1027Quantum-Chemical Descriptors in QSAR/QSPR Studies Mati Karelson* and Victor S. Lobanov Department of Chemistry, University of Tartu, 2 Jakobi Str., Tartu, EE 2400, Estonia Alan R. Katritzky Center for Heterocyclic Compounds, Department of Chemistry, University of Florida, P.O. Box 117200, Gainesville, Florida 32611-7200 Received August 21, 1995 (Revised Manuscript Received February 26, 1996) Contents Introduction Quantum Chemical Methods Quantum Chemical Descriptors QSAR/QSPR Results A. Biological Activities B. Chemical Reactivities C. Partition Coefficients D. Chromatographic Retention Indexes and Response Factors E. Other Physicochemical Properties F. Substituent Constants G. Solvational Characteristics H. CoMFA V. Conclusions VI. References I. II. III. IV. 1027 1027 1029 1033 1033 1035 1036 1036 1037 1038 1038 1040 1040 1041 I. Introduction Quantitative structure-activity and structureproperty relationship (QSAR/QSPR) studies are unquestionably of great importance in modern chemistry and biochemistry. The concept of QSAR/QSPR is to transform searches for compounds with desired properties using chemical intuition and experience into a mathematically quantified and computerized form. Once a correlation between structure and activity/property is found, any number of compounds, including those not yet synthesized, can be readily screened on the computer in order to select structures with the properties desired. It is then possible to select the most promising compounds to synthesize and test in the laboratory. Thus, the QSAR/QSPR approach conserves resources and accelerates the process of development of new molecules for use as drugs, materials, additives, or for any other purpose. While it is not easy to find successful structureactivity/property correlations, the recent exponential growth in the number of papers dealing with QSAR/ QSPR studies clearly demonstrates the rapid progress in this area. To obtain a significant correlation, it is crucial that appropriate descriptors be employed, whether they are theoretical, empirical, or derived from readily available experimental characteristics of the structures. Many descriptors reflect simple molecular properties and thus can provide insight into the physicochemical nature of the activity/ property under consideration. Recent progress in computational hardware and the development of efficient algorithms has assisted the routine development of molecular quantummechanical calculations. New semiempirical methods supply realistic quantum-chemical molecular quantities in a relatively short computational time frame. Quantum chemical calculations are thus an attractive source of new molecular descriptors, which can, in principle, express all of the electronic and geometric properties of molecules and their interactions. Indeed, many recent QSAR/QSPR studies have employed quantum chemical descriptors alone or in combination with conventional descriptors. Quantum chemistry provides a more accurate and detailed description of electronic effects than empirical methods.1 Quantum chemical methods can be applied to quantitative structure-activity relationships by direct derivation of electronic descriptors from the molecular wave function. In many cases it has been established that errors due to the approximate nature of quantum-chemical methods and the neglect of the solvation effects are largely transferable within structurally related series; thus, relative values of calculated descriptors can be meaningful even though their absolute values are not directly applicable.2 Moreover, electronic descriptors derived from the molecular wave function can be also partitioned on the basis of atoms or groups, allowing the description of various molecular regions separately. Most work employing quantum chemical descriptors has been carried out in the field of QSAR rather than QSPR, i.e. the descriptors have been correlated with biological activities such as enzyme inhibition activity, hallucinogenic activity, etc.3-6 In part this has been because, historically, the search for quantitative relationships with chemical structure started with the development of theoretical drug design methods. Quantum-chemical descriptors have also been reported to correlate the reactivity of organic compounds, octanol/water partition coefficients, chromatographic retention indices, and various physical properties of molecules.7-11 The present article reviews applications of quantum chemical descriptors in the development of QSAR/QSPR dealing with the chemical, physical, biochemical, and pharmacological properties of compounds. II. Quantum Chemical Methods Methods based on classical molecular force fields and quantum-chemical methods are each capable of N. However. Thus. available solutions of the respective Schrodinger ¨ equations are necessarily approximate and the computational time is proportional to a high exponential (N4. MP2.19-21 correlated pair many-electron theory (CPMET)22 and its various coupled-cluster approximations. in chemistry at the Moscow State University. He is actively engaged in research and teaching. in 1995. minimizing the potential energy of a molecular structure. this method provides better results the larger the basis set (i. For this reason.e.12 In principle.30. molecular orbital energies. in 1988. are less directly related to the size of the basis set. number of atomic orbitals) employed.g. these procedures are widely known as semiempirical methods. on QSAR/QSPR at the University of Tartu.1028 Mati Karelson (born in 1948) is the Professor and Head of Theoretical Chemistry at the University of Tartu. quantum-chemical theory should be able to provide precise quantitative descriptions of molecular structures and their chemical properties. industrial consulting.16-18 multiconfigurational self-consistent field (MC SCF). although approximate.31 . In 1994. and windsurfing.28.31 Experimental data on atoms and prototype molecular systems have often been used to estimate values of quantities used in the calculations as parameters. partial charges on atoms). 31. Estonia.29 Most of these methods are extremely time consuming and require large CPU memories and are therefore impractical for the calculation of extended sets of relatively large molecules (i. and the many other electronic descriptors of potential value to QSAR studies. thus. Both approaches can be used for thermo- dynamic and dipole moment calculations but only quantum-chemical methods can estimate atomic σ charges. MP3. Karelson is a member of the International Socity of Quantum Biology and Pharmacology and the New York Academy of Sciences. 37. Russia. received his M. more than 10 atoms). born in 1966 in Yekaterinburg. These include configuration interaction (CI). even within these limitations molecules may be described by ab initio methods with some degree of reliability after an accurate search on the potential energy surface(s) has been carried out at a lower level of theory. As an alternative to ab initio methods. London. have now become routine and have been demonstrated to provide results of real utility. Lobanov.15 and for such properties more attention has to be paid to the balance of the basis set used.D on physical organic chemistry in 1975. N5) of the number of electrons in the molecule. 569−602) and an overview of his scientific work in Heterocycles (1994.. He received his Ph. semiempirical quantum-chemical methods can be used for the calculation of molecular descriptors. He received his Ph. but based on simplifications and approximations introduced into the computational procedure which dramatically reduce the computational time. In general. Katritzky (born 1928. Russia. the foundations of the QSAR/QSPR. M. His research has dealt with the theory of solvent effects. editing. These methods have been developed within the mathematical framework of the molecular orbital theory (SCF MO).e. Victor S.D. due to mathematical and computational complexities this seems unlikely to be realized in the foreseeable future. he was nominated as a Courtesy Professor in Chemistry at the University of Florida.14 Other electronic properties. A light-hearted account of his life appeared in the Journal of Heterocyclic Chemistry (1994.16 A wide variety of ab intio methods beyond Hartree-Fock have been developed and coded to account for electron correlations in the molecule.30. Møller-Plesset theory of various orders.17.13 However. Alan R. researchers need to rely on methods which. UK) is Kenan Professor of Chemistry and Director of the Institute for Heterocyclic Compounds at the University of Florida. 3−130). MP4). although according to the variational principle this is strictly valid only for the total electron energy of the molecule.8 Most ab initio calculations have been based on the orbital approximation (Hartree-Fock method). Estonia. particularly those describing electron distribution in the molecule (dipole and higher moment.S.23-27 and perturbation theory (e. While the ab initio model Hamiltonian8 provides a complete representation of all nonrelativistic interactions between the nuclei and electrons in a molecule. and the development of the respective computer software. international travel. practical ab initio calculations are severely limited by the types of atoms and size of molecules. Several reviews have been published on the application of quantum-chemical descriptors in QSAR.35 modified neglect of diatomic overlap (MNDO). The MNDO and AM1 methods have the advantages of relatively short computational times (compared with ab initio calculations) and the availability of parametrization for a variety of atoms (all the elements of the second period. for example see refs 3 and 12 and references cited therein. readers are referred to a comprehensive.12 A summary of the most frequently used quantumchemical descriptors is given in Table 1. but consistent in their trends. S.33 intermediate neglect of differential overlap (INDO). This method is preferably used to describe qualitatively the electronic structure of molecular systems.35 MNDO.32.8 CNDO is the simplest (and most approximate) semiempirical method. dipoles. although there is some natural overlap. involving the neglect of both diatomic and single-atom atomic orbital overlap.36 AM1. For example.40 For additional background information on ab initio and semiempirical quantum chemical methods.13 It has already been mentioned that while the results produced by different semiempirical methods are not generally comparable. In most cases the direction but not the magnitude of the error is known. ¨ complete neglect of differential overlap (CNDO).13 Comparison of ab initio (STO-3G) and AM1 derived descriptors in the QSAR study of toxicity39 suggests that the use of AM1 values instead of STO-3G values is completely acceptable and even desirable. MNDO.37 and parametric model 3 (PM3).36 Austin model 1 (AM1). all chemical interactions are by nature either electrostatic (polar) or orbital (covalent). neglecting diatomic differential overlap only). although now somewhat outdated. Si.34. According to classical chemical theory. P. the AM1 method provides good descriptions even for anions and hydrogen bonded systems. the computational error is considered to be approximately constant throughout the series. Unlike experimental measurements there is no statistical error in quantum-chemical calculations. their use in the design of a training set in a QSAR study presents two main advantages: (a) the compounds and their various fragments and substituents can be directly characterized on the basis of their molecular structure only.243 III. electronic properties calculated by the MNDO method may well be less reliable than those calculated by the AM1 method. review by Loew and Burt. any semiempirical method can be used to calculate quantum chemical descriptors.12 In using quantum chemistry-based descriptors with a series of related compounds. To name but some of the most popular: