P-glycoprotein (P-gp) is a plasma membrane glycoprotein that has been signaled as a primary cause of multidrug resistance (MDR) in tumors. We performed a yeast 2-hybrid screen using the C-terminal domain of P-gp and identified 2 small GTPases involved in vesicular trafficking, Rab4 and Rab14, which complex with P-gp. The overexpression of GFP-Rab4, either transiently or stably, but not of Rab14, in K562ADR cells decreased the presence of P-gp in the cell surface. As a result, expression of this GTPase reduced the MDR phenotype of K562ADR cells, by augmenting the intracellular accumulation of daunomycin (DNM). This effect was mimicked by the constitutively active Rab4Q72L mutant, but not by the dominant negative Rab4S27N mutant. Rab4 regulated excocytotic P-gp trafficking to the plasma membrane from intracellular compartments, and this modulation required the interaction of both proteins and the GTPase activity. Noteworthy, K562ADR cells exhibited a significant reduction of Rab4 levels, but not of other Rab GTPases, as compared with the sensitive parental cell line, suggesting that the development of the MDR phenotype in these cells involves upregulation of P-gp and a concomitant downregulation of proteins that regulate its surface expression. Attenuation of endogenous Rab4 levels in K562ADR by RNA interference enhanced the expression of P-gp in the cell surface, and reduced the uptake of DNM. Accordingly, these findings substantiate the notion that modulation of the temporal and spatial distribution of P-gp in cancer cells may be a valid therapeutic strategy to alleviate the MDR phenotype, and signal to Rab4 as a potential target.
P-glycoprotein (P-gp or ABCB1) is a glycosylated integral membrane protein that belongs to the ATP binding cassette (ABC) transporter family encoded in humans by the MDR1 gene.1–3 P-gp is highly expressed in epithelial and endothelial cells such as those in the gastrointestinal track and the blood brain barrier.4–6 The physiological role of P-gp is the extrusion of a broad spectrum of structurally unrelated xenobiotics across cell membranes by acting as an energy-dependent drug-efflux pump.1, 7 In addition, this glycoprotein is overexpressed in tumor cell membranes critically contributing to the manifestation of the multidrug resistance (MDR) phenotype in human cancers.
The expression of P-gp in tumors could be intrinsic or induced by antichemotherapy drugs. In vivo, P-gp expression is associated with poor overall prognosis and response of tumors to current chemotherapy, in part, because of the protein capacity to extrude a broad range of compounds, including anthracyclines, vinca-alkaloids and taxanes.8 Chemotherapeutically induced expression of P-gp has been well documented in tumors such as acute leukemia and small-lung cancer, breast and ovarian cancer, head and neck tumors, Kaposi sarcoma, and child neuroblastoma.9 The central role of P-gp in clinical oncology has prompted the discovery and development of antagonists.10–16 These agents are known as chemosensitizers or revertants of the MDR phenotype. By antagonizing the active efflux of anticancer drugs, chemosensitizers promote an accumulation of these cytotoxic agents into tumor cells, thus augmenting the efficacy of the chemotherapeutic treatment.9, 17
Despite the important role played by P-gp in cellular physiology and in the manifestation of the MDR phenotype, limited information is available on cytosolic interacting partners that may modulate aspects of protein expression and function. Previous studies using the linker region of P-gp as a bait have shown the RING finger protein 2 (RNF2), an E3 ubiquitin ligase, as an interacting protein that regulates the cellular abundance of P-gp.18 A complementary approach using overlapping peptides of this linker region identified α- and β-tubulin as proteins that directly bind to P-gp.19 In addition, the interaction of P-gp with caveolin-1 has been documented.20, 21 Caveolin-1 appears to inhibit P-gp functionality and, notably, MDR tumors appear to downregulate caveolin-1 levels.21, 22 Furthermore, a recent yeast 2-hybrid screen using the N-terminal domain of P-gp demonstrated its interaction with a new isoform of an endoplasmic reticulum (ER)-bound Bap29 protein.23 Complexes of P-gp and Bap29 were primarily retained in the ER, suggesting that Bap29 influences the processing and, perhaps, the trafficking of P-gp to the plasma membrane. In this regard, it has also been described that manipulation of Rab5 levels and actin integrity modulate the endocytosis and recycling of P-gp.24–26 These studies have proposed the tenet that modulation of P-gp levels at the plasma membrane by altering its endocytosis-recycling may be a potential target for therapeutic intervention.
