Pooling Strategies to Improve Throughput of High Resolution Mass Spectrometry‐Based Analysis of Xenobiotic Metabolites Generated from Huh7 Human Hepatoma Cell Lines Transduced with P450 Enzymes

Abstract

As an approach to identify unknown mass spectral features corresponding to xenobiotic metabolites and simultaneously identify the enzyme(s) responsible, we developed a panel of lentivirally transduced human hepatoma cell lines (Huh7) that express xenobiotic metabolizing enzymes. However, screening thousands of chemicals individually against 40-50 different cell lines each would require immense analytical resources. We hypothesized that pooling samples can improve analytical efficiency and throughput without significantly affecting metabolite generation or detection, and designed two pooling strategies to generate and detect metabolites from 36 different chemicals (including a number of drugs whose metabolism is well characterized) and 8 cytochrome P450-expressing HuH7 cell lines. In the pre-pooling strategy, 36 chemicals were grouped into 12 pools of 6 randomized compounds, with each compound present in two separate pools. The 12 pools were each incubated with 8 different cell lines for 0 and 2h resulting in 192 samples for analyses. In the post-pooling strategy, 36 chemicals were incubated individually with 8 different cell lines. After incubation, the 36 chemicals were pooled by the same system as the pre-pooling strategy. 10 μL aliquots of sample extracts were analyzed using liquid chromatography (Dionex Ultimate 3000) and Fourier-transform high-resolution mass spectrometry (Thermo Scientific Fusion) in both HILIC+ and C18- modes. Expression of the P450 proteins was verified by immunoblotting. Results from the pre-pooled samples showed that the anti-cancer drug pazopanib was metabolized by the CYP3A4- and CYP2C8-expressing cell lines. Bupropion and β-naphthoflavone were metabolized mainly by CYP2B6 and CYP1A2, verifying that the cell lines express functional P450s. Next, we compared the metabolomic profiles of the xenobiotics from the two different pooling strategies. The post-pooling strategy generally showed more metabolite generation, although it required 3 times as many biological sample incubations. Hydroxybupropion and β-naphthoflavone dihydrodiol generation were similar between the two pooling strategies. Qualitatively, the generation of hydroxy trifluralin, a possible metabolite of the commonly used herbicide, showed no significant differences between the two pooling strategies, although the signal levels using pre-pooling were 13 times higher than in post-pooling samples. Thus, either pooling strategy can be used as screens to identify enzyme-compound pair reactions and to guide subsequent metabolite generation and characterization. The presence of each xenobiotic in two pools increases stringency for qualifying a feature as a metabolite, while achieving the same confirmation as technical replicates. While the pre-pooling strategy may be vulnerable to confounding effects of potential interactions among xenobiotics in a given pool, the presence of each substrate in two different pools mitigated this concern.