Chemotoxicity of doxorubicin and surface expression of P-glycoprotein (MDR1) is regulated by the Pseudomonas aeruginosa toxin Cif

Abstract

P-glycoprotein (Pgp), a member of the adenosine triphosphate-binding cassette (ABC) transporter superfamily, is a major drug efflux pump expressed in normal tissues, and is overexpressed in many human cancers. Overexpression of Pgp results in reduced intracellular drug concentration and cytotoxicity of chemotherapeutic drugs and is thought to contribute to multidrug resistance of cancer cells. The involvement of Pgp in clinical drug resistance has led to a search for molecules that block Pgp transporter activity to improve the efficacy and pharmacokinetics of therapeutic agents. We have recently identified and characterized a secreted toxin from Pseudomonas aeruginosa, designated cystic fibrosis transmembrane conductance regulator (CFTR) inhibitory factor (Cif). Cif reduces the apical membrane abundance of CFTR, also an ABC transporter, and inhibits the CFTR-mediated chloride ion secretion by human airway and kidney epithelial cells. We report presently that Cif also inhibits the apical membrane abundance of Pgp in kidney, airway, and intestinal epithelial cells but has no effect on plasma membrane abundance of multidrug resistance protein 1 or 2. Cif increased the drug sensitivity to doxorubicin in kidney cells expressing Pgp by 10-fold and increased the cellular accumulation of daunorubicin by 2-fold. Thus our studies show that Cif increases the sensitivity of Pgp-overexpressing cells to doxorubicin, consistent with the hypothesis that Cif affects Pgp functional expression. These results suggest that Cif may be useful to develop a new class of specific inhibitors of Pgp aimed at increasing the sensitivity of tumors to chemotherapeutic drugs, and at improving the bioavailability of Pgp transport substrates.

p-glycoprotein (Pgp, ABCB1/MDR1) encoded by the human MDR1 gene (accession no. NM000927) belongs to the adenosine triphosphate (ATP)-binding cassette (ABC) transporter family. Pgp, an integral membrane glycoprotein (170 kDa), is the first human ABC transporter cloned and characterized for its ability to confer a multidrug resistance (MDR) phenotype in cells (6, 22, 33). This ATP-dependent efflux pump is involved in cross-resistance to a variety of structurally unrelated cytotoxic agents (37). Substrates that are transported by Pgp include anticancer drugs doxorubicin, vinblastine, etoposide, and Taxol (for a review, see Ref. 5). Efflux of these drugs results in decreased intracellular drug concentration and, subsequently, reduced cytotoxicity. Pgp also transports endogenous substrates, including lipids and steroids, as well as many xenobiotics (for reviews, see Refs. 5, 10, 17).

Pgp is expressed in many human tissues, including capillary endothelial cells in the brain (8). Pgp is also present in the biliary canalicular surface of hepatocytes, on the apical surface of small biliary ductules, on the luminal/apical surface of columnar epithelial cells in the colon and small intestine, and on the apical surface of proximal tubular cells in the kidney (42). The presence of Pgp in pharmacological barriers (i.e., the blood-brain barrier and intestine) suggests a physiological role of Pgp in protecting vital organs by preventing endogenous substrates, xenobiotics, and chemotherapeutic drugs from entering the blood stream (35). Pgp in the intestine, liver, and kidney may also detoxify cells by actively excreting toxins from cells into the adjacent luminal space (35).

High expression levels of Pgp are found in many tumors derived from tissues that normally express this protein. However, Pgp is also highly expressed in tumors derived from tissues that do not normally express Pgp (36), suggesting that expression of the MDR1 gene may be activated during the conversion to malignancy (7). Moreover, increased Pgp protein and MDR1 gene expression in tumors in cancer patients correlates with resistance to chemotherapeutic agents (43). Indeed, the expression of Pgp in cancer cells has often been associated with poor prognosis and failure of chemotherapy (for reviews, see Refs. 4, 17).

Pgp is localized on the plasma membrane of cells, with a small amount detected in the Golgi and very low levels in the endoplasmic reticulum and nucleus in multidrug-resistant cells (for a review, see Ref. 24). Therefore, reducing cell surface Pgp expression is a potential strategy to increase the efficacy of therapeutic agents.

We have recently identified and characterized a 33-kDa protein toxin secreted from Pseudomonas aeruginosa (PA14) (27). This protein rapidly decreases the apical plasma membrane expression of cystic fibrosis transmembrane conductance regulator (CFTR) and inhibits CFTR-mediated chloride ion secretion in polarized human airway epithelial cells and kidney cells expressing wild-type CFTR (WT-CFTR) and the most common mutant variant of CFTR, ΔF508-CFTR (14), which accounts for 80% of cystic fibrosis (CF) alleles in Caucasians. Therefore, the secreted protein was designated CFTR inhibitory factor (Cif) (27). Cif inhibits CFTR expression in the apical membrane by reducing the endocytic recycling of CFTR without having general effects on protein trafficking. For example, neither fluid phase endocytosis nor the localization and expression of gp114, Na⁺-K⁺-ATPase, and the transferrin receptor were affected by Cif (41). CFTR (ABCC7), like Pgp, belongs to the ABC transporter family (34). CFTR and Pgp share a number of features with regard to their intracellular trafficking. Both proteins undergo constitutive endocytosis from the plasma membrane and recycling back to the plasma membrane (13, 23, 32), and endocytosis of CFTR (26) and Pgp (23) is mediated by the Rab5 and clathrin-dependent pathway (15, 23). Given these similarities, we hypothesized that Cif might also reduce Pgp trafficking and apical membrane expression in epithelial cells.

Therefore, the aim of the current study was to examine whether Cif reduces Pgp expression in the plasma membrane, an effect that could be exploited to develop a potential therapeutic strategy in cancer for increasing the sensitivity to chemotherapeutic drugs transported by Pgp. We report that the recombinant Cif protein reduced the apical membrane abundance of Pgp in a time-dependent manner in kidney (Madin-Darby canine kidney, MDCK), intestine (Caco-2), and airway (Calu-3) epithelial cells. In contrast, Cif had no effect on the membrane abundance of other two ABC transporters, multidrug resistance protein 1 (MRP1, ABCC1) and MRP2 (ABCC2). Furthermore, the MDCK cell line overexpressing green fluorescent protein-labeled Pgp (MDCK-GFP-Pgp) was preferentially sensitized to doxorubicin by Cif compared with the parental isogenic MDCK cell line (MDCK-C7 cells) that expresses low levels of Pgp. Moreover, Cif also increased intracellular accumulation of daunorubicin, a naturally fluorescent analog of doxorubicin. These results suggest that Cif could be useful for the development of a novel class of inhibitors of Pgp with improved specificity and affinity aimed at increasing the sensitivity of tumors to chemotherapeutic drugs and increasing the bioavailability of Pgp substrates by promoting their absorption by the gastrointestinal track and their uptake across the blood-brain barrier.

