Discussion

PBPK modeling approaches are increasingly used to predict transporter-mediated pharmacokinetics and related DDIs (Poirier et al., 2009; Gertz et al., 2013; Varma et al., 2013; Gertz et al., 2014; Jamei et al., 2014); however, the abundance of OATPs and other transporters is not adequately incorporated in the PBPK models as yet, primarily because of to lack of data (Zamek-Gliszczynski et al., 2013; Galetin, 2014). Despite emerging proteomic data, so far no systematic analysis of the differences in total membrane protein expression of OATP1B1, OATP1B3, and OATP2B1 reported in the liver tissue and cryopreserved hepatocytes has been performed. Following the inclusion criteria, eight individual studies were considered herein; the number of individual samples used to quantify the transporters investigated ranged from 1 to 64. Significant moderate to high heterogeneity between studies was detected per individual transporter for the liver tissue abundance data. In contrast, the degree of heterogeneity could not be assessed in cellular systems as the number of studies was either too small and/or only the average expression data of investigated transporters were quoted.

The current analysis revealed marginal differences in OATP expression levels between liver tissue and cryopreserved hepatocytes in suspension, as the relative expression factors for all OATPs were <1.5-fold for this cellular system (Fig. 1, Table 2). In the case of SCHH, relative expression factor of up to 2.5-fold is required for OATP1B3 and OATP2B1, in contrast to OATP1B1, where the weighted mean protein expression of this transporter was comparable between liver tissue and SCHH (based on n = 5 samples). The current analysis revealed substantial differences in the extent of variability of OATP1B3 abundance data (1.5- to 32-fold within individual studies). This trend was also evident in the study with the largest sample size, in which OATP1B3 expression varied 8.8-fold in 64 subjects (Prasad et al., 2014), in agreement with variability in OATP1B1 and OATP2B1 expression within the same study (up to 6.6-fold). In most of the studies, the extent of variability in protein expression exceeds the differences associated with the well-recognized genetic variations in the SLCO1B1 and SCLO1B3 genes encoding OATP1B1 and OATP1B3, respectively (Link et al., 2008; Niemi et al., 2011; Gong and Kim, 2013). However, direct analysis of any potential link between transporter genotype versus protein expression and, subsequently, functional activity was not possible because of the lack of data.

The present meta-analysis investigated for the first time the correlation between OATP expression levels in the same individuals; this analysis was only performed for liver tissue due to availability of data. The low and moderate correlations observed (Fig. 3) suggests limited dynamic coregulation of OATP transporters, as opposed to metabolizing enzymes (Achour et al., 2014). By aggregating the abundance data reported in three studies, the analysis identified a significant but weak correlation of hepatic tissue expression levels of OATP1B1 with age of the human donors, as opposed to the findings reported by each individual selected study (Balogh et al., 2012; Prasad et al., 2014; Vildhede et al., 2014). In addition, no differences in abundance data were apparent between male and female subjects, consistent with the outcomes of individual studies. This finding was in agreement with a recent meta-analysis of abundance data of drug-metabolizing cytochrome P450 enzymes (Achour et al., 2014).

There are several limitations to the analysis presented here that merit consideration when interpreting the findings. The first limitation is the relatively small number of eligible studies and also their sample size, which is particularly evident for data reported in SCHH. In addition, individual expression data of the selected OATP transporters were not quoted or extractable in some of the studies. Hence, average abundance data were incorporated in the analysis alongside individual data, lowering the statistical power of the analysis. In addition, most of the studies did not quantify the transporter expression levels in the liver tissue and cellular systems from the same donor. One exception is the study by Kimoto et al. (2012), in which OATP abundance data were reported for the hepatocyte systems (SCHH and cryopreserved hepatocytes in suspension) prepared from the same donor; however, no corresponding expression data for the liver tissue were available. The mean expression of OATP1B1 and OATP2B1 from the five donors in the Kimoto et al. study did not show a statistical difference between the two hepatocyte formats, with the exception of OATP1B3, which was significantly decreased in SCHH. These findings were partially in agreement with the outcomes of the meta-analysis, as no significant differences in expression levels of all OATP transporters investigated were detected between SCHH and cryopreserved hepatocytes in suspension (n = 17) (Fig. 1). In addition, no expression data are currently available for cryopreserved hepatocytes in monolayer format, highlighting the need for larger datasets to fully clarify differences in transporter expression under differing conditions of culture, associated variability between in vitro systems, and impact of these factors on transporter-mediated IVIVE. Finally, demographic information for the human donors (reported for n = 86 in total) allowed the analysis of correlation of OATP liver tissue expression with age or sex. The corresponding analysis was not feasible for either of the cellular systems, as no demographic information was reported.

