The following compounds were purchased commercially from the sources indicated: allopurinol, benzbromarone, cyclosporine, D-fructose, elacridar, furosemide, hydrochlorothiazide, nicotinic acid, oxypurinol, rosuvastatin calcium salt, salicylic acid, 4-hydroxy chalcone (Wako Fine Chemical, Osaka, Japan); atorvastatin, chlorothiazide, febuxostat, mizoribine, pyrazinecarboxylic acid, ribavirin, tacrolimus, xylitol (Tokyo Chemical Industry, Tokyo, Japan); ethambutol, losartan (LKT Laboratories, St Paul, MN, USA); fenofibrate, probenecid, sulfasalazine, Ko143, ATP, AMP, creatine phosphate disodium salt tetrahydrate, creatine phosphokinase type I from rabbit muscle (Sigma-Aldrich, St. Louis, MO, USA); pyrazinamide (ACROS ORGANICS, Geel, Belgium); theophylline (Nacalai Tesque, Kyoto, Japan); and topiroxostat (MedChem Express, Princeton, NJ, USA). The [8-¹⁴C]-uric acid (53 mCi/mmol) was from American Radiolabeled Chemicals (St. Louis, MO, USA). All other chemicals used were commercially available and of analytical grade.

Human embryonic kidney 293 cell-derived 293A cells were purchased from Life Technologies (Carlsbad, CA, USA) and cultured in DMEM (Nacalai Tesque) supplemented with 10% fetal bovine serum (Biowest, Nuaillé, France), 1% penicillin/streptomycin, 2 mM L-glutamine (Nacalai Tesque), and 1 × Non-Essential Amino Acid (Life Technologies) at 37°C in an atmosphere of 5% CO₂ as described previously (Toyoda et al., 2016a). All experiments were carried out with 293A cells at passages 10–20. To express human ABCG2 (NM_004827) fused with Myc-tag at its N-terminus (Myc-ABCG2) and EGFP (control), we used Myc-ABCG2 and EGFP-expressing adenoviruses constructed in our previous study (Ito et al., 2015), respectively. To express the URAT1, open reading frame of URAT1 (NM_144585.3) was cloned into a pcDNA3.1(+) vector (Life Technologies) with a FLAG tag at its N-terminus. To express mouse Abcg2 and EGFP (control), open reading frames of mouse Abcg2 (NM_011920) and EGFP were inserted into a pcDNA3.3 vector (Life Technologies), respectively.

The FVB/NJcl WT mice were purchased from CLEA Japan (Tokyo, Japan). The Abcg2 KO mice (Ichida et al., 2012) had been maintained in our laboratory as described previously. All the animals were housed in temperature- and humidity-controlled animal cages with a 12-h dark-light cycle, and free access to water and standard animal chow (MF, Oriental Yeast Company, Tokyo, Japan) as described previously (Ito et al., 2014; Takada et al., 2015). In the present study, all the in vivo experiments were conducted with 6–8-week-old male mice. The animal research protocols used in the present study were approved by the Animal Studies Committee of the University of Tokyo.

The membrane vesicles were prepared from 293A cells infected with EGFP- or Myc-ABCG2- expressing adenovirus as described previously (Hayashi et al., 2005). Similarly, mouse Abcg2-expressing plasma membrane vesicles were prepared from 293A cells 48 h after the plasmid transfection using Polyethylenimine “MAX” (PEI-MAX) (1 mg/mL in milliQ water, pH 7.0; Polysciences, Warrington, PA, USA) as described previously (Stiburkova et al., 2016). The [8-¹⁴C]-urate transport assay with the ABCG2/Abcg2-expressing membrane vesicles was performed using a rapid filtration technique (Matsuo et al., 2009; Toyoda et al., 2016b). The urate transport activity was calculated as an incorporated clearance (μL/mg protein/min): (incorporated level of urate [DPM/mg protein/min]/urate level in the incubation mixture [DPM/μL]). ATP-dependent urate transport was calculated by subtracting the urate transport activity in the absence of ATP from that in the presence of ATP.

The IC₅₀ values of each test compound against the urate transport by ABCG2 were calculated as follows. First, the urate transport activities were measured in the presence of SUA-affecting drugs at several concentrations. The transport activities were expressed as a percentage of control (100%). Then, the calculated values were fitted to the following formula using the least-squares methods (Powles et al., 2007) with the Excel 2007 (Microsoft, Redmond, WA, USA) program:

where, E_max is the maximum effect, C is the drug concentration, and n is the sigmoid-fit factor. The unbound concentrations of each drug in human plasma were calculated based on the general information provided by pharmaceutical industries. The unbound fraction in human plasma (f_u) was calculated by subtracting the ratio of bound drugs from total drugs. The maximum unbound concentration of each drug in human plasma (f_uC_max) was calculated by multiplying f_u and the reported C_max together.

