Experimental Procedures

Definitions of Nonsubstrate, Inhibitor, and Substrate

A nonsubstrate was negative in all three assays (Table2). An inhibitor was positive only in the calcein-AM assay. A transported substrate was positive in the efflux assay regardless of response in the other assays. A nontransported substrate was negative in the efflux assay but positive in the ATPase assay and the calcein-AM assay. An unambiguous substrate was positive in all three assays.

Materials

GlaxoSmithKline Chemical Registry supplied all test substances. All other chemicals except radiolabeled compounds and cell culture reagents were purchased from Sigma (St. Louis, MO). Cell culture reagents were purchased from Invitrogen (Carlsbad, CA).dl-[4-³H]Propranolol (15–30 Ci/mmol) and d-[1-¹⁴C]mannitol (50–63 mCi/mmol) were purchased from Amersham Pharmacia Biotech (Arlington Heights, IL). Transwells (12-well, 11-mm diameter, 0.4-μm pores) were purchased from Corning Costar (Cambridge, MA).

Monolayer Efflux Studies

MDR1-MDCKII cells were obtained from The Netherlands Cancer Institute (Amsterdam, The Netherlands). Cells were passaged and grown as described (Lentz et al., 2000), with cells split twice weekly at a ratio of 1:10. Cells were not grown in the presence of any selection agent to maintain Pgp expression. For transport studies, cells were seeded onto polycarbonate Transwell filter membranes at a density of 300,000 cell/cm², and monolayers were ready for studies 3 days later. Compounds were dissolved at 20 mM in 100% DMSO and dilutions for studies prepared in transport buffer (8.1 mM Na₂HPO₄, 138 mM NaCl, 0.5 mM MgCl₂, 1.47 mM KH₂PO₄, 2.67 mM KCl, 0.9 mM CaCl₂, 5.6 mM glucose, and 0.33 mM sodium pyruvate, pH 7.4). Compounds were tested at either 10 or 20 μM concentration and in two directions (A→B and B→A) in triplicate. Monolayer efflux studies were conducted at 37°C in a humidified incubator with shaking (90 rpm) for 60 min. Transendothelial electrical resistance was measured with an Endohm Meter (World Precision Instruments, Sarasota, FL). Markers for paracellular ([¹⁴C]mannitol), transcellular ([³H]propranolol), and Pgp (amprenavir) efflux were included in each experiment. Concentrations of [¹⁴C]mannitol and [³H]propranolol were measured by liquid scintillation counting with Ready Safe Liquid Scintillation Fluid (Beckman Coulter, Inc., Foster City, CA) by using a Beckman Coulter LS501 counter. Amprenavir was analyzed by cassette liquid chromatography with tandem mass spectrometry (LC/MS/MS) along with the test compounds.

Monolayer Efflux Studies: Bioanalysis by High-Throughput LC/MS/MS Analysis

All analyses were performed by dual LC/MS/MS and cassette analysis (Polli et al., 2000; Wring et al., 2000). High-performance liquid chromatography was performed on a Hewlett Packard 1100 (Hewlett Packard, Palo Alto, CA) equipped with a column-switching valve. The sample (injection volume, 10 or 20 μl) was loaded on column by means of a Gilson 215 autosampler (Gilson Medical Electronics, Middleton, WI) controlled using a proprietary software add-in to a Hewlett Packard Chemstation. Chromatography was performed on Phenomenex Aqua C₁₈ columns (30 × 2 mm i.d., 3 μm; Phenomenex, Torrance, CA) at a flow rate of 0.6 ml/min. The mobile phase consisted of two solvents: 10 mM ammonium formate, pH 3.5, with 1.5% (v/v) methanol (A), and 100% acetonitrile (B). The gradient profile was 0 to 2.0 min 1% (v/v) B; 2.0 to 3.0 min linear gradient to 95% (v/v) B; 3.0 to 3.9 min 95% (v/v) B; 3.9 to 4.0 min linear gradient to 1% (v/v) B; and 4.0 to 4.6 min 1% (v/v) B. Mass spectrometry was performed on a PE Sciex API2000 (PE Sciex, Toronto, ON, Canada) equipped with either a turbo ion spray source for electrospray ionization or a heated nebulizer source for atmospheric pressure chemical ionization; polarity was selected for optimum sensitivity. Detection by tandem mass spectrometry was based on precursor ion transitions to the strongest intensity product ions. Key instrumental conditions were optimized to yield best sensitivity. Selected ion monitoring was used if tandem MS afforded inadequate response. Simultaneous assay of analytical standards or test samples was performed using cassette analysis, where samples or standards containing test compounds were pooled prior to injection (3–6 compounds/cassette). This reduced the total number of injections required and allowed analytical work to typically be completed within 6 h of each permeability study. The calibration range was typically 0.5 nM to 1.5 μM (n = 6) for each compound. Dose and donor solutions were diluted in transport medium + acetonitrile (1 + 1, v/v) as required to bring their concentrations into this range. Concentration of drug in the samples was calculated from the chromatographic peak area by using proprietary software developed at GlaxoSmithKline.

