Chemistry Reference
In-Depth Information
environment and the ligand. The free energy of binding is expressed as a linear combination of these interactions
energies using MD or MC ensemble averages, either with the ligand bound to a solvated receptor or with the free
solvated ligand. It is argued that intramolecular conformational energies, entropies and receptor desolvation are
embedded in the linear response approximation and the adjustable parameters of the model. The treatment of long-
range electrostatic interactions can be a complicating factor which is a well known limitation for simulations with
explicit solvent [462].
There are also many unanswered questions regarding the empirical parametrization of the LIE method. The fact that
LIE is computationally less costly and can perform well with truly diverse scaffolds is an advantage over FEP/TI
methods. In addition large savings in calculation time can be obtained by replacing explicit hydration by a
continuum solvent model such as generalized Born or PB. The continuum model can also be used in combination
with MD, MC or MD/MC (HMC) protocols as well as energy minimization only. In addition to electrostatic and van
der Waals terms, LIE can also include the number of hydrogen bonds between ligand and its surroundings, accepted
or donated, apolar, aromatic or polar ligand SASA, receptor SASA and ligand or receptor intramolecular energy.
The presence or absence of a chemical functionality in the ligand as well as the number of ligand rotable bonds can
also be used. As in traditional QSAR, LIE/ELR models may be derived from a training set and used in prediction
modes. In general, at a cheaper computational cost, LIE can rival the accuracy of FEP/TI when a suitable training set
is available.
A general consensus indicates that polar desolvation which opposes binding in aqueous solution should not be
neglected. In continuum solvent models the hydrophobic effect is accounted for by a non-polar solvation SASA-
dependent term which favors binding. Its contribution provides in many cases a main driving force for binding. Of
course, the addition of apolar groups only favors binding if it does not introduce steric clashes or conformational
strain in the bound ligand and if the apolar groups do not further desolvate polar groups.
Among the a priori models which predict intermolecular interaction based on a few adjustable parameters and
molecular structure are COSMO-RS and COSMO-SAC which are extensions of a dielectric continuum-solvation
model to liquid-phase thermodynamics, i.e models which predict liquid-phase activity coefficients. COSMO-based
models require input in the form of a molecule-specific distribution of the surface-charge density, a sigma profile
which can be generated from single structures by performing computational expensive quantum-mechanical
calculations [452].
Conceptually, COSMO-based models create a cavity with the exact size of a molecule within a homogeneous
medium, or solvent, of a dielectric constant and then place the molecule inside the cavity. The solvation free energy
represents the change in Gibbs free energy associated with moving a molecule from a fixed position in an ideal gas
to a fixed position in a solution. The cavity-formation free energy represents the change in Gibbs free energy
required to form a cavity within a solution S of the exact size of the molecule. The charging free energy represents
the Gibbs free energy required to remove the screening charges from the surface of the molecular cavity. We
determine the solvation free energy from the sum of the cavity-formation free energy and the charging free energy.
A sigma profile is a probability distribution of the surface-charge density of a molecule or a mixture. COSMO-based
models construct the molecular shaped cavity within the perfect conductor according to a specific set of rules and
atom-specific dimensions. The molecule's dipole and higher moments withdraw charges from the surrounding
medium to the surface of the cavity in order to screen or cancel the electric field both inside the conductor and
tangential to the surface, allowing the molecule to move freely within the system without altering the system's
overall energy. The induced charges are calculated on the solute surface in discretized space from Poisson's
equation and the zero total potential boundary condition [452].
The average of the segment surface-charge densities from COSMO calculations yields new surface-charge densities.
The sigma profile for a molecule can be defined as the probability of finding a segment with a surface-charge
density. One assumption of a sigma-profile generation procedure is that the optimization geometry from the
calculation in the vapor phase is identical to the optimal geometry in the condensed phase which can save large
amounts of computational time for calculations on larger molecules. Factors such as solvent polarity, molecular size
