Planta Med 2013; 79 - PA36
DOI: 10.1055/s-0033-1351940

In silico prediction and experimental evaluation of furanoheliangolides as potent antitrypanosomal agents

TJ Schmidt 1, R Brun 2, M Kaiser 2, NP Lopes 3, FB Da Costa 3
  • 1University of Münster, Institute for Pharmaceutical Biology and Phytochemistry (IPBP), Corrensstr. 48, D-48149 Münster, Germany
  • 2Swiss Tropical and Public Health Institute (Swiss TPH), Socinstr. 57, CH-4002 Basel, Switzerland
  • 3Universidade de São Paulo, Faculdade de Ciências Farmacêuticas de Ribeirão Preto, Ribeirão Preto, Brazil

Continuing our studies on QSAR for the activity of sesquiterpene lactones (STL) against Trypanosoma brucei rhodesiense (Tbr) [1], we have extended the set of available IC50 data to 69, and generated a new QSAR model [2] in a similar way as reported previously. Descriptors were calculated from optimized 3D structure models. Compounds were divided into a training- and a test set (46/23). The training set descriptor matrix was submitted to genetic algorithm-based variable selection/multiple linear regression (GA-MLR). The best model regarding internal and external predictions (R2= 0.75, Q2= 0.65, P2(test set)= 0.35) was employed to predict the activities for a database of 1700 STL structures.

Quite notably, among the 71 compounds predicted to be highly active (IC50 ≤0.1µM), 15 were STL of the furanoheliangolide (FH) subclass, which had not been tested for anti-Tbr activity before. Hence, four FH (1-4) were tested in vitro.

Fig. 1

Goyazensolide (1), budlein A (2), 15-deoxy-4,5-dihydro-2',3'-epoxygoyazensolide (3) and 4,15-isoatriplicolide tiglate (4) displayed IC50 values of 0.07, 0.07, 0.20 and 0.02µM, respectively. 4 is the most active STL against Tbr found up to present. The lower activity of 3 is due to the absence of the α,β,γ,δ-unsaturated ketone moiety. In agreement with our earlier observations [1], the presence of a second Michael acceptor in addition to the α-methylene-γ-lactone group is required for high anti-Tbr activity.

The experimental results confirm the usefulness of our QSAR model to predict the activity of untested STL. FH represent a new class of very interesting lead compounds against Tbr which deserve further investigations.

References:

[1] Schmidt TJ et al. Molecules, 14, 2062 – 76 (2009)

[2] For all calculations: Molecular Operating Environment 2011.10 (MOE), Chemical Computing Group, Montreal, Canada, http://www.chemcomp.com/

This work is part of the activities of ResNetNPND: www.uni-muenster.de/ResNetNPND/

Support of CCG, Montreal, is gratefully acknowledged.