Brain Accumulation of Dasatinib Is Restricted by P-Glycoprotein (ABCB1) and Breast Cancer Resistance Protein (ABCG2) and Can Be Enhanced by Elacridar Treatment

Purpose: Imatinib, a BCR-ABL tyrosine kinase inhibitor, is a substrate of the efflux transporters P-glycoprotein (P-gp; ABCB1) and ABCG2 (breast cancer resistance protein), and its brain accumulation is restricted by both transporters. For dasatinib, an inhibitor of SCR/BCR-ABL kinases, in vivo interactions with P-gp and ABCG2 are not fully established yet.

Experimental Design: We used Abcb1a/1b^−/−, Abcg2^−/−, and Abcb1a/1b;Abcg2^−/− mice to establish the roles of P-gp and ABCG2 in the pharmacokinetics and brain accumulation of dasatinib.

Results: We found that oral uptake of dasatinib is limited by P-gp. Furthermore, relative brain accumulation, 6 hours after administration, was not affected by Abcg2 deficiency, but absence of P-gp resulted in a 3.6-fold increase after oral and 4.8-fold higher accumulation after i.p. administration. Abcb1a/1b;Abcg2^−/− mice had the most pronounced increase in relative brain accumulation, which was 13.2-fold higher after oral and 22.7-fold increased after i.p. administration. Moreover, coadministration to wild-type mice of dasatinib with the dual P-gp and ABCG2 inhibitor elacridar resulted in a similar dasatinib brain accumulation as observed for Abcb1a/1b;Abcg2^−/− mice.

Conclusions: Brain accumulation of dasatinib is primarily restricted by P-gp, but Abcg2 can partly take over this protective function at the blood-brain barrier. Consequently, when both transporters are absent or inhibited, brain uptake of dasatinib is highly increased. These findings might be clinically relevant for patients with central nervous system Philadelphia chromosome–positive leukemia, as coadministration of an inhibitor of P-gp and ABCG2 with dasatinib might result in better therapeutic responses in these patients.

Multidrug efflux transporters of the ATP-binding cassette (ABC) transporter family, such as P-glycoprotein (P-gp; ABCB1), breast cancer resistance protein (ABCG2), and multidrug resistance protein 2 (ABCC2), can have an important effect on chemotherapy. These efflux pumps have a broad and overlapping substrate specificity and are expressed at apical membranes of important epithelial barriers and at the canalicular membrane of hepatocytes. Consequently, these proteins can facilitate excretion of transported drugs via liver, intestine, and kidneys and restrict the intestinal uptake of orally encountered substrates. In addition, ABC transporters are localized at the so-called sanctuary site barriers, such as blood-brain, blood-testis, and blood-placenta barrier, where they restrict accumulation of potentially harmful compounds. Moreover, (over-)expression of these efflux pumps in tumor cells can lead to multidrug resistance through active efflux of cytostatic drugs (1).

Imatinib, a first-generation, orally active inhibitor of BCR-ABL kinase, is currently used as frontline therapy for Philadelphia chromosome–positive (Ph+) leukemia (defined by the fusion of the BCR and ABL genes), including chronic myeloid leukemia (CML) and Ph+ acute lymphoblastic leukemia. Imatinib has revolutionized the treatment of CML by its impressive therapeutic responses in the chronic phase of the disease. However, patients with advanced disease show generally lower and more transient responses to imatinib. In addition, accumulating evidence indicates that resistance to imatinib in a significant proportion of patients in all disease stages results in failure to achieve an optimal response (2–4). Thus far, several mechanisms of resistance have been reported, including point mutations in the BCR-ABL kinase domain, amplification of the gene, and imatinib binding to α₁-acid glycoprotein (5–9).

Dasatinib is a second-generation, oral, multitargeted inhibitor of BCR-ABL and SRC family kinases, rationally designed for CML treatment (10). Compared with imatinib, dasatinib has 325-fold greater in vitro potency against cells expressing “wild-type” (WT) BCR-ABL (11). Furthermore, dasatinib effectively inhibits the growth of clones of a leukemic cell line expressing all known imatinib-resistant BCR-ABL isoforms, with the exception of T315I (12). Importantly, dasatinib has shown significant activity in imatinib-resistant or imatinib-intolerant patients with CML or Ph+ acute lymphoblastic leukemia (13–17).

