Regulation of OATP1B1 Function by Tyrosine Kinase–mediated Phosphorylation

The organic anion transporting polypeptide OATP1B1 (SLCO1B1) is expressed at the basolateral membrane of hepatocytes and is involved in sodium-independent uptake of various endogenous and xenobiotic compounds where metabolism or excretion transpires (1). These compounds include a remarkably broad range of structurally diverse substrates, such as angiotensin II type 1 receptor antagonists, angiotensin converting enzyme inhibitors, taxanes, methotrexate, and statins (2). Considering this broad substrate recognition, it has been well established that OATP1B1 is a high-risk site of drug–drug interactions (DDI). For example, elevated systemic concentrations of statins and greater risk of myopathy or rhabdomyolysis occur in patients with reduced functional OATP1B1 genetic variants (3), or those who receive coadministration of OATP1B1 inhibitors, such as rifampicin (4). Because of the consequential pharmacokinetic changes to various xenobiotics when OATP1B1 function is altered, regulatory agencies, such as the FDA, have added this transporter as an important mediator of DDIs in new guidance for industry (5). This guidance advises that when hepatic metabolism and/or biliary secretion contribute to clearance by >25%, sponsors should evaluate their new molecular entities as potential OATP1B1 substrates or inhibitors in vitro. While progress has been made to implement recommendations in this guidance, the influence of most marketed and investigational drugs on OATP1B1 function remains poorly documented.

Recently, there have been several clinical reports of unexpected DDIs between tyrosine kinase inhibitors (TKI) and OATP1B1 substrates (6–8). TKIs inhibit tyrosine phosphorylation events mediated by kinases that contribute to growth factor signaling and are used in various diseases, including cancer, rheumatoid arthritis, pulmonary fibrosis, and macular degeneration (9). Although various in vitro studies have demonstrated that some TKIs reduce OATP1B1 activity (10–13), these agents themselves are not highly recognized substrates of OATP1B1 (14, 15), suggesting that the inhibitory properties are predominantly noncompetitive and potentially the consequence of a target-mediated mechanism. Multiple regulatory aspects of OATP1B1, including genetic and epigenetic mechanisms, have been rigorously evaluated (16). Surprisingly, phospho-tyrosine–mediated regulation of drug transporters has not been extensively studied, although data from global phospho-proteome studies have suggested that several drug transporters may have conserved tyrosine-phosphorylated sites (17). Against this background, we hypothesized that function of OATP1B1 is dependent on kinase-mediated tyrosine phosphorylation, and that certain FDA-approved TKIs potently inhibit this phosphorylation event and act as perpetrators of clinical DDIs.

Human embryonic kidney (HEK293) cells (ATCC) were transfected with FLAG-tagged OATP1B1, or vector control, as described previously (11). Cells were cultured in DMEM supplemented with 10% FBS at 37°C and 5% CO₂, although not beyond 30 passages. Uptake assays were conducted 48 hours following transient transfection using Lipofectamine 2000 with suggested methods (Thermo Fisher Scientific). TKIs were obtained from Sigma-Aldrich and Thermo Fisher Scientific. EZ-Link Sulfo-NHS-SS-Biotin and Streptavidin-agarose Beads were purchased from Thermo Fisher Scientific. Human plateable hepatocytes (Thermo Fisher Scientific) were thawed in a 37°C water bath for 2 minutes, transferred to 50 mL of Hepatocyte Thawing Media (Invitrogen), and centrifuged at 100 × g for 10 minutes. The supernatant was removed and cells were gently resuspended in Phenol red-free Williams Medium E (Thermo Fisher Scientific) with a serum-containing Hepatocyte Plating Supplement Pack (Thermo Fisher Scientific), and plated at a density of 0.8 × 10⁶ cells per well in 24-well collagen-coated Tissue Culture Plates (Thermo Fisher Scientific) at a volume of 0.5 mL per well. Plates were placed in a 37°C incubator for 6 hours, after which the media were aspirated from wells and replaced with serum-free Williams medium E supplemented with a hepatocyte maintenance supplement pack. Plates were then kept at 37°C and medium was replaced every 24 hours for 4 days.