extended Huckel theory (EHT). parameters for almost all types of atoms are available. and INDO methods are quite different in their absolute values. Hg. Sn. Al. Because of the neglect of important electronic interactions. some reports demonstrate that the AM1 method can be used to calculate electronic effects which are difficult to deal with by ab initio methods. they do often reproduce similar trends. associated with the assumptions required to facilitate the calculations. shape and binding properties of a complete molecule as well as of molecular fragments and substituents.38 Descriptors calculated by different methods may have different significance in obtaining correlations8 which strictly should be defined only within the framework of one particular method.31 The INDO and MINDO procedures were developed as a compromise in which the one-center overlaps are retained in onecenter integrals. and (b) the proposed mechanism of action can be directly accounted for in terms of the chemical reactivity of the compounds under study. The descriptors given in the table can be subdivided as follows: Atomic Charges. halogens. Quantum-chemically derived descriptors are fundamentally different from experimentally measured quantities. Electrical charges in the molecule are obviously the driving force of electrostatic interactions. and bond lengths are more reliable than those obtained from low-quality ab initio methods.37 and PM338 are the methods based on the correct inclusion of one-center overlap (i. as well as for a comparison of their applicability.3 The extended Huckel theory (EHT) is an extension ¨ of the Huckel π-electron approximation and treats ¨ all valence electrons in a molecule. AM1 semiempirical charges. the derived QSAR models will include information regarding the nature of the intermolecular forces involved in determining the biological or other activity of the compounds in question. Quantum Chemical Descriptors Quantum-chemical methods and molecular modeling techniques enable the definition of a large number of molecular and local quantities characterizing the reactivity. and Pb).37 The MNDO method tends to overestimate electronic repulsion. The variation in the molecular orbital indices is generally largest for AM1 and diminishes in the order AM1 > MNDO > INDO.8. A basic weakness of quantum-chemical descriptors is the failure to directly address bulk effects.39 Moreover. the charge distributions deduced are often unrealistic.e. it has been proven that local electron densities or charges are important in many chemical reactions and physico- .8 Overlap integrals are calculated over Slater-type atomic functions while electronic and nuclear repulsion are neglected. the net electronic charges calculated by the AM1.34 modified INDO (MINDO).32 The same principles apply to the improved version CNDO/2.41 Consequently. given that AM1 calculations are obtained more rapidly. Further arguments in favor of employing AM1 molecular descriptors include notably the general observation that minimal basis set ab initio results are frequently inferior both to the more sophisticated extended basis set calculations and to semiempirical calculations.1 A number of semiempirical methods have been developed over the last several decades. Indeed.39 In effect.31.1029 In principle. There is inherent error however.33 The CNDO approximation is incorrect in the sense that there is no justification to neglect single-atom differential overlap. Because of the large well-defined physical information content encoded in many theoretical descriptors. In contrast to MNDO. a) and over valence AOs (p.A ) 2 ∑ ∑ (C j m)1 NA A 2 jm) / j ∑S . 83. 225. 225. 198. 244-246 1 1. 13. 65. 79. 245 51. 235. 46 HOMO. 225 43 50. 99. 187 3. 43. 46. 225 πAA. LUMO. πAB ) 4 ∑ ∑∑∑ i a p r A B CpiCA CriCB pa ra i - ∑ a πAA 41. 40. 65 47 qA. 225.A. sum of net atomic charges over all atoms in a molecule polarizability tensor 1. 236 superdelocalizabilities Atom-Atom Polarizabilities self-atom polarizabilities and atom-atom polarizabilities (sum over MOs (i. LUMO HOMO. 42. 80. 225 1. 65. 41. 225 53.A QT.n) 2 LUMO. Qmax QAB ∑q2 A qE.n are the coefficients of the atomic orbital Xn in the LUMO indices of frontier electron density 1. CLUMO.SN. 41.L 43. 189. 8. 82. 232. 50. CHOMO... 42-44.and π-electron densities of the atom A HOMO/LUMO electron densities on the atom A electrophilic atomic frontier electron densities. 192.e. 244. 236 47. 100. 235. 79. 185. ∑S E.π QA.Rxx)2] P) r A)1 ∑ |Q |/N A N .Rzz)2 + (Rzz . 227. 55. R and T stand for reactant and transition state Orbital Electron Densities σ. 191. 9.n are the coefficients of the atomic orbital Xn in the HOMO nucleophilic atomic frontier electron densities. 226. 53.A N.A LUMO . 55.σ. 87.Ryy)2 + (Ryy . 68. 236 7. 39. 227-232. 242. 196. 10. 56 fE ) r fN ) r ∑(C 2 HOMO. 53.r)) sum of self-atom polarizabilities Molecular Polarizabilities molecular polarizability mean polarizability of the molecule anisotropy of the polarizability polarization of the molecule. 236 R R ) 1/3(Rxx + Ryy + Rzz) β2 ) 1/2[(Rxx . 7.A. NA)) sums of electrophilic and nucleophilic 41. 186.A. 42. qA. 225 1. . 42. 65. QA. 9. 98. 46. Quantum-Chemical Descriptorsa definition QA Qmin. qN. 69. B sum of squared charge densities on atoms of type A electrophilic and nucleophilic electronic charges calculated from the occupied and unoccupied orbitals. Q2 T A Qm name Charges net atomic charge on atom A net charges of the most negative and most positive atoms net group charge on atoms A.. 42.A Superdelocalizabilities electrophilic and nucleophilic superdel ocalizabilities (sum over the occupied (E) or unoccupied (N) MO (j) and over the number of valence AO in the atom A (m ) 1. respectively sum of absolute values of the charges of all the atoms in a given molecule or functional group sum of squares of the charges of all the atoms in a given molecule or functional group mean absolute atomic charge (i. 13.1030 Table 1. 188 51. 194. 39. 48. 68. 234-238.H. 41. 55. 100. 7-9. 79-81. 197. 65. 10 42. 55. 185. 193 51.ηT fraction of HOMO/LUMO energies arising from the atomic orbitals of the atom A HOMO and LUMO orbital energy difference absolute hardness activation hardness.HOMO η ) ( LUMO . 195. QA Q2 . 225. 56. 197. 234. 235 9.HOMO)/2 ∆η ) ηR . 40. 190 45 7 1. 41.n) ∑(C FE ) fE/ r r FN ) fN/ r r HOMO LUMO SE. the average of the absolute values of the charges on all atoms) HOMO and LUMO Energies energies of the highest occupied (HOMO) and lowest unoccupied (LUMO) molecular orbitals ref(s) 7. 53. 226. 65. 241. 41-44. 236 1. 41-43. overall electron densities and net charges on atoms are considered as nondirectional reactivity indices. 197.48 According to the frontier molecular orbital theory (FMO) of chemical reactivity.42 Moreover.ET heat of formation relative heat of formation ionization potential electron affinity.9. semiempirical methods are mostly parametrized to reproduce heats of formation.1. Many quantum-chemical descriptors are derived from the partial charge distribution in a molecule or from the electron densities on particular atoms. difference in total energy between the neutral and anion radical species energy of protonation. divided by the squares of the respective interatomic distances) local dipole index. Therefore the calculated atomic charges may be less reliable. the difference between the total energy of the protonated and neutral forms ∆H0 f ∆(∆H0) f IP EA ∆E a For other extensive lists of quantum-chemical descriptors see refs 41 and 225. chargebased descriptors have been widely employed as chemical reactivity indices or as measures of weak intermolecular interactions. It has been shown47 that these orbitals play a major role in governing many chemical reactions and determining electronic band gaps in solids.42.3. 13. the formation of a transition state is due to an interaction between the frontier orbitals (HOMO and LUMO) of reacting species. although none of them corresponds to any directly experimentally measurable quantity.3 The calculated σ. this definition of atomic charge is arbitrary and other definitions are available. Molecular Orbital Energies. In contrast. sum over all connected pairs of atoms quadrupole moment tensor Energies total energy binding energy Ai ref(s) 1.42 and the average absolute atomic charge.45 Other common charge-based descriptors are the most positive and the most negative net atomic charges. 234. 41. DY. 46. Thus. However. the treatment of the frontier molecular orbitals separately from the other orbitals is based on the general principles governing the nature of chemical reactions. Energies of the HOMO (highest occupied molecular orbital) and LUMO (lowest unoccupied molecular orbital) are very popular quantum chemical descriptors (see Table 1). In fact. µhybr µ2 DX. 236 8. 41 41. DZ ∆ name Dipole Moments and Polarity Indices molecular dipole moment charge and hybridization components of the dipole moment square of the molecular dipole moment components of dipole moment along inertia axes submolecular polarity parameter (largest difference in electron charges between two atoms) topological electronic index (sum of the absolute differences in electronic excess charges on all atomic pairs in a given molecule.and π-electron densities on a particular atom also characterize the possible orientation of the chemical interactions and. they are also responsible for the formation of many chargetransfer complexes. 236 65 10. Most modern semiempirical methods use Mulliken population analysis14 for the calculation of the charge distribution in a molecule. and/or geometric characteristics of the molecules. thus. 81 68 Eb ) ∑E i . 66. 236 67 9.qj| 2 rij D) τ ET ∑|Q A. chemical properties of compounds.10. 192. For these reasons the values of atomic charges calculated by different semiempirical methods are in sometimes poor agreement with each other. 50. 10. 186. 9. solute-solvent interactions. 68. 245 9. 244.1031 Table 1 (Continued) definition µ µchar. these numerical quantities are easy to obtain and they give at least a qualitative picture of the charge distribution in a molecule.B N A . e. 226.44 Various sums of absolute or squared values of partial charges have been also used to describe intermolecular interactions.QB|/NAB 42. 67. 42.g. 43 50 48.3.8 Atomic partial charges have been used as static chemical reactivity indices.3 Such calculated net atomic charges are suitable for characterizing interactions according to classical point-charge electrostatic model.i*j ∑ |qi . are often considered to be directional reactivity indices. 69. 88.43 The latter are obtained by subtracting the number of valence electrons belonging to the atom according to the classical valence concepts from the total electron density on the atom. 237. 53. 65. ionization potentials.49 Thus. 236 48 66 TE ) ij.46 Atomic charges are also used for the description of the molecular polarity of molecules. 