高效液相色谱法构建定量结构-保留关系预测中药肝肾毒性组分的正辛醇-水分配系数值

构建了定量结构-保留关系(QSRR)测定马兜铃酸A、马兜铃酸B、马兜铃内酰胺及白黎芦醇的正辛醇-水分配系数(K_(ow))。采用反相高效液相色谱(RP-HPLC)法,以甲醇-水为流动相,以16种已知K_(ow)值的酸性和中性苯系物为模型化合物,以保留时间两点校正法(DP-RTC)校正保留时间,并由Snyder-Soczewinski方程得100%水相保留因子k_w,建立了表观正辛醇-水分配系数K_(ow)″与k_w的定量关系(QSRR模型),并对模型进行了内、外部验证。结果显示,不同pH下的QSRR模型线性相关性良好(相关系数R~2为0.980~0.987),内部验证(交叉验证相关系数R_(cv)~2为0.982~0.988)和外部验证结果(6种验证化合物的相对误差RE为0.6%~10.9%)令人满意。将建立的QSRR模型应用于中药中4种潜在肝肾毒性化合物的K_(ow)测定,并与软件计算值、摇瓶法(SFM)实验值比较,结果显示该方法准确性更高,且简单快捷。该文提出的采用中性及酸性苯系物建立QSRR模型,通过对结构与性质相似的中药材组分进行RP-HPLC分析,得到各待测组分的保留时间即可获得其K_(ow)值的简便策略,解决了中药组分复杂且难以分离、无法通过SFM测定其K_(ow)值的问题,为通过定量-构效关系(QSAR)模型实现快速预测中药组分的肝肾毒性提供了可靠的K_(ow)数据。

Prediction of n-octanol-water partition coefficients of hepatotoxicity/nephrotoxicity components in traditional herbs using quantitative structure-retention relationship models by high performance liquid chromatography

A quantitative structure-retention relationship (QSRR) was proposed to determine n-octanol-water partition coefficients (K_ow) of aristolochic acid A, aristolochic acid B, aristololactam and resveratrol in this paper. Sixteen acidic and neutral benzene analogues with reliable experimental K_ow data were chosen as model compounds for establishing linear relationship between the apparent n-octanol-water partition coefficient (K_ow″) and retention factor of 100% aqueous phase (k_w), which corresponds to retention factors of solutes using neat aqueous fraction as mobile phase. Methanol-water mixture was used as mobile phase at various mobile phase pH, and retention time (t_R) was rectified by a dual-point retention time correction (DP-RTC). The established models were evaluated by internal and external validations, respectively. QSRR models had good correlation coefficients at different mobile phase pH with correlation coefficients (R²) ranged from 0.980 to 0.987. Both internal (the cross-validated correlation coefficients (R_cv²) of 0.982-0.988) and external (0.6% ≤ relative errors (RE)≤10.9% for all the six verification compounds) validations showed satisfactory results. K_ow values of four hepatotoxic/nephrotoxic ingredients in traditional herbs were predicted by QSRR, and compared with the ones obtained by shake-flask method (SFM) and calculated by software as well. The method provided a convenient and rapid way to obtain reliable K_ow data. In addition, the proposed protocol realized simultaneous prediction for K_ow values of structure-similar components in traditional herbs merely through retention times, thereupon solved the dilemma of K_ow measurement by SFM resulting from inaccessibility of separating single component from complex herbs. These K_ow data can be applied in toxicity prediction of components in traditional herbs through quantitative structure-activity relationships (QSARs) in a follow-up study.