MATERIALS AND METHODS

Purified proteins and chemicals.

Cif-His and mutant Cif(H269A)-His that has no effect on the cell surface expression of CFTR were purified as described previously (27). For clarity of presentation, in the remainder of the article, we will refer to Cif-His as Cif and to Cif(H269A)-His as Cif(H269A). Stocks of purified proteins in buffer (20 mM HEPES buffer, pH 7.5, containing 500 mM NaCl and 10% glycerol) at a concentration of 1 mg/ml were stored at 4°C and used within 3 wk. Doxorubicin, daunorubicin, cis-diammineplatinum(II) dichloride (cisplatin), verapamil hydrochloride, pepstatin A, leupeptin, aprotinin, E-64, and Thesit were all obtained from Sigma-Aldrich (St. Louis, MO). Propidium iodide (PI) was purchased from Invitrogen (Molecular Probes, Eugene, OR). Mouse anti-Pgp monoclonal antibody C219 was purchased from Calbiochem (EMD Biosciences, La Jolla, CA). Mouse anti-MRP1 monoclonal antibody MRPm5 and mouse anti-MRP2 monoclonal antibody M₂ III-6 were purchased from Chemicon International (Temecula, CA). Mouse anti-ezrin monoclonal antibody was obtained from BD Biosciences (San Jose, CA).

Cell lines and cell culture.

MDCK-C7 cells were a gift from Dr. Hans Oberleithner (Wurzburg, Germany). Cells were grown in minimal essential medium (MEM; GIBCO BRL, Gaithersburg, MD) supplemented with 10% fetal bovine serum (FBS; GIBCO BRL), 50 U/ml penicillin (Sigma-Aldrich), 50 μg/ml streptomycin (Sigma-Aldrich), and 2 mM l-glutamine (GIBCO BRL). Isogenic MDCK-C7 cells stably expressing human GFP-Pgp (MDCK-GFP-Pgp; clone 29, passages 6–10) were established and maintained in culture in the presence of 150 μg/ml geneticin (G418; GIBCO BRL) in a 5% CO₂-95% air incubator at 37°C as described previously (28). The parental MDCK-C7 cells (passages 86–91) express endogenous Pgp (19, 28). The total Pgp expression level (Pgp and GFP-Pgp) in MDCK-GFP-Pgp cells is four times higher than that in the MDCK-C7 cells (28). Polarized MDCK-GFP-Pgp and MDCK-C7 monolayers were obtained by seeding 0.2 × 10⁶ cells onto Transwell permeable supports (24-mm diameter; Corning, Corning, NY) coated with Vitrogen plating medium (VPM) as previously described (25). Cells were grown at 37°C for 7 days to establish polarized monolayers.

Calu-3 cells, a human subbronchial gland adenocarcinoma cell line, were obtained from the American Type Culture Collection (ATCC; Manassas, VA) and maintained in MEM containing 50 U/ml penicillin, 50 μg/ml streptomycin, 2 mM l-glutamine, 1 mM sodium pyruvate (GIBCO BRL), and 10% FBS in a 5% CO₂-95% air incubator at 37°C as previously described (41). To establish polarized monolayers, Calu-3 cells (passages 32–42) were seeded onto VPM-coated 24-mm-diameter Transwell permeable supports at a density of 1 × 10⁶. Cells were grown in an air-liquid interface culture at 37°C for 21 days to establish polarized monolayers.

Caco-2 cells, a human colonic adenocarcinoma cell line, were obtained from ATCC (Manassas, VA) and maintained in Dulbecco's modified Eagle's medium (Cellgro; Mediatech, Herdon, VA) supplemented with 50 U/ml penicillin, 50 μg/ml streptomycin, 2 mM l-glutamine, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids (GIBCO BRL), and 10% FBS in a 5% CO₂-95% air incubator at 37°C. Caco-2 cells (passages 26–40) were seeded onto VPM-coated 24-mm-diameter Transwell filters at a density of 0.5 × 10⁶ and grown in culture at 37°C for 21 days to establish polarized monolayers.

Cell surface biotinylation and Western blot analysis.

To determine the abundance of ABC transporters in the plasma membrane, polarized cells grown on 24-mm-diameter Transwell permeable supports were incubated at 37°C with Cif or Cif(H269A) or with equivalent dilutions of buffer in antibiotic-free MEM supplemented with 10% FCS. Cif or Cif(H269A) was added to the apical side of the monolayers. After incubation, cell surface biotinylation to monitor apical expression of Pgp was performed as described in detail previously by our laboratory (30, 40). Briefly, apical membrane proteins were selectively biotinylated using sulfosuccinimidyl-6-(biotinamido) hexanoate (sulfo-NHS-LC-biotin) (EZ-Link; Pierce Biotechnology, Rockford, IL) at 4°C. The cells were then solubilized in lysis buffer (25 mM HEPES, pH 8.0, 1% Triton X-100, and 10% glycerol) containing the Complete protease inhibitor cocktail tablet (Roche Diagnostic, Indianapolis, IN), and the biotinylated proteins were precipitated using streptavidin agarose beads (Pierce, Rockford, IL). The apical membrane expression of Pgp was determined by Western blotting using the mouse monoclonal anti-Pgp antibody C219 (Calbiochem). Cell surface biotinylation of the basolateral MRP1 and the apical MRP2 was also performed in MDCK-GFP-Pgp and MDCK-C7 cells to evaluate the effects of Cif on their plasma membrane expression, followed by Western blotting using mouse monoclonal anti-human MRP1 antibody MRPm5 (Chemicon International) and anti-human MRP2 antibody M₂ III-6 (Chemicon International), respectively. For the examination of Cif on the basolateral expression of MRP1, sulfo-NHS-LC-biotin was added to the basolateral side of the monolayers. Protein band intensity was analyzed as described previously (40, 41) using NIH ImageJ software (version 1.38; Wayne Rasband, National Institutes of Health, Bethesda, MD; http://rsb.info.nih.gov).

