INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

P-GLYCOPROTEIN is an abundant, 170,000-dalton plasma membrane glycoprotein that confers cellular multidrug resistance (mdr) to a wide variety of hydrophobic drugs, such as colchicine and vinblastine (29). P-glycoproteins are members of a superfamily of gene products that include ATP-binding cassette transporters like the cystic fibrosis transmembrane conductance regulator and bacterial permeases (32). In humans, P-glycoproteins are encoded by two genes, designated MDR1 and MDR3 (9, 60). Mdr3p is unable to confer cellular drug resistance, but it is responsible for specifically translocating phosphatidylcholine across the liver canalicular duct membrane into bile (56). Although Mdr1p can transport phosphatidylcholine, its substrate specificity is much broader than Mdr3p, since it translocates numerous xenobiotic drugs and short-chain lipid analogs (61), in addition to interacting with a wide variety of endogenous compounds like progesterone (64), prenylcysteines (65, 66), and possibly cholesterol (15). Mdr1p is expressed in a variety of normal tissues like the gastrointestinal tract and kidney (11, 59), where it may function to remove toxic compounds from the body. Renal P-glycoprotein forms a transepithelial drug transport pathway (34) that is responsible for the net urinary excretion of various xenobiotics in vivo (16, 17). By virtue of its broad substrate specificity, P-glycoprotein may also protect mucosal surfaces from the cytotoxic effects of various hydrophobic compounds in luminal fluids (35). The expression of Mdr1p in various nonexcretory cells like renal glomerular mesangial cells (1, 22) and tight endothelia, which form the blood-brain barrier (12, 38), also suggests that this isoform serves a protective function. Murine gene knockouts have confirmed P-glycoprotein's role in the pharmacokinetic handling of various drugs, particularly in the brain (53, 54). Mice lacking mdr1a/b genes, however, show normal fertility and viability (52). Expression of Mdr1p may still be important for the long-term health and propagation of various species, since P-glycoprotein will afford some tissues protection from absorbed cytotoxic and mutagenic agents.

Because Mdr1p is expressed in the apical membrane of renal epithelia (11, 42, 59), it seemed plausible that transported substrates of P-glycoprotein would be present in urine. Indeed, modulators of P-glycoprotein function have been detected in both human serum (36) and rat urine (7). With a competitive, cellular [³H]cyclosporin A accumulation assay to screen for human urinary compounds capable of reversing mdr (28), the synthetic surfactant nonylphenol ethoxylate (NPE) was isolated. NPE chemosensitizes multidrug-resistant C5 cells by increasing the cellular accumulation of various Mdr1p substrates including [³H]cyclosporin A. NPE also blocks [³H]azidopine photolabeling of P-glycoprotein in both plasma membranes from multidrug-resistant C5 cells and human kidney brush-border membranes (BBM). NPE also blocks the transepithelial transport of Mdr1p substrates like [³H]cyclosporin A by kidney epithelia, and transport studies using [³H]NPE show that urinary excretion of this absorbed surfactant is likely mediated by kidney P-glycoprotein.

Widely used synthetic surfactants like NPE are readily absorbed and rapidly removed from the body by gastrointestinal and renal excretion routes (20, 40, 47). P-glycoprotein is an ideal candidate for this excretory role, since it is expressed in the liver, gastrointestinal tract, and kidney and can transport a wide variety of hydrophobic compounds, including nonionic detergents (43).

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Cells and materials. Multidrug-resistant (C5, B30) and drug-sensitive (Aux-B1) Chinese hamster ovary (CHO) cell lines were obtained from Dr. V. Ling (British Columbia Cancer Institute, Vancouver, BC). Madin-Darby canine kidney (MDCK) cells were obtained from the American Type Culture Collection. Outdated human plasma was obtained from the Canadian Red Cross. Human renal cortical tissue was isolated from unused kidney donor transplants obtained from the Multiple Organ Retrieval Unit at the Toronto Hospital. [³H]cyclosporin A and [³H]azidopine were purchased from Amersham; [³H]NaBH₄ and [³H]digoxin were purchased from DuPont-NEN; tissue culture media, serum, and antibiotics were purchased from Life Technologies; verapamil and colchicine were from Sigma Chemical; and the liquid scintillation cocktail, Cytoscint ES, was from ICN. Cyclosporin A and Tergitol NP-9 were gifts from Sandoz and Union Carbide, respectively.

Detection of urinary NPE. For screening urinary components that interact with P-glycoprotein, a competitive drug accumulation assay was used (28). C5 cells were grown to confluence in 96-well microtiter plates in -minimum essential medium (-MEM) supplemented with 10% calf serum, 50 units/ml penicillin, and 50 µg/ml streptomycin. Cells were washed with 200 µl/well of -MEM, buffered with 25 mM HEPES (pH 7.0), then incubated at 37°C for 30 min in 100 µl of the above medium containing 50 nM [³H]cyclosporin A (7.5 Ci/mmol) and appropriate dilutions of individual C₁₈ HPLC fractions reconstituted in 100 µl dimethyl sulfoxide (DMSO). Plates were then washed twice with ice-cold PBS, and cells were released from wells by incubation at 37°C for 10 min with 200 µl/well 0.1% trypsin in PBS. Radioactivity of each well of cells was measured in a model LS7500 liquid scintillation spectrometer (Beckman Instruments).

Solid-phase extraction of human urine. Urine was collected in nitric acid-treated reagent bottles from a single, healthy, medication-free male donor over a 1-yr period. Each batch of 40-50 liters was filtered to remove particulate material, then chromatographed on 5-Mega Bond Elut (Varian) solid-phase extraction (SPE) columns containing 30 g C₁₈ resin each. Columns were washed with 250 ml each of 15% isopropanol/15% acetonitrile, then eluted sequentially with 50 ml each of 1) 25% isopropanol/25% acetonitrile, 2) 40% isopropanol/40% acetonitrile, and 3) methanol. Respective eluates were pooled, dried, then extracted with 30 ml methanol. Insoluble material was removed by centrifugation, and supernatants were further clarified by centrifugation in SR-3 concentrators (Amicon). Filtrates were dried, and extracts were reconstituted in 0.5 ml DMSO.

