Pharmaceutical Research

, Volume 24, Issue 12, pp 2281–2296

Multichannel Liquid Chromatography–Tandem Mass Spectrometry Cocktail Method for Comprehensive Substrate Characterization of Multidrug Resistance-Associated Protein 4 Transporter

Authors

  • Yasuo Uchida
    • Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical SciencesTohoku University
  • Junichi Kamiie
    • Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical SciencesTohoku University
    • SORSTJapan Science and Technology Agency (JST)
  • Sumio Ohtsuki
    • Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical SciencesTohoku University
    • SORSTJapan Science and Technology Agency (JST)
    • Division of Membrane Transport and Drug Targeting, Graduate School of Pharmaceutical SciencesTohoku University
    • SORSTJapan Science and Technology Agency (JST)
Research Paper

DOI: 10.1007/s11095-007-9453-7

Cite this article as:
Uchida, Y., Kamiie, J., Ohtsuki, S. et al. Pharm Res (2007) 24: 2281. doi:10.1007/s11095-007-9453-7

Abstract

Purpose

To develop a comprehensive substrate-screening method for the ATP-binding cassette (ABC) transporter, and identify new substrates for multidrug resistance-associated protein 4 (MRP4/ABCC4).

Methods

Human MRP4-expressing membrane vesicles were incubated with a mixture of 50 compounds, including methotrexate, a known MRP4 substrate. The amounts transported were simultaneously determined by liquid chromatography–tandem mass spectrometry.

Results

From 49 compounds, 12 were identified as substrate candidates for MRP4 in the first screening. The second screening was performed involving the uptake of mixture using single quadrupole multichannel mode, and the third screening was performed involving the uptake of individual compounds using multiple reaction monitoring multichannel mode. As a result, eight substrate candidates were additionally identified. Subsequently, in the fourth step, osmotic pressure-dependent transport was demonstrated for 18 compounds (cefmetazole, piperacillin, rebamipide, tetracycline, ampicillin, benzylpenicillin, bumetanide, cephalosporin C, enalapril, pipemidic acid, furosemide, ceftazidime, pravastatin, hydrochlorothiazide, sulbactam, baclofen, bezafibrate and alacepril) among the 20 substrate candidates, thereby confirming them as MRP4 substrates. By contrast, the uptakes of meloxicam and nateglinide did not depend on osmolarity, indicating that these compounds were not substrates, but bound to MRP4.

Conclusions

The new comprehensive substrate-screening method for ABC transporters allowed the identification of 18 new substrates for MRP4.

Key words

ATP-binding cassette transporterhuman multidrug resistance-associated protein 4liquid chromatography–tandem mass spectrometrysubstrate screeningvesicle uptake study

Abbreviations

ABC

ATP-binding cassette

ACE

angiotensin-converting enzyme

BBB

blood–brain barrier

BCSFB

blood–cerebrospinal fluid barrier

CMZ

cefmetazole

CNS

central nervous system

DHEAS

dehydroepiandrosterone sulfate

ESI

electrospray ionization

LC–MS

liquid chromatography–mass spectrometry

LC–MS/MS

liquid chromatography–tandem mass spectrometry

MRM

multiple reaction monitoring

MRP

multidrug resistance-associated protein

MTX

methotrexate

MUID

membrane uptake-index difference

OAT

organic anion transporter

Q1

single quadrupole

INTRODUCTION

The human multidrug resistance-associated protein 4 (MRP4/ABCC4) is a subtype of the ABC transporter superfamily, and it excretes substrates from the inside to the outside of cells by coupling with ATP hydrolysis. MRP4 is expressed in a number of organs, including the kidney and brain, and contributes to the distribution and elimination of predominantly organic anions (1). In the kidney, MRP4 is localized at the proximal tubule apical membrane, and is involved in tubular secretion into the lumen (2,3). The messenger RNA (mRNA) level of MRP4 in the kidney is the highest among all of the transporters that contribute to the apical efflux transport of organic anions in the proximal tubules, such as MRP2, MRP4, organic anion transporter 4 (OAT4), and ABC transporter G2 (ABCG2) (4). Furthermore, MRP4 plays an important role in restricting the distribution of drugs in the central nervous system (CNS). An in vivo study has demonstrated that MRP4, which is localized at the luminal membrane of the brain capillary endothelium and at the basolateral membrane of the choroid plexus epithelium, restricts penetration of the anticancer agent topotecan into the brain and cerebrospinal fluid (CSF) (5).

β-Lactam antibiotics, diuretics, non steroidal anti-inflammatory drugs (NSAIDs), and angiotensin-converting enzyme (ACE) inhibitors are among the most important groups of organic anionic drugs undergoing tubular secretion (69). Additionally, it has been reported that anionic drugs, such as β-lactam antibiotics, the 5-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitor pravastatin, centrally acting anionic NSAIDs, and baclofen, have poor brain distributions (1014). However, little is known about whether MRP4 is involved in the renal tubular secretion and/or the restricted distribution in the CNS of these various anionic drugs. To clarify whether MRP4 is involved in such processes, a comprehensive substrate screening procedure needs to be applied to these drugs.

Substrate screening for MRP4 has been performed by means of vesicle uptake studies, cell efflux transport assays, cytotoxicity assays, ATPase assays, and inhibition assays against vesicle uptake of a known substrate, and also by means of in vivo studies using knockout mice (5,1517). Among these assays, vesicle-uptake study only allows evaluation of the direct transport of a compound via MRP4. Therefore, vesicle-uptake study is suitable for the precise identification of substrates. However, the majority of test compounds have been limited to labeled compounds, and uptake studies are time-consuming, thereby preventing comprehensive substrate screening by means of vesicle-uptake studies.

Liquid chromatography–mass spectrometry (LC–MS) is one of the most widely used analytical techniques for non-labeled small compounds. LC–MS has become an essential methodology for drug metabolism and pharmacokinetic screening, including cytochrome P450 enzyme-inhibition screening and Caco-2-absorption screening (18). Recent advances in MS have allowed the simultaneous measurement of >50 molecules (19). Furthermore, by applying LC–tandem MS (MS/MS), analytes can be measured with a high sensitivity, comparable to that achieved using radioisotopes. Therefore, by taking advantage of this analytical technique, a comprehensive substrate screening based on vesicle uptake can be performed.

The purpose of the present study was to develop a comprehensive and rapid substrate-screening method for MRP4 by employing both vesicle-uptake study and LC–MS/MS, and to establish whether 49 compounds were in fact substrates of MRP4.

MATERIALS AND METHODS (LC–MS/MS COCKTAIL METHOD)

Reagents

Amoxicillin, azathioprine, (±)-baclofen, bezafibrate, bumetanide, caffeine, captopril, cefmetazole sodium salt (CMZ), ceftazidime hydrate, cephalosporin C zinc salt, dantrolene sodium salt, dehydroepiandrosterone sulfate (DHEAS), enalapril maleate salt, enoxacin, furosemide, hydrochlorothiazide, ketoprofen, meclofenamic-acid sodium salt, nalidixic acid, neostigmine bromide, norfloxacin, ofloxacin, pipemidic acid, piroxicam, salicylic acid, sulindac, tetracycline, theobromine, trans-4-(aminomethyl)cyclohexanecarboxylic acid (tranexamic acid), and warfarin were purchased from Sigma Chemical Co. (St. Louis, MO). Alacepril, allopurinol, ampicillin sodium, diclofenac sodium, ethacrynic acid, indomethacin, mefenamic acid, meloxicam, methotrexate (MTX), pravastatin sodium salt, sulbactam sodium salt, and sulpyrine monohydrate were purchased from Wako Pure Chemicals (Osaka, Japan). Ciprofloxacin hydrochloride, L-glutamine, and benzylpenicillin sodium salt were purchased from ICN Biomedicals Inc. (Aurora, OH). Alendronate, potassium canrenoate, and rebamipide were purchased from LKT Laboratories Inc. (St. Paul, MN). L-glutamic acid and theophylline were purchased from Nacalai Tesque Inc. (Kyoto, Japan). Isosorbide dinitrate was purchased from Alexis Biochemicals (San Diego, CA). Nateglinide was purchased from Toronto Research Chemicals Inc. (North York, Canada). Nicorandil was purchased from Tocris Bioscience (Ellisville, MO). Piperacillin sodium salt was purchased from MP Biomedicals Inc. (Eschwege, Germany). Control vesicles and human MRP4-expressing membrane vesicles were purchased from GenoMembrane (Yokohama, Japan) or were kindly supplied by GenoMembrane. All other chemicals were of reagent grade, and were available commercially.

