Pharmaceutical Chemistry Journal

Vol. 40, No. 5, 2006

EFFECT OF THE TYRAMINE FRAGMENT OF OPIOID RECEPTOR LIGANDS ON THEIR AGONIST AND ANTAGONIST PROPERTIES N. E. Kuz’mina,1 E. S. Osipova,1 V. S. Kuz’min,1 and V. B. Sitnikov1 Translated from Khimiko-Farmatsevticheskii Zhurnal, Vol. 40, No. 5, pp. 20 – 26, May, 2006. Original article submitted July 1, 2005.

A comparative analysis of correlations between the ED50 and AD50 values describing the agonist and antagonist properties of opioid receptor ligands, on the one hand, and molecular descriptors related to the electronic and geometric properties of the tyramine moiety of these ligands, on the other hand, has been carried out. The affinity of ligands to the complementary site of opioid receptors depends on the electronic properties of the tyramine moiety, such as the basicity of the TF nitrogen atom, the electron-acceptor properties of the “cation head,” the electron-donor properties of the aryl moiety, and the ability of substituent atoms in the aryl moiety to act as proton donors and form hydrogen bonds with the OR. The intrinsic activity of ligands depends on the coplanarity of atoms of the substituent in the aryl moiety to the phenyl ring plane. Substantial protrusion of substituent atoms out of the plane of the phenyl ring apparently prevents penetration of a ligand deep into “binding pocket” of the receptor, which leads to disappearance of the antagonist properties of this ligand.

unstudied. This circumstance is related to the fact that, until quite recently, the major hypothesis was that agonists and antagonists compete for binding with the same sites of ORs [7]. This paper opens a series of papers devoted to elucidation of the properties of OR ligands, which are significant from the standpoint of both their OR affinity and the intrinsic activity. Now it is commonly accepted that the main structural component in OR ligands, which is responsible for their interaction with the receptor, is the tyramine fragment (TF) [2, 7, 8]. The influence of the TF properties on OR ligand affinity has been extensively studied and it has been established that the protonated nitrogen atom in the TF (below, TF nitrogen) accounts for its ionic interaction with an anion group of the complementary OR site, supported by the hydrogen bond (H-bond) formation [9 – 12]. The aryl fragment forms a charge-transfer complex (CTC) with a certain complementary site of the OR [5, 13]. A substituent in the phenyl ring acts as a proton donor and forms an H-bond with the OR [6, 14, 15]. However, the ability of TFs to influence the intrinsic activity of the OR ligands has not been studied so far. In this context, the main aim of this study was to establish which of the TF properties of a given ligand account for the ligand – receptor affinity and which of them determine the ligand position in the binding pocket and, hence, the intrinsic activity of this ligand.

Modern notions about ligand – opiate receptor (OR) interactions are based on the experimentally established fact according to which the OR agonists and antagonists interact with different sites of these receptors [1]. It is commonly accepted that agonists occur near the membrane, while antagonists penetrate inside the OR cluster cavity called the binding pocket [2]. The results of numerical simulations of the OR structure and the OR ligand accommodation inside the binding pocket show that bulky ligands (bivalent, peptide, etc.) acquire a strictly determined orientation in this pocket, while small ligands (such as morphinans, benzomorphans, phenylpiperidines) admit various arrangements in the OR binding pocket and, hence, can exhibit both agonist and antagonist properties [2]. Evidently, the affinity of a ligand to an OR and the ability to occupy various positions in the binding pocket depend on the properties of both the entire ligand molecule and its particular structural fragments. While the relationship between the properties of whole molecules and their affinity has been studied rather extensively [3 – 6], the influence of these properties on the accommodation of a ligand in the binding pocket and, hence, on the intrinsic activity (i.e., on the ability of a given substance to stimulate the receptor) remain almost 1

State Research Institute of Organic Chemistry and Technology, Moscow, Russia.

254 0091-150X/06/4005-0254 © 2006 Springer Science+Business Media, Inc.

