ABCG2 is required to control the sonic hedgehog pathway in side population cells with stem-like properties

Authors


Abstract

The Sonic Hedgehog (Hh) pathway has been implicated in the maintenance of stem or progenitor cells in many adult tissues. Importantly, abnormal Hh pathway activation is also associated with initiation of neoplasia, but its role in tumor growth is still unclear. Here, we demonstrate that cyclopamine, a plant-derived alkaloid product used to inhibit the Hh signaling pathway, reduces the Side Population (SP) obtained by Hoechst 33342 (Ho342) dye measurements. In addition, cyclopamine is able to modulate, along with oxysterols and other products, the ABCG2 transporter by increasing Ho342 and mitoxantrone uptake. Therefore, if the SP is solely measured as a Ho342 dye extruding fraction, this may be significantly modulated by the inhibition of ABCG2 transport fraction, independently from the action of cyclopamine on the Hh pathway. Our results indicate that ABCG2 may act in the upstream regulation of the Hh signaling pathway to protect the stemness of the SP compartment, giving support to the cancer stem cell hypothesis and suggesting that ABCG2 is not only critical for increased resistance to anticancer agents. © 2011 International Society for Advancement of Cytometry

Cyclopamine is a plant-derived alkaloid product used as an Hedgehog (Hh) antagonist (1–7) with potential effectiveness in tumor therapies in several xenograft models of medulloblastoma (8), prostate (9), small cell lung cancer (10), and digestive tract tumors (11). It has been previously shown that the combination of low-dose chemotherapy and cyclopamine results in enhanced cell killing (8). Interestingly, the effects of oncogenic mutations in the Hh pathway can also be reversed by cyclopamine, which inhibits Hh signaling by targeting the product of the proto-oncogene Smoothened (SMO) (12).

Hh signaling pathway regulates cell proliferation, tissue polarity, and cell differentiation during normal development, and abnormal signaling of this pathway has been associated with cancer. Human tumors such as those of basal cell carcinoma, medulloblastoma, rhabdomyosarcoma, and others, are associated with activating mutations of SMO (13) or with inactivating mutations of Patched (PTCH) (14). SMO and PTCH are involved in the Hh pathway, and mutations of these proteins may lead to an increased activity of the Hh response pathway (15). It has been suggested that Hh pathway activity may play a role in the maintenance of the capacity of tumor stem cells, favoring self-renewal, and proliferation of their progeny (16). Because Hh signaling regulates progenitor cell fate in normal development and homeostasis, aberrant pathway activation might be involved in the maintenance of such populations in cancer. Moreover, it has also been proposed that in stem cell populations, downregulation of PTCH1 in the presence of high-levels of SMO expression renders such cells sensitive to Hh ligand and results in pathway activation, leading to stem cell population expansion (17). Thus, Hh pathway inhibition would deplete the clonogenic tumor fraction through terminal differentiation rather than cytotoxicity, and such activation of Hh signaling would have the capacity to expand cancer stem cells without triggering terminal differentiation.

On this basis, a number of different research groups have explored the potential role of cyclopamine in Hh signaling in cancer stem cell biology, showing that the Side Population (SP) can be depleted following cyclopamine treatment (18). SP cells are rare stem cells with expression of ABCG2, a multidrug ABC transporter discovered using mitoxantrone-resistant breast cancer cells (19). Although many ABC transporters have been identified as drug-resistance proteins, they are all expressed in normal tissues (20). ABCB1 and ABCG2 transporters are respectively highly conserved in normal CD34-positive (21) and CD34-negative (22) stem cells and may play a role not only in protecting these cells from cytotoxic agents, but also from endogenous or physiological substrates. ABCG2 confers the SP phenotype and enhances hypoxic cell survival by reducing heme or porphyrin accumulation and having a key role in hematopoietic stem cells concentrated in hypoxic areas (23). Moreover, enforced expression of ABCG2 inhibits hematopoietic development and it has been proposed as a regulator in extracellular signals that influence stem cell interactions with the microenvironment (22). Possible roles for ABC transporters expressed in normal cells have been suggested by disruption of their function (24).

Here, we show that ABCG2, independently of how Ptch regulates Smo phosphorylation, conformation, and trafficking, is likely to be modulating the Hh pathway, and maybe orchestrating some other gene expression programs by means of an active efflux of substrates involved in intercellular and/or intracellular cell signaling. The regulation of stem cell signaling by ABC transporters would be an additional way of understanding how complex cell signaling is, adding more complexity at the upstream signaling pathway.