240 51.49 . A similar idea has been utilized in the definition of superdelocalizability.1032 The energy of the HOMO is directly related to the ionization potential and chracterizes the susceptibility of the molecule toward attack by electrophiles.51 The concept of hard and soft nucleophiles and electrophiles has been also directly related to the relative energy of the HOMO/LUMO orbitals.50.1. however.1 Dipole Moment and Polarity Indices. Self-atom and atom-atom polarizabilities (πAA. the difference in energy between the HOMO and LUMO. while the energy level of the HOMO reflects the relative reactivity of different molecules.52 The HOMO-LUMO gap.e.61-63 Furthermore.64 The first-order polarizability tensor contains information about possible inductive interactions in the molecule. the HOMO density is critical to the charge transfer (electrophilic electron density fE) and in the case of an acceptor molecule r the LUMO density is important (nucleophilic electron density fN). the superdelocalizability calculated based only on the frontier orbital.49 In the case of a donor molecule. charges) describe isolated molecules in their ground state.54 The activation hardness distinguishes between the reaction rates at different sites in the molecule and thus is relevant for predicting orientation effects.3 These indices have been employed in r QSAR studies to describe drug-receptor interaction sites. however.7 Molecular Polarizability. reflects only the global polarity of a molecule. i.3 or to the ability of a reactant to form bonds through charge transfer. the former describing the interactions with the electrophilic center and the latter describing the interactions with the nucleophilic center in the second reactant. The concept of “activation hardness” has been also defined on the basis of the HOMO-LUMO energy gap (see Table 1).226 The total anisotropy of the polarizability (second-order term) characterizes the properties of a molecule as an electron acceptor.50 The total dipole moment.56 Superdelocalizabilities. For example.g. i. the electron density of the r HOMO at an atom is a measure of the relative reactivity of the HOMO at that atom within a single molecule. Frontier orbital electron densities on atoms provide a useful means for the detailed characterization of donor-acceptor interactions.54 Frontier Orbital Densities. thus molecules with smaller ionization potentials (. The superdelocal- izability (Sr) on an atom r is.53 A large HOMO-LUMO gap implies high stability for the molecule in the sense of its lower reactivity in chemical reactions.HOMO) are expected to be more reactive as nucleophiles.55 According to the frontier electron reactivity theory.55 and can be used to compare corresponding atoms in different molecules. For example. related to the contribution made by atom r to the stabilization energy in the formation of a charge-transfer complex with a second molecule.58 This parameter is frequently employed to characterize molecular interactions7.43 This concept.1.3. neglects the electronic reorganization in the excited state and therefore may often lead to conceptually incorrect results. the dynamic indices refer to the transition states of the reactions. Atom-Atom Polarizabilities.10.51. However. to some extent.53. the second-order term is called the first hyperpolarizability.47 The qualitative definition of hardness is closely related to the polarizability. The energy of the LUMO is directly related to the electron affinity and characterizes the susceptibility of the molecule toward attack by nucleophiles.49 To compare the reactivities of different molecules.9. since a decrease of the energy gap usually leads to easier polarization of the molecule.49 Superdelocalizabilities are so-called dynamic reactivity indices. the most significant property of the molecular polarizability is the relation to the molecular bulk or molar volume.226 The most obvious and most often used quantity to describe the polarity is the dipole moment of the molecule. molecular polarity accounts for chromatographic retention on a polar stationary phase. the electronic polarizability of molecules shares common features with the electrophilic superdelocalizability. the majority of chemical reactions take place at the position and in the orientation where overlap of the HOMO and LUMO of the respective reactants can reach a maximum.3 While the static indices (e.65.7.44 If the electrons of the frontier orbital predominate in those interactions. can be used.50.47 The HOMO-LUMO gap has also been used as an approximation to the lowest excitation energy of the molecule. the one-orbital analog of superdelocalizability. The polarization of a molecule by an external electric field is given in terms of the nth order susceptibility tensors of the molecular bulk. hard electrophiles have a high-energy LUMO.53 Polarizability values have been shown to be related to hydrophobicity and thus to other biological activities. Thus. etc. Both the HOMO and the LUMO energies are important in radical reactions.1. Hard nucleophiles have a low-energy HOMO.3. soft nucleophiles have a high-energy HOMO.59 A distinction is made between electrophilic and nucleophilic superdelocalizability (or acceptor and donor superdelocalizability respectively).7 These quantities are defined on the basis of perturbation theory and merely represent the effect of a perturbation at one atom on the electronic charge at the same (πAA) or a different atom (πAB). .43. frontier electron densities can strictly be used only to describe the reactivity of different atoms in the same molecule.60 The first-order term is referred to as the polarizability.57 which is an index of the reactivity of occupied and unoccupied orbitals (see Table 1).e. frontier electron densities have to be normalized by the energy of the corresponding frontier molecular orbitals: FE ) fE/ HOMO. πAB) have also been employed to describe chemical reactivity. FN ) r r r fN/ LUMO.10.43. The polarity of a molecule is well known to be important for various physicochemical properties and many descriptors have been proposed to quantify the polarity effects. and soft electrophiles have a low-energy LUMO.47.3. is an important stability index.43 Soft interactions are influenced by the ability to accept or donate electron density (superdelocalizability) through orbital rather than electrostatic charge transfer.49. such tensors depend on the choice of the coordinate system and therefore the orientation of the molecular root fragment must be the same for all molecules in the series.48. such as molecular polarizability and size-shape descriptors. and substituent reactivity indices (superdelocalizabilities). However.50 vectors of lone pair densities. (b) descriptors describing the bulk properties of the substituents (Verloop’s steric parameters76. dispersion interaction terms. F ) 109.76. and the HOMO/ LUMO energy difference.41 IV.225 According to their conclusions.8.1033 Local polarities can be represented by local dipole moments. s ) 0. F ) 123. the observed PCA clustering of 66 descriptors derived from the AM1 calculations was similar to that previously reported for the monosubstituted benzene series.78 n ) 28.8. while descriptors characterizing the molecule as a whole. The conclusions from this analysis41 were as follows: (a) Substituent induced chemical shifts. In particular. the stronger the bond) and can be used to determine the correct localization of the most favorable hydrogen bond acceptor site. The main contributions to the first component are from the net atomic charges on the carbon atoms. such as net atomic charges or frontier orbital electron densities.41 The authors also noted that the descriptors are well spread in the principal component space. s ) 0.73. are often preferrable for congeneric series.50 Energy. the anisotropy of the molecular polarizability. for example. (b) The molecular refractivity73 and Verloop steric parameters76.48 and the topological electronic index. The total energy calculated by semiempirical methods has been shown to be a good descriptor in a number of different cases. frontier orbital electron densities.77 together with the molecular weight and substituent van der Waals volumes are modeled by the second component to which contributions are made mainly by the polarizabilities.66 The reaction enthalpy can be accounted for by the difference in heats of formation between reactants and products or between conjugated species.909.74qO + 64.41 According to the PCA.899.84qSO2NH2 + 8.66 The quadrupole moment tensor can be also used as an index to describe possible electrostatic interactions. the dipole moments.9. and their squared terms. can be considered as a good measure of the strength of hydrogen bonds (the higher the energy.364. the only previous systematic comparison of quantum-chemical (AM1) descriptors with both empirical and experimental descriptors was made by De Benedetti and co-workers.72 analysis was performed to establish whether the theoretical molecular descriptors bear the same information content as the so-called physicochemical parameters.55 To our knowledge. see.78 The descriptors which contribute to this component are the solvent accessible surfaces of the substituents.69 overlap populations.70 the substituent descriptors considered clustered into three main groups: (a) descriptors recording the substituent effects on the aromatic ring (net atomic charges. The following charge based polarity indices have been proposed (see Table 1): the local dipole index. net atomic charges.77 and Taft and SwainLupton resonance constants73.74 n ) 28.336. the total energy has been used as a measure of nonspecific interactions between a solute and stationary phase in gas-chromatography.50 and free valence of atoms. and (c) theoretical and experimental molecular and substituent dipole moments and their square terms. have wider applicability. Biological Activities Quantum-chemical descriptors have been long used in quantitative structure-activity relationship studies in biochemistry.67 For example. resonance and field substituent constants.41 Fifty theoretical (quantum-chemical and others) descriptors calculated for 50 monosubstituted benzenes were analyzed by the principal component analysis (PCA) and partial least-squares method (PLS). and 12. and the electrophilic superdelocalizability of the substituents (the hydrophobicity parameter73 is also close to this cluster).2 log Π50 ) 119. compared the performance of quantum-chemical descriptors to model binding affinities of congeneric and noncongeneric R1-adrenoceptor antagonists.74 are modeled by the first component. The following descriptors have also been proposed but do not fall into the above-mentioned categories: atomic orbital electron populations. The descriptors discussed above constitute the majority of quantum chemical descriptors published and successfully used in QSAR/QSPR studies.50 partitioning of energy data into one-center and twocenter terms.245 The energy of protonation. In a recent study Cocchi et al.3 where qSO2NH2 is the charge of the -SO2NH2 group . PLS71. MO-derived indices describing atom or local fragment properties.43.75 Hammett substituent constants. the electrophilic superdelocalizability of the benzene ring.43 the differences between net charges on atoms. HOMO-LUMO energies. and superdelocalizabilities have been shown to correlate with various biological activities. refs 3-6. and the energy of the frontier molecular orbitals.77 and the molecular refractivity73) clustered together with theoretical descriptors defining the polarizability properties of substituents. Linear regression analysis has provided good correlations between the CNDO/2 calculated total net (σ + π) atomic and group charges and anhydrase inhibition by heterocyclic sulfonamides.79 log Π50 ) 37. r ) 0. but these are conceptually difficult to define. defined as the difference between the total energies of the protonated and neutral forms of the molecule. (c) The third component models the liphophilic73 and lipophobic parameters.68 Others.244.225 Significantly.74 and substituentinduced chemical shifts75). dispersion forces. There are several extensive reviews on QSAR studies. First approximations of these quantities can be obtained by considering the atomic charges in the localized regions of the molecule. r ) 0. QSAR/QSPR Results A. 57π8 .0.02 n ) 37.17. F ) 11. the 2. r ) 0.931) by a two parameter equation involving logP and either the HOMO energy or the electron density on the sp3hybridized N in the HOMO.91.17 .245.55 For example. and frontier orbital energies. However.17.80. CNDO/2 calculated charges. s ) 1. the best correlation was found using AM1 calculated net atomic charges on carbon atoms (q2) and the hydrophobic parameter (log P). whereas binding free energies are well described by the combination of molecular volume with various frontier orbital parameters. and π are physicochemical parameters of the corresponding substituents.229 and polycyclic aromatic nitro compounds.4 HOMO . In general.228. and 4-mono derivatives of 1.919 and r ) 0.0.65.14 log P .231 Mutagenic activities of aryltriazenes and heterocyclic triazenes were found to be equally well correlated (r ) 0. the following equation employs the LUMO energy and the total π-electron energy (ETπ): log(1/IC50) ) 0.2.244 Analogous relationships were observed for aromatic and heteroaromatic amines.232 Quantitative analysis of the antirhinoviral activity of 9-benzylpurines produced correlation equations involving Huckel MO generated electronic param¨ eters and physicochemical properties of substituents.289πR4 .85 where π is the Hansch hydrophobic constant. 