Protein kinase C-mediated phosphorylation does not regulate drug transport by the human multidrug resistance P-glycoprotein.

P-glycoprotein (P-gp) is an active transporter that can confer multidrug resistance by pumping cytotoxic drugs out of cells and tumors. P-gp is phosphorylated at several sites in the "linker" region, which separates the two halves of the molecule. To examine the role of phosphorylation in drug transport, we mutated P-gp such that it could no longer be phosphorylated by protein kinase C (PKC). When expressed in yeast, the ability of the mutant proteins to confer drug resistance, or to mediate [3H]vinblastine accumulation in secretory vesicles, was indistinguishable from that of wild type P-gp. A matched pair of mammalian cell lines were generated expressing wild type P-gp and a non-phosphorylatable mutant protein. Mutation of the phosphorylation sites did not alter P-gp expression or its subcellular localization. The transport properties of the mutant and wild type proteins were indistinguishable. Thus, phosphorylation of the linker of P-gp by PKC does not affect the rate of drug transport. In light of these data, the use of agents that alter PKC activity to reverse multidrug resistance in the clinic should be considered with caution.

Multidrug resistance is responsible for the failure of chemotherapy in tens of thousands of cancers each year. Overexpression of the 170-kDa P-glycoprotein (P-gp) 1 can confer multidrug resistance on otherwise drug-sensitive cells and tumors (reviewed by Gottesman and Pastan (1993)). P-gp is an active transporter that utilizes the energy of ATP hydrolysis to pump cytotoxic drugs out of cells, reducing their intracellular concentrations and, hence, their toxicity. The P-gp molecule is composed of two halves, which share significant similarity to each other. Each half-molecule consists of a transmembrane domain, which participates in substrate binding and forms the pathway through which substrate crosses the membrane; and a nucleotide-binding domain, which couples the energy of ATP hydrolysis to transport. Between the two halves of the molecule is a highly charged region, the "linker." The linker appears to be the only region of P-gp that is phosphorylated, either in vivo or in vitro (Chambers et al., 1990;Ma et al., 1991;Orr et al., 1993). Within the linker there are eight consensus sites for phosphorylation by protein kinase C (PKC) and/or protein kinase A. Only three of these sites in human, and two in mouse, normally appear to be accessible to phosphorylation (Chambers et al., 1993(Chambers et al., , 1994Orr et al., 1993).
Several studies have led to the suggestion that PKC-mediated phosphorylation of P-gp might modulate drug transport and, hence, might provide a novel target for reversing multidrug resistance in the clinic (reviewed by Germann et al. (1995)). However, all these studies are subject to alternative interpretations (see "Discussion"). In order to address the role of phosphorylation in a rigorous fashion, we mutated the phosphorylation sites such that P-gp could no longer be phosphorylated by PKC. Real-time measurement of drug transport following expression of the wild type and mutant P-gps in yeast and human cells showed that phosphorylation of the linker region has no significant effect on the rates of drug transport.

Construction of Mutations in the Human MDR1 Gene-Mutations
introduced into the linker of the human MDR1 gene are shown in Fig.  1. Construction of mutant MDR1 genes that express P-gp8A and P-gp8E (the eight consensus serine/threonine residues mutated to alanine or glutamate, respectively) has been described previously (Hardy et al., 1995). Mutant MDR1 genes in which three sites, Ser-661, Ser-667, and Ser-671, were altered to alanines or glutamates were constructed by oligonucleotide-directed mutagenesis by similar procedures. Mutagenesis was performed using oligonucleotides 5Ј-GATTCAAGA-TCCGCCCTAATAAGAAAAAGAGCCACTGTAGGGCCGTCCGTGG-ATCAC-3Ј to construct the P-gp3A mutant and 5Ј-GATTCAAGATCC-GAGCTAATAAGAAAAAGAGAGACTCGTAGGGAGGTCCGTGATC-AC-3Ј to construct the P-gp3E mutant. The entire mutated MDR1 DNA fragment was sequenced after mutagenesis to ensure that only the intended changes had been introduced. The construction of pMDR712, expressing the human MDR1 gene mutated in both ATP-binding domains (K433M and K1076M), has been described previously .