Monolayer Efflux Studies: Calculations

Apparent Permeability.

The apparent permeability (P_app) was calculated with the equation P_app = 1/AC₀ (dQ/dt), where A is membrane surface area, C₀ is donor drug concentration at t = 0, and dQ/dt is amount of drug transported within a given time period. Data are presented as the average P_app (nm/s) ± standard deviation from three monolayers. A ratio of the B→A/A→B P_app values was calculated. Involvement of a Pgp-mediated efflux mechanism is indicated if the B→A/A→B ratio is >2.0. For compounds with B→A/A→B ratios of 1.5 to 2.0, a follow-up experiment with 2 μM GF120918 (a potent, specific Pgp inhibitor) was completed to confirm that the compound was a Pgp substrate. In all cases, in the presence of GF120918, the B→A/A→B ratio was reduced to ∼1.0. Such results were interpreted to indicate that the compound underwent Pgp-mediated efflux.

Mass Balance.

Mass balance (MB) is the percentage of original compound mass accounted for at the end of the experiment (sum of amount in the A and B chambers). MB is calculated with the following equation: MB = [(C_At∗V_A) + (C_Bt∗V_B)]/(C₀∗V_D), where C_At andC_Bt are the concentrations in the A and B chambers at time (t),C₀ is the concentration of the donor at t = 0, V_A andV_B are the volumes of the apical and basolateral chambers, and V_D is the volume of the donor.

Western Blot Analysis

MDR1-MDCKII cell lysates were prepared as described (Evers et al., 1998) and protein concentrations determined with the bicinchoninic acid protein assay kit (Pierce Chemical, Rockford, IL). Cell lysates were resolved on 8% SDS-polyacrylamide gel electrophoresis precast mini-gels and transferred to nitrocellulose membrane following the protocol provided by the manufacturer (Novex, San Diego, CA). The blots were stained with Ponceau S solution (0.1% Ponceau S in 5% acetic acid) to confirm equal loading of protein. The blot was incubated with the C219 anti-Pgp antibody (Signet Laboratories, Dedham, MA) diluted 400-fold in Blotto blocking buffer (Pierce Chemical) overnight at 4°C. The blot was washed three times with Tris-buffered saline containing 0.1% Tween 20 and incubated with a horseradish peroxidase-conjugated secondary antibody (Amersham Pharmacia Biotech UK, Ltd., Little Chalfont, Buckinghamshire, UK) for 2 h at room temperature. After the blot was washed three times with Tris-buffered saline, immunoreactive bands were visualized by using the enhanced chemiluminescence plus (ECL+) detection system (Amersham Pharmacia Biotech UK, Ltd.).