3-.53EA .4840Qmax .54I13 .91.84 For substituted toluenes.873FR4 .230.56π .39 n ) 19.39q3 + 5.3. The pharmacological activities of the 2-.234 Another QSAR study of a series of 8-substituted xanthines provided correlation equations between classical and MNDO calculated parameters and the affinity toward A1 and A2 adenosine receptors:44 pKi(A1) ) 0. r ) 0. s2 ) 0. e.40 n ) 5. s ) 0.503.094ETπ . in which Ames TA98 and TA100 mutagenecities correlate linearly with the HOMO and LUMO energies of the amine. F ) 35.714.208 n ) 19.056RR2 + 0.958.41π1 + 61.00 where R.59.21.90 n ) 38.309 The correlations involving other AM1 electronic parameters of chlorofuranones were inferior.45. r ) 0. frontier orbital densities.8882 only.0418Ly + 2.e. but most often it is almost entirely masked by the hydrophobicity.5.and 3-positions of the phenyl ring are affected by electronic parameters.4-dihydropyridine and parasubstituted toluenes have been correlated with classical and various quantum-chemical (AM1) parameters such as net atomic charges.23 LUMO . it was determined that various serotypes of rhinovirus behave differently in terms of the electronic parameters that inhibit their action.32. s2 ) 0. F ) 37.66π8 .0.043. s ) 0.565. in the form of the charge at the most highly charged atom in the molecule.47I7 + 0.245 In another study.66 g/mL).66 pKi(A2) ) 0.40B1.246 A good linear correlation was found between the calculated LUMO energy of the nitrogen atom of the nitro group and the logarithm of the mutagenicities of nitroarenes. r2 ) 0.234 In the case of dihydropyridine derivatives. the stability of the corresponding anion radical) of the molecules log A ) 0. the correlation with the frontier electron densities of the LUMO at the R-carbon had r ) 0. r ) 0.852. and sizes. F ) 20.244.76q2 . F ) 14. F.9556 ln(TA100) ) -14. r ) 0.49Lm .01 where Ly is the length of an ellipsoid of uniform density (1.044 + 2.2.39 n ) 21. dipole moment. s ) 0.61.4 n ) 35. such as the HOMO energy and greatest populations in the HOMO and LUMO for methyl group hydrogen atoms.827.21q1 . Significant correlations were found51. and B1 and L are the Verloop sterimol parameters. F ) 19.13. r ) 0.p . the participation of frontier orbitals in mutagenic or carcinogenic activity seems to be essential. and the correlation analysis resulted in the following equation: ln(TA100) ) 12. the carcinogenic and mutagenic potencies of benzanthracenes were shown to correlate with LUMO energies. In a QSAR study on the mutagenicity of quinolines.81. although the HOMO and LUMO energies and electron densities were also considered:82 -log IC50(1B) ) 6.06I7 + 1.8 + 45.57SN.0. r ) 0.0.35. with the same principal moments of inertia as the molecule.81 between Ames TA100 mutagenicity and both the electron affinities (EA) or LUMO energies (i.1034 and qO is the charge on the oxygen atom of the SO2 group. in the form of the smallest dimension of the molecule.83. s ) 0. The correlations observed suggest a reaction mechanism in which chlorofuranones act as electron acceptors in the interaction with DNA. have both been shown to be important descriptors in quantitative structureactivity relationships of some arylalkylamine and arylalkylamino acid activators of carbonic anhydrase:42 log TA100 ) 1. it was found that the rate constant of hydroxylation can be described by a twoparameter expression involving the dipole moment.3.83 aromatic and heteroaromatic nitro compounds.9 The involvement of the net atomic charge on carbon atom number 2 (q2) suggests that the 2-position on the quinoline ring is the site of activation.g.323 LUMO n ) 50. 85 drug-receptor interactions are under either charge or orbital control.1035 where π is the substituent hydrophobicity parameter. r ) 0.04 n ) 7. According to the polyelectronic perturbation theory of Klopman and Hudson. which suggests that the enzymesubstrate interaction may be of the second order.1. led to chance correlations.43 n ) 13.79.21.046.42R(1) + 2.88. quantum-chemical parameters can easily be misused. F ) 30. e.19.162.25R/∆E + 0. Thus. the dipole moment does not correlate with the activity for this series. r ) 0.52 n ) 16. s ) 0. F ) 24.87.65µ + 1. F ) 42.431∑q0 . and acute toxicity) in the series of 20 nitriles:53 N The average acceptor superdelocalizability Sav was N obtained by averaging the sum of Si over the atoms involved in the π bonds.56 The acute toxicity of soft electrophiles such as substituted benzenes. F ) 15.18R(1)2 . SN is the donor superdelocalizability.58 A significant correlation was obtained between the inhibition potency of indanone-benzylpiperidine inhibitors of acetylcholinesterase and the MNDO HOMO energy:69 -log(IC50) ) 2.985. s ) 0.91 n ) 17.7SN av n ) 114. r ) 0. I is the substituent indicator variable. r ) 0. and 243.8 where C4 is the HOMO out-of-plane π-orbital coefficient of the ring carbon atom.4 The R/∆E parameter is an orbital energy weighted polarizability term and thus implies that the acute toxicity of nitriles is a function of molecular size/ polarity and electronic activation energy.227 n ) 20. rate of oxidative metabolism. F ) 387.94. the net atomic charges characterize electrostatic interactions.58 2 HOMO . F ) 14. r ) 0. which is equivalent to the ionization potential in MNDOPM3 calculations.g. positive charge on the pyridinium ring. An index of frontier orbital electron density derived from semiempirical molecular orbital calculations (MNDO-PM3) was found to correlate with fungicidal activity of ∆3-1.978.3 log EtOH/glucose ) -0. while the donor superdelocalizability characterizes the covalent component of the interaction.6.249 -log Ki ) -1. In some cases.887∆(∆H0) .2.18µ2 .76.81.47 n ) 13. such as hydrophobicity or a Hammett constant. 12. s ) 0.HOMO × 102) was derived from the HOMO electron density at the sulfur atom (fr(1)) and the HOMO energy of the molecule ( HOMO) in electronvolts measured from the zero level. F ) 39. r ) 0.1 -log LD50 ) -1. r2 ) 0.247.14R(1) .5 -log LD50 ) -0. s ) 0.915 ∆(∆G0) ) 0.87. e.17.891 n ) 14. r ) 0. and with the differences in heats of formation between the acids and their anions. see refs 3-6.58 B.199. F ) 261.56 log P + 13.328 HOMO + 81. and anisotropic polarizability.8 log(1/LC50) ) -1.396 f n ) 14.46 pEC50 ) 2.44 The CNDO/2 calculated molecular polarizability (R) and the HOMO/LUMO energy difference (∆E) have been successfully correlated with a number of biological activities (ethanol inhibition. s ) 0.0. s2 ) 0.248 or misinterpreted.006R + 0.1 Overall.53 However.981. F ) 238.76 An index R(1) ) fr(1)/(.939.757.8.2.199.g.13 ∆(∆G0) ) -316. s ) 0. with the energies of the highest occupied molecular orbital (HOMO) of the corresponding benzoate anions. s ) 0. r ) 0. r ) 0.95 n ) 13. F ) 302. s ) 0. µ is the total dipole moment.56 pEC50 ) 0.217 n ) 14. s ) 0. However.25.8.7 .235 The dissociation constants of β-adrenergic active compounds were found to be governed mainly by the CNDO/2 calculated total anisotropy of polarizability which characterizes the properties of the molecule as an electron acceptor. a quantum-chemical descriptor has appeared to be important due to a high correlation with a conventional descriptor.9 HOMO .03R + 0. and anilines has been correlated with MNDO calculated descriptors of soft electrophilicity for aromatics: average superdelocalizability and LUMO energy.853 ∆(∆G0) ) 17.4-thiadiazolines. r ) 0. The hydrophobicity (log P) and soft electrophilicity descriptors were shown to be orthogonal for the 114 compounds studied. the importance of quantum-chemical descriptors in QSAR studies is well recognized.69R/∆E + 0.758 Analogous relationships were found in a related study: the AM1 calculated energy differences between the acids and their conjugate bases and the anion HOMO energies correlate satisfactorily with the experimental (condensed phase) acidity of phenols and of aromatic and aliphatic carboxylic acids.0.359.8.21C4 .49 + 0. Chemical Reactivities The gas-phase acidity of substituted benzoic acids was related linearly with the AM1 calculated net charges on the oxygen atoms. s ) 0. and q is the net atomic charge.282. F ) 42. whereas the latter would be more appropriately used on other considerations. phenols. The inhibition potency of charged sulfonamide inhibitors of carbonic anhydrase has also been correlated with the HOMO energy. 1036 The best correlation employed four descriptors and included calculated atomic charge densities.1. and amido groups.2.863 It was proposed that superdelocalizability apparently includes both electrostatic and perturbational effects.090 + 1. r2 ) 0. NN. The equation employs sums of squared atomic charge D.868 σ ) 2.C n ) 25 (substituted benzenes). the sums of absolute values of atomic charges and the charge dispersions. amines.051 + 7.92Q2 . esters.119NN . defined as the difference between the absolute hardness of the reactant and transition state. 0.16Hf + 0. the density of electrons at an atom and a measure of their instability. alcohols.4.551Q4 + N O 17.47 Activation hardness. was used to predict successfully orientations in electrophilic aromatic substitutions. and oxygen atoms.13.167 × 10-4S2 . which includes molecular mass. and QO is the square root of the sum of the squared charges on oxygen atoms.4.O1 n ) 19 (benzoic acids). energies of HOMO and LUMO. s ) 1.599 .417SN.1 where S is the molecular surface. nitrogen.273 O n ) 118.591SE. QN is the square root of the sum of the squared charges on nitrogen atoms. NT. The method yields better results than the fragment approach and requires fewer parameters.093NC . r2 ) 0. charge on the oxygen atom.4.7 σ ) 8.15 where NH.57 × 10-3Mw . and amides.389∑q2 C where d1 and d11 are the atomic charges on acidic oxygen atoms in the acid and anion. polarizability. r ) 0.9 Ialkane is the indicator variable for alkanes. Partition Coefficients The octanol/water partition coefficient (log P) is the standard quantity to characterize the hydrophobicity/ hydrophilicity of a molecule. a property of major importance in biomedical applications.17Q4 + N 26. D is the calculated dipole moment. s ) 0.e. NC. for example ref 86.864 σ ) 3.1. and energy parameters. The activation hardness. melting point.9. and polarizability.9986Ialkane + 9.106 × 10-2S + 14. molecular volume.7.910NT + N O 1. respectively.48. calculated from Huckel ¨ MO theory (see Table 1).929QON .6.226.344 + 0.10 These quantum . Over the years a number of procedures have been proposed for calculating partition coefficents from the molecular structure. ethers.13D .237 Superdelocalizabilities.937NO .28∑q2 + 0. and QT2) parameters. refractivity.2078NH + pKa ) 33.67O + 0.985.709NM n ) 61.12 n ) 183. The natural logarithm of the rate of glutathione Stransferase-catalyzed conjugation of the series of fluoronitrobenzenes also correlates with the LUMO energy and with the relative heat of formation of the Meisenheimer complex. are minimized. Atomic charge densities has been proposed as the basis for calculating octanol/water partition coefficients. O is the ovality of the molecule.565SE. and NA. qC.88.241 C.45 The method is based on the linear correlation equation obtained for the set of 61 compounds including hydrocarbons.500SE. In general.12.812 + 2. cyano.81Q2 . and qO are the corresponding atomic charges.87O2 .and bifunctional molecules on four stationary phases of different polarity were best described by a combination of