Immunoprecipitation of Cif.

Studies were conducted to determine whether Cif enters cells and to determine whether intracellular Cif is degraded during the experiment. To this end, 50 μg of Cif or Cif(H269A) at 100 μg/ml was added to the apical media overlying polarized MDCK-GFP-Pgp cells grown on Transwell permeable filters at 37°C for various lengths of time. Subsequently, cells were washed extensively with ice-cold PBS and lysed in immunoprecipitation buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 10 mM MgCl₂, 1 μM pepstatin A, 10 μM leupeptin, 10 μg/ml aprotinin, 10 μM E-64, and 3% Thesit). Cell lysates were precleared by centrifugation at 14,000 rpm for 30 min. Intracellular Cif was immunoprecipitated by combining precleared lysates with 1.5 μg of a rabbit anti-Cif antibody (27) for 2 h at 4°C under agitation before the addition of immobilized protein A beads (Thermal Scientific, Rockford, IL) to the mixture overnight at 4°C. Protein A beads were washed three times with wash buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 10 mM MgCl₂, and 0.3% Thesit), and proteins were eluted from the beads in Laemmli sample buffer (Bio-Rad Laboratories, Hercules, CA) containing 50 mM DTT at 100°C for 5 min, followed by SDS/PAGE. Cif was detected by Western blotting using a rabbit anti-Cif antibody as described previously (27).

Colony-forming assays.

Colony-forming assays were conducted to examine the ability of Cif to alter the cellular toxicity of doxorubicin. MDCK-GFP-Pgp and MDCK-C7 cells were grown in six-well plates at 200 cells/well for 24 h. Cells were preincubated with 50 μg/ml Cif, Cif(H269A), or buffer (1:20 dilution) for 60 min or with 20 μM verapamil for 90 min, followed by treatment with increasing concentrations of doxorubicin (0.2 to 10 μM) for 60 min, in serum-free medium. Subsequently, drug-containing medium was replaced with normal growth medium, and cells were grown for an additional 48 h. Cells were then washed with PBS once, followed by 100% methanol for 5 min, and air-dried. Cells were visualized by staining with 0.4% (wt/vol) Giesma (Sigma-Aldrich) in 70% methanol for 90 min. Colonies containing >20 cells were counted. Data were obtained from four individual experiments performed in triplicate for each experimental treatment. Triplicate observations were averaged to produce a single experimental value. Data are expressed as a percentage of control (mean ± SE). EC₅₀ values were calculated as the concentrations resulting in 50% inhibition of colony formation using the sigmoidal dose-response equation of GraphPad Prism (version 4.03 for Windows; GraphPad Software, San Diego, CA; www.graphpad.com).

Cytotoxicity assays.

Cytotoxicity assays of cisplatin (a substrate of MRP2) were performed as described previously (28) with minor modification. MDCK-GFP-Pgp cells and isogenic parental MDCK-C7 cells were plated in 96-well plates (Falcon; Becton Dickinson) at a density of 1,000 cells/well. Twenty-four hours after plating, cells were preincubated with 50 μg/ml Cif or Cif(H269A) proteins or vehicle control (buffer, 1:20 dilution) for 60 min or with 20 μM verapamil for 90 min at 37°C in 100 μl of serum-free medium. After the pretreatment, increasing concentrations of cisplatin (0.5 to 30 μM) were added to the medium for 2 h. The medium was then replaced with drug-free complete growth medium, and cells were incubated for 48 h at 37°C. Cell death was assessed using the CellTiter 96 AQ_ueous One solution reagent (Promega, Madison, WI) according to the manufacturer's instructions. Cytotoxicity was assessed by monitoring the absorbance at 490 nm using a Synergy HT Multi-Detection microplate reader (BioTek Instruments, Winooski, VT). Viable cell number is linearly correlated with absorbance at 490 nm. The EC₅₀ value was defined as the drug concentration resulting in 50% cell death. Cytotoxicity of doxorubicin was also assessed in MDCK-GFP-Pgp cells pretreated with 100 μg/ml Cif for 1 and 2 h.

Flow cytometric detection of intracellular daunorubicin.

To assess substrate transport by cells expressing Pgp, MDCK-GFP-Pgp cells and the parental MDCK-C7 cells were seeded in 60-mm culture plates at a density of 0.5 × 10⁶ cells. After 24 h of growth, cells were pretreated with or without 100 μg/ml Cif, Cif(H269A), or buffer (1:10 dilution) for 1 h or 20 μM verapamil for 90 min in serum-free medium. Cells were then incubated with 1 μM daunorubicin, a naturally fluorescent analog of doxorubicin, for 1 h. Cell viability following the treatment with Cif, Cif(H269A), verapamil, or vehicle control was evaluated by staining a fraction of cells with PI. Cells were washed in PBS and trypsinized in 0.25% (wt/vol) Trypsin-0.53 mM EDTA (trypsin/EDTA; GIBCO BRL). Cells were pelleted by centrifugation at 3,000 rpm for 3 min, resuspended in 1 ml of PBS, and analyzed immediately. At least 10,000 cells were analyzed per sample for daunorubicin fluorescence (excitation at 488 nm, emission at 575 nm) in the FACScan analyzer (Becton Dickinson Immunocytometry Systems, San Jose, CA) with CellQuest software (version 3.3; BD Biosciences). Identical detection settings were used for all studies. Data were processed using FlowJo software (Tree Star, Ashland, OR).

Statistics.

Differences between means were compared using unpaired two-tailed Student's t-test with GraphPad Prism (version 4.03 for Windows; GraphPad Software). Data are means ± SE. Statistical significance is ascribed for P < 0.05.

RESULTS

Cif reduces apical membrane expression of Pgp in a time-dependent manner.

Previously, we reported that Cif reduced apical membrane levels of CFTR in human bronchial epithelial cells (CFBE41o⁻) stably transfected with WT-CFTR (WT-CFBE cells) and MDCK cells stably expressing GFP-WT CFTR (WT-MDCK cells) (27). The goal of this study was to determine whether Cif also decreases the apical membrane expression of Pgp in epithelial cells.