Monolayer Efflux Studies: Characterization of MDR1-MDCKII Cell Line

To determine the robustness and reliability of the MDR1-MDCKII cell line, Pgp expression and activity levels were monitored over passage numbers 29 to 49, which covered 2.5 months of cell culture. Expression levels were similar between passages 33 and 49 (Fig.1). Pgp protein levels at passages 29 and 31 were lower because these were the initial passages after being seeded directly from liquid nitrogen (P28 was the original frozen passage number). The initial low expression is probably a result of the cells acclimating to the cell culture conditions. Pgp activity, as determined by transport of the human immunodeficiency virus protease inhibitor amprenavir, was very stable from passages 30 to 50 (Table3). The P_app values and B→A/A→B ratio varied less than 15% over these passages. Pgp functional activity was more consistent between passages than Pgp protein levels detected by Western blotting. For efflux studies, functional activity appears to be the best way to monitor the robustness of the cell line. Markers of transcellular (propranolol) and paracellular (mannitol) transport are also consistent over passages 30 to 50 (Table 3). Our current protocol is to use the cells for 3 months (passages 33–52) and then begin a new culture. However, we have not observed any loss in activity, alterations in cellular morphology, or other deleterious changes that would prohibit using the cell line as far as passage 64. Electron microscopic studies demonstrated that under these plating and culturing conditions the cells form a monolayer (J. W. Polli, unpublished data).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Pgp expression levels over 20 passages. Pgp levels were monitored over passages 29 to 49 by Western blotting with the C219 antibody. Pgp expression was stable after passage 33.

Drug-Stimulated Pgp ATPase Activity

Pgp expressing Spodoptera frugiperda (Sf9) membranes were purchased from GENTEST (Woburn, MA) and drug-stimulated Pgp ATPase activity was estimated by measuring inorganic phosphate released from ATP according to the manufacturer's protocol. The test concentration of 20 μM was selected based on previous published studies that demonstrated that this concentration would provide good ATPase activation for the majority of compounds (Litman et al., 1997). Briefly, membranes (25 μg) were incubated at 37°C for 5 min in 30 μl Tris-MES buffer, pH 6.8 (50 mM Tris-MES, pH 6.8, 50 mM KCl, 5 mM sodium azide, 2 mM EGTA) and 20 μM each test compound in the presence or absence of 200 μM sodium orthovanadate in duplicate wells of a 96-well plate. The reaction was started by the addition of 30 μl 10 mM ATP (magnesium salt) and was stopped 20 min later by addition of 30 μl of 10% SDS containing antifoam A. Detection reagent (180 μl, 1:4 35 mM ammonium molybdate in 15 mM zinc acetate, pH 5.0, and 10% ascorbic acid, pH 5.0) was added to all wells and incubated at 37°C for 20 min. The absorbance at 800 nm was measured via a microplate spectrophotometer. The drug-stimulated ATPase activity (nmol/min/mg of protein) was determined as the difference between the amounts of inorganic phosphate released from ATP in the absence and presence of vanadate. Phosphate standards were prepared in each plate and verapamil served as a positive control. Drug-stimulated Pgp ATPase activity was reported as fold-stimulation relative to the basal Pgp ATPase activity in the absence of drug (DMSO control). A compound was classified as an activator if the fold-stimulation was greater than 2-fold over the DMSO control.

Calcein Inhibition Assay

The calcein-AM assay was optimized from the original methods reported in the literature (Liminga et al., 1994; Tiberghien and Loor, 1996). The optimization steps included determining cell culture conditions for plating, using the MDR1-MDCKII cell line (e.g., plating density, media, number days in culture), and maximizing the fluorescent signal-to-noise ratio (e.g., calcein-AM concentrations, use of black-sided 96-well plates, time of assay, compound concentrations). Cells were seeded at 70,000 cells/well (200 μl of culture medium) in 96-well black plates with clear bottoms (Packard Instrument Co., Meridian, CT). Cells were fed 24 h after seeding and the assay performed 48 h later. Medium was removed and monolayers washed three times with transport buffer. Test compounds were added to monolayers in 50 μl of transport buffer containing 1% DMSO as solvent. DMSO concentration was constant in test and control wells (each n = 2). Plates were incubated at 37°C for 10 min. Calcein-AM (Molecular Probes, Eugene, OR) was added at 10 μM in 50 μl of transport buffer to give a final concentration of 5 μM. Plates were immediately placed in a SpectraMax Gemini cytofluorimeter (Molecular Devices, Sunnyvale, CA) for 60 min and read at 15-min intervals at 485-nm excitation and 530-nm emission.

