Intracellular ABC Transporter A3 Confers Multidrug Resistance in Leukemia Cells by Lysosomal Drug Sequestration

Intracellular ABC transporter A3 confers multidrug resistance in leukemia cells by lysosomal drug sequestration

B Chapuy1, R Koch1, U Radunski1, S Corsham1, N Cheong2, N Inagaki3, N Ban4, D Wenzel5, D Reinhardt6, A Zapf7, S Schweyer8, F Kosari9, W Klapper9, L Truemper1 and GG Wulf1

1Department of Hematology and Oncology, Georg-August-University Goettingen, Goettingen, Germany; 2Department of Physiology, University of Pennsylvania, Philadelphia, PA, USA; 3Department of Diabetes and Clinical Nutrition, Graduate School of Medicine, Kyoto University, Kyoto, Japan; 4Department of Physiology, Akita University School of Medicine, Akita, Japan; 5Department of Neurobiology, Max Planck Institute for Biophysical Chemistry, Goettingen, Germany; 6Department of Pediatrics, Medizinische Hochschule Hannover, Hannover, Germany; 7Department of Medical Statistics, Georg-August-University Goettingen, Goettingen, Germany; 8Department of Pathology, Georg-August-University Goettingen, Goettingen, Germany and 9Department of Hematopathology and Lymph Node Registry, University of Schleswig- Holstein, Kiel, Germany