Transport activity was measured either by using 8-(2-[fluoresceinyl]-aminoethylthio)-adenosine-3′,5′-cyclic-monophosphate (8FcA; 5–40 μmol/L), ³H-estradiol-β-glucuronide (³H-EβG; 2 μmol/L), or ³H-lapatinib (1 μmol/L, Moravek Biochemicals), as described previously (11, 18, 19). All cellular accumulation experiments were conducted using phenol red- and serum-free conditions. Substrate concentrations were normalized to total protein content with a Pierce Protein Assay (Thermo Fisher Scientific). OATP1B1-mediated transport was assessed under various conditions, including 15-minute preincubation with various inhibitor concentrations (0.01–10 μmol/L), followed by coincubation with OATP1B1 substrates, as described previously (11). The initial screen to identify potential OATP1B1 inhibitors was conducted using 10 μmol/L to establish a maximum potential of inhibition. Inhibition of transport activity by TKIs was determined by comparing accumulation of 8FcA or ³H-EβG with DMSO alone (control). Recovery of transporter activity was measured by exposing cells to nilotinib (10 μmol/L) for 15 minutes, after which cells were washed three times with PBS and incubated with fresh DMEM for up to 24 hours, when cellular accumulation of 8FcA was measured.

Hepatocytes were exposed to nilotinib, ruxolitinib (10 μmol/L), or vehicle (DMSO) for 15 minutes, followed by measurement of ³H-EβG (1 μmol/L) at 4°C or 37°C. Cells were washed three times with ice-cold PBS and lysed with 1 N NaOH. Cellular concentrations of ³H-EβG were measured using a scintillation counter, as described previously (11), and normalized to total protein concentration as described above. The net active transport of ³H-EβG was calculated by subtracting uptake at 4°C from uptake at 37°C.

Tyrosine to phenylalanine mutants of OATP1B1 were generated using the QuikChange XL Site-Directed Mutagenesis Kit, following the manufacturer's instructions (Agilent Technologies). Mutagenesis primers were designed using QuikChange Primer Design software. DNA was extracted from bacterial colonies, selected with kanamycin, using a QIAprep Spin Miniprep Kit (Qiagen). Successful mutagenesis was confirmed by Sanger sequencing before utility in transient transfection experiments.

A structural model of OATP1B1 was generated by utilizing the intensive mode in Phyre2 (20). Using six structures of major facilitator superfamily transporters as templates with a confidence of 99% (fold library IDs: c2gfpA, d1pw4a, c4cl5B, c3o7pA, c6e8jA, and c4ldsB), 74% of residues were successfully placed at >90% confidence. Because of the lack of representative structures in the Protein Data Bank or a great evolutionary distance between the sequence of OATP1B1 and any solved structures, 144 of 691 residues were modeled ab initio. Most of these residues consist of the extracellular domains that span putative transmembrane domains TM3/TM4 and TM9/TM10.

Whole-cell lysates were prepared with modified RIPA buffer as described previously (21). Coimmunoprecipitation was executed using anti-FLAG beads following the manufacturer's suggestions (EZview Red ANTI-FLAG M2 Affinity Gel, Sigma-Aldrich). Western blot analysis was performed using Invitrogen Bis-tris Gradient Mini-gels, followed by detection with ECL Reagent (Cell Signaling Technology). Primary antibodies were purchased from Cell Signaling technology, and used at a 1:1,000 dilution. Horseradish peroxidase–conjugated goat anti-rabbit antibodies (Cell Signaling Technology) were used as a secondary antibody (1:10,000). Surface expression of biotinylated protein following treatment with nilotinib (10 μmol/L) or clear DMEM was measured following the manufacturer's guidelines for biotinylation of cell surface proteins (Thermo Fisher Scientific). Cells were washed with cold PBS, and then incubated for 1 hour at 4°C with sulfo-N-hydroxysuccinimide-SS-biotin (1 mg/mL in PBS). Cells were washed again with cold PBS containing 100 mmol/L glycine, followed by a 10-minute incubation at 4°C in a PBS-glycine mixture. Cells were then lysed in buffer (10 mmol/L Tris, 150 mmol/L NaCl, 1 mmol/L EDTA, 0.1% SDS, 1% Triton X-100, and 0.2% Protease Inhibitor Cocktail, Sigma-Aldrich, P8340) for 1 hour at 4°C with shaking, and centrifuged at 13,000 rpm for 2 minutes to collect the supernatant. The supernatant was incubated with streptavidin-agarose beads for 1 hour at room temperature under constant agitation, and then washed with lysis buffer used above prior to analysis.