Cloning of the Mutated MDR1 Genes into a Yeast Expression Vector-
The entire MDR1 open reading frame was cloned into the yeast expression vector pVTU101 (Vernet et al., 1987) under the control of the alcohol dehydrogenase promoter as a 3.9-kilobase pair PvuII-XhoI fragment isolated from pMDR7S , generating plasmid pVTMDR. The mutant genes were introduced into pVTMDR as 2199base pair EcoRI-PstI fragments to generate pVT737 (P-gp8A), pVT738 (P-gp8E), pVT739 (P-gp3A), and pVT740 (P-gp3E). These plasmids were then checked for introduction of the relevant mutations by DNA sequencing.
Measurement of Vinblastine Accumulation and Drug Cytotoxicity in Yeast-Saccharomyces cerevisiae strain SY1 (MAT␣,112, was grown at 25°C on supplemented medium (0.7% yeast nitrogen base without amino acids, 0.1 mM histidine, 0.2 mM leucine, 0.2 mM uracil) containing 2% glucose. Yeast strain SY1 carries a temperature-sensitive mutation and, at the restrictive temperature, accumulates secretory vesicles within the cytoplasm (Walworth and Novick, 1987). These vesicles are exclusively inside-out and of uniform size. Secretory vesicles were prepared and [ 3 H]vinblastine (Amersham Corp.) accumulation measured as described (Ruetz and Gros, 1994). The effect of valinomycin (20 g/ml) on yeast growth was assessed by measurement of absorbance at 600 nm, using strain JPY201 transformed with each plasmid, grown at 30°C in YPD broth.
Generation of Cell Lines Expressing Wild Type and Mutant P-gps-GM0637 is an SV40-transformed lung fibroblast cell line expressing negligible levels of P-gp and was obtained from the Imperial Cancer Research Fund cell culture facility. Cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum. Cells were transfected with derivatives of the expression plasmid pcDNA3 (InVitrogen Corp.). The MDR1 cDNA was cloned into pcDNA3 as a 3.9-kilobase pair SmaI-XhoI fragment (nucleotides 425-4265 from pMDR7S  to give pcMDR, which expresses wild type P-gp. To construct an equivalent plasmid expressing P-gp8A, the 2199-base pair EcoRI-PstI fragment from pMDR737 (nucleotides 1601-3800) was cloned between the EcoRI-PstI sites of pcMDR to generate plasmid pcMDR737. These two plasmids were transfected into GM0637 cells grown to 50% confluence in 15-cm dishes. pcMDR or pcMDR737 DNA (30 g) was mixed with DC-Chol lipid vesicles (150 g) in 18 ml of Opti-MEM (Life Technologies, Inc.) at room temperature for 10 min, added to the cells, and incubated for 3 h. Cells were washed once with PBS and then returned to normal media. When confluent, the cells were split 1:3 and replated in media containing Geneticin (Life Technologies, Inc.) (400 g/ml). Resistant colonies were allowed to reach confluence and then cultured sequentially in media containing increasing concentrations of vinblastine (10, 20, 40, 75, and 100 ng/ml). Cells transfected with pcMDR and pcMDR737 were designated GM.P-gp and GM.P-gp8A, respectively, and were maintained in media supplemented with vinblastine (100 ng/ml) and Geneticin (400 g/ml).
Transient expression of P-gp in S1 cells was carried out using the vaccinia-T7 hybrid expression system (Fuerst et al., 1986;Elroy-Stein et al., 1989) as described previously Gill et al., 1992;Sardini et al., 1994). The plasmids used were based on the wild type P-gp-expressing plasmid pMDR7, which has been described . pMDR712 is a derivative of pMDR7 in which Lys 3 Met changes have been introduced in the "Walker motif" of each of the nucleotide-binding domains. This mutant protein is severely impaired in its drug transport activity (Sardini et al., 1994). pMDR737 and pMDR738 are derivatives of pMDR7, which express P-gp8A and P-gp8E, respectively.