Cif reduced the apical membrane expression of Pgp in MDCK-GFP-Pgp cells in a time-dependent manner (Fig. 1, A and B). The apical membrane expression of Pgp was unchanged by Cif after 10 min. Pgp levels were reduced by 28% at 30 min compared with buffer control and by 44% at 90 min after treatment with Cif (Fig. 1B). Interestingly, no reduction was observed at 2 h, suggesting that the effects of Cif are transient and reversible. Buffer alone had no effect on cell surface Pgp expression at all time points tested.

Fig. 1.

CFTR inhibitory factor (Cif) reduces apical membrane expression of P-glycoprotein (Pgp) in Madin-Darby canine kidney (MDCK)-GFP-Pgp cells in a time-dependent manner. MDCK-GFP-Pgp cells were incubated with 100 μg/ml Cif for 10, 30, 60, and 90 min (′) and 2 h. A: representative Western blots demonstrate apical membrane expression of Pgp detected by cell surface biotinylation, as well as total expression levels of Pgp and ezrin in MDCK-GFP-Pgp cells following Cif treatment. B: summary of cell surface biotinylation experiments to detect Pgp expression in the apical membrane. C: Cif also decreased the total expression levels of Pgp in cell lysates in a time-dependent manner. D: summary of Western blot analysis of ezrin levels in whole cell lysates. Pgp data in B and C were normalized against whole cell lysate ezrin and expressed as a percentage of Pgp expression in buffer control. Data were obtained from 5 individual experiments. *P < 0.05; **P < 0.01, significantly different from cells treated with buffer control (Student's t-test).

The effect of Cif on total Pgp protein (i.e., plasma membrane and intracellular stores) was also examined. Total Pgp protein was also reduced by Cif in a time-dependent manner. There was no change at 10 min, but Pgp was reduced by 33% at 90 min (Fig. 1, A and C). At 2 h, Cif reduced total Pgp protein by 56%, an observation consistent with the view that Cif causes a reduction of the cytoplasmic pool of Pgp protein despite the restoration of apical Pgp protein. Because ezrin expression was not altered by Cif or buffer, it was used as a reference protein for analysis in Fig. 1 and in all subsequent studies.

To determine whether the effects of Cif on the plasma membrane expression of Pgp are cell line or tissue specific, we also examined two other cell lines, Caco-2 cells (20) and Calu-3 cells (18), both of which express endogenous Pgp. Similar to MDCK-GFP-Pgp cells, Cif, but not the buffer control, reduced the apical membrane expression of Pgp in a time-dependent manner in both Caco-2 (Fig. 2) and Calu-3 cells (Fig. 3). Reduction of the apical membrane Pgp was observed as early as 10 min after treatment in Caco-2 cells (29%; Fig. 2, A and B) and Calu-3 cells (33%; Fig. 3, A and B). The maximum inhibition occurred at 60 min in both Caco-2 (52%; Fig. 2B) and Calu-3 cells (44%; Fig. 3B). Total Pgp protein also was decreased by Cif as early as 60 min in Caco-2 (24%; Fig. 2C) and 30 min in Calu-3 cells (14%; Fig. 3C) with the maximum effect noted at 90 min following treatment. Taken together, these results suggest that the effect of Cif on apical membrane Pgp expression is not cell type dependent but is more rapid in Caco-2 (intestine) and Calu-3 (airway) cells compared with MDCK (kidney) cells.

Fig. 2.

Cif reduces apical membrane expression of Pgp in Caco-2 cells in a time-dependent manner. Polarized Caco-2 cells were incubated with 100 μg/ml Cif for 10, 30, 60, and 90 min(′) and 2 h. A: representative Western blots demonstrate apical membrane expression of Pgp, as well as total Pgp and ezrin expression in Caco-2 cells. C, Cif; B, buffer. B: summary of experiments examining the effect of Cif on the apical membrane expression of Pgp. C: summary of experiments examining the effects of Cif on Pgp expression in whole cell lysates. D: summary of experiments examining the effects of Cif on ezrin expression in whole cell lysates. Expression of Pgp reported in B and C was normalized against whole cell lysate ezrin and expressed as a percentage of Pgp expression in buffer-treated cells. Data were obtained from 4 individual experiments. *P < 0.05; **P < 0.01, significantly different from cells treated with buffer control (Student's t-test).

Fig. 3.

Cif reduces apical membrane expression of Pgp in Calu-3 cells in a time-dependent manner. Polarized Calu-3 cells were incubated with 100 μg/ml Cif for 10, 30, 60, and 90 min(′) and 2 h. A: representative Western blots showing apical membrane expression of Pgp as well as total Pgp and ezrin expression in Calu-3 cells. B: summary of experiments examining the effect of Cif on the apical membrane expression of Pgp. C: summary of the effects of Cif on total Pgp expression. D: summary of the effects of Cif on ezrin expression in whole cell lysates. Expression of Pgp reported in B and C was normalized against whole cell lysate ezrin and expressed as a percentage of Pgp expression in buffer-treated cells. Data were obtained from 4 individual experiments. *P < 0.05; **P < 0.01, significantly different from cells treated with buffer control (Student's t-test).

Cif reduces apical membrane expression of Pgp in a dose-dependent manner and requires a residue critical for the epoxide hydrolase activity of Cif.

Previous studies have demonstrated that Cif belongs to the α/β-hydrolase superfamily and exhibits epoxide hydrolase activity in vitro (27). A mutant Cif protein, Cif(H269A), containing a single mutation (His269 to Ala mutation) was generated and purified (27). This mutant lacks epoxide hydrolase activity and does not decrease apical membrane expression of CFTR (27).

To examine the ability of Cif(H269A) to reduce the apical membrane expression of Pgp, increasing concentrations of Cif and Cif(H269A) proteins were applied to polarized MDCK-GFP-Pgp cells for 30 min. As shown in Fig. 4, Cif reduced the apical membrane expression of Pgp in a dose-dependent manner. By contrast, Cif(H269A) had no effect on either the apical membrane expression of Pgp or the total Pgp expression at any concentration tested (Fig. 4). Ezrin expression was not altered by either Cif or Cif(H269A) and thus was used, as described above, as a reference protein (Fig. 4).

Fig. 4.