Here, we have further pursued the identification of intracellular regulatory proteins of P-gp function. For this task, we used the yeast 2-hybrid system to identify proteins that associate with the C-terminal of P-gp (Ct-P-gp) (residues 984M-1276S) containing the ATP binding site. The rationale for using this domain considered that, as for the cystic fibrosis transmembrane conductance regulator (CFTR),27 an ABC transporter (ABCC7) homolog of P-gp, it could be involved in regulating aspects of the glycoprotein cellular trafficking. We report the identification of 2 GTPases, Rab4 and Rab14 involved in membrane protein trafficking,28–32 as interacting proteins of Ct-P-gp. We show that Rab4, but not Rab14, colocalizes with P-gp in cytoplasmatic compartments of K562ADR cells, and that this interaction reduces the surface expression of P-gp and the extent of the MDR phenotype. Modulation of surface expression of P-gp by Rab4 required both the interaction of both proteins and the GTPase activity of Rab4. Our findings lend support to the notion that Rab4-mediated regulation of P-gp trafficking may be a therapeutic strategy to attenuate the MDR phenotype.
ADR: doxorubicin; Ct-P-gp: C-terminal domain of P-gp; DNM: daunomycin; EE: early endosomes; MDR: multidrug resistance; P-gp: P-glycoprotein; Tf-TRITC: transferrin-TRITC
Material and Methods
Yeast 2-hybrid screening
The pGBKT7 and pACT2 vectors, yeast strains PJ69-2A and Y187 and the procedures of yeast 2-hybrid analysis were derived from MATCHMAKER GAL4-based 2-hybrid system (Clontech, Mountain View, CA), and the screening was carried out essentially as described previously.33 Briefly, the rat brain cDNA library (Clontech), subcloned in pACT2 vector, was transformed into Y187 to generate the prey. The cDNA encoding Ct-P-gp (amino acids 984–1,276) was amplified by PCR from the mouse mdr1-b, subcloned into the pGBKT7 vector and then transformed into the yeast PJ69-2A to generate the bait strain. The diploid yeast colonies resulting from the mating of both strains were grown on quadruple dropout (trp-, leu-, ade- and lys-) SD minimal agar medium and β-galactosidase activity on X-α-Gal (5-bromo-4-chloro-3-inoyl-α-D-galactopyranoside). The pACT2 library plasmids from the positive clones were isolated and sequenced. Encoded proteins were identified from the NCBI sequence database using the BLAST network service.
Plasmids and constructs
cDNAs of human Rab4 and Rab14 were obtained by PCR with specific primers from a human brain library (Clontech) and cloned into the pEGFP-C1 plasmid (Clontech). Rab4S27N and Rab4Q72L mutants and Rab14S25N and Rab14Q70L mutants were generated by PCR mutagenesis using wild-type GFP-Rab4 and GFP-Rab14 as templates, respectively. Ct-P-gp (amino acids 984–1,276) was cloned into the pGEX-4T1 plasmid (Amersham Biosciences, Arlington Heights, IL) as a GST fusion protein.
GST pull down assay
Bacterially expressed GST-Ct-P-gp was affinity purified on a glutathione Sepharose 4B column (Amersham Biosciences) following a previously described protocol.34 Immobilized fusion protein (7.5 μg) was incubated with HEK293 whole-cell extracts overexpressing, GFP, GFP-Rab4 and GFP-Rab14 (250 μg total protein) at 4°C for 2 hr. After 3 washes with lysis buffer (50 mM Tris·HCl, pH 7.4, 150 mM NaCl, 0.25% sodium deoxycholate, 1% Nonidet P-40, 1 mM EDTA, 1 mM phenylmethylsulfonylfluoride, 1 mM aprotinin, 1 mM leupeptin and 1 mM pepstatin), GST-bound complexes were eluted from the resin with 25 mM glutathione and denaturated with SDS-PAGE sample buffer at 90°C for 8 min. Protein complexes were resolved and analyzed by SDS-PAGE electrophoresis.