No published studies so far have reported protein expression, genotype, mRNA levels and activities of OATP1B1, OATP1B3, and OATP2B1 transporters in the same samples. Similarly, only two studies so far have investigated a potential correlation between expression of OATP transporters and metabolizing enzymes in liver tissue and SCHH (Ohtsuki et al., 2012; Schaefer et al., 2012). However, considering that the abundance data were determined using the plasma membrane fraction, these two studies were excluded from the current analysis. Ohtsuki et al. (2012) reported higher expression of an ion transporter protein Na/K-ATPase (considered to be a basal membrane marker) in the crude total membrane fraction compared with the plasma membrane fraction obtained from the same liver samples (n = 17). However, a higher abundance should be expected, at least theoretically, in the plasma membrane fraction, which is anticipated to be more transporter-enriched and functionally pertinent. Arguably, losses attributable to plasma membrane preparation methods could be responsible for this contradiction, as highlighted recently (Harwood et al., 2014). Plasma membrane may contain only a fraction of the total expressed transporter attributable to potential internalization reported for several transporter proteins (Bow et al., 2008; Kock et al., 2010; Choi et al., 2011). In addition, impact of post-translational modifications on cell surface trafficking and expression of OATPs is still not well defined. For these reasons, it seemed more appropriate to use studies that quantified transporters in crude total membrane as opposed to plasma membrane.

Heterogeneity between studies highlighted in this analysis may be a consequence of methodological diversity and raises questions of whether the differences in expression levels of OATP transporters may be attributable to the method of tissue procurement (organ donors versus resections from diseased livers), tissue storage conditions (frozen versus fresh tissue), and/or diversity in subject demographics. In addition, sample preparation, i.e., membrane fractionation methods, differences in LC-MS/MS protein quantification, selection of the standard peptide, variation in proteolytic digestion of membrane proteins, and technical recovery, may all contribute to the heterogeneity observed. Although there seemed to be an overall agreement between studies in the selection of internal standards and the use of proteolytic enzyme, differences were identified in the method of extraction of membrane fractions, the calibration process, proteolytic conditions, and the type of quantification (details in Supplemental Table 2). Interestingly, Karlgren et al. (2012) and Vildhede et al. (2014) used a label-free relative quantification method that does not use internal standards and requires no calibration. In all the studies, the calibration process was carried out using light peptide standards of known concentrations, with the exception of Ji et al. (2012), who used purified OATP proteins previously quantified by amino acid analysis. Calibration using protein standards is expected to improve the quality of the quantification process; however, this study only quantified one sample of cryopreserved hepatocytes, which did not permit comparison with other data. The discussed technical and analytical differences may account for some of the interstudy variability identified here. It is suggested, therefore, that the interpretation of the heterogeneity noted in the current analysis cannot be attributed solely to true biologic variability in OATP expression without any additional studies with respect to consistency of proteomic LC-MS/MS methodologies and cross-laboratory comparisons of abundance data using the same biologic samples. A recent report (Kumar et al., 2014) suggested that normalization of absolute abundance levels of OATPs with the abundance of the membrane protein marker Na/K-ATPase can reduce the effect of variability introduced by technical and analytical procedures. The authors demonstrated strong linear correlation between normalized OATP1B1 abundance and uptake of estradiol 17-β glucuronide in vitro, supporting further this concept.

In conclusion, quantitative protein levels of OATP transporters in different human hepatocyte formats relative to liver tissue were assessed in a systematic way, emphasizing caution against arbitrary use of transporter proteomic data reported in the literature. The analysis highlighted weak correlation between OATP1B1 expression and age and no direct relationship between sex and OATP expression levels. The current analysis reports for the first time weak to moderate correlation in expression of different transporters (e.g., OATP1B1 versus OATP1B3) in the same individual and highlights the need for corresponding data on transporter-enzyme expression (e.g., OATP1B1 versus CYP3A4). Differences in OATP expression between liver tissue and cellular systems should be accounted for in the IVIVE of transporter-mediated drug disposition in the form of relative expression factor. However, it is evident that minimal differences seen in the currently available OATP expression data cannot solely rationalize and improve underprediction of active uptake clearance reported so far. Additional studies with larger sample size and transporter and enzyme expression data reported in the same individuals are imperative to further refine our understanding of quantitative expression differences between the in vitro cellular systems and human liver tissue and their impact on PBPK-IVIVE of transporter kinetic data.