Multidrug resistance (MDR) seriously limits the efficacy of cells exhibiting progenitor cell properties both in vitro and chemotherapy in patients with cancer and leukemia. Active in vivo in numerous other types of cancer.7 As for the transport across membranes is essential for such cellular drug resistance, largely provided by ATP-binding cassette (ABC) mechanism of dye and drug exclusion, we first discovered the transport proteins. Intracellular drug sequestration contributes overexpression of the ATP-binding cassette transporter A3 to MDR; however, a genuine intracellular ABC transport protein (ABCA3) in leukemic SP cells in an in-vivo leukemia model. with MDR function has not yet been identified. Analyzing the Those data were confirmed in series of malignant SP cells from intrinsic drug efflux capacity of leukemic stem cells, we found both patients with neuroblastoma and AML.7–9 Here we the ABC transporter A3 (ABCA3) to be expressed consistently demonstrate that ABCA3 was expressed in virtually all cases in acute myeloid leukemia (AML) samples. Greater expression of ABCA3 is associated with unfavorable treatment outcome, of AML from a series of 119 pediatric and adult patients, and and in vitro, elevated expression induces resistance toward a that high levels of expression are associated with a significantly broad spectrum of cytostatic agents. ABCA3 remains localized impaired prognosis. We discovered that expression of ABCA3 within the limiting membranes of lysosomes and multivesicular induces a phenotype of broad multidrug resistance (MDR), bodies, in which cytostatics are efficiently sequestered. In mediated by subcellular drug sequestration to lysosomes and addition to AML, we also detected ABCA3 in a panel of finally gathered evidence of ABCA3 expression in all major lymphohematopoietic tissues and transformed cell lines. In conclusion, we identified subcellular drug sequestration types of malignant lymphohematopoietic disease. mediated by the genuinely intracellular ABCA3 as being a clinically relevant mechanism of intrinsic MDR. Leukemia (2008) 22, 1576–1586; doi:10.1038/leu.2008.103; Patients, materials and methods published online 8 May 2008 Keywords: drug resistance; ABCA3; lysosome Patients For the AML cohort of 119 patients, the bone marrow (n ¼ 112) and peripheral blood (n ¼ 7) specimens analyzed in this study represented archival diagnostic material collected from children Introduction treated within the German multicenter BFM98 protocol, and adults treated according to the German HD98A and B protocol Despite many improvements in the efficacy of anticancer at the University Hospital in Goettingen, Germany. The study agents, the resistance of tumor cells to cytostatic drugs remains was submitted to the Ethics Committee of the University of the major cause of treatment failure in patients with dissemi- Goettingen, and no objections were raised. The clinical nated malignant diseases. As a paradigm, multiagent chemo- characteristics of the AML study population are reported in therapy induces complete hematological remission in most Supplemental Table 1. As control samples, adult human bone patients with acute myeloid leukemia (AML), but the majority of marrow progenitor cells were isolated from routine diagnostic such patients still experience relapse and will finally succumb to posterior iliac crest aspirates of individuals without marrow the disease.1 On the cellular level, such leukemic cells initiating disease involvement, again following internal review board (IRB) relapse must combine in one cell the resistance to cytostatic approval. The tissue array of aggressive lymphomas was collected drugs with the capacity for clonal outgrowth.2–4 Applying the from patients participating in the German multicenter studies Hoechst 33342 staining technique to isolate stem cells with side NHLB1, NHLB2 and the Mabthera International Trial of the population (SP) properties,5 we described such a leukemic German high-grade lymphoma study group. Informed consent to progenitor cell with intrinsic drug efflux capacity some years the scientific evaluation of the tissue samples was obtained from ago.6 Consecutively, we and others have detected malignant SP the patients on study inclusion, and permission for this analysis was obtained from the IRB of the Deutsche Studiengruppe Hochmaligne Non-Hodgkin-Lymphome study group. Correspondence: Professor Dr GG Wulf, Department of Hematology and Oncology, Georg-August-University Goettingen, Robert-Koch-Str 40, Goettingen 37075, Germany. E-mail: [email protected] Reagents, cells, plasmids and antibodies Received 12 February 2008; revised 28 March 2008; accepted 28 The stable ABCA3-eGFP and green fluorescent protein (GFP) March 2008; published online 8 May 2008 transfectants HEK293-ABCA3/eGFP and HEK293-eGFP were Multidrug resistance in leukemia cells B Chapuy et al 1577 obtained from the embryonic kidney cell line HEK293 as 58 1C) and elongation (60 s at 72 1C) followed by a melting curve described previously and were routinely propagated in Dulbecco’s analysis. Subsequently, the threshold PCR cycle number (CT) modified Eagle’s medium supplemented with 300 mg/ml G418. was obtained when the increase in the fluorescence signal of the For cytotoxicity assays, the transfected cell lines were propa- PCR product indicated exponential amplification. This value gated without G418 for four passages, without losing transgene was then normalized to the threshold PCR cycle number expression as evaluated by fluorescence microscopy. The obtained for b-actin mRNA from a parallel sample. The hABCA3 pEGFP-N1-ABCA3 wild-type (wt) and pEGFP-N1-ABCA3 primer (us 50-TTCTTCACCTACATCCCCTAC-30;ds50-CCTTTCG N568D mutant plasmids were described previously.10 The CCTCAAATTTCCC-30) yielded an amplicon of 139 bp, the monoclonal mouse antibody to early endosomal antigen 1 b-actin primer (us 50-CACACTGTGCCCATCTACGA-30;ds50-TGA (EEA1) was obtained commercially (Transduction Laboratories), GGATCTTCATGAGCTAGTCAG-30) and amplicon of 99 bp. the antibodies to lysosomal-associated membrane protein 1 Transcripts of mRNA from the cell line HL60, which displayed (LAMP1, code H4A3) and LAMP2 (code H4B4) were from the weak expression of hABCA3 both on the RNA and protein Developmental Studies Hybridoma Bank (Iowa City, IA, USA). levels, were run in parallel with all reactions, and the values of Human MPR46 was detected using the monoclonal antibody ABCA3 levels were reported using HL60 with the expression (code 10C6), and human ABCA3 with a polyclonal rabbit anti- level of 1 as standard. human ABCA3 antibody.11 The myeloid leukemia cell line HL60 (DSMZ, Braunschweig, Germany) was propagated in RPMI 1640 supplemented with 25 mM 4-(2-hydroxyethyl)-1- Microscopy piperazineethanesulfonic acid, GlutaMAX I (Gibco-BRL, For the routine detection of ABCA3 on sorted cell populations, Karlsruhe, Germany), penicillin/streptomycin (Sigma, Stein- we performed indirect immunocytology following cytocentrifu- heim, Germany and Biochrom, Berlin, Germany) and 10% gation for 10 min at 800 g and fixation with 4% paraformalde- heat-inactivated fetal calf serum (Gibco-BRL). The T-cell hyde (PFA)/phosphate-buffered saline (PBS). The primary lymphoma/leukemia cell lines Jurkat and HUT78 (DSMZ) and antibody to ABCA3, a polyclonal rabbit serum, was diluted the variant PM1 of HUT78, as well as the natural killer (NK) cell 1:200 in PBS and incubated overnight. Cells were fixed in 4% line YT (DSMZ) were also propagated in RPMI 1640 supple- PFA/PBS, followed by blocking of endogenous peroxidase and mented as above. The NK cell line NK92 (kindly provided by T visualized using the horseradish peroxidase reaction with the Tonn, Frankfurt, Germany) was cultured in X-vivo medium Envision system according to the manufacturer’s instructions (BioWhittaker, Taufkirchen, Germany), containing 5% human (Vector, Peterborough, UK). Likewise, deparaffination, fixation AB plasma supplemented with 100 U/ml interleukin -2 (R&D with 4% PFA/PBS, antigen retrieval in citrate-based buffer, and Systems, Minneapolis, MN, USA). The aggressive B-cell antibody stains as above were performed on paraffin tissue lymphoma/leukemia cell lines Balm3, Raji, Karpas 422, Ramos, sections arranged in multi-tissue arrays according to standard SuDHL4 (DSMZ) as well as the Hodgkin’s cell lines HDLM2, protocols. ABCA3 cytoplasmic staining levels were quantified L1236 and L428 (DSMZ) and the mantle cell lymphoma cell by panel agreement of three independent reviewers as negative lines Granta 519, Mino and Sp53 (ATCC, Manassas, VA, USA) (0), low ( þ ), intermediate ( þþ) or high ( þþþ). Pappen- were maintained in RPMI 1640 supplemented as above. heim’s panoptic and myeloid peroxidase stains of hematopoietic cells, and hematoxylin/eosin stains of tissue sections were performed according to the standard protocols. In order to carry Cell separation and flow cytometry out fluorescence colocalization work, HEK293-ABCA3/eGFP Mononuclear cells (MNC) were separated from whole blood or cells were grown on 24-well cover slips to 50% confluency prior bone marrow aspirates using density centrifugation with Ficoll to experimentation. At room temperature, cells were fixed using (Pharmacia, Uppsala, Sweden). A FACScan cytometer (Becton 3.7% PFA for 20 min, with the subsequent quenching of any Dickinson, Heidelberg, Germany) was used for the flow unspecific binding using 50 mM NH4Cl for 15 min and cytometric analysis resulting from the excitation of eGFP permeabilization with 0.05% Triton X-100 in PBS for 15 min. (fluorescein isothiocyanate) and daunorubicin (DNR) at Primary antibodies to EEA1, LAMP1 and MPR46 for the 488 nm. Data were analyzed with CellQuest (Becton Dickinson) colocalization study were diluted 1:100 in PBS for 1 h. After or WinMDI (Trotter, 1999) software. Cell sorting by flow washing twice with PBS and incubating with 10% goat serum, cytometry was performed on the FACSVantage SE. Purity was the primary antibodies were visualized using goat secondary determined with FACS, and only samples exceeding 90% antibodies at a dilution of 1:500 in PBS coupled to Cy3 marker expression after positive selection were used in further (Dianova, Hamburg, Germany). All samples were mounted in experiments. Fluoromount (DAKO, Hamburg, Germany) and analyzed with the TCS-2 AOBS confocal laser scanning microscope (Leica) with a Â 63 inversion oil objective (Leica). The data were RNA preparation and PCR exported as TIFF files and arranged using Adobe PhotoShop Total RNA was isolated from MNC or cell line preparations with without further modification of the primary image. All images the RNeasy kit (Qiagen, Hamburg, Germany) according to the from confocal sections were obtained under identical micro- manufacturer’s instructions, suspended in RNAase-free-distilled scopic settings. For the supravital lysosome staining and tracking water, and quantified using a spectrophotometer. Superscript II of DNR (Sigma), cells were grown on 24-well cover slips to 50% reverse polymerase (Invitrogen, Karlsruhe, Germany) was used confluency prior to experimentation (in the case of adherent for cDNA synthesis, and quantitative reverse transcriptase cells) or to a density of 2 Â 106 cells per ml growth medium (in polymerase chain reactions (qRT-PCR) of hABCA3 and b-actin the case of suspension cultures). Lysotracker Red (50–100 nM, transcripts were performed in duplicates on a TaqMan cycling Molecular Probes, Karlsruhe, Germany) was incubated with the machine (ABI Prism 7900HT Sequence Detection System, cells at 37 1C for 1 h. In order to visualize subcellular DNR, cells Applied Biosystems, Darmstadt, Germany). The SYBR green kit were incubated with 2.5 mM DNR at 37 1C for 2 h, followed by (Qiagen) was used according to the manufacturer’s protocols, washing with PBS. The viable cells were analyzed for with 40 cycles of denaturation (15 s at 95 1C), annealing (30 s at subcellular DNR and eGFP fluorescence directly after washing,