Drug Cytotoxicity Assays-Cytotoxicity assays were based on published methods (Cano-Gauci and Riordan, 1987). Briefly, cells were seeded in 24-well dishes at a density of 4 ϫ 10 3 cells/well in 1 ml of DMEM with 10% fetal calf serum. Drug was added at the indicated concentrations and cells grown until untreated cells reached confluence. Cell growth is expressed as a percentage relative to cells grown without drug. Verapamil, vinblastine, colchicine, and doxorubicin were from Sigma.
Calcein-AM Assay for P-gp Activity-Calcein accumulation was measured in constantly stirred, detached cell suspensions (1 ϫ 10 6 cells) in 1 ml of transport buffer (see above), as described previously (Homolya et al., 1993). Calcein-AM (Molecular Probes) was at 2.5 M. Fluorescence was measured using a Shimadzu RF1001PC fluorimeter (37°C, ex 493 Ϯ 1.5 nm; em 515 Ϯ 1.5 nm). Calcein fluorescence in individual cells was measured by epifluorescence confocal microscopy (Sardini et al., 1994) with ex ϭ 488 nm and em Ͼ 515 nm. Fluorescence changes were detected as a function of time by collecting sequential images, usually at 20-s intervals, with the cells maintained in the presence of calcein-AM (2.5 M). Cells were plated 48 h before experiments on coverslips coated with poly-L-lysine and were maintained in DMEM with 10% FCS containing vinblastine (100 ng/ml) and Geneticin (400 g/ml). Vinblastine was removed 24 h before coverslips were used. Experiments were performed at 37°C with a perfusion solution con-taining 140 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 0.5 mM MgCl 2 , 10 mM Hepes, 5.5 mM D-glucose, pH 7.3.
Accumulation of calcein, measured by fluorescence increase, is directly related to the rate of transport by P-gp as follows. We define the influx of calcein-AM into the cell as I, and the concentrations of calcein-AM and its fluorescent product calcein to be [C-AM] and [C], respectively. Calcein-AM is either removed by transport from the cell, with rate constant ␣, or is hydrolyzed to calcein in the cytoplasm by endogenous esterases with rate constant ␤. Therefore, at equilibrium, From these two equations, and with the assumption that ␣ Ͼ Ͼ ␤, the following relationship can be derived between the P-gp transport rate ␣ and [C], which is proportional to cell fluorescence.
This argument is not changed if P-gp transports drug from the inner leaflet of the lipid bilayer instead of from the cytoplasm (Higgins and Gottesman, 1992;Homolya et al., 1993), provided that partition of the drug between membrane and cytoplasm is rapid. The concentration of drug in the membrane, which determines the efflux rate by P-gp, is then a constant multiple of the concentration of drug in the cytoplasm, and the equations above are unchanged. Western Blotting-Proteins were extracted from cell membranes in 2% SDS and protein quantified using the BCA assay (Pierce). Proteins was separated by electrophoresis on a 7.5% SDS-polyacrylamide gel (Laemmli, 1972) and transferred to a nitrocellulose membrane (Hybond, Amersham) (Towbin et al., 1979). P-glycoprotein was identified by probing with the monoclonal antibody C219 (1 g in 10 ml) (Centocor Diagnostics) and detected by chemiluminescence (ECL, Amersham). Anti-PKC␣ antibody was a kind gift of Peter Parker (Imperial Cancer Research Fund).
