The syntheses of the phosphane complexes M(CO)5PH3 (M = Mo, W), W(CO)5PD3, and W(CO)5PF3 and the results of X-ray structure analyses of W(CO)5PH3 and Mo(CO)5PCl3 are reported. Quantum-chemical DFT calculations of the geometries and M−P bond dissociation energies of M(CO)5PX3 (M = Cr, Mo, W; X = H, Me, F, Cl) have been carried out. There is no correlation between the bond lengths and bond dissociation energies of the M−P bonds. The PMe3 ligand forms the strongest and the longest M−P bonds of the phosphane ligands. The analysis of M−PX3 bonds shows that PCl3 is a poorer σ donor and a stronger π(P) acceptor than the other phosphanes. The energy decomposition analysis indicates that the M−P bonds of the PH3 and PMe3 complexes have a higher electrostatic than covalent character. The electrostatic contribution is between 56 and 66% of the total attractive interactions. The orbital interactions in the M−PH3 and M−PMe3 bonds have more σ character (65−75%) than π character (25−35%). The M−P bonds of the halophosphane complexes M(CO)5PF3 and M(CO)5PCl3 are nearly half covalent and half electrostatic. The π bonding contributes ∼50% to the total orbital interaction.

No abstract

1We present a combination of semiempirical quantum-mechanical (SQM) calculations in the conductor-like screening model with the MM/GBSA (molecular-mechanics with generalised Born and surface-area solvation) method for ligand-binding affinity calculations. We test three SQM Hamiltonians, AM1, RM1, and PM6, as well as hydrogen-bond corrections and two different dispersion corrections. As test cases, we use the binding of seven biotin analogues to avidin, nine inhibitors to factor Xa, and nine phenol-derivatives to ferritin. The results vary somewhat for the three test cases, but a dispersion correction is mandatory to reproduce experimental estimates. On average, AM1 with the DH2 hydrogen-bond and dispersion corrections gives the best results, which are similar to those of standard MM/GBSA calculations for the same systems. The total time consumption is only 1.3-1.6 times larger than for MM/GBSA.Keywords: MM/PBSA, semiempirical calculations, ligand binding, continuum solvation, dispersion, hydrogen-bond corrections.2 Introduction Most drug molecules exert their action by binding to a receptor, typically a protein, forming a complex, as described by the reaction P + L → PL (1) where P is the protein, L the ligand (the drug), and PL the complex. The binding is governed by the binding free energy, ΔG bind . Much effort has been spent on developing computational methods to estimate this quantity. 1,2If ΔG bind could be accurately calculated, important parts of drug development could be performed in the computer. Thereby, the number of drug candidates that needs to be synthesised could be strongly reduced, which would allow pharmaceutical companies to save vast amounts of money and time.Computational methods to estimate binding affinities range from statistical scoring functions to simulation-based methods that are exact in theory but require extensive sampling of unphysical intermediate states. 1 An attractive alternative is the so-called end-point methods, which sample only the protein, the ligand, and the complex. One of the most popular endpoint methods is MM/GBSA (molecular mechanics with generalised Born and surface-area solvation). This method estimates the binding free energy as the difference in free energy between the complex, the protein, and the ligand, viz., ΔG bind = G(PL) -G(P) -G(L). Each free energy is estimated from the sum 4,5where the first two terms are the electrostatic and van der Waals energies of the system, estimated at the molecular mechanics (MM) level, G solv is the polar solvation free energy, G np is the non-polar solvation free energy, and the last term is the absolute temperature multiplied by an entropy estimate, obtained at the MM level. The brackets in Eqn. 2 indicate an average over snapshots from a molecular dynamics (MD) or Monte Carlo simulation. The sampling of snapshots is performed at the MM level because the number of atoms in the system is typically more than 10000, including explicit solvent molecules. However, when the energies in Eqn. 2 are computed, the solvent molecules...

Models for the prediction of blood-brain partitioning (logBB) and human serum albumin binding (logK(HSA)) of neutral molecules were developed using the set of 5 COSMO-RS sigma-moments as descriptors. These sigma-moments have already been introduced earlier as a general descriptor set for partition coefficients. They are obtained from quantum chemical calculations using the continuum solvation model COSMO and a subsequent statistical decomposition of the resulting polarization charge densities. The model for blood-brain partitioning was built on a data set of 103 compounds and yielded a correlation coefficient of r2 = 0.71 and an rms error of 0.40 log units. The human serum albumin binding model was built on a data set of 92 compounds and achieved an r2 of 0.67 and an rms error of 0.33 log units. Both models were validated by leave-one-out cross-validation tests, which resulted in q2 = 0.68 and a qms error of 0.42 for the logBB model and in q2 = 0.63 and a qms error of 0.35 for the logK(HSA) model. Together with the previously published models for intestinal absorption and for drug solubility the presented two models complete the COSMO-RS based set of ADME prediction models.