Materials and Methods

Cell Lines and ABCG2 cDNA

The human KB line (ATCC, Rockville, MD) was originally thought to be derived from an epidermal carcinoma of the mouth, but was subsequently found as a HeLa contaminant. Parental cells and the derived resistant sublines were maintained in complete medium consisting of DMEM (Gibco BRL, Grand Island, NY) supplemented with 20% fetal calf serum (Gibco BRL), 2 mM L-glutamine, 1 mM sodium pyruvate, 15 U ml−1 penicillin, and 15 ng ml−1 streptomycin (Biological Industries, Kibbutz Beth Haemek, Israel). In this study we have used different human brain astrocytoma cell lines. MOG-G-CCM and U87-MG cells were obtained from the European Collection of Cell Cultures (ECACC). GOS3 cells were obtained from the Deutshche Sammlung von Mikroorganismen und Zellkulturen GMBH (DSMZ). LN405 and SW1783 were obtained from the American Type Culture Collection (ATCC). For the transfection experiments, KB cells were transfected with the full-length ABCG2 cDNA (R482 variant; MXRA cells) by a cationic liposome-mediated method (GeneShuttle-20, Q-BIOgene) (following the manufacturer's instructions). KB-transfected cells were continuously maintained in complete medium containing the selected drug. Prior to the flow cytometry experiments, exponentially growing cells were detached using trypsin/EDTA (PAA Laboratories, Pasching, Austria), rinsed twice at 1,500 rpm for 5 min, resuspended in cold PBS supplemented with 0.1% NaN3 and 2% bovine serum albumin (PBS-BSA) and kept on ice. The ABCG2 DNA construct above mentioned was kindly provided by Dr. S. E. Bates (National Cancer Institute, Bethesda, MD).

Chemicals, Antibodies, and Drugs

Mitoxantrone (MTX), Hoechst 33342 (Ho342), propidium iodide, cyclopamine, temozolomide (TMZ), staurosporinone (K252c), 11-ketocyclopamine (jervine), 20(alpha)-hydroxycholesterol, 22(S)-hydroxycholesterol and 25-hydroxycholesterol were purchased from Sigma-Aldrich (St. Louis, MO). The 24(S)-hydroxycholesterol was purchased from Enzo Life Sciences. ABCG2 siRNAs, RIPA buffer, acrylamide/bys acrylamide 30% BioReagent 29:1 solution and Tween® 20, were purchased from Sigma Aldrich (St. Louis, MO). Lipofectamine RNAiMAX and Opti-MEM® I Medium were purchased from Invitrogen Life Science. Monoclonal anti-ABCG2 (BXP-53, ab24115), polyclonal anti-PTCH (ab39266), polyclonal anti-SMO (ab38686), rabbit polyclonal secondary antibody to rat IgG - H&L (HRP, ab6734), and Goat polyclonal Secondary Antibody to Rabbit IgG-H&L (HRP, ab6721), were purchased from Abcam plc (Cambridge, UK). ECL Plus Western Blotting Detection System and Full Range Rainbow™ Molecular Weight Markers were purchased from Amersham (GE Healtchare). Pierce® BCA Protein Assay Kit was purchased from Thermo Scientific (Rockford, IL).

Western Blot

Cells were detached using trypsin/EDTA (PAA laboratories, Pashing, Austria), rinsed at 400g for 5 min at room temperature, and the pellet was frozen with dry ice and kept at −80°C until their use. Pellet was thawed and RIPA buffer was immediately added (1 ml for each 10 × 106 cells) with freshly added protease inhibitor cocktail and PMSF (10 μl of 10 mg ml−1 PMSF stock for each ml of RIPA buffer). Samples were gently resuspended and incubated at 4°C for 30 min under shaking. Next, samples were centrifuged at 12,000g for 20 min at 4°C. The supernatant represented the total cell lysate and was transferred to a new microfuge tube (previously cooled at 4°C). Total protein concentration was measured in the extract by Pierce® BCA Protein Assay Kit (Thermo Scientific, Rockford, IL).

