Antiepileptic Drugs Reduce Efficacy of Methotrexate Chemotherapy By

Leukemia (2009) 23, 1087–1097 & 2009 Macmillan Publishers Limited All rights reserved 0887-6924/09 $32.00 www.nature.com/leu ORIGINAL ARTICLE

Antiepileptic drugs reduce efficacy of methotrexate chemotherapy by downregulation of Reduced folate carrier transport activity

S Halwachs1, I Scha¨fer 2, P Seibel2 and W Honscha1

1Faculty of Veterinary Medicine, Institute of Pharmacology, Pharmacy and Toxicology, University of Leipzig, Leipzig, Germany and 2Department of Molecular Cell Therapy, Center for Biotechnology and Biomedicine, Faculty of Medicine, University of Leipzig, Leipzig, Germany

Concurrent treatment with methotrexate (MTX) and antiepileptic with ALL develop seizures during therapy.5–9 Treatment failure drugs, such as phenobarbital (PB), reduces the efficacy of MTX has been shown to result from intrinsic or acquired MTX chemotherapy in childhood acute lymphoblastic leukemia resistance of tumor cells, including decreased intracellular (ALL). This can result from defective Reduced folate carrier 10 (Rfc1)-dependent cellular uptake of MTX. Indeed, we have MTX accumulation or impaired polyglutamylation. However, shown that functional Rfc1 activity is significantly reduced by no mechanistic information was as yet available how anti- clinically relevant concentrations of the anticonvulsant drugs convulsants compromise the efficacy of MTX chemotherapy. PB or carbamazepine in an adequate in vitro model. As PB is Reduction of Reduced Folate Carrier/Reduced folate carrier known to regulate carrier-associated transport by the nuclear (RFC/Rfc1; SLC19A1/Slc19a1)-dependent MTX influx in the receptor constitutive androstane receptor (CAR), we investi- gated the involvement of the CAR signaling cascade and the presence of PB or CBZ may be an important mode of acquired mode of PB-induced downregulation of Rfc1 activity. CAR MTX resistance as RFC/Rfc1 provides the major route for cellular activation by PB or the CAR agonist 1,4-bis[2-(3,5-dichloro- uptake of MTX in various tissues and a number of tumor pyridyloxy)]-benzene resulted in translocation of Ca2 þ -depen- cells,1,11,12 including leukemic blasts.13,14 Actually, in pre- dent protein kinase Ca (cPKCa) to the plasma membrane related liminary in vivo and in vitro studies on the involvement of Rfc1 to significantly elevated PKC activities. In contrast, subcellular in the PB-induced reduction of intracellular MTX levels, we localization of Ca2 þ -independent PKCd was only marginally have shown that clinically relevant concentrations of PB or CBZ altered. Studies on intracellular distribution of the Rfc1 protein 11,15 indicated that PB-induced activation of cPKCa was associated caused a significant decrease in Rfc1 uptake activity. with carrier internalization from the plasma membrane into the For the regulation of carrier-associated drug transport by PB- cytosol independent of the Rfc1 phosphorylation status. In type inducers such as PB or CBZ, several studies demonstrated conclusion, we identified for the first time the molecular that the MTX efflux carrier multidrug resistance-associated mechanism of this clinically relevant drug resistance in patients protein 2 (MRP2/Mrp2; ABCC2/Abcc2)16 is coordinately regu- with ALL concurrently receiving MTX chemotherapy and antiepileptic drugs. lated with phase I and II xenobiotic metabolizing enzymes by Leukemia (2009) 23, 1087–1097; doi:10.1038/leu.2009.6; the nuclear receptor constitutive androstane receptor (CAR, 17 published online 12 February 2009 NR1I3). CAR is typically stimulated either directly by the Keywords: drug resistance; methotrexate; antiepileptic drugs; pesticide contaminant 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene reduced folate carrier; transport; regulation (TCPOBOP)18 or allusively by PB-type cytochrome P450 (CYP450) inducers19 associated with a poorly characterized signaling cascade that involves several phosphorylation/ dephosphorylation steps resulting in regulation of target gene expression. Up to now, the expression of CAR has been shown in human peripheral blood mononuclear cells and CD4 þ Introduction 20,21 T cells, respectively. Our previous results suggested that the CAR signaling cascade is also involved in the regulation of Rfc1 The folate antagonist methotrexate (MTX; amethopterin) is an activity as TCPOBOP induced a significant reduction in Rfc1- essential component in the treatment of malignancies, including 15 dependent MTX uptake similar to PB. However, this reduction leukemia, lymphoma or head and neck cancers.1 Interestingly, in carrier activity was not due to altered gene expression or clinical studies suggest that concomitant treatment with MTX decreased Rfc1 protein levels. and the antiepileptic drugs, such as phenobarbital (PB) and The presence of putative protein kinase C (PKC) phosphoryla- carbamazepine (CBZ), reduces the efficacy of MTX chemo- 11,22 tion sites in the RFC/Rfc1 protein suggested that the carrier therapy in childhood acute lymphoblastic leukemia (ALL)2 and may be subject to phosphorylation-induced functional regula- in patients with high-grade glioma (HGG).3 Seizures are a tion. Actually, our preliminary data indicated a PKC-dependent common complication in cancer patients that require ongoing regulation of Rfc1 activity on the posttranscriptional level as the treatment with anticonvulsant agents.4 In ALL and brain tumors, PB-induced downregulation of Rfc1-mediated MTX uptake was the commonest malignant disorder and solid tumor in children, almost completely reversed in the presence of the specific PKC up to 50% of patients with brain tumor and one-fifth of patients 15 inhibitor bisindolylmaleimide (BIM). The PKC family of serine/threonine protein kinases is grouped into classical, Correspondence: Dr S Halwachs, Faculty of Veterinary Medicine, Ca2 þ -dependent (cPKC: a, bI, bII, g), novel, Ca2 þ -independent Institute of Pharmacology, Pharmacy and Toxicology, University of (nPKC: d, e, Z, y, m) and atypical (aPKC: z, i/l) PKC isoforms Leipzig, An den Tierkliniken 15, 04103 Leipzig, Germany. 23 E-mail: [email protected] according to differences in their domain composition. In Received 27 August 2008; revised 11 December 2008; accepted 7 leukemic blasts expression of PKCa, b, g, d, e, y and z has been January 2009; published online 12 February 2009 verified so far.24–27 Recent studies demonstrated that PKC is Resistance to MTX chemotherapy by antiepileptic drugs S Halwachs et al 1088 involved in short-term regulation of MTX efflux transporter protein and 30 min compared to untreated control cells. The MRP2 through induction of carrier retrieval from the plasma protein concentration was determined using the bicinchoninic membrane into the cytosol associated with a decreased MRP2 (BCA) protein assay (Pierce, Rockford, IL, USA). transport activity.28 However, no information is yet available on To investigate the influence of PB-type CYP450 inducers and the molecular mechanism of PKC-mediated regulation of the PKC modulators on Rfc1 uptake activity, HPCT-1E3 cells were influx carrier Rfc1 by PB-type CYP450 inducers. treated with PB (430 mM), the CAR agonist TCPOBOP (2.5 nM) In this study, we demonstrate that treatment with the over 48 h or with 1 mM of the PKC inductor phorbol 12-myristate antiepileptic drugs PB or CBZ results in activation of Ca2 þ - 13-acetate (PMA; Calbiochem, Bad Soden, Germany) and/or dependent cPKC isoforms associated with a redistribution of 50 nM PKC inhibitor BIM (Calbiochem) for 2 h. Respective Rfc1 protein from the plasma membrane into the cytosol by a concentrations of the PB-type inducers and PKC modulators mechanism linked to the CAR signaling cascade. Hence, the were chosen as described previously.15 Measurements of reduction of intracellular MTX levels associated with a reduced Rfc1-mediated [3H] methotrexate ([3H]-MTX; specific activity: efficacy of cancer chemotherapy is due to a decrease in the 6.18 Â 108 kBq/mmol; Moravek Biochemicals, Brea, CA, USA) amount of the transmembrane protein Rfc1 but appears to be uptake in HPCT-1E3 cells or HPCT-1E3-rRfc1-EGFP cells were independent of the Rfc1 phosphorylation state. carried out as recently delineated.15 In brief, cells were harvested and washed three times in Tyrode buffer or sodium- 3 5 free choline-Tyrode buffer. [ H]-MTX (2 Â 10 dpm/ml in 5 mM Materials and methods nonlabeled MTX) was added to 1.2 ml cell suspension and aliquots (100 ml) were removed at different time points (15, 45, Reagents 75, 180, 300, 600 and 1200 s) following separation of cells by All chemicals including media and supplements were obtained centrifugation through a silicon oil cushion. The pelleted cells from Sigma-Aldrich (Deisenhofen, Germany) unless stated were dissolved in 4 M KOH overnight and the cell-associated otherwise. radioactivity was measured by liquid scintillation counting (LS 6500; Beckman, Fullerton, CA, USA). Protein concentration of the cell suspension was measured using the BCA protein assay. Cell culture Functional uptake activity of Rfc1 is sodium dependent at The rat hepatocytoma cell line HPCT-1E3 was donated by neutral pH.11,30 Hence, the intracellular accumulation of the E Petzinger (Justus-Liebig-University, Giessen, Germany) and Rfc1 marker substrate MTX was initially measured in Tyrode maintained in Dulbecco’s modified Eagle’s medium (DMEM; buffer containing 137 mM sodium as the overall (sodium- PAA, Coelbe, Germany) containing 10% (v/v) fetal calf serum dependent and sodium-independent) MTX uptake or in (FCS; Gibco, Karlsruhe, Germany), 2 mM glutamine (PAA), sodium-free choline-Tyrode buffer representing the sodium- 10 mg/ml insulin, 10 mg/ml inosine and 1.5 mM dexamethasone independent MTX accumulation rate. Subsequently, Rfc1 as previously published.15 activity was defined indirectly as the difference of [3H]-MTX uptake over 20 min in the presence or absence of extracellular Generation of HPCT-1E3 cells expressing rRfc1-EGFP sodium. Finally, sodium- and time-dependent Rfc1 activity was For transfection HPCT-1E3 cells were grown on 94-mm cell expressed in pmol per mg protein. culture dishes until they displayed a subconfluent phenotype. To obtain the Rfc1-mediated MTX influx velocity (Vinflux), a Subsequently, 8 Â 106 cells were collected by trypsinization, curve fitting was performed by nonlinear regression for the total resuspended in HBS buffer (21 mM 4-(2-hydroxyethyl)-1-piper- cellular MTX uptake in the presence of extracellular sodium or azineethanesulfonic acid (pH 7.5), 137 mM NaCl, 5 mM KCl, the sodium-independent uptake of MTX. The curve fitting was carried out using a mathematical model applicable to describe 0.7 mM Na2HPO4,6mM glucose, 0.5 mg/ml salmon sperm DNA) and electroporated in the presence of 20 mg linearized the uptake of MTX in HPCT-1E3 cells. The respective parameters pEGFP-RK-Rfc129 at 170 V and 950 mF (Bio-Rad Gene pulser). to calculate the Vinflux-values were obtained by means of After incubation for 10 min at room temperature, the HPCT-1E3 SigmaPlot 10 (Systat Software Inc., San Jose, CA, USA). cells were resuspended in culture medium supplemented with Afterward, the Vinflux-values for the sodium-dependent Rfc1- 20% FCS and were split in 94-mm cell culture dishes. After 4 h mediated MTX uptake were calculated as the difference in the cells were washed with medium and incubated for 36 h. After 21 Vinflux-values for the overall MTX uptake and the sodium- days of selection in culture medium containing 800 mg/ml G418 independent MTX accumulation and expressed in pmol per mg (PAA) resistant colonies were picked and grown individually for protein and minute. Finally, we investigated whether there are analyzing positive clones by immunofluorescence. significant differences in the Rfc1-mediated MTX influx (Vinflux) after pretreatment with PB-type CYP450 inducers or PKC modulators compared to unstimulated control cells. Uptake assays in HPCT-1E3 cells and HPCT-1E3 cells expressing rRfc1-EGFP To determine the time-dependent influence of PB on Rfc1- Pentoxyresorufin O-deethylase assay mediated intracellular MTX accumulation, cells were cultured HPCT-1E3 cells were cultured on 24-well plates and pretreated on 96-well plates and incubated with PB (130 mM) for 6, 12, 24 with PB (0.04, 0.43, 1.08 and 2.15 mM) or TCPOBOP (0.25, 2.5, or 48 h. Measurements of Rfc1-mediated fluoresceinated 25 and 250 nM) for 48 h. CYP2B1 activity was determined using methotrexate (FMTX) uptake in HPCT-1E3 cells were carried a modified method described previously.31 In brief, cells were 15 out as described previously. In brief, 10 mM FMTX was added incubated with 15 mM pentoxyresorufin O-deethylase (PROD) for 30 min and cells were lysed with 1% SDS in phosphate- and 10 mM dicumarol in DMEM supplemented with 10 nM buffered saline (PBS) after two washing steps with ice-cold dexamethasone for 2 h at 37 1C. Subsequently, 2 Â 90 ml aliquots incubation buffer. Rfc1-mediated sodium-dependent intracellu- of medium was withdrawn from each well and transferred to a lar FMTX accumulation was assayed in a fluorescence micro- 96-well plate. After addition of 30 ml 0.1 M sodium acetate plate reader (485/535 nm) and expressed in % pmol per mg buffer, pH 4.5, containing 15 Fishman units of b-glucuronidase

