Reviews in Computational Chemistry Volume 18
Reviews in Computational Chemistry Volume 18
Reviews in Computational Chemistry Volume 18
- No tags were found...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
difference <strong>in</strong> conformational energy between the global m<strong>in</strong>imum structure<br />
and the current conformation of the ligand <strong>in</strong> the complex. However, force<br />
field estimates of energy differences between <strong>in</strong>dividual conformations are not<br />
reliable for all systems. In practice, better correlations with experimental<br />
b<strong>in</strong>d<strong>in</strong>g data are obta<strong>in</strong>ed when stra<strong>in</strong> energy is used as a filter to weed out<br />
unlikely b<strong>in</strong>d<strong>in</strong>g geometries rather than when stra<strong>in</strong> energy is added to the<br />
f<strong>in</strong>al score. Estimation of ligand stra<strong>in</strong> energy based on force fields can be<br />
time consum<strong>in</strong>g, and so alternatives such as empirical rules derived from<br />
small-molecule crystal structure data are often employed. 144 Conformations<br />
generated by such programs are, however, often not stra<strong>in</strong> free, because<br />
only one torsional angle is treated at a time. Some stra<strong>in</strong>ed conformations<br />
can be excluded when two consecutive dihedral angles are simultaneously<br />
taken <strong>in</strong>to account, however. 78<br />
Empirical Scor<strong>in</strong>g Functions<br />
The underly<strong>in</strong>g idea of empirical scor<strong>in</strong>g functions is that the b<strong>in</strong>d<strong>in</strong>g<br />
free energy of a noncovalent receptor–ligand complex can be <strong>in</strong>terpreted as<br />
a sum of localized, chemically <strong>in</strong>tuitive <strong>in</strong>teractions. Such energy decompositions<br />
can be a useful tool to understand b<strong>in</strong>d<strong>in</strong>g phenomena even without analyz<strong>in</strong>g<br />
3D structures of receptor–ligand complexes. Andrews, Craik, and<br />
Mart<strong>in</strong> 126 calculated average functional group contributions to b<strong>in</strong>d<strong>in</strong>g free<br />
energy from a set of 200 compounds whose aff<strong>in</strong>ity to a receptor had been<br />
experimentally determ<strong>in</strong>ed. These average functional group contributions<br />
can then be used to estimate a receptor-<strong>in</strong>dependent b<strong>in</strong>d<strong>in</strong>g energy for a compound<br />
that can be compared to experimental values. If the experimental value<br />
is approximately the same as or higher than the calculated value, one can <strong>in</strong>fer<br />
a good fit between receptor and ligand and essentially all functional groups of<br />
the ligand are <strong>in</strong>volved <strong>in</strong> prote<strong>in</strong> <strong>in</strong>teractions. If the experimental energy is<br />
significantly lower, one can <strong>in</strong>fer that the compound can not fully form its<br />
potential <strong>in</strong>teractions with the prote<strong>in</strong>. Experimental b<strong>in</strong>d<strong>in</strong>g aff<strong>in</strong>ities have<br />
also been analyzed on a per atom basis <strong>in</strong> quest of the maximal b<strong>in</strong>d<strong>in</strong>g aff<strong>in</strong>ity<br />
of noncovalent ligands. 145 It was concluded that for the strongest b<strong>in</strong>d<strong>in</strong>g<br />
ligands, each nonhydrogen atom on average contributes 1.5 kcal/mol to the<br />
total b<strong>in</strong>d<strong>in</strong>g energy.<br />
With 3D structures of receptor–ligand complexes at hand, the analysis of<br />
b<strong>in</strong>d<strong>in</strong>g phenomena can of course be much more detailed. The b<strong>in</strong>d<strong>in</strong>g aff<strong>in</strong>ity<br />
G b<strong>in</strong>d<strong>in</strong>g can be estimated as a sum of <strong>in</strong>teractions multiplied by weight<strong>in</strong>g<br />
coefficients Gi<br />
Scor<strong>in</strong>g Functions for Receptor–Ligand Interactions 53<br />
G b<strong>in</strong>d<strong>in</strong>g<br />
X<br />
i<br />
Gi fiðr l; rpÞ ½2Š<br />
where each fi is a function of the ligand coord<strong>in</strong>ates r l and the prote<strong>in</strong> receptor<br />
coord<strong>in</strong>ates rp, and the sum is over all atoms <strong>in</strong> the complex. Scor<strong>in</strong>g schemes