The assay was initially calibrated via dose response for GF120918, verapamil, vinblastine, vincristine, and ritonavir (Fig.2). Maximum calcein fluorescence measured in relative fluorescence units (RFU) was seen at 1.0 μM GF120918. Therefore, 1 μM GF120918 was used in all subsequent assays as a positive control and considered to give maximum response. Based on the results for verapamil, vinblastine, vincristine, and ritonavir, test concentrations of 50, 100, and 200 μM were selected for the remaining compounds. Quantification of Pgp inhibition was completed using the following equation: % maximum = (RFU_comp − RFU_background)/(RFU_GF120918− RFU_background)∗100, where RFU_comp is fluorescence in the presence of 100 μM test compound (comp), RFU_GF120918 is fluorescence in the presence of 1 μM GF120918, and RFU_background is fluorescence in absence of the compound (typically 45–65 RFU). A compound was determined to inhibit Pgp when % maximum was >10%.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Dose response of various compounds in the calcein-AM assay. Data are reported as the RFU values from a 60-min incubation with 5 μM calcein-AM. Background fluorescence was 45 to 65 RFU.

Results

Overview of Individual Assay Results

Monolayer Efflux Results.

The P_{app A→B} values ranged from 0.94 to 916 nm/s and B→A/A→B ratios from 0.67 to 165 nm/s for the 66 tested compounds (Tables4-11). Thirty-nine compounds underwent efflux and were classified as Pgp substrates based on the B→A/A→B ratio >2.0, whereas 27 compounds did not undergo efflux (B→A/A→B ratio <1.5). Discrimination between noneffluxed and effluxed compounds was clear. Four of the 39 compounds that were transported (BW1351W91, diltiazem, loratadine, and neostigmine) had a B→A/A→B ratio between 1.5 and 2.0. Of these four, three had P_{app A→B} values >260 nm/s and one had a P_{app A→B} value <10 nm/s. The B→A/A→B ratios for these compounds was reduced to ∼1.0 in the presence of GF120918, a specific and potent Pgp inhibitor. Hence, each was classified as a transported substrate.

Drug-Stimulated Pgp ATPase Assay.

Of the 66 compounds tested, 35 stimulated ATPase activity, 30 were inactive and 1 inhibited basal ATPase activity (GF120918, ratio = 0.23, Table 8). The range of activity was 0.23- to 12.1-fold and verapamil served as a positive control (Tables 6-11). There was clear separation between stimulators of ATPase from those that were inactive. Only one compound, daunorubicin (Table 10), fell into the nonconfident range of ATPase stimulation between 1.5 and 2.0, a region where it is difficult to classify a compound as a stimulator. Because the stimulation ratio was below 2-fold, daunorubicin was classified as inactive in the ATPase assay.

Calcein-AM Assay.

Of the 66 compounds tested, 32 increased calcein fluorescence and 34 did not (Tables 6-11). The % maximum response, defined by dividing the test compound's RFU response by the GF120918 RFU response, ranged from −3.50 to 102. Only three compounds (vinorelbine, vincristine, and propranolol) had % maximum response between 6 and 12%, the nonconfident zone. Because all these compounds had a response <10% of maximum, they were classified as being negative in the assay.

Comparison of Results from Monolayer Efflux, ATPase, and Calcein-AM Assays

Results from the three assays were grouped according to whether there was concordance between the assays (Tables4 and 5). Category I had concordance across the three assays and contained 14 unambiguous nonsubstrates (Table 6) and 18 unambiguous substrates (Table 7) that had mean P_{app A→B} values of 240 and 131 nm/s, respectively. There was a striking contrast in P_{app A→B} distribution between the nonsubstrates and substrates. For nonsubstrates, 12 of 14 (85.7%) compounds had P_{app A→B} values either <20 or >300 nm/s, whereas only 4 of 18 (22.2%) substrates had P_{app A→B} in this range. Category I also included three possible nonsubstrate inhibitors, i.e., negative for efflux and ATPase activity but positive for calcein-AM inhibition (Table8).

Category II compounds illuminated differences among the assays (Table5) and segregated into two groups defined by the absence of monolayer efflux (group IIA, 10 of 31 compounds, individual compounds listed in Table 9) or presence of monolayer efflux (group IIB, 21 of 31 compounds, individual compounds listed in Tables10 and11). Group IIB was subdivided further based on discordance between monolayer efflux with both ATPase and calcein-AM results (group IIB₁, Table 10), or with either the ATPase or calcein-AM results (groups IIB₂ and IIB₃, Table 11).