C₁₈ HPLC purification of urinary NPE. Reverse-phase C₁₈ HPLC on a Micosorb MV column (Rainin) was carried out on 100-µl aliquots of the solid-phase extract obtained with solvent mixture 2 described above. Optical density was monitored at 280 nm. The C₁₈ HPLC column was washed isocratically at 1 ml/min for 160 min, using 15% isopropanol/15% acetonitrile. Bound material was eluted at 1 ml/min over an 80-min period, using a linear solvent gradient ending with 50% isopropanol/50% acetonitrile. Eighty 1-ml fractions were collected, dried, then reconstituted in 100 µl DMSO. Dilutions equivalent to 1:5 of the original fraction that were capable of enhancing [³H]cyclosporin A accumulation in C5 cells greater than twofold were pooled. Enriched fractions from 10 individual preparations were rechromatographed, using the same linear solvent gradient as described above. Partially purified material was finally chromatographed on a new C₁₈ HPLC column at 1 ml/min, using an isocratic solvent mixture of 25% isopropanol/25% acetonitrile to elute bound material over a 160-min period (Fig. 2A).

To examine individual differences in urinary NPE excretion levels, 4-liter samples of urine from three healthy, medication-free male donors and a parallel sample of distilled water were processed. Samples were chromatographed on 6 ml, C₁₈ SPE columns containing 1 gm C₁₈ resin (J. T. Baker). Columns were washed with 100 ml 15% isopropanol/15% acetonitrile, then eluted with 9 ml methanol. The methanol eluates were dried and reconstituted in 100 µl DMSO. Extracts were chromatographed on a new reverse-phase C₁₈ HPLC column, and bound material was eluted using the same linear gradient of solvents and conditions as described above.

Structural determination of urinary NPE. UV/Vis spectra of purified urinary NPE were obtained using a model DU-640b spectrophotometer (Beckman Instruments). NMR spectra were obtained on a Unity 500-MHz spectrometer (Varian). To obtain spectra of individual C₁₈ HPLC fractions of purified urinary NPE, samples were dissolved in 200 µl C²HCl₃ each and added to 2.5-mm (OD) microprobe NMR tubes (Wilmad). Mass spectral analyses of partially purified and pure urinary NPE were performed on a model APIII electrospray instrument (Sciex Instruments).

Synthesis of [³H]NPE. The terminal glycol of the ethoxylate chain of Tergitol NP-9 was oxidized to the aldehyde by the acetic anhydride/DMSO method for derivatizing polyethylene glycols (30). The reaction products were subjected to C₁₈ HPLC, as described above, for purification of urinary NPE. A mixture corresponding to ~30% oxidized Tergitol NP-9 (as determined by ¹H-NMR) was obtained. Radioisotopically labeled NPE was obtained by reducing 5 µmol of the aldehyde-containing detergent mixture with 5 mCi of [³H]NaBH₄ (0.2 Ci/mmol). [³H]NPE was purified by chromatographing the reaction mix on a 1-ml C₁₈ SPE column, followed by elution with 1 ml methanol. The resulting [³H]NPE had a specific activity of 21 mCi/mmol.

Plasma protein binding of NPE. Two-milliliter aliquots of human plasma were incubated for 30 min at 22°C with 1 or 10 µM [³H]NPE in the presence or absence of 10 µM cyclosporin A. For comparison, plasma samples were also incubated with 0.1 µM [³H]digoxin or 1 µM [³H]cyclosporin A in the presence or absence of 10 µM Tergitol NP-9. Four 5-µl aliquots of the incubation mix were counted for radioactivity. The samples were centrifuged for 30 min at 5,000 rpm in Centricon 30 (Amicon) concentrators, using a model J2-21 centrifuge (Beckman Instruments). Four 5-µl samples of the filtrate were counted for radioactivity, and the differences between the total and free amounts of radioisotopes were used to estimate the bound fraction of drug or surfactant.

Mdr-reversing properties of NPE. The growth of Aux-B1 and C5 cells in media containing various concentrations of colchicine in the presence or absence of 10 µM Tergitol NP-9 was measured after 5 days. Cell viability was assessed using the tetrazolium dye reduction assay (5).

Drug transport studies. The effects of cyclosporin A, verapamil, Tergitol NP-9, or urinary NPE on the accumulation of [³H]cyclosporin A by C5 cells were assessed using the microtiter plate drug-accumulation assay. The effects of 10 µM cyclosporin A, verapamil, or Tergitol NP-9 on the accumulation of 100 nM [³H]NPE by Aux-B1 and C5 cells were examined by the competitive drug-accumulation assay, described above, except that cells were grown in 24-well plates and removed with 0.5 ml/well 0.1% trypsin.

Transepithelial transport of [³H]cyclosporin A was measured using confluent MDCK cells grown on Anopore semipermeable inorganic membrane supports in 10-mm culture inserts (Nunc) in DMEM supplemented with 10% calf serum, 50 units/ml penicillin, and 50 µg/ml streptomycin. Cells were washed once with HEPES-buffered -MEM, then 0.4 ml of media containing 100 nM [³H]cyclosporin A with or without 10 µM NPE were added to either the basolateral (lower culture dish) or apical (upper insert) surfaces of cells. The same volume of media containing 100 nM unlabeled cyclosporin A, with or without 10 µM NPE present, was added to the opposite surface, and cells were incubated at 37°C. Four 5-µl aliquots of media from the corresponding trans side of cells were counted for radioactivity at various times thereafter.