Effect of the Tyramine Fragment

255

TABLE 1. Characteristics of OR Agonists and Antagonists N

R1 R3

R2 R4 Compound

1 2 3 4 5 6 7 8 9

Substituents (R1, R2, R3, R4)

10 11 12 13 14 15

CH2-cyclo-Pr, Me, Me, OH CH2-cyclo-Pr, Me, Me, NH2 CH2-cyclo-Pr, Me, Me, NHMe CH2-cyclo-Pr, Me, Me, NHEt CH2-cyclo-Pr, Me, Me, NHPr CH2-cyclo-Pr, Me, Me, NHBu CH2-cyclo-Pr, Me, Me, N(Me)2 CH2-cyclo-Pr, Me, Me, NHBz CH2-cyclo-Pr, Me, Me, NHCH2cPr CH2-cyclo-Pr, Me, Me, NO2 CH2-cyclo-Pr, Me, Me, CONH2 H, Me, Me, NH2 Pr, Me, Me, NH2 i-Bu, Me, Me, NH2 Me, Me, Et, OH

16 17 18

CH2-cyclo-Pr, Me, Et, OH CH2-cyclo-Bu, Me, Et, OH C5H11, Me, Et, OH

19

C6H13, Me, Et, OH

20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35

Me, Me, Me, SCOPh Me, Me, Me, OH Me, H, H, OH Me, Me, H, OH Me, Et, H, OH Me, Pr, H, OH Me, Et, Me, OH Me, Pr, Me, OH Me, Et, Et, H Me, Et, Et, OH Me, Et, Et, OMe Me, Et, Et, NO2 Me, Me, Me, H Me, Me, Me, F H, H, H, H Me, H, H, OMe

ED50, mg/kg

AD50, mg/kg

Ref.

Compound

0.15a 0.8a 0.44a 1.3a 1.2a 1.0a 0.79a 1.6a 0.7a

0.03 2.7 1.35 80.0 40.0 40.0 80.0 1.0 8.0

30 30 30 30 30 30 30 30 30

36 37 38 39 40 41 42 43 44

24.0a 0.06a 6.1a 5.6a 1.5a 1.2b, 0.2a 2.1a b 1.8 , 0.31a 1.8a, 4.2b 6.2b, 5.7a 0.98a 3.0b 4.4b 1.8b 2.3b 2.1b 4.9b 2.9b 2.5b 4.2b 4.6b 9.4b 13.5b 22.7b 10.2b 7.2b

9.6 n/a n/a 2.0 5.4 n/a

30 14 30 30 30 31

— — —

Substituents (R1, R2, R3, R4)

ED50, mg/kg

AD50, mg/kg

Ref.

Me, H, H, OAc Me, Me, H, OAc Me, Me, H, H C2H4F, Me, Me, OH C2H3F2, Me, Me, OH C7H15, Me, Et, OH Me, Me, Me, OCH3 CH2CH=CH2, Me, Me, NH2 CH2CH=CH2, Me, Me, OH

7.5b 3.1b 11.0b 9.1b 9.7b 7.3a 5.7b 2.5a 0.1a

— — — — — — — 1.5 0.047

36 36 36 37 37 31 32 30 38

45 46 47 48 49 50

CH2CH=C(CH3)2, Me, Me, OH Et, Me, Et, OH Bu, Me, Et, OH CH2CH=CH2, Et, Me, OH (CH2)3CH=CH2, Me, Et, OH CH2CH=C(CH3)2, Me, Et, OH

– 21.3a 13.0a – – –

6.3 5.5 2.38 0.049 1.18 3.1

30 39 39 38 39 39

31 31 31

51 52 53

CH2-cyclo-Pr, Me, Et, OH CH2-cyclo-Bu, Me, Et, OH CH2-cyclo-Pr, Me, Me, OMe

– – –

0.19 0.477 0.146

39 39 38

—

31

54

CH2CºCH, Me, Me, OH

–

0.78

38

— — — — — — — — — — — — — — — —

32 32 33 34 34 34 34 34 35 34 33 35 35 35 35 36

55 56 57 58 59 60 61 62 63 64 65 66 67 68 69

CH2C(Cl)=CH2, Me, Me, OH CH2C(CH3)=CH2, Me, Me, OH cis-CH2CH=CHCl, Me, Me, OH CH2CH=CÑl2, Me, Me, OH (CH2)2-cyclo-Pr, Me, Me, OH CH2-cyclo-Bu, Me, Me, OH CH2-cyclo-C5H9, Me, Et, OH CH2-cyclo-Pr, Et, Me, OH Pr, Me, Me, OH (CH2)2CH=CH2, Me, Me, OH CH2CN, Me, Me, OH C7H15, Me, Me, OCH3 Me, Me, Me, OAc Pr, Me, Et, OH CH2CH=CH2, Me, Et, OH