Imatinib is a substrate of the efflux transporters P-gp and ABCG2 and penetration of imatinib through the blood-brain barrier is markedly restricted by both transporters (18–27). This might explain why up to 20% of the imatinib-treated patients with either lymphoid or myeloid blast crisis CML or Ph+ acute lymphoblastic leukemia develop central nervous system (CNS) relapses (28, 29). In contrast, dasatinib was recently found to have antitumor activity in a mouse model of intracranial CML, and in patients with CNS Ph+ leukemia, dasatinib induced substantial responses (30). Moreover, in K562 leukemic cells that overexpress P-gp, dasatinib showed significant antiproliferation activity, in contrast to imatinib (31). These findings may indicate that dasatinib is not seriously affected by P-gp and/or ABCG2 at the blood-brain barrier or by P-gp in the K562 cells. Nonetheless, it was recently found that the accumulation of dasatinib was decreased in P-gp–overexpressing and ABCG2-overexpressing leukemic cells and that dasatinib was transported in P-gp–overexpressing MDCK-II cells, suggesting that dasatinib is a substrate of P-gp and ABCG2 (32, 33).

To further study the interaction of dasatinib with drug transporters, we used in vitro Transwell transport assays. In addition, Abcb1a/1b^−/−, Abcg2^−/−, and Abcb1a/1b;Abcg2^−/− mice were used to investigate the in vivo roles of P-gp and ABCG2 in the pharmacokinetics and brain accumulation of dasatinib.

Chemicals. Dasatinib monohydrate originated from Sequoia Research Products. GlaxoSmithKline kindly provided Elacridar (GF120918). [¹⁴C]Inulin was from Amersham. Methoxyflurane (Metofane) was from Medical Developments Australia. Bovine serum albumin, fraction V, was from Roche. All other chemicals and reagents were obtained from Sigma-Aldrich.

Transport assays. Polarized canine kidney MDCK-II cell lines were used in transport assays. MDCK-II cells transduced with human ABCB1 or ABCC2 or murine Abcg2 were described previously (34, 35). Transepithelial transport assays using Transwell plates were carried out as described previously with minor modifications (36). Experiments with cells transfected with human ABCC2 were done in the presence of 5 μmol/L elacridar to inhibit any endogenous P-gp activity. Elacridar does not affect ABCC2 activity. When elacridar was applied, it was present in both compartments during 2-h preincubation and during the transport experiment. After preincubation, experiments were started (t = 0) by replacing the medium in either the apical or basolateral compartment with fresh Opti-MEM medium (Invitrogen), either with or without 5 μmol/L elacridar and containing 5 μmol/L of dasatinib. Cells were incubated at 37°C in 5% CO₂, and 50 μL aliquots were taken at t = 2 and 4 h. Transport was calculated as the fraction of drug found in the acceptor compartment relative to the total amount added to the donor compartment at the beginning of the experiment. Transport is given as mean percentage ± SD (n = 3). Membrane tightness was assessed in parallel, using the same cells seeded on the same day and at the same density, by analyzing transepithelial [¹⁴C]inulin (∼3 kBq/well) leakage. Leakage had to remain <1% of the total added radioactivity per hour.

Relative transport ratio. Active transport was expressed by the relative transport ratio (r), defined as r = percentage apically directed transport divided by percentage basolaterally directed translocation, after 4 h (37).

Animals. Mice were housed and handled according to institutional guidelines complying with Dutch legislation. Animals used were male WT, Abcb1a/1b^−/− (38), Abcg2^−/− (35), and Abcb1a/1b;Abcg2^−/− (39) mice, all of a >99% FVB genetic background, between 10 and 12 wk of age. Animals were kept in a temperature-controlled environment with a 12-h light/12-h dark cycle and received a standard diet (AM-II, Hope Farms) and acidified water ad libitum.