Samples were prepared in loading buffer containing 50 mM Tris pH 6.8, 2% (w/V) SDS, 10 mM DTT, 10% (v/v) glycerol, and 0.2% (w/v) β-mercaptoethanol, incubated at 95°C for 5 min and centrifuged (short spin at 14,000 rpm) for suitable collection of the sample in the tube. Samples were applied to a SDS-PAGE (SDS acrylamide gel electrophoresis 10%) and electrophoresis was performed for ∼60 min at 160 V. Stained proteins of known molecular mass (range 12–225 kDa) were run simultaneously as standards. The electrophoretically separated proteins were transferred onto PVDF membranes (BioRad) by humid electrotransference with the Mini Trans-Blot Cell System (BioRad) at 300 mA for 1 h. After blotting, PVDF membranes were dried and stained with Ponceau red Staining solution and immediately washed twice with TBS-Tween (0.1%) (TBST). Membranes were blocked for 2 h at room temperature in TBST (containing 8% of non-fat dried milk) and then washed three times during 10 min in TBST at room temperature. Membranes were incubated with the appropriate dilution of the corresponding primary antibody in TBST (1% of non-fat dried milk). Incubation dilutions were as follows: Polyclonal anti-SMO (1:1,000 dilution), Polyclonal anti-Patched/PTCH (1:1,000 dilution) and monoclonal anti-ABCG2 (1:250 dilution). All incubations were performed overnight at 4°C. After washing three times during 10 min in TBST, membranes were incubated with a secondary antibody in TBST (1% of non-fat dried milk), HRP-labelled anti-rabbit IgG (1:8,000 dilution) for Polyclonal anti-Smo and Polyclonal anti-PTCH and with secondary antibody HRP-labeled anti-rat IgG (1:10,000) for Monoclonal anti-ABCG2 at room temperature for 1 h. Membranes were washed with TBST again before detecting bound antibodies using an ECL Plus Western Blotting Detection System (GE Healthcare) according to the manufacturer's recommended procedure. Immunoreaction signals were captured and analyzed using an ImageQuant™ LAS 4000 system by Fujifilm Life Science, USA.

siRNA Forward Transfection

Cells were seeded in a six-well tissue culture test plates (Nunc™) at a final concentration of 5 × 104 cells per well in 2.5 ml antibiotic-free normal growth medium supplemented with FBS. Cells were incubated at 37°C in a CO2 incubator until 50–60% confluence for 24 h. The following solutions were prepared: Solution A: For each transfection, 1.5 μl of each siRNA duplex (Mix of three siRNA duplexes) was diluted into 250 μl of Opti-MEM® I Medium without serum (Invitrogen Life Science), Solution B: For each transfection, 5 μl of lipofectamine RNAiMAX (Invitrogen Life Science) was diluted into 250 μl Opti-MEM® I Medium. The siRNA duplex solution (Solution A) was then added directly to the diluted transfection reagent (Solution B), mixed gently and then incubated for 20 min at room temperature. After incubation, the final volume for each transfection (500 μl) was added to the cells (final RNA duplex concentration of 50 nM) in a total volume of 3 ml of antibiotic-free normal growth medium supplemented with FBS. Cells were incubated at 37°C in a CO2 incubator and the medium was changed after 6 h of treatment. Finally, cells were subsequently incubated for 48 h, at which time the medium was aspirated and cells were tripsinized immediately for RNA isolation.

siRNA sequences used were as follows:

  • Oligo 4247837 SASI_Hs02_00338396, sense: 5′-GUCAACUCCUCCUUCUACA-3′

  • Oligo 4247838 SASI_HsO2_00338396_AS: 5′-UGUAGAAGGAGGAGUUGAC-3′

  • Oligo 4247839 SASI_HsO1_00136088, sense: 5′-GUCUAAGCAGGGACGAACA-3′

  • Oligo 4247840 SASI_HsO1_00136088_AS: 5′-UGUUCGUCCCUGCUUAGAC-3′

  • Oligo 4247841 SASI_HsO1_00136087, sense: 5′-GGUUAUCACUGUGAGGCCU-3′

  • Oligo 4247842 SASI_HsO1_00136087_AS: 5′-AGGCCUCACAGUGAUAACC-3′

MISSION siRNA Universal negative control #1 was used for the siRNA transfection assay.

Flow Cytometry

Data were collected using a MoFlo® (Beckman-Coulter) equipped with three lasers operating at 351, 488, and 633 nm. Filter combination for Ho342 analysis was: 440DLP, BP405/30 (blue), and BP670/40 (red). Ho342 signal was measured using linear scale at 30 mW. Acquisition was stopped when 100,000 live-gated events were collected. Propidium iodide was collected through a band pass filter of 630/30. MXR fluorescence was measured using a 740LP filter. For drug uptake and retention assays, acquisition was stopped when 20,000 gated events were collected in the fluorescence-cell count histogram. Gating was based on forward scatter and side scatter dot plots, by encircling populations with amorphous regions and then excluding dead cells (life-gate) by propidium iodide counter-staining. Data were stored as listmode files and analyzed using FlowJo Software (Tree Star).

Ho342 Labeling

Cells were resuspended in prewarmed Dulbecco's modified Eagle Medium (DMEM) supplemented with 2% heat-inactivated fetal calf serum (FCS) and 10 mM HEPES at a concentration of 1–2 × 106 cells ml−1. Ho342 was added at a concentration of 5 μg/106 cells. Samples were incubated in a water bath at 37°C for 2 h in agitation, in the dark. Cells were centrifuged at 483 rcf for 6 min at 4°C and resuspended in cold Hank's Buffered Salt Solution (HBSS)/FCS/HEPES at a concentration of 1–2 × 107 cells ml−1. Samples were kept in the dark at 4°C until analysis. Propidium iodide was added at a concentration of 5 μg ml−1 to exclude dead cells. In order to remove cell aggregates, samples were filtered through a 50-μm nylon mesh prior to acquisition.