Rapid and Conventional Substrate Screening by Single Quadrupole (Q1) Multichannel LC–MS Analysis Following a Cocktail Transport Study of Membrane Vesicles

  1. 1.

    Simultaneous vesicle-uptake study to determine the MRP4 transport rate for a cocktail of compounds

     
Uptake experiments were performed using the rapid-filtration method, as described previously (20), and these were carried out in medium containing membrane vesicles, 0.25 M sucrose, 10 mM Tris–HCl (pH 7.4), 10 mM MgCl2, 4 mM ATP, 10 mM phosphocreatine, 100 µg/ml creatine phosphokinase, a mixture of test compounds, and MTX, which is a known substrate for MRP4 (21), in a total volume of 220 µl. The reactions were carried out at 37°C and stopped by the addition of 700 µl ice-cold stop solution (0.25 M sucrose, 10 mM Tris–HCl (pH 7.4), and 100 mM NaCl). The samples were passed through 0.22 µm Durapore membrane filters (Millipore, Bedford, MA) under vacuum. The filters were then washed three times with 3 ml ice-cold stop solution, and extracted with 1 ml methanol. The extracts were concentrated to 50 µl by centrifugation under vacuum. Then, 150 µl 0.1% formic acid in water was added to 50 µl of the concentrated extracts. A sample of 100 µl was used for the LC–MS quantification, as described below.
  1. 2.

    Simultaneous quantification of intravesicular uptake for multiple compounds by Q1 multichannel LC–MS

     
The sample analysis was automated by coupling a triple quadrupole mass spectrometer (4000QTRAP, Applied Biosystems) to an Agilent 1100 high-performance liquid chromatography (HPLC) system (Agilent Technologies). Briefly, a 100 µl uptake sample was injected onto a reversed-phase HPLC column (XDB-C18; 2.1 × 150 mm; particle size 5 µm; Agilent). Mobile phases A and B consisted of 0.1% formic acid in water and 0.1% formic acid in acetonitrile, respectively. In the first screening, the compounds were separated and eluted from the column using a linear gradient with a 110-min run time at a flow rate of 0.2 ml/min, and the sequence was as follows: A–B (100:0) for 5 min after injection, 0:100 at 65 min and up to 75 min, 100:0 at 85 min and up to 110 min. In the second screening, the compounds were separated and eluted from the column using a linear gradient with a 60-min run time at a flow rate of 0.2 ml/min, and the sequence was as follows: A–B (100:0) for 5 min after injection, 0:100 at 30 min and up to 32 min, 100:0 at 35 min and up to 60 min. The ion source was set as the electrospray ionization (ESI) positive-ionization mode. The MS operating conditions were as follows: curtain gas, 40; ionspray voltage, 5,000 V; temperature, 600°C; ion source gas 1, 50 p.s.i.; ion source gas 2, 80 p.s.i.; and declustering potential, 60 V. The compounds were simultaneously and selectively determined using 50 m/z channels, which were specific for each compound, in the Q1 multiple ions mode (Table I).
  1. 3.

    Evaluation of the MRP4-dependent transport rate

     
Table I

m/z channels of 50 compounds in Q1 multichannel LC–MS and MRM channels of several compounds in MRM multichannel LC–MS/MS

Compound

Monoisotopic mass

m/z channel

MRM channels

Compound

Monoisotopic mass

m/z channel

MRM channels

Allopurinol

136.03

137.03

 

Nateglinide

317.20

318.20

318.3/165.8

Salicylate

138.03

139.03

139.1/79.1

Norfloxacin

319.13

320.13

 

Glutamine

146.07

147.07

 

Enoxacin

320.13

321.13

 

Glutamic acid

147.05

148.05

 

Furosemide

330.01

331.01

328.9/205.0

Tranexamic acid

157.11

158.11

 

Piroxicam

331.06

332.06

 

Theobromine

180.06

181.06

 

Ciprofloxacin

331.13

332.13

 

Theophylline

180.06

181.06

 

Benzylpenicillin

334.10

335.10

335.2/91.1

Caffeine

194.08

195.08

 

Ampicillin

349.11

350.11

350.2/106.1

Nicorandil

211.06

212.06

 

Meloxicam

351.03

352.03

352.0/115.1

Baclofen

213.05

214.05

214.1/151.1

Sulindac

356.08

357.08

357.2/233.2

Captopril

217.08

218.08

218.2/56.1

Canrenoate

358.21

359.21

 

Neostigminea

223.14

223.14

 

Bezafibrate

361.11

362.11

362.1/139.1

Nalidixic acid

232.08

233.08

233.2/104.1

Ofloxacin

361.14

362.14

 

Sulbactam

233.04

234.04

232.2/140.1

Bumetanide

364.11

365.11

365.1/240.1

Isosorbide dinitrate

236.03

237.03

 

Amoxicillin

365.10

366.10

366.1/114.1

Mefenamic acid

241.11

242.11

 

Rebamipide

370.07

371.07

371.0/216.0

Alendronate

249.02

250.02

 

Enalapril

376.20

377.20

377.2/234.1

Azathioprine

277.04

278.04

278.0/142.0

Alacepril

406.15

407.15

407.2/242.1

Meclofenamic acid

295.02

296.02

 

Cephalosporin C

415.10

416.10

416.1/312.1

Hydrochlorothiazide

296.96

297.96

295.9/268.9

Pravastatin

424.25

425.25

423.2/321.2

Ethacrynic acid

302.01

303.01

303.1/257.0

Tetracycline

444.15

445.15

445.2/410.1

Pipemidic acid

303.13

304.13

304.1/217.1

MTX

454.17

455.17

455.2/308.0

Warfarin

308.10

309.10

 

CMZ

471.05

472.05

472.0/356.1

Sulpyrine

311.09

312.09

312.1/77.1

Piperacillin

517.16

518.16

518.1/142.9

Dantrolene

314.07

315.07

313.0/227.9

Ceftazidime

546.10

547.10

547.1/468.2

The m/z channels were set at the monoisotopic masses plus one. The MRM channels were obtained by direct infusion of the compound solution into 4000QTRAP.

The uptake was calculated from the peak area on the mass chromatogram, and was expressed as the vesicle-to-medium ratio, which was obtained by dividing the amount taken up into the membrane vesicles by the substrate concentration in the uptake medium. To evaluate the MRP4-dependent transport rate, the membrane uptake-index difference (MUID) of each compound was calculated from the following equation:
$${\text{MUI}}_{\text{D}} = {{{\text{vesicle}}} \mathord{\left/{\vphantom {{{\text{vesicle}}} {{{{\text{medium}}\,\,{\text{ratio}}_{{\text{MRP}}4} - {\text{vesicle}}} \mathord{\left/{\vphantom {{{\text{medium}}\,\,{\text{ratio}}_{{\text{MRP}}4} - {\text{vesicle}}} {{\text{medium}}\,\,{\text{ratio}}_{{\text{Mock}}{\text{.}}} }}} \right.\kern-\nulldelimiterspace} {{\text{medium}}\,\,{\text{ratio}}_{{\text{Mock}}{\text{.}}} }}}}} \right.\kern-\nulldelimiterspace} {{{{\text{medium}}\,\,{\text{ratio}}_{{\text{MRP}}4} - {\text{vesicle}}} \mathord{\left/{\vphantom {{{\text{medium}}\,\,{\text{ratio}}_{{\text{MRP}}4} - {\text{vesicle}}} {{\text{medium}}\,\,{\text{ratio}}_{{\text{Mock}}{\text{.}}} }}} \right.\kern-\nulldelimiterspace} {{\text{medium}}\,\,{\text{ratio}}_{{\text{Mock}}{\text{.}}} }}}}$$

Here, the vesicle-to-medium ratioMRP4 and the vesicle-to-medium ratioMock are the uptake of compounds into membrane vesicles prepared from MRP4-transfected and parental vector-transfected Sf9 cells, respectively.

Highly Sensitive Substrate Screening by MRM Multichannel LC–MS/MS Analysis Following an Individual Transport Study or Osmotic Pressure Dependency Study

  1. 1.

    Individual vesicle-uptake study to determine the MRP4 transport rate for a single compound

     

Uptake experiments were performed as described above, and were carried out with a single test compound in the medium containing membrane vesicles, 0.25 M sucrose, 10 mM Tris–HCl (pH 7.4), 10 mM MgCl2, 4 mM ATP, 10 mM phosphocreatine, 100 μg/ml creatine phosphokinase, and the test compound, in a total volume of 110 μl. The reactions were carried out at 37°C, and were stopped by the addition of 800 μl ice-cold stop solution. The samples were then passed through 0.22 μm Durapore membrane filters under vacuum. The filters were washed three times with 3 ml ice-cold stop solution, and were extracted with 1 ml methanol. The extracts were concentrated to 50 μl by centrifugation under vacuum. Then, 150 μl 0.1% formic acid in water was added to 50 μl of the concentrated extracts. Next, 200 μl samples of several test compounds were mixed. An aliquot of 100 μl was used for the LC–MS/MS quantification, as described below.