Plasma pharmacokinetics. Dasatinib was dissolved in DMSO (25 mg/mL) and 25-fold diluted with 50 mmol/L sodium acetate buffer (pH 4.6). For oral studies, dasatinib was dosed at 10 mg/kg body weight (10 mL/kg). To minimize variation in absorption, mice were fasted 3 h before dasatinib was administered by gavage into the stomach using a blunt-ended needle. For plasma pharmacokinetic studies after systemic exposure, dasatinib was applied by i.p. injection because i.v. injection into a tail vein compromises multiple blood sampling from the tail. For i.p. studies, dasatinib was dosed at 5 mg/kg body weight (5 mL/kg). We used a 2-fold lower dose for i.p. than for oral studies to obtain plasma concentrations and area under plasma concentration-time curve (AUC) values that are in the same range for both administration routes. Multiple blood samples (∼50 μL) were collected from the tail vein at 7.5 min (i.p.) or 15 min (oral) and 30 min and 1, 2, 4, and 6 h (both series) using heparinized capillary tubes (Oxford Labware). At the last time point, mice were anesthetized with methoxyflurane and blood was drawn by cardiac puncture. Immediately thereafter, mice were sacrificed by cervical dislocation and brains were rapidly removed, homogenized on ice in 4% bovine serum albumin, and stored at −20°C until analysis. Blood samples were centrifuged at 2,100 × g for 6 min at 4°C, and the plasma fraction was collected, completed to 200 μL with drug-free human plasma, and stored at −20°C until analysis.

Brain accumulation of dasatinib in combination with elacridar. Elacridar was dissolved in a mixture of ethanol, polyethylene glycol 200, and 5% glucose (2:6:2) and i.v. injected into a tail vein at 10 mg/kg (2.5 mL/kg). Dasatinib dissolved in DMSO (25 mg/mL) was 25-fold diluted with 50 mmol/L sodium acetate buffer (pH 4.6) and injected as a single i.v. bolus at 5 mg/kg body weight (5 mL/kg). Dasatinib was administered 15 min after an injection with either elacridar or with the vehicle used to dissolve elacridar. Blood and brains were isolated 60 min after dasatinib administration and processed and stored as described above.

Relative brain accumulation. Brain concentrations were corrected for the amounts of drug in the brain vasculature (i.e., 1.4% of the plasma concentration right before the brains were isolated; ref. 22). Subsequently, brain accumulation after oral or i.p. dasatinib administration was calculated by determining the dasatinib brain concentration relative to the AUC from 0 to 6 h (AUC_0-6). For studies with i.v. dasatinib in combination with elacridar, blood was collected only just before the brains were isolated, 60 min after dasatinib administration. Therefore, brain-to-plasma ratios were calculated to assess the relative brain accumulation.

Drug analysis. Dasatinib concentrations in Opti-MEM cell culture medium and in plasma samples and brain homogenates were determined using a sensitive and specific liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay. Chromatography was carried out using a Solvent delivery system LC-20AD Prominence (Shimadzu), consisting of a binary pump, autosampler, degasser, and column oven. Chromatographic separations of the analytes were carried out on a Gemini C18 column, 50 × 2.0 mm internal diameter, 5 μm (Phenomenex). A mobile phase consisting of eluent A (10 mmol/L ammonium hydroxide in water) and eluent B (1 mmol/L ammonium hydroxide in methanol) was pumped through the column with a flow of 0.25 mL/min. The following gradient was used:

The respective retention time for dasatinib was 3.8 min. The mass spectrometric analyses were done using a Finnigan TSQ Quantum Ultra Triple Quadrupole Spectrometer equipped with an electrospray ion source (Thermo Fisher). The mass spectrometers were operating in positive electrospray selective multiple reaction monitoring, and three multiple reaction monitoring channels were monitored at unit resolution corresponding to dasatinib m/z 488 to 401, and the stable ¹³C isotope of imatinib m/z 498 to 394 (used as internal standard). Samples were pretreated by protein precipitation with acetonitrile, and the supernatants were diluted 1:1, v/v (sample extract: eluent A) before injection (10 μL).