Leukemia Multidrug resistance in leukemia cells B Chapuy et al 1578 and after a 24 h period of washout in fresh medium in cell 1050 mF, 99 O, t: 104 ms) and incubated for 3 h in 400 ml culture. Confocal sections were obtained as described above. OptiMEM before further expansion with 1.5 ml of complete Immunoelectron microscopy was performed according to the growth medium overnight. For the transfer of specific, double- Tokuyasu method. Wild-type HEK293, transfectants overexpres- stranded, predesigned siRNA against ABCA3 (Qiagen; sense 50- sing ABCA3, and MNC of AML samples were fixed with 2% PFA CGGAGAAAUUACAUGUGGA-dTdT-30; antisense sequence, and 0.1% glutaraldehyde in 0.1 M sodium phosphate (pH 7.4) at 50-UCCACAUGUAAUUUCUCCG-dTdG-30; coding region room temperature for 30 min, before the cells were postfixed starts at position 1035 relative to the start codon) in suspension with 4% PFA and 0.1% glutaraldehyde on ice for 2 h. After cell cultures, 50 nM siRNA against hABCA3 were electroporated being washed twice with PBSF0.02% glycine, the cells were with a program optimized for RNA transfer (300 V, 150 mF, embedded in 10% gelatin, cooled on ice and cut into small 1540 O, t: 231 ms). Transfer efficiency, as measured with the blocks. The blocks were infused with 2.3 M sucrose overnight transfer of Alexa488 tagged scrambled siRNA (Qiagen; sense 50- and stored in liquid nitrogen. Ultrathin cryosections were cut UUCUCCGAACGUGUCACG-UdT-30; antisense 50-ACGUGA from the frozen samples and labeled with primary antibodies CACGUUCGGAGA-AdT-30), routinely exceeded 80%. In the detected with protein A conjugated to gold (protein A gold, case of adherent cells, chemical transfection (Hyperfect, PAG). In the case of monoclonal antibodies, a polyclonal rabbit Qiagen) was used yielding similar degrees of efficiency anti-mouse bridging antibody (Sigma) was used prior to according to the manufacturer’s recommendations. Using detection with PAG. Sections were contrasted with uranyl siRNA knockdown, the nadir of ABCA3 mRNA and protein acetate methyl cellulose on ice for 10 min, embedded in the expression was documented as occurring 72 h after electropora- same solution and examined with a Phillips CM120 electron tion, as tested by qRT-PCR and FACS. microscope. In preparation of whole cell imaging, cells were fixed in 2% glutaraldehyde in 0.1 M PBS, pH 7.4, for 2 h, postfixed in 1% OsO4 for 1 h, dehydrated in ethanol and then Viability and clonogenicity assays, daunorubicin embedded in Epon. measurements For viability testing by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphe- nyltetrazolium bromide (MTT) test, cells were seeded in Transfer of plasmid DNA and siRNA triplicates on 96-well culture plates at a density of 1 Â 105 cells For plasmid transfer, 1 Â 106 cells in 150 ml OptiMEM (Gibco- per well and were treated with the indicated concentrations BRL) were mixed in a prechilled cuvette (Biozym, Hess. of DNR (Sigma), mitoxantrone (Wyeth, Mu¨nster, Germany), Oldendorf, Germany) with 10 mg of either pmax-GFP, pEGFP- etoposide (Hexal, Holzkirchen, Germany), vincristine (Cell N1-ABCA3 wt or pEGFP-N1-ABCA3 N568D, pulsed in an Pharm, Bad Vilbel, Germany), cytosine arabinoside (Ara-C, Cell EasyjecT (Equi Bio, Middlesex, UK) electroporator (300 V, Pharm) and gemtuzumab ozogamicin (GO, Wyeth) for 3 h each,

Figure 1 Expression of ATP-binding cassette transporter A3 (ABCA3) in acute myeloid leukemia (AML) blast cellsABCA3 expression was detected on the protein and mRNA levels in primary AML samples. Western blot analysis (a) of seven representative cases with a polyclonal antibody detecting ABCA3 at a molecular size of 188 kDa revealed marked inter-individual heterogeneity. In the myeloid cell line HL60, ABCA3 protein was detectable at only low levels, compared to strong expression after transfection with an ABCA3-expression vector. Immunocytology against ABCA3 further revealed notable heterogeneity within the blast cell population as a consistent ﬁnding in the clinical samples, with a representative example shown in (b), scale bar 2 mm. mRNA levels, measured by qRT-PCR normalized to b-actin housekeeping gene controls and represented on a logarithmic scale, are shown in a box-whisker plot (c) for clinical samples from the 109 AML patients, again documenting remarkable inter- individual differences in ABCA3 expression. Analyzed in parallel with the AML samples, ABCA3 was not detectable in 7 of 11, or very low levels falling outside the scale of this plot in 4 of 11 bone marrow samples from normal volunteers (#).