Leukemia Resistance to MTX chemotherapy by antiepileptic drugs S Halwachs et al 1089 and 120 Roy units of arylsulfatase (Roche Applied Science, supernatant representing the cytosolic fraction, the pellet was Mannheim, Germany) supernatants were incubated in a shaking washed with homogenization buffer and resuspended in a buffer incubator at 37 1C to allow hydrolysis of potential resorufin containing 0.5% (v/v) Triton X-100. The resulting suspension was conjugates. After 2 h, 240 ml of ethanol was added to each well incubated on ice for 45 min and then centrifuged at 50 000 g (4 1C) and plates were centrifuged at 850 g for 10 min. Formation of for 30 min. The resulting supernatant was used to determine resorufin was measured in a fluorescence microplate reader membrane-associated PKC protein levels in HPCT-1E3 cells. The (540/595 nm; Tecan, Crailsheim, Germany). Cells were lysed protein concentration present in cytosol and membrane fractions with 200 ml 1% SDS in PBS and the protein concentration was was measured using the BCA protein assay. Subsequently, equal determined using the BCA protein assay. Formation of resorufin protein amounts of the particular fractions were separated on in HPCT-1E3 cells was expressed relatively to the control level SDS/10% polyacrylamide gel and blotted onto nitrocellulose as % pmol per mg protein and minute. Background fluorescence membrane. Ca2 þ -dependent cPKC isoforms and Ca2 þ -indepen- was considered by incubation of cells with 15 mM PROD and dent nPKC isoforms were detected using a specific monoclonal 10 mM dicumarol in DMEM supplemented with 10 nM dexa- anti-cPKCa antibody (1:1000; BD Biosciences, Heidelberg, methasone for 5 min at 37 1C. A standard curve of resorufin was Germany) and a polyclonal anti-nPKCd antibody (1:1000; prepared and standards were processed as described for generous gift from D Eisinger, University of Munich, Germany), samples. respectively. cPKC-antibody complexes were visualized with alkaline-phosphatase (AP)-conjugated goat anti-mouse IgG (1:7000; Promega) and nPKC-antibody complexes were demon- PKC activity assay strated with AP-conjugated goat anti-rabbit IgG (1:2000; Dako- HPCT-1E3 cells were treated with PB (430 mM) for 12, 24 or 48 h. Cytomation, Hamburg, Germany). Densitometric quantification of Besides, cells were incubated with PB (430 mM; 48 h), TCPOBOP membrane-bound cPKCa was performed with 1D Image Analysis (2.5 nM; 48 h) or with 1 mM PMA (2 h) in the presence or absence Software (GeneTools, Syngene, Cambridge, UK). The amount of of the PKC inhibitors BIM (50 nM)orGo¨6976 (200 nM; membrane-bound cPKCa wasexpressedasthex-fold change Calbiochem) for 2 h. HPCT-1E3 cells (1 Â 107) were harvested compared to untreated control cells. and dissolved in hypotonic lysis buffer (25 mM Tris, pH 7.4, 1 Â protease inhibitor cocktail). After elimination of cell debris and 1 the nuclear fraction by centrifugation at 1000 g for 5 min (4 C), Western blot analysis of HPCT-1E3 cells expressing membrane and cytosol fractions of HPCT-1E3 cells were rRfc1-EGFP separated by centrifugation at 50 000 g (4 1C) for 30 min. HPCT-1E3 cells expressing rRfc1-EGFP were cultured until Subsequently, the supernatant representing the cytosolic frac- subconfluence. Cells were washed with ice-cold PBS and tion was collected and the pellet representing the membrane incubated in lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM fraction was washed in lysis buffer following resuspension in NaCl, 1% Nonidet P-40, 0.5% sodium desoxycholate, 2 mM buffer containing 25 mM Tris, pH 7.4, 0.05% (v/v) Triton X-100, sodium vanadate, 1 Â protease inhibitor cocktail) for 60 min on 0.5 mM EDTA, 0.5 mM EGTA and 10 mM b-mercaptoethanol. ice by gentle rocking. Cells were scraped and collected in a Cytosolic and membrane-associated PKC activity was deter- microcentrifuge tube. Nuclei and cell debris were pelleted at mined using the PepTag Non-radioactive PKC Assay (Promega, 1000 g for 5 min at 4 1C. Proteins were separated on SDS/10% Madison, WI, USA) according to the manufacturer’s protocol. polyacrylamide gel and blotted onto nitrocellulose membrane. PKC activity was measured in a fluorescence microplate reader Unspecific binding was blocked with 3% bovine serum albumin (485/535 nm) as the amount of phosphorylated peptide substrate (BSA) in TBS-Tween (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, resolved in 0.8% Tris-HCl agarose (pH 8.0) and expressed 0.05% (v/v) Tween 20) for 1 h at room temperature. Membranes relatively to untreated control cells in % units per 107 HPCT-1E3 were probed with a mixture of two monoclonal IgG1 anti-GFP cells and 30 min. The linear relationship between phosphoryla- antibodies (1:1000; Roche Applied Science) following visuali- tion and PKC activity was verified using a standard curve. zation with AP-conjugated goat anti-mouse IgG (1:7000). Standards were processed as described for samples. Protein samples of native HPCT-1E3 cells probed with anti- GFP-antibodies served as a negative control. Western blot analysis of cPKCa and nPKCd in HPCT-1E3 cells To investigate the influence of PB-type CYP450 inducers and Immunocytochemical analysis PKC modulators on translocation of cPKC and nPKC isoforms, For indirect immunolocalization of Rfc1, HPCT-1E3 cells HPCT-1E3 cells were treated with PB (430 mM) or TCPOBOP expressing rRfc1-EGFP were cultured on glass coverslips and (2.5 nM) for 48 h in the presence or absence of the PKC inhibitors treated with PB (430 mM; 48 h), TCPOBOP (2.5 nM; 48 h) or with BIM (50 nM)orGo¨6976 (200 nM) for 2 h. Incubation with the 1 mM PMA (2 h) in the presence or absence of the cPKC inhibitor prototypical PKC inducer PMA (1 mM; 2 h) in the presence or Go¨6976 (200 nM) for 2 h. Cells were fixed with 2% (w/v) absence of BIM or Go¨6976 served as a positive control for the paraformaldehyde (10 min at 20 1C) and permeabilized with involvement of PKC in the regulation of Rfc1 by PB-type 0.1% (v/v) Triton X-100. Unspecific binding was blocked with inducers. Cytosolic and membrane-associated PKC were deter- 3% (v/v) BSA in PBS. Nuclear staining was achieved by addition mined using a modified method described previously.32 Cells of 4,6-diamidino-2-phenyl indole (DAPI; 0.5 mg/ml) and the were scraped in homogenization buffer (25 mM Tris-HCl, pH trans-Golgi network (TGN) was visualized by means of a 7.4, 0.5 mM EDTA, 0.5 mM EGTA, 2 mM sodium vanadate, 1 Â specific polyclonal antibody to TGN38 (1:200) following protease inhibitor cocktail) and the cell suspension was lysed by incubation with Texas-red-conjugated goat anti-rabbit IgG ultrasound. After elimination of cell debris by centrifugation at (1:400; Vector Laboratories, AXXORA, Lo¨rrach, Germany). 1000 g for 5 min (4 1C), membrane and cytosol fractions of Finally, the specimens were washed and mounted on slides. HPCT1E3 cells were separated by centrifugation of the cell Control incubations of TNG were performed by omission of the suspension for 30 min at 50 000 g (4 1C). After collection of the primary anti-TGN38 antibody. Confocal images of HPCT-1E3