Cell lines, culture conditions and transfections
HEK293 cells were cultured in DMEM medium. The drug-sensitive human erythroleukemia K562 cells and the resistant K562ADR subline were grown in RPMI 1640 supplemented with 2 mM L-glutamine, 10% (v/v) fetal bovine serum, and antibiotics (penicillin, 100 units/ml and streptomycin, 100 μg/ml). Cells were maintained at 37°C in a humidified incubator with 5% CO2. HEK293 cells (1.2 × 106 cells) were transfected with pEGFP, pEGFP-Rab4 and pEGFP-Rab14 using the Lipofectamine™ 2000 reagent (Invitrogen, Carlsbad, CA) following the manufacturer's instructions. To express recombinant proteins and to transfect the siRNA against Rab4A, K562ADR cells were electrotransfected.35 Briefly, K562ADR cells (2 × 107 cells) were suspended in RPMI 1640 medium (500 μl) and placed in the electroporation chamber (0.4 cm) (Bio-Rad) with 80 μg/ml cDNA encoding GFP, GFP-Rab4 or GFP-Rab14, or 500 nM siRNA (AMBION, Austin, TX), and subjected at a 270 V and 1,050 μF single pulse in a Bio Rad Gene Pulser. After pulse application, cells were transferred immediately into an eppendorf tube and incubated at 37°C for 30 min to facilitate the resealing of the electropores. After postpulse incubation, each sample was diluted in culture medium (15 ml). Cells were analyzed 48 hr post-transfection.
To obtain the cellular clones K562ADR/GFP and K562ADR/GFP-Rab4 stably expressing the proteins, transfected cells were selected with 1 mg/ml G418, and cellular clones were sorted by flow cytometry coupled to a cell sorter (Epics Altra, Beckman-Coulter, Boca, CA). The stable cell lines were continuously grown in culture medium supplemented with 0.4 mg/ml G418.
Intact cell chemical cross-linking immunoprecipitation
Cross-linking reaction was carried out essentially as described.36 Briefly, K562ADR cells (∼8 × 106 cells) were washed twice in PBS and incubated with 1 mM dithiobissuccinimidyl propionate (DTSP, Pierce, Rockford, IL) at 22°C for 30 min following the manufacturer's instructions. The cross-linking reaction was stopped with 20 mM Tris-HCl pH 7.4. After incubating the reaction mixture for an additional 15 min, cells were washed twice with ice-cold PBS, harvested and homogenized in lysis buffer for 15 min at room temperature, followed by incubation at 0°C for 15 min, and one freeze-thaw cycle. Thereafter, cells extracts were centrifuged at 14,000 rpm for 30 min at 4°C. Cross-linked complexes were immunopurified with anti-GFP (1:100) at 4° C for 12 hr (Clontech). Subsequently, 30 μl of protein G agarose beads (Pierce) were added and incubated for 2 hr at 4°C. Immunoprecipitates were denatured with SDS-PAGE sample buffer (90°C for 8 min), separated by SDS-PAGE, and analyzed by Western immunoblotting using an anti-GFP (Santa Cruz Biotechnology, Santa Cruz, CA) and an anti-P-gp antibody (C-219, Calbiochem, Darmstadt, Germany) (1:1,000).
Real time RT-PCR
Total RNA from K562 and K562ADR cells was isolated using the RNeasy Mini Kit (Qiagen GmbH, Hilden, Germany). To eliminate potential DNA contamination, total RNA was treated with RQ1 DNase (Promega Corp., Madison, WI) at 37°C for 30 min, followed by 2 min at 94°C. Reverse transcription of 1 μg RNA was performed by the TaqMan Reverse Transcription Reagents kit (Applied Biosystems, Foster City, CA) according to the manufacturer's instructions. Real-time quantitative PCR was done to amplify 20 ng of cDNA using the ABI Prism 7700 Sequence Detector System (Applied Biosystems). The MGB TaqMan probes were MDR1 (Hs00184491_m1); RAB4A (Hs00190157_m1); RAB14 (Hs00249440_m1); RAB5A (Hs00991290_m1); RAB7 (Hs00245005_m1); RAB8A (Hs00180479_m1); RAB11A (Hs00366449_m1) and for GAPDH (432631E) all of them supplied by Applied Biosystems. GAPDH was used as an endogenous reference in singleplex PCR. The mRNA relative gene expression was calculated by the comparative Ct method referred to the GAPDH housekeeping gene expression (ABI Prism 7700 Sequence Detection System: User Bulletin # 2, Applied Biosystems).
After 24 hr of plasmid electrotransfection, K562ADR cells were plated onto poly-D-lysine-coated coverslips (2.5 × 105 cells/ml) for immunofluorescence, and incubated for 24 hr. For the internalization of TRITC-conjugated transferrin (Tf-TRITC, SIGMA, St. Louis, MO), cells were incubated with 50 μg/ml of the dye in RPMI-1640 medium at 37°C for 30 min. Cells were washed with RPMI-1640 medium and processed for immunocytochemistry. After extensive washing with PBS, cells were fixed in 4% paraformaldehyde and 4% saponin for 20 min at 4°C, and incubated with 1:50 dilution of anti-P-gp antibody (4E3-16, Calbiochem) or with 1:100 dilution of anti-EEA1 (Abcam, Cambridge, MA) for 24 hr at 4°C. Cells were then washed and incubated with TRITC anti-mouse IgG (Sigma) for 1 hr at room temperature. Cells were washed again, embedded and analyzed by confocal microscopy (LSM5 Pascal; Zeiss, Gottingen, Germany).