Transepithelial transport of [³H]NPE was measured as described above, except that MDCK cells were grown in 25-mm culture inserts and the apical or basolateral surfaces of cells were incubated at 37°C with 3.5 ml of media containing 100 nM [³H]NPE. The same volume of media containing 100 nM unlabeled NPE was added to the opposite surface. For time course measurements, 0.4-ml aliquots from the corresponding trans side of cells were counted for radioactivity at various times. For competition studies, cells were incubated in the presence or absence of 10 µM cyclosporin A for 2 h at 37°C. Four 0.5-ml aliquots of media from the corresponding trans side of cells were counted for radioactivity. Experiments were conducted in triplicate, with respective measurements pooled for statistical analysis.

[³H]azidopine photolabeling of P-glycoprotein. C5 plasma membranes and human renal BBM vesicles at 2 mg/ml protein in 100 mM mannitol and 10 mM HEPES/Tris (pH 7.5) were prepared as previously described (6, 8). C5 membranes or BBM, corresponding to 50 µg or 0.5 mg of protein, respectively, were incubated on ice for 5 min with an equal volume of dilution buffer containing 1 µM [³H]azidopine (47 Ci/mmol, final concentration = 0.5 µM) and various concentrations of NPE. Samples were photolyzed as previously described (6) and either used directly for electrophoresis (C5 membranes) or immunoprecipitated (BBM) with a specific antisera to Mdr1p, as previously described (6). Electrophoresis and gel fluorography were conducted as previously described (6). The amounts of 170-kDa P-glycoprotein photolabeled with [³H]azidopine were quantitated by densitometric scanning of fluorographs as previously described (6).

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

Purification of urinary NPE. The presence of P-glycoprotein in the apical membranes of kidney epithelial cells suggests that Mdr1p substrates should be present in urine as a consequence of this transporter's function. To screen for potential Mdr1p substrates in urine, a competitive cellular drug uptake assay, based on the ability of P-glycoprotein to specifically reduce the accumulation of [³H]cyclosporin A in multidrug-resistant CHO-C5 cells (28), was used. Mdr1p substrates act as competitive inhibitors of P-glycoprotein and reverse the multidrug-resistance phenotype of C5 cells. By blocking the efflux of [³H]cyclosporin A from C5 cells, other Mdr1p substrates increase the cellular accumulation of this cyclic peptide. At 10 µM, the Mdr1p substrates, verapamil and cyclosporin A, increased the accumulation of [³H]cyclosporin A by C5 cells about fourfold (Fig. 1B). At equivalent concentrations, these substrates only marginally increased the accumulation of [³H]cyclosporin A by drug-sensitive Aux-B1 cells (Fig. 1A), consistent with their much lower P-glycoprotein expression levels (6).

View larger version (31K): [in this window] [in a new window]   Fig. 1.   Mdr-reversing activities of P-glycoprotein substrates and nonylphenol ethoxylates. [3H]cyclosporin A accumulation was measured in drug-sensitive Aux-B1 cells (A) and in drug-resistant C5 cells (B) in the presence or absence of 10 µM concentrations of cyclosporin A, verapamil, Tergitol NP-9, or purified urinary NPE. C18 HPLC fractions 30-50 of purified urinary NPE (Fig. 2A) were pooled, and the concentration was determined spectophotometrically at 280 nm, using a standardized Tergitol NP-9 solution in methanol. Amounts of [3H]cyclosporin A accumulated by C5 cells treated with the various agents were statistically greater (n = 4, P < 0.005) than untreated C5 cells. Error bars represent ±SE. Tergitol NP-9 and urinary NPE slightly increased the amount of [3H]cyclosporin A accumulated by Aux-B1 cells relative to untreated Aux-B1 cells. However, these [3H]cyclosporin A levels were not statistically different (n = 4, P < 0.05) from those observed when B1 cells were treated with the Mdr1p substrates, cyclosporin A, or verapamil.

Because Mdr1p substrates are quite hydrophobic, they bind to reverse-phase resins like C₁₈. Human urine was therefore fractionated using C₁₈ hydrophobic interaction chromatography. The initial fractionation of urine using C₁₈ SPE columns revealed that the eluate obtained with the 40% isopropanol/40% acetonitrile solvent mixture had the highest mdr-reversing activity. The urinary components responsible for most of the mdr-reversing activity were purified further from the 40% isopropanol/40% acetonitrile fraction by differential solvent extraction and reverse-phase C₁₈ HPLC, as described in EXPERIMENTAL PROCEDURES. Although some mdr-reversing material was also present in the 25% isopropanol/25% acetonitrile elution fraction, the components responsible for this activity were not purified further, due to the high degree of contamination of this fraction with other urinary compounds.

Figure 2A shows a preparative reverse-phase C₁₈ HPLC elution profile obtained with the isocratic solvent mixture, 25% isopropanol/25% acetonitrile of mdr-reversing material, isolated from over 400 liters of urine collected from a single individual over a 1-yr period. The chromatograph shows a broad peak of absorbance (280 nm) and mdr-reversing activity. An amount equivalent to a 2.5-fold dilution of the peak fraction 36 increased the accumulation of [³H]cyclosporin A in C5 cells ~10-fold (Fig. 2A), indicating the presence of a potent inhibitor of P-glycoprotein. Rechromatography of individual fractions produced single, sharper absorbance (280 nm) peaks that retained mdr-reversing activity (Fig. 2B). Gel filtration HPLC of these urinary multidrug reversing components in chloroform on a molecular-sizing, JORDI GPC 500 Å, 5 µm column (Alltech Associates) produced a single elution peak of coincident absorbance (280 nm) and multidrug reversing activity with an average molecular mass of 500 Da (data not shown). These results suggest that the mdr-reversing material isolated from human urine (Fig. 2A) consists of a number of low-molecular-weight compounds with similar but not identical chromatographic properties.