Drug Uptake and Retention Assays

MXRA cells were seeded in six-well tissue culture testplates (Nunc™) at a final concentration of 105 cells ml−1, 1 ml total volume per well, 24 h before the dye uptake/retention experiments to allow attachment to the surface and to achieve the conditions allowing for optimal growth. Drug uptake was determined by adding either Ho342 or mitoxantrone to the culture medium for 3 h at 37°C, in the presence or absence of staurosporinone (K252c), 11-ketocyclopamine (jervine), 20 (alpha)-hydroxycholesterol, 22(S)-hydroxycholesterol, 24(S)-hydroxycholesterol, and 25-hydroxycholesterol in separate wells, at different concentrations. After 3-h incubation, cells were washed and resuspended in dye-free culture medium and the other drugs were maintained to evaluate their effect on Ho342 and mitoxantrone (MTX) retention. Concentrations used were 0.2–20 μM for K252c, and the rest of the molecules ranged from 1 to 10 μM. All experiments were run in triplicate.

Real-Time PCR Analysis

RNA was isolated using Illustra RNAspin Mini RNA Isolation Kit (GE Healthcare) and converted to cDNA using SuperScript™ II RNase H Reverse Transcriptase (Invitrogen). Quantitative rtPCR was performed using an IQ5 Multicolor Real-Time PCR Detection System (BioRad), IQTM SYBR Green Supermix (BioRad) and gene specific primers. All real-time data was normalized respect to the housekeeping gene HPRT. Primers sequences used were as follows:

  • ABCG2, sense: 5′-CGACAGCTTCCAATGACCTGA-3′

  • ABCG2, antisense: 5′-ACCAGGTTTCATGATCCCATTGAT-3′

  • SMO, sense: 5′- CAGCTTCCGGGACTATGTGCTATG-3′

  • SMO, antisense: 5′-GAAGGCTCGGGCGATTCTTG-3′

  • PTCH, sense: 5′-GGCAGCGGTAGTAGTGGTGTTC-3′

  • PTCH antisense: 5′-TGTAGCGGGTATTGTCGTGTGG-3′

To further validate the suitability of the primers, the PCR products were sequenced.

Results

Cyclopamine Modulates ABCG2 Efflux Activity

We first evaluated the modulatory effect of cyclopamine on ABCG2 function by measuring the intracellular accumulation of two fluorescent substrates of ABCG2: Ho342 and mitoxantrone. As shown in Figure 1, 10 and 30 μM cyclopamine blocked the efflux function of ABCG2 and thereby increased the accumulation of Ho342 and MTX in ABCG2-expressing MXRA cells. We used cyclopamine as a potential substrate competitor with Ho342 and MTX in independent experiments. Thus, we confirmed that cyclopamine is modulating ABCG2 activity.

Figure 1.

Effect of cyclopamine on Ho342 and mitoxantrone retention. The function of the MXRA cells (KB cells overexpressing the ABCG2 transporter) was examined. We first investigated Ho342 efflux ability in the presence of cyclopamine. KB cells transfected with ABCG2- R482A were incubated with 1 μg ml−1 Ho342 for 3 h at 37°C in the presence or absence of 10 or 30 μM cyclopamine. Afterward, cells were incubated for 12 h at 37°C in absence of Ho342 but maintaining cyclopamine. Then, cells were analysed using flow cytometry and showed a heterogeneous fluorescent profile. MXRA cells incubated with 30 μM cyclopamine showed higher intracellular accumulation of Ho342 when compared with 10 μM cyclopamine treated cells, indicating a progressive slow decrease in Ho342 fluorescence when cyclopamine was used at the highest concentration (a). Then we tested the reversal efficacy of cyclopamine in the presence of 0.8 μM MXR (mitoxantrone) for 3 h at 37°C. The same procedure as when using Ho342 was followed for MXR experiments and flow cytometry confirmed its reversal efficacy on ABCG2 (b).