– – – – – – – – – 0.3a 12.0b 5.5b 0.97b – –

4.2 0.094 0.018 5.1 0.092 0.37 0.28 0.024 0.019 — — — — 7.0 0.6

38 38 38 38 38 38 38 38 38 16 16 40 39 39

Notes: ED50 determined by convulsant (a) and hot-plate (b) tests (solid and dashed curves, respectively, in Figs 1 – 4); (n/a) no antagonist activity; (—) no data on the activity.

256

N. E. Kuz’mina et al. (a)

(b)

30

12

20

8

AD50, mg/kg

ED50, mg/kg

10

10

6 4 2 0

0 162

186

170 174 OSNN, kcal/mole

178

–2

182

170

172

174 176 178 OSNN, kcal/mole

180

Fig. 1. Plots of (a) ED50 and (b) AD50 versus basicity of the TF nitrogen atom.

(a)

(b) 12

25

10

AD50, mg/kg

ED50, mg/kg

20 15 10

8 6 4 2

5 0 0

–2 0.235 0.240 0.245 0.250 0.255 0.260 0.265 0.270 0.275 HOMO , FAr

0.25

0.26 HOMO FAR ,

eV

0.27 eV

Fig. 2. Plots of (a) ED50 and (b) AD50 versus electron-acceptor properties of the aryl moiety in a protonated ligand molecule.

As the objects for this study, we have selected compounds of the benzomorphan group (Table 1). This choice was related to the fact that most of these molecules are capa-

TABLE 2. Coefficients of Pair Correlation between Molecular Descriptors of TF Properties and ED50 (AD50) Values Descriptor

A

B

C

D

*

– 0.76 – 0.63*

*

– 0.72 – 0.67*

*

– 0.69 – 0.67*

– 0.36 – 0.24

LUMO Fkat

– 0.76*

– 0.78*

– 0.72*

– 0.22

LUMO FAr

– 0.12

– 0.08

– 0.18

– 0.19

OSNN HOMO FAr

HH-cb Vsub Ssub Osub DA

– 0.60 – 0.25 – 0.23 – 0.16 – 0.31

*

– 0.60 – 0.16 – 0.22 – 0.33 – 0.17

*

– 0.70 – 0.04 – 0.02 – 0.03 – 0.03

* Correlation significant on a level of p < 0.05.

*

– 0.13 0.39 0.40 0.35 0.71*

ble of interacting with all OR types [14, 16, 17] and exhibit mixed agonists – antagonist behavior. These compounds were involved in numerous syntheses and pharmacological studies aimed at determining the influence of the type of substituents in the phenyl ring of the TF fragment on the opiate activity. Compounds of the benzomorphan series are conformationally “rigid,” which facilitates their modeling and favors analysis of the structure – property relationships. In order to establish the “TF property – OR affinity” and “TF property – intrinsic activity” relationships, we used the available homogeneous experimental data on the pharmacological activity in terms of the ED50 (dose of the agonist that produces an analgesic effect in 50% of animals in the test group) and AD50 (dose of the antagonist that produces a 50% antagonism of the agonist effect induced by ED50) [18]. The objects (benzomorphans) were subdivided into several reference series: (i) series A included compounds exhibiting agonist properties, for which the ED50 values were determined by the convulsant test upon subcutaneous administration (1 – 20, 41, 43, 44, 46, 47, 64, Table 1);

Effect of the Tyramine Fragment

257

(a)

(b)

25

12 10

AD50, mg/kg

ED50, mg/kg

20 15 10 5

8 6 4 2 0

0

0.060

0.060 0.062 0.064 0.066 0.068 0.070 0.072 0.074 0.076 LUMO Fkat ,

0.064

0.068

LUMO Fkat ,

eV

0.072

0.076

eV

Fig. 3. Plots of (a) ED50 and (b) AD50 versus electron-acceptor properties of the protonated TF nitrogen atom and its surrounding.

