Two- and three-dimensional quantitative structure–permeability relationship of flavonoids in Caco-2 cells using stepwise multiple linear regression (SMLR), partial least squares regression (PLSR), and pharmacophore (GALAHAD)-based comparative molecular similarity index analysis (COMSIA)
- First Online:
- Cite this article as:
- Gonzales, G.B., Van Camp, J., Zotti, M. et al. Med Chem Res (2015) 24: 1696. doi:10.1007/s00044-014-1241-4
Limited oral bioavailability has hindered the widespread use of flavonoids as bioactive substances. Several studies have been performed to evaluate the transport characteristics of flavonoids in the intestines using cell models, such as Caco-2 cells, but information regarding the key structural features of flavonoids that influence intestinal uptake is still limited to date. In this study, quantitative structure–permeability relationship (QSPR) models were developed to study the permeability of 36 flavonoids through Caco-2 cells using both 2D and 3D approaches. For the 2D model, stepwise multiple linear regression (SMLR) and partial least squares regression (PLSR) resulted in good internal (RSMLR2 = 0.8, RPLSR2 = 0.93) and external (RSMLR2 = 0.93, RPLSR2 = 0.90) predictability using a set of 409 molecular descriptors. The high cross-validated (leave-one-out) R2 values (Q2) for both 2D models (QSMLR2 = 0.77, QPLSR2 = 0.67) suggest that the models are robust and predictive. A pharmacophore (GALAHAD)-based COMSIA analysis was used to generate the 3D QSPR model, which yielded a predictive and robust model (Rtraining2 = 0.96, Rtest2 = 0.95, Q2 = 0.625) composed of hydrogen bond acceptor and donor fields. According to the contour plots, the locations of hydrogen bond acceptors and donors play a crucial role in determining Caco-2 permeability of flavonoids. The models provide deeper insight into the QSPR of flavonoids on intestinal absorption using Caco-2 cell models and could be useful for the high-throughput screening of flavonoids and/or flavonoid-like drugs with health-promoting activities.