topological and quantum chemical (QA. Mw is the molecular weight. nitriles. see. give good predictive correlations of the Hammett constants of aromatic compounds. r ) 0.03QO + 27.01d1 + 0.43. and NO are the number of hydrogen. the above-mentioned approach has been extended by involving more molecular descriptors including the calculated dipole moment. which has been expressed in terms of dipole moment. Chromatographic Retention Indexes and Response Factors It is generally accepted that chromatographic retention of a compound on a polar stationary phase depends on the polarity of the compound. The predictive power of the model has been demonstrated by the accurate estimation of log P for complex molecules.9388. is an excellent index for predicting orientation effects. calculated by the CNDO/2 method.1.0. r ) 0. QON is the sum of absolute values of atomic charges on nitrogen an oxygen atoms.66. the reaction coordinate is such that changes in the HOMOLUMO gap. densities calculated by the MINDO/3 procedure: log P ) 0. ketones. r ) 0.844NA + 0. i.N n ) 19 (phenyl amines).74d11 . and NM are the numbers of carboxy. acids. F ) 115. or hardness. QA2. log P ) -1. carbon. In another study. A model has been derived to estimate log P of substituted phenols.416QN . qN. net atomic charges.900 σ ) 1.87 All quantum chemical parameters were calculated using the AM1 procedure and regression equations were extended to a larger set of partition coefficient data. r ) 0.01 HOMO 17.C n ) 13 (simple aromatics).227 The gas chromatographic retention indexes of 43 mono.7316∑q2 + 2. total surface area. 39ET .90-96 there are very few papers employing quantum chemical descriptors for this purpose.6. r ) 0. and polychlorophenols.0.48 The equation involves the CNDO/2 calculated total energy (ET) and the originally proposed submolecular polarity parameter (∆.96RWCeff .33 + 10.5 .55VH. RF ) -2. RWCeff is the relative weight of “effective” C atoms.80∆ n ) 22. IOV-101 ) 301.min + 0.+ 0. s ) 35. s ) 67.66 This index is calculated as the sum of the absolute differences in excess electronic charges on all atomic pairs in a given molecule.515 .230A+ + 8.88 Good six-parameter correlations have been obtained with nonpolar stationary phase retention times and response factors for a set of 152 diverse organic compounds (R2 ) 0.87ST/N + 0.93. Table 1). s ) 0.min .1. polymethylphenols. This equation was derived from the theoretical principle that many bulk properties of liquids are driven by the intermolecular interactions which can be expressed through the molecular surface energy.1.1016.min is the minimum valency of a H atom. provides a better regression equation for the retention data of a diverse group of aliphatic and heterocyclic amines. s ) 0.32VH.9NCH . QA2) and global polarities (QT2) of the molecules. it was shown that these parameters are more closely related to the retention indexes on polar stationary phases than simple or squared dipole moments.88 .min is the minimal total bond order of a C atom. while the submolecular polarity parameter is a measure of the solute’s ability to take part in polar solute-stationary phase interactions.66ET .829.67 The energies ET and Eb had been calculated by CNDO/2 method.21. AHB is the sum of the surface areas of hydrogen-bonding hydrogen atoms multiplied by their corresponding scaled net atomic charge.6 .89 The response factors are correlated by the minimum total bond order of a carbon atom and the total molecular one-center electron-electron repulsion.015A. The topological electronic index. As the polarity of the stationary phase increases.13. ST/N is the total entropy of the molecule at 300 K divided by the number of atoms.16RNCeff .11.97 The latter can be presented as a sum of dispersion tR ) 26. in combination with the total energy.89 BP ) 127. The most significant quantumchemical descriptors correlated with the retention times are the molecular polarizability (characterizing molecular bulk and dispersion interaction with the media) and the minimum valency of a hydrogen atom (related to hydrogen bonding interactions). AOPmax is the maximum atomic orbital electronic population.718A .054 where NCH is the relative number of C-H bonds. F ) 745. To characterize differences in polar properties of molecules. An approach for the estimation of several physical properties of organic compounds (critical temperature.29 The ionization potential.0. The total energy (ET) or binding energy (Eb) for a homologous series of esters were both shown to give straight line correlations with the gas chromatographic retention indexes for both nonpolar and polar stationary phases. s ) 14.66 E. calculated by MOPAC (AM1). µhybr is the total hybridization component of the molecular dipole. VH. R2 ) 0.03µhybr n ) 152.959 and R2 ) 0. r ) 0. A+ is the sum of the surface areas of positively charged atoms multiplied by their corresponding scaled net atomic charge. respectively) using a combination of conventional and AM1 calculated quantum-chemical descriptors. a greater differentiation between the homologous series appears.892.979.0. R2 ) 0. By means of factor analysis. RNCeff is the relative number of “effective” C atoms.45 According to these authors.981.019Mw . boiling point) based on the computation of the molecular surface interactions (MSI) has been proposed. molar critical volume.min + 3. Other Physicochemical Properties Although numerous attempts have been made to correlate physical properties of organic compounds (particularly boiling point) with structural parameters. the following correlation equation has been proposed for the boiling point: IOV-101 ) 160.959.1037 chemical parameters represent different sums of CNDO/2 calculated atomic charges (see Table 1) and account for both the local (QA. and bC. the total energy represents a bulk measure of the solute’s ability to participate in nonspecific interactions with the stationary phase. a two-parameter regression equation was derived which satisfactorily describes the retention of structurally different polar solutes on a relatively non-polar stationary phases.149. A.93AOPmax n ) 152. see Table 1). In another study.is the sum of the surface areas of negatively charged atoms multiplied by their corresponding scaled net atomic charge.046R + 0. For instance. divided by the squares of the respective interatomic distances.800AHB n ) 137. a topological electronic index TE has been proposed (cf.6TE n ) 22.1 where A is the total molecular surface area.0. was found to be the most suitable property to adjust the capacity ratios (liquid chromatography retention) of polychlorobenzenes.97 This approach employs MSI descriptors calculated from atomic surface areas and EHT net atomic charges.7 + 0. R is the polarizability.21bC. ET is the total moEE lecular one-center electron-electron repulsion energy. Mw is the molecular weight.0003ET EE 1. r ) 0. which reflect the inclination of the thermally cracked products to undergo “chemiionization” in the flame ionization detector of the chromatograph. F ) 214. while polar interactions were represented by dipole moment components in these correlations. Politzer et al. SFHA is the fractional hydrogen acceptors surface area. n ) 18) for σ was with the calculated electronic charge on the oxygen of the anion (qO-).12 n ) 18 These correlations indicate that the Hammett and Taft constants for substituted benzenes and also the equilibrium constants for the ionization of substituted benzoic acids are controlled by the anion structure. results similar to mentioned above were obtained in independent correlations of boiling points.105-110 and we therefore consider here only those studies of direct correlations of empirical substituent constants (σ.278Gi + 19. the following six parameter correlation equation was derived for boiling points: σo also regressed well against the same two parameters although the preferred parameter was 1/ HOMO: σo ) 34.4 + 0. and hr.115 The most significant correlation (r ) 0.can be considered as the “enthalpic” descriptors and therefore. as the ionization equilibria of carboxylic acids is known to be an isoentropic reaction series.116 The correlations obtained are remarkably good. An analogous result was obtained using ab initio calculations with STO-3G basis sets augmented by diffuse functions for all oxygen atoms.4f r. For example. σI. except the gravitation index.min on I I a ring carbon atom of a monosubstituted benzene and Hammett σ-constant. in terms of the quantum-chemical characteristics of the molecules has been of substantial theoretical and pragmatic interest.117 Good correlations of the inductive substituent constants with more elaborate enthalpic (energetic) descriptors have also been reported.946 S ) 0. SRN is the relative negatively charged surface area.113 constants were examined using the MNDO114 quantum-chemical characteristics of a series of benzoic acids and benzoate anions.min r ) 0.119 invented the average local ionization energy (I(r)) of a molecule h as useful characterstics of atomic energetic properties in molecule.min for the I conjugated bases of substituted acetic acids and the inductive substituent constants σI:120 BP ) .) with quantum-chemical characteristics of the corresponding substituents or molecules. The operational value of such correlations lies in the possible extension of known correlations involving conventional electronic substituent constants to compounds for which such constants are unknown or difficult to define. G.min . Notably. and therefore it has been of the highest interest to develop quantitative structure property/activity relationships which reflect intermolecular interactions in dense media. polar interactions (related to the first-order term of the electrostatic interactions A. resonance. Charged surface areas were calculated from net atomic charges derived by AM1. Substituent Constants The theoretical analysis of various empirical substituent constants related to intramolecular inductive. etc.2SFHA + 136.87 r ) 0. s ) 14.of benzoate anions correlate even better with a somewhat different set of Hammett σ-constants.247.1038 interactions (proportional to the molecular surface area A). despite numerous attempts to ascribe definite physical meanings to empirical substituent constants through secondary (multi)linear correlation analysis with quantum-chemically derived molecular descriptors. bN.6467 + 0. melting points.17.N hN n ) 85.62SRN + 2503. n ) 18) was observed with the reciprocal of the highest occupied molecular orbital energy (1/ HOMO) of the corresponding substituted benzoate anion. P. involving descriptors corresponding (1) . AM1 calculated partial charges on the oxygen atom qO.and A+). also discussion in section V). Notably. Already Jaffe101-103 found that the values of Hammett σ-constant were linearly related with the π-electron densities on the respective substituents. and flash points of substituted pyridines. S ) 0.25µchar + 1713.120 A good relationship was also demonstrated to exist between the hs. considering the neglect of solvation effects which are known to be of considerable importance in acid-base equilibria. S ) 0. the correlation of the pKa with this type of molecular characteristics is not completely unexpected. All the descriptors involved. from electrostatic considerations. and hydrogen bonding interactions (AHB). Remarkably. or steric effects. An excellent correlation was found between the lowest of h(r) value.19. µchar is the total point-charge component of the molecular dipole. It was defined as the average energy required to remove an electron from any point r or region dr in the molecule.899.5. The Hammett σ 111 and Taft σo 112.min is the minimal total bond order of a nitrogen atom.921.5 where Gi is the gravitation index.11 Hydrogen-bonded related surface areas of the molecules were found to correlate with all three properties.4962Is.N is the average nucleophilic reactivity fN index for a nitrogen atom. it was not surprising that h σI ) -3.1bN. hs.108 However. were calculated by the AM1 method. such QSAR/QSPR correlation equations are usually multiparametric. Since σ and σo are well correlated with each other.118.38(1/ HOMO) + 7. R2 ) 0.26. Another linear relationship was reported early on between the σ-constants and π-electron localization energies. these empirical constants may possess substantial entropic components and thus in principle cannot be described solely by essentially energetic quantum-chemical descriptors (cf.104 The development of the concept of the electronic substituent effects has been extensively reviewed. Because of the complexity of solvent effects. F.943. a comprehensive theoretical treatment of these constants is still absent. both HOMO and qO. Solvational Characteristics The solvent effects play a key role in many chemical and physical processes in solution.97 In conclusion. A nearly equally good linear correlation for σ (r ) 0. how molecular structures. However.136-139 Also. The electrostatic basicity contribution. it is assumed that the solvent does not change significantly the geometrical and electronic structure of the molecules. as discussed above. hydrolysis rates of organophosphorus compounds.133. is simply the most negative atomic charge in the solute molecules. it has been observed that in numerous cases this assumption is no longer valid. TLSER regressions have been developed for over 100 solute/solvent-based properties. supercritical carbon dioxide solutions.127 and the polarizability term π* is derived from the polarization volume computed by the method of Kurtz. The linear solvation energy relationship (LSER) descriptors based on LFER (linear free energy relationships) are demonstrated to be successful in correlating a wide range of chemical and physical properties involving solute-solvent interactions as well as biological activities of compounds.239 and others. However. spectral.130 six physicochemical properties. that is. whereas q+ is the most positive charge of a hydrogen atom in the solute molecule. and pKa of organic acids. b. than does LSER. permitting a much greater degree of a priori prediction once a correlation is derived. These properties. by the quantum-chemical calculation of the solute molecule together with the first solvent coordination sphere or with the . polarizability. the predominant structure of a tautomeric molecule may be altered in different solvents and in the gas phase. and by increasing the ease in which the parameters can be computed. how changes in molecular structures affect the observed property. The most general form of TLSER is as follows126 log(γ) ) co + c1Vmc + c2π* + c3 a + c4 b + c5q+ + c6q- where Vmc is the molecular van der Waals volume calculated according to the method of Hopfinger. identified. it is not necessary to synthesize a new compound and measure specified descriptors in order to generate a prediction.132 acute toxicities.e. The original LSER descriptors (also called the solvatochromic descriptors) were derived from UVvis spectral shifts of indicator dyes. or more accurately. The specific. In this way. (2) to the solvent’s ability to act as a hydrogen bond acceptor or donor. TLSER uses a single set of descriptors (six for the TLSER) and each parameter describes a single. orthogonal molecular event or characteristics. is represented as the difference in energy between the lowest unoccupied molecular orbital ( LUMO) of water and the highest occupied molecular orbital ( HOMO) of solute. each of these terms has been derived from semiempirical molecular orbital methods. because TLSER descriptors are computed.135.134 Several TLSER correlations have been reported for the chemical and physical properties of substances in solution.128 The hydrogen-bonding effects are separated into donor and acceptor components.131 gasphase acidities. the resulting correlations relate an empirical (macroscopic) property to molecular (microscopic) parameters. and hydrogen bonding terms. Second.123 The coefficients of the descriptors in the correlation equation are expected to provide insight into the physical nature of the solute-solvent interactions related to the experimentally observed plenomena or data. it is important to apply in these cases the quantum-chemical descriptors.129 opiate receptor activities of some fenantyl-like compounds. called the theoretical LSER (TLSER). because the parameters are orthogonal. and toxicological properties. including adsorption on charcoal. In particular. chemical.140-148 Therefore. and rationalized. Attempts to correlate computationally derived structural and electronic descriptors with the solvatochromic parameters have met with only moderate success. Analogously. can be deduced.222-224 In LSER.126 activities of some local anesthetics. Further. the log P hydrophobicity parameter discussed above can also be considered as a solvational characteristic since it is directly related to the change of the free energy of solvation of a solute between two solvents (water and octanol). denoted as q-. Although there exist tables of LSER parameters and predictive relationships to help in their estimation. Hafkensheid retention indices. Last but not least. and (3) to short-range dispersion and repulsion interactions. hydrogen-bonding or solute-solvent charge-transfer effects on the solute molecular structure can be usually satisfactorily accounted for by using the supermolecule approach.125. runs the gamut of physical. octanol/water partition coefficient. The advantage of using this single set of descriptors has been demonstrated in the ability to compare disparate properties and data sets. The covalent contribution to Lewis basicity. Like LSER. The TLSER descriptors have led to good correlations and physical interpretations for variety of biological activities of chemical compounds: nonspecific toxicities. i. Thus their ability to make a priori predictions has been somewhat limited because of their empirical origin.121 Indeed. LSER descriptors for new materials are not easily defined.1039 to the polarity and the polarizability of the solvent. First. they incorporate steric. its ability to calculate easily the parameters of almost any chemical species significantly increases the correlative ability of this method by increasing the number of compounds that can be used in the data set. calculated by using one of the available methods to account for the solvent effects. TLSER has been demonstrated to be useful in three regards.122. Famini and coworkers have developed a new set of computationally derived descriptors. the intramolecular resonance effect can be substantially affected by the different dielectric media.126 The TLSER attempts to maintain the same relationship between property and parameters.124 In an attempt to circumvent the problems associated with the LSER parameters. problems of cross correlation between independent variables are usually minimized. like the LSER before it. the hydrogen-bonding donating ability is divided into two components: a is the energy difference between the HOMO of water and LUMO of solute. The CoMFA approach is based on the assumption that since most drug-receptor interactions are noncovalent. CoMFA Since its introduction in 1988. we feel it appropriate to include a short overview of this methodology in the present review.200 Gaussian92201) have an option to calculate the vibrational. Different schemes (arithmetic average. Usually a sp3 carbon atom with a positive unit charge (+1) is moved on a rectangular grid that encompasses the aligned molecules. it should be kept in mind that a quantum chemical calculation is performed for a single structure at an energetic minimum.171-177 and carcinogenic and toxicological properties of compounds. most standard quantum-chemical program packages (AMPAC. However. In most cases the quantum-chemically derived descriptors have definite physical meaning. the quantum chemical descriptors cannot in principle account for entropic and temperature effects. Whereas examples of the use of those theoretically calculated thermodynamic molecular characteristics in QSAR/QSPR are practically unknown. the zero-point vibrations of the molecule are neglected.199 MOPAC. organic. H. and translational partition functions of the molecule at the given temperature and their respective contributions to the molecular enthalpy.150 comparative molecular field analysis (CoMFA) has become rapidly one of the most powerful tools for three-dimensional quantitative structure-activity relationship (3DQSAR) studies. however. It should be noted that the q2 is different from the crossvalidated correlation coefficient in multilinear regression and a q2 > 0. this is to be encouraged in future studies of thermodynamically complex and presumably temperature-dependent properties of compounds.184 V.and intermolecular interaction mechanisms determining the molecular property or chemical process under study.152 A CoMFA QSAR table of thousands of columns is formed thereafter from the numerical values of the fields at each grid point which is subsequently analyzed using special multivariate statistical analysis procedures.163-170 antiviral activities. quantumchemical descriptors can be derived solely from the theoretical structure of the molecule.3 is already considered significant. Also. for the conformationally flexible molecules with several energetically close conformational minima. as most chemical reactions and all biochemical reactions refer to condensed (mostly liquid) media.151-169 Because the molecular conformation optimization and charge density calculations to obtain the molecular field are often performed by using quantum-chemical methods.157 A cross-validated R2 (q2) obtained as a result of this analysis serves as a quantitative measure of the predictability of the final CoMFA model. When such effects are dominant for a given property or process. all molecules under investigation are first structurally aligned and the steric electrostatic fields around them sampled with probe atoms. Also. AM1. the use of descriptors applying to a singular conformation is required. in contrast to empirical substituent and solvent effect constants. The fields arising from the charge distribution on the frontier molecular orbitals (HOMO-s) have also been suggested for CoMFA analysis.156 and cross-validation. This enables applications of QSAR/QSPR correlation equations involving quantum chemical descriptors to hypothetical structures that have been never synthesized or are otherwise unavailable. Boltzmann average) could be used for this purpose. In such cases. a preliminary averaging of the molecular descriptors is therefore advisable. Finally. such as partial leastsquares (PLS) analysis155. it has also been applied for the description of the chemical reactivity of compounds. particularly in the studies of biological activity of the compounds. it should be advantageous to use molecular descriptors calculated using some quantum-chemical scheme which accounts for specific and nonspecific . Thus. and biomedical chemistry. rotational. and other thermodynamic functions. quantum-chemical descriptors are not adequate for their description and any correlation obtained using them can be regarded as accidental. MNDO. only one conformation of the compound may be active in the given process. changes in the biological activity of compounds should correlate with the steric and electrostatic fields of these molecules. In order to develop the numerical representation of those fields.179 The CoMFA approach has been mostly used in biomedical QSAR studies.180-183 Notably.1040 solvent molecules attached to the hydrogen-bonding centers of the solute. Therefore.159 CoMFA has been used for the quantitative description of enzyme inhibition activities of compounds.178. analytical. may have serious drawbacks. and thus they have proven to be especially useful in the clarification of the detailed intra. and PM3 calculated Mulliken charges have been used most widely for this purpose. Thus it corresponds to the hypothetical physical state of the gas at 0 K and infinitely low pressure. depending on the nature of the chemical structures or processes involved.158 In most cases the molecular field is developed from the quantum-chemically calculated atomic partial charges of the molecule under investigation. Conclusions We have demonstrated above that the molecular descriptors derived from the quantum-chemically calculated molecular total wave function and charge distribution have wide applicability for the development of quantitative structure activity/property relationships in numerous areas of physical. provided it has been geometry optimized.160-162 receptor antagonist and agonist activities. entropy. CoMFA has been used to correlate log k for the SN2 reaction of benzyl benzenesulfonates and p-methoxybenzylamines. it should be mentioned that these descriptors are not completely universal and.149 The nonspecific solvent effects can be described using one of the possible dielectric reaction field models. the thermodynamic functions provided by the quantum-chemical program packages mentioned above still refer to but a single conformation of a single molecule. However. However. First. However. in some applications. Vartiainen. B. 1980. Chem. Menziani. 