(a)

(b)

14

25

12

20

10

AD50, mg/kg

ED50, mg/kg

30

15 10

8 6 4

5

2

0 0

1

2

3

4

5

0

6

0

1

2

EH-bond, kcal/mole

3

4

5

6

EH-bond, kcal/mole

Fig. 4. Plots of (a) ED50 and (b) AD50 versus the ability of the aryl fragment to act as a proton donor and form H-bonds with ORs.

(a)

90

(b) 1.0

AD50, mg/kg

70 50 30 10

0 –10 –0.2

0.2

0.6

1.0 1.4 DA, Å

1.8

2.2

2.6

3.0

0.5

1.0

1.5 D A, Å

2.0

2.5

Fig. 5. Pots of (a) AD50 and (b) logistic function describing the presence of OR antagonism versus noncomplanarity of substituent atoms to the phenyl ring.

(ii) series B included compounds exhibiting agonist properties, for which the ED50 values were determined by the hot-plate test upon subcutaneous administration (compounds 15, 17 – 19, 21 – 40, 42, 65 – 67, Table 1);

(iii) series C included compounds exhibiting antagonist properties, for which the AD50 values were determined by the tail-flick test upon subcutaneous administration (compounds 1 – 3, 8, 10, 1 – 14, 43 – 63, 68, 69, Table 1);

258

N. E. Kuz’mina et al.

(iv) series D included compounds exhibiting affinity to ORs and considered as “potential antagonists” (i.e., containing typical antagonist substituents – cyclopropylmethyl and allyl – at the TF nitrogen atom), for which the AD50 values were determined by the tail-flick test upon subcutaneous administration (compounds 1 – 11, 43, 44, 53, Table 1). In compounds of this series, only substituents in position 8 of the phenyl ring were varied and, hence, a change in the antagonist activity could be related to this modification only. The following properties of TFs in compounds of the series A – D were analyzed: (i) the basicity of the TF nitrogen atom, (ii) the electron-acceptor properties of a “cation head” (comprising the TF nitrogen atom and its nearest surrounding) in a protonated molecule, (iii) volume; (iv) surface area; (v) oval conformation of the substituent in the aryl moiety and its coplanarity to the phenyl ring plane; (vi) the ability of substituent atoms in the aryl moiety to act as proton donors and form H-bonds; and (vii) the electron-donor properties of the aryl moiety. Previously, we have demonstrated [19] that the electronic properties of the aryl moiety involved in the formation of a CTC with a complementary site of an OR depend on the order in which the active centers of the TF interact with the receptor. If the cation head and the aryl moiety simultaneously interact with the corresponding complementary regions of the receptor, the aryl moiety probably acts as the electron donor in the CTC. In the case of a sequential interaction, whereby the electrostatic interaction of the TF nitrogen atom precedes the formation of the CTC, the aryl moiety probably acts as the electron acceptor. The oval conformation of a substituent with the surface area Ssub and the volume Vsub in the aryl moiety was characterized by the ratio Osub of its surface area to that of a sphere with the same volume [20]. The Ssub and Vsub values of substituents were calculated using additive schemes, as sums of the contributions due to separate atoms of the given fragment [20, 21]. The spatial orientation of an aryl moiety was characterized in terms of its noncoplanarity to the median plane of the phenyl ring plane, which was defined as the sum of the maximum deviations of atoms: D A = d 1 + d 2, where d1 and d2 are the maximum deviations (expressed in angströms) in the opposite directions from the median plane. The donor – acceptor properties of the aryl moiety in opiate-active ligands were characterized by the normalized electron density contributions of atoms of the phenyl ring and substituents (coplanar with this ring and involved in the conjugation) to the highest occupied molecular orbital HOMO (HOMO) of the protonated molecule (FAr ) and the lowest unoccupied molecular orbital (LUMO) of the LUMO LUMO ligand – aryl moiety associate(FAr ). The FAr values were calculated using a model of the ligand – tripeptide OR