Leukemia Resistance to MTX chemotherapy by antiepileptic drugs S Halwachs et al 1090 cells expressing rRfc1-EGFP were taken by sequential scanning Standard deviation of sodium-dependent [3H]-MTX uptake of optical sections of about 1 mm thickness. was determined as the pooled standard deviation (s.e.m.) in For live cell imaging analysis, HPCT-1E3 cells expressing consideration of the sample size. Consecutively, t-values were rRfc1-EGFP were cultured on glass bottom dishes (MatTek obtained using the standardized mean method (dt) to elucidate Corporation, Ashland, MA, USA) for 72 h. Subsequently, dishes significant differences in sodium-dependent MTX uptake after were placed in a small object incubator that provides a constant pretreatment with PB-type inducers, TCPOBOP or PKC temperature of 37 1C and flow of 5% CO2. After addition of modulators in comparison with an untreated control. The culture medium containing PMA (1 mM) with or without BIM critical value of t can be obtained from the t-table (two-sided (50 nM)orGo¨6976 (200 nM), respectively, confocal pictures of test) and is written as tdf,p, where p is the significance level and cells were taken after 0, 60 and 120 min by sequential scanning df the degrees of freedom. of the entire depth of the section in 1 mm z-size steps. Cell nuclei were visualized by addition of Hoechst 33342 (3 mg/ml; AppliChem, Darmstadt, Germany). Results

Time-dependent influence of PB on Rfc1 uptake activity Confocal laser scanning microscopy To determine the time-dependent influence of PB on Rfc1 influx Cell samples were analyzed using the inverted confocal laser activity, HPCT-1E3 cells were incubated with PB for 6, 12, 24 or scanning microscope Leica TCS SP5 equipped with an HCX PL 48 h and the intracellular accumulation of the Rfc1 marker APO 63 Â 1.4 oil immersion objective (Leica Microsystems, substrate FMTX was measured. Pretreatment of cells with PB up Wetzlar, Germany). All images were acquired in the sequential to 24 h resulted only in a slight decrease in MTX accumulation scan mode using an argon laser at 488 nm wavelength (to (Supplementary Figure 1A) whereas incubation with PB for 48 h visualize Rfc1-bound EGFP), a helium-neon laser at 594 nm (to significantly reduced FMTX uptake at about one-third compared visualize Texas Red) or a UV-diode laser at 405 nm (to visualize to an untreated control. On the basis of these results an DAPI or Hoechst 33342). Image stacks were processed and incubation period of 48 h was chosen for all subsequent Rfc1 analyzed with the LAS AF1.7.0 software (Leica Microsystems) regulation studies. and Adobe Photoshop 6.0.