Leukemia Multidrug resistance in leukemia cells B Chapuy et al 1579 washed and then incubated for a period of 24 h, 3mM imidazole/HCl, 1 mM EDTA pH 7.3 and centrifuged for except vincristine (48 h) and Ara-C (42 h). After 24 h at 37 1C, 30 min at 35 000 g without brakes in a vertical tube rotor Vti the culture volume of 100 ml was supplemented with MTT in 65.1 (Beckmann Instruments, Krefeld, Germany). After removal PBS to achieve a final concentration of 0.5 mg/ml. After 4 h of percoll, the supernatant and membranes were prepared by incubation, suspension cells were spun, the supernatant centrifuging each of the 23 fractions for 1 h at 100 000 g. discarded and the pellet resuspended in 30% (v/v) dimethyl Membranes were resuspended in 0.5% Triton X-100/PBS and sulfoxide, 5% (v/v) formic acid and 1% (w/v) Triton-X100 (all used for further analysis, such as western blotting, determination from Sigma) dissolved in isopropanol. The light absorbance from of enzymatic activity and protein concentrations. DNR was formazan was measured at 540 nm on a Tecan SLT photometer measured within samples on Tecan Ultra Microplate Reader (Tecan SLT Spectra). We expressed the effect on viability as the (TECAN, Mainz, Germany). The enzymatic activity of ratio of values from treated versus untreated samples, that is, b-hexosaminidase, b-glucuronnidase and a-mannosidase was specific viability is the ratio of absorbance with drug to measured using either 10 mM p-nitrophenyl-2-acetamido-2- absorbance of untreated control. IC50 was defined as the deoxy-b-D-glucopyranoside, 20 mM p-nitrophenyl-b-D-glucuro- concentration of drug causing a 50% inhibition of cell growth nide or 10 mM p-nitrophenyl-a-D-mannopyranoside (all Sigma) as compared with untreated control. To analyze clonogenicity as substrate, which was prepared at 10 mM in 0.1 M sodium of HEK293 wt and transfectants, cells were seeded on a 10 cm citrate (pH 4.6). Usually 5–10 ml of the sample was mixed at 4 1C diameter dish and cultured to 20% confluency 24 h preceding with 100 ml of the substrate, followed by 0.5–4 h of incubation at drug exposure for 3 h as indicated. After drug removal by three 37 1C. The reaction was stopped by adding 0.5 ml of 0.4 M rounds of washing, cells were propagated in culture medium for glycine/NaOH (pH 10.4) and the specific activity was calcu- the time points indicated. Prior to macroscopic counting of cell lated after measuring the absorbance at 405 nm. clones, the culture dishes were fixed with 3.7% PFA/PBS for 30 min, washed and stained with hematoxylin for 5 min. All experiments were done in triplicate, with representative experiments for at least three independent assays being portrayed in the result section. In preparation of ionizing 1.0 5 PFS (g-radiation) and nonionizing (UVC light) irradiation, 1 Â 10 0.9 cells per well were plated either in 96-well plates (MTT assay) 0.8 P = 0.00691 or 500 cells on a 10 cm diameter dish (clonogenicity assay) low ABCA3 24 h prior to irradiation. The doses of UVC radiation were 0.7 0, 10, 20, 30 or 40 J/m2, and those for g-radiation were 0, 0.6 2.5, 5, 7.5, 10 and 30 Gy, as indicated. Viability was assayed by MTT 48 h, clonogenicity 10 days after irradiation as 0.5 described. 0.4 For DNR measurements, HEK293 and HEK293-ABCA3 cells 0.3 were exposed to DNR for 1 h at the concentrations indicated. Probability of survival [%] high ABCA3 Following three washes with PBS, nuclei were isolated using the 0.2 Nuclei EZ Prep Isolation Kit (Sigma) according to the manu- 0.1 facturer’s recommendations. Briefly, cells were harvested in 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 prechilled PBS after trypsination, and an aliquot of the cells was time [d] used to determine total cellular DNR retention. Thereafter, the cells were lysed for 5 min at 4 1C in lysis buffer, before the nuclei were collected by centrifugation at 500 g for 5 min. The pooled 1.0 low ABCA3 OS supernatant was collected as the organelle/cytosol pool. The 0.9 P = 0.013 nuclei were resuspended in PBS and the nuclear DNR 0.8 fluorescence was analyzed at 560 nm after excitation at 488 nm on the flow cytometer. For subcellular fractionation, 0.7 HEK293 and HEK293-ABCA3 cells were incubated for 3 h with 0.6 2.5 mM DNR. Subsequently, after washing with ice-cold PBS, 0.5 cells were harvested from three 15 cm dishes from each cell 0.4 line, homogenized and subcellular fractionation of the post- nuclear supernatant (PNS) performed as detailed. Western blot 0.3 high ABCA3 and quantification of western blot results with ImageJ quantifi- Probability of survival [%] 0.2 cation program (rsb.info.nih.gov/ij/) were performed according 0.1 to standard protocols. 0.0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 time [d]

Subcellular fractionation Figure 2 ATP-binding cassette transporter A3 (ABCA3) expression After washing with ice-cold PBS, cells were harvested from three levels and chemotherapy treatment outcome in patients with acute 15 cm dishes from each cell line, in 0.25 M sucrose buffered with myeloid leukemia (AML). Kaplan–Meier product limit estimates of 3mM imidazole/HCl, pH 7.3, 1 mM EDTA and a 1:500 dilution progression-free survival (PFS, a), overall survival (OS, b) for patients of a protease inhibitor cocktail (Sigma) using five strokes of a (n ¼ 87) with newly diagnosed AML treated by pediatric multicenter 1 and adult protocols, without stem cell transplant recipients. Low levels G27 syringe. The PNS was prepared (5 min/500 g/4 C) and 6 ml of ABCA3 expression (gray line) compared to high ABCA3 levels of this protein normalized PNS was adjusted to a final (black line) were significantly associated with both improved concentration of 20% Percoll solution (12 ml) in 0.25 M sucrose, progression-free (P ¼ 0.0069) and OS (P ¼ 0.013) by log-rank test.

Leukemia Multidrug resistance in leukemia cells B Chapuy et al 1580 Statistical evaluations differences in survival from the time of diagnosis through To carry out survival analysis on the patient data in the AML progression (progression-free survival, PFS) or death of any cohort, Kaplan–Meier analysis was employed, analyzing cause (overall survival, OS). Multivariate analysis of confounding

3h DNR 48h ARA-C 1 1 * * 0.8 0.8 * * 0.6 * 0.6 0.4 0.4 0.2 0.2 0 0 0 1 10 50 0 15 10 50

3h Mitoxantron 3h Etoposide * 1 1 * 0.8 0.8 0.6 * 0.6 * * 0.4 0.4 0.2 0.2 * Viability [% of control] 0 0 0 1 10 50 100 010 20 50 100

3h GO 3h Vincristine * 1 1 * * 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 0 0.5 1 5 011020 5

Concentration [µM]

HEK293 HEK293 + ABCA3

1 200 160 0.8 * 120

0.6 80 40 Viability [% of total]

0.4 ABCA3-GFP [MFI] 0 05 HEK299 HEK299 ABCA3-siRNA DNR [µM] ABCA3-GFP

HEK293 HEK293+ABCA3 HEK293+ABCA3+scrumbled siRNA HEK293+ABCA3+siRNA ABCA3

Figure 3 Induction of cellular multidrug resistance in vitro by ATP-binding cassette transporter A3 (ABCA3). In HEK293 cells, stable ABCA3 expression led to significantly increased cell survival following exposure to soluble toxins. Viability differences (a) in the response of HEK293 (gray lines) and HEK293-ABCA3 (black lines) to incubation with daunorubicin, Ara-C, mitoxantrone and etoposide reaching statistical significance by Student’s t-test were marked *. HEK293 cells do not express CD33, the target of the immunotoxin gemtuzumab ozogamicin (GO, bottom left panel in a). Knockdown of ABCA3 expression by specific siRNA diminished the expression levels of ABCA3 (b, right panel, mean values of 30 000 events each) and significantly increased the susceptibility of HEK293-ABCA3 cells to daunorubicin toxicity (b, left panel). All the results are representative of at least three independent experiments performed in triplicate, with results portrayed as mean values±s.e.m.