fragment (Ile-Ast-Typ) with terminal acyl and methylamide groups, in which the protonated N atom of the ligand forms an H-bond with the carboxy group of the Asp residue [19]. The choice of the sequence of amino acid residues in the OR fragment was based on the available experimental data concerning residues essential for the binding to OR ligands. These data were obtained by methods of directed mutation and numerical simulation of receptor – ligand complexes [2, 9 – 12]. The electron density contributions were normalized to the energy of the opposite boundary orbital as [22] HOMO C(i ) = (CHOMO(n ))2/ |ELUMO|, FAr LUMO C(i ) = (CHOMO(n ))2/ |EHOMO|, FAr

where CHOMO(n ) is the coefficient of the n th atomic orbital of C(i ) atom to HOMO and CHOMO(n ) is the same for LUMO. The electron-acceptor properties of a cation head were evaluated by calculations of the normalized contributions of its atoms to the corresponding LUMO of a given ligand in LUMO the protonated form (Fcat ). The ability of substituent atoms in the aryl moiety to act as proton donors and form H-bonds was evaluated as the energy gain upon the H-bond formation in the “protonated ligand – water “ model system, where the water molecule was considered as the proton donor. It was natural to assume that the nature of interactions of ORs with agonists and antagonists is generally the same, since a large number of OR ligands exhibit simultaneously the properties of both agonists and antagonists. In this case, molecular descriptors correlated simultaneously with AD50 and ED50 values characterize the properties determining the affinity of ligands, while the molecular descriptors correlated with only one of these values characterize the properties determining the ligand orientation in the “binding pocket” and, hence, the presence or absence of intrinsic activity. The analysis of “biological activity – molecular descriptor” relationships was performed in two steps. First, the presence and degree of such a relationship between molecular descriptors, on the one hand, and the AD50 and ED50 values, on the other hand, was established by the method of correlation analysis. Then, the forms of particular relations were determined by means of regression analysis. In the course of the correlation analysis, the Pearson coefficient (r ) was calculated, which reflected the presence of a linear dependence between a given pair of variables against the background of all other variables (r falls in the interval from –1 to +1, zero corresponding to the absence of correlation, +1 to total positive dependence, and –1 to total negative dependence). The degree of relationship was characterized in terms of the Chaddock scale (see, e.g., [23]). According to this scale, the correlation is considered significant for r = 0.5 – 0.7 and high for r = 0.7 – 0.9. The significance of pair correlation coefficients was verified using the Student t-criterion. The val-

Effect of the Tyramine Fragment

ues of pair correlation coefficients for the molecular descriptors characterizing the TF properties and the AD50 and ED50 values for all series of objects are presented in Table 2. As can be seen from Table 2, compounds possessing the OR agonist properties (series A and B) exhibit a significant inverse relation between ED50 and the molecular descriptors characterizing the electronic properties of TFs such as the basicity of the TF nitrogen atom, the electron-donor properties of the aryl moiety, and the electron-acceptor properties of the cation head in protonated molecules, and the ability of substituent atoms in the aryl moiety to act as proton donors and form H-bonds with the OR sites. At the same time ED50 is weakly correlated with the molecular descriptors characterizing the TF geometry. An analysis of the form of relations between ED50 and significant molecular descriptors by means of regression analysis showed that the correlations are most adequately described in terms of the exponential function (Figs. 1 – 4). It should be noted that, although the pair correlation coefficients between ED50 and molecular descriptors are generally close, the rate of exponential decay depends on the method of ED50 determination. In compounds exhibiting antagonist activity (series C), the AD50 value is correlated to the same molecular descriptors as the ED50 value for agonists, and these AD50 correlations are also most adequately described by the exponential function. It should be noted that some of the TF properties influencing the affinity of ligands are polycollinear, that is, mutually interrelated with large correlation coefficients. Examples are offered by (i) the basicity of the TF nitrogen atom and the electron-acceptor properties of the cation head and (ii) the electron-donor properties of the aryl moiety and the electron-acceptor properties of the TF nitrogen atom and its surrounding in the protonated molecule. An analysis of the results obtained for objects in the series A – C leads to the conclusion that the affinity of agonists and antagonists to ORs in these series is determined entirely by the electronic properties of TFs. The fact that the parameters characterizing biological activity are correlated with the electron-donor properties of the aryl moiety in the protonated molecule, rather than with the electron-acceptor properties of this moiety in the ligand – OR complex, indicates that the interactions between the complementary sites of TF and OR take place simultaneously. The results obtained for series D (containing potential agonists possessing affinity to ORs) indicated that only one molecular descriptor was significant, namely, the parameter characterizing the degree of noncoplanarity (DA) of substituents to the phenyl ring plane. An analysis of the relationship between AD50 and DA showed that this correlation appears as an S-shaped curve (Fig. 5a ) that is most adequately described by the logistic function. Since this function has only two values (0 and 1), we assumed that 0 corresponds to the