Immunohistochemistry-Two monoclonal antibodies were used to detect the expression of P-gp: JSB1 (Boehringer Mannheim), which recognizes an internal epitope; and UIC2, which recognizes an external epitope. For JSB1, cells were fixed at Ϫ20°C for 4 min in methanol followed by 4 min in acetone. This treatment completely permeabilizes the cells, making the internal epitope accessible to the antibody. After three washes in PBS ϩ 0.2% fish skin gelatin (Sigma), coverslips were incubated for 30 min at ambient temperature with antibody (5 g/ml). For UIC2, cells were fixed in PBS plus 3% paraformaldehyde. The cell membrane is not permeabilized, and, on exposure to UIC2 (5 g/ml), only surface membrane P-gp is detected. Cells were then washed three times with PBS ϩ 0.2% fish skin gelatin. Antibody labeling was visualized by incubation with a fluorescently labeled secondary antibody (5 g/ml for 20 min at room temperature) raised against the specific class of primary antibody (anti-mouse IgG 1 conjugated to Texas Red for JSB1, and anti-mouse IgG 2a conjugated to rhodamine for UIC2; both from Euro-Path). After thorough washing with PBS, coverslips were mounted onto slides with anti-fading agent (Vectashield mounting medium H-1000) and images acquired using the confocal microscope. From confocal optical sections, it was verified that all UIC2 label was located FIG. 1. Mutation of the consensus phosphorylation sites of P-gp. Diagram of P-gp indicating the two transmembrane domains and two nucleotide-binding domains. The linker region is located between the two halves of the molecule and the protein sequence containing the consensus phosphorylation sites (amino acids 660 -684 of human P-gp) is shown. The residues altered in the mutants described in this study are shown. The names of the yeast plasmids expressing these mutant proteins are given.
on the surface membrane.
Phosphorylation of P-gp in Cell Membranes-Membranes were prepared as detailed by Cornwell et al. (1986). The membranes were enriched for integral membrane proteins by sodium carbonate treatment as described by Fujiki et al. (1982). In vitro phosphorylation of P-gp by PKC in membranes was as described (Chambers et al., 1992). Briefly, membranes (250 g) in phosphorylation buffer (10 mM Tris-Cl, 25 mM sucrose, 10 mM MgCl 2 , 0.5 mM CaCl 2 , 0.25% bovine serum albumin, 10 M ATP, 100 g/ml phosphatidylserine, 20 g/ml 1,2dioleolyl-sn-glycerol, 50 ng/ml calyculin A) and 10 Ci of [ 32 P]orthophosphate (3000 Ci/mmol; Amersham) were incubated with PKC (0.1 unit) at 30°C for 10 min. The reaction was stopped by addition of 5 ϫ Laemmli buffer and proteins separated by SDS-polyacrylamide gel electrophoresis. Phosphorylated proteins were detected and quantitated using a PhosphorImager (Molecular Dynamics). PKC (rat brain) was from Calbiochem.

Mutation of the Consensus PKC Phosphorylation
Sites of P-gp-There are eight serine/threonine residues in the linker of P-gp that could, potentially, be phosphorylated (Fig. 1). However, it has been demonstrated that three of these, serines 661, 667, and 671, are the major sites of phosphorylation by PKC both in vivo and in vitro (Chambers et al., 1993(Chambers et al., , 1994. Thus, site-directed mutagenesis was used to change these three serine residues to alanines in order to prevent phosphorylation (P-gp3A). In a second mutant (P-gp3E), the serines were replaced by glutamate; this also prevents phosphorylation but, in addition, mimics the negative charge of phosphorylation. To exclude the possibility that secondary sites within the linker region might be phosphorylated when the primary sites were mutated, two additional mutations were generated in which all eight serine/threonine residues were changed, either to alanine (P-gp8A) or to glutamate (P-gp8E). The sequences of the linkers in these mutant proteins are presented in Fig. 1.