2. Simultaneous quantification of intravesicular uptake for several compounds by MRM multichannel LC–MS/MS

A 100-μL uptake sample mixture of several test compounds was injected onto a reversed-phase HPLC column (XDB-C18; 2.1 × 150 mm; particle size 5 µm). The compounds were separated and eluted from the column using a linear gradient with a 30-min run time at a flow rate of 0.2 ml/min, and the sequence was as follows: A–B (100:0) for 5 min after injection, 0:100 at 15 min, 100:0 at 16 min and up to 30 min. The eluted compounds were simultaneously and selectively determined by using the multiplexed MRM channels in the ESI positive or negative ionization mode. Each MRM channel was specific for a compound, as listed in Table I. The conditions of the MRM channel were optimized using automatic optimization software (Analyst, ABI) by the direct infusion of 1–100 μM compound solution at a flow rate of 5 μl/min with a syringe pump (Harvard) into 4000QTRAP. The uptake was expressed as the vesicle-to-medium ratio, and the MUID was determined as described above.

Data Analysis

For kinetic studies, the Km and Vmax of CMZ by MRP4-expressing membrane vesicles were estimated from the following equation, using the nonlinear least-squares regression analysis program MULTI (22):
$$v = {{V_{\max } \times \left[ S \right]} \mathord{\left/{\vphantom {{V_{\max } \times \left[ S \right]} {\left( {K_{\text{m}} + \left[ S \right]} \right).}}} \right.\kern-\nulldelimiterspace} {\left( {K_{\text{m}} + \left[ S \right]} \right).}}$$

Here, v and [S] are the uptake rate and the concentration of CMZ, respectively.

Unless otherwise indicated, all data represent the mean ± standard error of the mean (SEM). An unpaired, two-tailed, Student’s t test was used to determine the significance of differences between the means of two groups. One-way analysis of variance (ANOVA) followed by the modified Fisher’s least-squares difference method was used to assess the statistical significance of differences among the means of more than two groups.

RESULTS

Scheme 1 presents a flow chart of the LC–MS/MS cocktail method, which allows the rapid and comprehensive screening and identification of MRP4 substrates by means of conventional Q1 multichannel LC–MS analysis following cocktail transport studies, and then highly-sensitive MRM multichannel LC–MS/MS analysis following individual transport studies or osmotic pressure-dependency studies. The substrate/non-substrate classification was established for all of the compounds by four steps of independent transport studies. The 49 test compounds selected were mainly organic anions, including those known to be extensively excreted from the kidney and/or to have restricted distribution into the brain, but not previously reported as substrates of MRP4.
https://static-content.springer.com/image/art%3A10.1007%2Fs11095-007-9453-7/MediaObjects/11095_2007_9453_Sch1_HTML.gif
Scheme 1

Strategy for rapid and comprehensive substrate and non-substrate screening for MRP4 using the LC–MS/MS cocktail method. The internal reference was MTX, a known substrate of MRP4

The First Screening; Rapid and Conventional Substrate Screening by Q1 Multichannel LC–MS Analysis Following a Cocktail Transport Study

The m/z channels of all compounds examined, including MTX, are listed in Table I, and were simultaneously analyzed by Q1 multichannel LC–MS. MTX was used as an internal reference to evaluate the inhibitory effect of compounds on MRP4-transport activity. The first screening was performed twice, primary and secondary screenings. The primary first screening was performed on a mixture of 50 compounds (Table I) at a concentration of 10 µM in 100 µg membrane vesicles prepared from MRP4-transfected Sf9 cells for 30 min at 37°C. At this concentration, the inhibitory effect of MTX on substrate transport via MRP4 would be negligible, because the Km value of MTX is 220 µM. Membrane vesicles prepared from the parent vector-transfected Sf9 cells were used for a control study in the same manner. Table II summarizes the results, listing the vesicle/medium ratioMRP4 and the vesicle/medium ratioMock as the apparent transport rate into MRP4-expressing membrane vesicles and into the control membrane vesicles, respectively. The apparent MRP4-dependent transport rate was determined for each compound by subtracting the value of the vesicle/medium ratioMock from that of the vesicle/medium ratioMRP4, and is listed as the MUID in Table II. A significant apparent MRP4-dependent transport rate was obtained for CMZ, piperacillin, MTX, rebamipide, tetracycline, ampicillin, benzylpenicillin, and bumetanide (Table II). In the absence of the compounds listed in Table II, the MRP4-dependent transport rate of MTX was determined as 38.4  ±  1.0 µl/(mg protein·30 min) (Table III). A significant inhibition (by 85.5%) in the presence of the 49 compounds (5.55  ±  0.27 µl/(mg protein·30 min)) suggested that the MRP4-dependent transport rate of the tested compounds was underestimated due to mutual inhibition (Table II). As isosorbide dinitrate, allopurinol, alendronate, and glutamine were not detected at 10 pmol by LC–MS, we were unable to determine their transport rates. Hydrochlorothiazide was not detected by LC–MS in the MRP4-expressing and control membrane vesicles, though it was detected at the 10 pmol level by LC–MS.
Table II

MRP4-dependent uptake of 50 compounds into membrane vesicles in the primary first screening

Compound

Vesicle/medium ratio (µl/(mg protein·30 min))

MUID

 

MRP4

Mock

µl/(mg protein·30 min)

MTX

6.93  ±  0.27

1.37  ±  0.05

5.55  ±  0.27***

CMZ

7.06  ±  0.14

0.374  ±  0.027

6.68  ±  0.14***

Piperacillin

6.55  ±  0.38

0.669  ±  0.019

5.88  ±  0.38***

Rebamipide

7.52  ±  0.05

3.64  ±  0.10

3.88  ±  0.05***

Tetracycline

17.3  ±  0.8

15.7  ±  0.5

1.55  ±  0.81***

Ampicillin

1.99  ±  0.06

0.556  ±  0.039

1.44  ±  0.06***

Benzylpenicillin

1.69  ±  0.23

0.314  ±  0.053

1.37  ±  0.23**

Bumetanide

3.66  ±  0.21

2.50  ±  0.20

1.17  ±  0.21*

 