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Purification of human urinary NPE. A: human urinary NPE was purified extensively by reverse-phase C18 HPLC, as described in EXPERIMENTAL PROCEDURES. Chromatograph was obtained by isocratic elution, using the solvent mixture 25% isopropanol/25% acetonitrile. Absorbance profile of fractions (solid line) corresponded with their ability to enhance the uptake of [3H]cyclosporin A into C5 cells (). B: rechromatography of peak fraction 34, using the same isocratic solvent conditions, resulted in a sharper peak of coincident absorbance (280 nm) and mdr-reversing activity. Chromatography of a 100-µg sample of Tergitol NP-9 produced a single peak of absorbance (280 nm) that coeluted with rechromatographed sample of urinary NPE in fraction 34 (arrowhead).

Chemical identification of urinary NPE. ¹H-NMR, UV/Vis spectroscopy, and mass spectrometry identified the urinary mdr-reversing components purified by C₁₈ HPLC as the synthetic surfactant NPE. ¹H-NMR spectra revealed the urinary mdr-reversing components had three distinct proton chemical environments characteristic of NPE. Figure 3 shows a one-dimensional ¹H-NMR spectrum of the peak fraction 36, which had the highest mdr-reversing capability (Fig. 2A). Major resonances residing at 7.2 ppm and 1.6 ppm were assigned to solvent protons from residual CHCl₃ and to hydrogen-bonded H₂O in the deuterated C²HCl₃ solvent, respectively. Only solvent resonances were detected in C₁₈ HPLC fractions that lacked mdr-reversing activity. The UV/Vis spectral properties of the C₁₈ HPLC fractions that had mdr-reversing activity indicated a single aromatic group was present in the urinary compound (absorbance max, 280 nm). The aromatic region of the ¹H-NMR spectrum (Fig. 3), which exhibits doublet resonances at 6.8 and 7.15 ppm, with a coupling of 7.5 Hz, was assigned to protons in a para-substituted phenol. These resonances were rather complex and suggested that some ortho-substituted or disubstituted forms may also be present in minor amounts.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Comparative 1H-NMR spectra of purified human urinary NPE and Tergitol NP-9. 1H-NMR spectrum of C18 HPLC fraction 36 (see Fig. 2A) was compared with that of 0.5 mM Tergitol NP-9; 128 scans were accumulated for human urinary NPE, whereas 32 scans were required to obtain the spectrum of Tergitol NP-9. Note similar features and integral areas associated with resonances in each spectrum. Integrated areas for resonances were normalized to the two ortho-protons at 6.8 ppm in the predominantly para-substituted phenol. Although spectral noise levels prevented more accurate estimates of the number of protons present in alkyl substituents, integrated areas of aliphatic resonances indicated that ~17-22 alkyl protons were present. Top: chemical structure for NPE.

The major ¹H-NMR resonance residing at 3.63 ppm was assigned to a polymer of ethylene oxide (EO). The neighboring ¹H-NMR resonances slightly downfield at 3.8 and 4.1 ppm were assigned to a single EO unit directly bonded to the phenolic oxygen. Correspondingly, this EO unit had -OCH₂- resonances which integrated to two protons each. The remaining resonances between 3.5 and 3.8 ppm accounted for the remaining EO units in the chain, including the terminal unit, which had smaller resonances at 3.71 and 3.59 ppm on either side of the major in-chain resonance at 3.63 ppm. The total integrated area of all of the EO units accounted for ~40 protons, corresponding to an average chain length of 10 EO units in the compound present in fraction 36.

Aliphatic ¹H-NMR resonances between 0.40 and 1.4 ppm indicated several different alkyl substituents were present at ortho- and para-positions of the phenol. The integral areas and complex splitting patterns of these ¹H resonances suggested that the urinary compound we had isolated was a member of the nonyl (C₉H₁₉) series of branched alkylphenol ethoxylates. A ¹H-NMR spectrum (Fig. 3, bottom) of a commercial preparation of this type of detergent, Tergitol NP-9 (average EO = 9), was remarkably similar, thereby establishing the identity of the urinary component as NPE. As expected, a two-dimensional coupled nuclear Overhauser enhanced spectrum of the urinary compound showed numerous short-range correlations among individual protons within the aromatic, polyoxyethylene and aliphatic groups but no definitive long-range connections among protons residing in each of the three separate chemical environments. The upfield region between 15 and 55 ppm of a proton-decoupled ¹³C-NMR spectrum of Tergitol NP-9 contained over 100 lines, confirming the highly branched nature of the nonyl alkyl substituent.

Structural elucidation of urinary NPE by ¹H-NMR was corroborated by mass spectrometry. A mass spectrum (Fig. 4A) obtained of C₁₈ HPLC fraction 32 (Fig. 2A)revealed a prominent ion peak at m/z = 678.0, corresponding to the ammoniated parent ion peak of NPE (EO = 10). Some larger ion species were due to the presence of higher order EO adducts, whereas smaller ones were in part generated by fragmentation of individual EO units (m/z = 44). A mass spectrum of Tergitol NP-9 (Fig. 4B) had similar features to the spectrum obtained for urinary NPE. Although fragmentation ions were also generated by loss of individual EO units from the parent ion (m/z = 639.2), absolute masses differed from that detected in the urinary NPE sample because sodium ions of Tergitol NP-9 were predominantly formed.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   Mass spectral analysis of urinary NPE and Tergitol NP-9. A: mass spectrum of C18 HPLC peak fraction 32 showed a characteristic fragmentation pattern due to loss of individual EO units of m/z = 44.0 from the ammoniated NPE parent ion present at m/z = 678.0 (EO = 10). Larger ion species were also detected since some larger EO adducts of NPE were present in this fraction as well. B: mass spectrum of Tergitol NP-9 also showed the loss of individual EO units of m/z = 44.0 from both the sodiated parent ion present at m/z = 639.2 (EO = 9) and larger EO adducts.