Mutation of the PKC Phosphorylation Sites of P-gp Does Not Alter Drug Transport Rates When Expressed in Yeast-Expression in yeast provides a convenient means of assessing the transport properties of mutant P-gps (Ruetz and Gros, 1994). Yeast plasmids encoding the wild type or mutated P-gps (pVTU101, pVTMDR, pVT737, pVT738, pVT739, pVT740; see Fig. 1) expressed the respective proteins at identical levels, as assessed by Western blotting (Fig. 2A). As expected, expression of wild type P-gp conferred resistance on yeast cells to 20 g/ml valinomycin (Fig. 2B). The level of resistance conferred by the four mutant P-gps was indistinguishable from that of wild type P-gp. Thus, mutation of the phosphorylation sites had no effect on the ability of P-gp to confer valinomycin resistance on yeast.
Panels C and D, vinblastine uptake into SVs prepared from yeast transformed with plasmids expressing wild type or mutant P-gps. SVs (1 mg of protein/ml) were incubated with 1 M [ 3 H]vinblastine in transport buffer for the times indicated. Verapamil (25 M) was included in the transport buffer where indicated. Data are the average of duplicates from at least three independent experiments. C, Ⅺ, pVTMDR (wild type P-gp); E, pVTU101 (vector, no P-gp); f, pVTMDR ϩ verapamil; q, pVTU101 ϩ verapamil. D, Ⅺ, pVT737 (P-gp8A); E, pVTU738 (P-gp8E); f, pVT739 (P-gp3A); q, pVT740 (P-gp3E); å, pVT737 ϩ verapamil. Verapamil inhibited transport by all mutant P-gps to similar extents. 1994). All four mutant P-gps confer rates of vinblastine uptake, and maximum levels of drug accumulation, indistinguishable from each other and from those of wild type P-gp (66 -73 Ϯ 14 pmol/mg protein). Verapamil (25 M), a known blocker of P-gp function, abolished vinblastine uptake by the wild type and mutant P-gps. Thus, mutation of the phosphorylation sites of P-gp had no effect on its transport activity when expressed in yeast.
Generation of Human Cell Lines Expressing P-gp Mutated in the Phosphorylation Sites-To investigate the role of P-gp phosphorylation in human cells, a matched pair of cell lines was generated that express the wild type and P-gp8A proteins (see "Experimental Procedures"). Western blotting and quantitation by densitometry showed that the two cell lines, GM.P-gp and GM.P-gp8A, express identical levels of wild type or mutant P-gp (Fig. 3A). The subcellular distribution of P-gp in the two cell lines was quantitated by fluorescence-linked immunocytochemistry. Two antibodies were used: UIC2, which binds to an external epitope; and JSB1, which binds to an internal epitope. In non-permeabilized cells, UIC2 assesses the amount of P-gp at the cell surface; in permeabilized cells, both JSB1 and UIC2 assess the total amount of P-gp. There was considerable heterogeneity of P-gp expression between individual cells (Fig. 3B) because the plasmids are maintained episomally and, consequently, the number of copies of the plasmid inherited by each daughter cell varies following cell division. However, the mean total fluorescence signals from large populations of individual permeabilized cells was similar for both cell lines. No significant difference in subcellular location of P-gp was detected between the two cell lines and the amount of P-gp in the cytoplasmic membrane, detected by UIC2 in non-permeabilized cells, was similar (Fig. 3B). Thus, phosphorylation of the linker of P-gp does not affect its subcellular localization or its trafficking to the plasma membrane.
Mutation of the Phosphorylation Sites in the Linker of P-gp Abolishes PKC-mediated Phosphorylation-Previous studies have shown that the linker is the only site of P-gp phosphorylation (Chambers et al., 1993(Chambers et al., , 1994Orr et al., 1993). To confirm that the mutated protein was not phosphorylated in the linker, or elsewhere, membrane vesicles prepared from GM0637, GM.P-gp, and GM.P-gp8A cells were converted to sheets by carbonate treatment, and incubated with radiolabeled orthophosphate and purified PKC (Fig. 4). As expected, no phosphorylated P-gp was detected in membranes from the non-P-gp-expressing parental cells (GM0637), while P-gp from cells expressing wild type P-gp (GM.P-gp cells) was phosphorylated. Importantly, P-gp in membranes expressing the mutant P-gp (GM.P-gp8A cells) could not be phosphorylated, even though the amount of P-gp in the membranes was equivalent to that of the cell line expressing wild type P-gp (see Fig. 3). Autophosphorylation of PKC (the band at 82 kDa) and of other membrane proteins provides an internal control for the phosphorylation reaction. Thus, the mutant P-gp cannot be phosphorylated, either in the linker region or at alternative, secondary sites.