Cells were treated with either DMSO or 10 μmol/L nilotinib for 15 minutes. Samples were washed with ice-cold PBS containing 1% sodium orthovanadate and exposed to lysis buffer (50 mmol/L Tris-HCl, 150 mmol/L NaCl, 5 mmol/L EDTA, 1% NP-40, 0.2% sodium orthovanadate, 0.2% sodium fluoride, 0.2% beta-glycerophosphate, and 0.2% sodium pyrophosphate), containing 0.2% protease inhibitor cocktail and 0.1% Nuclease (Sigma-Aldrich). Lysates were centrifuged at 4°C to collect supernatants, which were added to anti-FLAG immunobeads (EZview Red ANTI-FLAG M2 Affinity Gel, Sigma-Aldrich) that were washed previously in nuclease-free lysis buffer, and incubated overnight at 4°C. Samples were then centrifuged at 4°C to collect supernatants, after which immunobeads were washed five times in buffer. Protein was eluted by adding 100 mmol/L glycine to the immunobeads at 37°C, and neutralized after 30 minutes with 1 mol/L Tris-base. The supernatants were collected and proteins were digested by a surfactant-aided precipitation/on-pellet digestion protocol (22). Derived peptides were analyzed by a highly sensitive and robust nano LC-Orbitrap MS system using a 3-hour gradient, as described previously (23). OATP1B1 peptides were identified by database searching of LC-MS files generated against the Homo sapiens OATP1B1 sequence using Sequest HT (embedded in Proteome Discoverer 1.4). Prediction of phosphorylation probability was conducted simultaneously with PhosphoRS. Identified peptides were sorted by peptide confidence, and only peptides with high or medium confidence (X_corr ≥ 0.9) were included in this analysis. Percentage of tyrosine phosphorylation was calculated by comparing the number of phosphorylated peptides at tyrosine residues (spectral counts) with the total number of spectral counts associated with peptides possessing the same tyrosine residues. Data are presented as the mean and standard error (SEM) of the three independent runs for samples exposed to nilotinib or DMSO.

A high-throughput siRNA screen of 779 protein kinases was performed to identify regulators of OATP1B1 function using methods similar to a previous study (21). Briefly, HEK293 cells transfected with OATP1B1 or vector control that were generated previously (24) were plated in 384-well plates and reverse transfected with 25 nmol/L siRNA using Lipofectamine RNAiMAX Reagent (Life Technologies). After 24 hours of exposure to siRNA, cells were exposed to 1 μmol/L doxycycline to induce OATP1B1 expression. Following 24 hours of induction, OATP1B1 activity was assessed by exposing cells to 8FcA (25 μmol/L) for 30 minutes. Cell viability was assessed by the CellTiter-Glo Luminescent Cell Viability Assay (Promega). Any tyrosine kinase impacted by siRNAs that reduced OATP1B1 function by more than 75% of control siRNA, deemed sensitive to OATP1B1 inhibiting TKIs, and expressed in HEK293 cells or human hepatocytes, were subjected to secondary assessment with 50 nmol/L targeting siRNAs, followed by measurement of OATP1B1 function, as described above.

Eight TKIs capable of inhibiting OATP1B1 by more than 75%, together with acalabrutinib, a negative control, were weighed out and numbered as 1–9 before transfer to conduct a Kinase Hotspot Assay (Reaction Biology Corp). The IC₅₀ value of the nine TKIs (eight OATP1B1 inhibitors and acalabrutinib) was extracted from a 10-dose inhibitory curve with 3-fold serial dilution starting at 50 μmol/L against LYN kinase. All reactions were carried out at 10 μmol/L ATP.

