field analysis (CoMFA) showed that MAO-B activity is nonlinearly (parabolically or bilinearly) correlated to the lipophilicity of MPTP analogues and influenced negatively by steric effects exerted by bulky substituents in the ortho position. With regard to MAO-A activity, while lipophilicity was shown to play no relevant role, electrostatic and steric fields led to a 3D-QSAR model with an acceptable predictive value (cross-validated r2 = 0.571). The results of this study bring evidence at a quantitative level that the MAO-B and MAO-A catalytic sites differ in their hydrophobic, steric, and stereoelectronic requirements.
Introductlon l-Methyl-4-phenyl-l,2,3,6-tetrahydropyridine (MPTP;’ 1) is a potent neurotoxin producing symptoms very similar to those observed in Parkinsonism (1-4). MPTP must be activated before it exerts its neurotoxic actions, and several studies have established that the toxication route of MPTP involves oxidation by cerebral monoamine oxidase, principally of type B (MAO-B), to the corresponding 2,3-dihydropyridinium intermediate which then oxidizes nonenzymatically to form l-methyl-4-phenylpyridiniumion, the neurotoxin that causes cell death or disproportionates to form MPTP and MPP+ (5-9). The importance of MAO-B in the toxication of MPTP and analogues is indicated by the observation that the dopaminergic deficits associated with their administration can be prevented by treating experimental animals with selective inhibitors of this enzyme, including deprenyl (10). Recently some of us,studying a large number of MPTP analogues, derived a predictive 3D-QSAR model of MA0 reactivity (11)by means of the CoMFA approach (12,13). Some structural features influencing the binding to the MA0 active site were postulated to be as follows: (1)The N-methyl group has the ideal size and elicits ideal interaction within the MA0 active pocket, while larger groups or hydrogen appear less favorable; (2) para substituents on the phenyl ring produce steric hindrance unfavorable to reactivity; (3) ortho and meta substituents may have stabilizing interactions within the active pocket, increasing reactivity. However, this study (11)was based on biological data taken from various literature sources and expressed
* Correspondence should be addressed to this author. + Universitg f
de Lausanne. Universita di Bari.
in simDlified form bv three reactivitv levels Le., 1-3). The use oisuch reactivity levels was imposed by the heterogeneous origin of the data but did not make it possible to obtain information on possible modulating effects of molecular properties on the enzymatic reactivity, nor did it reveal the structural features eliciting MAO-A and MAO-B selectivity. It is known, in fact, that MAO-A is also relevant beside MAO-B in the bioactivation of some MPTP congeners, especially those bearing a substituent in position 2’ of the phenyl ring, and consequently in the development of Parkinson disease. For instance, going from the 2’-H in MPTP to 2’-methyl or 2’-ethyl results in compounds that are considerably better MAO-A substrates than MPTP. Hence the 2’ position was proposed as a site contributing a favorable hydrophobic interaction which increases the binding of MPTP analogues to MAO-A (14). During the submission of this paper, a new study (15) based on the structural features of 19 MPTP derivatives defined the size and the topography of MAO-A and MAO-B binding sites and confirmed the existence of (a) a hydrophobic pocket to accommodate the CZtsubstituent and (b) an additional smaller pocket compatible with small electron-withdrawingsubstituents at C3,. However, neither physicochemical parameters, especially partition data, nor structural investigations were apparently reported in order to substantiate this hypothesis and to place the observed ~~
~
Abbreviations: MPTP (l-methyl-4-phenyl-l,2,3,6-tetrahydropyridine), MPDP+ (l-methy1-4-phenyl-2,3-dihydropyridiniumion), MPP’ (l-methyl-4-phenylpyridinium ion), MA0 (monoamine oxidase), CoMFA (comparativemolecular field analysis), QSMR (quantitative structuremetabolism relationships), QSAR (quantitative structure-activity relationship), MRA (multipleregression analysis),CPC (centrifugalpartition chromatography),TMS (tetramethylsilane),PLS (partial least squares), SF (shake flask), MR (molar refractivity).