Cyclopamine Effect on SP Studies

The effect of cyclopamine on the accumulation of Ho342 was tested to characterize its modulatory effect on Ho342 mediated transport using functional flow cytometry. The accumulation of Ho342 was initially studied in ABCG2 overexpressing MXRA cells. The cells were incubated as we (25) and others (26) previously described in the presence of either 10 or 30 μM of cycloplamine. This drug (30 μM) and temozolomide (100 μM) were added to MXRA cells during the whole incubation process (2 h) in the presence of Ho342, and showed that only cyclopamine had an effect on the accumulation of Ho342 in MXRA cells (Fig. 2). Since MXRA cells were enforced to express ABCG2, the observed SP cells were not true stem cells. Hence, we resorted to characterizing the accumulation of Ho342 using human brain cancer cell lines previously described to be enriched in stem-like cells. We used GOS3, LN405, MOG-G-CCM, SW1783, and U87-MG cells for cell sorting experiments. Primary cells were stained with Ho342 as described. Relative numbers of initially detected SP cells ranged from 0.002 to 2.86% of the PI-negative cells within the life gate. Subsequently, cells were sorted to obtain enriched products in SP cells (ranging from 40 to 80%). Accordingly, for the enrichment or purification of SP cells, a representative experiment using MOG-G-CCM cells is shown (Fig. 3). SP cells were initially detected in low numbers in a small percentage (0.99%) prior to cell sorting experiments. Isolated SP cells were cultured for 2 weeks following sorting under the same conditions as unsorted cells, giving 5.42% of SP cells. Sorted cells were cultured again for 2 weeks, giving 27.8% of SP cells.

Figure 2.

Efflux of Ho342 is impaired by cyclopamine and gives inappropriate SP profiles in MXRA cells. Representative analyses of the SP profiles using MXRA cells. KB cells were driven to have an enforced ABCG2 expression, giving the MXRA cells (see Materials and Methods). MXRA cells were incubated with Ho342 as usually for SP analyses and in the presence of 30 μM cyclopamine or with the same concentration of cyclopamine and 100 μM temozolomide. Percentages of SP cells are shown in the cytograms containing the relevant cell population. Although ABCG2-transfected KB cells cannot be considered true SP cells, they gave a very similar fluorescent profile to these cells (a). (b) shows the effect of cyclopamine on cells that classically appear within the SP, giving a decrease of 27.8% of SP cells. Temozolomide did not have a significant effect on the accumulation of Ho342 (c).

Figure 3.

Representative sorting experiments with cells supposed to have true SP cells. SP cells were initially detected in low numbers (0.99%). Isolated SP cells were cultured for 2 weeks following sorting under the same conditions as unsorted cells, giving 5.42% of SP cells. Sorted cells were cultured again for 2 weeks, giving 27.8% of SP cells. For competitive experiments using cyclopamine, MOG-G-CCM cells were sorted again to give a minimum of 40% of SP cells.

For competitive experiments using cyclopamine, MOG-G-CCM and LN405 cells were sorted again. Only cell cultures with a minimum of 40% of SP cells were prepared in order to study the effect of cyclopamine on Ho342 uptake. SP-enriched MOG-G-CCM and LN405 cells were stained with Ho342 as described in the presence or absence of 30 μM cyclopamine or with the same concentration of this drug and 100μM temozolomide. MOG-G-CCM cells incubated with Ho342 gave 41.2% SP cells, whereas cells incubated either with cyclopamine alone or in combination with temozolomide, gave 11.2 and 15.6% of SP cells respectively (Fig. 4a). LN405 cells incubated with Ho342 gave 67.7% SP cells, whereas cells incubated either with cyclopamine alone or in combination with temozolomide, gave 6.22 and 5.8% of SP cells, respectively (Fig. 4b). Hence, this significant decrease of SP cells was a consequence of adding cyclopamine, whereas temozolomide had no effect on Ho342 efflux. This phenomenon probably occurred independently of the cytotoxic mechanisms of the drugs we used, indicating that the decrease in MOG-G-CCM and LN405 SP cells was not a consequence of the depletion of the SP compartment, but also a direct consequence of the molecular mechanisms involved in the affinity of cyclopamine and Ho342 for ABCG2. Furthermore, we analyzed the effect of cyclopamine in unsorted populations with a low content of SP cells. We incubated SW1783 cells again with Ho342 in the presence or absence of 30 μM cyclopamine or with the same concentration of this drug and 100 μM temozolomide. SW1783 cells incubated with Ho342 gave 0.48% SP cells, whereas cells incubated either with cyclopamine alone or in combination with temozolomide both gave 0.16% of SP cells (Fig. 5a). Taken together, our results showed a significant decrease in the SP compartment as a consequence of adding 30 μM cyclopamine, ranging from 57.47 to 90.81% (median = 69.74%) (Fig. 5b). These results were consistent and reproducible even when low numbers of SP cells were detected. Cells were incubated again in the presence of temozolomide alone (not in combination with cyclopamine), confirming that temozolomide had no effect on Ho342 efflux (Fig. 5c).

Figure 4.