Group IIA compounds did not undergo observable monolayer efflux but were positive in both the calcein-AM and ATPase (Table 9). These compounds were thus classified as nontransported substrates. A distinguishing feature of IIA compounds is exceptionally high permeability. The mean P_{app A→B} value for this group was 37- and 26-fold larger than that for groups IIB₁ and IIB₂ (Table 5), groups where compounds had measurable efflux (Tables 10 and 11). No compound in group IIB had a P_{app A→B} >75 nm/s, whereas the lowest P_{app A→B} for IIA was 316 nm/s. Also, the mean P_{app A→B} for IIA was 4-fold larger than that for category I substrates, which had a mean P_{app A→B} = 131 nm/s (Table 4). Only one category I compound, diltiazem, had a P_{app A→B}> 316 nm/s, which was the lowest P_{app A→B} in group IIA (ketoconazole, Table 9). High permeability is a key characteristic of compounds in group IIA.

The notable feature of group IIB compounds is poor permeability. Group IIB₁ compounds were positive for efflux but negative in both the ATPase and calcein-AM assays (Table 10). Group IIB₁ had a mean P_{app A→B}9-fold lower than that for category I substrates (Table 4), compounds that were positive in all three assays. Only 3 of 18 category I substrates had P_{app A→B} <20 nm/s compared with 10 of 13 group IIB₁ compounds. The mean B→A/A→B ratio was similar for category I and group IIB₁ substrates (Tables 7 and 10), suggesting that the relative contribution of Pgp-mediated efflux to the P_{app A→B} value was similar.

Group IIB₂ substrates (n = 7) had concordance between monolayer efflux and ATPase (Table 11) and had low permeability (mean P_{app A→B} = 20.4 nm/s) as seen for group IIB₁. Cyclosporin A was the only compound that underwent efflux, was a calcein-AM inhibitor, and was inactive in the ATPase assay (group IIB₃, Table11).

In this set of 66 compounds, none were positive in the ATPase assay and negative in both the monolayer efflux and calcein-AM assays (NYN, group IIC, Table 5).

Discussion

Recombinant expression technology has provided unprecedented access to human Pgp. Assay formats for classification of substrates and inhibitors are varied in the literature and as yet not standardized. The abundance of functional Pgp, accessibility of the compound to Pgp, and analytical method are factors influencing the sensitivity of these assays. Of the 66 compounds tested, 52 (78.8%) were positive in one or more of the assays and only 18 compounds (27.3%) were positive in all three assays. A similar finding has been reported by Scala et al. (1997) who demonstrated that Pgp substrates and antagonists cluster into two groups, and that only 7 of 84 compounds positive in at least one Pgp assay were positive in both assays. Thus, whether compounds are observed to be Pgp substrates depends on which assay is used. The consistency in performing the assays also determines the outcome. In this article, we provide a consistent approach by using a large set of compounds with diverse chemical structure, affinity for Pgp, and membrane permeability. We have directly evaluated assay formats and identified compounds that are suitable as benchmarks for classification of Pgp substrates.

Unambiguous substrates and nonsubstrates (compounds in category I) represented a wide range of structure and membrane permeabilities. A significant proportion (14 of 18, 77.8%) of the unambiguous substrates had P_{app A→B} values between 20 and 300 nm/s. This defines the optimal overlapping P_app range of these assays and provides a guideline for estimating whether these assays are likely to provide concordant results for substrates. Loperamide, terfenadine, and quinidine are representative positive controls for all three assays that are readily available and convenient to measure analytically. Two of the 14 (14.3%) unambiguous nonsubstrates with P_{app A→B} values in the optimal range are amantidine and triampterin, and these compounds are suitable negative controls. A number of category I nonsubstrates that fall outside the optimal P_{app A→B} range (doxorubicin, itraconazole, methotrexate, propranolol, ranitidine, and yohimbine) has been classified as substrates in the literature (Seelig, 1998). This is arguably due to the difference in assay sensitivity and/or threshold defined as a positive response.

