Assessment of contribution of BCRP to intestinal absorption of various drugs using portal‐systemic blood concentration difference model in mice

Abstract Prediction of the intestinal absorption of new chemical entities (NCEs) is still difficult, in part because drug efflux transporters, including breast cancer resistance protein (BCRP) and P‐glycoprotein (P‐gp), restrict their intestinal permeability. We have demonstrated that the absorptive quotient (AQ) obtained from the in vitro Caco‐2 permeability study would be a valuable parameter for estimating the impact of BCRP on the intestinal absorption of drugs. In this study, in order to assess the correlation between the in vitro AQ for BCRP and in vivo contribution of BCRP on drug absorption, we evaluated the oral absorption of various compounds by portal‐systemic blood concentration (P‐S) difference method in wild‐type (WT), Bcrp(−/−), and Mdr1a/1b(−/−) mice. In addition, we also calculated a rate of BCRP contribution (Rbcrp). Ciprofloxacin and nitrofurantoin showed the low Rbcrp value (0.05 and 0.15), and their apparent fractions of intestinal absorption in WT mice were 46.5% and 63.7%, respectively. These results suggest that BCRP hardly affects their intestinal absorption in mice. On the other hand, the apparent fraction of intestinal absorption of topotecan and sulfasalazine was significantly lower in WT mice than in Bcrp(−/−) mice. Moreover, their Rbcrp values were 0.42 and 0.79, respectively, indicating the high contribution of BCRP to their oral absorption. Furthermore, in vivo Rbcrp calculated in this study was almost comparable to in vitro AQ obtained from Caco‐2 permeability study. This study provides useful concepts in assessing the contribution of BCRP on intestinal absorption in drug discovery and development process.


| INTRODUC TI ON
Oral drug administration has been most frequently used in clinical because it has several advantages against other administration routes, such as easy to use, high safety, good patient compliance, and low cost. Therefore, in the development of new drug, it is very important to make many new chemical entities (NCEs) to be an orally available dosage form.
However, most of the NCEs, which have been discovered recently, tend to have disadvantageous characteristics for oral administration, that is, poor water solubility, low membrane permeability, and substrate for various efflux drug transporters. In particular, at the early drug discovery stage, it is important to estimate whether each NCE is recognized by drug efflux transporter and its intestinal permeability is restricted.
In drug efflux transporters, breast cancer resistance protein (BCRP; ABCG2) expression level in human intestine has been reported to be equal to or even higher than that of MDR1. 1,2 BCRP has one adenosine 5'-triphosphate (ATP)-binding cassette and six transmembrane domains and is, therefore, so called a half-ABC transporter, which forms homodimers to obtain functional activity. 3,4 Since Bcrp(−/−) mice were developed by Schinkel et al, 5 a lot of in vivo studies using Bcrp(−/−) mice have been carried out to evaluate the effect of BCRP on the oral absorption of drugs. [6][7][8] In most of these reports, systemic plasma concentration of drugs after oral administration was compared between Bcrp(−/−) mice and wild-type (WT) mice. In case of BCRP substrate drug, its bioavailability (BA) in Bcrp(−/−) mice is tended to be higher than that in WT mice, because Bcrp is highly expressed in liver and kidney, relatively high expressed in small intestine. [6][7][8][9] We have evaluated the Caco-2 permeability of various BCRP and/ or P-glycoprotein (P-gp) substrates and defined an absorptive quotient (AQ) for estimating the specific contribution of BCRP to intestinal permeability of drugs. This in vitro assay system using Caco-2 cells for calculating AQ might be an efficient approach to estimate the oral absorption of NCEs, particularly with respect to the contribution of BCRP. In order to demonstrate this expectation, it is required to investigate whether the estimated contribution of BCRP to intestinal permeability from in vitro study correlates with the in vivo study.
In this study, we evaluated the contribution of BCRP, as well as P-gp, which is a representative drug efflux transporter, to intestinal drug absorption using a recirculatory model for portal-systemic blood concentration (P-S) difference method ( Figure 1) in Bcrp(−/−) and Mdr1a/1b(−/−) mice. 10,11 This method was developed to separately evaluate the rate and extent of absorption from the gastrointestinal tract into the portal system and disposition of a drug in the body. We here applied this method for various model compounds, and estimated the apparent local absorption ratio from the gastrointestinal tract into the portal system (F a F g ) in WT, Bcrp(−/−), and Mdr1a/1b(−/−) mice. Then, we calculated the in vivo AQ values for BCRP and P-gp, and ratios of contribution (R), which indicate the contribution of BCRP and P-gp on the intestinal absorption. Furthermore, we also assessed the correlation of in vivo AQ with in vitro AQ obtained from in vitro Caco-2 permeability studies.