¹H-NMR and mass spectral analyses of individual C₁₈ HPLC fractions that contained mdr-reversing activity (Fig. 2A) confirmed the presence of NPE and showed multiple EO chain-length species of this surfactant were contained in the preparation. The urinary NPEs contained different amounts of EO, with the more hydrophobic, shorter EO chain-length species having longer C₁₈ HPLC retention times. The heterogeneous elution pattern of the urinary mdr-reversing components during C₁₈ HPLC (Fig. 2A) was therefore attributable to the different chromatographic properties of the various EO chain-length species of NPEs present. Rechromatography of fraction 34, which consists of NPEs having average chain lengths of 10 EO units, produced a single peak of absorbance (280 nm) that coeluted with Tergitol NP-9 (arrowhead in Fig. 2B).

To confirm that NPE is present in human urine, a number of control experiments were performed. Crude solid-phase extracts, prepared from 4-liter batches of urine donated by three different individuals, all contained a peak of mdr-reversing activity that coeluted on C₁₈ HPLC with Tergitol NP-9 (data not shown). Mass spectrometry confirmed the presence of NPE in this peak. No mdr-reversing activity was detected in any C₁₈ HPLC fractions when either a DMSO injection vehicle or a control sample, obtained by putting 4 liters of distilled water through the same SPE procedure as the urine samples, were chromatographed. These results show that mdr-reversing activity is present in the urines of different individuals and that the urinary NPE is derived by excretion.

Plasma binding of NPE. Because the molecular weight of NPE is low, its presence in urine could arise as a consequence of glomerular filtration. The extent to which this would occur would be greatly dependent on the amount of free surfactant present in plasma. Binding of [³H]NPE to human plasma proteins was measured by a centrifugation filtration assay, using a 10-kDa molecular weight cutoff. The resulting filtrate simulates the protein-free fluid within Bowman's space (44). At total concentrations of either 1 or 10 µM, similar to [³H]cyclosporin A, >99% of [³H]NPE was bound by plasma proteins (15 and 150 pmol of NPE/mg protein, respectively). In contrast, [³H]digoxin, which is known to be transported by kidney P-glycoprotein in vivo (16), was found to be only 38% bound by plasma proteins. Addition of 10 µM cyclosporin A did not affect the amounts of [³H]NPE bound to plasma proteins. Similarly, addition of 10 µM Tergitol NP-9 did not affect the binding of [³H]cyclosporin A or [³H]digoxin to plasma proteins. These results indicate that, like cyclosporin A and at concentrations below its critical micelle concentration (CMC = 0.1 mM, Ref. 31), very little free NPE would be available in blood for glomerular filtration. Instead, most urinary NPE would be expected to originate as a consequence of transepithelial transport.

Reversal of Mdr by urinary NPE. Ethoxylate-containing surfactants like Triton X-100 have previously been shown to reverse mdr by enhancing the accumulation of a variety of Mdr1p substrates in multidrug-resistant cells (3, 19, 43). To further characterize urinary NPE as a substrate of P-glycoprotein, C₁₈ HPLC fractions 40-50 (Fig. 2A) were pooled, concentrated, and tested for their ability to reverse the mdr phenotype of C5 cells. Similar to cyclosporin A and verapamil, urinary NPE and Tergitol NP-9 marginally increased the accumulation of [³H]cyclosporin A by drug-sensitive Aux-B1 cells (Fig. 1, top), due to some low-level P-glycoprotein expression in this parental cell line (6). At a total concentration of 10 µM, however, urinary NPE increased the amount of [³H]cyclosporin A accumulated by drug-resistant C5 cells by threefold (Fig. 1, bottom). Tergitol NP-9 was a more effective inhibitor of P-glycoprotein, since, like cyclosporin A and verapamil, an equivalent concentration raised [³H]cyclosporin A levels in C5 cells by more than fivefold.

Dose-response curves (Fig. 5) confirmed urinary NPE [half-maximal inhibitory constant (K_0.5) = 7.5 µM] was slightly less effective at preventing [³H]cyclosporin A efflux from C5 cells than Tergitol NP-9 (K_0.5 = 2.5 µM). The urinary NPEs, in the pooled C₁₈ HPLC fractions tested, contained a more heterogeneous assortment of shorter EO chain lengths than Tergitol NP-9. Low-molecular-weight alkylphenol ethoxylates are known to be less effective substrates of P-glycoprotein (43). Nonylphenol (10 µM), which lacks the EO moiety, was tested, and it did not inhibit the efflux of [³H]cyclosporin A from drug-resistant C5 or B30 cells (data not shown). Similar to effect of the mdr-reversing agent, verapamil (K_0.5 = 5 µM), on [³H]cyclosporin A accumulation (Figs. 1 and 5), urinary NPE and Tergitol NP-9 comparably increased the accumulation of several other Mdr1p substrates (e.g., colchicine, vinblastine, dexamethasone) in C5 cells (data not shown). These results demonstrate that NPE is capable of blocking the cellular drug efflux mediated by P-glycoprotein.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Dose-response effects of NPE and verapamil on [3H]cyclosporin A accumulation. Amount of [3H]cyclosporin A (50 nM) accumulated by C5 cells was measured in the presence or absence of different concentrations of urinary NPE (), Tergitol NP-9 (), or verapamil () as described in EXPERIMENTAL PROCEDURES. Half-maximal inhibition constants (K0.5) for urinary NPE, Tergitol NP-9, and verapamil were estimated from [3H]cyclosporin A accumulation curves as previously described (8). Note similar effects of urinary NPE and Tergitol NP-9 on [3H]cyclosporin A accumulation by C5 cells over the same concentration range as mdr-reversing agent, verapamil. Error bars represent ±SE; n = 4.