0 1992 American Chemical Society
Toxication of MPTP Analogues
Chem. Res. Toxicol., Vol. 5, No. 3, 1992 367
structuremetabolism relationships on a more quantitative basis. With the aim to further explore the spatial and physicochemical requirements for the oxidation catalyzed by MA0 (A and/or B), we investigated a number of toxic and nontoxic analogues of MPTP, measuring their partition coefficients and ionization constants, and assessed some structural factors. Combinations of physicochemical properties and structural descriptors were correlated with MA0 reactivity data taken from a single literature source (16). Quantitative structure-activity relationships were thus obtained using simple and multiple regression analysis and comparative molecular field analysis.
Materials and Methods Chemicals. MPTP (1) and its analogues (2-14) used for the physicochemical measurements were hydrochlorides. MPTP (1) and a'-methyl-MPTP (7) were purchased from Research Biochemicals (Wayland, MA). PTP (2), 4'-F-PTP (3), and N-(hydroxyethy1)-PTP (5) were purchased from Aldrich Chemical Co. (Milwaukee, WI). 4'-Amino- (9) and 3'-amino-MPTP (10) (Dr. S. Markey, Bethesda, MD), N-ethyl-PTP (4), N-phenethyl-PTP (6), 4'-chlor+MPTP (8), 3-methyl-MPTP (ll),li-pyrid-l'-yl-MTP (13), NJW-dimethylcinnamylamine(14) (Dr. W. Gessner and Dr. A. Brossi, N M , Bethesda, MD), and 4-thien-2'-yl-MTP (12) (Lilly & Co., IN) were obtained as kind gifts. Their purity was checked by HPLC. Caution: Compounds 1-14 are tonic or potentially toxic. All contact with skin, eyes, and lungs must be avoided. All solvents were of analytical grade and were purchased from Merck (Darmstadt, Germany). Lipophilicity Measurements. Distribution coefficients (log D)in the 1-octanol/water (0.01 M, phosphate buffer, pH = 7.46) system were measured by centrifugal partition chromatography. Measurements were performed at room temperature (25 & 1 "C) using an IT0 multilayer coil separator-extractor (P.C. Inc., Kim Place, Potomac, MD 20854) equipped with a preparative coil, No. 10 (i.d. 2.6 mm, total capacity 350 mL). The rotation speed of the rotor was about 1000 rpm. To measure the distribution Coefficient of all MPTP analogues reported in Table I, the volume ratio between the stationary and mobile phase and the flow rate (1.b8.0 mL/min) was adjusted depending on solute lipophilicity. The retention time of potassium dichromate or anthracene was taken as column dead time (to) when water or 1-octanol, respectively, was used as the mobile phase. Other instrumental details were as reported (1 7). The log D value of each compound was determined a t least three times and was concentration-independent; relative standard deviations were always 0.6). scale as log P. In parentheses are estimated values. dloltho:
MAO-B = 2.97(f1.14) log P = 0.56(f0.22)(10gP)’ - 0.98(f1.43) (2)
n = 26
rcv2 = 0.46 r = 0.746 F = 28.9 log Po = 2.63(*0.19) s = 0.367
MAO-B = 1.03(*0.42) log P 3.10(f1.20) log (PP + 1) + 0.57(&0.92) (3) n = 26 rcv2 = 0.46 R = 0.756 F = 29.3 log Po = 2.80 log /3 = -3.11 s = 0.369
Recently, Selassie et al. (32) suggested that the bilinear model can be helpful in exploring the limits of the interaction of ligands with the receptor sites. Specifically, they proposed that, if the slope on the right side of the bilinear part of the QSAR is essentially zero, this would correspond to substituents extending beyond the enzyme surface into aqueous space. In contrast, if the slope of the right side of the bilinear slope is clearly negative (as in eq 3), a steric hindrance effect is probably involved. At some point, the steric constraints of a hydrophobic pocket will begin to limit the optimum benefit of the hydrophobic interaction of large substituents. A combination of log P and MR terms will be useful in QSAR equations to solve this problem.
