A rapid screening system evaluates novel inhibitors of DNA methylation and suggests F-box proteins as potential therapeutic targets for high-risk neuroblastoma
Livius Penter & Bert Maier & Ute Frede &
Benjamin Hackner & Thomas Carell &
Christian Hagemeier & Matthias Truss
Abstract After extensive research on radiochemotherapy, 5- year survival rates of children with high risk neuroblastoma still do not exceed 50%, owing to adverse side-effects exem- plified by doxorubicin-induced cardiomyopathy. A promising new approach is the combination of conventional therapies with specific modulation of cell signaling pathways promot- ing therapeutic resistance, such as inhibition of aberrant kinase activity or re-expression of silenced tumor suppressor genes by means of chromatin remodeling. In this regard, we established a system that allows to identify potential drug targets as well as to validate respective candidate inhibitors in high-risk neuroblastoma model cell lines. Cell culture, drug exposure, shRNA-mediated knockdown and pheno- type analysis are integrated into an efficient and versatile single well-based protocol. By utilizing this system, we assessed RG108, SGI-1027 and nanaomycin A, three novel DNA methyltransferase inhibitors that have not been tested in neuroblastoma cell lines so far, for their potential of synergistic anti-tumor activity in combination with doxoru- bicin. We found that, similarly to azacytidine, SGI-1027 and nanaomycin A mediate synergistic growth inhibition with
doxorubicin independently of N-Myc status. However, they display high cytotoxicity but lack global DNA demethylation activity. Secondly, we conducted a lentiviral shRNA screen of F-box proteins, key regulators of protein stability, and identi- fied Fbxw11/β-TrCP2 as well as Fbxo5/Emi1 as potential therapeutic targets in neuroblastoma. These results comple- ment existing studies and underline the reliability and versa- tility of our single well-based protocol.
Keywords Neuroblastoma . F-box proteins . DNA methyltransferase inhibitors . RNA-interference . Screening system . Doxorubicin
Introduction
Even after extensive research, the 5-year survival rate of children with high-risk neuroblastoma still does not exceed 50 %. Standard therapeutic regimens include high-dose che- motherapy, irradiation, myeloablative chemotherapy with au-
Electronic supplementary material The online version of this article (doi:10.1007/s11523-014-0354-5) contains supplementary material, which is available to authorized users.
L. Penter (*) : U. Frede : C. Hagemeier : M. Truss (*) Labor für Pädiatrische Molekularbiologie, Charité – Universitätsmedizin Berlin, 10117 Berlin, Germany
e-mail: [email protected] e-mail: [email protected]
B. Maier
Institut für Medizinische Immunologie, AG Chronobiologie, Charité – Universitätsmedizin Berlin, 10115 Berlin, Germany
B. Hackner : T. Carell
CIPSM, Fakultät für Chemie und Pharmazie,
Ludwig-Maximilians-Universität, 81377 München, Germany
tologous, hematopoietic stem cell transplantation, retinoic acid, and anti-GD2 immunotherapy. Main limitations of these treatment options are therapy-induced complications like in- fections and adverse side effects of employed drugs [1, 2].The anthracycline doxorubicin is an example of an effective che- motherapeutic agent whose clinical usage is dose-limited due to concomitant cardiotoxicity following perturbations of cal- cium homeostasis, formation of iron complexes as well as radical oxygen species, mitochondrial dysfunction and dam- age to cell membranes [3]. As the risk of doxorubicin-induced congestive heart failure increases in a dose-dependent manner, it would be highly desirable to selectively increase doxorubi- cin sensitivity of neuroblastoma cells in order to improve the therapeutic efficacy of this compound [4].
Targeted therapy with small molecules offers an approach to reverse drug resistance by interference with oncogenic signaling. This was first demonstrated with imatinib, which improved the outcome of patients with chronic myeloid leu- kemia dramatically by inhibition of the causative Bcr-Abl tyrosine kinase [5]. The most promising drug target in neuro- blastoma so far is anaplastic lymphoma receptor tyrosine kinase (ALK). Crizotinib, an inhibitor of ALK, is currently being evaluated in a phase II clinical trial. However, only 6 to 10 % of patients will potentially be eligible for this treatment option as it only applies to cases of familial neuroblastoma resulting from heterozygous ALK mutations [6]. Therefore, in the future, it will be necessary to identify novel therapeutic targets for patients with high-risk neuroblastoma, for instance, using systematic functional screening. F-box proteins, key regulators of protein stability that mediate proteosomal deg- radation through ubiquitination, are a family of genes that might provide additional therapeutic drug targets as they are involved in numerous cellular signaling pathways controlling proliferation, cell growth, and differentiation. Given the can- cer relevant functions of this subclass of genes and their poor systematic characterization so far, it is impertinent to assess them as potential drug targets for neuroblastoma patients.