Drug accumulation assays
Daunomycin (DNM) accumulation measurements in tumor cells were carried out as described.16 Briefly, cells were counted, collected, washed with PBS and resuspended at 1 × 106 cells/ml. Cellular suspensions were incubated with 3 μM DNM (Sigma) at 37°C for 1 hr. Cells were analyzed in a FACS flow cytometer (Epics XL, Beckman Coulter) and data were collected in the list mode. Mean fluorescence of DNM was evaluated for each condition (control and DNM samples) in 10,000 viable cells. DNM uptake was expressed as the ratio of arithmetic mean fluorescence intensity of DNM/arithmetic mean fluorescence intensity of control condition. Values were normalized with respect to those obtained for each condition with cells expressing GFP. Data are presented as mean ± sd, with n = 4.
P-gp detection by flow cytometry
Membrane and total P-gp detection was performed as described.16 After 1 wash with PBS, K562ADR cells were fixed in 4% paraformaldehyde and 4% saponin for 20 min at 4°C. Following 3 washes with 1% BSA in PBS, cellular pellets were suspended in 1% BSA in PBS at 5 × 106 cells/ml. To detect total P-gp, the pellet was suspended in the reaction buffer plus 0.1% NP-40 at the same cellular concentration. One hundred microliters of cell suspension were incubated with P-gp extracellular antibody (2 μg/tube) (4E3-16, Calbiochem) for 1 hr at room temperature. Cells were washed 3 times with 1% BSA in PBS and diluted again in 100 μl of corresponding reaction buffer. Two microliters of TRITC-conjugated rabbit anti-mouse IgG (Sigma) were added as secondary antibody and incubated for 30 min at room temperature avoiding light exposure. Cells were then washed 3 times as described above and diluted in 500 μl of cold PBS. Cells were analyzed by flow cytometer and P-gp labeling was measured on a red fluorescence detector. Data acquisition and analysis were performed on 10,000 viable cells and values were normalized with respect to those obtained for total P-gp condition with cells expressing GFP. Data are presented as mean ± sd, with n = 3.
For blockade of protein translation, cells (20 × 104 cells/ml) were incubated with 20 μg/ml of the inhibitor of protein translation cicloheximide (CHX) at 37°C in a humidified incubator with 5% CO2 for 24 hr. Thereafter, cells were processed to detect P-gp in the plasma membrane as described above.
Internalization and recycling of P-gp
K562ADR/GFP and K562ADR/Rab4 cells were incubated with monoclonal anti-P-gp 4E3-16 for 1 hr at 4°C, washed extensively with PBS, and incubated in medium at 37°C for the indicated times. The amount of surface P-gp was measured by flow cytometry as described above.
An unpaired t-test was used for statistical analysis. A p < 0.05 was considered significant.
Rab4 and Rab14 appear as interacting proteins of P-gp
To identify proteins that interact with P-gp, we performed a yeast 2-hybrid to screen a rat brain complementary DNA library using the C-terminal of murine P-gp (residues 948–1,276, including the ATP binding site) as the bait. The mouse mdr1b gene shares 91% identity with rat mdr1b and 83% with human mdr1. At the level of the C-terminal domain, the identity augments up to 97 and 94% with the rat and human orthologs, respectively. A brain-derived library was employed because P-gp is highly enriched in the brain, primarily in the blood brain barrier.5, 6 Several diploid yeast colonies resulting from the mating were positive for histidine and adenine biosynthesis and β-galactosidase activity. DNA sequence analysis of the positive clones revealed the presence of several known gene products including traslocon subunits as well as DNA fragments whose sequences are yet to be defined. Notably, we identified 2 cDNA in a few diploid yeast clones whose sequences were highly homologous to 2 members of the Ras oncogene family, Rab4 and Rab14 (Fig. 1a). Rab proteins are highly conserved GTPases of 21–25 kDa (Fig. 1b) that control intracellular vesicular protein transport as well as cellular endocytosis and exocytosis.38, 39 Both proteins show a 58% identity and 67% similarity (Fig. 1b). Although these proteins appear ubiquitously expressed in virtually all tissues, they exhibit a distinct intracellular localization and regulate discrete steps of the endocytotic and secretory pathways. Rab4 is mainly found in early endosomes (EEs) and recycling endosomes, and primarily modulates the sorting and transport from the EEs to the plasma membrane.28, 30, 32 In contrast, Rab14 is localized in the Trans-Golgi Network and has been implicated in the regulation of membrane trafficking between the Golgi complex and the endosomal compartments.29, 31
Rab4 and Rab14 associate with recombinant Ct-P-gp, and with P-gp in K562ADR cells
To verify that both Rab proteins associate with the C-terminal of P-gp, we performed a pull down assay using recombinantly produced GST-Ct-P-gp fusion protein. Incubation of GST-Ct-P-gp with cell extracts from HEK293 cells transfected with human GFP-Rab4 and GFP-Rab14 yielded the formation of complexes that could be isolated by glutathione-based affinity chromatography (Fig. 2a). This interaction was specific since GST alone was unable to pull down the Rab proteins. Similarly, GFP and GST did not interact (Fig. 2a). Thus, the recombinant Ct-P-gp interacts with heterologously expressed human Rab4 and Rab14 proteins.
