Variability in protein quantification using different mass spectrometry-based proteomics workflows

The aim was to evaluate whether quantified protein levels of membrane transporters and enzymes relevant to drug disposition, obtained from different proteomics methodologies, can be effectively compared and combined.
Protein concentrations of hepatic drug transporters and metabolizing enzymes were quantified in ten identical human liver samples in six laboratories using their respective in-house methods. Nine phase I-metabolizing enzymes (CYPs), four phase II-metabolizing enzymes (UGTs), five uptake transporters (SLCs), and five efflux transporters (ABCs, and SLCs) were included in the analysis. The protein quantification was performed using five targeted and one global, label free approach, respectively. Three laboratories determined the membrane proteins in whole cell lysates while the other three laboratories analyzed proteins after subcellular fractionation. Comparison across all six methodologies were enabled by scaling protein levels obtained in subcellular fractions to those in whole cell, with the assumption of 100% yield in the fractionation process. The contribution of differently quantified transporter proteins to the intrinsic uptake clearance (CLint,uptake) of atorvastatin in vivo were estimated using a static mathematical model.

The protein levels of the higher expressed drug metabolizing enzymes, obtained by the different laboratories, were overall in good agreement. The protein levels obtained from the methods using whole tissue lysate differed on average two-fold, while levels quantified in subcellular fractions and scaled to whole cell were four to seven-fold lower. The protein levels of the lower expressed transporting proteins displayed a larger variation between the methodologies. The transporters were in general quantified with a three-fold average difference across five of the laboratories. However, scaled protein levels from plasma membrane fractions were on average ten to 38-fold lower compared to the other laboratories. By using the protein levels obtained from the different methodologies, the CLint,uptake of atorvastatin was calculated. Consistently in all laboratories, SLCO2B1 and SLCO1B1 contributed mostly to the CLint,uptake, while SLCO1B3 contributed to a lower extent. Despite this consistency, large differences in the actual CLint,uptake values were demonstrated for the transporters, obtained from the different methodologies (cumulative CLint,uptake from the three transporters ranging from 156-411 µl/g liver/min across the laboratories).

Thus, methodological differences such as subcellular fractionation procedures influence protein quantification immensely and are demonstrated to have a large impact on prediction of hepatic uptake clearance of drugs.