Reversal of multidrug resistance by NPE. Because NPE inhibits the drug transport function of P-glycoprotein, it should chemosensitize multidrug-resistant cells similar to verapamil. To test this possibility, drug-sensitive B1 and multidrug-resistant C5 cells were grown in various concentrations of colchicine in the presence or absence of NPE. As shown in Fig. 6A, Aux-B1 cells were two orders of magnitude more sensitive to the cytotoxic effects of colchicine than C5 cells (Fig. 6B). Ten micromolar NPE reduced the colchicine resistance of C5 cells to levels observed in Aux-B1 cells. Furthermore, this surfactant completely abolished any resistance to colchicine observed in Aux-B1 cells due to their low-level P-glycoprotein expression (6). NPE also chemosensitized C5 cells to vinblastine and daunomycin (data not shown), thereby confirming its ability to reverse mdr.

View larger version (13K): [in this window] [in a new window]   Fig. 6.   Reversal of mdr by NPE. Effects of 10 µM Tergitol NP-9 on colchicine resistance of drug-sensitive Aux-B1 (A) and multidrug-resistant C5 cells (B) were assessed as described in EXPERIMENTAL PROCEDURES. Note the 100-fold higher concentrations of colchicine required to reduce the viability of C5 cells relative to Aux-B1 cells in the absence of Tergitol NP-9 (). In the presence of 10 µM Tergitol NP-9 (), resistance of C5 cells to colchicine is comparable to that observed in Aux-B1 cells in the absence of this surfactant. Error bars represent ±SE; n = 4.

View larger version (30K): [in this window] [in a new window]   Fig. 7.   Cellular accumulation of [3H]NPE. Amount of [3H]NPE (100 nM) accumulated by drug-sensitive AuxB1 cells (A) and multidrug-resistant C5 cells (B) was measured in the presence or absence of 10 µM of either cyclosporin A, verapamil, or Tergitol NP-9. A: in the absence of a competitive Mdr1p substrate or reversing agent, 2-fold higher levels of [3H]NPE were taken up by Aux-B1 cells relative to the C5 cell line (*** P < 0.005, significantly different from C5 cells). B: cyclosporin A and Tergitol NP-9 increased levels of [3H]NPE accumulated by Aux-B1 cells only slightly. Levels of [3H]NPE accumulated by C5 cells were significantly increased when any of these agents were present (*** P < .005, significantly different from untreated C5 cells). Error bars represent ±SE; n = 6.

NPE transport by multidrug-resistant cells. To directly establish whether NPE is a Mdr1p substrate, a tritiated form of Tergitol NP-9 was synthesized and used to study the transport properties of NPE in multidrug-resistant cells. C5 and Aux-B1 cells were incubated with 100 nM [³H]NPE for 30 min. A twofold-lower accumulation of NPE by multidrug-resistant C5 cells (Fig. 7B) relative to the parental Aux-B1 cell line (Fig. 7A) was observed. If the efflux of [³H]NPE from C5 cells is mediated by P-glycoprotein, this transport process should be blocked by other Mdr1p substrates. Cyclosporin A, verapamil, and Tergitol NP-9 significantly increased the amount of [³H]NPE accumulated by C5 cells (Fig. 7B). Cyclosporin A and Tergitol NP-9 also marginally increased the amount of [³H]NPE accumulated by Aux-B1 cells (Fig. 7A), due to some low-level P-glycoprotein expression in this parental cell line (6). These findings support the conclusion that NPE is transported by P-glycoprotein and that Mdr1p drugs inhibit the efflux of this surfactant from multidrug-resistant cells. Also consistent with this conclusion was the finding that B30 cells, which have higher expression levels of P-glycoprotein than C5 cells, accumulate still lower levels of [³H]NPE (data not shown).

Transepithelial transport properties of NPE. Polarized MDCK cells specifically express P-glycoprotein in their apical membranes (33). With a specific antisera to Mdr1p, the expression of P-glycoprotein in apical membranes isolated from MDCK cells was independently confirmed (data not shown). [³H]cyclosporin A, an Mdr1p substrate, was preferentially transported (basolateral to apical) by epithelia formed by MDCK cells grown on semipermeable membrane supports. The basolateral-to-apical flux of [³H]cyclosporin A was approximately tenfold faster than apical to basolateral flux (Fig. 8). This preferential transport of [³H]cyclosporin A by MDCK cells is similar to that previously reported for other Mdr1p drugs like vinblastine (33). At 10 µM, Tergitol NP-9 reduced the basolateral-to-apical transport of [³H]cyclosporin A by MDCK cells, such that the fluxes became equivalent (Fig. 8). Tergitol NP-9 also inhibited the basal-to-apical transport of other Mdr1p substrates like vinblastine (data not shown). These results indicate NPE can interfere with the net excretion of Mdr1p substrates by inhibiting kidney P-glycoprotein function.

View larger version (18K): [in this window] [in a new window]   Fig. 8.   Effect of NPE on the transepithelial transport of [3H]cyclosporin A. Madin-Darby canine kidney (MDCK) cell monolayers were grown to confluence on semipermeable membrane supports, then incubated either basolaterally or apically with 100 nM [3H]cyclosporin A in the presence or absence of 10 µM Tergitol NP-9 as described in EXPERIMENTAL PROCEDURES. Note preferential flux of [3H]cyclosporin A from basolateral to apical media () and its nearly complete exclusion from basolateral media when added to the apical surface of epithelia (). Exposure of cells to NPE significantly reduced basolateral-to-apical transport () and increased apical-to-basolateral () movement of [3H]cyclosporin A.

To see whether NPE was excreted by kidney epithelia, MDCK cells were incubated either apically or basolaterally with 100 nM [³H]NPE. The time course (Fig. 9A) shows that the net transepithelial transport of NPE to the apical side, in the absence of an Mdr1p blocker, was small and reached a steady-state rate within 30 min. This is in contrast to the net transport of [³H]cyclosporin A, which exhibited a log period and continued for over 3 h (Fig. 8). The small differential net transepithelial transport rate (basal to apical) of NPE by MDCK cells is comparable to that of verapamil (33). The passive permeability of epithelia to surfactants may be significant, due to their intrinsic abilities to flip across lipid bilayers very rapidly (23).