Cif reduces apical membrane expression of Pgp in MDCK-GFP-Pgp cells in a dose-dependent manner. MDCK-GFP-Pgp cells were incubated with increasing concentrations of Cif or the mutant Cif(H269A) (50, 100, 200, and 300 μg/ml) for 30 min. A: representative Western blots illustrate the effects of Cif and Cif(H269A) on apical membrane Pgp, whole cell lysate Pgp, and ezrin. C, Cif; B, buffer; H, Cif(H269A). B: summary of experiments described in A. Expression of Pgp reported in B was normalized against whole cell lysate ezrin and expressed as a percentage of Pgp expression in buffer-treated cells. Data were obtained from 4 individual experiments. *P < 0.05; **P < 0.01, significantly different from cells treated with buffer control (Student's t-test).

Internalization of Cif and Cif(H269A) in MDCK-GFP-Pgp cells.

To investigate the fate of Cif in MDCK-GFP-Pgp cells, cells were treated apically with a total of 50 μg of Cif or Cif(H269A) at 100 μg/ml for various lengths of time ranging from 10 min to 24 h. After incubation, the amount of Cif and Cif(H269A) inside cells was immunoprecipitated and analyzed by performing SDS-PAGE and Western blot analysis. As shown in Fig. 5, Cif (∼33 kDa), which was absent at time 0, was detected intracellularly 10 min after addition of Cif to the apical medium. Compared with the 10-min time point, the amount of Cif inside cells did not change significantly over the next 24 h (Fig. 5A). Approximately 13% of Cif added to the apical medium was internalized (Fig. 5B). Cif(H269A) was also rapidly internalized, and by 10 min, ∼20% of the 50 μg of protein added to the apical culture medium were detected intracellularly (Fig. 5, C and D). Although intracellular levels of Cif(H269A) were not as constant as Cif, at all time points the amount of intracellular Cif was equal to or greater than the amount of intracellular Cif(H269A) (Fig. 5, B and D). Taken together, these observations demonstrate that both Cif and Cif(H269A) entered cells rapidly. The amount of intact Cif inside cells was relatively constant between 10 min and 24 h after addition of Cif to the apical medium. Thus the transient effect of Cif on plasma membrane Pgp abundance was unlikely to be due to the degradation of intracellular Cif (see discussion).

Fig. 5.

Internalization of Cif in MDCK-GFP-Pgp cells. Polarized MDCK-GFP-Pgp cells were incubated with or without 50 μg of Cif or Cif(H269A) at 100 μg/ml added to the apical culture medium for various lengths of time at 37°C as indicated. Intracellular Cif and Cif(H269A) were immunoprecipitated and measured as described in materials and methods. A and C: representative Western blots illustrating intracellular Cif (A) and Cif(H269A) (C). B and D: summary of the amount of intracellular Cif (B) and Cif(H269A) (D). Values were obtained from densitometric analysis of 2 individual experiments and expressed as a percentage of 50 μg of Cif or Cif(H269A) added to the extracellular medium at time 0. *P < 0.05 (Student's t-test).

Cif increases the cytotoxic effects of doxorubicin in MDCK-GFP-Pgp cells.

Pgp is one of the major ABC transporters thought to be involved in the resistance of tumor cells to anticancer agents. To assess the functional consequences of Cif treatment, we measured the cytotoxicity of doxorubicin, a transport substrate of Pgp, in both MDCK-C7 cells and MDCK-GFP-Pgp cells. The cytotoxicity of doxorubicin was assessed by a colony-forming assay 48 h after Cif exposure. Verapamil was used as a positive control. Verapamil is a competitive inhibitor of Pgp and thereby inhibits Pgp-mediated substrate transport (44). In these studies, Cif was added to cells at a concentration of 50 μg/ml (60-min treatment), which reduced apical membrane Pgp to levels similar to those reported above using 100 μg/ml (30-min treatment; see Fig. 4B) (Fig. 6D) before the addition of doxorubicin.

Fig. 6.

Cif enhances the cytotoxic effects of doxorubicin (Dox) on MDCK-GFP-Pgp and MDCK-C7 cells. Cytotoxicity to Dox (0.2–10 μM) was measured using a colony-forming assay as described in materials and methods. Cytotoxicity dose-response curves for doxorubicin demonstrate that Cif increased the cytotoxic effects of Dox in MDCK-GFP-Pgp (A) and MDCK-C7 cells (B). The effect of Cif was similar to that of verapamil at 20 μM in both cell lines (see A and B). Mutant Cif(H269A) or buffer alone had no effect compared with untreated cells. C: in the absence of Dox, neither Cif, Cif(H269A), buffer, nor verapamil had a cytotoxic effect. D: representative Western blots demonstrating a similar reduction of apical membrane and whole cell lysate Pgp in MDCK-GFP-Pgp cells treated with 100 μg/ml Cif for 30 min (lane 1) or 50 μg/ml Cif for 60 min (lane 2). Cells treated with buffer at 1:10 dilution for 30 min (lane 3) or at 1:20 dilution for 60 min (lane 4) were used as controls. Data were obtained from 3 individual experiments performed in triplicate.

Cif increased the ability of doxorubicin to kill MDCK-C7 and MDCK-GFP-Pgp cells. In MDCK-C7 cells, Cif reduced the EC₅₀ of doxorubicin by fivefold compared with the buffer control (from 1.3 to 0.3 μM; Fig. 6B and Table 1). Cif reduced the EC₅₀ of doxorubicin in MDCK-GFP-Pgp cells by approximately 10-fold compared with the buffer control (from 4.1 to 0.4 μM; Fig. 6A and Table 1). Thus Cif reduced the EC₅₀ of doxorubicin in both cell lines to approximately the same value (0.4 μM in MDCK-GFP-Pgp cells and 0.3 μM in MDCK-C7 cells). Cif(H269A) or buffer alone did not alter doxorubicin cytotoxicity in either cell line (Table 1). It is important to note that these two cell lines are isogenic and that MDCK-GFP-Pgp cells express four times more Pgp than MDCK-C7 cells. Doxorubicin was uniformly more cytotoxic to MDCK-C7 cells than to MDCK-GFP-Pgp cells (Table 1), most likely because of lower Pgp expression in the MDCK-C7 cell line. The dose-response curves revealed that cytotoxicity evoked by doxorubicin at ≥5 μM was similar in both MDCK-GFP-Pgp cells (Fig. 6A) and MDCK-C7 cells, regardless of the treatment condition. Verapamil, Cif, Cif(H269A), and buffer had no cytotoxic effect (i.e., in the absence of doxorubicin; Fig. 6C).

Table 1.