The 293A cells were seeded on a 12-well plate (Becton Dickenson & Co., Franklin Lakes, NJ, USA) at 2 × 10⁵ cells/well. Twenty-four hours after seeding, the cells were transiently transfected with the FLAG-URAT1 pcDNA3.1(+) vector or empty vector using PEI-MAX (0.4 μg plasmid/4 μL PEI-MAX/100 μL serum free DMEM/2 × 10⁵ cells) as described previously (Stiburkova et al., 2016). Forty-eight hours after the transfection, the cells were washed twice with Buffer T2 (125 mM Na-gluconate, 4.8 mM K-gluconate, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 1.3 mM Ca-gluconate, 25 mM HEPES, 5.6 mM D-glucose, and pH 7.4) and preincubated in Buffer T2 for 15 min at 37°C. Then, the buffer was exchanged to fresh Buffer T2 containing 2 μM [8-¹⁴C]-urate, and the cells were further incubated for the indicated periods. Subsequently, the cells were washed with ice-cold Buffer T2 twice and then lysed with 500 μL 0.2 M NaOH on ice with gently shaking for 1 h. The resulting lysates were transferred to 1.5-mL tubes, neutralized with 100 μL 1 M HCl, and then the radioactivity were measured using a liquid scintillator. The protein concentrations were determined using a bicinchoninic acid assay kit (Life Technologies) according to the manufacturer’s instruction. The urate transport activity was calculated as the incorporated clearance (μL/mg protein/min): (incorporated level of urate [DPM/mg protein/min]/urate level in the incubation mixture [DPM/μL]). URAT1-dependent urate transport activity was calculated by subtracting the urate transport activity of mock cells from that of the URAT1-expressing cells.

The expression of Myc-ABCG2 on membrane vesicles and FLAG-URAT1 in transiently transfected-293A cells was confirmed using western blot analysis. To obtain the cell lysate samples of URAT1-expressing 293A cells, 48 h after the transfection, the cells were washed twice with phosphate-buffered saline without Ca²⁺ and Mg²⁺ and then dissolved in RIPA buffer (0.1% SDS, 0.5% deoxycholate, 1% NP40, 150 mM NaCl, 50 mM Tris-HCl, and pH 7.4) containing protease inhibitor cOmplete, EDTA-free (Roche, Basel, Switzerland). After centrifugation at 15,000 × g for 10 min at 4°C, the resulting supernatants were collected as the cell lysates. The membrane vesicles prepared from the adenovirus-infected 293A cells or cell lysates of vector-transfected 293A cells were subjected to 8.5% SDS-polyacrylamide gel electrophoresis and transferred to PVDF membranes (Immobilon, Millipore Corporation, Billerica, MA, USA). After blocking with 3% bovine serum albumin in 0.05% Tween 20 containing Tris-buffered saline for 1 h at room temperature, the membranes were treated with following antibodies as described previously (Matsuo et al., 2009); mouse anti-ABCG2 antibody (BXP-21) (Santa Cruz Biotechnology, Santa Cruz, CA, USA) (1:1000), a mouse anti-FLAG M2 antibody (Sigma Aldrich) (1:1000), a rabbit anti-Na⁺/K⁺-ATPase antibody (Santa Cruz Biotechnology) (1:1000), anti-mouse IgG or anti-rabbit IgG antibody labeled with horseradish peroxidase (GE Healthcare, Piscataway, NJ, USA) (1:2500). The immunoblotted membranes were treated with ECL Prime (GE Healthcare) and analyzed using a Chemidoc XRS (Bio-Rad Laboratories, Richmond, CA, USA).

We conducted an in vivo Abcg2 inhibition test to examine the effect of febuxostat on the intestinal absorption of sulfasalazine, an ABCG2 substrate. Febuxostat was dissolved in 100 mM sodium bicarbonate buffer, pH 10. Prior to the experiment, the mice were fasted overnight, and then they were orally administered febuxostat by gavage at a dose of 150 mg/kg body weight (b.w.) with the expectation of complete inhibition of intestinal Abcg2. Twenty minutes later, sulfasalazine (20 mg/kg b.w.) was orally administered by gavage. At the indicated periods, the mice were anesthetized with diethyl ether and blood was collected from the jugular veins using heparinized syringes, followed by centrifugation at 3,000 × g for 10 min. The resulting supernatant (plasma) was collected and stored at -80°C until the LC/MS/MS analysis was performed.