259

absence of antagonism (AD50 >> 10 mg/kg) while 1 corresponds to the presence of antagonism (AD50 < 10 mg/kg). A curve of the logistic dependence of the antagonist activity on the DA is depicted in Fig. 5b. As can be seen, an average protrusion of substituent atoms from the phenyl ring plane in excess of 2Å leads to the disappearance of antagonist properties. At the same time, the agonist properties are not influenced by the noncoplanarity of a substituent in position 8 of the phenyl ring. This conclusion agrees well with the assumption that an OR binding site for antagonists is situated much deeper in the “binding pocket” than the binding site for agonists [2]. The noncoplanarity of a substituent to the phenyl ring plane probably hinders the penetration of a ligand inside the “binding pocket,” which leads to the disappearance of antagonism. Thus, the affinity to ORs and the intrinsic activity of ligand molecules depend on different properties of their TFs. The affinity is influenced by the electronic characteristics of TF regions simultaneously interacting with the complementary sites of ORs (basicity of the TF nitrogen atom, electron-acceptor properties of the cation head, electron-donor properties of the aryl moiety, and the ability of substituent atoms to act as proton donors and form H-bonds with the OR site). The position of a ligand in the OR “binding pocket” (and, hence, the intrinsic activity of this ligand) depends on the features of the TF spatial structure, primarily, on the coplanarity of substituent atoms to the phenyl ring plane. EXPERIMENTAL PART All calculations were performed using the HyperChem (Ver. 7) program package [24] and ChemOffice Ultra-8.0 software [25]. The molecular geometry was optimized in two stages: first, by the molecular mechanics method using the MM+ parametrization [26] and second, by the quantum-mechanical method using the PM3 parametrization [27]. The molecular geometry optimization was terminated upon reaching a gradient norm level below 0.1 kcal/(mole Å). The electron structure characteristics were calculated within the framework of the AM1 approximation [28]. REFERENCES 1. K. Raynor, H. Kong, S. Law, et al., Proc. Natl. Acad. Sci. USA, 91(6), 8042 – 8046 (1994). 2. I. D. Pogozheva, A. L. Lomize, and H. I. Mosberg, Biophys. J., 75(8), 612 – 634 (1998). 3. B. Bellaw, T. Conway, and F. R. Ahmed, J. Med. Chem., 17(8), 907 – 908 (1974). 4. J. P. Tollenaere and H. Moereels, Eur. J. Med. Chem., 15, 337 – 340 (1980). 5. K. S. Razzak and K. A. Hamid, J. Pharm. Sci, 69(7), 796 – 799 (1980). 6. M. P. Wentland, X. Sun, R. Lou, et al., Bioorg. Med. Chem. Lett, 13(11), 1911 – 1914 (2003). 7. P. V. Sergeev and N. L. Shimanovskii, Receptors [in Russian], Meditsina, Moscow (1987), pp. 18 – 20.