Reversal effect of cyclopamine on Ho342 accumulation using SP-enriched MOG-G-CCM and LN405 cells. Only cell cultures with a minimum of a 40% of SP cells were prepared for studying the effect of cyclopamine on Ho342 uptake. SP-enriched cells were stained with Ho342 in presence or absence of 30 μM cyclopamine or with the same concentration of this drug and 100 μM temozolomide. MOG-G-CCM cells incubated with Ho342 gave 41.2% SP cells, whereas cells incubated either with cyclopamine alone or in combination with temozolomide, gave 11.2 and 15.6% of SP cells respectively (a1 and a2). LN405 cells incubated with Ho342 gave 67.7% SP cells, whereas cells incubated either with cyclopamine alone or in combination with temozolomide, gave 6.22 and 5.8% of SP cells, respectively (b1 and b2). Both figures, a2 and b2, summarize the inhibition of ABCG2-mediated transport of Ho342 in the presence of 30 μM cyclopamine or with the same concentration of cyclopamine and 100 μM temozolomide.

Figure 5.

Reversal effect of cyclopamine on Ho342 accumulation using non SP-enriched cells. SW1783 cells were incubated with Ho342 in the presence or absence of 30 μM cyclopamine or with the same concentration of this drug and 100 μM temozolomide. SW1783 cells incubated with Ho342 gave 0.48% SP cells, whereas cells incubated either with cyclopamine alone or in combination with temozolomide both gave 0.16% of SP cells (a). (b) summarizes the inhibition of ABCG2-mediated transport of Ho342 in the presence of 30 μM cyclopamine or with the same concentration of cyclopamine and 100 μM temozolomide. Data from four representative experiments are expressed as median and interquartile range. (c) show representative SP cytograms of MXRA, LN405, and MOG-G-CCM cells incubated either with cyclopamine or with temozolomide, showing that temozolomide alone did not have a significant effect on the accumulation of Ho342.

Effect of Hydroxycholesterols (HCs), Jervine, and K252c on ABCG2 Activity

We next evaluated the modulatory effect of 20(alpha)-HC, 22(S)-HC, 24(S)-HC, 25-HC, 11-ketocyclopamine (jervine), staurosporinone (K252c) on ABCG2 function by measuring the intracellular accumulation of mitoxantrone in ABCG2-expressing MXRA cells. Each of the drugs tested had an effect on mitoxantrone transport. As shown in Figure 6, 25-HC, jervine and K252c blocked the efflux function of ABCG2 and thereby increased the accumulation of MTX in ABCG2-expressing MXRA cells. We used these drugs as potential substrate competitors with MTX and Ho342 (data not shown) in independent experiments. Thus, we confirmed that HCs, 11-ketocyclopamine and staurosporinone also modulate ABCG2 activity.

Figure 6.

Effect of hydroxycholesterols, jervine and K252c on mitoxantrone retention in MXRA cells. We next investigated mitoxantrone efflux ability in the presence of hydroxycholesterols, jervine and K252c. Drug uptake was determined by adding 0.8 μM mitoxantrone to the culture medium for 3 h at 37°C, in the presence or absence of 25-hydroxycholesterol, 11-ketocyclopamine (jervine), and staurosporinone (K252c) in separate wells. After 3-h incubation, cells were washed and resuspended in dye-free culture medium and the other drugs were maintained in order to evaluate their effect on mitoxantrone retention. Concentrations used were K252c (0.2–2 μM) and the rest of the molecules ranged from 1 to 10 μM, showing that hydroxycholesterols, 11-ketocyclopamine, and staurosporinone were also modulating ABCG2.

ABCG2, SMO, and PTCH1 Expression Levels in Isolated SP Cells

LN405, MOG-G-CCM, and SW1783 cells were used to determine whether ABCG2 expression (mRNA) was increased in SP cells isolated using flow cytometric sorting. Real-time PCR (rtPCR) using specific primers for the transporter was carried out, showing that ABCG2 was expressed at higher concentrations in SW1783 SP-cells, and progressively decreasing in MOG-G-CCM and LN405 SP-cells. Decreased expression levels of Smo and Ptch were observed, and appeared to be dependent on the differential expression pattern of ABCG2 (Figs. 7a and 7b). These results support the hypothesis that ABCG2 regulates Smo levels, whereas the decline in the expression of Ptch may represent a compensatory mechanism as a result of Smo downregulation. Furthermore, LN405, MOG-G-CCM, and SW1783 SP cells were incubated in the presence of 10 and 30 μM cyclopamine for 24 h. Using rtPCR, ABCG2, SMO, and PTCH1 expression levels remained almost unchanged after cyclopamine treatment (Fig. 7c). In addition, we obtained near identical flow cytometric cell cycle profiles of LN405, MOG-G-CCM, and SW1783 SP cells (Fig. 7d). For comparison, parallel cell cycle analyses were performed on identical conditions to those used for rtPCR. All together, these experiments suggest that ABCG2 may play a role in cyclopamine resistance.