Ceftazidime

2.72  ±  0.17

1.90  ±  0.05

0.823  ±  0.175

Canrenoate

0.908  ±  0.102

0.103  ±  0.005

0.806  ±  0.102

Cephalosporin C

1.69  ±  0.01

0.942  ±  0.146

0.744  ±  0.013

Alacepril

1.07  ±  0.01

0.461  ±  0.031

0.613  ±  0.006

Amoxicillin

1.84  ±  0.20

1.25  ±  0.29

0.589  ±  0.199

Sulbactam

0.548  ±  0.050

0.0768  ±  0.0088

0.471  ±  0.050

Nalidixic acid

1.65  ±  0.04

1.18  ±  0.04

0.468  ±  0.042

Enalapril

0.785  ±  0.025

0.311  ±  0.029

0.474  ±  0.025

Mefenamic acid

1.83  ±  0.13

1.49  ±  0.04

0.343  ±  0.135

Captopril

1.04  ±  0.05

0.764  ±  0.059

0.276  ±  0.052

Baclofen

1.01  ±  0.02

0.758  ±  0.057

0.250  ±  0.020

Pipemidic acid

0.620  ±  0.014

0.404  ±  0.026

0.217  ±  0.014

Norfloxacin

0.536  ±  0.018

0.409  ±  0.040

0.127  ±  0.018

Bezafibrate

0.345  ±  0.019

0.244  ±  0.033

0.101  ±  0.019

Furosemide

3.28  ±  0.25

3.21  ±  0.38

0.0649  ±  0.2466

Piroxicam

1.38  ±  0.03

1.32  ±  0.02

0.0607  ±  0.0326

Sulindac

0.355  ±  0.031

0.319  ±  0.018

0.0361  ±  0.0309

Ciprofloxacin

0.358  ±  0.037

0.328  ±  0.038

0.0299  ±  0.0367

Enoxacin

0.465  ±  0.030

0.437  ±  0.032

0.0281  ±  0.0297

Ofloxacin

0.160  ±  0.009

0.176  ±  0.032

−0.0167  ±  0.0087

Nateglinide

0.630  ±  0.025

0.657  ±  0.065

−0.0273  ±  0.0250

Theobromine

0.502  ±  0.116

0.544  ±  0.005

−0.0419  ±  0.1154

Warfarin

0.675  ±  0.058

0.734  ±  0.031

−0.0594  ±  0.0584

Caffeine

0.732  ±  0.017

0.833  ±  0.075

−0.101  ±  0.017

Azathioprine

0.0844  ±  0.0093

0.194  ±  0.020

−0.110  ±  0.009

Ethacrynic acid

1.47  ±  0.05

1.65  ±  0.06

−0.183  ±  0.052

Pravastatin

3.61  ±  0.55

3.95  ±  0.46

−0.332  ±  0.553

Tranexamic acid

2.07  ±  0.18

2.48  ±  0.11

−0.407  ±  0.177

Meloxicam

3.75  ±  0.09

4.24  ±  0.06

−0.487  ±  0.086

Theophylline

0.184  ±  0.036

0.746  ±  0.038

−0.562  ±  0.036

Dantrolene

1.25  ±  0.05

2.26  ±  0.16

−1.00  ±  0.06

Salicylate

3.15  ±  0.22

4.61  ±  0.76

−1.46  ±  0.22

Meclofenamic acid

5.54  ±  0.56

7.58  ±  0.24

−2.05  ±  0.56

Neostigmine

3.17  ±  0.17

5.51  ±  0.07

−2.34  ±  0.17

Sulpyrine

4.04  ±  0.30

6.58  ±  0.30

−2.54  ±  0.30

Glutamic acid

2.32  ±  0.36

5.47  ±  0.07

−3.15  ±  0.36

Nicorandil

0.230  ±  0.032

4.08  ±  0.58

−3.86  ±  0.03

 

Isosorbide dinitratea

Allopurinola

Alendronatea

Glutaminea

Hydrochlorothiazideb

Membrane vesicles (100 µg) prepared from MRP4-transfected or parental vector-transfected Sf9 cells (Mock) were incubated at 37°C for 30 min in uptake medium containing 50 compounds (each 10 µM) in the presence of 4 mM ATP. Each value represents the mean ± SEM (n = 4).

*p < 0.05, **p < 0.01, ***p < 0.001, significantly greater than 0 µl/(mg protein·30 min) of MUID, Dunnett’s test.

aThe authentic samples (10 pmol) were not detected by LC–MS, and the compounds were also not detected in uptake samples of membrane vesicles prepared from MRP4-transfected and parental vector-transfected Sf9 cells.

bThe authentic sample (10 pmol) was detected, but the compound was not detected in uptake samples.

Table III

MRP4-dependent uptake of 39 compounds into membrane vesicles in the secondary first screening

Compound

Vesicle/medium ratio (µl/(mg protein·30 min))

MUID

MRP4

Mock

µl/(mg protein·30 min)

MTX (single)

39.6  ±  1.0

1.13  ±  0.04

38.4  ±  1.0

 

MTX

7.44  ±  0.51

0.963  ±  0.243

6.48  ±  0.51***,†††

Cephalosporin C

1.93  ±  0.12

0.561  ±  0.145

1.37  ±  0.12***

Enalapril

1.16  ±  0.03

0.139  ±  0.005

1.02  ±  0.03***

Nateglinide

1.51  ±  0.08

0.548  ±  0.061

0.964  ±  0.083**

Pipemidic acid

1.14  ±  0.06

0.321  ±  0.099

0.818  ±  0.058*

Meloxicam

3.25  ±  0.10

2.43  ±  0.42

0.816  ±  0.097*

 

Sulpyrine

6.17  ±  0.13

5.50  ±  0.89

0.670  ±  0.135

Ciprofloxacin

0.864  ±  0.056

0.251  ±  0.037

0.612  ±  0.056

Alacepril

1.12  ±  0.02

0.507  ±  0.050

0.609  ±  0.021

Norfloxacin

0.920  ±  0.063

0.323  ±  0.107

0.597  ±  0.063

Bezafibrate

0.729  ±  0.030

0.212  ±  0.029

0.517  ±  0.030

Baclofen

0.846  ±  0.025

0.336  ±  0.063

0.509  ±  0.025

Enoxacin

0.845  ±  0.053

0.352  ±  0.108

0.493  ±  0.053

Ethacrynic acid

1.65  ±  0.06

1.16  ±  0.09

0.491  ±  0.062

Dantrolene

1.37  ±  0.06

0.909  ±  0.088

0.465  ±  0.064

Piroxicam

1.38  ±  0.05

0.919  ±  0.141

0.465  ±  0.050

Tranexamic acid

1.39  ±  0.19

1.01  ±  0.25

0.384  ±  0.191

Canrenoate

1.62  ±  0.19

1.26  ±  0.22

0.361  ±  0.187

Warfarin

0.816  ±  0.016

0.561  ±  0.036

0.255  ±  0.016

Nalidixic acid

0.475  ±  0.023

0.299  ±  0.076

0.176  ±  0.023

Glutamic acid

1.02  ±  0.19

0.860  ±  0.311

0.160  ±  0.190

Caffeine

0.867  ±  0.069

0.711  ±  0.116

0.156  ±  0.069

Neostigmine

1.52  ±  0.16

1.38  ±  0.37

0.138  ±  0.159

Sulindac

0.375  ±  0.007

0.240  ±  0.012

0.135  ±  0.007

Captopril

0.809  ±  0.182

0.686  ±  0.177

0.123  ±  0.182

Salicylate

3.18  ±  0.26

3.09  ±  0.11

0.0900  ±  0.2612

Ofloxacin

0.144  ±  0.014

0.103  ±  0.008

0.0412  ±  0.0139

Theophylline

0.409  ±  0.104

0.402  ±  0.124

0.00715  ±  0.10374

Azathioprine

0.370  ±  0.067

0.382  ±  0.111

−0.0115  ±  0.0669

Amoxicillin

1.85  ±  0.25

1.96  ±  0.16

−0.109  ±  0.246

Sulbactam

0.685  ±  0.093

0.799  ±  0.175

−0.114  ±  0.093

Mefenamic acid

1.68  ±  0.14

1.80  ±  0.13

−0.123  ±  0.145

Furosemide

2.29  ±  0.15

2.44  ±  0.15

−0.141  ±  0.146

Nicorandil

0.724  ±  0.078

0.879  ±  0.062

−0.156  ±  0.078

Pravastatin

3.01  ±  0.24

3.18  ±  0.47

−0.171  ±  0.240

Theobromine

0.523  ±  0.104

0.720  ±  0.089

−0.198  ±  0.104

Ceftazidime

2.41  ±  0.16

3.03  ±  0.69

−0.630  ±  0.161

Meclofenamic acid

5.22  ±  0.44

6.94  ±  0.53

−1.72  ±  0.44

 

Hydrochlorothiazidea

Membrane vesicles (100 µg) prepared from MRP4-transfected or parental vector-transfected Sf9 cells (Mock) were incubated at 37°C for 30 min in uptake medium containing 39 compounds (each 10 µM) or 10 µM MTX only in the presence of 4 mM ATP. Each value represents the mean ± SEM (= 4).

*p < 0.05, **p < 0.01, ***p < 0.001, significantly greater than 0 µl/(mg protein·30 min) of MUID, Dunnett’s test.

p < 0.001, significantly different from MUID of MTX (single), Student’s t test.

The authentic (10 pmol) sample was detected by LC–MS, while the compound was not detected in uptake samples of membrane vesicles prepared from MRP4-transfected and parental vector-transfected Sf9 cells.

The secondary first screening was performed for a mixture of 39 compounds, excluding seven substrate candidates and four undetectable compounds from the list shown in Table II, and the results are summarized in Table III. A significant apparent MRP4-dependent transport rate was obtained for MTX, cephalosporin C, enalapril, nateglinide, pipemidic acid, and meloxicam (Table III). The MTX-transport rate was 6.48  ±  0.51 μl/(mg protein·30 min) in the presence of the 38 compounds, which was 16.1% of the control value, again suggesting an underestimation of the MRP4-transport rate for all of the compounds (Table III).