| Pharmacokinetic studies
All the mice were fasted overnight with free access to tap water. In the intravenous administration studies, model compounds were administered via the tail vein at doses of 1 mg/kg (n = 3). Following administration, blood samples were collected from the abdominal vein of the anesthetized mice at 0.083, 0.17, 0.5, 1, 2, 4, and 8 hours. In the oral administration study, ciprofloxacin, nitrofurantoin, topotecan, and sulfasalazine were administered by gavage at a dose of 1, 2, 1, and 5 mg/kg, respectively (n = 2).
Following administration, blood samples were taken from the portal and abdominal veins of the anesthetized mice at 0.083, 0.17, 0.5, 1, 2, 4, and 8 hours. The plasma samples were separated by centrifugation at 14 000g for 10 minutes at 4°C and stored at −30°C until analysis.

| Determination of blood/plasma concentration ratio (R b )
The model compounds were spiked into fresh whole blood collected from FVB mice at final concentrations of 1 µg/mL. After the incubation at 37°C for 15 minutes, the plasma samples were obtained by centrifugation at 14 000g for 10 minutes at 4°C. Similarly, the model compounds were added to plasma, and reference blood samples were obtained according to the same procedure. These concentrations of drugs in each sample were analyzed using HPLC (C B and C P , respectively). R b value was calculated by dividing C B by C P .

| Analytical methods
Ciprofloxacin and nitrofurantoin were extracted from the plasma with dichloromethane and ethyl acetate, respectively. After organic layer was evaporated at 60°C, the resultant residues were dissolved in a mobile phase. For the determination of topotecan and sulfasalazine, plasma samples were mixed with acetonitrile, centrifuged at 750 g for 10 minutes at 4°C, and the supernatants were collected. After the evaporation of the supernatants, the residues were dissolved in a mobile phase, and acidified with phosphoric acid for topotecan. All drugs were analyzed by

| Pharmacokinetic analysis
Elimination rate constant (k e ) was determined by the least squares re- where AUMC iv and AUC iv mean AUMC and AUC after intravenous administration, respectively.
Absorption rate constant (k a ) after oral administration was calculated by the nonlinear least squares fitting with program MULTI. 12 Apparent F a F g (F a , absorption ratio; F g , intestinal availability) in P-S difference model was calculated by Eq.4: where Q pv is the portal blood flow (106.6 mL/min/kg, 13,14 AUC pv is the AUC in portal vein, and AUC sys is the AUC in systemic circulation). BA was calculated by Eq.5: where AUC oral is AUC after oral administration. Dose iv and Dose oral are administered dose in the intravenous and oral administration study,

respectively.
Hepatic availability (F h ) was calculated by Eq.6: In vivo AQ was defined by the following equation using k a in WT, Bcrp(−/−), and Mdr1a/1b(−/−) (k a,WT , k a,BCRP , k a,P-gp ) ( Figure 2): In addition, we defined a rate of contribution (R), which indicates the contribution of P-gp or BCRP on the intestinal absorption, by the following equation: where F a F gWT , F a F gBcrp , and F a F gP-gp are F a F s in WT, Bcrp(−/−), and Mdr1a/1b(−/−) mice, respectively.

| Ciprofloxacin
We also evaluated the plasma concentration of ciprofloxacin fol- Mdr1a/1b mice

| Nitrofurantoin
The plasma concentration-time curve of nitrofurantoin after intravenous and oral administration in WT, Bcrp(−/−), and Mdr1a/1b(−/−) mice is shown in Figure 4, and the corresponding pharmacokinetic parameters are listed in Table 2. The  On the other hand, there were no significant differences in AUC sys and CL tot values between bcrp knockout (KO) and p-gp KO mice, indicating that BCRP and p-gp hardly affect the elimination process of nitrofurantoin.