Figure 7.

Expression of ABCG2, SMO and PTCH in SP-isolated cells, cell cycle analysis, and rtPCR following ABCG2 siRNA treatment of SP-enriched cells. Data from two representative experiments are reported as the mean of triplicate determinations ± S.D. Real-time PCR (rtPCR) of mRNA from LN405, MOG-G-CCM and SW1783 cells using specific primers for ABCG2, Smo and Ptch. a, rtPCR for ABCG2, SMO and PTCH in unselected cells. b, rtPCR for ABCG2, SMO and PTCH in SP fractions isolated from LN405, MOG-G-CCM and SW1783 cells. c, rtPCR for ABCG2, PTCH and SMO in SP-sorted cells incubated with 10 and 30 μM cyclopamine. d, Flow cytometric cell cycle analyses of SP-enriched LN405, MOG-G-CCM and SW1783 cells treated for 24h with 10 and 30μM cyclopamine. e, Representative experiment of ABCG2 reversal using siRNAS in LN405 SP cells (n = 4). f, rtPCR for SMO in KB and MXRA cells and in SP-enriched LN405, MOG-G-CCM and SW1783 cells following treatment with siRNA for ABCG2. g, Basal expression of PTCH and SMO by Western Blot analyses of KB, MXRA, LN405, MOG-G-CCM, SW1783 and SP-enriched LN405, MOG-G-CCM and SW1783 cells. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Reversion of ABCG2 Levels by Forward Transfection Using siRNAs

For ABCG2 knock-down experiments, Smo mRNA levels of KB and MXRA cells were also measured, since this would show how a controlled increase of ABCG2 expression affects Smo levels. First of all, SP cells were transfected with siRNAs and resulted in a partial decrease in ABCG2 protein expression, with a decrease in the mean fluorescence intensity from 36.35 ± 2.59 to 23.4 ± 1.88 a.u. (arbitrary units; Fig. 7e). Treated cells were then processed in parallel for the isolation of total RNA. Both KB and MXRA cells gave very low Smo mRNA levels and remained unchanged for KB treated cells when compared with its siRNA negative control used for the transfection assay. LN405, MOG-G-CCM, and SW1783 SP-cells gave the higher concentrations of Smo mRNA when compared with KB and MXRA cells, but Smo levels remained almost unchanged following siRNA treatment (Fig. 7f).

Protein Levels for Smo and Ptch by Western Blot Analysis

Since mRNA levels are not necessarily reliable indicators of true expression profiles, the protein levels of SMO and PTCH were also measured in KB, MXRA, LN405, MOG-G-CCM, and SW1783 cells. No significant differences in protein expression levels for KB and MXRA cells were observed. Protein levels were also tested in SP cells isolated using flow cytometric sorting. PTCH was overexpressed in MOG-SP and SW1783-SP cells. SMO was overexpressed in MOG cells and slightly overexpressed in SW1783 cells. In general, Smo protein was underexpressed in SP cells, whereas Patch was overexpressed, with the exception of LN405 cells (Fig. 7g).

Discussion

It has been reported that cyclopamine may act to deplete the SP compartment (17, 18, 27). Consistent with this idea is our demonstration that cyclopamine can markedly reduce the SP, an observation that would support the argument that SP cells are Hh pathway-dependent. Testing this concept, we looked at the possibility that cyclopamine might interact with ABCG2 and limit the optimal examination of SP cells, using the Hoechst-based assay. Our results clearly show that ABCG2 efflux function is modulated dose-dependently by cyclopamine when human cell models are used, either with an enforced expression of wild-type ABCG2 or with true SP cells. However, it cannot be ruled out that cyclopamine could have a dual mode of action, independently inhibiting Hh signaling and ABCG2-dependent efflux at the same time (28). Furthermore, this observation not only has a direct effect on SP measurements, but may also have a long after-effect through the modulation of signaling cascades in stem cells.

To date, SP cells can only be accurately identified using functional flow cytometry, based on the efflux of a fluorescent dye known as Ho342. Although cyclopamine and other drugs have long served for inhibiting the Hh pathway, there is no doubt that ABC transporters may play an important role apart from regulating drug toxicity in both normal and cancer cells (29). The steroidal nature of cyclopamine and its ability to disrupt cholesterol synthesis or transport suggested to us that it might affect the action of ABCG2. Cyclopamine can inhibit Hh pathway activity in the absence of Ptch function (12). Thus, ABCG2 may protect the proto-oncogene SMO, being a new potential “mechanism-based” associated regulator with the Hh pathway. In addition to some activating mutations that render Smo proteins resistant, the oncogenic SmoA1 protein has been reported as resisting suppression by Ptch (30), indicating that oncogenic Smo proteins may not be subject to normal regulation. Under these circumstances, our results support the idea that ABCG2 also regulates Smo proteins by inhibiting the mechanism of action of cyclopamine and maybe other molecules that may act upon Smo.