Leukemia Multidrug resistance in leukemia cells B Chapuy et al 1581 variables, that is, risk group according to karyotype (high-, individual patient proved to be ABCA3 positive, the intercellular intermediate- and low-risk according to Grimwade12 was tested distribution within the blast population was significantly by Cox regression using Statistica 7.1 software (StatSoft) or R heterogeneous (Figure 1b). In questioning the impact of ABCA3 (R-Project). The threshold level of 0.5 arbitrary units in the on the efficacy of chemotherapy of patients with AML, we qRT-PCR to separate low and high levels of expression of ABCA3 analyzed patients at first presentation of disease and who had in a given sample was based on biological considerations. As not undergone any immunotherapy in the form of allogeneic determined on an array of cell lines, cell populations with stem cell transplantation during the observation period (n ¼ 87). transcript levels of 0.5 compared to HL60 were scored negative Thus, we found high levels of ABCA3 significantly associated for ABCA3 protein detection by western blot or immunocyto- with both reduced PFS (Figure 2a; P ¼ 0.0069) and OS of the logy. In the statistical evaluation of the in-vitro experiments, patients (Figure 2b; P ¼ 0.013). These differences were still differences between cohorts of samples with and without significant when stem cell transplant recipients were included experimental intervention were tested using the two-tailed (n ¼ 109, Supplementary Figure 1). The survival differences Student’s t-test, and differences with Po0.05 were considered between groups of high and low levels of ABCA3 expression significant. were also evident within the prognostic cytogenetic subgroups (Supplementary Figure 1). Performing multivariate analysis using Cox regression, ABCA3 levels separated outcome independently of karyotype implications (Supplementary Figure 1). Beyond Results AML, we discovered a significant expression of ABCA3 also in a broad spectrum of transformed hematopoietic cells. Among the Expression and clinical relevance of ABCA3 in acute hematopoietic malignancies, ABCA3 levels were highest in the myeloid leukemia aggressive B-cell lymphomas, both on the level of cell lines and Following from the identification of ABCA3 expression in clinical tissues (Figure 7). Thus, we found significant expression leukemic progenitor cells with the SP phenotype in an in-vivo 8 of ABCA3 in the blasts of patients with AML, and high levels of model system, here we analyzed the expression of ABCA3 in a expression were found to be associated with poor treatment cohort of 81 pediatric and 38 adult patients at the time of first outcome. diagnosis with AML (Supplementary Table 1). Transcripts of the transporter were detectable in all samples, with expression levels varying over a wide range of four log scales, whereas ABCA3 transcripts were not, or only at very low levels, Induction of a multidrug resistance phenotype by detectable in normal whole bone marrow samples (Figure 1c). ABCA3 in HEK293 and HL60 cells Interestingly, AML patients displayed remarkable differences Addressing the question as to whether ABCA3 expression is able between individuals (Figures 1a and b). Furthermore, when an to mediate drug resistance in a similar fashion to other ABC

Figure 4 Localization of ATP-binding cassette transporter A3 (ABCA3) to organelles of the late endosomal system. Corresponding confocal sections after indirect immunofluorescence stains of HEK293-ABCA3/enhanced green fluorescent protein (eGFP C) depicted the colocalization of ABCA3 with LAMP1 (a, upper row), a membrane protein of the late endosomal system. Likewise, Lysotracker red staining, which predominantly accumulates in lysosomes, colocalized with ABCA3/eGFP fluorescence (a, lower row). HEK293-ABCA3 cells retained Lysotracker red approximately four times more efficiently than HEK293 wild-type (wt) cells (b, lower and upper panels) as detected by fluorescence-activated cell sorting (FACS). Furthermore, ABCA3 overexpression in HEK293 was associated with elevated levels of late endosomal membrane proteins LAMP1 and 2 (c), with relative increases of 41 and 24%, respectively, comparing wt HEK293 versus HEK293-ABCA3 blots by densitometric quantification.

Leukemia Multidrug resistance in leukemia cells B Chapuy et al 1582 transporters, we compared the susceptibility of a stable multidrug resistance phenotype, both in the HEK293 and HL60 transfectant expressing ABCA3, HEK293-ABCA3/eGFP, to the leukemic cell models. parental, ABCA3-negative reporter cell line HEK293 after exposure to a panel of cytostatic drugs. As shown in Figure 3, ABCA3 overexpression induces a significant reduction in Localization of ABCA3 to the late endosomal cytotoxicity observed following exposure to DNR, mitoxan- compartment trone, etoposide, Ara-C and vincristine (Figure 3a). The Under high-power magnification, we had already observed a protective effect of ABCA3 expression was not observed on cytoplasmic staining pattern in primary leukemic cells stained exposure to the immunotoxin GO, which would not bind to for ABCA3 (Figure 1b). Correspondingly, ABCA3 is found CD33-negative HEK293 cells (Figure 3a). The ABCA3-mediated delineating cytoplasmic organelles in the HEK293-ABCA3/eGFP protection of HEK293 cells from cytotoxic damage by soluble variant cells (Figure 4a). Such subcellular fluorescence coloca- substances could be significantly reduced by functional knock- lizes with markers for late endosomal and lysosomal mem- down of ABCA3 with a specific siRNA (Figure 3b, here shown branes, that is, the LAMP1 (Figure 4a), but not with markers of for DNR). Beyond direct cytotoxicity, we also analyzed the the early endosomal compartment such as EEA1 or Golgi/early effects of ABCA3 on clonogenicity in the HEK293 in-vitro endosomes (MPR46, data not shown). Furthermore, ABCA3 model. In correlation with the viability assays, ABCA3 expres- colocalizes clearly with fluorescent agents staining the lumen of sion markedly increases the fraction of cells with the capacity of acidic organelles, such as the vital dye Lysotracker Red clonal regrowth following exposure to chemotherapy in (Figure 4a). Staining with Lysotracker Red also reveals a fourfold clonogenicity tests, and does not alter the susceptibility to augmentation of dye retention per cell in the cells expressing g- and UV-radiation (Supplementary Figure 2A). A nonfunc- ABCA3 when compared to the wt cell line (Figure 4b). In tional, mutated variant N568D-eGFP of ABCA3 does not induce addition, the extended lysosomal volume in ABCA3-positive drug resistance, and specific siRNA against ABCA3 increases the cells is associated with an increase in lysosomal membrane cytotoxic efficacy of DNR, etoposide and Ara-C (Supplementary proteins such as LAMP1 and 2 (Figure 4c), as well as some, but Figure 2C, D). Thus, ABCA3 function is associated with a not all lysosomal matrix enzyme activity (data not shown). To