| Topotecan
The plasma concentration-time profiles of topotecan after intravenous and oral administration in WT, Bcrp(−/−), and Mdr1a/1b(−/−) mice were also investigated ( Figure 5, Table 3). The k a value of topotecan in WT mice was 3.18 per hour, indicating that topotecan is rapidly absorbed from the upper intestine after oral administration. In addition, its BA in WT mice was approximately 37%, and this is similar to the human BA (40%). 15  higher than that in WT mice (57%), and its k a value was 5.18 per hour.
These results indicate that the intestinal absorption of topotecan in mice is dominated by BCRP.
The higher level of AUC sys and slightly lower CL tot value were observed after intravenous injection of topotecan in Bcrp(−/−) mice, compared with WT mice. These results suggest that BCRP is also involved in the elimination process of topotecan.
On the other hand, there were no differences in the pharmacoki-

netics of topotecan between oral and intravenous administration in
Mdr1a/1b(−/−) mice, indicating that p-gp has no effect on the intestinal absorption and elimination of topotecan.

| Sulfasalazine
The time course of plasma concentration of sulfasalazine after intravenous and oral administration in WT, Bcrp(−/−), and Mdr1a/1b(−/−) mice is shown in Figure 6, and the corresponding pharmacokinetic  parameters are given in Table 4. The F a F g and BA values in WT mice were estimated to be 16.9% and 10.2%, respectively. These are almost similar to the human F a F g and BA (12% and < 15%, respectively). [17][18][19] These results indicate that the intestinal absorption of  was hardly observed, and the V dss was low (0.19 L/kg). Taken together, it is considered that the late t max and small V dss values cause the low k a value of sulfasalazine despite its F a F g value was approximately 100%.
On the other hand, the F a F g value of sulfasalazine in Mdr1a/1b(−/−) mice (F a F g : 30%) was also higher than that in WT mice. However, the influence of p-gp on the intestinal absorption of sulfasalazine is considered not to be so high compared with BCRP. In addition, the C 0 and V dss values were not different between Mdr1a/1b(−/−) mice and WT mice.
Sulfasalazine is degraded to sulfapyridine and 5-aminosalicylic acid by bacteria in the large intestine. 17 Although sulfapyridine is well absorbed from the intestine, its plasma concentration in Bcrp(−/−) mice was much less than that in WT and Mdr1a/1b(−/−) mice (data not shown). This may be because sulfasalazine is highly absorbed from the intestine without degradation in Bcrp(−/−) mice, whereas sulfasalazine is degraded to sulfapyridine because of its low F a F g in WT and Mdr1a/1b(−/−) mice.
We summarized the k a values and calculated AQ bcrp and AQ P-gp values of model drugs in Table 5. In sulfasalazine, AQ value could not be estimated because the k a value was much lower in Bcrp(−/−) mice despite its F a F g value was significantly higher than WT mice.
Then, we calculated the rate of contribution (R) value on the intestinal absorption using F a F g values in each mice (Table 6).
Ciprofloxacin and nitrofurantoin showed low R bcrp and R p-gp values, indicating that the contribution of both BCRP and P-gp to their intestinal absorption would be little. On the other hand, topotecan and sulfasalazine showed relatively high R bcrp in contrast to low R pgp . These results indicate that BCRP mainly acts as a barrier to their intestinal absorption.

| Evaluation of the in vitro-in vivo correlation
We have demonstrated that the R value would be a valuable alternative parameter to in vivo AQ for estimating the contribution of efflux transporters to drug absorption. Therefore, we investigated the relationship between in vivo R and in vitro AQ estimated from Caco-2 permeability in our previous study. We have clarified that the drugs,

| D ISCUSS I ON
In this study, we defined the R value for estimating the quantitative contribution of BCRP and P-gp to the intestinal absorption of drugs,  pharmacokinetic parameters, including AUC, F a F g , and BA, of several drugs by P-S difference method in rats, and they have demonstrated that these parameters can be more strictly defined than those by the simplified models. 20,21 In addition, the pharmacokinetic parameters of drugs evaluated by P-S difference method were in good accordance with the experimental values obtained from other recirculatory models, such as bile duct cannulation method. Moreover, P-S difference method can define the drug pharmacokinetics on a physiological basis without significant experimental variability. Based on these reasons, we used P-S difference method here to determine the local drug absorption.