1183.. Chem.. D. 1247.. Radecki. J. (83) Klopman. J. Suto. (44) Doichinova. F. Taft. 1285. (70) Wold. D. 1990. Adv. Klopman. G. 19.. Phys. Curciani. A. Segal. Computational Aspects for Large Chemical Systems. Grendze. J. E. Supuran. Springer-Verlag.. R. D. S. Rev.. Acedemic Press: London. Lobanov. Chromatogr.-Act. C. 1987. (69) Cardozo. A. Chim. 16. Sci.. Clementi. Sys. N. M. Phys. 1979. A. John Wiley & Sons: New York. V. Am. Toomey. Dieter... G. Geladi. J. 1967. T. 1979. 525. Am.. Magn. L. M.. 29. Rivail. Phys. 83.. Quantum Chem. (27) Bartlett. G. Kaliszan. Vol. S. 23. R. Bartlett. ¨ (51) Tuppurainen. Mutat. (35) Bingham. John Wiley & Sons: Chichester. T. Med. 205. 361. 306.. M. J. 126. pp 115-123. A. 37. 24. Comput. Santry. (79) De Benedetti. Kaiser.-L. Handy. 1983.. 31.. G. 1992. C. Frontier Orbitals and Organic Chemical Reactions.. (76) Ebert. L.. Eur. Struct. 545. P.. Springer-Verlag: Berlin. Quant. 165.199-201 but it is advisable to consult the original papers before doing the calculations since most of the quantum chemical reaction field models involve empirical parameters (dielectric permittivity of the medium. 1994. (80) Tuppurainen. Y. Mol. H. J. 1981. Murata. J. A. P. G. In summary. R. C. C.. P. Kluwer Academic Publ. C..-K. 209. Relat. 1967. Chem. A. J. 32. K. J. 1084. Chem. J. Chem. Chem. Yonezawa. 3783. 1697. U... Ed. B. 1976.. C.. Quantum Chem. 1423. R. O. III. 2026. H. G. 1980. 359. Specific effects. Org. (38) Stewart. R. D. L. Chem. pp 85-225. Am. P. L. Simas. Eds. Theory of Orientation and Stereoselection. 1987. N. J. 8. Chem. 1991.1041 (bulk) solvation effects in these media. 1977. Lopez de Compadre. M. 766. 54. Quant. 227. Wilson. D.. R. 82. J. J. K.. (19) Goddard. R. Chem. Chim. ¨ ¨ Maran. Coats. Anal. 256. Relat. R....-Act. pp 1-83. Cocchi. (71) Geladi. Parr. 1988. 16. 1984. (34) Pople. 16. (10) Buydens. 413 (PBC). J. 4899. Chemom.. 27. Chem. Schlegel. Chim. E. (14) Murrell. (29) Gauss. S. Sugimoto. Sys. Escobar. Cruciani. O. 12. Nagata. Chem. F. Quant.. 22... 181. P. C. (54) Pearson. (15) Clementi. Kettle. G. (37) Dewar. 1109. 1987. G. Chem. 62. M.. Pullman. J. Relat. 1985. R. V..-Act.. C. SpringerVerlag: New York. (18) Supercomputer Simulations in Quantum Chemistry. D.. S. Sci. Berlin. 1987. Rev. P. 24. Lotjonen. 29.. (73) Hansch. Iroff. A. Soc. 1989. 1110. A. 1992.. Chem. 55. 1037. A. Stewart. 1987.. Garcıa-Raso. (47) Zhou. P. R. Latrofa. (Theochem) 1992. D. Jr. 93. Struct. A. 1990. G.. Med.. E. K. J. K. Science 1975. Mihailova. J. Chemom. N. A.. J. (84) Coburn. Velazquez. R. D. A. (32) Pople.. Urban.. ACS Symp. (16) Carsky P. C. D. Vol. 3902. G.. 37. J. C. Carotti. M. (25) Bartlett. 14. Dewar. Chemosphere 1989. 1990. 1983. (30) Pople. C. 1525. Testa. 1988... S. P. M. 4654. 1991. 584. J. S. 107. Chem. 1989. D. P. R. Chem. Res. J. E.. 59. 1992... Acta (Berl. M.. Buss. Hudson. J. cavity size. C. 10. Phys. J. G. 3289. Genchi. (22) Kucharski. Chem. 12. Kaliszan. J.-Act. References (1) (1) Cartier. Relat. 1994. 1989. J. Thiel. A. Soc. Binkley.. 401. 1987.. Chemom. 1992. Yamanishi. 179. Chem.. D. 1974. Ordonica. 1974. P. 1987. Laatikainen. J. K. Elsevier: Amsterdam. (52) Fleming.. Approximate Molecular Orbital Theory. B. Xenobiotica 1994. 499. S. 6. B. G. Segal. Tedder J. (75) Ewing. Intell. VI. Y. J. Raso. Rev. Massart. Leo. 247. M. G.. Ed. 1994. Z. R. J. A. (59) Klopman G. Chem. and shape of the solute molecule). Pople. M. (43) Kikuchi. Y. S. J. 1993. 22. Hashimoto. M. Rev. Struct. 231. (46) Ordorica. Korolkovas. 35. F. Lecture Notes in Chemistry. Intell. Acta 1986. 94. F.. Lab. 12. W. (11) Murugan. Struct. J.. (78) Van De Waterbeemd. W. Int. (39) Gough. Iimura. 10. J. Stewart. F. Alunni. 1983. Anal. M. III. R. S.: Dordrecht. 1984. 99.. 10. 133. Shimoda. 1966.. Chem. 731. Theoretical Drug Design Methods. C.. A. Singh. Menziani.: Dordrecht. Phys. C.. P. Dupuis M. 63. Topsom. (42) Clare. 142. Quant. H. (9) Bodor. S. J. (77) Skagerberg. D. Purvis. J. J.. (13) Sotomatsu.. Chemom. (53) Lewis. pp 34-39. on the molecular structure can be accounted for using the supermolecule approach where the solute molecule is treated together with the specifically coordinated solvent molecules. 1969. 1989. M. Zoebisch. 112. 1993. Am. Med. Weirzba. Intell.. (67) Saura-Calixto.. J. E. P. J. Diercksen.. Hansch. R. 1986. D. J. K. Chem. Ital. M. K. Chem.. (50) Sklenar. A. (56) Nakayama. F. Cremer. Chim. pp 55-165.. M. (36) Dewar. 12.. Delivery 1991.. 1965.) 1982. Schlegel. Struct. 13. D.. J. 1978. K. Y. 273... Halkiewicz. C. (20) Krishnan. (49) Fukui. John Wiley & Sons: New York. Y. G. Chem. 46. R.. (31) Dewar. R.. G. (3) Franke.-Act. Mekenyan. (65) Gaudio. Res. Sys.. J. I. Mutat. M.. 1979. (48) Osmialowski. Soc. Springer-Verlag: Berlin.. D. 1993. S. T. N. Testa. J. 281. Eds. John Wiley & Sons: New York. (57) Fukui. 2. R. T. (17) Methods in Computational Molecular Physics. 120. J. H. Vol. M.. Quantum Chem. 14. primarily hydrogen bonding. 97. 467.. 2.. Res. 185. C. 251. A. 8. 346. De Benedetti. 1989. 266.. M. Wong. Drug Des. F. 1595. S. Co. Lotjonen. Struct.. 1985. Several of them are available in the standard program packages. A. 1966. (2) Redl. H. Natcheva. I. ¨ ¨ Mutat. 1. Belohorcova. McGrawHill: New York. 1934. Gaz. Chem. Pharm. 12. Acta 1967. Chromatogr.. M. 3. Ser. A. F. Quant. Triggle. Church. 5720. 1987. A. (60) Kurtz. V. A. L. (74) Taft.. T. S. T.. 18.. J.. (72) Geladi. P..-Act. M.. 1.. 2103.. G. Ed. A. Substituent Constants for Correlation Analysis in Chemistry and Biology.. Bonelli. Int. 83. (62) Leo. 47. 1992.. 633.. Lo. G. (7) Brown. 1989. 111. Kowalski. 16. Lab. Struct. Phys. 1979.. M.. Takahata. Comput. Comput. Holloway. Relat. P. Phys. Hagiwara.. Quantum Chem. Soc. A. V. In Progress in Physical Organic Chemistry.. G... R. Clementi. C. ´ Sci. J. Halkiewicz.. M. K. Res.. G. Reidel Publ. Org. In Advances in Drug Research. J. A. In Chemical Reactivity and Reaction Paths. S. (61) Cammarata.. 1975. 19. (8) Gruber. J. Cramer. 28. A. Chem. E. P. 1. Vartiainen. Annu. Comput.. Lecture Notes in Chemistry. J. Lehmann. Linda. K. (64) Lewis. F. J. S. K. J. Chem. B.. H. Chromatogr. Vol. 4256.. V. Adv. Res.. 14. 2. Esbensen. (85) Klopman. J. Bindal. Eur. 71. Beveridge. J. Rev. Berkoff.. C. D. Phys.. A. M. W. J. Med. J. R. S. R. S. 44... 349. Chem. J. Cruciani. Soc. 618. Relat.. 45. 1970.. 53.. (41) Cocchi. Laatikainen. P. 1957. Chem. P. Jager. Int. (81) Tuppurainen. Mutat. 6. Solo. J. B. J. Chem. J. 17. (4) Gupta. Hansch. J.. 32... John Wiley & Sons: New York. J. 1981. M. it is clear that quantum chemical descriptors have tremendous applicability and potential in QSPR/QSAR studies in diverse areas of chemistry and biomedicine provided that their use is critically analyzed and justified for a given property or phenomenon. Phys. Med. D. S. Katritzky. C. 1975. Theor. E. (21) Perspectives in Quantum Chemistry. (28) Møller. Karelson. 97. C. G. Strandberg. W. J. A.. 335. . Chem. Fujita. Sci. S. D. Am. Schaefer. ´ L. G. Phys. D.. A. Rachwal. 43. C. Krishnan. Quant. P. 1989.. (33) Pople.. G. J. N. M. 246. Healy. J.. 72. B. K. J. Total Environ. Quantum Chem. A. 231. S. 1989. Triggle. 1986.. Plesset. Chem. Geerlings. Rev. S. Franco. (5) Gupta. Gabanyi. S.. Vol. R. G. A. R. A. G. (Theochem) 1994. 1980. T. J... (63) Hansch. (23) Cizek. 49. ˇ ˇ (24) Pople. S129.. (55) Prabhakar. 91. 87. Pharm. 83. 139. 768. S. P. Ed. 280. (68) Trapani. 1970. Parke. Ab initio Calculations. Mol. A. Chem.. Rosenkranz.. (58) Veith. Frassineti.. M. Soc. Pharm. Liso. (45) Klopman. 561. Theor. Hopfinger. Comput. J. J. J. 157. R. 51. S. 1985. 1994. CHEMTECH 1994.. Santini. Z. J. B. J. 7... (66) Osmialowski. K. J. The Chemical Bond. 44. R.. Ebert. Lecture Notes in ˇ Chemistry. Sci.. Tonucci. (6) Gupta. J. H. R. 179. P. (12) Magee. A.. K. Struct. L.. H. 1978. 209. J. H. M. Lab. A. J. Jortner. 55... J. G. 1993. S. 738.. (40) De Benedetti. Ioannides. (26) Bartlett. 1986. E. Med. A number of different calculation schemes202-221 are available for the description of the solvent bulk (reaction field) effects on the solute geometrical and electronic structure. 187.. Tamm. (82) Debnath. P. H. Dobosh. 11. J. 1984. D. 8. R. W. (164) Loughney. L. Comput. A. A. J. Mickiewicz. F. J.. Abraham.. Morita. Conti. Barcock. Karelson. P. 1992. P.. R. Frenkel. Inf.. C. J. Taylor. B. 339. F. H. J. J.. Abboud. J. A. (139) Tamm. J. R. 1992.. A. Patterson. Stat.. Wilson. J. Quant. 2390. M. P. 36. A. Recent Advances. (172) Diana. Lipkowitz.. H. 36. (99) Greco.. 71.. 306. 1007. T. J. H. Sherwood. J. G. Med. Comput. Bremont. C. Med. 1994.-C.-Aided Mol. 3511. Y. R.... J. C. 2206. Org. A. Chem. A. C. D. Menziani. Rouselle. Development and Engineering Center: Aberdeen Proving Ground. 1. (174) Waller. W.. J. Sci. (127) Hopfinger. 1281. J. C. P. Chem. Egolf. J. F. 56. Fesen. 1993. Med.. Zerner. W. 35. U. Pharmacol. J.. Chem. K. (156) Wold. Chem. H. 1980.. P.. B. J. Topsom. 36. 1993. Des. (165) Calder.. In Correlation Analysis in Chemistry. Murray. Phys. 659. R. C. (130) Famini. 1992. (120) Murray. J. R. 3660. 1992. 8325. 5959. (140) Karelson. L. C. 34. Jurs. Pearson. Vittoria. Am. W. J. H. (170) Burke.. M. 569. 1991.. 1994. 14. R. 1991. Chem. C. Prog. I. Capps. W. Stewart. (157) Cramer. 31.. Can. 1987. J. J. Prog. (94) Stanton. 1978. Tresaurywala. (89) Katritzky.. (162) Raghavan.. Am. 277. D. Perkin Trans. G. K. Comput. 16. 37. L. Dunn. Comput. Wiese. R. B. (152) Also. Kim. Comput. S. (180) Kim. Fujita. Y. J. Ashman. Y. G. Famini. Marshall. C. H. A. Marquez. 479. G. Wilson. Physical Organic Chemistry. J. Chem. 4186. M. Ignatchenko. Am. (159) Waller. Y. L. DeBenedetti.. 493. Quant. 4006. K.. Inf. (111) Hammett. E.-Aided Mol. Eur.154 (153) Kim.. T. R. R. W. T. Casida. J. W. L. M. Hicks. (169) Tokarski. (129) Famini.. H. Patterson. K. Mayer. Des. Hopfinger. Nimura. 7126. J. Analyst 1994. Dieter. (176) Horwitz. K. J. 1993.. T. L. B. Nelson. Leopold. (144) Rzepa. J. Cha. 1992. (154) Debnath.. (163) Rault. (114) Dewar. M. Mishra. Hunt. Chem. J. Bureau. H. King. V. J. Sci. (158) Agarwal. Proc.. W. (148) Gould. M. (106) Charton. F.. S. 3639. Med. R.. J. 339. 115. (119) Politzer. (136) Tamm. H. G. 279. 2056. J. Chem. (131) Famini. Andreoli. Burke. Inf. 37. 32.. Jr. 11.. . Brinck. Eds. P. 295. Zerner. 276. S. Soudijn. Hellberg. Des. 299... Quantum Chem.. P. Kuthan. W. Soc. (100) Vervoort. 1991. (88) Hanai.Act. 1992. Plenum Press: New York. Chem. E. 1993.. Phys. J. A. 2436. R. H. J. OH-) have been proposed as field detectors in CoMFA. Am. Med. Bunce. 1994.T. J. M. A. R.. 93. 71.. 487. Soc. 1992. J. (184) Yoo. Soc. Comput. Chem. (102) Jaffe. Cocchi. Med. J. 102. M. Quant. G. 1992. 1992. Y. M. J. 38. N. Med.. Marshall. 1989. Chem. Chem. J.. R. J. 1985. T. Y. 6. J.. 1990. E. P. Taft. J. (145) Cramer. M. L. Schwender. 