To evaluate the association of Rab4 and Rab14 proteins with human P-gp we employed a coimmunoprecipitation strategy. For this experiment, we used the anthracycline-resistant erythroleukemia cell line K562ADR, which highly overexpresses P-gp in the plasma membrane as compared with the parental sensitive subline K562 (Figs. 2b and 2c). K562ADR cells were transiently transfected with GFP or GFP-Rab4 or GFP-Rab14, and their protein complexes were immunopurified with an anti-GFP antibody. In the absence of a crosslinker agent, we could neither immunopurify P-gp with the anti-GFP antibody, nor immunoisolate GFP-Rab4 nor GFP-Rab14 with an anti-P-gp antibody (data not shown), suggesting that the interaction between the full length glycoprotein and the Rab GTPases may be of low affinity or very dynamic. In support of this notion, when we used a crosslinker to stabilize the protein complexes we could immunopurify P-gp with the anti-GFP antibody (Fig. 2d), implying that Rab4 and Rab14 associated with the glycoprotein in these cells. Nonetheless, the extent of the interaction appeared to be low as evidenced by the poor intensity of the immunoprecipitated P-gp as compared with the expression level of the glycoprotein in these tumor cells (Fig. 2d). This result may also reflect the relatively modest expression levels of the GFP-Rab proteins, compared to that of P-gp (Fig. 2d). Notice that the blurry band that appears in cells transfected with GFP displays a higher electrophoretic mobility than P-gp, suggesting that may be a contaminant rather than the glycoprotein. Taken together, these results imply that Rab4 and Rab14 interact with P-gp in cells.
To further substantiate the association of Rab proteins with P-gp, we next investigated the codistribution of P-gp with GFP-Rab4 and GFP-Rab14 in K562ADR by immunocytochemistry using a specific anti-Pgp antibody and the fluorescence of the GFP protein. Akin to previous reports,31, 32, 40 we observed that the cellular distribution of the Rab proteins was not drastically altered by their fusion to the GFP protein, as evidenced by the colocalization of the GFP-Rab4 protein with the internalized transferrin (Tf-TRITC) and the EEA1 endosomal marker (Figs. 3d and 3e).
As illustrated in Figure 3a, P-gp displayed preferential plasma membrane localization in K562ADR cells transfected with the GFP protein. A similar result was obtained when the GFP-Rab14 protein was expressed in these cells, namely we could not detect a clear colocalization of this GTPase with P-gp. In marked contrast, P-gp exhibited a cytosolic and surface distribution in K562ADR cells transfected with the GFP-Rab4 (Fig. 3b). Note that both proteins displayed a strong colocalization in the cytosol but not in the plasma membrane, suggesting that the expression of GFP-Rab4 may partly relocalize P-gp into an endosomal fraction. Therefore, all these findings corroborate that P-gp associates with Rab4 in cells, and to a lesser extent with Rab14. Furthermore, the interaction with Rab4 appears to influence the subcellular localization of P-gp.
Transient expression of Rab4 increases the intracellular accumulation of DNM in K562ADR cells
To examine whether the intracellular presence of P-gp in K562ADR cells transiently transfected with Rab4 affected their MDR phenotype, we measured the intracellular accumulation of DNM, a fluorescent substrate of P-gp, by flow cytometry. As shown in Figure 4a, the expression of Rab4, but not of Rab14, significantly augmented the intracellular accumulation of DNM in K562ADR cells. This effect was emulated by the expression of the constitutively active Rab4Q72L mutant, but not by the dominant negative, GDP-blocked, inactive Rab4S27N mutant (Fig. 4b). As expected, neither the active (Rab14Q70L) nor the inactive Rab14S25N mutants had an effect on the intracellular concentration of the drug, similar to the Rab14 wild type protein (Fig. 4c). Note that all proteins display similar levels of expression indicating that the differences in drug accumulation were not due to variations in the amount of the recombinant protein expressed (Fig. 4d). Collectively, our data indicate that expression of Rab4 in K562ADR cells reduces the MDR phenotype, presumably by altering the functionality or expression of P-gp, and that this attenuation of the phenotype requires the GTPase activity of Rab4.