Pharmacokinetic calculations and statistical analysis. Pharmacokinetic parameters were calculated by noncompartmental methods using the software package WinNonlin Professional version 5.0. The AUC was calculated using the trapezoidal rule, without extrapolating to infinity. Elimination half-lives (t_{1/2, el}) were calculated by linear regression analysis of the log-linear part of the plasma concentration-time curves. The peak plasma concentration (C_max) and the time of maximum plasma concentration (T_max) were estimated from the original data. Apparent clearance (CL_app or CL/F) was calculated by the formula CL_app = dose/AUC. The relative oral bioavailability (F_rel) was calculated by the formula F_rel = (AUC_oral × Dose_i.p.)/(AUC_i.p. × Dose_oral) × 100%. The two-sided unpaired Student's t test was used for statistical analysis. Data obtained with single and combination knockout (KO) mice were compared with data obtained with WT mice. Differences were considered statistically significant when P < 0.05. Data are presented as means ± SD.

In vitro transport of dasatinib. We used monolayers of the polarized canine kidney cell line MDCK-II and its subclones transduced with human ABCB1 or ABCC2 or murine Abcg2 to study transepithelial vectorial transport of 5 μmol/L dasatinib. Dasatinib was modestly but actively transported in the apical direction in the MDCK-II parental cell line (Fig. 1A), which is known to have relatively high expression of functional endogenous canine P-gp (40). Transport in the parental cells could be completely inhibited with elacridar, a potent inhibitor of P-gp (Fig. 1D), suggesting that indeed endogenous P-gp was responsible for dasatinib transport in the parental cells. In cells transfected with murine Abcg2 or human ABCB1 P-gp, transport of dasatinib was markedly increased, resulting in 10.8- and 8.4-fold higher transport ratios (r, defined in Materials and Methods) compared with parental cells, respectively (Fig. 1B and C). In cells transduced with human ABCC2, no transport of dasatinib was observed (Fig. 1E).

Effect of ABCG2 and P-gp on plasma pharmacokinetics of dasatinib in mice. Because our data indicated that both ABCG2 and P-gp have a profound effect on dasatinib transport in vitro, we further investigated their separate and combined effect in vivo. In the clinic, dasatinib is administered orally to cancer patients, and therefore, we first studied plasma dasatinib concentrations after oral administration of 10 mg/kg to WT, Abcb1a/1b^−/−, Abcg2^−/−, and Abcb1a/1b;Abcg2^−/− mice. Abcb1a/1b^−/− mice displayed a 1.7-fold higher AUC_oral compared with WT mice (P < 0.05; Fig. 2A; Table 1), and the maximal plasma levels in Abcb1a/1b^−/− mice were reached markedly faster than in WT mice. In WT mice, the relatively late C_max, ∼2 hours after oral administration, might partly be explained by enterohepatic circulation of dasatinib. The fact that this late C_max was absent in P-gp–deficient animals may indicate that P-gp restricts rapid absorption from the intestine and/or mediates substantial hepatobiliary excretion and thus contributes to enterohepatic circulation of dasatinib. Deficiency of only Abcg2 did not affect plasma pharmacokinetics after oral administration (Fig. 2A; Table 1). Abcb1a/1b;Abcg2^−/− mice had 2.0-fold higher AUC_oral (P < 0.01), and the C_max was 1.9-fold higher compared with WT mice (P < 0.05; Fig. 2A; Table 1). Collectively, these results indicate that P-gp, and not Abcg2, restricts the oral uptake of dasatinib in mice.

To study the plasma pharmacokinetics after systemic administration, dasatinib was given by i.p. injection at 5 mg/kg. At this dose, plasma concentrations were in the same range as in oral experiments with 10 mg/kg. Abcb1a/1b^−/− mice had slightly, but significantly, lower plasma concentrations 0.5 and 1 hour after administration, resulting in a 1.3-fold lower AUC_i.p. compared with WT mice (P < 0.05; Fig. 2B; Table 1). The AUC_i.p. values for Abcg2^−/− and Abcb1a/1b;Abcg2^−/− mice were not different from WT mice. These results indicate that both transporters do not strongly affect the systemic distribution and elimination of dasatinib. Because we do not know if the availability after i.p. administration is complete, we used the AUC_i.p. as a relative standard to calculate the relative oral bioavailability (F_rel) in the different mouse strains. WT and Abcg2^−/− mice displayed a relative oral bioavailability of ∼30% and P-gp deficiency increased the relative oral availability in both Abcb1a/1b^−/− and Abcb1a/1b;Abcg2^−/− mice to ∼68% (P < 0.01; Table 1).