The Second Screening; Substrate Screening of Compounds Divided Into 4 Groups by Q1 Multichannel LC–MS Analysis Following a Cocktail Transport Study

The compounds that failed to demonstrate significant apparent MRP4-dependent transport in Table III were divided into four groups as shown in Table IV. The second substrate screening was performed for each group using a 5 μM mixture with MTX for 60 min. No significant MRP4-dependent transport rate (MUID) was observed for any compound examined, except MTX (Table IV). No significant inhibitory effect on the MTX-transport rate was obtained for groups 3 and 4, confirming that the compounds in these groups are not effective MRP4 substrates. In contrast, in groups 1 and 2, there was a significant reduction in the MTX transport rate to 12.7 and 21.9% of that of the control study (60.7  ±  1.1 µl/(mg protein·60 min) without mixture), respectively (Table IV). Therefore, the possibility cannot be ruled out that a significant inhibitory effect was causing an underestimation of the MRP4-dependent transport activity in these groups.
Table IV

MRP4-dependent uptake of 34 compounds in four groups in the second screening

 

Compound

Vesicle/medium ratio (μl/(mg protein·60 min))

MUID

MRP4

Mock

μl/(mg protein·60 min)

MTX (single)

64.3  ±  1.1

3.62  ±  0.38

60.7  ±  1.1

1

MTX

12.3  ±  0.7

4.62  ±  0.32

7.68  ±  0.66*,**

Ethacrynic acid

1.06  ±  0.16

0.807  ±  0.162

0.252  ±  0.160

Baclofen

2.17  ±  0.12

1.93  ±  0.24

0.241  ±  0.120

Azathioprine

0.458  ±  0.127

0.550  ±  0.145

−0.0928  ±  0.1273

Bezafibrate

1.42  ±  0.13

1.56  ±  0.39

−0.143  ±  0.130

Sulbactam

0.937  ±  0.011

1.12  ±  0.30

−0.187  ±  0.011

Dantrolene

1.46  ±  0.09

1.84  ±  0.23

−0.384  ±  0.090

Sulpyrine

7.00  ±  0.40

7.44  ±  0.48

−0.446  ±  0.404

Nalidixic acid

2.70  ±  0.12

3.89  ±  0.84

−1.19  ±  0.12

 

2

MTX

18.6  ±  0.9

5.30  ±  0.30

13.3  ±  0.9*,**

Ceftazidime

3.33  ±  0.08

2.56  ±  0.32

0.772  ±  0.083

Furosemide

5.41  ±  0.28

4.70  ±  0.25

0.713  ±  0.282

Alacepril

2.53  ±  0.09

2.01  ±  0.19

0.522  ±  0.088

Amoxicillin

3.23  ±  0.10

2.76  ±  0.60

0.474  ±  0.101

Pravastatin

5.78  ±  0.35

5.50  ±  1.00

0.280  ±  0.352

Sulindac

1.79  ±  0.12

1.69  ±  0.12

0.102  ±  0.118

Captopril

0.630  ±  0.038

0.683  ±  0.040

−0.0529  ±  0.0377

Salicylate

3.29  ±  0.04

4.03  ±  0.35

−0.744  ±  0.042

Hydrochlorothiazidea

 

3

MTX

60.9  ±  2.7

3.97  ±  0.09

56.9  ±  2.7*

Ofloxacin

1.17  ±  0.10

1.00  ±  0.31

0.180  ±  0.101

Ciprofloxacin

1.11  ±  0.02

1.00  ±  0.19

0.110  ±  0.017

Enoxacin

2.61  ±  0.27

2.60  ±  0.26

0.00587  ±  0.26727

Canrenoate

0.769  ±  0.009

1.06  ±  0.08

−0.293  ±  0.009

Mefenamic acid

7.86  ±  0.72

8.35  ±  0.18

−0.491  ±  0.722

Norfloxacin

2.05  ±  0.20

2.72  ±  0.20

−0.661  ±  0.196

Piroxicam

3.99  ±  0.28

4.80  ±  0.07

−0.805  ±  0.279

Meclofenamic acid

10.7  ±  0.4

13.4  ±  0.7

−2.79  ±  0.38

 

4

MTX

67.1  ±  1.5

4.99  ±  0.16

62.1  ±  1.5*

Theophylline

1.37  ±  0.07

1.15  ±  0.14

0.214  ±  0.068

Caffeine

1.21  ±  0.07

1.13  ±  0.09

0.0791  ±  0.0674

Nicorandil

0.521  ±  0.089

0.605  ±  0.131

−0.0833  ±  0.0893

Glutamic acid

1.32  ±  0.35

1.59  ±  0.37

−0.274  ±  0.347

Warfarin

1.80  ±  0.10

2.13  ±  0.11

−0.328  ±  0.097

Tranexamic acid

1.13  ±  0.09

1.50  ±  0.15

−0.368  ±  0.085

Theobromine

0.732  ±  0.139

1.34  ±  0.38

−0.613  ±  0.139

Neostigmine

4.02  ±  0.25

5.46  ±  0.34

−1.44  ±  0.25

Membrane vesicles (40 μg) prepared from MRP4-transfected or parental vector-transfected Sf9 cells (Mock) were incubated at 37°C for 60 min in uptake medium containing nine or ten compounds (each 5 μM) or 5 μM MTX only in the presence of 4 mM ATP. Each value represents the mean ± SEM (n = 4).

*p < 0.001, significantly greater than 0 μl/(mg protein·60 min) of MUID, Dunnett’s test.

**p < 0.001, significantly different from MUID of MTX (single), Dunnett’s test.

aThe authentic (10 pmol) sample was detected by LC–MS, while the compound was not detected in uptake samples of membrane vesicles prepared from MRP4-transfected and parental vector-transfected Sf9 cells.

The Third Screening; Highly Sensitive Substrate Screening by MRM Multichannel LC–MS/MS Analysis Following an Individual Transport Study

A third substrate screening was performed by individual transport studies of the compounds in groups 1 and 2 shown in Table IV, and the results are summarized in Table V. After an individual transport study, the extracts of 12 (positively ionized) or five (negatively ionized) compounds from membrane vesicles were combined, which allowed us to shorten the analysis time, and the intravesicular amounts of the respective compounds were simultaneously determined by highly sensitive LC–MS/MS in the MRM multichannel mode. A significant apparent MRP4-dependent transport was seen using furosemide, ceftazidime, pravastatin, hydrochlorothiazide, sulbactam, baclofen, bezafibrate, and alacepril, while no significant MRP4-dependent transport was obtained for dantrolene, amoxicillin, sulpyrine, ethacrynic acid, azathioprine, sulindac, captopril, salicylate, or nalidixic acid.
Table V

MRP4-dependent single uptake of 17 compounds into membrane vesicles in the third screening

Compound

Vesicle/medium ratio (μl/(mg protein·30 min))

MUID

Detection mode

MRP4

Mock

µl/(mg protein·30 min)

Furosemide

31.7  ±  2.6**

3.22  ±  0.14

28.5  ±  2.6

Negative

Ceftazidime

4.18  ±  0.09**

1.16  ±  0.10

3.02  ±  0.09

Positive

Pravastatin

3.19  ±  0.04**

0.856  ±  0.048

2.34  ±  0.04

Negative

Hydrochlorothiazide

2.97  ±  0.16**

1.52  ±  0.01

1.45  ±  0.16

Negative

Sulbactam

1.10  ±  0.14*

0.197  ±  0.014

0.901  ±  0.145

Negative

Baclofen

1.56  ±  0.05**

0.730  ±  0.035

0.827  ±  0.054

Positive

Bezafibrate

1.19  ±  0.06**

0.487  ±  0.017

0.700  ±  0.060

Positive

Alacepril

0.522  ±  0.014**

0.284  ±  0.017

0.237  ±  0.014

Positive

 

Dantrolene

3.30  ±  0.20

3.05  ±  0.07

0.247  ±  0.196

Negative

Amoxicillin

0.601  ±  0.013

0.438  ±  0.062

0.163  ±  0.013

Positive

Sulpyrine

2.97  ±  0.11

2.83  ±  0.06

0.137  ±  0.115

Positive

Ethacrynic acid

2.26  ±  0.02

2.16  ±  0.19

0.100  ±  0.018

Positive

Azathioprine

0.0959  ±  0.0077

0.0566  ±  0.0190

0.0393  ±  0.0077

Positive

Sulindac

0.301  ±  0.017

0.269  ±  0.028

0.0317  ±  0.0167

Positive

Captopril

0.162  ±  0.003

0.173  ±  0.002

−0.0116  ±  0.0027

Positive

Salicylate

1.02  ±  0.04

1.05  ±  0.05

−0.0363  ±  0.0395

Positive

Nalidixic acid

0.359  ±  0.073

0.415  ±  0.061

−0.0568  ±  0.0725

Positive

Membrane vesicles (10 µg) prepared from MRP4-transfected or parental vector-transfected Sf9 cells were incubated at 37°C for 30 min in uptake medium containing 10 µM single compound in the presence of 4 mM ATP. The uptakes of 12 compounds or five compounds were simultaneously determined by MRM multichannel LC–MS/MS in the ESI positive or negative mode, respectively. Each value represents the mean ± SEM (n = 3).