Figure 5 Morphology of ATP-binding cassette transporter A3 (ABCA3)-positive organelles in HEK293 and primary leukemic cells. In comparison with HEK293 wild-type (wt) cells (a), HEK293-ABCA3/enhanced green ﬂuorescent protein (eGFP) cells (b, detailed in c) displayed an abundance of late endosomal organelles, such as lysosomes, multivesicular bodies and multilamellar bodies (closed arrowhead in b and c), depicted by Epon- ﬁxed electron micrographs. In cryoimmunostaining experiments, ABCA3 colocalized with LAMP1 in the lysosomal membranes (d). Here, ABCA3 protein is labeled with 10 nm gold particles (closed arrowhead), and LAMP1 with 5 nm gold particles (open arrowhead). The size of the scale bars is indicated on the respective panels.

Leukemia Multidrug resistance in leukemia cells B Chapuy et al 1583 further characterize the ABCA3-positive organelles, we exam- nuclear accumulation during drug exposure. After 24 h washout, ined the wt and transfectant HEK293, as well as primary the ABCA3-positive vesicles still retained the drug, whereas the leukemic cells under an electron microscope. In HEK293 cells, nucleus was found cleared of DNR fluorescence (Figure 6a). ABCA3 overexpression leads to a significant increase in the Accordingly, subcellular fractionation of such cells revealed number of organelles of the late endosomal system, that is, DNR in the fraction of lysosomal content with lysosomal lysosomes, multivesicular bodies and multilamellar bodies b-hexosaminidase activity and expression of the lysosomal (Figures 5b and c). Correspondingly, we also detected an membrane protein LAMP1, and only to a very minor degree in excessive accumulation of lysosomes in the cytoplasm of the endosomal fractions with expression of the early endosomal leukemic blast cells isolated from patients with AML, yet not marker MRP46 (Figure 6c). Finally, together with the increased as many multilamellar or multivesicular bodies (data not lysosomal retention of DNR, we observed a greater accumula- shown). In line with the observations using confocal micro- tion of DNR in the ABCA3-positive cells, albeit associated with scopy, we again noticed colocalization of ABCA3 and LAMP1 a reduction in nuclear DNR levels in cells overexpressing on the limiting membrane of the lysosome (Figure 5d). Thus, ABCA3 (Figures 6b and 7). Our data are consistent with the ABCA3 is expressed in the membranes of the late endosomal efficient sequestration of DNR in the lysosomal space, thus system in leukemic and ABCA3-transfected cells, and increased reducing the amount of drug reaching the nucleus. transporter expression augments the endosomal system.

Discussion ABCA3-mediated subcellular drug sequestration Subcellular drug sequestration with subsequent reduction of the We report here on the expression of the genuine intracellular active substance at the intracellular site of action may be viewed ABCA3 in AML, document its association with poor outcome of as contributing to multidrug resistance.13 Exploiting the fluores- disease in AML, present evidence of its capacity to protect cells cence of DNR in order to track its location in the cell using against the toxic effects from an array of cytotoxic drugs and confocal microscopy, we depicted DNR fluorescence concen- propose drug sequestration into lysosomes/lysosomal-related trated in vesicular organelles showing ABCA3/eGFP fluores- organelles as the mechanism for its protective effects. cence (Figure 6a). These experiments revealed that DNR Since the initial description of drug resistance by the accumulated in the ABCA3-positive vesicles in parallel to the permeability glycoprotein (MDR1), drug extrusion by ABC

Figure 6 The effects of ATP-binding cassette transporter A3 (ABCA3) overexpression on subcellular distribution of daunorubicin (DNR). Following exposure of HEK293-ABCA3/enhanced green fluorescent protein (eGFP) cells to DNR, which fluoresces red following excitation at 488 nm, confocal imaging located the drug (a, left panels) in cytoplasmic vesicular structures (arrowheads), congruent with the ABCA3/eGFP fluorescence (a, middle panels), as shown by the yellow fluorescence in the merge (a, right panels). Directly following DNR exposure (a, upper row, Â 400 magnification), the drug was detected in the nucleus and in the eGFP-positive cytoplasmic vesicles. After 24 h washout, (a, lower row, Â 630 magnification), the drug was still retained in the vesicles, whereas the nucleus was cleared of DNR fluorescence. Flow cytometry revealed an increase in total cellular drug uptake at higher DNR concentrations in HEK293 cells expressing ABCA3 (b, left panel), and a lower nuclear accumulation of the drug, shown in absolute values (b, middle panel) and the ratio of nuclear versus total cellular fluorescence (b, right panel). The mean fluorescence intensity (MFI) of 30 000 events each in representative experiments is depicted. Accordingly, subcellular fractioning documented the accumulation of DNR (c, upper panel) in those fractions exhibiting high b-hexosaminidase activity (c, middle panel), high expression of LAMP1 (c, bottom panel upper row) and low expression of MPR46 (c, bottom panel lower row), consistent with the accumulation of DNR predominantly in lysosomes.