The protein Rab4 modulates the presence of P-gp in the plasma membrane in K562ADR
We next questioned if the lessening of the MDR phenotype by Rab4 was due to the overexpression of the protein rather than a modulatory activity of the GTPase. To address this issue, we stably expressed GFP-Rab4 in the K562ADR cell line. We obtained the stable cell lines K562ADR/GFP and K562ADR/GFP-Rab4 (named K562ADR/Rab4). To minimize cloning artifacts of the G418-based clone selection, we decided to work with a pool of antibiotic-resistant clones showing GFP fluorescence rather than using a single cellular clone. Confocal immunocytochemistry of the clones showed that GFP did not display significant colocalization with P-gp (Fig. 5a). However, a clear intracellular codistribution of the glycoprotein with the GFP-Rab4 is discerned (Fig. 5b). As for the transient expression, GFP-Rab4 appears localized in the cytosol of the cell, presumably in endosomes. Virtually all K562ADR/Rab4 cells displayed the presence of intracellular P-gp that colocalized with the GTPase. These data substantiate a similar expression and colocalization profile of GFP-Rab4 and P-gp in cells transiently and stably transfected with Rab4. Thus, we next performed all the experiments in the stable cell line.
As shown in Figure 6a, the stable K562ADR/Rab4 cells displayed a significantly higher accumulation of DNM than K562ADR/GFP, akin to transient transfectants. To investigate the role of Rab4 protein in the attenuation of the MDR phenotype in these cells, we first evaluated the expression levels of P-gp. The total and surface expressed glycoprotein was measured by flow cytometry using the anti-P-gp 4E3-16 antibody that recognizes an extracellular epitope. The stable expression of GFP-Rab4 produced a measurable leftward shift of the P-gp fluorescence cytogram as compared with that seen in cells stably expressing GFP (Fig. 6b). The position and shape of the cytogram are proportional to the surface expressed P-gp, and a leftward change appears related to a decrease in the expression level of proteins in the plasma membrane.37 As displayed in Figure 6c, a quantitative assessment of the cytograms reveals that the expression of GFP-Rab4 decreased by a significant ≈20% the presence of the glycoprotein in the cell surface as compared with K562ADR/GFP cells, consistent with the confocal images (Fig. 5). This decrement in the surface expressed protein occurred without an alteration in the total P-gp production. This data, along with the colocalization of P-gp and Rab4 in cytosolic compartments, imply that the attenuation of the MDR phenotype by Rab4 is due to a decrease of the amount of glycoprotein in the plasma membrane of tumor cells.
To further evaluate this hypothesis, we hypothesized that blockade of protein translation may induce an increment in the surface expression of P-gp in K562ADR/Rab4 cells. As illustrated in Figure 6d, blockade of protein translation with cycloheximide slightly reduced the amount of surface expressed P-gp in K562ADR/GFP, plausibly because of a decrease in P-gp expression due to protein degradation. In marked contrast, the treatment with CHX significantly incremented the surface P-gp in K562ADR/Rab4 cells, most likely due to mobilization of the P-gp form the Rab4-retained endosomal fraction as a consequence of a decrement in Rab4 levels. This augment in the expression of P-gp in the cell surface resulted in an increment of the MDR phenotype of these cells, as concluded form the higher uptake of DNM in the presence of cycloheximide (data not shown). Taken together, these results support the notion that a significant fraction of P-gp is retained by Rab4 in an endosomal compartment, and that this population may be recruited to the plasma membrane. These observations further indicate that Rab4 specifically influences the localization of P-gp in the plasma membrane by promoting its intracellular accumulation.