Immunoprecipitation Involvement of PKC on PB- and TCPOBOP-induced HPCT-1E3 cells expressing rRfc1-EGFP were incubated with reduction of Rfc1-dependent MTX uptake 1 mM PMA and 100 nM okadaic acid for 2 h. Subsequently, cells As shown previously, repression of Rfc1-dependent MTX uptake by the PB-type CYP450 inducers PB or CBZ, the CAR agonist were washed with ice-cold PBS und incubated in lysis buffer 15 (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, TCPOBOP and PKC activators exhibit a similar pattern. 0.5% sodium desoxycholate, 2 mM sodium vanadate, 1 Â Therefore, we aimed to determine the involvement of PKC in protease inhibitor cocktail) for 60 min on ice by gentle rocking. regulation of Rfc1 activity by these CYP450 inducers by Cells were scraped and collected in a microcentrifuge tube. examining the potential effect of the specific PKC inhibitor Nuclei and cell debris were pelleted at 1000 g for 5 min at 4 1C. BIM on reduction of Rfc1-mediated MTX uptake by PB or After incubation of the supernatant with 4 mg anti-GFP antibody TCPOBOP. (Roche Applied Science) on ice for 60 min, 40 ml of protein Hence, Vinflux within the initial uptake interval over 3 min, G-agarose beads (Pierce) were added and the samples were representing the main criteria of Rfc1 activity, was determined incubated at 4 1C over night. After centrifugation at 2000 g at (Figures 1A and B). Incubation of cells with 430 mM PB or 2.5 nM 4 1C for 2 min the pelleted beads were washed twice by rotating TCPOBOP over 48 h was associated with a significant decrease the tubes with 1 ml of lysis buffer for 10 min at 4 1C following in the Rfc1-mediated, sodium-dependent MTX influx velocity (Figure 1, a). Simultaneous treatment (2 h) with BIM, a potent two washing steps with 1 ml of wash buffer (50 mM Tris-HCl, pH 2 þ 2 þ M inhibitor of classical Ca -dependent and novel Ca -inde- 7.4, 0.25 NaCl, 0.1% Nonidet P-40, 0.05% sodium deso- 33 xycholate) and a final wash with 1 ml of distilled water. pendent PKC isoforms, almost completely reversed the Immunocomplexes were resuspended in 80 ml Laemmli buffer repression of Rfc1 influx activity by PB or TCPOBOP (Figure 1, b b). Incubation of cells with PMA, which activates both cPKCs without -mercaptoethanol and incubated for 20 min at room 34 temperature. Samples were loaded on 10% SDS–polyacryla- and nPKCs through binding of their diacylglycerol site, and/or mide gel electrophoresis and transferred onto nitrocellulose with BIM served as a validation for the involvement of PKC in membranes. Membranes were probed with an anti-phos- regulation of Rfc1 activity by the antiepileptic drug PB. phoserine antibody (1:200; Qiagen, Hilden, Germany) follow- The addition of PMA resulted in a significant decrease in ing visualization with AP-conjugated goat anti-mouse IgG the Rfc1-mediated MTX influx velocity (Figure 1, c). This (1:7000). The amount of phosphorylated Rfc1 was corrected to reduction was totally abolished by concurrent treatment with the amount of immunoprecipitated Rfc1. BIM (Figure 1 , d). The initial velocity of the sodium-independent MTX accumu- lation rate, which is particularly due to the influx activity of 15,35 Statistical analysis import carrier organic anion transporter 2 (Oat2, Slc22a7), Statistical analysis of all data was carried out using Microsoft was generally not significantly affected (Figure 1). Excel software (Office 2000). Curve fittings were performed by means of Prism 3.0 (GraphPad Software Inc., San Diego, CA, USA). Student’s t-test was used to determine the significance of Induction of CYP2B1 activity by PB-type CYP450 differences between respective test series after elimination of inducers outliers with a G-test. Statistical significance was assumed at The HPCT-1E3 hybrid cell line provides a suitable in vitro model P-values of p0.05. for the study of Rfc1 regulation by PB-type inducers as it possess

Leukemia Resistance to MTX chemotherapy by antiepileptic drugs S Halwachs et al 1091 A PB To investigate the time-dependent influence of PB-type * inducers on PKC activity in the membrane fraction or in the +Na+ cytosol, subcellular fractions of HPCT-1E3 cells were obtained and enzyme activity was measured as described under Materials 8 * -Na+ and methods section. Pretreatment of cells with PB for 24, 48 or 72 h resulted in a time-dependent significant increase in 6 membrane-associated PKC activity compared to unstimulated

) control cells (Supplementary Figure 1B). In this context, induction of PKC activity in the membrane fraction was most influx 4 pronounced after pretreatment of cells with PB for 48 h and (V reached a plateau phase when PB was added for a longer

MTX influx rate incubation period as there was no significant difference in 2 membrane-associated PKC activity after treatment of cells with PB for 48 or 72 h. In contrast, the enzyme activity in the cytosol was considerably decreased after treatment of cells with PB over 24, 48 or 72 h. Control a b c d Consecutively, we determined the effect of PB-type inducers and PKC modulators on cytosolic or membrane-associated PKC B TCPOBOP + +Na activity. As shown in Table 1, membrane-associated PKC activity was significantly induced after incubation with PB for -Na+ * 48 h by on average 4.5-fold compared to untreated control cells. 8 ** A similar effect was observed after 48 h with CAR agonist TCPOBOP or after treatment with PKC inducer PMA (2 h), ) 6 which led to a significant increase in membrane-associated

influx enzyme activity by on average 4- and 3.4-fold, respectively. This

(V induction mediated by PB, TCPOBOP or PMA was completely 4 abolished in the presence of the PKC inhibitor BIM (2 h). MTX influx rate Concurrent treatment of cells (2 h) with the carbazole derivate 2 Go¨6976, a selective inhibitor of Ca2 þ -dependent cPKC iso- forms,39 was related with a significant inhibition of PB-induced enzyme activity by B62% whereas the cPKC inhibitor Control a b c d completely diminished the TCPOBOP-mediated induction of Figure 1 Effect of phenobarbital (PB)-type cytochrome P450 membrane-associated PKC activity. In the presence of Go¨6976 B (CYP450) inducers and modulators of protein kinase C (PKC) activity PMA-induced enzyme activity was inhibited by 44%. on sodium-dependent, Rfc1-mediated and sodium-independent MTX In relation to an untreated control, PKC activity in the influx velocity (Vinflux). (A) HPCT-1E3 cells were pre-treated with 430 cytosolic fraction was significantly reduced after treatment mM PB (a) over 48 h and/or BIM (b) for 2 h and with 1 mM PMA (c) in with PB, TCPOBOP or PMA by on average 63% (Table 1). In the presence or absence of BIM (d) for 2 h. (B) HPCT-1E3 cells were the presence of BIM, the PB-mediated reduction of cytosolic pre-treated with 2.5 nM TCPOBOP (a) over 48 h and/or BIM (b) for 2 h and with 1 mM PMA (c) in the presence or absence of BIM (d) for 2 h. enzyme activity was completely abolished whereas the decrease Cells were generally harvested and the culture medium was replaced in PKC activity mediated by TCPOBOP or PMA was reversed to by Tyrode buffer or sodium-free Choline-Tyrode buffer and 5 mM MTX B54 and B62%, respectively, by the PKC inhibitor. Concurrent containing 2 Â 105 dpm [3H] MTX was added. Uptake was stopped by treatment of cells with the cPKC inhibitor Go¨6976 not only centrifugation of aliquots through a silicon oil cushion and the inhibited the decline in enzyme activity mediated by PB, incorporated radioactivity was measured by liquid scintillation TCPOBOP or PMA but further associated with a significant counting. Rfc1 activity was defined as the sodium- and time- increase in cytosolic activity of PKC. dependent uptake of MTX (5 mM) over 20 min. Vinflux-values were obtained by non linear regression of sodium-dependent or sodium- independent uptake of MTX in HPCT-1E3 cells and expressed in pmol per mg protein and minute. The results represent the mean±s.e.m. of Effect of PB-type CYP450 inducers and PKC modulators at least triplicate measurements (*Po0.05, **Po0.01). on subcellular distribution of cPKCa and nPKCd To investigate the influence of PB-type inducers or PKC a phase I and II metabolism36 and the endogenous expression of modulators on subcellular distribution of cPKCa and nPKCd, MTX transporters, including Rfc1.37 HPCT-1E3 cells were separated into a membrane (M) and a To investigate the inducibility of the CAR receptor in our in cytosolic (C) fraction as described under Materials and methods vitro model HPCT-1E3, cells were pretreated with various section as the activation of PKC isoforms is associated with their concentrations of PB or CAR agonist TCPOBOP for 48 h. translocation from the cytosol to the plasma membrane.38 Addition of selected concentrations of both substances sig- In untreated control cells, cPKCa (Figure 2) was distributed in nificantly induced CAR-dependent CYP2B1-mediated PROD favor of the cytosol whereas nPKCd was almost equally shared activity by 1.5-fold compared to untreated control cells (Figure 3). Pretreatment of cells with PB or TCPOBOP over 48 h (Supplementary Figure 2). was associated with a significant translocation of cPKCa to the plasma membrane and a reversal of the enzyme distribution in relation to control cells (Figure 2). In the presence of PKC inhibitor BIM (2 h), translocation of cPKCa mediated by PB or Activation of PKC activity by PB-type CYP450 inducers TCPOBOP was completely diminished. The cPKC inhibitor Activation of PKC isoforms is associated with their translocation Go¨6976 reversed the enzyme redistribution mediated by PB or from the cytosol to the plasma membrane.38 PMA to a large part, and TCPOBOP-induced cPKCa