The collected plasma was deproteinized with a fourfold volume of methanol containing 2 μg/mL 4-hydroxy chalcone as an internal standard. After vortexing for 10 min, the samples were centrifuged at 20,000 × g for 15 min at 4°C. Then, the supernatants were analyzed using the LC/MS/MS technique.

The LC/MS/MS analysis was conducted using an ultra-performance LC system connected to a Xevo TQ-S mass spectrometer (Waters Corporation, Milford, MA, USA). The samples were separated using a 1.7-μm particle ACQUITY C18 column (2.1 mm × 100 mm, Waters), maintained at 50°C, under gradient mobile phase conditions with a mixture of 0.1% formic acid in water and 0.1% formic acid in acetonitrile as solvents (0–1 min 70:30 v/v, 1–3 min 70:30 to 2:98 v/v, 3–5 min 2:98 v/v, and 5–6.5 min 70:30 v/v) with a flow rate of 0.3 mL/min. The separated samples were introduced into an MS in the positive and negative electrospray ionization mode for sulfasalazine and 4-hydroxy chalcone, respectively. Each compound was quantified in the multiple reactions monitoring mode (399.24 > 119.02, Cone 20 V, Collision 46 eV for sulfasalazine; 223.15 > 117.00, Cone 50 V, and Collision 36 eV for 4-hydroxy chalcone).

The AUC of sulfasalazine was calculated as the total area of the trapezoids formed by the points of the concentration and time in the concentration-time plots.

All statistical analyses were performed using the Excel 2007 program. Significant differences were identified using a two-way ANOVA followed by Tukey-Kramer post hoc test. The values were considered significant when P < 0.05.

To examine the effects of 25 SUA-affecting drugs (Table 1) on ABCG2 function, we measured the urate transport activity of ABCG2 in the presence of each SUA-affecting drug. Expression of ABCG2 in the plasma membrane vesicles was confirmed using western blot analysis (Figure 1A). According to our previous reports (Matsuo et al., 2009; Stiburkova et al., 2016), we used 20 μM of urate in reaction mixtures in the following experiments. First, we examined the time-dependent increase in ATP-dependent urate transport into the ABCG2-expressing plasma membrane vesicle (Figure 1B). The urate transport activities of ABCG2 were higher than those of the mock (EGFP), and linearly increased for 20 min. Therefore, we examined the effect of SUA-affecting drugs on ABCG2 for 10 min at the indicated concentrations in Table 1. In this experiment, the concentrations of each drug (3–1000 μM) were determined as the maximum soluble levels in the transport buffer.

The in vitro screening result revealed that 10 of the 25 drugs investigated decreased the urate transport activity of ABCG2 to <20% of that of the vehicle control (Figure 1C). Although 12 drugs tended to increase the urate transport activity of ABCG2, their effects were slight compared to the drastic inhibitory effects of the 10 drugs. Therefore, we focused on the most active 10 drugs in the further analyses.

Since the in vitro screening was conducted in the presence of maximum concentrations of each drug, we subsequently estimated the clinical effects of the selected 10 drugs. First, to determine the IC₅₀ values of each drug against the urate transport activity of ABCG2, we measured the urate transport activities in the presence of several concentrations of each drug (Figure 2). Then, the IC₅₀ values were calculated from the results obtained using the least-squares methods (Table 2). Among the 10 drugs we examined, three (benzbromarone, topiroxostat and febuxostat) potently inhibited ABCG2 with IC₅₀ values at submicromolar concentrations (0.20, 0.18, and 0.027 μM, respectively). Indeed, these values were considerably lower than those of four drugs that have been reported to interact with ABCG2, atorvastatin (4.3 μM), cyclosporine (3.2 μM), tacrolimus (3.1 μM) and rosuvastatin (2.3 μM).

To further characterize the potent inhibition of ABCG2 by febuxostat, we compared the activity of ABCG2 in the presence of febuxostat and two well-known ABCG2 inhibitors; Ko143, a fumitremorgin C analog, and elacridar (GF120918). In the presence of 100 nM of each compound, the urate transport activities of ABCG2 were 20 ± 3% with febuxostat, 75 ± 10% with Ko143 and 92 ± 10% with elacridar compared with the vehicle. These results indicate that the inhibitory effect of febuxostat against ABCG2 is stronger than those of these well-known ABCG2 inhibitors.