View larger version (16K): [in this window] [in a new window]   Fig. 9.   Transepithelial transport of [3H]NPE. MDCK cells were incubated either basolaterally () or apically () with 100 nM [3H]NPE as described in EXPERIMENTAL PROCEDURES. A: samples of media were recovered from the trans surfaces of cells at various times thereafter and counted for radioactivity. Note preferential accumulation of labeled surfactant in apical media when the basolateral surface of epithelia is exposed to [3H]NPE. B: following a 2-h incubation, significantly more [3H]NPE accumulated in apical media (left, open bar, P < 0.005) than in basolateral media (right, open bar) when the trans sides of untreated epithelia were incubated with labeled surfactant. This net excretion pattern for [3H]NPE (basolateral to apical) was significantly reduced (P < 0.005) in MDCK cells treated with 10 µM cyclosporin A (left, solid bar). [3H]NPE accumulated at higher levels in basolateral media of cyclosporin A-treated epithelia when the labeled surfactant was initially exposed to apical sides of MDCK cells (right, solid bar, P > 0.005). Error bars represent ±SE; n = 12.

To confirm that the transepithelial transport of NPE is mediated by kidney P-glycoprotein, the effect of cyclosporin A on the excretion of this surfactant was tested. In the presence of 10 µM cyclosporin A, basolateral-to-apical transport of [³H]NPE was reduced, whereas apical-to-basolateral flux of [³H]NPE was increased (Fig. 9B). This increased apical-to-basolateral flux of [³H]NPE observed in the presence of cyclosporin A suggests that P-glycoprotein may play a protective role by preventing reabsorption of this surfactant. Unexpectedly, however, this apical-to-basolateral flux of [³H]NPE was even greater than the movement of surfactant in the opposite direction when P-glycoprotein was not blocked by cyclosporin A (Fig. 9B). It is possible that a different transport mechanism is responsible for the apical-to-basolateral movement of NPE seen in cyclosporin A-treated MDCK cells. Alternatively, a net passive movement of NPE from apical to basal surfaces could occur if this surfactant is capable of differentially "flip-flopping" the apical membrane. The unique glycosphingolipid outer lipid leaflet composition of the apical membrane (55) may slow surfactant movement from the inner leaflet, thereby enhancing lateral movement of NPE to the basolateral surface. Despite its unique epithelial permeability properties, the results are consistent with NPE being transported by renal P-glycoprotein, since a specific Mdr1p substrate like cyclosporin A reduces its net excretion from the apical surface. Consequently, P-glycoprotein, localized in the apical membrane of renal epithelia, likely mediates the net transepithelial transport of this surfactant into urine and also reduces its reabsorption.

Inhibition of [³H]azidopine photolabeling of P-glycoprotein by NPE. The drug binding site on kidney P-glycoprotein has previously been photolabeled with the calcium antagonist, [³H]azidopine, and the interactions of various Mdr1p substrates have been studied (6, 8, 37). To determine whether NPE interacts directly with the drug binding site on P-glycoprotein, competitive [³H]azidopine photolabeling experiments on C5 plasma membranes and human renal BBM vesicles were performed. [³H]azidopine interacts with the drug binding site on P-glycoprotein, and Mdr1p substrates block its ability to covalently photolabel the protein (7, 51). Total concentrations of 5-10 µM urinary NPE half-maximally blocked [³H]azidopine photolabeling of P-glycoprotein in both these membrane preparations (Fig. 10). At similar concentrations, Tergitol NP-9 also half-maximally inhibited [³H]azidopine photolabeling of P-glycoprotein (data not shown). These results show NPE interacts directly with the drug binding site on P-glycoprotein in both multidrug-resistant cells and human kidney epithelia. Because the concentrations of NPE that block [³H]azidopine photolabeling of P-glycoprotein are below the CMC for this surfactant, it is unlikely that nonspecific detergent effects on membranes are directly responsible for the results obtained.

View larger version (33K): [in this window] [in a new window]   Fig. 10.   Effect of urinary NPE on [3H]azidopine photolabeling of P-glycoprotein. Shown are [3H]azidopine displacement curves and gel fluorographs (inset) of photolabeled 170-kDa P-glycoprotein in C5 plasma membranes () or immunoprecipitates prepared from human kidney brush-border membranes (BBM) (). Membranes were photolabeled with 0.5 µM [3H]azidopine in the presence or absence of serial dilutions of purified urinary NPE as described in EXPERIMENTAL PROCEDURES. K0.5 values for urinary NPE were calculated as previously described (8). Note decrease in [3H]azidopine photolabeling of P-glycoprotein in both membrane preparations over the same urinary NPE concentration range.

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

Chemical characterization of urinary NPE. In this study, the synthetic surfactant NPE was identified by UV/Vis, ¹H-NMR, and mass spectrometry as a prominent P-glycoprotein substrate in human urine. A complex array of ¹H-NMR resonances was observed within both the aliphatic and aromatic regions due to the heterogeneous forms of NPE present. Because tripropylene is used for the synthesis of nonylalkylphenols, a variety of branched di- and trisubstituents are generated at the para- (70%) and ortho (30%)-positions of the phenol (25). The commercially available surfactant Tergitol NP-9 also contains a range of EO chain lengths with an average content of 9 EO units. Urinary NPE was more heterogeneous with an assortment of shorter EO adduct sizes, ranging from a maximum of 10 to less than 7 EO units. This suggests that there may be greater environmental exposure to shorter EO chain-length forms of NPE or catabolism of larger EO adducts prior to excretion.