Leukemia Resistance to MTX chemotherapy by antiepileptic drugs S Halwachs et al 1092 Table 1 Effect of PB-type CYP450 inducers and PKC modulators on C M CMM C C M PKC activity in HPCT-1E3 cells 1.00±0.04 1.21±0.04 ** 1.09±0.08 1.10±0.11

Relative protein kinase C 82 kDa activity PB PB Treatment Membrane Cytosol Control PB +BIM +Gö6976 Control 100±26 100±9 Control+1 mM PMA 340±61*** 30±8*** C M C M C M C M Control+1 mM PMA/+50 nM BIM 94±962±6*** 1.00±0.04 1.26±0.07** 1.11±0.10 1.09±0.14 ± ± Control 100 20 100 12 82 kDa Control+431 mM PB 371±36*** 42±9*** Control+431 mM PB/+50 nM BIM 119±30 99±9 TCPOBOP TCPOBOP ± ± Control TCPOBOP Control 100 29 100 12 +BIM +Gö6976 Control+2.5 nM TCPOBOP 309±51*** 33±4*** Control+2.5 nM TCPOBOP/+50 nM BIM 106±554±6*** Figure 2 Effect of phenobarbital (PB), 1,4-bis[2-(3,5-dichloro- pyridyloxy)]benzene (TCPOBOP) and phorbol 12-myristate 13-acetate Control 100±54 100±2 (PMA) on subcellular localization of cPKCa. HPCT-1E3 cells were Control+1 mM PMA 340±9* 44±6** treated with PB (430 mM; 48 h), with TCPOBOP (2.5 nM; 48 h) or with Control+1 mM PMA/+200 nM Go¨ 6976 191±21 161±52 1 mM PMA (2 h) in the presence or absence of the protein kinase C (PKC) inhibitors bisindolylmaleimide (BIM) (50 nM)orGo¨6976 Control 100±17 100±12 (200 nM) for 2 h. Equal protein amounts were separated on an SDS/ Control+431 mM PB 535±140** 42±9*** 10% polyacrylamide gel and cytosol (C) or membrane-associated (M) Control+431 mM PB/+200 nM Go¨ 6976 202±53 145±3** cPKCa proteins were detected by a specific anti-PKCa antibody and are apparent at B82 kDa (-). The western blot shown is representa- Control 100±17 100±11 tive of at least three independent experiments. x-Fold changes in Control+2.5 nM TCPOBOP 477±54*** 32±7*** membrane-bound cPKCa are given from three measurements. Control+2.5 nM TCPOBOP/ 103±24 147±26* Mean±s.d. (**Po0.01). +200 nM Go¨ 6976 Abbreviations: BIM, bisindolylmaleimide; PB, phenobarbital; PMA, phorbol 12-myristate 13-acetate; TCPOBOP, 1,4-bis[2-(3,5-dichloro- CMM CMCMC pyridyloxy)]benzene. Cells were pretreated with PB or TCPOBOP for 48 h as well as PMA 92 (2 h) in the presence or absence of BIM or Go¨ 6976 over 2 h. Consecutively, subcellular fractions were obtained and PKC activity 75 was measured as described under Materials and methods section. Enzyme activity in cytosolic and membrane fractions was expressed relatively to untreated control cells in % units per 107 HPCT-1E3 cells ± and 30 min. The results represent the mean s.d. of quadruplicate Control PB PB PB measurements. +BIM +Gö6976 *Po0.05; **Po0.01; ***Po0.001. C M C M C M C M

translocation was totally abolished. In PMA-treated cells, cPKCa 92 protein was almost entirely detected in the membrane fraction (not shown). In PMA-treated cells, BIM and Go¨6976 (2 h) 75 considerably inhibited cPKCa translocation as almost equal amounts of cPKCa protein was detected in the cytosolic and the membrane fraction. TCPOBOP TCPOBOP Control TCPOBOP In contrast, pretreatment of cells with CAR agonist TCPOBOP +BIM +Gö6976 did not alter the subcellular localization of nPKCd, and the membrane-associated enzyme amount was only slightly in- Figure 3 Effect of phenobarbital (PB), 1,4-bis[2-(3,5-dichloro- creased in PB-induced cells compared to an untreated control pyridyloxy)]benzene (TCPOBOP) and phorbol 12-myristate 13-acetate d (Figure 3). In TCPOBOP-treated cells, co-incubation neither (PMA) on subcellular distribution of nPKC . HPCT-1E3 cells were treated with PB (430 mM; 48 h), with TCPOBOP (2.5 nM; 48 h) or with with BIM nor with Go¨6976 resulted in an altered subcellular 1 mM PMA (2 h) in the presence or absence of the protein kinase C distribution of nPKCd in relation to untreated control cells. In the (PKC) inhibitors bisindolylmaleimide (BIM) (50 nM)orGo¨6976 presence of BIM, redistribution of nPKCd mediated by PB was (200 nM) for 2 h. Equal protein amounts were separated on an SDS/ inhibited to a minor extent whereas specific cPKC inhibitor 10% polyacrylamide gel and cytosol (C) or membrane-associated (M) d nPKCd proteins were detected by a specific anti-PKCd antibody and Go¨6976 had no effect on PB-induced nPKC translocation. B - However, nPKCd translocated to the plasma membrane after are apparent at 78 kDa ( ). The western blot shown is representa- tive of two independent experiments. incubation with the prototypical PKC inducer PMA (not shown). Addition of BIM resulted in a slight inhibition of PMA-induced nPKCd redistribution whereas specific cPKC inhibitor Go¨6976 Effect of PKC inducer PMA and protein phosphatase 2A had no effect on PMA-stimulated nPKCd translocation. Taken inhibitor okadaic acid on the level of Rfc1 together, our data indicate that treatment with the antiepileptic phosphorylation drug PB leads to an induction of the CAR regulatory pathway We have shown previously that treatment with PKC inducer resulting in activation of Ca2 þ -dependent cPKC isoforms. PMA and/or protein phosphatase 2A (PP 2A) inhibitor okadaic