Comparison of the cytotoxic effects of doxorubicin in MDCK-GFP-Pgp cells and MDCK-C7 cells

Treatments	EC50, μM
	MDCK-GFP-Pgp	MDCK-C7
Untreated	3.1±0.2a	1.3±0.2
Buffer	4.1±0.6b	1.3±0.02
Cif(H269A)	3.4±0.3b,f	1.3±0.2g
Cif	0.4±0.1d	0.3±0.1d
Verapamil	0.3±0.03b,e	0.2±0.01c

Drug concentrations resulting in 50% cell death (EC₅₀) are means ± SE of data from 3 individual experiments performed in triplicate. Cif, CFTR inhibiting factor; Cif(H269A), Cif with His269 to Ala mutation; MDCK-GFP-Pgp, Madin-Darby canine kidney (MDCK)-C7 cells stably expressing human green fluorescent protein-labeled P-glycoprotein (GFP-Pgp). Statistical analysis was performed using Student's t-test.

^aP < 0.05, untreated MDCK-GFP-Pgp cells vs. MDCK-C7 cells.

^bP < 0.05, MDCK-GFP-Pgp cells vs. MDCK-C7 cells with the same treatment.

^cP < 0.05, verapamil-treated vs. untreated MDCK-C7 cells.

^dP < 0.01, Cif vs. buffer.

^eP < 0.001, verapamil-treated vs. untreated MDCK-GFP-Pgp cells.

^fP < 0.001, Cif(H269A) vs. Cif in MDCK-GFP- Pgp cells.

^gP < 0.05, Cif(H269A) vs. Cif in MDCK-C7 cells.

Parallel experiments were also performed to assess the cytotoxic effects of doxorubicin in MDCK-GFP-Pgp and MDCK-C7 cells using the CellTiter 96 AQ_ueous One solution reagent (Promega) as described in materials and methods. Similar results were obtained for the cytotoxicity of doxorubicin in these two cell lines with the use of the colony-forming assay and the CellTiter assay (data not shown).

In the experiments described above, doxorubicin was added to cells 1 h after the addition of Cif, at a time when the apical expression of Pgp was reduced by ∼40% (Fig. 1). Thus, when apical membrane Pgp was reduced by Cif (1 h), the EC₅₀ for doxorubicin also decreased, as noted above. However, 2 h after the addition of Cif to the apical medium, apical membrane Pgp returned to control levels (Fig. 1). Thus it can be predicted that if doxorubicin were added 2 h after Cif treatment, at a time when apical membrane Pgp is similar to control levels, Cif should not affect the EC₅₀ for doxorubicin. As predicted, the EC₅₀ for doxorubicin was similar to control when cells were treated with doxorubicin at 2 h after Cif treatment (Fig. 7 and Table 2). These results confirm that cell surface Pgp plays a pivotal role in determining drug sensitivity to Pgp transported substrates.

Fig. 7.

Time-dependent effects of Cif on the cytotoxicity of Dox. MDCK-GFP-Pgp cells were pretreated with 100 μg/ml Cif for 1 or 2 h before the addition of increasing concentrations of Dox (0.2–10 μM). Cytotoxicity of Dox was measured using the CellTiter assay as described in materials and methods. Cytotoxicity dose-response curves demonstrate that pretreatment with Cif for 1 h increased the cytotoxic effects of Dox similarly to that of verapamil at 20 μM, whereas pretreatment of cells with Cif for 2 h or with buffer (for 1 or 2 h) had no effect on the cytotoxic effects of doxorubicin compared with untreated cells. Data were obtained from 2 individual experiments performed in triplicate.

Table 2.

Comparison of the cytotoxic effects of doxorubicin in MDCK-GFP-Pgp cells pretreated with 100 μg/ml Cif for 1 and 2 h

Treatments	EC50, μM
Untreated	3.4±0.2
Buffer (1 h)	3.0±0.1
Cif (1 h)	0.3±0.04*
Buffer (2 h)	4.5±0.1
Cif (2 h)	3.0±0.1
Verapamil	0.4±0.06*

EC₅₀ values are means ± SE of data from 2 individual experiments performed in triplicate. Statistical analysis was performed using Student's t-test.

^*P < 0.01, Cif vs. buffer and verapamil-treated vs. untreated MDCK-GFP-Pgp cells.

Cif does not affect MRP1 or MRP2 expression.

In addition to Pgp, MDCK cells also express two other ABC drug transporters, the basolateral MRP1 and the apical MRP2 (12, 16). Thus studies were conducted to determine whether Cif also reduces the cell surface expression of MRP1 and MRP2 (9, 17). In these studies, the concentration of Cif used (100 μg/ml for 60 min) was chosen because it produced a maximum inhibition of Pgp levels. As shown in Fig. 8, Cif had no effect on cell surface expression of MRP1 or MRP2 in MDCK C7 or MDCK-GFP-Pgp cells (Fig. 8, A and B). Moreover, total MRP1 and MRP2 abundance was similar in MDCK C7 and MDCK-GFP-Pgp cells. In addition, total MRP1 and MRP2 abundance also was not altered by Cif compared with the buffer control (Fig. 8, A and C). These results demonstrate that Cif specifically downregulates the plasma membrane expression of Pgp and does not affect the plasma membrane expression of MRP1 or MRP2 in MDCK cells.

Fig. 8.

Cif does not alter the expression of multidrug resistance proteins 1 (MRP1) and 2 (MRP2). MDCK-GFP-Pgp and MDCK-C7 cells were incubated with 100 μg/ml Cif or buffer control for 60 min. This concentration and time were chosen because they elicited maximum effects on Pgp membrane abundance of Pgp. A: representative Western blots show basolateral membrane expression of MRP1 and apical membrane expression of MRP2 detected by cell surface biotinylation, as well as total expression levels of MRP1, MRP2, and ezrin following the treatment of Cif or buffer. B: summary of cell surface expression of MRP1 and MRP2 with or without Cif treatment. C: summary of Western blot analysis of total MRP1 and MRP2 expression. D: summary of Western blots analysis of total expression level of ezrin. Expression of MRP1 and MRP2 reported in B and C was normalized against whole cell lysate ezrin and expressed as a percentage of MRP1 or MRP2 expression in buffer-treated cells. Data were obtained from 3 individual experiments.

