Discussion

In this study, we compared the hepatic disposition of a microdose of the model Abcg2 substrate [¹¹C]erlotinib in EGFR^∆hep and EGFR^fl/fl mice to assess the EGFR-mediated regulation of hepatic Abcg2 transport activity. EGFR^∆hep mice have a cell type-specific deletion of EGFR in hepatocytes, whereas EGFR^fl/fl mice have normal EGFR expression levels in all tissues. EGFR^∆hep mice are healthy and do not display any phenotypical abnormalities except for a reduction in body weight (Natarajan et al., 2007). As there is evidence for sex specific differences in EGFR pathways in the liver of mice (Wang et al., 2016) and as the assessment of these sex differences was not the subject of the present study, we focused on male mice in our experiments. Several previous studies have provided evidence for a regulatory link between EGFR signaling and ABCG2 expression (Fig. 1). It has been shown that the epidermal growth factor (EGF) can induce cell-surface expression of ABCG2 via the MAPK pathway (Meyer zu Schwabedissen et al., 2006). Moreover, the PI3K/Akt pathway, a downstream signal transduction pathway of EGFR, plays a crucial role in the posttranscriptional regulation of ABCG2 expression and its subcellular localization (Mogi et al., 2003; Takada et al., 2005; Pick and Wiese, 2012; Porcelli et al., 2014). For instance, treatment of ABCG2-overexpressing LLC-PKI cells with a PI3K inhibitor resulted in an internalization of ABCG2 from the apical surface and a decrease in the relative expression level of ABCG2 on the cell surface (Takada et al., 2005). These regulatory pathways may either be exploited, for instance, to overcome transporter-mediated chemoresistance of cancer cells, or they could also lead to unwanted DDIs, when a drug that interferes with EGFR signaling is combined with another drug that is transported by ABCG2. Most studies examining the regulation of ABC transporters have used cellular systems, and few in vivo data are available to assess whether modulation of regulatory pathways translates into alterations of transporter-mediated disposition of probe substrates (Slosky et al., 2013; Wang et al., 2014). Recent work has shown that PET with radiolabeled transporter substrates is a powerful tool to noninvasively measure the activities of different ABC and solute carrier transporters in the living organism (Kusuhara, 2013; Langer, 2016).

Erlotinib is a first-generation EGFR-targeting TKI and is approved for the treatment of non–small cell lung and pancreatic cancer. Erlotinib has also been suggested as potential treatment of hepatocellular carcinoma, although a recent phase 3 study was unable to demonstrate survival improvement with erlotinib in advanced-stage HCC (Zhu et al., 2015). Erlotinib is excreted predominantly via the hepatobiliary route; in humans, 83% of an i.v. dose was excreted in feces and only 8% in urine (Ling et al., 2006). Erlotinib is a substrate of ABCG2 and ABCB1 (Kodaira et al., 2010) and has been shown to inhibit these transporters at higher concentrations (Shi et al., 2007). Moreover, there is evidence that erlotinib downregulates cellular ABCG2 expression levels via EGFR inhibition (Porcelli et al., 2014). We recently demonstrated that k_bile of [¹¹C]erlotinib was 1.3-, 2.3-, and 2.8-fold reduced in Abcb1a/b^(−/−), Abcg2^(−/−), and Abcb1a/b^(−/−)Abcg2^(−/−) mice, respectively, relative to wild-type mice (Traxl et al., 2015). Moreover, we found that most of the radioactivity in plasma, liver, and bile was composed of unmetabolized [¹¹C]erlotinib after i.v. injection of the radiotracer into wild-type mice, whereas only radiolabeled metabolites were detected in urine. These data provided evidence that hepatobiliary excretion of [¹¹C]erlotinib is mediated in mice by Abcg2 and, to a lesser extent, by Abcb1a/b, suggesting that [¹¹C]erlotinib may be used as a PET probe substrate to measure Abcg2 transport activity in the liver. In the present study, we used [¹¹C]erlotinib to investigate whether the deletion of EGFR in hepatocytes leads to changes in hepatic Abcg2 transport activity. To obtain quantitative pharmacokinetic parameters of hepatic disposition of [¹¹C]erlotinib, we estimated the transfer rate constants of radioactivity from blood into the liver (k_uptake,liver) and from the liver into bile (k_bile) using integration plot analysis (Shingaki et al., 2015; Traxl et al., 2015) (Fig. 5). K_bile is a parameter that has been used in other studies to assess canalicular ABC transport activities, such as that of Abcg2, in the liver of mice (Takashima et al., 2013).