Although it has been suggested that Ptch might regulate lipid trafficking to repress Smo, no endogenous lipid regulators of Smo have yet been identified, nor has it ever been shown that Ptch controls lipid trafficking. In fact, Ptch may regulate Smo degradation by modifying the lipid composition of the endosomes through which Smo passes (31). Alternatively, we propose that ABCG2 is likely to be operating as a regulator of Smo independently of how its structural form is preserved. Therefore, a hypothetical role for ABCG2 in retranslocating Smo proteins to intracellular compartments cannot be ruled out. Whether or not ABCG2 also protects Ptch and regulates lipid trafficking was not evaluated in this work and further studies will be needed to discover how ABCG2 may also regulate the binding of different substrates to Ptch.

On the basis of our cyclopamine measurements, we explored the effects of different compounds classically used for Hh signaling pathway studies. In particular, oxysterols represent a class of potent regulatory molecules with remarkably diverse and important biological actions (32). Interestingly, oxysterols are activators of the Hh signaling pathway in pluripotent mesenchymal cells (33), which have not been reported to contain SP cells (34). In addition to the oxysterols that we tested, jervine and K252c also modulated mitoxantrone transport. Given the effect of oxysterol 20(S)-HC (35) and curcumin (36) in Notch pathway signaling, and given that oxysterols and curcumin are ABCG2 substrates (37), it is possible that ABCG2 may act not only as a firewall to regulate Hh pathway signaling, but also to regulate Notch signaling as well as other important signaling pathways highly conserved in stem cells, such as Wnt. Our results seem to contradict the findings of Bleau et al (38) regarding glioma tumor stem-like cells, and a reasonable alternative explanation of their observations could be that ABCG2 may also act as a regulator of the PTEN/PI3K/Akt pathway in SP cells.

The following model (Fig. 8) would be consistent with what is already known about Hh regulation and with our new observations. ABCG2 may efflux different substrates able to interact with Smo and maybe alters lipid trafficking through cell membranes. Hence, the mechanisms by which ABCG2 differentially activates low- or high-level Hh signaling may rely on the efflux of substrates able to bind Smo and other proteins, and even explain different activation levels in the presence of mutated or truncated Smo proteins. The association of Hh and ABCG2 would be an additional way of understanding how complex cell signaling is in stem cell regulation, adding more complexity at the upstream signaling pathway.

Figure 8.

Hypothetical model for ABCG2 regulation of Hedgehog Pathway in SP stem cells. PTCH, SMO, and GLI have been identified in the Hedgehog signaling cascade, playing a key role in the control of stem cell proliferation. Moreover, aberrant activation of the Hh pathway has been associated with different cancers. Here we propose a model for Hh regulation in light of our data, with the unexpected role of ABCG2. SMO activity can be modulated by synthetic molecules (i.e., cyclopamine) and endogenous metabolites (i.e., oxysterol derivatives). Thus, cyclopamine results led us to investigate the modulation of SMO by ABCG2. Our results suggest that ABCG2 is directly acting as a firewall on SMO and maybe also acting on other potential targets involved in Hh signaling cascade, preventing their binding in a concentration-dependent manner, and adding more complexity to Hh regulation. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

We propose that Hh pathway activity in SP stem-like cells can be regulated through the activity of ABCG2. Thus, the ABC transporters expressed in stem cells, such as ABCG2 in SP cells or ABCB1 in human hematopoietic CD34-positive cells may protect not only against xenobiotics or anticancer compounds, but also be able to protect the stem compartment from a series of ligands involved in transduction signaling. In fact, ABC transporters may have an additional and unexpected role in preserving stem cells and cancer stem cells from physiological (i.e., oxysterols) and synthetic substrates resulting in pathway activation or inactivation. This hypothesis should be relevant for preclinical and clinical studies of Hh antagonist therapies (and others) aimed at avoiding self-renewal and long-term survival of tumor stem cells. If true, ABCG2 expression will be crucial for maintaining the SP compartment undifferentiated by inhibiting transduction signaling through ABCG2 activity. These results can be relevant for understanding how stem cells but also how cancer stem cells can proliferate while maintaining the stem cell phenotype.

Acknowledgements

The authors thank Drs. S. Bates and Rob Robey (National Cancer Institute, Bethesda, MD, USA) for providing the ABCG2 DNA constructs and for all their help. They also thank Noèlia Purroy for technical assistance in preparing samples.

Ancillary