*p < 0.01, **p < 0.001, significantly different from the vesicle-to-medium ratioMock, Student’s t test.

The Fourth Step; Precise Identification of MRP4 Substrates by MRM Multichannel LC–MS/MS Analysis Following an Osmotic Pressure Dependency Study

To clarify whether the apparent MRP4-dependent transport represented transport via MRP4 or binding to MRP4, the osmotic pressure dependency was examined for the compounds identified by the first, second, and third screening studies.

As shown in Fig. 1(a, b and h), the uptake of CMZ, rebamipide and piperacillin by MRP4-expressing membrane vesicles in the presence of ATP (closed circle) was greater than that in the presence of AMP (open circle) under isotonic conditions. The uptake of CMZ and rebamipide decreased linearly with the increase in medium osmolarity. The y-intercept indicates the amount of compounds binding to the vesicle surface, MRP4 and/or filters. The y-intercept values for CMZ and rebamipide were less than 0 µl/(mg protein·min), indicating that the binding to the vesicle surface, MRP4 and/or filters is negligible. The uptake of piperacillin decreased linearly with the increase in medium osmolarity, and the y-intercept was 0.114 µl/(mg protein·min) (Table VI), indicating that 12% of piperacillin is bound to the vesicle surface, MRP4 and/or filters. Thus, these results indicate that CMZ, rebamipide and piperacillin were transported into the intravesicular space by MRP4 in an ATP-dependent manner.
https://static-content.springer.com/image/art%3A10.1007%2Fs11095-007-9453-7/MediaObjects/11095_2007_9453_Fig1_HTML.gif
Fig. 1

Osmotic pressure dependency of the uptake of substrate candidates into MRP4-expressing membrane vesicles. The uptake of 1 µM of each substrate (CMZ (a), rebamipide (b), MTX (c), and piperacillin (h)) was measured at 37°C for 3 min by incubating membrane vesicles (10 µg) prepared from MRP4-transfected (circles) Sf9 cells in uptake medium containing 250 mM sucrose and various concentrations of raffinose (0–0.3 M, but 0–0.5 M for CMZ). The uptake of 5 µM of each substrate (benzylpenicillin (e), tetracycline (g), bumetanide (i), and ampicillin (p)), 10 µM substrate (furosemide (d), cephalosporin C (f), alacepril (j), pravastatin (k), hydrochlorothiazide (l), baclofen (m), bezafibrate (o), enalapril (q), pipemidic acid (r), sulbactam (s), meloxicam (t), and nateglinide (u)) or 100 µM ceftazidime (n) was measured at 37°C for 10 min by incubating membrane vesicles (10 µg) prepared from MRP4-transfected (circles) or parental vector-transfected (squares) Sf9 cells in uptake medium containing 250 mM sucrose and various concentration of raffinose (0–0.3 M). Closed symbols, uptake in the presence of 4 mM ATP; open symbols, uptake in the presence of 4 mM AMP. Each point represents the mean ± SEM (n = 3–4)

Table VI

Evaluation of osmolarity-dependent transport activities for the MRP4 substrate candidates into MRP4-expressing membrane vesicles

Compound

Screening step

Slope

Intercept

Apparent MRP4-dependent transportisotonic

Apparent MRP4-dependent transporthypertonic

Osm·µl/(mg protein·min)

µl/(mg protein·min)

µl/(mg protein·min)

µl/(mg protein·min)

CMZ

Primary first

2.76 × 10

<0

8.06 × 10  ±  0.22 × 10a

 

Rebamipide

Primary first

2.10 × 10

<0

4.98 × 10  ±  0.20 × 10a

 

MTX (reference)

 

8.76 × 10−1

<0

  

Furosemide

Third

7.93 × 10−1

4.89 × 10−1

1.74  ±  0.06

 

Benzylpenicillin

Primary first

5.72 × 10−1

<0

1.14  ±  0.23

 

Cephalosporin C

Secondary first

4.81 × 10−1

<0

1.22  ±  0.04

 

Tetracycline

Primary first

4.15 × 10−1

2.67 × 10−1

8.66 × 10−1  ±  0.66 × 10−1

 

Piperacillin

Primary first

2.86 × 10−1

1.14 × 10−1

5.45 × 10−1  ±  0.57 × 10−1a

 

Bumetanide

Primary first

1.66 × 10−1

5.29 × 10−2

2.99 × 10−1  ±  0.93 × 10−1

 

Alacepril

Third

7.35 × 10−2

<0

8.12 × 10−2  ±  0.05 × 10−2

 

Pravastatin

Third

7.13 × 10−2

<0

1.18 × 10−1  ±  0.07 × 10−1

 

Hydrochlorothiazide

Third

5.43 × 10−2

9.92 × 10−2

1.02 × 10−1  ±  0.14 × 10−1

5.71 × 10−2  ±  0.32 × 10−2*

Baclofen

Third

5.00 × 10−2

1.91 × 10−2

1.09 × 10−1  ±  0.04 × 10−1

 

Ceftazidime

Third

4.24 × 10−2

<0

8.00 × 10−2  ±  0.55 × 10−2

 

Bezafibrate

Third

3.71 × 10−2

<0

5.04 × 10−2  ±  0.38 × 10−2

 

Ampicillin

Primary first

3.62 × 10−2

<0

7.67 × 10−2  ±  0.50 × 10−2

 

Enalapril

Secondary first

3.11 × 10−2

3.03 × 10−3

5.87 × 10−2  ±  0.24 × 10−2

 

Pipemidic acid

Secondary first

2.01 × 10−2

4.36 × 10−3

2.18 × 10−2  ±  0.70 × 10−2

 

Sulbactam

Third

6.09 × 10−3

<0

1.47 × 10−2  ±  0.06 × 10−2

 

 

Meloxicam

Secondary first

3.72 × 10−2

4.26 × 10−1

9.62 × 10−2  ±  2.91 × 10−2

8.14 × 10−2  ±  6.79 × 10−2

Nateglinide

Secondary first

8.75 × 10−3

1.41 × 10−1

6.94 × 10−2  ±  1.28 × 10−2

5.85 × 10−2  ±  1.80 × 10−2

The apparent MRP4-dependent transportisotonic was obtained by subtracting the vesicle-to-medium ratiomock from the vesicle-to-medium ratioMRP4 in the absence of raffinose. The apparent MRP4-dependent transporthypertonic was obtained in the presence of 0.3 M raffinose. Each value was determined from the same data shown in Fig. 1. Each value of the apparent MRP4-dependent transport represents the mean ± SEM.

*p < 0.05, significantly different from the apparent MRP4-dependent transport isotonic, Student’s t test.

aThe apparent MRP4-dependent transport isotonic was obtained by subtracting the vesicle-to-medium ratio MRP4 in the presence of AMP from that in the presence of ATP, in the absence of raffinose.

We found that 17 compounds shown in Fig. 1(d–g, i–u) exhibited greater uptake by MRP4-expressing membrane vesicles (circles) than that by the control vesicles (squares) under isotonic conditions, and the uptake decreased linearly with the increase in medium osmolarity. Among these 17 compounds, the y-intercept values for eight compounds (Fig. 1; e, f, j, k, n, o, p and s) were less than 0 µl/(mg protein·min) (Table VI), indicating that these compounds (benzylpenicillin, cephalosporin C, alacepril, pravastatin, ceftazidime, bezafibrate, ampicillin and sulbactam) were transported into the intravesicular space by MRP4.

The y-intercept values for six other compounds, furosemide, tetracycline, bumetanide, baclofen, enalapril and pipemidic acid (Fig. 1; d, g, i, m, q and r) were 0.489, 0.267, 0.0529, 0.0191, 0.00303 and 0.00436 µl/(mg protein·min), respectively (Table VI), which indicate the binding amount of these compounds to the vesicle surface, MRP4 and/or filters. The apparent MRP4-dependent transport is the sum of the amount of compounds transported into the intravesicular space by MRP4 and that bound to MRP4, and is determined by subtracting the vesicle-to-medium ratioMock from the vesicle-to-medium ratioMRP4. As shown in Table VI, the y-intercept values of these six compounds were less than 31% of the apparent MRP4-dependent transport under isotonic conditions (1.74, 0.866, 0.299, 0.109, 0.0587 and 0.0218 μl/(mg protein·min) for furosemide, tetracycline, bumetanide, baclofen, enalapril and pipemidic acid, respectively). Therefore, the major part of the apparent MRP4-dependent transport does represent transport into the intravesicular space by MRP4, and so these six compounds are also substrates of MRP4.