It has been reported that expression of PKC may be increased in multidrug-resistant cells (Palyoor et al., 1987;Fine et al., 1988). The levels of total PKC, and of the PKC␣ isoform that is believed to be responsible for P-gp phosphorylation (Yu et al., 1991;Ahmad et al., 1994), were unchanged following expression of the wild type mutant P-gp, as assessed by immunoblotting (data not shown).
Mutation of the PKC Phosphorylation Sites Does Not Alter the Ability of P-gp to Confer Drug Resistance-The cytotoxic effect of several different drugs was assayed for the three cell lines (Table I). The IC 50 for vinblastine in untransfected GM0637 cells was 0.75 ng/ml. In GM.P-gp cells, expressing wild type P-gp, the IC 50 for vinblastine was 318 Ϯ 61 ng/ml (n ϭ 3), a 425-fold increase in resistance. Cells transfected with the mutant P-gp8A (GM.P-gp8A cells) showed levels of resistance similar to those of wild type P-gp (IC 50 ϭ 283 Ϯ 51 ng/ml; n ϭ 3). The resistance to colchicine and doxorubicin conferred by wild type P-gp and P-gp8A was also indistinguishable (Table  I). Thus, mutation of PKC phosphorylation sites in the linker region of P-gp has no effect on its ability to confer drug resistance or its drug resistance profile.

Mutation of PKC Phosphorylation Sites of P-gp Does Not Affect Rates of Vinblastine Accumulation or Doxorubicin Efflux-[ 3 H]Vinblastine accumulation by cell monolayers was
used as a direct assessment of P-gp transport activity. Vinblastine accumulation by the parental cells (GM0637), which do not express P-gp, was rapid, reaching a plateau of 40.5 nmol/g protein after 60 min of incubation. In contrast, cells expressing wild type P-gp (GM.P-gp cells) or the mutant P-gp8A protein (GM.P-gp8A cells) showed identical, severely reduced rates of vinblastine accumulation (Ͻ2.5 nmol/g protein). This reduced drug uptake for both cell lines was reversed by verapamil (50 M).
Doxorubicin efflux was measured by confocal fluorescence microscopy as described previously (Sardini et al., 1994). S1 cells were transiently transfected with wild type and mutant P-gps and preloaded with doxorubicin (see "Experimental Procedures"). In cells transfected with plasmid pMDR7, expressing wild type P-gp, the time constant for doxorubicin efflux was 827 Ϯ 74 s (mean Ϯ S.E.; n ϭ 14). As a control, cells were transfected with pMDR712, which is mutated in the ATP-binding sites such that drug transport is impaired Sardini et al., 1994). The time constant for drug efflux (3078 Ϯ 404 s, n ϭ 4) was significantly different from that of wild type P-gp (p Ͻ 0.01). In contrast, the time constants for doxorubicin efflux from cells transfected with pMDR737 (expressing P-gp8A, time constant 1481 Ϯ 230 s, n ϭ 5) or pMDR738 (expressing P-gp8E; time constant 1085 Ϯ 87 s, n ϭ 10) were similar to that of cells expressing wild type P-gp. Thus, phosphorylation of the linker does not alter the ability of P-gp to mediate vinblastine accumulation or doxorubicin efflux.