11. J. C. 237.. Chem. A. 8. 3-D QSAR in Drug Disign. 1952. 1995. Relat. Mol. T. L. Boyd. S. R.. Chem. D. W. G. J. G.. J. Li. Rietjens. E. 1990. Med. Chem.. S. M. B. Struct. 122. I. P. Soc. R. G. D. M. 1969 91. D. H. Purcell. 195. 1989. Politzer. Chem.. Th. Penski. J. 233. Chapman. I. Hansch. H. N. 211. Robba. A. Biochem.1042 (86) Leo. (113) Taft. J. 3768. J. D. 301. Zerner. O. J. 162. P.. T. W. (90) Needham. D.. J. J. A. 395. 4899.-Aided Mol.. P. P. (155) Wold. Martin. Cotta Ramusino. 1992. W.. M. Relat. Army Chemical Research. Giolitti.. R.. (125) Famini. H. M. M. Quant. Struct. D. 2nd ed.. M. M. Chapman. H. H.. R. (105) Godfrey. Org. Struct. F. M. Jurs. 287. T. 1993..-Act. C. C. P. (96) Egolf. III. 119. Chem. C. Perkin Trans. R.. Chem. (185) Venturelli. Yang. T. B. R. A. Comput. 1990. Chem. C. (118) Sjoberg. M.. 1984. ACS Symp. 553. 57. (179) Debnath. 1993. (THEOCHEM) 1992. Phys. D. J. C.Y. D. Chem.Act. G. Des. ¨ M. H. Struct. 773. 11. J. L.. V. J. S. Des. Seybold. M.. Lobanov. Peaver.K. P... R. (121) Reichardt. Sjostrom. W. J. S. A. A. Hillier. III.-Aided Mol... M. 12. S. 1799. 1992.. T. A. Quant. H.. Hall. J. A. Chem. Chem.. J. E. W. Chem. G. (128) Kurtz. Tamm. M. In Reviews in Computational Chemistry... Corbett. J. P.. (135) Famini. 20. 1993. Shorter. VCH Publishers: New York. D. Pommier. Soc. M. 1988. 1958. 16.. 6030... Y. Relat. 1994. D. Org. Est. J. D. R. Med.. Eds. R. G.S. Tulp. J.. Chem. J. A.. Brinck.. L. A. Chem.-L. S. (95) Egolf.. P. R... M. 1987. H. Karelson. A.. M. (107) Reynolds. Methodol. B. McKinney. Dunn. C.. 36. Comput. J. Kowalski. Tamm. 16. (166) Langlois. C. Cato. Chem. R. (183) Kim. H. 206. Chem. Wilson. R.. E.. 735.. Chem.. 2390. Kim. Am. T. W. Org.. Joshi. C.. 22.. R. Fanelli.. W.. Veeger. SIAM J. J. Struct.. 1992. 1668. (110) Charton. C.. A. 4434. ESCOM: Leiden. 1992. Chem. Szafran. Joshi. R. 1995. 1977.. Chem. Org. Chem.. L.. 11. Tetrahedron Comput. 94. 1993.. 1983. 1984.. Med. L.. Med. G.. G. 1771. 635. C. Soc. K. C. Soc.-P. Med. W. 7. J. Y. 1989.. J. (101) Jaffe. E. Chem. A. C. Comput.. J. Res. McGraw-Hill: New York. 117. III. Kinoshita. (173) Kumar. 1994. I. C. 1554. D. T. D. M. Lambert. P. J. P. J. A. N... H. Huang. Relat. Phys. J. (142) Karelson. 110. Bunce.Act.. H. 1002. R. G. Y.. 1167. 5. Soc.. A. React. Johansson. 68. Marshall. 1990. J. C... Taft. Beck. Am.. J. (171) Krystek. (181) Kim. Struct. Carere. A. Katritzky.. (112) Taft.. Am. 66.. Ashman. K. (103) Jaffe. W. Chem.. M.. (109) Taft. H. M. Marshall. Katritzky. Quant. Oprea. M. 72. Vyplel. Chem. Solvents and Solvent Effects in Organic Chemistry. Kier. C. Nussbaumer.. Oglesby. Eur. J. R. 2 1990. J.. E. 33.. Comput. Phys. Yi. M. 1959. (133) Famini.. T. 1994. 542. J. Buolamwini. Szafran. M. C. 45. H. J. Pilo. 14. I. E. 5. 1984. R. Katritzky.. Y. Chem. Dahlgren. V. Green. Vincent.. J. Relat. Kier. Politzer. Y. Struct.. S.. 1990 (122) Kamlet. B. Chem.. A. 1989. D. P. Wilson.. C. R. Tetrahedron Lett. B. F. Perkin Trans. 82.. Ser. Sci. Med.. C. D. Y. (134) Famini. Shorter. J... D. M. H. Tamm. T. N. B. Quant. 37... 1182. 2151. 1988.. MD.. Quant. Chem... Med..G. Conti.. Stein. J. J. 1991. Chem. van Berkel. J. 40... J. Y. Chem. W. (151) Kubinyi. Kohn. (97) Grigoras.. Inf. 395. S. 1993. Int. Hansch. 1993. Soc. Relat. J. R. 1992. V. Struct-Act. M. Guest. Chem. Med. T.. 113. Kassel. Ruhe. S. J. 271.. A. 81.. Res. 34. R. S. 35. Chem. J. Comput. J. M. (147) Woodcock. C. 81. (132) Famini. 77. W. (91) Stanton. L. 1995. Stouch. (93) Balaban. J.. Phys. 1991.. 99.Act. 37. S. M. H.-Aided Mol. Relat. C. Chem. (104) Sixma.. Martin. M.. A.. Comput.. C.. Soc. R. M. J. Sci. Chem. D. 1991. Mol.. 6. 7. S28.. 1989. 947.. van Wijngaarden. 31. Sci. H. Quantum Chem. Struct. Phys. P. 673. 327. Albano. 26. Recueil 1953. Murray. F. P. 1993. Hicks. P.. Using Theoretical Descriptors in Quantitative Structure. Med. J. A. 2723. L. (123) Kamlet. Zerner.. Plenum Press: New York. J. M.. S. J. P. 1978. A. (115) Gilliom. Acad. Young. M.. CRDEC-TR-085. P. S. R. 1991. M. 1991. Chem. G. J. R. M. J. 11. S. Int. Struct. Geladi. J. 127. 1. D. Comput. R. Gaudy. 247. Wei. J. 36. J. G.. 7. 1983. C. 449.. A. Soc. J. T. J. Chem. 38. Chem. Martin. 890. 1993. 1988. Karelson. (92) Balaban.. Hopfinger. M. Chem.. Karelson. Benigni. 1007. S.. 344. Lett. Wold.. (116) Bohm.. M. Phys. (175) Oprea.. M. G. 1994. 110. T. G. Herslof.-Act. 162. Jurs. Doherty.. 4343. Tropsha. In CHEMOMETRICS: Mathematics and Statistics in Chemistry. H. M. Am. Quant. Comput. 1991. Chem. Chem. I. J.. P. Med. 37. G. 10. Sci. VCH Publishers: New York. K. Lindberg. (87) Bodor. Martin. Org. L. T.. Wozniak. L.. 80. U. Reidel: Dordrecht. A. J. Milne. Zerner. I. 6. Anal. M. K. J. Hillier.. 34. Wilson. Soc. Soc. Naylor. (108) Taft.. Rev. R. D.. Comput.. R. Sci. S. I. Karelson. M. Chem. 10. Comput. Recent Advances. L. Int.-J.P. J. J. Relat. Chem.. 2 1991.. T. Chem. Jurs. Novellino.. 2 1993. B. 74. J. (160) Nicklaus. 1959. 1992. Phys. T. Perkin Trans. K. H. 461. Chem. L.. (161) DePriest. K. (146) Gould. S. 2987. S. G. 165. (182) Kim. W. M. Comput.. 1990.. R. Weinstein. (98) Hecht.. Org. Mickiewicz. Chem. Chapter 7. Berner. 778. 1952. Karelson.. K. (138) Katritzky. Chem. 21.. 1561. J. 18.. 1995. J. D. 5372. J... M. 36. 2 1993.. J. L. Chem. P. C. M. 1. 1060. Am. Karelson. C. J. 11. (143) Katritzky.. L. (177) Sung. Hatano. Katritzky. C. J. M. Org. D. Chem. Y. L. 7. C. Chr. Chem. Struct.. D. M. In Correlation Analysis in Chemistry. 54. Med. 255. A. 20. Wilson. 1994. Chem. Chem. 1440. Comput.. 266. Martin. Eds. 4152. 11. Thiel. R. A. P... Massova.153. Esbensen. J. H. (THEOCHEM) 1992... B. Ed. Soc. 2761.. (THEOCHEM) 1992. (150) Cramer.. Chem.. Org.. Regan. (168) van Steen..Activity Relationships V. M. N. Dunn. Kowalczyk. (186) Grebelli. Sci. J. Perkins Trans. C. R. 1084. J. 38.. Chem. Wyatt.. Hillier.. 1993. M. Sebolt-Leopold. K. Green. R. C. 1988. J. (178) Waller. J. Dutko.-Act. H. D. 1992. 437. (137) Karelson. 20. DeVito. 32. R. J. Edlund. Mutat. Chem. M. Zerner. Inf.. 2 1992. 1940. L. 2. Truhlar. Soc.. B. M. 616. Comput.. I. The Netherlands. 244. E. G. M. R. H. T. C. K.-E. Am. S. Marshall. 6. 1992. (124) Lewis. Silipo. P. 259. W. (167) Waller.. the molecular probes (H2O. Inf. (149) Diercksen. F. 32. (141) Karelson. S. I. ¨ (117) Sotomatsu. 1994. J... 1991.. Prog. Chem.. 1992. 5 (3). D. (126) Wilson... 1952. Wilson. B.. Wessel. J. F. 198. G. Waller. G. A. A. 36.. D. Ed. Med. Mol. 1992. T. A. D. Tamm.. Chem. V. G. Gill. J. Mutat. Tomasi. 1991. B. Phys. (THEOCHEM) 1993. 139. 48.. L. (215) Cramer. Science 1992. W. Interact. C. C. Org. Struct.-L. R. G. J. (THEOCHEM) 1991. Baker. Rizzi... (209) Rivail. Rev. R. Cotta-Ramusino. Ioannides.. (237) Rietjens. P. (226) Grunenberg. O. G. (211) Hoshi. 11901. J. 149. Phys. V.. J. 1995. Yoshio.. G. M.. Gaussian. B. Sci. Chujo. Am. A. 24. R. J. G. Foresman. A. (249) Hansch. 2 1994. G. (194) Ramos. M. 40. OrvilleThomas. A.. (207) Karelson. 12... Pergamon Press: Oxford. Bruggemann. Debnath. J. R..: Pittsburgh. (233) Lewis. 1975. C. S. Frisch. B. Crit. 1199. 539. G. A. Aust. B. B. 1994. Am. 1385. M. C. I. Th. Veeger. Revision A.. Raghavachari. (240) Kourounakis. 481.. Med. H. A. Phys. J. Mol. Head-Gordon. Acta 1995. Replogle. Chem. 1107. 213. B. 193. Terryn. Comput.. 37. M. Soffers. Cocchi. Struct. O. 690.. Mol. M. 96. (232) Lewis.. Struct.... Loew. Persico. 87. (243) Loew. Menziani. (220) Karelson. Comput.. H.. Chem. B. M. Truhlar. J. A. Inf. 1993. J. Menziani.. J. Tomlinson. V. Wong. W. Toxicol. Ed. 1645. Chim. 50. Vol.1043 (187) Rastelli. L. 1995. G. G. O. Chem. M. D. Quintero. R. Lake. P. K. (222) Famini. Chem. Fanelli.. J. Binkley. S.. Environ.. 63.. C. H.. 55. A. 25. N. Chem. 687. Rinaldi. A. A.. (227) Azzaoui. MOPAC Program Package.. III. QCPE No. Chem... Mol. R. Terryn. Perkin Trans. (239) Sixt.. Eur.. Parke. Chem. Chem. Pharm. C. Y. Besler.. Mutat. Hooiveld.. Org... D. Ganellin. M. J. Villar. Lopez de Compadre. A. (197) Benigni. Zerner. M. A.... p 514. J. J. Menziani. S. F. Chem. Scrocco. Chim.. D. D. Environ.. J.. M... (236) Clare. L.. 3536. Ruiz-Lopez. J. Mutagen. 1995. 1987. (246) Zhang. Herges. M. M. (THEOCHEM) 1995. 37. 3. C. 88. J. 1991. (218) Angyan. Struct. Defrees. (198) Seri-Levy.. Phys. J. Mol. M. V. Andres. M. Burt. P. 1994. Ramsden. G. 455. 29. Pharmacol.. 1. 67.. Chemosphere 1995. C.. Chromatography 1995. A. 30. K. Hansch C. Truhlar. Phys. J. Chem.. F. Soc. J. G. G. S. Wilson. D. J. Vol. C. C. 1994. C. 20. Chem. Chem. 629.. R. 79. Tute. 624. S. Chem. A. 1995. Telzer. K. Chem. (204) Tapia. Struct. Res. 327. P. Robb. U. Environ. (242) Clare. 1982... J.. 1980. J. Phys.. In Molecular Interactions. M. C. (238) Galanakis. J. Ed. 343. In Comprehensive Medicinal Chemistry. J. T. PA. Schlegel.... Xenobiotica 1994. (224) Famini. (188) Langenaeker. 55. Phys. A. 1992. R. Mol.. (208) Miertus. Stewart. 79. M. Rev. 12. (235) Supuran. A. H. J. (196) Horwitz. J. S. Gonzales. L. Fanelli. Med. Acta 1980. 17.. Phys. 309. M. (THEOCHEM) 1991. (217) Wang... Mutagen. K. 177. Shawnee. Struct. 6. G. C. Res. Bodor. 681. G. J. G. M. 829. 55. A. Cocchi.. J. K. G. 21. (206) Karelson. Biochem. Mol. K. 256. R. (228) Debnath.. E. C. G. B. C. F. W. Theor. H. I. (244) Shusterman. Sakurai. H. 81. G. 786. Calder. Chem. 6. J. Moro. Chem. D. 1992. 37. F... 1990. 1995.. Goscinski. P. 1994. Mol.. Wiese. Ratajczak. F. Mol. C. M... 234. (199) AMPAC 5. W. M. (THEOCHEM) 1995. 8040. Sci. J. 1992.. C.. 20. (248) Kubinyi... S.. 1988.. 1992. Huang. 140. 293. Org..Biol. Mutagen. J. (203) Constanciel. Mol. C. Neto. 17.. M. J. Shusterman. R. GAUSSIAN 92. E. H. J. In Burger's Medicinal Chemistry and Drug Discovery. (229) Debnath. R. 357. G. G.0.-L. 1995. 10. W. T. 233. Comber. N. J. B. L. Johnson. DeBenedetti. 127. J. (201) Frish. S. 2027.. Chem.. J. Toxicol. Dunn.. pp 105-203. J. L. R.. 307. 275.. M. React.. 1993.. J.. M. (THEOCHEM) 1991. Geerlings. p 183.. 34.. C. Shusterman. 94. 8552. Herget. (231) Debnath. 1995. Chim. (THEOCHEM) 1991. T. G. A. 8305. React. Vervoort.-J. C. J. 93. 113. J.. G. J. 233. Owen. 1992. Fox. J.. Pople. Mol. J. E. 569... 6949. M. Gallo. P.. Wiberg. 1995. 1653. Math. Acta 1980. W. Comperts. B. 259. Giorgi. G. V. J. T. J. DeBenedetti. 53. 19. Phys. Parke. B. Chem. (200) Stewart. M.. A. -Tapia. Struct. J. M. Ioannides. J. J. J. D. Hansch. P.. 1315. Shusterman.. (223) Headley.. M. Ford... (234) Lewis. 11. 92.. 1995. (245) Debnath.. P. van Duijnen. D. Mol. Res. 1727. G. Wilson.. F.. Inf. L. M. K. Lopez de Compadre. Lopez de Compadre. Wiley: New York. A. Debnath.. V. (THEOCHEM) 1991. Zhang... J.. C. Webb. K. 331. T. J.. A. L. H.. (216) Wong. ¨ (214) Cramer. Giraldo. Chim. 1995.. 1992. Med. M. 5014... B. Chem. Hansch. Org. R. (212) Drummond. Morin-Allory. Eds. L. N. 1992. (191) Mitchell. Soc. Hansch. B. J. Med. Menziani. Starnes. (230) Debnath. J. J. 307.. A. P. S. J. D. J. G. 1983.. (205) Thole. Cocchi. Rastelli. . Massova.. 1983. (213) Karlstrom. M.. B. Zerner. E. CHEMTECH 1991. Altschuh. R. B. E. 8. Y. J.. Phys. Yu. C. Shusterman. Med. Comput. West. W. 1994. D. F. 387. (190) Sokolov. J. R. (193) Cosentino. C. Chem.. B. Y. Sannes. 97. B. Chem. (221) Tomasi. 114. 329. E. (192) Bodor. 1981. D. J. A... Hansch.. M. J.. Y. M. M. L. Anal. ¨ 2397. J. E. Chem. 1992. Inc. 286. KS 66216. Struct. 1992. Parke. Struct. 207. 30. (202) Tapia. Richards. (195) Kantola. Chem. 4. Corbett. (241) Makovskaya. 328. Dunn. V.. 4162. C. M. 1983. Famini. O. Chem.. H. L. R. O. Trucks. 23. 905. 38. P. 539. D. Ä ´ (219) Karelson. Clare. J. Comput. J.. J. G. P. F. Dean. Demel. de B. 120.. 38. G. D. K. Theor. A. Res. Mol. Chem. T. Mol. ˆ 1987. M. (225) Cocchi. (247) Wold. Soc.. 227. D. Martin. J. (THEOCHEM) 1994. K. J. Wolff. 59. 305. Schiavi.W. 19. G. 1995... K.. J. 35. M. J. 619. M. J. Struct.. 315. Wilson. DeBenedetti. Turconi. 1990. W. 251. Med. Chem. A. J. J. W. 97. Hansch. W. Shusterman.. (THEOCHEM) 1985. M. 1991.. 337. M. Wiley & Sons: Chichester. Mol.. S. 781. W. Semichem: 7128 Summit. 1994. (210) Rivail.. (189) DeBenedetti. L.
Report "Quantum Chemical Descriptors in QSAR QSPR Studies"