Abstract

Overexpression of membrane bound, ATP-dependent transport proteins is one of the predominant mechanisms leading to multiple drug resistance in tumor therapy as well as in the treatment of bacterial and fungal infections. In tumor therapy, P-glycoprotein (P-gp, ABCB1) is responsible for transport of a wide variety of natural product toxins out of tumor cells leading to decreased accumulation of cytotoxic drugs within the cells. Inhibition of P-gp thus gives rise to a resensitization of multidrug resistant tumor cells and represents a versatile approach for modulation of multidrug resistance. Within this paper, a set of propafenone-type inhibitors of P-gp were analyzed using both interaction field based methods such as CoMFA and CoMSIA and Hologram QSAR. With both methods, highly predictive models with q2-values > 0.65 were obtained. Models using logP as additional descriptor generally yielded higher predictive power. On basis of unfavorable steric and favorable electrostatic and hydrophobic interaction fields, these models were able to explain all outlayers identified in previous Hansch-analyses. For HQSAR analysis, models with q2-values up to 0.72 were obtained. Positive influences were found for electron donating groups on the aromatic systems. Highly negative influences were found for diphenylalkylamine substituents, which is a further hint for steric hindrance. The models with highest predictive power were used for screening of a small virtual library. Synthesis and pharmacological testing of a sub set of this library showed that the external predictivity of the HQSAR models generally is lower than the internal one.

Keywords: p-glycoprotein, multidrug resistance, comfa, comsia, hologram qsar, propafenone, virtual screening