Effect of ABCG2 and P-gp on brain accumulation of dasatinib in mice. As shown in Fig. 3, the relative brain accumulation, determined 6 hours after administration and corrected for the plasma AUC_0-6, was not different between Abcg2^−/− and WT mice, either after oral (Fig. 3A) or i.p. (Fig. 3B) administration. However, in Abcb1a/1b^−/− mice, the relative brain accumulation was 3.6-fold increased after oral and 4.8-fold higher after i.p. administration. Abcb1a/1b;Abcg2^−/− mice had the most pronounced increases in relative brain accumulation, 13.2-fold after oral and 22.7-fold after i.p. administration. The relative brain accumulation values as well as the uncorrected brain concentrations are listed in Table 2. Correction for plasma concentration at t = 6 hours yielded qualitatively the same results for relative brain accumulation as correction for AUC_0-6 (data not shown). These results show that P-gp almost completely controls the dasatinib brain accumulation no matter whether Abcg2 is present or not. However, in Abcb1a/1b^−/− mice, Abcg2 can partly but not completely take over the function of P-gp at the blood-brain barrier. These results further show that when both transporters are absent, brain accumulation of dasatinib is highly increased.

Effect of the dual P-gp and ABCG2 inhibitor elacridar on dasatinib brain accumulation. Because both P-gp and ABCG2 markedly restricted the brain accumulation of dasatinib, we investigated if inhibition of both efflux transporters at the blood-brain barrier would result in an increased brain accumulation of dasatinib. We used the dual P-gp and ABCG2 inhibitor elacridar. Figure 4A shows that the dasatinib plasma concentration, 60 minutes after i.v. administration, was 1.7-fold higher in Abcb1a/1b;Abcg2^−/− compared with WT mice (P < 0.05). In the presence of elacridar, plasma concentrations of dasatinib in WT and Abcb1a/1b;Abcg2^−/− mice were not different anymore, suggesting that elacridar inhibits the systemic elimination of dasatinib via P-gp and/or Abcg2. In contrast to the modest effect on plasma concentrations, elacridar drastically increased brain concentration in WT mice by 10-fold (P < 0.001; Fig. 4B). Brain concentrations in Abcb1a/1b;Abcg2^−/− mice seemed also somewhat increased when elacridar was applied, but the difference was not statistically significant (P = 0.11; Fig. 4B) and probably reflects the modestly higher plasma levels (Fig. 4A). Figure 4C shows the brain-to-plasma ratios, which were calculated to correct for the differences in plasma concentrations. For WT mice, the ratio was 4.4-fold increased with elacridar, resulting in a similar dasatinib brain accumulation as observed for Abcb1a/1b;Abcg2^−/− mice. For Abcb1a/1b;Abcg2^−/− mice, elacridar did not affect brain-to-plasma ratios. Overall, these results show that elacridar can completely inhibit P-gp and ABCG2 at the blood-brain barrier, leading to highly increased dasatinib concentrations in the brain.

In this study, we show that the second-generation tyrosine kinase inhibitor dasatinib is efficiently transported in vitro by P-gp and Abcg2. We extended these observations in vivo and found that relative oral availability is limited by P-gp. We further show that the brain accumulation of dasatinib is primarily restricted by P-gp but that Abcg2 can partly take over this protective function at the blood-brain barrier. Consequently, when both efflux transporters are absent, brain uptake of dasatinib is highly increased. And finally, we show that the brain accumulation of dasatinib in WT mice can be markedly increased by applying the dual P-gp and ABCG2 inhibitor elacridar.

It was recently reported that the cellular accumulation of dasatinib in P-gp–overexpressing or ABCG2-overexpressing leukemic cells was significantly lower than in control cells (32). In addition, active transport of dasatinib in P-gp–overexpressing MDCK-II cells was recently shown (33). Our in vitro data on the transport of dasatinib by human P-gp are consistent with these observations, and we further show that, in addition to the reported transport by human ABCG2 (32), also mouse Abcg2 is an efficient transporter for dasatinib. Based on the reported in vitro data and our own in vitro results, we further investigated the effect of P-gp and Abcg2 on the pharmacokinetics of dasatinib. We found that Abcb1a/1b^−/− mice had 1.7-fold higher plasma AUC_oral than WT mice. In a recent study, no difference in AUC_oral between WT and P-gp–deficient mice was observed (41). However, the limited number of animals used in that study (n = 3) may have resulted in substantial variation (the variation was not reported). In addition, we know from experience that variation in absorption after oral administration can be markedly reduced by food deprivation for 3 hours before administration. We did so and we used five animals per group, resulting in more statistical power to detect differences in AUC_oral.