We next investigated whether Rab4 modulated the endocytosis of plasma membrane P-gp. For this task, we tagged surface P-gp with the monoclonal anti-P-gp 4E3-16 and measured the uptake of the P-gp-antibody complex as a function of time.26 Cells were tagged at 4°C for 1 hr, and incubated at 37°C for the indicated times (Fig. 6e). Thereafter, the amount of P-gp-antibody complex present in the plasma membrane was determined using a fluorescent anti-mouse antibody by cytometry. As seen in Figure 6e, K562ADR/GFP cells displayed an internalization of P-gp that at 30 min amounted to ≈25% of the prebound monoclonal anti-P-gp 4E3-16. After 4 hr, all P-gp-antibody complexes were recycled to the membrane. Similarly, in K562ADR/Rab4 cells, tagged P-gp exhibited an extent and time course of internalization akin to control cells (Fig. 6e), suggesting that the overexpression of Rab4 did not affect the constitutive endocytosis of the protein in these cells.
K562ADR cells exhibit a reduced expression of Rab4 and its downregulation with siRNA decreases intracellular drug accumulation and increases P-gp in the plasma membrane
The effect produced by Rab4 in the MDR phenotype in K562ADR cells through the modulation of the P-gp level at the cell surface, let us to question which were the expression levels of small GTPases in these tumor cells. The analysis of the mRNA levels by quantitative RT-PCR illustrated that K562ADR cells display a significant 50% reduction of the Rab4A mRNA with respect to that present in the sensitive parental K562 subline (Fig. 7a). This decrement was specific to Rab4 protein, since the mRNA levels of Rab5A, Rab7, Rab8A, Rab11A and Rab14 were identical in the sensitive K562 and resistant K562ADR cell lines. As expected, the lower level of Rab4A mRNA was translated into a reduced amount of expressed protein as evidenced by the Western immunobloting using an anti-Rab4 specific antibody (Fig. 7b). Notice that K562ADR displayed virtually all P-gp at the cell surface (Figs. 2c and 3a). Therefore, it appears that an increment in the expression of the P-gp in K562ADR cells is accompanied by a decrease in Rab4 protein levels.
Because K562ADR cells tend to accumulate low amounts of DNM, we questioned whether further downregulation of Rab4 levels had an impact on the intracellular concentration of the cytotoxic drug, strengthening the degree of the MDR phenotype. For this task, we evaluated the effect of dropping Rab4 levels with a specific siRNA. As illustrated in Figure 8, partial elimination of endogenous Rab4 protein in K562ADR by RNA interference (Fig. 8a) resulted in a significant decrease in the accumulation of DNM in these tumoral cells (Fig. 8b). Notably, downregulation of Rab4 expression resulted in a significant rightward shift of the P-gp cytogram indicating a significant increment in the surface expression of P-gp (Figs. 8c and 8d). These results are consistent with the observations obtained when Rab4 was overexpressed, namely a reduced MDR phenotype due to a lower expression of P-gp in the cell surface. Therefore, our findings lend support to the tenet that Rab4 attenuates the trafficking of P-gp to the plasma membrane and that intervention on this molecular pathway may be a valuable therapeutic strategy.
P-gp is an efflux pump, highly expressed in epithelial and endothelial cells, which plays a key role in the extrusion of xenobiotics.7 In addition, this cellular transporter may be naturally expressed or chemotherapeutically induced in tumor cells, leading to the MDR phenotype.42–44 The main consequence of having high expression levels of P-gp is a decrease in the intracellular concentration of cytotoxic drugs. The key role of this protein in both human physiology and pathology has prompted intense investigations to understand its mechanism of action, as well as the molecular details involved in its biogenesis and trafficking to the cell surface. The identification of proteins that interact with P-gp is important for our further understanding on the function of P-gp, as well as its mechanism of action. Previous studies have shown that this glycoprotein associates with caveolin-1,20–22, 45, 46 Bap29varP,23 RNF 218 and α and β tubulins.19 The interaction of P-gp with caveolin-1, mediated through residues 37–45 at the N-terminal of the transporter, was found in both endothelial and MDR cells, and resulted in the inhibition of its transport activity.21 Bap29varP also interacted with N-terminal of the protein, and appeared to function as a chaperone that influences the processing and trafficking of P-gp.23 In contrast, the E3 ubiquitin ligase RNF2 binds to a motif found in the middle of the protein (aa 633–709) and regulates the cellular abundance of the protein.18
Despite this information, the numbers of interacting partners of P-gp appears limited and, intriguingly, there are not known proteins that bind to the C-terminal of P-gp, a domain holding one of the ATP binding domains. Thus, we addressed this question and used a yeast 2-hybrid approach to isolate proteins that interact with this key region of the protein and may have important functional implications. The most salient contribution of our study is the finding that P-gp directly binds to 2 GTPases, Rab4 and Rab14, which serve as regulators of vesicular transport. Rab4 is typically located in EEs and it is associated with sorting, recycling and transport to the plasma membrane.28, 30, 47–49 Rab14 is a recently identified close homolog of Rab4 (58% identity) that localizes in the Golgi and endosomal compartments where it appears involved in membrane trafficking between the Golgi complex and endosomes.29, 31 Both proteins were found to associate with P-gp through their interaction with the protein C-terminal.