All animals were housed and handled according to approved Institutional Animal Care and Use Committee protocols. Wild-type DBA/1lacJ mice (8–12 weeks old males) were housed in a temperature-controlled environment with a 12-hour light cycle and given standard chow diet and water ad libitum. Rosuvastatin, formulated in 0.5% hydroxypropyl methylcellulose, or methotrexate USP isotonic liquid clinical injection solution (Pfizer) diluted in normal saline, were administered by oral gavage or intravenous bolus in the tail vein, respectively, at a dose of 10 mg/kg. Oral nilotinib (formulated with 10% 1-methyl-2-pyrrolidinone and 90% of polyethylene glycol 300) was administered 30 minutes before rosuvastatin or methotrexate at doses of 50 or 37.5 mg/kg, respectively, by oral gavage. At select timepoints after rosuvastatin or methotrexate administration, blood samples (30 μL each) were taken from individual mice from the submandibular vein, the retro-orbital venous plexus vein, and cardiac puncture at the terminal timepoint in a manner similar to methods published previously (25). Isoflurane was used as an anesthetic. All blood samples were centrifuged at 1,500 × g for 5 minutes, and plasma was separated and stored at −80°C until analysis.

Rosuvastatin was quantified using LC/MS-MS, following a modified procedure published previously (26). A Vanquish Ultra-HPLC coupled with a Quantiva Triple Quadrupole Mass Spectrometer from Thermo Fisher Scientific was used for analysis. An Accucore Vanquish C18 Column (50 × 2.1 mm, dp = 1.5 μm, Thermo Fisher Scientific) with an ACQUITY UPLC BEH Shield RP18 VanGuard Precolumn (2.1 mm × 5 mm, dp = 1.7 μm, Waters Corporation) was employed for analyte separation. The column and autosampler were maintained at 30°C and 4°C, respectively. The mobile phase was composed of solvent A (0.1% acetic acid in H₂O) and solvent B (0.1% formic acid in acetonitrile), and a gradient elution was used for 5 minutes at a flow rate of 0.35 mL/minute. The gradient conditions were as follows: 0–1 minutes, 10%–60% B; 1–3 minutes, 60%–90% B; 3–4 minutes, 90% B; and 4–5 minutes, 90%–10% B. Aliquots of 5 μL of the extracted samples were injected. The following parameters were set for the mass spectrometer: 40 Arb, 20 Arb, 10 Arb, 342°C, and 358°C for sheath gas, aux gas, sweep gas, ion transfer tube, and vaporizer temperature, respectively. The ion source was operated using heated ESI with the ion spray voltage set at 3,500 V in positive ion mode. The collision gas argon was used at a pressure of 1.5 mTorr. An optimized selective reaction monitoring mode was applied for the quantitation with the following parameters: m/z 482.22>258.11 and collision energy at 32.15 V for rosuvastatin, and m/z 488.23>264.11 and collision energy at 33.06 V for the internal standard, rosuvastatin-d6. Protein precipitation extraction was used to isolate rosuvastatin from mice plasma. Prior to analysis, frozen samples were thawed at room temperature, an aliquot of 10 μL of plasma was transferred into a 0.5 mL Eppendorf tube, followed by the addition of 50 μL of internal standard working solution and 20 μL of methanol. The samples were briefly vortex mixed and centrifuged at 13,000 rpm for 7 minutes at 4°C. Next, 65 μL aliquots of the organic layer were transferred to crimp top Polypropylene Vials (Agilent Technologies), and a 5 μL volume of each was injected into the LC/MS-MS system. The lower limit of quantitation (LLOQ) for rosuvastatin was determined to be 5 ng/mL, and the calibration curve ranged from 5 to 200 ng/mL. The within-run and between-run precisions were within 2.56% (15.34% for LLOQ), while the accuracy ranged from 91.39% to 102.88% (96.68% for LLOQ). Concentrations of methotrexate were determined by a Fluorescence Polarization Immunoassay (Abbott, TDxFLx System).

Data are presented as mean, unless otherwise stated. Differences among groups were determined using ANOVA followed by Dunnett test and unpaired t tests. Statistical analyses were conducted using GraphPad Prism version 8.1.2. P < 0.05 was considered to be statistically significant.