We found no difference in AUC_oral between Abcg2^−/− and WT mice, despite our observation that dasatinib is a good Abcg2 substrate in vitro. Moreover, Abcb1a/1b;Abcg2^−/− mice had a similar AUC_oral compared with Abcb1a/1b^−/− mice, indicating that P-gp, and not Abcg2, restricts the oral uptake of dasatinib in mice. We have shown earlier that there is not a higher level of P-gp RNA in the intestine of Abcg2^−/− mice,³

so possible P-gp up-regulation cannot explain the lack of effect of Abcg2. We further show that the relative oral bioavailability of dasatinib of ∼32% in WT mice was more than doubled to ∼68% in P-gp–deficient mice. This might be clinically relevant because dasatinib is taken orally by cancer patients, and intestinal uptake in humans might be limited by P-gp as well.

In addition to the effect of P-gp on the relative oral availability, we show that P-gp, together with ABCG2, drastically restricts the brain accumulation of dasatinib in mice. It is interesting to note that P-gp seems to have a dominant role compared with Abcg2 in both intestine and blood-brain barrier, whereas in vitro both transporters seem to efficiently transport dasatinib. A possible explanation of this apparent discrepancy could be that Abcg2 is more lowly expressed in intestine and blood-brain barrier than P-gp, something that is difficult to assess with the currently available tools. This would make P-gp the dominant player, and, at least in the blood-brain barrier, only when P-gp is absent the contribution of Abcg2 becomes visible. As mentioned above, we have shown earlier that there is not a higher level of P-gp RNA in Abcg2 KO intestine. Although it therefore seems unlikely, we cannot completely exclude that there is increased expression of blood-brain barrier P-gp in Abcg2 KO mice. Be that as it may, it is clear that both P-gp and Abcg2 can contribute to decreased brain accumulation of dasatinib. The relevance of this finding may be put in perspective by the recent findings of Porkka et al. (30), who showed that dasatinib has antitumor activity in a mouse model of intracranial CML. Dasatinib also induced substantial responses in patients with CNS Ph+ leukemia and showed a considerably higher CNS accumulation than imatinib in mice and humans (30). Interestingly, imatinib failed to show any antitumor activity in their mouse model of intracranial CML, which is consistent with several clinical and preclinical findings (22, 28, 42, 43). This difference between dasatinib and imatinib cannot simply be explained by the fact that brain accumulation of imatinib is limited by P-gp and ABCG2 (21, 22, 27, 42, 43). In fact, our data show that P-gp and Abcg2 also profoundly restrict the brain accumulation of dasatinib in mice. Moreover, compared with other agents that have a good accumulation of the blood-brain barrier, such as cytarabine, the CNS accumulation of dasatinib is still low (44). Therefore, other factors, such as its far higher potency (325-fold) compared with imatinib (11), may explain the antitumor effects of dasatinib within the CNS.

We further showed that the brain accumulation of dasatinib in WT mice could be markedly increased with the dual P-gp and ABCG2 inhibitor elacridar. Moreover, WT mice pretreated with elacridar had similar dasatinib brain-to-plasma ratios as observed for Abcb1a/1b;Abcg2^−/− mice that received either vehicle or elacridar. These results suggest that elacridar can fully inhibit P-gp and ABCG2 at the blood-brain barrier. We realize that results obtained in mice cannot simply be transferred to the clinical situation. However, we believe that our observations provide a rationale for combining dasatinib with a dual P-gp and ABCG2 inhibitor to increase the dasatinib brain accumulation and further improve its therapeutic efficacy in patients with CNS Ph+ leukemia. Clearly, further preclinical and clinical research is warranted to assess the feasibility of this approach.

No potential conflicts of interest were disclosed.

We thank Evita van de Steeg and Marijn Vlaming for critical reading of the manuscript.