The y-intercept values of the remaining three compounds, hydrochlorothiazide, meloxicam and nateglinide (0.0992, 0.426 and 0.141 μl/(mg protein·min)) were 97, 443 and 203% of the apparent MRP4-dependent transport under isotonic conditions (0.102, 0.0962 and 0.0694 μl/(mg protein·min)), respectively (Table VI). Thus, it appears that the apparent MRP4-dependent transport of these compounds can be mainly attributed to the binding to MRP4. To confirm this, the uptake by the control vesicles was examined under hypertonic conditions, and the apparent MRP4-dependent transport amounts were compared between isotonic and hypertonic conditions, since the binding to MRP4 is independent of osmotic pressure. As shown in Table VI, the apparent MRP4-dependent transport of hydrochlorothiazide under hypertonic conditions was significantly lower than that under isotonic conditions, indicating that hydrochlorothiazide is transported into the intravesicular space by MRP4. In contrast, for meloxicam and nateglinide, the apparent MRP4-dependent transport was similar under hypertonic and isotonic conditions, suggesting that meloxicam and nateglinide are not transported into the intravesicular space via MRP4, but rather bind to it.

Table VI summarizes the slope of the osmotic pressure dependency shown in Fig. 1 in order of the slope value. CMZ and rebamipide both showed elevated slopes of ATP-dependent transport into MRP4-expressing membrane vesicles (Table VI), confirming them to be strong substrates of human MRP4. The slopes of the osmotic pressure dependency for furosemide, benzylpenicillin, cephalosporin C, tetracycline, piperacillin, and bumetanide were of a similar order to that of MTX, which is a known substrate of MRP4. Alacepril, pravastatin, hydrochlorothiazide, baclofen, ceftazidime, bezafibrate, ampicillin, enalapril, pipemidic acid, and sulbactam were found to be relatively weak substrates of MRP4.

Characterization of CMZ Transport by Human MRP4

As CMZ exhibited the highest transport activity among the compounds examined, its activity was further characterized. Fig. 2a shows the time-course of CMZ transport by MRP4-expressing membrane vesicles in the presence of ATP, demonstrating a linear phase for at least 3 min with an initial uptake rate of 60.6 μl/(mg protein·min). The vesicle/medium ratio of CMZ at 10 min by MRP4-expressing membrane vesicles in the presence of ATP (409  ±  9 μl/mg protein) was significantly greater than that in the presence of AMP (13.9  ±  0.5 μl/mg protein) or into the control vesicles in the presence of ATP (5.80  ±  0.04 μl/mg protein). A concentration-dependence was established for CMZ transport by MRP4-expressing membrane vesicles, and is illustrated in Fig. 2b. The Km and Vmax values (mean ± standard deviation (SD)) obtained for CMZ transport were 28.5  ±  3.0 µM and 0.914  ±  0.029 nmol/(mg protein·min), respectively. Table VII summarizes the inhibitory effects exerted by MTX and DHEAS, which were selected as known MRP4 substrates, and indomethacin, ketoprofen, and diclofenac, which were selected as known MRP4 inhibitors. All of the compounds had significant inhibitory effects on the transport of CMZ into human MRP4-expressing membrane vesicles (Table VII). MTX (30 µM) inhibited CMZ transport by 39.4%, suggesting that the IC50 value of MTX for MRP4-mediated CMZ transport is in the range of 30 µM to 100 µM. This range is similar to the reported IC50 value (100 to 200 µM) of MTX for MRP4-mediated estradiol 17 β-D-glucuronide transport (23).
https://static-content.springer.com/image/art%3A10.1007%2Fs11095-007-9453-7/MediaObjects/11095_2007_9453_Fig2_HTML.gif
Fig. 2

Time-dependent and concentration-dependent transport of CMZ by MRP4. Membrane vesicles (10 µg) prepared from MRP4-transfected or parental vector-transfected Sf9 cells (Mock) were incubated at 37°C in uptake medium containing CMZ in the presence of 4 mM ATP or 4 mM AMP. a, time dependence of 1 µM CMZ uptake by MRP4. b, concentration-dependence of CMZ uptake by MRP4. The uptake rates were measured at 3 min, and were calculated by subtracting the values obtained in the presence of 4 mM AMP from those obtained in the presence of 4 mM ATP. Inset: Eadie–Hofstee plot of the same data. Each point represents the mean ± SEM (n = 3–4)

Table VII

Inhibitory effects of known substrates and inhibitors for MRP4 on CMZ uptake into MRP4-expressing membrane vesicles

Inhibitor

Concentration

Vesicle/medium ratio

% of control

μM

µl/(mg protein·3 min)

Control

 

208  ±  4

100  ±  2

Indomethacin

50

11.9  ±  0.3

5.72  ±  0.14*

Ketoprofen

50

66.2  ±  2.9

31.7  ±  1.4*

MTX

30

127  ±  5

60.6  ±  2.4*

DHEAS

3

143  ±  4

68.7  ±  2.0*

Diclofenac

100

166  ±  8

79.3  ±  3.9*

Membrane vesicles (10 µg) prepared from MRP4-transfected Sf9 cells were incubated at 37°C for 3 min in uptake medium containing 1 µM CMZ and 4 mM ATP in the presence or absence of the indicated inhibitors. The CMZ uptake into membrane vesicles prepared from parental vector-transfected Sf9 cells in the presence of 4 mM ATP was subtracted from all uptake values. Each value represents the mean ± SEM (n = 4).

*p < 0.001, significantly different from the control uptake, Dunnett’s test.

DISCUSSION

The present study describes the establishment of a comprehensive and rapid substrate-screening method (LC–MS/MS cocktail method) for MRP4. By employing this sequential screening method, 18 new substrates for MRP4 were identified from a set of 49 compounds; furosemide and hydrochlorothiazide, which were recently reported to be substrates of MRP4 (24), were also confirmed to be substrates. Substrate screening for ABC transporters has previously been performed rapidly by means of ATPase assays and inhibition assays against the vesicular uptake of a known substrate. Azidothymidine, 9-(2-phosphonylmethoxyethyl)adenine, and MTX are transport substrates for MRP4, but they do not stimulate ATP hydrolysis in MRP4-expressing membrane preparations (17). Urate is a transport substrate of MRP4, but does not inhibit the transport of cAMP into MRP4-expressing membrane vesicles (25). Therefore, these assays may detect false positives as substrates. By contrast, a vesicle-uptake study evaluates the direct transport of the substrate via the transporter. Therefore, the LC–MS/MS cocktail method described here is more suitable for substrate identification than either the ATPase assay or the inhibition assay. Since vesicle-uptake studies have been performed for various ABC transporters, the LC–MS/MS cocktail method should be applicable for the identification of substrates of other ABC transporter subtypes.

However, standards (10 pmol) of four compounds were not detected (Table II), suggesting that the LC–MS/MS cocktail method may not be applicable to all the compounds tested. The molecular weights of these four compounds are less than 250 (Table I), and such compounds are often difficult to detect by LC–MS/MS because of high background noise. In addition, isosorbide dinitrate lacks functional groups producing positive ions, such as amino groups. Alendronate has an amino group, but also has double phosphate groups. Such physicochemical properties make these compounds difficult to ionize in the positive-ionization mode.

The LC–MS/MS cocktail method is characterized as follows. The first screening is used to comprehensively identify the potent and moderate substrates in a mixture of many compounds. The second screening is used to efficiently characterize non-substrates as well as substrates, since dividing the compounds into groups results in a reduction in the effect of mutual inhibition. The third screening classifies all the remaining compounds as substrates or non-substrates. However, even if MRP4-dependent transport appears to be present, if the effect of osmotic pressure on the uptake by membrane vesicles is not evaluated, a compound binding to MRP4 might be mistakenly identified as a substrate. Therefore, the fourth step is necessary to demonstrate that the candidate is a substrate. Indeed, meloxicam and nateglinide were apparently transported by MRP4 in the first screening procedure. However, they were finally identified as compounds that were merely bound to MRP4, since no sensitivity of the apparent MRP4-dependent transport to osmotic pressure was observed (Table VI). Higgins et al. proposed the ATP switch model for ABC transporters, in which the transport cycle of ABC transporters is initiated by substrate binding, leading to an increase of the affinity of the nucleotide binding domain for ATP (26). Based on this model, the binding of meloxicam and nateglinide is likely to be ATP-independent. However, ATP-dependent binding can not be ruled out, and further studies will be needed to elucidate in detail the binding mechanisms.