To comprehensively evaluate the ability of TKIs to alter OATP1B1 function, we first performed studies using in vitro and ex vivo cell-based models to measure OATP1B1 activity in the presence and absence of FDA-approved TKIs. Our study revealed that although many TKIs, such as acalabrutinib, had no impact on OATP1B1, 29 of 46 tested TKIs significantly reduced OATP1B1-mediated uptake of the prototypical substrate 8FcA in HEK293 cells overexpressing OATP1B1, while exposure to the TKI, larotrectinib, led to an apparent increase in transport activity (Fig. 1A; Supplementary Table S1). These results were confirmed when assessing OATP1B1-dependent uptake of another substrate, EβG, using the same cells (Supplementary Fig. S1A), with select positive and negative OATP1B1-inhibiting TKIs. Further investigation indicated that the most potent inhibitor of OATP1B1 identified from the screen, namely nilotinib, which has also been reported by others to be a strong inhibitor (13, 10), reduces OATP1B1-dependent transport ex vivo in human hepatocytes (Fig. 1B). Nilotinib can also increase plasma concentration of the known OATP1B substrate, rosuvastatin (a drug predominantly eliminated by the liver), after 30-minute pretreatment in mice (Fig. 1C) by increasing the AUC 3.42-fold. A similar outcome occurred with methotrexate (a drug predominantly eliminated by the kidney; Supplementary Fig. S1B) where a 1.33-fold increased AUC in the presence of nilotinib was observed. Next, we found that nilotinib influenced OATP1B1 function at concentrations that can be observed in patients (IC₅₀ ∼1 μmol/L; Fig. 2A; ref. 27), and that the mechanism of inhibition is predominantly noncompetitive, as indicated by kinetic analysis and its effects following preincubation (Fig. 2B and C). Inhibition of OATP1B1 by nilotinib was reversible (Fig. 2D) and not associated with significantly reduced membrane surface expression (Fig. 2E).

On the basis of the predominant noncompetitive and reversible nature of OATP1B1 inhibition by nilotinib, we hypothesized that effects of TKIs on transport function involve direct tyrosine phosphorylation events. This hypothesis was verified by the demonstration that OATP1B1 is indeed tyrosine phosphorylated (Fig. 3A), and that this event occurs at 23 different tyrosine (Y) sites (Supplementary Fig. S2). As a result of the large number of tyrosine phosphorylated residues, exposure to nilotinib was observed to have no significant effect on the overall phosphorylation of OATP1B1. However, among the phosphorylated residues, Y640 and Y645 were found to be consistently highly phosphorylated (80.2% and 74.2%, respectively) and often together (71.3% of peptides detected), while exposure to nilotinib significantly reduced this posttranslational modification at Y645 (P < 0.05; Fig. 3B). Phosphorylation of Y640 alone was not significantly reduced by nilotinib, although a trend to reduced phosphorylation was observed (P = 0.11).

To date, there are no solved structures of OATP transporters, or bacterial homologs with similar sequences that could enable generation of high-confidence homology models encompassing all OATP1B1 residues. However, both a 3D model of OATP1B1 generated using the predictor Phyre2 (Supplementary Fig. S3), and the Consensus Constrained TOPology prediction server (28), which predicts that TM12 ends with A646, indicates that Y645 is located at the membrane interface end of the last intracellular transmembrane helix (TM12), or just at the beginning of the intracellular C-terminal flexible tail. The Y640 residue appears to reside deeper in the membrane. While the lack of experimental structural information for OATP1B1 prevents an accurate depiction of the extent in which Y640 or Y645 phosphorylation would alter the structure of OATP1B1, we subsequently constructed and systematically characterized site-specific tyrosine-to-phenylalanine (Y-F) mutants of OATP1B1. These studies revealed that Y640F and Y645F are associated with reduced OATP1B1 function (Fig. 3C).

