Additional CoMFA Fields

The straightforward nature of the CoMFA paradigm makes it flexible for introducing different fields in the 3D-QSAR model. Whereas steric and electrostatic properties of molecules are major physicochemical properties related to biological activity, they are purely enthalpic (see Eqs. [6]-[8]). It is desirable, and often important, to characterize additional properties on a three-dimensional basis. Efforts to include entropic properties within a 3D-QSAR framework have been applied to the characterization of the hydrophobic nature of molecules, as reviewed by Folkers and Merz. More recently, reactivity-based fields based on molecular orbitals have also been imported into CoMFA studies (see below). At present, the type of field to be generated and included in a CoMFA model is limited only by the theory used to develop the field. 

One of the earliest efforts directed toward including hydrophobic effects in drug-receptor interactions is the hydropathic interaction (HINT) technique of Kellogg et al. The HINT formalism is strongly rooted in the CLOGP 

Here S is the solvent-accessible surface area for atom /, a, is the hydrophobic atom constant for atom /, and R , = e, where r is the distance between atom i and grid point t. In addition to supplementing standard CoMFA fields, HINT has been demonstrated to effectively model experimentally determined log P values in cases where CLOGP values fail. A different fragment-based, distance-dependent method to estimate lipophilicity, the molecular lipophilic potential (MLP), has been recently shown to be useful in docking and as a third field in CoMFA studies. 

This approach has been applied successfully to model the hydrophobic effect of substituents on the aromatic ring of several series of compounds with respect to alterations in pharmacodynamic (pKj) as well as chemical equilibrium (pKm) constants. In the three cases presented by Kim, the 3D-QSAR models were consistently more robust statistically than the corresponding classical QSAR equations based on t substituent constants. 

Goodford recently added a hydrophobic (DRY) probe ° to the GRID program. This DRY probe could be used to explore hydropathic interactions and is likely to replace the water probe in GRID/CoMFA studies. 

---

In CoMFA, field contours are typically graphed using the following convention green and blue denote favorable interactions for steric and electrostatic fields, respectively, whereas yellow and red mark unfavorable interactions for the same fields. An intuitive approach to these conventions was proposed by Cramer. - In addition to this convention, we proposed the following... 

T. I. Oprea, unpublished work (1996). Because CoMFA fields are derived from a three-dimensional grid placed in Cartesian space, one can empirically estimate that CoMFA models should exhibit three PLS components. A requirement for higher complexity to fit the data set may indicate that mechanisms in addition to ligand fit (e.g., transport, difhision) are included in the target property. 

There are now a few hundred practical applications of CoMFA in drug design. Most applications are in the field of ligand-protein interactions, describing affinity or inhibition constants. In addition, CoMFA has been used to correlate steric and electronic parameters.Less appropriate seems the application of CoMFA to in vivo data, even if lipophilicity is considered as an additional parameter. As most CoMFA applications in drug design have been comprehensively reviewed in three books and in some reviews, " Table 1 gives only an overview of some typical applications that have been reported in the last few years. 

Molecular lipophilicity potential (MLP) has been developed as a tool in 3D-QSAR, for the visualization of lipophilicity distribution on a molecular surface and as an additional field in CoMFA studies [49]. MLP can also be used to estimate conformation-dependent log P values. 

Waller et al. (237) performed a CoMFA study to analyze the metabolic rates of CYP2E1 in rodents as intrinsic clearance of a 12 chlorinated volatile organic compounds (VOCs). After superimposition, the steric and electrostatic field interaction energies, the HINT (/jydropathic interactions) energy (238), and molecular orbital field were calculated in addition to clogP. The best model... 

Additionally, computational chemists often use the resulting output alignment of the molecules as input for 3D-QSAR modeling. As already stated, most field-based 3D-QSAR approaches (such as CoMFA) need a pre-aligned set of molecules and the pharmacophore method is certainly one of the best ways to obtain an objective alignment of the compounds. Klabunde et al., for instance, have recently reported the use of a pharmacophore model of human liver glycogen phosphorylase inhibitors together with 3D information from inhibitor-enzyme complexes to derive a predictive CoMFA model [98]. 

The foundation of the CoMFA approach lies in the fact that the interaction between the biotarget and organic ligand is usually non-covalent and substantially controlled by the shape of molecules. In addition, van der Waals and Coulomb forces in most cases provide an adequate description of non-covalent interactions within a molecular mechanics framework. Thus, the authors assumed that the biological action of compounds can be explained by the shape and electrostatic field of their molecules. 

In addition to the steric and electrostatic descriptors, it was proposed to use other 3D molecular fields characterized by the sampling over the rectangular grid - in particular, the hydrophobic field/molecular lipophilic potential (MLP), ° hydrogen bonding and quantum-chemical parameters, e.g., orbital densities.Descriptor selection techniques are often recommended to enhance the stability, predictivity and interpretability of the CoMFA models. ... 

In addition to an extensive summary provided previously on this moiety (8), Brouillette et al. (209) employed comparative molecular field analysis (CoMFA), a three-dimensional structure-activity technique, to provide a new potential anticonvulsant, 2-hydroxy-2-phe-nylnonanamide (40), whose Na-i-channel inhibition (IC50 = 9 fiM) compared favorably to 40 yM for phenytoin (1). This study suggested that the hydantoin ring system is not necessary in Na+channel binding. Research on water-soluble prodrugs of phenytoin has continued since the work by Stella, which led to the synthesis of fosphenytoin (Id) (8,209-215). A... 

The studies described above revealed that BE values can be reliable predictors of ligand binding affinities for target proteins. The BE value can also serve as an additional descriptor to supplement QSAR models. A recent example demonstrated that the statistical quality of a 3D-QSAR model based on comparative molecular field analysis (CoMFA) improved dramatically—from R2 = 0.65 to R2 = 0.93—by inclusion of calculated BE values [86,87]. 

In addition to the descriptors mentioned above, grid-based methods have been frequently used in the field of quantitative structure-activity relationships (QSAR). In one of the most successful approaches, the comparative molecular field analysis (CoMFA), the molecule is placed in a box and the interaction energy values between this molecule... 

In this chapter only QSAR methods which use physicochemical or structural features of molecules will be discussed, while in Chapter 25 3D-QSAR approaches will be presented. These so-called 3D-QSAR techniques, e.g. CoMFA, use the basic statistical principles, such as partial least squares (PLS), of QSAR methods, but in addition use the three-dimensional characteristics of a molecule specifically related to electronic, steric and lipophilic field effects. In these methods the molecular superposition believed relevant to binding to the target is crucial. 

Other fields than those implemented in the CoMFA program have been proposed for 3D QSAR analyses, e.g. different interaction fields calculated by the program GRID [33, 909, 910] or hydrophobic fields derived from HINT [918 — 921] (chapter 9.2). In addition, any other parameters, e.g. physicochemical properties like log P or quantum-chemical indices, may be added to the X block, if they are properly... 

---