In the present study, 12 and eight substrate candidates were identified in the first and third screenings, respectively. In the primary first screening, potent substrates (such as CMZ and rebamipide) were mainly identified (Tables II and VI). In the secondary first screening, moderate substrates were mainly identified (Tables III and IV). Although eight substrates were identified in the third screening, most had a weak ability to be transported by MRP4 (Tables V and VI). These results demonstrated that the first screening was useful for comprehensively identifying potent substrates. In contrast, the entire process (including the third screening) was necessary to identify all the substrates, since not all of substrates were identified in the first screening.

For the second screening, the compounds were divided into four groups. If the compounds were divided randomly, substrates, inhibitors and non-substrates would be equally distributed in all groups, and it is expected that none of the compounds is characterized as substrate candidates or non-substrates due to mutual inhibition in all groups. As a result, all compounds would have to be tested individually in the third screening. To minimize this problem, compounds that were expected to inhibit MRP4 were classified into group 2 or 3, so that the compounds of groups 1 and 4 could be efficiently characterized as substrate candidates or non-substrates.

β-Lactam antibiotics (amoxicillin and ceftazidime), ACE inhibitors (alacepril and captopril) and furosemide may inhibit MRP4 transport, because their structural analogues were identified as substrate candidates in the first screening. NSAIDs (sulindac, salicylate, mefenamic acid, piroxicam and meclofenamic acid), quinolone antibiotics (ofloxacin, ciprofloxacin, enoxacin and norfloxacin) and pravastatin may also inhibit MRP4 transport, since their structural analogues have been reported to be inhibitors of MRP4 and/or to interact with MRP2, which is a major subtype of MRPs (2729). These compounds were classified into group 2 or 3. As a result, 16 compounds in groups 3 and 4 were identified to be non-substrates, since no significant MRP4-dependent transport was observed for any of them under the condition of no inhibition of MTX transport via MRP4 (Table IV). Although the actual inhibitory effect in each group was contrary to our expectation, the second screening efficiently identified the non-substrates and decreased the number of compounds for the third screening. There still remains a possibility that these groups included compounds that did not inhibit MTX transport, but did inhibit the transport of other substrates via MRP4. Thus, the possibility of missing substrates cannot be completely ruled out.

The first and second screenings were performed by Q1 multichannel LC–MS, which did not require optimization of the measurement conditions for each compound. A simultaneous measurement was achieved simply by setting a common ionization condition and m/z channels assumed to be [monoisotopic mass + 1]. Thus, Q1 multichannel LC–MS is a simple and rapid quantification method, and it is suitable for substrate screening requiring speed rather than accuracy. In contrast, MRM multichannel LC–MS/MS, which was used in the third and fourth screenings, requires optimization of the MRM channel of each analyte before the simultaneous measurement. Despite requiring time for optimization, it has an approximately 100-fold higher sensitivity than Q1 multichannel LC–MS, allowing the use of small amounts of membrane vesicles. Therefore, MRM multichannel LC–MS/MS is suitable for the precise identification of substrates by individual transport studies for many compounds.

This study showed that CMZ and rebamipide are potent substrates of MRP4 compared with MTX and 16 further substrates. The reported MRP4 substrates (with Km values <100 µM) were mainly endogenous compounds, such as steroids, prostaglandins, and cyclic nucleotides, and included only one drug, topotecan (Km = 1.66 µM) (1,30). Transport of CMZ by MRP4 exhibited a Km value of 28.5 µM, indicating that CMZ has a relatively high affinity for MRP4 among drugs. The intravesicular volume of membrane vesicles prepared from various species was reported to be from 0.60 to 10 µl/mg protein (3133), although that of Sf9 cell membrane vesicles is not known. These results suggest that the uptake of CMZ and rebamipide into membrane vesicles is an uphill process against a concentration gradient, since the vesicle-to-medium ratios of these compounds are much greater than the reported intravesicular volume, as shown in Figs. 1b and 2a (rebamipide, 155 µl/mg protein at 3 min; CMZ, 409 µl/mg protein at 10 min). This supports the conclusion that these drugs are potent substrates of MRP4.

Based on the creatinine clearance (approximately 100 ml/min) and the binding ratio to serum protein (84% for CMZ and 98% for rebamipide) (34,35), the clearance rates of CMZ and rebamipide attributable to glomerular filtration alone can be calculated to be 16 and 2 ml/min respectively. These values are far smaller than the apparent renal clearances of CMZ and rebamipide, which have been reported to be 111 and 287 ml/min, respectively (36,37). This implies that CMZ and rebamipide are mainly eliminated by tubular secretion, and that high expression of MRP4 at the apical membrane of the proximal tubules plays an important role in the elimination process.

A recent study showed that the mRNA level of MRP4 in the kidney is the highest among the transporters (including MRP2, MRP4, OAT4, and ABCG2) that contribute to the apical efflux transport of organic anions in the proximal tubules (4). However, the identified drug substrates for MRP4 have been limited to a few types of compound, such as nucleotide, camptothecin, and folate analogs (1). The present study has identified 18 drugs as substrates for MRP4, and revealed that MRP4 transports various anionic drugs, including β-lactam antibiotics, diuretics, and ACE inhibitors. It was previously reported that these drugs are eliminated by tubular secretion from the kidney (6,7,9), and Mrp4 is involved in the apical efflux of anionic drugs, such as adefovir and tenofovir, in vivo (3). The present study suggests that MRP4 is, at least partially, involved in the apical efflux of various anionic drugs, including β-lactam antibiotics, diuretics, and ACE inhibitors.

The concentration of the centrally acting antispastic agent, baclofen, in the brain interstitial fluid (ISF) was 27-fold lower than that in the plasma, and the restricted distribution in the brain ISF is ascribed to the efficient efflux from the brain through the blood–brain barrier (BBB) (11). We suggested that OAT3, localized at the abluminal membrane of the brain capillary endothelium, promotes the entry of baclofen into the endothelium from the brain (38). Our present study suggested that MRP4, localized at the luminal membrane (39), is involved in the efflux transport of baclofen into the plasma from the endothelium. Although P-glycoprotein (P-gp) plays an important role in the efflux transport of various drugs at the luminal membrane, it has not been reported to transport baclofen. Additionally, it has been reported that the concentration of baclofen in the brain ISF was markedly increased by administration of probenecid, which is an inhibitor not of P-gp, but of MRP (11,40). Consequently, MRP4 might be one of the important efflux routes of baclofen at the BBB.

The distribution of β-lactam antibiotics in the brain is restricted, resulting in reduced neurotoxicity and difficulties in treating bacterial CNS infections (13,41). Recent studies have suggested that both OAT3, which is localized at the abluminal membrane of the BBB and the apical membrane of the blood–cerebrospinal fluid barrier (BCSFB), and peptide-transporter 2 (PEPT2), which is localized at the apical membrane of the BCSFB, contribute to the restricted distribution of β-lactam antibiotics in the brain (4244). However, no efflux transporter of β-lactam antibiotics has been identified on the blood side. In the present study, we identified β-lactam antibiotics as substrates of MRP4, which is an efflux transporter localized on the blood side of the BBB and BCSFB. Therefore, our present findings will improve our understanding of the restricted distribution of β-lactam antibiotics in the brain.

In conclusion, the present study demonstrated that the LC–MS/MS cocktail method is useful for comprehensive substrate screening for MRP4, and identified 18 compounds as new substrates of human MRP4. Furthermore, it identified CMZ and rebamipide as potent substrates. Using this methodology, we established a new substrate profile for MRP4. Our results indicate that MRP4 is a new efflux route for various anionic drugs in the kidney, the BBB and the BCSFB. Our approach is therefore expected to be widely useful for the characterization of ABC transporter function. Furthermore, it should also be useful for lead optimization from compound banks during early drug development.

Acknowledgments

We thank H. Yabuuchi (GenoMembrane, Yokohama, Japan) for kindly supplying the human MRP4-expressing membrane vesicles and the control vesicles. We also thank N. Funayama for secretarial assistance. This study was supported, in part, by a Grant-in-Aid for Scientific Research on Priority Area 17081002 from The Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan, and by a 21st Century Center of Excellence (COE) Program Grant from the Japan Society for the Promotion of Science. This study was also supported, in part, by the Industrial Technology Research Grant Program from New Energy and the Industrial Technology Development Organization (NEDO) of Japan.

Copyright information

© Springer Science+Business Media, LLC 2007