Because multiple, structurally and biologically diverse TKIs strongly suppress the function of OATP1B1, we hypothesized that the underlying mechanism is due to inhibition of one or more common tyrosine kinases that are essential for normal OATP1B1 function. To identify a tyrosine kinase(s) responsible for OATP1B1 activity, we performed an RNAi kinase library screen using reverse transfection of transporter-overexpressing HEK293 cells with siRNAs for 779 kinase genes. From this analysis, we obtained 36 hit leads of genes that, once downregulated, reduced the transport activity of OATP1B1 to less than 25% without resulting in more than 25% reduced cell viability (Fig. 4A; Supplementary Table S2). Among these 36 kinase genes, the only known tyrosine kinases identified were PTK7, MERTK, LYN, MAP2K1, MAP2K6, and FGFR2, where knockdown reduced function to 17.5%, 17.5%, 17.7%, 19.7%, 21.4%, and 22.2% of baseline values, respectively. On the basis of gene expression signatures in HEK293 cells, detected protein expression in human hepatocytes (Supplementary Fig. S4A), kinomescan databases, and previous literature (29–31) on tyrosine kinase inhibitory properties of TKIs at concentrations relevant to our study (Supplementary Table S3), LYN was identified as the potential candidate for further functional validation studies. Using a kinase assay, we verified that sorafenib, axitinib, dabrafenib, cabozantinib, ibrutinib, and nilotinib, all of which reduce OATP1B1 function by >75%, also potently inhibit LYN activity with IC₅₀ values of 15–685 nmol/L (Fig. 4B; Supplementary Fig. S4B and S4C). Two of these TKIs, lapatinib and trametinib, did not inhibit the activity of LYN kinase, suggesting that their mechanism of OATP1B1 inhibition is unrelated to an effect on tyrosine phosphorylation. Subsequent studies demonstrated that lapatinib is itself a transported substrate of OATP1B1 and that the inhibitory mechanism of trametinib is competitive (Supplementary Fig. S5), which is in line with previous observations (32, 33). Finally, a more focused assessment confirmed that approximately 50% reduced LYN expression following exposure to siRNA was associated with 51% reduced OATP1B1-mediated transport of EβG (Fig. 4C and D).

In this study, we found that function of the liver uptake transporter, OATP1B1, is partially dependent on LYN, a kinase expressed in hepatocytes, and tyrosine phosphorylation. This enzymatic process was also found to be highly sensitive to nilotinib, which was identified as the most potent inhibitor of OATP1B1 among the currently investigated TKIs. Moreover, considering that nilotinib is orally administered and is susceptible to a first pass effect that would indicate higher exposure than standard systemic levels, the inhibitory effect on OATP1B1 is expected to occur at clinically relevant concentrations (34). These findings not only have substantial clinical ramifications, but also highlight a previously unknown mechanism that sheds light on DDIs that result in increased systemic concentrations and/or side effects in patients receiving polypharmacy regimens involving TKIs and OATP1B1 substrates (6–8). Furthermore, these results indicate that the tyrosine phospho-proteome should be considered during drug development to predict DDI liabilities and avoid unfavorable outcomes when potent kinase inhibitors are coadministered with agents undergoing OATP1B1-dependent elimination pathways.

Although global phospho-proteome and prediction data (17), as well as nonspecific investigations, previously indicated that OATP1B transporters are phosphorylated proteins (35, 36), details regarding the true status or biological impact of tyrosine phosphorylation sites were lacking. Using highly sensitive and comprehensive LC/MS proteomic analysis, we found that multiple OATP1B1 tyrosine residues are phosphorylated, and that Y640 and Y645 were among the most highly phosphorylated amino acids. The importance of these tyrosine residues was further demonstrated by the loss of OATP1B1 activity when substituted for phenylalanine by mutagenesis, and reduced phosphorylation of Y645 in the presence of nilotinib. Consistent with our findings, previous studies have reported that OATP1B1 is predicted to contain multiple binding sites for LYN, a Src tyrosine kinase family member identified from a kinase-knockdown screen that is highly sensitive to nilotinib and other TKIs with OATP1B1 inhibitory properties (37, 38). It is noted that one of these predicted sites is at Y640, a site that is often phosphorylated with Y645 and displays a trend of reduced phosphorylation in the presence of nilotinib. These collective findings imply that LYN facilitates OATP1B1 activity via phosphorylation, and that TKIs capable of inhibiting LYN tyrosine kinase function disrupt these events and render the transporter inactive (Fig. 5). It is under such conditions that systemic concentrations of coadministered OATP1B1 substrates are expected to increase because of reduced hepatic uptake and clearance.

The notion that tyrosine phosphorylation is essential to the function of xenobiotic uptake transporters is not unprecedented. Activity of the organic cation transporter, OCT2 (SLC22A2), and the apical sodium-dependent bile acid transporter, ASBT (SLC10A2), is reportedly mediated by tyrosine phosphorylation (21, 39), and similar to OATP1B1, both OCT2 and ASBT have conserved Src homology binding domains (37, 38). Indeed, we reported previously that the function of OCT2 is dependent on tyrosine phosphorylation by the Src kinase, YES1, and that transport activity is reduced by various TKIs with YES1 inhibitory properties (21). The recognition of OATP1B1 as a tyrosine-phosphorylated protein that is functionally dependent on a related Src kinase provides credence to the thesis that these enzymes are major regulators of solute carrier function, and thereby, the disposition of an increasingly large number of xenobiotics.