Leukemia Resistance to MTX chemotherapy by antiepileptic drugs S Halwachs et al 1093 acid significantly reduced Rfc1-dependent MTX influx.15 As the of cells with PMA (2 h) and selective cPKC inhibitor Go¨6976 not Rfc1 protein possesses two putative PKC binding sites,11 only abolished the PMA-induced Rfc1 translocation into the reduction of Rfc1 activity after PB-mediated PKC activation cytosolic fraction but also resulted in an increase in membrane- may be due to carrier phosphorylation. To address this issue, associated carrier protein (Figure 5c). Moreover, in the presence HPCT-1E3 cells expressing Rfc1 tagged with a cEGFP epitope of Go¨6976 a considerable amount of Rfc1 was found around were generated (Supplementary Figure 3A). Functional activity cell nuclei. In regard to the congruent pattern of Go¨6976- of Rfc1-EGFP was verified by measuring the sodium- and time- induced Rfc1 distribution and TGN38 immunoreactivity dependent uptake of [3H]-MTX over 20 min using the initial (Figure 4c, Supplementary Figure 4C and 5C), these findings uptake interval over 3 min as an essential criteria for Rfc1 suggest an additional colocalization of Rfc1 and the Golgi activity.15 Initial velocity of rRfc1-EGFP-mediated MTX uptake apparatus in living cells in the presence of the cPKC inhibitor. corresponded two times the functional activity of wild-type HPCT-1E3 cells (Supplementary data Figure 3B). HPCT-1E3 cells expressing rRfc1-EGFP were treated with Discussion 1 mM PMA and 100 nM okadaic acid for 2 h and the level of Rfc1 phosphorylation was determined. Western blot analyses of anti- Recent clinical studies suggest that concurrent long-term anti- GFP immunoprecipitates with anti-GFP antibody served as an convulsant therapy of patients with ALL or HGG with internal control for loading and for normalization. The Rfc1 PB-type CYP450 inducer PB or CBZ may cause a decrease in protein is found slightly phosphorylated in untreated control intracellular MTX levels associated with a reduced efficacy of cells. However, incubation of cells with PKC inducer PMA and cancer chemotherapy and accordingly an aggravated prog- PP 2A inhibitor okadaic acid did not result in elevated levels of nosis.2,3 However, this observation could not be easily carrier phosphorylation compared to unstimulated cells (not attributed to CYP450 enzyme induction as metabolism of MTX shown). is mediated by hepatocellular aldehyde oxidase and folylpo- lyglutamate synthetase.42 Previous studies from our laboratory on the involvement of MTX influx carrier Rfc1 in this clinically Effect of PB-type CYP450 inducers and PKC modulators important drug interaction suggested that therapeutically rele- on Rfc1 localization vant concentrations of PB-type CYP450 inducers cause a It has been recently demonstrated that activation of PKC results significant reduction in Rfc1-mediated MTX uptake by PKC.15 in internalization of MTX import carrier protein OAT140 or MTX In this study, we have therefore systematically investigated the efflux transporter MRP228 from the plasma membrane into the mode of PKC-mediated Rfc1 downregulation by the anti- cytosol associated with a reduction in their transport activity. epileptic drug PB. Hence, we asked whether PB-induced PKC-mediated reduction In line with previous data,15 we observed that CAR agonist of Rfc1-dependent MTX uptake is based on a decrease in the TCPOBOP or PKC inducer PMA caused a significant reduction amount of membrane-associated Rfc1 protein. of Rfc1-mediated MTX uptake similar to PB. This repression was In untreated control cells (Figure 4a, Supplementary Figures almost completely reversed by selective PKC inhibitor BIM. 4A and 5A), Rfc1 was almost completely localized to the plasma Thus, our results indicate the involvement of nuclear receptor membrane. Besides, a small amount was detected in the cytosol CAR and PKC, respectively, in PB-induced reduction of Rfc1 predominately at the TGN. This Golgi apparatus-associated activity. compartment,41 delineated by immunoreactivity against the Several studies have demonstrated that the CAR signaling TGN marker protein TGN38, was generally found around cell cascade comprises a complex network of phosphorylation/ nuclei. Treatment of cells with PB (Figure 4b) or with CAR dephosphorylation events.19,43,44 However, the mechanisms agonist TCPOBOP (Supplementary Figure 4B) for 48 h resulted required to fully activate CAR have not yet been defined.45 To in a significant decrease in membrane-associated Rfc1 and a investigate whether induction of the CAR signaling cascade dispersed allocation of the carrier protein within the cell. The involves activation of PKC, the effect of PB and the CAR agonist effect of PB (Figure 4c) or TCPOBOP (Supplementary Figure 4C) TCPOBOP on enzyme activity and on subcellular distribution of on Rfc1 redistribution was completely inhibited by the specific PKC isoforms was determined. We observed a significant cPKC inhibitor Go¨6976. Moreover, addition of the cPKC induction in membrane-associated along with a simultaneous inhibitor was related to a considerable increase in Golgi- decrease in cytosolic PKC activity by PB and TCPOBOP similar associated Rfc1. HPCT-1E3 cells incubated with PKC inducer to the specific PKC inducer PMA. Our results regarding the PMA demonstrated the most pronounced redistribution of Rfc1 correlation between drug resistance and PKC induction are into the cytosol (Supplementary Figure 5B) in relation to similar to previous data demonstrating a two- and sevenfold pretreatment with PB or TCPOBOP. Go¨6976 almost completely increase of PKC activity in the doxorubicin resistant HL-6046 diminished the PMA-mediated effect on Rfc1 localization and the adriamycin-resistant K562 human leukemia cell line,47 (Supplementary Figure 5C). Moreover, a prominent colocaliza- respectively. tion of Rfc1 and TGN38 was additionally found in the presence As for the effect of PB-type inducers on subcellular distribu- of the cPKC inhibitor. tion of PKC isoforms, our results display that induction of the To investigate the effect of the two PKC inhibitors BIM and CAR signaling pathway by PB or TCPOBOP primarily involves Go¨6976 on PKC-mediated redistribution of Rfc1 in living cells, activation of Ca2 þ -dependent cPKCs as shown by concurrent subcellular Rfc1 allocation was examined in rRfc1-EGFP- incubation with PKC inhibitors BIM or Go¨6976. BIM represents transfected HPCT-1E3 cells cultured on glass bottom dishes. In a potent inhibitor of Ca2 þ -dependent classical PKC and novel untreated cells, Rfc1 is localized almost completely to the Ca2 þ -independent PKC isoforms33 whereas carbazole derivate 2 þ 39 plasma membrane (Figures 5a–c). Incubation of cells with 1 mM Go¨6976 is a selective inhibitor of Ca -dependent cPKCs. In PMA induced a significant translocation of Rfc1 from the plasma contrast to PB-type inducers, phorbol ester PMA stimulates membrane into the cytosol within 2 h (Figure 5a). PKC inhibitor Ca2 þ -dependent cPKCa and Ca2 þ -independent nPKCd sup- BIM totally repressed the PMA-induced redistribution of Rfc1 porting previous observations on substrate specificities of PKC into the cytosol compartment (Figure 5b). Concurrent treatment isozymes.48 Consistently, selective cPKC inhibitor Go¨6976 only

Leukemia Resistance to MTX chemotherapy by antiepileptic drugs S Halwachs et al 1094

Figure 4 Effect of phenobarbital (PB) on subcellular localization of Rfc1 in HPCT-1E3 cells. In untreated control cells (a) Rfc1-EGFP (green) was almost completely localized to the membrane fraction. Besides, a small amount was detected in the cytosol predominately at the Golgi apparatus- associated trans-Golgi network (TGN) delineated by the TGN marker protein TGN38 (red). After pretreatment with 430 mM PB for 48 h, Rfc1 was found in large part in the cell (b). Go¨6976 (200 nM, 2 h) completely inhibited the appearance of Rfc1 in the cytosol and considerable amounts of Rfc1 protein are additionally colocalized with the TGN (c). Colocalization results in yellow/orange. Cell nuclei (blue) were visualized by means of 4,6-diamidino-2-phenyl indole (DAPI) (a–c, left: Rfc1 in green; middle: TGN38 in red; right: overlay; bars, 10 mm).

partially suppressed PMA-induced PKC activity whereas en- CYP450 inducers such as the anticonvulsant agent CBZ or the zyme stimulation by PB-type inducers was almost totally antipsychotic drug chlorpromazine.15 Finally, this would be in abolished. line with the finding that cotreatment with the CAR agonist In consideration of the distinct mode of PKC isoform TCPOBOP and PMA had an additive effect on activation of the activation by PB or TCPOBOP compared to PMA, the human CYP2B6 gene in vitro.50 mechanism of cPKC stimulation by PB-type CYP450 inducers The presence of putative PKC phosphorylation sites on the may be distinct from the known tumor-promoting effect of PB.49 Rfc1 protein11 suggested that downregulation of Rfc1-dependent This assumption is corroborated by the similar repression pattern MTX uptake may be due to carrier phosphorylation associated of Rfc1 activity with other non-tumor-promoting PB-type with a modified substrate affinity. However, our results indicate

Leukemia Resistance to MTX chemotherapy by antiepileptic drugs S Halwachs et al 1095

Figure 5 Protein kinase C (PKC) inhibitor bisindolylmaleimide (BIM) and cPKC inhibitor Go¨6976 inhibit the PMA-induced redistribution of Rfc1. Incubation of Rfc1-EGFP (green) transfected HPCT-1E3 cells with 1 mM PMA-induced translocation of Rfc1 from the plasma membrane into the cytosol within 2 h (a). Concurrent treatment with 50 nM BIM (b) or 200 nM Go¨6976 (c) almost completely inhibited the PMA-induced translocation of Rfc1 into the cytosol. Cell nuclei (blue) were visualized by addition of Hoechst 33342 (a–c, representative cells are shown from three independent experiments; bars, 10 mm).