Next, we estimated the f_uC_max based on the general information from pharmaceutical industries as described in Section “Materials and Methods” (Table 2), and calculated the ratio of f_uC_max/IC₅₀, as an indicator of the possible in vivo clinical inhibition of ABCG2 by each drug. A high f_uC_max/IC₅₀ value suggests that the drug has the potential to inhibit ABCG2 in humans. Among the 10 drugs we examined, benzbromarone and febuxostat had a relatively high f_uC_max/IC₅₀ value (benzbromarone and febuxostat, 0.99 and 3.4, respectively). Since febuxostat had the highest f_uC_max/IC₅₀ value and was expected to inhibit ABCG2 in humans, we proceeded to further analyze this drug.

Our in vitro findings strongly suggested that febuxostat would inhibit ABCG2 in vivo. To examine this possibility, we investigated the inhibitory effect of febuxostat on the intestinal absorption of an ABCG2 substrate drug in WT and Abcg2 KO mice. Prior to the in vivo analyses, we confirmed that febuxostat inhibited mouse Abcg2 in vitro in a similar manner to human ABCG2 as described above. The IC₅₀ value of febuxostat against the urate transport activity of mouse Abcg2 was 35 nM (Figure 3A), which was comparable to that of human ABCG2 (IC₅₀, 27 nM, Table 2), indicating that febuxostat inhibited mouse Abcg2 as strongly as it did human ABCG2.

To examine whether febuxostat inhibited Abcg2 in vivo, we focused on its effect on the absorption of sulfasalazine, an ABCG2 substrate, in the intestine where ABCG2 acts as a gatekeeper for xenobiotics. It is noteworthy that the intestinal absorption of sulfasalazine majorly depends on the ABCG2 function in both humans and mice (Zaher et al., 2006; Yamasaki et al., 2008). First, we addressed the time-dependent changes in the plasma concentration of orally administered sulfasalazine in WT mice. In WT mice pretreated with febuxostat, the plasma concentration of sulfasalazine was significantly higher than that in control mice at every time point (Figure 3B).

We then performed similar experiments using Abcg2 KO mice and calculated the AUC of sulfasalazine. In WT mice, the administration of febuxostat increased the AUC of sulfasalazine (Figure 3C). On the other hand, in Abcg2 KO mice, febuxostat did not increase the AUC of sulfasalazine. Moreover, there was no significant difference in the AUC of sulfasalazine among three groups; febuxostat-treated WT mice, vehicle-treated Abcg2 KO mice and febuxostat-treated Abcg2 KO mice. These results suggest that febuxostat enhanced the absorption of orally administered sulfasalazine by inhibiting Abcg2 in vivo.

Some of the SUA-affecting drugs we investigated hardly affected ABCG2 function (Figure 1C), while their effects on other urate transporters remained unclear. Therefore, we focused on URAT1, another physiologically important urate transporter recognized as a pharmacological target of SUA-lowering drugs, and compared the effects of the drugs on ABCG2 and URAT1.

To evaluate the effect of SUA-affecting drugs on the urate transport activity of URAT1, we established an in vitro assay system for the urate transport activity with 293A cells transiently expressing URAT1. The expression of URAT1 in the 293A cells was confirmed by western blot analysis (Figure 4A). Then, we measured the amount of [8-¹⁴C]-urate incorporated into URAT1-expressing 293A cells for 0–120 s (Figure 4B). The result showed that the urate transport activity in URAT1-expressing cells was more than fivefold higher than that in the mock cells, and the amount of incorporated urate increased in a time-dependent manner. Then, we evaluated the effect of drugs on the URAT1 activity for 20 s in subsequent investigations.

We compared the effect of each SUA-affecting drug (final concentration 3 μM) on the urate transport activity of ABCG2 and URAT1 (Figure 4C; Supplementary Figure S1). Among the 25 SUA-affecting drugs, only benzbromarone strongly inhibited both ABCG2 and URAT1. The urate transport activities of ABCG2 and URAT1 in the presence of 3 μM benzbromarone were 13 ± 2% and 12 ± 5%, respectively, compared to the vehicle control. On the other hand, febuxostat and topiroxostat did not affect the urate transport activity of URAT1, although they potently inhibited ABCG2. The rest of the investigated drugs had little effect on both ABCG2 and URAT1 at 3 μM.