Consistent with these biochemical studies revealing that MDCK-C7 and MDCK-GFP-Pgp cells expresses similar levels of MRP2 and that Cif did not change the membrane abundance of MRP2, Cif had no effect on the cytotoxicity of cisplatin, a substrate of MRP2, in either of these two cell lines (Fig. 9, A and B, and Table 3). Neither Cif(H269A) nor the buffer altered the EC₅₀ values of cisplatin (Table 3). Moreover, the EC₅₀ for cisplatin was similar in MDCK-C7 and MDCK-GFP-Pgp cells. These observations are consistent with the fact that these two cell lines have equal amounts of MRP2 in the apical plasma membrane (Fig. 8A) and the observation that Pgp does not transport cisplatin (16, 28). Because there is no substrate that is specific for MRP1, and because Cif did not change plasma membrane MRP1 abundance, there was no reason to conduct cytotoxicity studies to assess possible differences in MRP1 function in MDCK-C7 and MDCK-GFP-Pgp cells. Taken together, the present data suggest that Cif specifically downregulates Pgp expression but not MRP1 or MRP2 expression. Moreover, the data also suggest that the enhanced sensitivity of MDCK-GFP-Pgp cells to doxorubicin upon the treatment with Cif is due, at least in part, to the specific reduction of apical membrane Pgp expression and did not involve changes in MRP1 and MRP2 expression or in the subcellular distribution of MRP1 and MRP2.

Fig. 9.

Effects of Cif on the cytotoxicity of cisplatin. The cytotoxic effects of cisplatin on MDCK-GFP-Pgp (A) and MDCK-C7 cells (B) were determined by incubating cells with cisplatin (0.5–30 μM) in the absence (untreated) or presence of Cif, Cif(H269A), buffer, or verapamil. Data were obtained from 4 individual experiments performed in triplicate.

Table 3.

Comparison of the cytotoxic effects of cisplatin in MDCK-GFP-Pgp and MDCK-C7 cells

Treatments	EC50, μM
	MDCK-GFP-Pgp	MDCK-C7
Untreated	10.1±2.5	12.4±4.9
Buffer	14.9±3.4	16.7±4.0
Cif	22.8±6.1	20.9±2.6
Cif(H269A)	15.7±1.5	15.1±0.1
Verapamil	10.1±0.9	8.7±0.9

EC₅₀ values are means ± SE of data from 4 individual experiments performed in triplicate. Statistical analysis was performed using Student's t-test.

Cif increases the cellular accumulation of daunorubicin.

The effect of Cif on the expression of functional Pgp also was assessed by measuring the accumulation of a naturally fluorescent derivative of doxorubicin, daunorubicin, using flow cytometry. Inhibition of Pgp by verapamil served as a positive control. Cif increased the accumulation of daunorubicin by 1.6- and 1.9-fold compared with the buffer control in MDCK-C7 and MDCK-GFP-Pgp cells, respectively (Fig. 10A). Neither Cif(H269A) nor buffer alone had an effect on daunorubicin accumulation (Fig. 10).

Fig. 10.

Cif increases the intracellular accumulation of daunorubicin in MDCK-GFP-Pgp and MDCK-C7 cells. Cells were pretreated with 100 μg/ml Cif or Cif(H269A) or buffer (1:10 dilution) for 60 min or 20 μM verapamil for 90 min in serum-free medium before incubation with 1 μM daunorubicin for 1 h at 37°C. A: mean intracellular red fluorescence intensity of daunorubicin was measured by flow cytometry in untreated cells or cells treated with Cif, Cif(H269A), buffer or verapamil. *P < 0.05; **P < 0.01 (Student's t-test; n = 4). #P < 0.05, intracellular accumulation of daunorubicin in treated or untreated MDCK-GFP-Pgp cells was significantly less than that in comparably treated or untreated MDCK-C7 cells (Student's t-test). B and C: representative histograms chosen from 5 independent experiments showing the accumulation of daunorubicin in MDCK-C7 cells (B) and MDCK-GFP-Pgp cells (C) following various treatments.

As expected, more daunorubicin accumulated in the untreated parental MDCK-C7 cells compared with untreated MDCK-GFP-Pgp cells due to lower Pgp expression in the MDCK-C7 cell line (Fig. 10A). Consistent with a previous report (28), accumulation of daunorubicin in MDCK-C7 cells was two times more than that in MDCK-GFP-Pgp cells (Fig. 10A). As expected, verapamil increased daunorubicin accumulation in both MDCK-C7 cells (2.7-fold; Fig. 10, A and B) and MDCK-GFP-Pgp cells (3.4-fold; Fig. 10, A and C) compared with untreated cells.

DISCUSSION

The major new finding in this report is that Cif, a toxin secreted by Pseudomonas aeruginosa, selectively and reversibly inhibits the apical membrane expression of Pgp in polarized airway, intestinal, and renal epithelial cells. Moreover, Cif increases the cytotoxicity to doxorubicin in MDCK-GFP-Pgp cells and increases cellular daunorubicin accumulation. These results suggest that Cif could be useful to develop a novel class of specific inhibitors of Pgp aimed at increasing the sensitivity of tumors to chemotherapeutic drugs and at increasing the bioavailability of Pgp transport substrates across barrier epithelial cells, such as the intestine and blood-brain barrier, by reducing the membrane abundance of Pgp. Cif was as effective as verapamil in increasing the cytotoxicity of doxorubicin. However, successful development of a Cif analog requires an increase in efficacy and specificity.