Leukemia Multidrug resistance in leukemia cells B Chapuy et al 1584

Figure 7 Detection of ATP-binding cassette transporter A3 (ABCA3) expression in all major types of lymphohematopoietic malignancies. ABCA3 was detected by qRT-PCR in a panel of hematopoietic cell lines (a), yielding highest expression levels in aggressive lymphoma and Hodgkin’s lymphoma cell lines, exceeding the levels in the myeloid leukemia cell lines substantially. Correspondingly, immunohistology revealed ABCA3 protein expression in all 240 tissue sections from aggressive non-Hodgkin’s lymphoma (NHL) specimens (b) of the trials the Mabthera International (MINT), NHL B1 and B2, graded as high-, intermediate- and low-expression levels, and represented in % relative proportions (b, lower panel) in the respective patient cohorts. An example of ABCA3 staining in an aggressive B-cell lymphoma is shown in (b) (lower panels, Â 100 magnification left, Â 400 magnification right), together with a normal (reactive) lymph node showing no ABCA3 reactivity (c, upper panels, Â 100 magnification left, Â 400 magnification right).

transporters has been studied extensively.13-15 It was in AML toxic effect. We speculate that ABCA3 overexpression has an that MDR1 expression was first found to be associated with poor impact on the cellular detoxification of all drugs, the extrusion response to treatment.16,17 Out of a family of 48 human ABC of which involves late endosomal/lysosomal transporters and transporters, further proteins such as MRP1, its homologues pathways. This would explain the broad spectrum of drug MRP2, MRP3, MRP5 and the breast cancer resistance protein resistance associated with ABCA3 overexpression. (BCRP or ABCG2) were detected in leukemic blasts at variable The question still remains as to how the concept of ABCA3- degrees of expression, and have been linked to clinically mediated drug resistance via augmentation of the late endo- relevant drug resistance in AML. These transporters are located somal system fits with previous reports on ABCA3. in the plasma membrane, thus making reductions in total cell ABCA3 was cloned originally from the cDNA of a human drug retention a conceivable explanation for the resistant cancer cell line.21 It is highly conserved between mammalian phenotype.18 Beyond total cell drug levels, however, the species, and was found expressed predominantly in the alveolar phenomenon of cytostatic drug sequestration in cytoplasmic type II cells of the lung.11,22 In alveolar cells, ABCA3 function is vesicles has been described in several drug-resistant cell lines, essential to the formation of lamellar bodies, a lysosomal-related and atypical intracellular activity of genuine plasma membrane organelle, and indispensable for pulmonary surfactant produc- transporters such as MDR1 and MRP1 has been implicated.19,20 tion and/or secretion. A loss of ABCA3 function leads to fatal It was subsequently observed that the cytoplasmic structures respiratory stress syndrome in neonates.23 In vitro, ABCA3 is that retained drugs were acidic in nature, that is, lysosomes and necessary, but not sufficient for the formation of mature lamellar endosomes.20 Following this concept, weak bases such as bodies in alveolar cells, leading to the concept that, in addition anthracyclines accumulate in acidic compartments become to ABCA3, other surfactant proteins such as SP-B are required for membrane impermeable through protonation and are thus kept the transition of lysosomes to distinct lamellar bodies.10,24 apart from the nucleus. Besides its function in the lungs, several reports also provided Our findings on the genuine intracellular transporter ABCA3 first evidence of a possible role of ABCA3 in cancer cell drug are congruent with this model and may at least in part explain resistance. In a matched pair analysis on a panel of 22 cell lines some of the experimental observations cited above: a high and their resistant variants on the genomic level, ABCA3 was degree of ABCA3 transporter expression leads to augmentation found to be overexpressed in three resistant variants with an of the late endosomal/lysosomal storage capacity, which is increase in gene copy numbers as the mechanism of ABCA3 acidic in nature, and improved efficacy of drug sequestration, abundance.25 Likewise, drug-resistant variants of human thus protecting the nuclear or cytoplasmic drug targets from any T-lymphoblastic cell lines displayed an increase in ABCA3