Table 111. Squared Correlation Matrix of Parameters Used in QSAR MRy MR4t MRR Zodho MAO-A MAO-B log P 0.18 0.06 0.04 0.15 0.27 0.02 MR2. 0.05 0.03 0.67 0.40 0.08 MR4, 0.01 0.03 0.25 0.02 0.02 0.09 0.19 MRR 0.30 0.16 Iortho MAO-A 0.06
Residual analysis (Le., the difference between calculated and observed activities) in eq 3 revealed a bad fit P 1 . 5 s) of compounds 4 (ethyl-PTP) and 15 (propyl-PTP), with stronger-than-predicted negative deviations reflecting the detrimental steric effect of N-alkyl groups larger than N-methyl. An higher-than-expected activity (about 1.8 s) was observed for compound 29 (3’-Br-MPTP) and is difficult to explain at this stage. Also sterically limited should be the space around the ortho substituents since the introduction of substituents larger than C1 decreases MAO-B activity. To account for these possible detrimental steric effects, we used molar refractivity values taken from literature (18) and scaled by a factor of 10. While MRR &e., MR for N-alkyl groups) proved significant in the multiparameter equation, attempts to use MR2, (Le., MR for ortho sub-
370 Chem. Res. Toxicol., Vol. 5, No.3, 1992
Altomare et al. *
stituents in the phenyl ring), also in bilinear or parabolic form, were not statistically as successful as the introduction of the indicator variable lortho, with a value of 1 for substituents larger than C1 (MR > 0.6) and 0 for the other ones. The statistically best equations obtained by MRA are as follows:
-N& \ /
I1
MAO-B 2.46(*0.75) log P - 0.43(*0.15)(10g P)’ 1.37(*0.49)MR~- 0.52(*0.28)1,,~h0 + 0.23(*0.98) (4)
n = 26
f
rcv2= 0.80 r = 0.920 F = 28.9 s = 0.226 log PO = 2.86(*0.20)
MAO-B = 1.05(*0.28) log P - 2.21(*0.81) X log (PP + 1)- 1.36(*0.51)MR~0.49(*0.30)10,~0+ 1.32(*0.69) (5) n = 26 rcv2 = 0.78 r = 0.918 F = 27.6 s = 0.235 log Po = 2.91 log /3 = -2.95
The squared correlation matrix is reported in Table 111. To emphasize the relative contribution of each independent variable, eqs 4 and 5 were also standardized (27), giving respectively (MAO-B)’ = 3.39(10g P)’- 3.07[(10g P)2]’ - 0.52(MR~)’- 0.40(10,th0)’ (6) (MAO-B)’ = 1.28(10g P)’- l.O9[lOg (PP + l)]’O.~~(MRR)’ - 0.35(1,,,h0)’ (7) The goodness of statistical parameters in both eqs 4 and 5 does not allow one to decide unequivocally if the dependence of MAO-B activity on lipophilicity is parabolic or bilinear. What is clearly demonstrated by eqs 4-7 is that hydrophobic interactions, as accounted for by the log P term, play a dominant role in modulating MAO-B oxidation of MPTP analogues and that, due to the nonlinear (parabolic or bilinear) relationship between enzyme activity and partition coefficients, an optimum value of log P around 3 is suggested. This value, within the limits of the confidence interval, is quite close to the log P value of the most active compounds (7,29,37: 3.0 < log P < 3.4). Besides log P, a minor but significant importance can be ascribed to detrimental steric effects exerted especially by N-alkyl groups and bulkier substituents in the ortho position of the 4-phenyl ring. Two compounds deviate from the eqs 4-7, namely the 4’-NO2 (32) and 4-tert-butyl(35) analogues. Perhaps the negative deviation of 32 may be due to some detrimental polar effect of the nitro group which is difficult to explain within the limits of the examined data set. The lowerthan-predicted activity of 35 could well be due to the peculiar shape of the tert-butyl moiety which does not allow it to reach as deeply as the phenyl group and elicit as much hydrophobic interaction. With the same set of independent variables only marginally interesting equations were obtained for MAO-A activity. Excluding again 4’-N02-MPTP (32) as a strong outlier, the “best” four-parameter equation and the corresponding standardized regression coefficients (27)obtained by the stepwise procedure of MRA are as follows: MAO-A = 0.28(*0.22) log P - 0.85(&0.73)MR~+ 0.39(*0.38)MR2