Mdr-reversing activity of NPE. The current study shows NPE reverses the mdr phenotype of C5 cells by increasing the accumulation of various P-glycoprotein substrates. The reduced potency of shorter EO chain-length forms of urinary NPE in reversing mdr (Fig. 2A) was also previously found for a series of alkylphenol ethoxylates (43). Additionally, the very hydrophobic alkylphenols like nonylphenol, which lack EO, did not act as P-glycoprotein substrates. Thus the polyoxyethylene moiety is essential for rendering these compounds Mdr1p substrates. Several different types of EO-containing surfactants like NPE are known to chemosensitize multidrug-resistant cells to variety of cytotoxic drugs (3, 4, 10, 19, 43, 49, 63). Because the toxicities of some alkylphenol ethoxylates are low (24, 41), they might therefore be useful as chemosensitizing agents in the treatment of multidrug-resistant tumors.

Alkylphenol ethoxylates as P-glycoprotein substrates. The accumulation of [³H]NPE was reduced in multidrug-resistant C5 cells relative to drug-sensitive B1 cells, and Mdr1p substates like cyclosporin A and verapamil increased the accumulation of [³H]NPE by C5 cells. Similar to NPE, decreased cellular accumulation of the octylphenol ethoxylate Triton X-100 correlates with increased levels of P-glycoprotein expression and multidrug resistance (43). Both urinary NPE and Tergitol NP-9, at doses comparable with other Mdr1p substrates (6), blocked [³H]azidopine photolabeling of P-glycoprotein. Triton X-100 also specifically displaces [³H]azidopine from the drug binding site of P-glycoprotein (43) and at low concentrations stimulates the ATPase activity of partially purified P-glycoprotein (18). These findings confirm that alkylphenol ethoxylates are P-glycoprotein substrates capable of reversing mdr.

The accumulation of surfactants like NPE (Fig. 5) and Triton X-100 (43) is only twofold lower in multidrug-resistant C5 cells than drug-sensitive Aux-B1 cells. By comparison, the accumulation of many other Mdr1p substrates (e.g., cyclosporin A, vinblastine) is much lower in drug-resistant cell lines relative to their parental drug-sensitive counterparts. Similarly, the transepithelial flux of NPE in the basal-to-apical direction in MDCK cells is only marginally greater than its movement in the opposite direction. The high apical-to-basal flux of NPE in MDCK cells treated with cyclosporin A underscores the potentially important role P-glycoprotein has in protecting epithelia from this very permeant surfactant. Alkylphenol ethoxylates likely have membrane permeability properties similar to modulators like verapamil that passively cross membranes almost as fast as P-glycoprotein can remove them (23). Interestingly, detergents also increase membrane fluidity and permeability to a variety of Mdr1p substrates (19). The possibility that NPE reversed mdr indirectly by altering the properties of cell membranes was excluded by experiments that utilized [³H]NPE to directly study the transport properties of this surfactant at nanomolar concentrations, well below the CMC of this detergent. These studies demonstrate unequivocally that NPE is a Mdr1p substrate that is transported out of multidrug-resistant cells and across kidney epithelia by P-glycoprotein.

Sources of NPE absorption. Alkylphenol ethoxylates were developed in the 1950s as a low-cost alternative to environmentally damaging tripolyphosphate detergents. Their annual production exceeds 600,000 metric tons (14). Alkylphenol ethoxylates are formulated for household and industrial use as hard surface cleansers (13) and textile cleansers (48), as well as wetting agents or emulsifiers in pesticide formulations (62). These surfactants are also used as intestinal permeability enhancers in drug delivery systems (58) and in paperboard products that may come into contact with food. Although we have not yet identified the particular sources of NPE ingestion/absorption in our subjects, individual differences in lifestyle or metabolism of this surfactant will likely be contributory factors affecting urinary excretion levels.

Excretion properties of NPE. The excretion of NPE in rats has been studied, and the clearance rate has been found to be highly dependent on the EO chain length (40). NPEs containing larger-size EO adducts were excreted more rapidly and predominantly in feces, whereas shorter EO adducts were recovered primarily in urine. In the present study, shorter EO chain-length forms of NPE were also found in human urine. As appears to be the case with respect to their incomplete biodegradation in the environment (26, 27, 46), branched alkyl and short EO-chain substituents of NPE, like those found in human urine, may be incompletely catabolized. Excretion of orally or dermatologically administered linear alkyl ethoxylates in rats and humans also occurs rapidly (20). Like phenolic detergents, these surfactants are recovered in feces, primarily as a consequence of biliary excretion (20, 47). Nonionic surfactants are therefore taken into the systemic circulation and require removal by renal or hepatic clearance mechanisms. Based on the current findings, we predict that synthetic detergents like NPE are eliminated from the body through the action of P-glycoproteins present in the liver, gastrointestinal tract, and kidney. Although other membrane transporters might also interact with NPE, our findings indicate that this particular surfactant is likely excreted into urine by renal P-glycoprotein.

Effects of human exposure to NPE. There is considerable concern that long-term exposure to compounds like NPE may be harming reproductive systems in various animal species, including humans. As a potent spermicidal agent, NPE (nonoxynol-9) is used in contraceptive foams and condoms. P-glycoprotein, expressed in endothelial cells of small blood vessels in the testis and throughout the female reproductive system (2), is responsible for forming a protective blood-tissue barrier (12, 53, 54). High local concentrations of NPE could potentially interfere with the normal function of endothelial P-glycoprotein. A property that P-glycoprotein shares with the estrogen receptor is its ability to recognize a broad range of structurally unrelated compounds (21). Although NPE is weakly estrogenic (39), alkylphenols that can be generated from NPE (27) are more estrogenic (45, 50, 57). In fact, it is the para-substituted, branched alkylphenols used to generate the NPE that are most estrogenic (50). Alkylphenols, which were not found to be P-glycoprotein substrates, would have to be excreted by a different transporter or they could accumulate in the body.

Conclusions