Mutation of the P-gp Phosphorylation Sites Does Not Alter Rates of Calcein-AM Transport-The "calcein" assay (Homolya et al., 1993) was used to measure rates of P-gp-mediated transport in real time; Calcein-AM is a non-fluorescent, hydrophobic compound, which is hydrolyzed rapidly by cytoplasmic esterases to release the membrane-impermeable, fluorescent calcein moiety. The rate of calcein-AM uptake into a population of cells can be measured by monitoring the rate of fluorescence increase. Calcein-AM is a substrate for P-gp, and in P-gp expressing cells its rate of uptake (increase in fluorescence) is reduced. The rate of calcein-AM uptake in cells expressing wild type P-gp was considerably reduced (0.02 units/s) compared with that of the parental, non-P-gp-expressing cells (0.42 units/s) (Fig. 5). Importantly, transport by the mutant P-gp8A protein (GM.P-gp8A cells) was identical to that of wild type P-gp (GM.P-gp cells). Addition of verapamil (50 M) restored the rate of calcein accumulation by cells expressing wild type P-gp and P-gp8A to that of the untransfected, parental cells.
Heterogeneity in P-gp expression was observed between individual cells within a population (see Fig. 3B). This might mask differences in drug transport rates between the wild type and P-gp8A proteins measured for a population of cells. To exclude this possibility, the rate of drug transport was measured in individual cells by an adaptation of the calcein assay. Fig. 6 shows false-color images of calcein accumulation by FIG. 5. Accumulation of calcein-AM in cells expressing wild type (GM.Pgp) or mutant (GM.P-gp8A) P-gp. 10 6 cells were exposed to C-AM (2.5 M) at 37°C, and the increase in fluorescence monitored as the drug was hydrolyzed intracellularly. Relative fluorescence units were used. Verapamil (50 M) was added as shown. E, GM0637 cells; ϩ, GM.P-gp cells; f, GM.P-gp8A cells. individual cells of the untransfected parental line (GM0637; panel A) and of cells expressing wild type P-gp (GM.P-gp cells; panel B). As expected, the increase in fluorescence was much slower for the P-gp-expressing cells and this was reversed by verapamil (50 M). The quantified fluorescence data from an individual cell expressing wild type P-gp is shown in Fig. 7A (continuous line). The transport rate of P-gp (solid circles), which is proportional to the reciprocal of the rate of change of fluorescence, was rapidly and reversibly inhibited by verapamil. Fig. 7B shows the mean rates of P-gp transport rate derived from such fluorescence measurements made on a large number of individual cells. The rates of transport in cells expressing wild type P-gp and P-gp8A (GM.P-gp and GM.P-gp8A, respectively) were significantly greater than in the non-expressing parental line (GM0637). Rates of transport by wild type P-gp and P-gp8A were similar. The rates of transport by P-gp and P-gp8A were reduced by verapamil to the background rates measured in non-P-gp-expressing (GM0637) cells. Thus, in this single-cell assay, mutation of the phosphorylation sites of P-gp has no major effect on its transport activity. DISCUSSION P-glycoprotein was first shown to be phosphorylated nearly 20 years ago (Carlsen et al., 1977). Subsequently, P-gp was shown to be phosphorylated by PKC at several discrete sites in the linker region, both in vivo and in vitro. This has led to the suggestion that phosphorylation might regulate drug transport by P-gp and, therefore, provide a target for reversal of MDR in the clinic. However, in this study we show that PKC-mediated phosphorylation has no role in regulating the drug transport activity of human P-gp.
This conclusion is in apparent contradiction to several published studies, which have indicated a role for PKC phosphorylation in regulating drug transport by P-gp. However, these published studies can be interpreted in other ways. Thus, although compounds that modulate PKC activity can alter drug transport (Fine et al., 1988;Sato et al., 1990;Bates et al., 1992Bates et al., , 1993Miyamoto et al., 1993), these PKC modulators frequently lack specificity and some (e.g. staurosporine) bind to P-gp and compete directly for drug transport (Sato et al., 1990;Miyamoto et al., 1992;Wakusawa et al., 1992Wakusawa et al., , 1993Lelong et al., 1994). Similarly, although an increase in PKC expression is characteristic of MDR cells (O'Brian et al., 1989;Lee et al., 1992;Blobe et al., 1993), this could be a consequence rather than a cause of multidrug resistance. Finally, although overexpres-

FIG. 6. Calcein-AM accumulation measured in individual cells.