Leukemia Multidrug resistance in leukemia cells B Chapuy et al 1585 expression levels in comparison with the parental lines, in this 3 Lapidot T, Sirard C, Vormoor J, Murdoch B, Hoang T, Caceres- study due to transcriptional upregulation.26 Furthermore, utilizing Cortes J et al. A cell initiating human acute myeloid leukaemia RT/PCR Steinbach et al.27 detected ABCA3 expression in a after transplantation into SCID mice. Nature 1994; 367: 645–648. cohort of 42 pediatric AML samples, and reported significantly 4 Jordan CT, Guzman ML, Noble M. Cancer stem cells. N Engl J Med 2006; 355: 1253–1261. higher levels of ABCA3 in 21 patients with poor treatment 5 Goodell MA, Brose K, Paradis G, Conner AS, Mulligan RC. outcome, compared to 21 patients who benefited from Isolation and functional properties of murine hematopoietic stem chemotherapy. Finally, in addition to the data of this report, cells that are replicating in vivo. J Exp Med 1996; 183: 1797–1806. significant ABCA3 expression was identified in cells of epithelial 6 Wulf GG, Wang RY, Kuehnle I, Weidner D, Marini F, Brenner MK neoplasms such as neuroblastoma and breast cancer.7 In this et al. A leukemic stem cell with intrinsic drug efflux capacity in context, Stahlman et al.28 recently detected ABCA3 expression acute myeloid leukemia. Blood 2001; 98: 1166–1173. 7 Hirschmann-Jax C, Foster AE, Wulf GG, Nuchtern JG, Jax TW, in a number of nonpulmonary tissues from both adults and Gobel U et al. A distinct ‘side population’ of cells with high drug children, such as the epithelia of the liver, pancreas, stomach, efflux capacity in human tumor cells. Proc Natl Acad Sci USA trachea, larynx, adrenal cortex, proximal renal tubules and the 2004; 101: 14228–14233. Purkinje cells in the cerebellum. We thus speculate that ABCA3 8 Norwood K, Wang RY, Hirschmann-Jax C, Andreeff M, is an essential factor in the formation of late endosomes/ Brenner MK, Goodell MA et al. An in vivo propagated human lysosomes in many different cell types at defined stages of acute myeloid leukemia expressing ABCA3. Leuk Res 2004; 28: 295–299. differentiation, with the transition to high levels of expression 9 Wulf GG, Modlich S, Inagaki N, Reinhardt D, Schroers R, and a role for the formation of specialized organelles and Griesinger F et al. ABC transporter ABCA3 is expressed in acute functions in a few tissues, such as the formation of lamellar myeloid leukemia blast cells and participates in vesicular bodies and surfactant excretion in type II pneumocytes. transport. Haematologica 2004; 89: 1395–1397. Associated with malignant transformation, however, transcrip- 10 Cheong N, Madesh M, Gonzales LW, Zhao M, Yu K, Ballard PL tional control of ABCA3 expression is lost, leading to high levels et al. Functional and trafficking defects in ATP binding cassette A3 of ABCA3 expression. From a clinical perspective, our findings mutants associated with respiratory distress syndrome. J Biol Chem 2006; 281: 9791–9800. suggest the interference with lysosomal sequestration mechanisms 11 Yamano G, Funahashi H, Kawanami O, Zhao LX, Ban N, Uchida Y as a promising avenue to be explored in order to modulate the et al. ABCA3 is a lamellar body membrane protein in human lung susceptibility of leukemia cell toward cytostatic agents. We alveolar type II cells. FEBS Lett 2001; 508: 221–225. envision the interference with ABCA3 by a specific inhibitor, or 12 Grimwade D, Walker H, Oliver F, Wheatley K, Harrison C, the inhibition of lysosomal function by lysosomotropic sub- Harrison G et al. The importance of diagnostic cytogenetics on stances, for example, chloroquine as such options. outcome in AML: analysis of 1, 612 patients entered into the MRC AML 10 trial. The Medical Research Council Adult and Children’s In summary, the ABCA3 is expressed in the cells of AML, Leukaemia Working Parties. Blood 1998; 92: 2322–2333. and mediates clinically relevant multidrug resistance by drug 13 Simon SM, Schindler M. Cell biological mechanisms of multidrug sequestration in late organelles of the endosomal system. We resistance in tumors. Proc Natl Acad Sci USA 1994; 91: view the interference with detoxification pathways related to 3497–3504. ABCA3 in tumor cells to be a very promising approach to 14 Gottesman MM, Pastan I. Biochemistry of multidrug resistance improving the efficacy of cytostatic treatment. mediated by the multidrug transporter. Annu Rev Biochem 1993; 62: 385–427. 15 Juliano RL, Ling V. A surface glycoprotein modulating drug Acknowledgements permeability in Chinese hamster ovary cell mutants. Biochim Biophys Acta 1976; 455: 152–162. 16 Marie JP, Zittoun R, Sikic BI. Multidrug resistance (mdr1) gene BC designed and performed research, and analyzed data; RK, UR expression in adult acute leukemias: correlations with treatment and SC performed research; NC, NI and NB contributed vital new outcome and in vitro drug sensitivity. Blood 1991; 78: 586–592. reagents; DW performed research; DR provided vital reagents; AZ 17 Leith CP, Kopecky KJ, Chen IM, Eijdems L, Slovak ML, McConnell analyzed data; SS, WK, FK and LT contributed vital reagents and TS et al. Frequency and clinical significance of the expression of analyzed data; GGW designed research, analyzed data and wrote the multidrug resistance proteins MDR1/P-glycoprotein, MRP1, the paper. We thank Birgit Juenemann and Boguslawa Sadowski and LRP in acute myeloid leukemia: a Southwest Oncology Group Study. Blood 1999; 94: 1086–1099. of the flow cytometry core facility at the University of Goettingen 18 Cole SP, Bhardwaj G, Gerlach JH, Mackie JE, Grant CE, Almquist and York Hildebrandt for technical assistance, as well as Andrew KC et al. Overexpression of a transporter gene in a multidrug- Entwistle for his editing and proofreading of thepaper. Financial resistant human lung cancer cell line. Science 1992; 258: support for this research was provided to GW from the Deutsche 1650–1654. Jose-Carreras Leukaemiestiftung and to BC from the research 19 Hurwitz SJ, Terashima M, Mizunuma N, Slapak CA. Vesicular funding program of the Faculty of Medicine, Georg-August- anthracycline accumulation in doxorubicin-selected U-937 cells: participation of lysosomes. Blood 1997; 89: 3745–3754. University Goettingen. 20 Altan N, Chen Y, Schindler M, Simon SM. Defective acidification in human breast tumor cells and implications for chemotherapy. Disclosure/Conflict of Interest J Exp Med 1998; 187: 1583–1598. None of the authors have any financial interest related to this 21 Klugbauer N, Hofmann F. Primary structure of a novel ABC work. transporter with a chromosomal localization on the band encoding the multidrug resistance-associated protein. FEBS Lett 1996; 391: 61–65. 22 Mulugeta S, Gray JM, Notarfrancesco KL, Gonzales LW, Koval M, Feinstein SI et al. Identification of LBM180, a lamellar body References limiting membrane protein of alveolar type II cells, as the ABC transporter protein ABCA3. J Biol Chem 2002; 277: 22147–22155. 1 Tallman MS, Gilliland DG, Rowe JM. Drug therapy for acute 23 Shulenin S, Nogee LM, Annilo T, Wert SE, Whitsett JA, Dean M. myeloid leukemia. Blood 2005; 106: 1154–1163. ABCA3 gene mutations in newborns with fatal surfactant 2 Fialkow PJ, Gartler SM, Yoshida A. Clonal origin of chronic deficiency. N Engl J Med 2004; 350: 1296–1303. myelocytic leukemia in man. Proc Natl Acad Sci USA 1967; 58: 24 Nagata K, Yamamoto A, Ban N, Tanaka AR, Matsuo M, Kioka N 1468–1471. et al. Human ABCA3, a product of a responsible gene for abca3 for

Leukemia Multidrug resistance in leukemia cells B Chapuy et al 1586 fatal surfactant deﬁciency in newborns, exhibits unique ATP transporters in childhood T-cell acute lymphoblastic leukemia. hydrolysis activity and generates intracellular multilamellar Mol Cancer Ther 2006; 5: 1986–1994. vesicles. Biochem Biophys Res Commun 2004; 324: 262–268. 27 Steinbach D, Gillet JP, Sauerbrey A, Gruhn B, Dawczynski K, 25 Yasui K, Mihara S, Zhao C, Okamoto H, Saito-Ohara F, Tomida A Bertholet V et al. ABCA3 as a possible cause of drug resistance in childhood acute myeloid leukemia. Clin Cancer Res 2006; 12: et al. Alteration in copy numbers of genes as a mechanism for 4357–4363. acquired drug resistance. Cancer Res 2004; 64: 1403–1410. 28 Stahlman MT, Besnard V, Wert SE, Weaver TE, Dingle S, Xu Y 26 Efferth T, Gillet JP, Sauerbrey A, Zintl F, Bertholet V, de et al. Expression of ABCA3 in developing lung and other tissues. Longueville F et al. Expression proﬁling of ATP-binding cassette J Histochem Cytochem 2007; 55: 71–83.

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

Leukemia