that cPKC activation by PB-dependent induction of the CAR In consideration of our results and the literature data, we signaling pathway does not result in elevated Rfc1 phosphory- altogether conclude that PB-, TCPOBOP- or PMA-dependent lation levels. activation of cPKC isoforms exerts distinct, cell compartment- PKC-dependent retrieval of Rfc1 from the plasma membrane specific effects. At the Golgi apparatus, cPKC isoforms stimulate might be another explanation for reduction of Rfc1 uptake the generation of Rfc1 transport vesicles associated with an activity by the antiepileptic drug PB. In untreated HPCT-1E3 increased targeting of intracellular Rfc1 molecules from the cells, Rfc1 was almost completely localized to the plasma Golgi compartment to the basolateral membrane. This process membrane. In line with a recent study on the subcellular does not require de novo synthesis of Rfc1 protein as shown distribution of the MTX influx carrier,51 small amounts of Rfc1 previously.15 In contrast, at the plasma membrane cPKCs induce protein additionally colocalized juxtanuclear with the TGN, the the internalization of Rfc1 protein into the cytosol. The retrieval last station of the Golgi apparatus.41 The TGN mediates of RFC/Rfc1 from the plasma membrane may be regulated transport vesicle biogenesis and sorting of transmembrane through Ca2 þ -dependent cPKCs by altering the organization of proteins primarily to the basolateral surface.41 PB, the CAR cytoskeletal proteins56 interacting with integral RFC/Rfc1 agonist TCPOBOP and PKC inducer PMA were all shown to protein domains. This hypothesis is supported by the observa- considerably induce translocation of Rfc1 from the plasma tion of frequent movements of RFC transport vesicles away from membrane into the cytosol. Due to the fact that the PKC or toward the cell surface.51 inhibitor BIM and the specific cPKC inhibitor Go¨6976 similarly Inhibition of PB-, TCPOBOP- or PMA-induced cPKC activity prevented PB-, TCPOBOP- or PMA-induced Rfc1 internaliza- by Go¨6976 results in retention of Rfc1 in the Golgi apparatus. In tion, a cPKC-mediated regulatory mechanism can be assumed. contrast, inhibition of cPKC activity in the membrane compart- Interestingly, addition of the cPKC inhibitor Go¨6976 was ment was related with a localization of Rfc1 protein at the further related to a significant increase in TGN-associated Rfc1. plasma membrane similar to untreated control cells. In this Our data agree with the finding that only slight amounts of the context, the cPKC-dependent induction of Rfc1 membrane efflux carrier bile salt export pump (BSEP; ABCB11) were targeting may represent a compensatory mechanism after cPKC- retained in the Golgi apparatus in untreated isolated rat mediated retrieval of Rfc1 protein from the plasma membrane. hepatocytes whereas addition of a specific PKC inhibitor Taken together it is concluded that pretreatment with PB-type induced a significant increase in TGN-associated BSEP.52 inducers causes induction of the CAR regulatory pathway, Moreover, recent studies showed that treatment of rat basophilic resulting in activation of Ca2 þ -dependent cPKC isoforms that leukemia or HepG2 cells with the PKC inducer PMA stimulated involve internalization of Rfc1 protein from the plasma the formation of transport vesicles at the TGN due to an membrane. A functional clinical significance is suggested by increased binding of cPKCa to Golgi membranes associated the observed tight association between appropriate carrier with cPKCa-dependent phosphorylation of Golgi-associated localization and Rfc1-mediated MTX uptake. In consideration proteins.53–55 These processes were strongly inhibited in the of its wide tissue distribution, including leukemic blasts, PB- presence of specific PKC inhibitors.53,54 induced downregulation of pivotal Rfc1 activity may result in

Leukemia Resistance to MTX chemotherapy by antiepileptic drugs S Halwachs et al 1096 alteration of total MTX pharmacokinetics. In this context, our associated with decreased reduced folate carrier expression. Blood data demonstrate for the first time the molecular mechanism of 1997; 89: 1013–1018. the reduced clinical efficacy of MTX cancer chemotherapy 15 Halwachs S, Kneuer C, Honscha W. Downregulation of the when a combined therapy with the antiepileptic drugs PB and reduced folate carrier transport activity by phenobarbital-type 2 cytochrome P450 inducers and protein kinase C activators. CBZ is administered. In consideration of our results, adminis- Biochem Biophys Acta 2007; 1768: 1671–1679. tration of newer nonenzyme/CAR-inducing antiepileptic drugs 16 Borst P, Evers R, Kool M, Wijnholds J. A family of drug transporters: such as gabapentin or levetiracetam4 may contribute to improve the multidrug resistance-associated proteins. J Natl Cancer Inst the frequently indicated anticonvulsant therapy of cancer 2000; 92: 1295–1302. patients. Finally, provision for drug interactions by regulation 17 Xu C, Li CY, Kong AN. Induction of phase I, II and III drug of carrier-associated MTX transport as another significant metabolism/transport by xenobiotics. Arch Pharm Res 2005; 28: 249–268. prognostic feature can help to increase the cure rate above 18 Tzameli I, Pissios P, Schuetz EG, Moore DD. The xenobiotic 80% in children with ALL. compound 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene is an agonist ligand for the nuclear receptor CAR. Mol Cell Biol 2000; 20: 2951–2958. Acknowledgements 19 Sueyoshi T, Negishi M. Phenobarbital response elements of cytochrome P450 genes and nuclear receptors. Annu Rev We thank C Lakoma for skillful technical assistance. We also Pharmacol Toxicol 2001; 41: 123–143. thank E Petzinger (Justus-Liebig-University, Giessen, Germany) for 20 Siest G, Jeannesson E, Marteau JB, Samara A, Be´range`re M, Pfister donating the HPCT-1E3 cell line and D Eisinger (University of M et al. Transcription factor and drug-metabolizing enzyme gene expression in lymphocytes from healthy human subjects. Drug Munich, Munich, Germany) for generously providing the anti- Metab Dispos 2008; 36: 182–189. PKCd antibody. This work was supported by the Deutsche 21 Hariparsad N. Hepatic and Extra-Hepatic Induction of Drug Forschungsgemeinschaft (HO 2103/2-1). Metabolizing Enzymes and Drug Transporters by Antiretrovirals, in the Presence and Absence of Viral Infection [Dissertation Doctor of Philosophy]. University of Cincinnati: Cincinnati, 2006. 22 Nguyen TT, Dyer DL, Dunnin DD, Rubin SA, Grant KE, Said HM. References Human intestinal folate transport: cloning, expression, and distribution of complementary RNA. Gastroenterology 1997; 1 Matherly LH, Goldman ID. Membrane transport of folates. Vitam 112: 783–791. Horm 2003; 66: 403–456. 23 Carmena D, Sardini A. Lifespan regulation of conventional protein 2 Relling MV, Pui CH, Sandlund JT, Rivera GK, Hancock ML, Boyett kinase C isotypes. Biochem Soc Trans 2007; 35: 1043–1045. JM et al. Adverse effect of anticonvulsants on efficacy of 24 Beck JF, Bru¨gger D, Brischwein K, Liu C, Bader P, Niethammer D chemotherapy for acute lymphoblastic leukaemia. Lancet 2000; et al. Anticancer drug-mediated induction of multidrug resistance- 356: 285–290. associated genes and protein kinase C isozymes in the 3 Riva M, Landonio G, Defanti CA, Siena S. The effect of T-lymphoblastoid cell line CCRF-CEM and in blasts from patients anticonvulsant drugs on blood levels of methotrexate. J Neuro- with acute lymphoblastic leukemias. Jpn J Cancer Res 2001; 92: oncol 2000; 48: 249–250. 896–903. 4 Vecht CJ, Wagner GL, Wilms EB. Interactions between anti- 25 Balasubramanian N, Advani SH, Zingde SM. Protein kinase C epileptic and chemotherapeutic drugs. Semin Oncol 2003; 30: 49–52. isoforms in normal and leukemic neutrophils: altered levels in 5 Ochs JJ, Bowman WP, Pui CH, Abromowitch M, Mason C, Simone leukemic neutrophils and changes during myeloid maturation in JV. Seizures in childhood lymphoblastic leukaemia patients. chronic myeloid leukemia. Leuk Res 2002; 26: 67–81. Lancet 1984; 2: 1422–1424. 26 Alkan S, Huang Q, Ergin M, Denning MF, Nand S, Maududi T et al. 6 Maytal J, Grossman R, Yusuf FH, Shende AC, Karayalycin G, Survival role of protein kinase C (PKC) in chronic lymphocytic Lanzkowsky P et al. Prognosis and treatment of seizures in children leukemia and determination of isoform expression pattern with acute lymphoblastic leukemia. Epilepsia 1995; 36: 831–836. and genes altered by PKC inhibition. Am J Hematol 2005; 79: 7 Winick NJ, Bowman WP, Kamen BA, Roach ES, Rollins N, 97–106. Jacaruso D et al. Unexpected acute neurologic toxicity in the 27 Jiffar T, Kurinna S, Suck G, Carlson-Bremer D, Ricciardi MR, treatment of children with acute lymphoblastic leukemia. J Natl Konopleva M et al. PKC alpha mediates chemoresistance in acute Cancer Inst 1992; 84: 252–256. lymphoblastic leukemia through effects on Bcl2 phosphorylation. 8 Mahoney Jr DH, Shuster JJ, Nitschke R, Lauer S, Steuber CP, Leukemia 2004; 183: 505–512. Camitta B. Intensification with intermediate-dose intravenous 28 Kubitz R, Huth C, Schmitt M, Horbach A, Kullak-Ublick G, methotrexate is effective therapy for children with lower-risk Ha¨ussinger D. Protein kinase C-dependent distribution of the B-precursor acute lymphoblastic leukemia: a Pediatric Oncology multidrug resistance protein 2 from the canalicular to the Group study. J Clin Oncol 2000; 18: 1285–1294. basolateral membrane in human HepG2 cells. Hepatology 2001; 9 Pace A, Bove L, Innocenti P, Pietrangeli A, Carapella CM, Oppido 34: 340–350. P et al. Epilepsy and gliomas: incidence and treatment in 119 29 Schrader S, Kneuer C, Honscha W. Alternative transcripts of rat patients. J Exp Clin Cancer Res 1998; 17: 479–482. slc19a1: defective sorting and inefficient expression of alternative 10 Kamen B. Folate and antifolate pharmacology. Semin Oncol 1997; splicing productsFan addendum. DNA Seq 2006; 16: 1–6. 24: 30–39. 30 Kneuer C, Honscha W. The H(+)-dependent reduced folate carrier 11 Honscha W, Do¨tsch KU, Thomsen N, Petzinger E. Cloning and 1 of humans and the sodium-dependent methotrexate carrier-1 of functional characterization of the bile acid-sensitive methotrexate the rat are orthologs. FEBS Lett 2004; 566: 83–86. carrier from rat liver cells. Hepatology 2000; 31: 1296–1304. 31 Donato MT, Gomez-Lechon MJ, Castell JV. A microassay for 12 Kneuer C, Honscha KU, Honscha W. Rat reduced-folate carrier-1 measuring cytochrome P450IA1 and P450IIB1 activities in intact is localized basolaterally in MDCK kidney epithelial cells and human and rat hepatocytes cultured on 96-well plates. Anal contributes to the secretory transport of methotrexate and Biochem 1993; 213: 29–33. fluoresceinated methotrexate. Cell Tissue Res 2005; 320: 517–524. 32 Hodges NJ, Orton TC, Strain AJ, Chipman JK. Potentiation of 13 Zhang L, Taub JW, Williamson M, Wong SC, Hukku B, Pullen J epidermal growth factor- induced DNA synthesis in rat hepato- et al. Reduced folate carrier gene expression in childhood acute cytes by phenobarbitone: possible involvement of oxidative stress lymphoblastic leukemia: relationship to immunophenotype and and kinase activation. Carcinogenesis 2000; 21: 2041–2047. ploidy. Clin Cancer Res 1998; 4: 2169–2177. 33 Gekeler V, Boer R, Uberall F, Ise W, Schubert C, Utz I et al. Effects 14 Gorlick R, Goker E, Trippett T, Steinherz P, Elisseyeff Y, Mazumdar of the selective bisindolylmaleimide protein kinase C inhibitor GF M et al. Defective transport is a common mechanism of acquired 109203X on P-glycoprotein-mediated multidrug resistance. Br J methotrexate resistance in acute lymphocytic leukemia and is Cancer 1996; 74: 897–905.