TA B L E 5 Data summary for k a and in vivo AQ values
In ciprofloxacin, P-gp, not BCRP, was likely to mainly contribute to its intestinal absorption (Figure 3, Tables 1 and 5). However, its secretion. 23 However, the main elimination pathway of ciprofloxacin is urinary excretion in human, and BCRP has been reported not to be expressed in human kidney. 24 Therefore, it is conceivable that the drug-drug interaction in BCRP is unlikely to occur through the elimination process.
Then, it is suggested that both BCRP and P-gp affect the intestinal absorption of nitrofurantoin in mice (Figure 4, Tables 2 and 5).
However, their contribution to the intestinal absorption could be ignored because nitrofurantoin showed high F a F g value in WT mice and there are no clinical reports about the involvement of P-gp in its absorption.
On the other hand, the intestinal absorption of topotecan was highly affected by BCRP ( Figure 5, Tables 3 and 5), although it has been reported that the distribution of topotecan is restricted by P-gp, rather than BCRP, in brain. 25 Moreover, the possibility of the involvement of BCRP in the elimination process was also demonstrated in the present study. It has been reported that the urinary excretion is the main elimination pathway of topotecan in mice and human. 26,27 On the other hand, Jonker et al have shown that GF120918, a BCRP inhibitor, decreases the biliary excretion of topotecan after intravenous administration, while its urinary excretion is hardly affected by GF120918. 28 Taking these findings into consideration, it is conceivable that the involvement of BCRP in biliary excretion of topotecan results in its lower CL tot in Bcrp(−/−) KO mice.
In human clinical studies, it has been demonstrated that topotecan shows the poor BA after oral administration (about 40%), and it is hardly metabolized. 16 However, the BA of topotecan has been reported to significantly increase to 100% when GF120918 is orally coadministered. 30,31 Furthermore, Sparreboom et al have shown that the oral BA of topotecan is 1.3-fold higher in patients who are heterozygous variant for the BCRP single-nucleotide polymorphism (SNP) than in patients with the normal BCRP. 31 These results are in accordance with our present results. Taken together, BCRP would act as a barrier for oral absorption of topotecan in human.
Similar to topotecan, the intestinal absorption of sulfasalazine was highly influenced by BCRP, rather than P-gp ( Figure 6, Table 3 and 5). Interestingly, the smaller V dss was observed in Bcrp(−/−) mice than WT mice, despite the V dss is assumed to become higher in investigation is required to clarify this event. In human study, the AUC value of sulfasalazine after oral administration in patients who are heterozygous variant for the BCRP SNP has been reported to be approximately 2-fold higher than that in patients with the normal BCRP. 35 Since there are no differences in the elimination of sulfasalazine between those patients, BCRP would affect the oral absorption of sulfasalazine not only in mice but also in human.
Thus, we have revealed that in vivo pharmacokinetic parameters of topotecan and sulfasalazine, which showed relatively high R bcrp value, could well reflect the human situation. Therefore, we finally  Table 7). Ciprofloxacin and nitrofurantoin, which showed low R bcrp value in vivo, showed low AQ BCRP values in vitro. In contrast, topotecan and sulfasalazine, which have been shown to be greatly influenced their oral absorption by BCRP in human, showed both in vivo R bcrp and in vitro AQ BCRP of more than 0.4. However, their absolute values were different in each drug, suggesting that the substrate recognition property of BCRP differ between mice and human. On the other hand, in vitro AQ BCRP value of topotecan was comparable to human AQ BCRP value estimated from the clinical data (0.61 vs 0.58). 30,31 In conclusion, we demonstrate that the accurate prediction of the contribution of BCRP in human intestinal drug absorption could TA B L E 7 Comparison between in vitro AQ and in vivo rate of contribution (R) for BCRP and P-gp

D I SCLOS U R E
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. This work did not involve studies with human subjects.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.