K_uptake,liver values were not significantly different in EGFR^∆hep and EGFR^fl/fl mice and ranged from 0.72 to 0.80 mL/min per gram of liver tissue, corresponding to a hepatic extraction ratio of 0.72–0.80 (assuming a hepatic blood flow rate in mice of approximately 1.0 mL/min per gram of liver tissue) (Davies and Morris, 1993) (Fig. 5B). K_uptake,liver values in EGFR^fl/fl mice were comparable to those measured in an earlier study in another mouse strain (FVB wild-type mice, k_uptake,liver: 0.734 ± 0.106 mL/min per gram of tissue), indicating good reproducibility of our analytical method (Traxl et al., 2015). There is evidence that hepatic uptake of [¹¹C]erlotinib is transporter-mediated, as reflected by a decrease in k_uptake,liver in wild-type mice coinjected with a therapeutic dose of unlabeled erlotinib compared with mice that only received a PET microdose of [¹¹C]erlotinib (Traxl et al., 2015); however, in contrast to certain low-permeability drugs, for which hepatic disposition is to a large extent transporter-mediated (e.g., statins), erlotinib is a highly lipophilic compound that can be expected to also penetrate cellular membranes by passive diffusion. K_bile was 3.0-fold lower in EGFR^∆hep mice (Fig. 5D), suggesting a decrease in the biliary excretion of [¹¹C]erlotinib. The decrease in total, as well as in membrane-bound, hepatic Abcg2 expression levels in EGFR^∆hep mice as revealed by Western blot analysis (Fig. 7D) suggested transcriptional regulation of Abcg2 protein levels by EGFR rather than an alteration in the membrane localization of Abcg2. Interestingly, there was an increase in hepatic Abcb1a/b expression levels (Fig. 7C), which is in line with another study that assessed changes in transporter expression levels in human conditionally immortalized proximal tubule epithelial cells after treatment with the EGFR recombinant antibody cetuximab; significant decreases occurred in ABCG2 mRNA and increases were seen in ABCB1 mRNA (Caetano-Pinto et al., 2017). Our data thus support the premise that hepatobiliary excretion of [¹¹C]erlotinib is, to a larger extent, dependent on Abcg2 than on Abcb1a/b, as k_bile was decreased in EGFR^∆hep mice despite an apparent upregulation of canalicular Abcb1a/b. Our data also provide evidence that [¹¹C]erlotinib PET is a sensitive tool to measure changes in Abcg2 transport activity in the liver.

Despite the decrease in radioactivity excreted into bile, liver profiles were parallel (Fig. 3B) and liver-to-blood AUC ratios were not significantly different in both mouse models, with a tendency for lower values in EGFR^∆hep mice (Fig. 4D). On the other hand, blood AUC values were significantly higher in EGFR^∆hep mice (Fig. 4A). This result could be related to an increase in the transfer of [¹¹C]erlotinib across the basolateral (sinusoidal) membrane from the liver into the blood in EGFR^∆hep mice, possibly involving basolateral efflux transporter(s), as potential compensatory mechanism for the decrease in biliary excretion. However, this remains speculative as the rate constant for transfer of radioactivity from liver into blood could not be determined with the presently used analysis approach.

Because erlotinib partly undergoes urinary excretion (Ling et al., 2006), we also measured distribution of [¹¹C]erlotinib to the kidneys and urinary bladder (Fig. 6). In a previous study, we found that urinary excretion of radioactivity after i.v. injection of [¹¹C]erlotinib was quite low in wild-type mice and markedly increased in Abcg2^(−/−) and Abcb1a/b^(−/−)Abcg2^(−/−) mice (Traxl et al., 2015), suggesting a shift from hepatobiliary to renal excretion (which was apparently not mediated by renal Abcg2 and Abcb1a/b), when hepatobiliary excretion was impaired owing to knockout of Abcg2. A similar phenomenon was observed in the present study, in which k_urine was 2.2-fold higher in EGFR^∆hep compared with EGFR^fl/fl mice (Fig. 6D).

Our in vivo data confirm previous in vitro data showing that EGFR signaling can regulate ABCG2 expression levels (Mogi et al., 2003; Takada et al., 2005; Meyer zu Schwabedissen et al., 2006; Pick and Wiese, 2012; Porcelli et al., 2014). We provide, to our knowledge, the first evidence that EGFR deletion in hepatocytes translates in vivo into a decrease in Abcg2 transport activity leading to changes in the disposition of an Abcg2 probe substrate. Our findings may have clinical relevance as they raise the possibility that treatment with EGFR inhibitors, such as TKIs or antibodies, may alter hepatic ABCG2 transport activity and thereby lead to changes in hepatobiliary clearance of concomitantly administered ABCG2 substrate drugs, which potentially could lead to hepatotoxicity of drugs but also could prove beneficial in the treatment of liver tumors by prolonging the residence times of anticancer drugs in the liver, if their biliary excretion is dependent on ABCG2 transport activity. It should be noted that our study was conducted with a microdose of erlotinib (∼120-fold lower than a human oral therapeutic dose of 2 mg/kg). As erlotinib itself is a potent inhibitor of ABCG2 (IC_50: 0.13 µM) (Noguchi et al., 2009), ABCG2 transport activity may be saturated in the liver at therapeutic erlotinib doses, which may result in less pronounced effects of EGFR deletion on the disposition of a therapeutic dose of erlotinib. Transgenic mouse models with cell type-specific deletion of regulatory pathways or transporters may serve as valuable tools to assess the impact of transporters on drug disposition.