Multivariate modeling of PCB bioaccumulation in three-spined stickleback (Gasterosteus aculeatus)

Three-spined sticklebacks were exposed to 20 polychlorinated biphenyls (PCBs) through gastrointestinal intake. The PCBs, viz. 2,2',3,4- TeCB, 2,2',4,6'-TeCB, 2,3,3',5'-TeCB, 2,3,4,4'-TeCB, 2,3',4,5'-TeCB, 3,3',4,5-TeCB, 2,2',3,4',6-PeCB, 2,2',4,4',5-PeCB, 2,2',4,6,6'-PeCB, 2,3,3',5,6-PeCB, 2,3,4,4',6-PeCB, 3,3'4,4',5-PeCB, 2,2',3,4,5,6'-HxCB, 2,2',4,4',5,5'-HxCB, 3,3',4,4',5,5'-HxCB, 2,2',3,3',4,5,6-HpCB, 2,2',3,4,4',6,6'-HpCB, 2,2',3,4',5,6,6'-HpCB, 2,3,3',4,4',5,6-HpCB, and 2,3,3',4',5,5',6-HpCB, were selected by means of a full factorial design in combination with principal component analyses based on several physicochemical properties of tetra- through hepta-PCBs. After exposure to the training set, pooled samples of the sticklebacks were analyzed by GCMS after lipid determination and cleanup using Florisil open-column chromatography. Bioaccumulation factors (BAFs) were calculated on a lipid basis by dividing the PCB concentrations in the fish by the respective concentrations in the feed. Higher chlorinated PCBs showed, in general, higher bioaccumulation than the lower chlorinated congeners. Polychlorinated biphenyls with vicinal hydrogens in the meta- and para-positions exhibited low bioaccumulation. Finally 18 of the 20 measured BAFs were used to calculate a multivariate quantitative structure-activity relationship (QSAR) by means of partial least squares fitting to latent structures. By applying this QSAR, the bioaccumulation potentials of 136 nontested tetra- through hepta PCB congeners were predicted.