The 10 compounds not effluxed, but positive in the ATPase and calcein-AM assays (category IIA), have high P_{app A→B} values (mean = 535 nm/s) and thus also have high passive permeability. Many have been reported to be Pgp modulators (Seelig, 1998; Ferte, 2000). This indicates that the monolayer efflux assay tends to be insensitive to highly permeable compounds (P_{app A→B} > 300 nm/s), yielding false negative results. This is likely due to rapid flux of compound through the plasma membrane leading to either insufficient drug concentrations within the inner membrane leaflet or a saturation of Pgp activity (Eytan et al., 1996). When compounds in category IIA were tested at a 10-fold lower concentration (2 μM), one compound (verapamil) had undergone efflux, three compounds did not undergo efflux (midazolam, mebendazole, and nifedipine), and five were indeterminate due to poor mass balance (<50%; data provided to reviewers). Thus, the ATPase and calcein-AM assays are best suited for chemical series with high membrane permeability (>300 nm/s) and high cellular accumulation (associated with good permeability), or to corroborate a borderline monolayer efflux result (e.g., diltiazem, Table 7). The distinct advantage of calcein-AM assay is that it can used to identify compounds that inhibit Pgp, and we recommend test concentrations of 1, 10, and 100 μM (focus on inhibition) versus the 50, 100, and 200 μM (focused on substrates) used in this study. Category IIA defines a set of high-permeability compounds to further investigate the relationship between nominal test concentration, concentration at the active site, and affinity for Pgp.

Category IIB substrates were positive for monolayer efflux but were not positive in the ATPase and/or calcein-AM assays. Category IIB₁ compounds were negative in both ATPase and calcein-AM assays. One feature of these compounds is very low P_{app A→B} (mean 14.6 nm/s) coupled with modest P_{app B→A} (mean 105 nm/s; mean B→A/A→B ratio = 11.6), i.e., low membrane permeability and transcellular flux. In contrast, group IIB₂ substrates, which stimulated ATPase activity and were negative in calcein-AM, had low P_{app A→B} (mean 20.4 nm/s), high P_{app B→A} (mean = 255 nm/s; 6 of 7 compounds having P_{app B→A} >135 nm/s), and larger B→A/A→B ratios (mean = 21.7). These data in conjunction with preliminary studies with GF120918 to determine membrane permeability suggest that group IIB₂ compounds have better passive permeability and transcellular flux than group IIB₁ compounds (data provided to reviewers). The striking association of high flux (group IIB₂) coinciding with ATPase response, and low flux (group IIB₁) coinciding with nonresponse, suggests that membrane concentrations of group IIB₁ substrates critical for activation of ATPase are not attained. Group IIB₁ compounds define a region of poor response (due to limited permeability) for ATPase and calcein-AM, whereas group IIB₂ defines the region of poor response for calcein-AM. Critical membrane concentration is a function of binding affinity. These compounds may have low binding affinities, and so are unable to displace other substrates (e.g., endogenous lipids or calcein-AM). Further studies are needed to determine what is unique to these compound sets (groups IIB₁ and IIB₂), and to determine the role membrane permeability and affinity has on ATPase stimulation and calcein-AM response.

There was no instance of a compound that was positive in the ATPase assay and negative in the other assays (NYN, group IIC). An explanation for this is that a false N in the monolayer efflux assay requires high permeability, whereas a false N in calcein-AM requires low permeability, which are mutually exclusive.

This study provides rational criteria for selecting a Pgp substrate assay and interpreting results. The compound set was sufficiently large and diverse to elucidate the strengths and weaknesses of the individual assays. Because Pgp efflux is more relevant to in vivo drug disposition when passive permeability is limiting, the selection of an assay must be biased toward responsiveness to compounds with low/moderate permeability. Therefore, we have chosen the monolayer efflux assay as our primary Pgp substrate screen, despite the significant analytical requirement. Offsetting the analytical burden is the simultaneous determination of the compound's passive membrane permeability (P_{app A→B} in the presence of GF120918) and a more “functionally” relevant endpoint (efflux). The tendency of the efflux assay to fail with highly permeable substrates is acceptable as Pgp is not likely to be a significant barrier for compounds with high permeability (category IIA). This study has provided a sound basis for a strategy for deploying these Pgp substrate assays in the early stages of drug discovery.