During the course of our study, we also identified several FDA-approved TKIs as previously unrecognized, potent inhibitors of OATP1B1 function, including ibrutinib, gilteritinib, and upadacitinib, and confirmed the inhibitory activity of some other TKIs, including pexidartinib. Interestingly, we found that preincubation with larotrectinib was associated with an apparent increase in OATP1B1 function in our in vitro models, while the prescribing information reports that larotrectinib is neither a substrate nor an inhibitor of OATP1B1. Further investigation is clearly warranted. However, the notion that transport function can be enhanced by certain xenobiotics has been reported previously, and could be mechanistically linked to inhibition of negative regulators of OATP1B1 activity, or the result of heterotropic cooperativity in the binding affinity of a substrate (21, 40). Regardless of the exact mechanism, the majority of our findings with TKIs are consistent with data available in prescribing information of individual TKIs, including prior observation that lapatinib and lenvatinib can reduce OATP1B1 function. Conflicting observations identified in our study are likely associated with differences in experimental design. The validity of our main findings is further strengthened through confirmation of the inhibitory potential for various other TKIs that were previously implicated as OATP1B1 inhibitors (10, 40, 41). In fact, two of these TKIs, pazopanib and dabrafenib, are also known to increase the systemic concentrations of the OATP1B1 substrate, rosuvastatin, as well as risk of treatment-related rhabdomyolysis (6, 14, 42). Our identification of nilotinib as the most potent OATP1B1 inhibitor among current FDA-approved TKIs is also consistent with most (10, 13), but not all previous studies (40). The apparent discrepancy with the latter study may be attributed to use of different substrates and uptake assay conditions (5, 15). Importantly, our kinetic analysis and preincubation data revealed that nilotinib is a potent predominant noncompetitive inhibitor that does not influence OATP1B1 surface expression. This is consistent with the notion that nilotinib reduces transport activity by altering its tyrosine phosphorylation status. The inhibitory mechanism also aligns with previous findings that nilotinib and many other TKIs are themselves not highly recognized transported substrates of OATP1B1 (15). Regardless, this inhibitory mechanism has potentially significant relevance in vivo, especially when considering that nilotinib can also diminish activity of the redundant transporter, OATP1B3. Furthermore, the findings observed here with nilotinib also highlight the importance of the recently revised FDA guidelines, whereby transport inhibition by new molecular entities should be investigated even if such agents are not transported substrates.

Interestingly, in addition to LYN, our in vitro siRNA screen identified several other tyrosine kinases, including PTK7, MERTK, MAP2K1, MAP2K6, and FGFR2, that may contribute to the regulation of OATP1B1. These kinases have not been implicated previously as regulators of transport activity and the exact mechanism and extent by which these kinases mediate OATP1B1 function, if any, remain unknown. In addition to the OATP1B1 residues, Y640 and Y645, we detected several additional OATP1B1 phospho-tyrosine sites of potential interest. Although the phosphorylation status of most of these tyrosine residues was unchanged by nilotinib, further investigation is warranted to assess whether these posttranslational modifications contribute to protein stability, function, or localization and whether they can be influenced by any of the other identified kinases in our screen. In this context, it is worth noting that protein kinase C (PKC) has been associated with modulating OATP1B1 cell surface recycling (16), a process that is distinct from the mechanism proposed for the TKIs used in this study as evident from the fact that concentrations of nilotinib used here do not affect PKC activity or cell surface expression of OATP1B1 (31). Collectively, the role of these kinases and OATP1B1 posttranslational modifications need to become a focal point of future studies in vivo, to confirm their true clinical impact.