Functional analysis was carried out in the tumor cell line K562ADR that expresses high levels of P-gp. Intriguingly, these MDR cells expressed quite low levels of Rab4, compared with the parental sensitive K562 cell line, while exhibiting similar expression of other Rab proteins including Rab14. We observed that transient overexpression of GFP-Rab4, but not of Rab14, in this cell line regulated the surface expression of the glycoprotein by promoting its retention in an endosomal fraction where it colocalized with Rab4. As a result, K562ADR cells that overexpress GFP-Rab4 exhibited a decrease in MDR, as evidenced by the higher intracellular concentrations of the anti-tumoral drug DNM, a well-known substrate of the pump. This MDR sensitizing effect was dependent of on the GTP-GDP status since it was emulated by the constitutively active, GTPase deficient Rab4Q72L mutant, and inhibited by the dominant negative, GDP-blocked Rab4S27N mutant.
Because transient expression of proteins in cells normally yields high amounts of the polypeptides, it raises concerns of whether the functional effects may be due to the excess of protein production. To address this issue, we generated a stable K562ADR cell line expressing Rab4. These cells display protein levels for the GTPase lower than those transiently transfected with the cDNA. However, functional characterization of the stable cell line revealed results akin to those gathered with transient overexpressing of the GTPase. Hence, stable expression of Rab4 significantly sensitized the MDR phenotype of K562ADR tumor cells. This sensitization was primarily due to a reduction of the amount of P-gp in the cell surface. The decrement of P-gp expression in the plasma membrane was accompanied by an increase in an intracellular endosomal fraction from where it could be transported to the plasma membrane. In support of this notion, inhibition of protein translation by cycloheximide or downregulation of Rab4 levels with siRNA selectively induced an increment of the surface expressed P-gp in K562ADR stably transfected with GFP-Rab4. Taken together, our findings indicate that Rab4 modulates P-gp function by impairing the protein expression at the plasma membrane. A similar role of Rab4 modulating the expression of amiloride-sensitive sodium channels, and CFTR at the cell surface has been reported.50, 51 These studies also show that overexpression of Rab4 in the cancer cell line HT29 reduces the surface expression of these 2 proteins by retaining them in an intracellular, endosomal fraction.
Cumulative evidence is showing that the plasma membrane localization of P-gp in tumor cells could be readily modified by modulating its endocytic/recycling traffic. In this context, treatment with chloroquine or overexpression of Rab5 or Ral A notably increased the intracellular levels of P-gp, by facilitating its redistribution from the cell surface to intracellular compartments.24, 26 These molecular manipulations exhibited a reduced multidrug resistant phenotype. Our results further substantiate these findings and indicate that Rab4 plays a key role in the trafficking and distribution of P-gp. Furthermore, our findings are consistent with a specific regulation of P-gp exocytosis by the small GTPase, as we could not observe impairment of the glycoprotein internalization in cells overexpressing Rab4. Noteworthy, the observation that K562ADR cells exhibit a lower level of expression of Rab4 than their sensitive counterpart, suggests the exciting hypothesis that downregulation of Rab4 levels could be a molecular event associated with the development of the MDR phenotype in some tumors. Interestingly, resistant MCF7ADR cells also exhibit an apparent loss of the E3 ubiquitin ligase RFN2, a protein that also interacts with P-gp,18 further substantiating the concept that acquisition of the MDR phenotype is associated to an increase of P-gp that involves upregulation of the transporter and concomitant downregulation of proteins that may influence its cellular abundance or its expression in the cell surface. Future experiments are required to better understand the role of Rab4 in the cellular distribution of P-gp and other related drug transporters, as well as in the manifestation MDR.
In conclusion, we have reported that Rab4 downregulates the surface expression of P-gp in cancer cells, by promoting its localization in a cytosolic endosomal fraction. As a result, expression of Rab4 attenuates the degree of MDR. Our findings imply that modulation of the temporal and spatial location of P-gp within cancer cells has an impact in the extent of the MDR phenotype and, therefore, suggest that it may be a clinically relevant strategy to sensitize chemotherapeutically resistant tumor cells.
This work was supported by grants from the Spanish Ministry of Science and Innovation (to A.F.-M.), the Fundación La Marató de TV3 (to A.F.-M).