Leukemia Resistance to MTX chemotherapy by antiepileptic drugs S Halwachs et al 1097 34 Adams PD, Parker PJ. TPA-induced activation of MAP kinase. 45 Goodwin B, Moore JT. CAR: detailing new models. Trends FEBS Lett 1991; 290: 77–82. Pharmacol Sci 2004; 25: 437–441. 35 Sekine T, Cha SH, Tsuda M, Apiwattanakul N, Nakajima N, 46 Aquino A, Hartman K, Knode M, Grant S, Huang K, Niu C et al. Role Kanai Y et al. Identification of multispecific organic anion of protein kinase C in phosphorylation of vinculin in adriamycin- transporter 2 expressed predominantly in the liver. FEBS Lett resistant HL-60 leukemic cells. Cancer Res 1988; 48: 3324–3329. 1998; 429: 179–182. 47 Sugawara S, Kumano N, Saijo S, Numata Y, Sato G, Motomiya M. 36 Utesch D, Arand M, Thomas H, Petzinger E, Oesch F. Xenobiotic- Protein kinase C activity in human leukemia cell lines with metabolizing enzyme activities in hybrid cell lines established by reference to sensitivity to antineoplastic agents. Tohoku J Exp Med fusion of primary rat liver parenchymal cells with hepatoma cells. 1992; 168: 393–396. Xenobiotica 1992; 22: 1451–1457. 48 Kazanietz MG, Areces LB, Bahador A, Mischak H, Goodnight J, 37 Halwachs S, Kneuer C, Honscha W. Endogenous expression of Mushinski JF et al. Characterization of ligand and substrate liver-specific drug transporters for organic anions in the rat specificity for the calcium-dependent and calcium-independent hepatocytoma fusion cell line HPCT-1E3. Eur J Cell Biol 2005; protein kinase C isozymes. Mol Pharmacol 1993; 44: 298–307. 84: 677–686. 49 Jirtle RL, Meyer SA, Brockenbrough JS. Liver tumor promoter 38 Newton AC. Protein kinase C: structural and spatial regulation by phenobarbital: a biphasic modulator of hepatocyte proliferation. phosphorylation, cofactors, and macromolecular interactions. Prog Clin Biol Res 1991; 369: 209–216. Chem Rev 2001; 101: 2353–2364. 50 Inoue K, Negishi M. Nuclear receptor CAR requires early growth 39 Martiny-Baron G, Kazanietzs MG, Mischak H, Blumberg PM, response 1 to activate the human cytochrome P450 2B6 gene. Kochs G, Hug H et al. Selective inhibition of protein kinase C J Biol Chem 2008; 283: 10425–10432. isozymes by the indolocarbazole Go¨6976. J Biol Chem 1993; 268: 51 Marchant JS, Subramanian VS, Parker I, Said HM. Intracellular 9194–9197. trafficking and membrane targeting mechanisms of the human 40 Wolff NA, Thies K, Kuhnke N, Reid G, Friedrich B, Lang F et al. reduced folate carrier in mammalian epithelial cells. J Biol Chem Protein kinase C activation downregulates human organic anion 2002; 277: 33325–33333. transporter 1-mediated transport through carrier internalization. 52 Kubitz R, Su¨tfels G, Ku¨hlkamp T, Ko¨lling R, Ha¨ussinger D. J Am Soc Nephrol 2003; 8: 1959–1968. Trafficking of the bile salt export pump from the Golgi to the 41 Traub LM, Kornfeld S. The trans-Golgi network: a late secretory canalicular membrane is regulated by the p38 MAP kinase. sorting station. Curr Opin Cell Biol 1997; 9: 527–533. Gastroenterology 2004; 126: 541–553. 42 Josting A, Diehl V, Engert A. Methotrexat: Pharmakologie, 53 Westermann P, Knoblich M, Maier O, Lindschau C, Haller H. Toxizita¨t und Rescuemassnahmen bei hochdosierter Behandlung. Protein kinase C bound to the Golgi apparatus supports the Der Onkologe 1998; 4: 974–985. formation of constitutive transport vesicles. Biochem J 1996; 320: 43 Joannard F, Galisteo M, Corcos L, Guillouzo A, Lagadic-Gossmann 651–658. D. Regulation of phenobarbital-induction of CYP2B and CYP3A 54 Radau B, Otto A, Mu¨ller EC, Westermann P. Protein kinase C genes in rat cultured hepatocytes: involvement of several serine/ alpha-dependent phosphorylation of Golgi proteins. Electro- threonine protein kinases and phosphatases. Cell Biol Toxicol phoresis 2000; 21: 2684–2687. 2000; 16: 325–337. 55 Buccione R, Bannykh S, Santone I, Baldassarre M, Facchiano F, 44 Marc N, Galisteo M, Lagadic-Gossmann D, Fautrel A, Joannard F, Bozzi Y et al. Regulation of constitutive exocytic transport by Guillouzo A et al. Regulation of phenobarbital induction of the membrane receptors. A biochemical and morphometric study. cytochrome P450 2b9/10 genes in primary mouse hepatocyte J Biol Chem 1996; 271: 3523–3533. culture. Involvement of calcium- and cAMP-dependent pathways. 56 Keenan C, Kelleher D. Protein kinase C and the cytoskeleton. Cell Eur J Biochem 2000; 267: 963–970. Signal 1998; 10: 225–232.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Leukemia