Our findings that OATP1B1 function is dependent on LYN kinase activity and is sensitive to TKIs raise several new questions. One involves whether loss of LYN kinase activity alone is sufficient to abolish OATP1B1 function and promote clinically relevant DDIs, which could be addressed in LYN-deficient models. Another relates to the role of LYN, or its own regulatory machinery, in the interpatient variability of handling OATP1B1 substrates under diverse environmental, physiologic, and pharmacologic conditions, which should be further explored in the future. The influence of OATP1B1 inhibitory TKIs other than nilotinib on transport function in vivo also warrants further investigation because 14 other TKIs that are capable of reducing OATP1B1-dependent transport by 75% and 19 TKIs that reduce activity by 50% were identified. Many of the TKIs identified recognize relatively few kinases, such as cabozantinib, ibrutinib, and dabrafenib. Some of these small molecules have been reported to inhibit OATP1B1-mediated transport (10, 12, 13, 40), however, many have not, and are not known potent LYN inhibitors at clinically relevant concentrations (43). Future investigation into how phosphorylation influences the structure and function of OATP1B1 is also warranted. Phosphorylation often occurs at transporter tails (44). According to our model, Y645 is located at the terminus of, or immediately following the last transmembrane helix of OATP1B1 (TM12), and, therefore, is likely accessible to tyrosine kinases. Y640 is located deeper within the transmembrane domain, but it still remains probable that this transmembrane helix can unwind and undergo phosphorylation by a mechanism similar to that outlined for threonine phosphorylation in the serotonin transporter, SERT (45), or by another mechanism. Unfortunately, the detailed process of phosphorylation and its impact or dependence on conformational changes that would contribute to transport activity are limited because of lack of experimentally solved OATP structures. Therefore, it is pertinent to make substantial advances in the knowledge of OATP transporter structures and apply these findings to improve our understanding of the transport process. For example, such investigation may clarify whether partial loss of Y640 and Y645 each is sufficient to abolish transport function, or whether nilotinib may act by additional mechanisms. Finally, our study describes a dramatic increase of rosuvastatin AUC with nilotinib, while a less prominent effect was observed with methotrexate. This may be due to methotrexate's predominant renal elimination, a lower nilotinib dose, or a weaker affinity for OATP1B1 compared with rosuvastatin. Clarity on these variations and their impact on methotrexate clearance, as well as other relevant OATP1B1-recognized drugs will require further study to establish specificity and dose dependence on TKI-OATP1B1 substrate DDIs.

In conclusion, our findings have identified a novel regulatory mechanism for the uptake transporter OATP1B1 that involves tyrosine phosphorylation and LYN, a Src kinase, that is highly sensitive to inhibition by multiple approved TKIs, including nilotinib. These studies are of direct human relevance because the identified regulatory pathway provides a mechanistic explanation for previously reported, unanticipated clinical DDIs between TKIs and OATP1B1 substrates, and may aid in the future development of refined combinatorial strategies by avoiding OATP1B1 substrates, such as statins and methotrexate, with TKIs capable of transporter inhibition, to optimize treatment outcome in a diverse array of diseases.

No disclosures were reported.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

E.R. Hayden: Data curation, formal analysis, writing–original draft, writing–review and editing. M. Chen: Data curation. K.Z. Pasquariello: Data curation. A.A. Gibson: Data curation. J.J. Petti: Data curation. S. Shen: Data curation. J. Qu: Conceptualization, resources, methodology. S.-S. Ong: Data curation. T. Chen: Data curation. Y. Jin: Data curation. M.E. Uddin: Data curation. K.M. Huang: Formal analysis. A. Paz: Data curation, formal analysis, visualization. A. Sparreboom: Conceptualization, supervision, methodology, writing–review and editing. S. Hu: Conceptualization, data curation, formal analysis, methodology, writing–original draft, writing–review and editing. J.A. Sprowl: Conceptualization, resources, supervision, investigation, methodology, writing–original draft, writing–review and editing.

Research reported in this article was supported, in part, by the National Center for Advancing Translational Sciences of the NIH under award number KL2TR001413 to the University at Buffalo, start-up funding provided (to J.A. Sprowl) by the Department of Pharmaceutical Sciences, University at Buffalo, and the NIH grants R01CA187176 (to A. Sparreboom) and R01CA215802 (to S. Hu). Support was also received from AHA (grant no., 20PRE35200228 to M.E. Uddin), the OSU Comprehensive Cancer Center Pelotonia program (to K.M. Huang), and the OSU College of Pharmacy Inaugural Dean's Innovation Award (to S. Hu and A. Sparreboom).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.