Fluorescence intensity is proportional to the concentration of accumulated calcein, the breakdown product of calcein-AM, and is shown color-coded on a scale from blue (low) to white (high). Cells were prescreened with a brief exposure to calcein-AM to allow those cells within the population expressing low levels of P-gp to be identified and excluded. Panel A, four representative frames showing the parental GM0637 cell line incubated with calcein-AM (2.5 M). The time delay between successive frames was 80 s. Panel B, eight representative frames showing the P-gp-expressing cell line (GM.P-gp) incubated with calcein-AM (2.5 M). The time delay between frames was 80 s. Note the substantially lower rate of calcein accumulation in frames 1-4 when compared to the parental cell line. Between frames 7 and 8, verapamil (50 M) was applied, inhibiting P-gp function and leading to a rapid increase in fluorescence. Filled circles (right-hand ordinate): P-gp transport rate calculated as inverse of rate of change of fluorescence (see "Experimental Procedures"). Panel B, mean P-gp activity derived from experiments on many individual cells expressing P-gp (GM.P-gp), P-gp.8A (GM.P-gp8A), or the non-expressing parental cells (GM0637). Experiments on each cell type were carried out in the presence and absence of verapamil. Mean P-gp activity was calculated from data obtained from many experiments similar to that shown in A. Bars show Ϯ S.E. Each mean value is from a total of at least 20 cells from at least seven independent experiments. sion of PKC in P-gp-expressing cells can increase multidrug resistance (Yu et al., 1991;Fan et al., 1992), this is probably via altered MDR gene expression (Abraham et al., 1990;Chaudhary and Roninson, 1992;Chin et al., 1992;Sampson et al., 1993;Uchiumi et al., 1993).
It is generally accepted that PKC normally phosphorylates the linker of P-gp in vivo. Compounds that alter PKC activity alter P-gp phosphorylation (Bates et al., 1992;Chambers et al., 1990Chambers et al., , 1992, and PKC␣ co-immunoprecipitates with P-gp (Ahmad et al., 1994). Consistent with previous studies (Chambers et al., 1993(Chambers et al., , 1994Orr et al., 1993), the data here confirm that PKC does not phosphorylate P-gp outside the linker. Thus, a role for PKC-mediated phosphorylation of P-gp in regulating its transport activity can be excluded. It has also been reported that other kinases may phosphorylate P-gp (Staats et al., 1990;Sampson et al., 1993;Germann et al., 1995;Lelong et al., 1994). However, as none of these enzymes appear to phosphorylate P-gp outside the linker, and as we have shown phosphorylation of the linker has no role in drug transport, it seems unlikely that these other kinases modulate drug transport. Nevertheless, the possibility remains that, in specific cell types, one or more of these enzymes might phosphorylate P-gp at a site outside the linker and, thereby, modulate its transport activity.
If PKC-mediated phosphorylation of the linker of P-gp has no role in modulating the rate of drug transport, then what is the role of the linker and of phosphorylation? Besides its role as an active transporter, P-gp has a second, distinct activity as a regulator of cell swelling-activated chloride channels Gill et al., 1992;Luckie et al., 1994). PKC-mediated phosphorylation of the linker modulates the efficacy with which P-gp regulates the heterologous channel (Hardy et al., 1995). Thus, phosphorylation of the linker appears to modulate the activity of P-gp as a channel regulator, rather than as an active transporter. Other ABC transporters also act as channel regulators (Higgins, 1995). As for P-gp, the ability of the cystic fibrosis gene product CFTR to regulate heterologous channels also depends on phosphorylation of a domain (the "R-domain"), which separates the two halves of the molecule and is the major site of phosphorylation (Riordan et al., 1989). Thus, the linker of P-gp and the R-domain of CFTR may serve analogous roles in modulating the ability of these ABC proteins to act as regulators of heterologous channels.