Cytotoxicity assays demonstrated that Cif increases doxorubicin cytotoxicity in Pgp-overexpressing cells (i.e., MDCK-GFP-Pgp cells) compared with that observed in MDCK cells expressing lower, endogenous levels of Pgp (i.e., MDCK-C7 cells) (Table 1). It is important to note that MDCK-C7 cells and MDCK-GFP-Pgp cells are isogenic, except that the latter express four times more Pgp (28). The isogenic nature of these two cell lines were confirmed by cell surface biotinylation studies showing that MDCK-GFP-Pgp and MDCK-C7 cells have similar levels of cell surface MRP1 and MRP2 (Fig. 8), an observation consistent with the view that overexpression of Pgp does not alter the expression of other MRPs. Therefore, the Cif-induced increased doxorubicin cytotoxicity in MDCK-GFP-Pgp and MDCK-C7 cells is most likely due to the reduced apical expression of Pgp caused by Cif and not to changes in other transporters. The more pronounced effects of Cif in MDCK-GFP-Pgp cells compared with MDCK-C7 cells suggest that Cif preferentially sensitizes Pgp-overexpressing cells. Cif reduced the EC₅₀ of doxorubicin. Thus it can be predicted that a Cif analog with high specificity for Pgp coadministered with anti-cancer drugs would allow the use of lower doses of drugs and thereby cause less cytotoxicity to nontumor cells. It also can be predicted that Cif would enhance the cellular accumulation of other Pgp substrates into cells and enhance the accumulation of excluded substrates, including antibiotics (1, 21, 39) and anti-inflammatory agents (2, 45), across barrier epithelia and the blood-brain barrier. Proof-of-principle for this effect is the observation that coadministration of Pgp inhibitors enhances the absorption and bioavailability of orally administered tobramycin (1) and amikacin (21), antibiotics that are poorly absorbed by the gastrointestinal tract in mouse models. Thus an analog of Cif that lacks the antigenicity of a bacterially derived protein would enhance anti-cancer therapy and could improve the bioavailability of anti-bacteria agents, as well as other drugs normally excluded by Pgp from cells, barrier epithelia, and the blood-brain barrier.

It is interesting to note that Cif reduced the amount of Pgp in the plasma membrane by ∼50% and increased the accumulation of daunorubicin in cells by twofold. Thus, as expected, there is a good inverse relationship between the membrane abundance of Pgp and daunorubicin accumulation in MDCK cells. However, the Cif-induced twofold increase in daunorubicin accumulation inside cells resulted in an ∼10-fold reduction in the EC₅₀ for doxorubicin. This observation is consistent with the view that the relationship between intracellular doxorubicin concentration and cytotoxicity (i.e., the dose response) is not linear, like many other therapeutic compounds (11). It also should be considered that the reduction in intracellular Pgp by Cif may also contribute to decreased resistance to doxorubicin, since cytoplasmic Pgp may contribute to drug resistance by sequestering drugs in cytoplasmic vesicles, preventing them from reaching cellular targets (29, 38).

The effect of Cif on reducing the apical membrane abundance of Pgp was transient and reversible. The apical Pgp abundance was maximally reduced 60 min after treatment with Cif but was restored to control levels by 2 h, possibly because Cif was degraded by epithelial cells. However, studies presented in Fig. 5 demonstrated that the intracellular levels of Cif were similar 1 and 2 h after Cif was applied to the apical culture medium. These results suggest that intracellular Cif is either inactivated or sequestered inside cells or that the cell compensates by an unknown mechanism for the inhibitory effects of Cif on Pgp expression. The observation that the apical membrane Pgp returned to control levels by 2 h after Cif treatment despite a significant reduction in total Pgp expression suggests that there was a redistribution of Pgp from an intracellular compartment to the cell surface (Fig. 1). This is likely a compensatory mechanism of the cell to maintain a level of Pgp in the membrane sufficient to protect the cell against toxins.

In preliminary studies, we observed that Cif enters cells via lipid rafts and that inhibition of Cif uptake via lipid rafts completely eliminates the effect of Cif on Pgp (Bomberger JM, personal communication). Together with studies in Fig. 5 demonstrating that Cif enters cells, these observations suggest that Cif may inhibit Pgp trafficking by modulating an intracellular target.

Treatment with doxorubicin during the period of reduced Pgp membrane abundance was accompanied by increased intracellular accumulation of doxorubicin, which resulted in greater cytotoxicity to cells. It is well known that the cytotoxicity of doxorubicin is due to its ability to irreversibly intercalate into DNA and thereby inhibits DNA replication and repair, RNA synthesis, and protein synthesis, leading to apoptosis (46). Therefore, even a brief increase in the intracellular accumulation of doxorubicin is irreversibly cytotoxic in sensitive cells. By contrast, as anticipated, the EC₅₀ for doxorubicin was unchanged with 2-h Cif pretreatment at a time when cell surface Pgp levels were restored.

In this and in previous studies, we have demonstrated that Cif selectively reduces the apical membrane levels of Pgp and CFTR (27) but does not affect the apical membrane expression of other ABC transporters, including MRP1 and MRP2. Furthermore, Cif does not affect apical membrane fluid phase endocytosis or the endocytosis of gp135 or the transferring receptor (41). Cif reduces the apical expression of CFTR by inhibiting the endocytic recycling of CFTR (41) and redirects CFTR from recycling endosomes to the lysosome for degradation (3). Cif also reduces the cell surface expression of Pgp and decreases the cellular levels of Pgp, most likely by enhancing the lysosomal degradation of Pgp. Both CFTR (26, 32) and Pgp (23) are constitutively internalized in clathrin-coated vesicles and efficiently recycle back to the plasma membrane from the endosomal compartment (23, 31). Because Pgp and CFTR trafficking in epithelial cells is similar, it is tempting to speculate that Cif also modulates the apical membrane expression of Pgp by reducing the endocytic recycling of Pgp back to the plasma membrane and redirecting Pgp to the lysosome for degradation. Additional studies, beyond the scope of this report, are required to elucidate how Cif decreases the plasma membrane expression of Pgp and to identify the intracellular target of Cif. The lack of effect of Cif on MRP1 and MRP2 cell surface expression suggests that the intracellular trafficking of these ABC transporters may be significantly different from that of CFTR and Pgp.

In conclusion, we have demonstrated that Cif, a secreted toxin from Pseudomonas aeruginosa, selectively reduces the apical plasma membrane expression of Pgp in a variety of epithelial cells, increases the cytotoxic effect of doxorubicin, and enhances the intracellular accumulation of its fluorescent analog, daunorubicin. Hence, Cif, Cif analogs, or the molecular target(s) of Cif may be candidates for developing a new class of specific Pgp inhibitors to enhance the ability of chemotherapeutic agents to kill tumor cells and to improve the bioavailability of other Pgp transport substrates, including antibiotics and steroids.

GRANTS

This study was supported by National Institutes of Health (NIH) Grants R01-HL0741745 (to B. A. Stanton) and R01-DK45881 (to B. A. Stanton) and a Research Development Program grant from the Cystic Fibrosis Foundation (to B. A. Stanton), the Rosalind Borrison Memorial Pre-doctoral Fellowship (to D. P. MacEachran), and NIH Training Grant Predoctoral Fellowship T32-DF007301 (to D. P. MacEachran).