260

8. S. S. Krylov, Khim.-Farm. Zh., 26(7 – 8), 122 – 124 (1992). 9. K. Befort, L. Tabbara, S. Bausch, et al., Mol. Pharm, 49(2), 216 – 223 (1996). 10. J. G. Li, C. Chen, J. Yin, et al., Life Sci, 65(2), 175 – 185 (1999). 11. A. Mansour, L. P. Taylor, and J. L. Fine, J. Neurochem., 68(1), 344 – 353 (1997). 12. V. M Kolb and S. Scheiner, J. Pharm. Sci, 73(6), 719 – 723 (1984). 13. K. Befort, D. Tabbara, D. Kling, et al., J. Biol. Chem., 271(17), 10161 – 10168 (1996). 14. M. P. Wentland, R. Lou, Y. Ye, et al., Bioorg. Med. Chem. Lett., 11(5), 623 – 626 (2001). 15. M. P. Wentland, W. Duan, D. J. Cohen, et al., J. Med. Chem., 43(19), 3558 – 3565 (2000). 16. E. L. May, A. E. Jacobson, M. V. Mattson, et al., J. Med. Chem, 43(26), 5030 – 5036 (2000). 17. M. P. Wentland, Y. Ye, C. L. Cioffi, et al., J. Med. Chem., 46(5), 838 – 849 (2003). 18. S. P. Welch, J. W. Huffman, and J. Lowe, J. Pharm. Exp. Ther., 286(3), 1301 – 1308 (1998). 19. N. E. Kuz’mina, E. S. Osipova, V. S. Kuz’min, et al., Khim.-Farm. Zh., 39(12), 24 – 28 (2005). 20. N. Bodor, Z. Gabanyi, and C-K. Wong. J. Am. Chem. Soc, 111(11), 3783 – 3786 (1989). 21. A. Gavezotti, J. Am. Chem. Soc, 105(16), 5220 – 5225 (1983). 22. M. Karelson and V. S. Lobanov, Chem. Rev, 96(3), 1027 – 1043 (1996). 23. V. N. Salin and É. Yu. Churilova, Statistics (A Practical Guide) [in Russian], Perspektiva, Moscow (2002), p. 75. 24. http: // www.hyper.com. 25. http: // www.cambrigesoft.com.

N. E. Kuz’mina et al.

26. U. Burkert and N. L. Allinger, Molecular Mechanics. ACS Monograph. (1982), p. 177. 27. J. P. Stewart, J. Comp.-Aid. Mol. Des., 4(1), 1 – 105 (1990). 28. K. B. Lipkowitz, D. B. Boyd (eds.), Reviews in Computational Chemistry, VCH Publishers, New York (1990). 29. http: // www.statsoftinc.com. 30. M. P. Wentland, N. F. Albertson, and A. K. Pierson, J. Med. Chem., 23(1), 71 – 74 (1980). 31. I. M. Uwaydah, E. L. May, and L. S. Harris, J. Med. Chem, 20(11), 1374 – 1377 (1977). 32. G. R. Lenz, S. M. Evans, and D. E. Walters, Opiates, Academic Press., New York (1986), pp. 259 – 303. 33. E. L. May and M. Takeda, J. Med. Chem, 13(5), 805 – 807 (1970). 34. J. H. Ager, S. E. Fullerton, and E. L. May, J. Med. Chem, 6(3), 322 – 325 (1963). 35. E. L. May and A. E. Jacobson, J. Med. Chem, 8(5), 563 – 566 (1965). 36. K. Kanematsu, M. Takeda, and A. E. Jacobson, J. Med. Chem, 12(3), 405 – 408 (1969). 37. W. Reifenrath, E. B. Rocher, and W. A. Al-Turk, J. Med. Chem., 23(9), 985 – 990 (1980). 38. S. Archer, N. F. Albertson, L. S. Harris, et al., 7(2), 123 – 127 (1964). 39. E. L. May, L. S. Harris, and I. M. Uwaydah, J. Med. Chem, 20(11), 1374 – 1377 (1977). 40. A. E. Jacobson, Biological Evaluation of Compounds for Their Physical Dependence Potential and Abuse Liability, Drug Evaluating Committee of the College on Problems of Drug Dependence, Bethesda (1997).