Besides targeted therapy by pathway interference, epige- netic modifications underlying tumorigenesis have moved into the scope of oncologic research in recent years. They offer an innovative strategy for overcoming drug resistance through reexpression of silenced tumor-suppressor genes by reversal of DNA hypermethylation or repressive chromatin marks [7]. The clinical feasibility of this concept was first demonstrated by azacytidine, an inhibitor of DNA methyl- transferases, which has improved therapeutic outcome of pa- tients with myelodysplastic syndrome significantly [8]. Re- cent in vivo studies have shown that azacytidine is also able to sensitize neuroblastoma cell lines to treatment with doxorubi- cin [2]. Unfortunately, efforts to utilize azacytidine for patients with pediatric malignancies including neuroblastoma had to be abrogated prematurely in a phase I trial due to adverse side effects at a clinically ineffective dose [9].
As the concept of interference with DNA methylation patterns remains valid, the identification of new inhibitors of DNA methyltransferases with less toxicity would mark im- portant progress. Several novel inhibitors of DNA methyl- transferases have been identified, which to our knowledge have not yet been tested in a neuroblastoma context: RG108 inhibits DNA methyltransferases with reportedly little adverse cytotoxicity [10]. SGI-1027 inhibits DNMT1, DNMT3a, as well as DNMT3b, again without significant cytotoxicity [11]. Nanaomycin A selectively inhibits DNMT3b and mediates anti-tumor activity in different cancer cell lines [12].
Here, we describe a flow cytometry-based approach that allows to rapidly assess the potential of novel compounds and therapeutic targets to increase doxorubicin-induced cytotoxicity
in neuroblastoma cell lines in a single-well format. Using this system, we were able to demonstrate anti-proliferative activity of the aforementioned DNA methyltransferase inhibitors in the neuroblastoma cell lines SH-EP and SK-N-AS (N-Myc single copy) as well as LAN-1 and SK-N-BE(2) (N-Myc amplifica- tion). Combination of SGI-1027 and nanaomycin Awith doxo- rubicin led to an increase of cytotoxicity. A functional screen of 275 short hairpin RNA (shRNA)-mirs against F-box proteins identified Fbxo5/Emi1 and Fbxw11/β-TrCP2 as potential ther- apeutic targets for neuroblastoma patients.
Methods and materials Cell culture
SH-EP and SK-N-AS cells provided by C. Dame, Charité – Universitätsmedizin Berlin, were grown in RPMI 1640 (Gibco). LAN-1 and SK-N-BE(2) cells provided by K. Schmelz, Charité – Universitätsmedizin Berlin, were grown in RPMI 1640 (Gibco). HEK293T cells provided by A. Kramer, Charité – Universitätsmedizin Berlin, were grown in DMEM (Gibco). Media were supplemented with 10 % fetal bovine serum (FBS) (Lonza), 1 % penicillin-streptomycin, and 1 % Glutamax (both Gibco).
During assays with inhibitors, cells were grown in presence of 600 nM azacytidine (Sigma), 10 μM RG108, 1 μM SGI- 1027 (Miltenyi), or 30 nM nanaomycin A (NCI), if not stated otherwise. Media were replaced daily. Exposure to doxorubi- cin (Sigma-Aldrich) was performed for 24 h during experi- ments with inhibitors of DNA methyltransferases and for 72 h during experiments with RNA-interference.
For inhibition of DNA methylation, cells were incubated with inhibitors for 3 days prior to doxorubicin treatment as previously described [2]. Low-dose doxorubicin was applied on SK-N-AS, LAN-1, SK-N-BE(2), and SH-EP cells at con- centrations of 30 and 300 nM for the latter, in order to study the effects of additional compounds. After 24 h of exposure to doxorubicin, cells were analyzed by flow cytometry. In the screen, SH-EP cells were incubated with 300 or 1000 nM doxorubicin for 3 days without replacement of media.
RNA-interference
shRNA-mir expression vectors from the Open Biosystems pGIPZ shRNA-mir library were obtained from A. Kramer, Charité – Universitätsmedizin Berlin, the PLL p53 shRNA expression vector from R. Wäsch, Universitätsklinikum Frei- burg [13], and packaging vectors psPAX2, as well as pMD2.G from D. Trono, École Polytechnique Fédérale de Lausanne.
For virus production, approximately 20,000 HEK293T cells were seeded per 96-well 24 h prior to transfection. Cells were transfected with 0.1 μg psPAX2, 0.06 μg pMD2.G, and
0.14 μg vector plasmid in 52 μl Opti-MEM (Gibco) and 0.52 μl Lipofectamine 2000 (Invitrogen). Viral supernatant (150 μl per 96-well) was harvested from HEK293T cells cul- tured in EMEM (Lonza), supplemented with 0.1 % gentamycin, 10 % FBS (both Lonza), 1 % non-essential amino acids, 1 % sodium pyruvate, 1 %L-alanyl-L-glutamine, and 2 % sodium bicarbonate (all Biochrome) 36 h after transfection.
Transduction of reporter cells in 96-well plates was per- formed with 20 μl viral supernatant in the presence of 8 μg/ml polybrene (Sigma) for 8 h after seeding at a density of 10,000 cells per 96-well, followed by antibiotic selection in the pres- ence of 8 μg/ml puromycin for 3 days (Sigma) in the same well. Subsequently, cells were grown for three additional days in fresh medium prior to analysis by flow cytometry or a cell viability assay.
Flow cytometry
Prior to staining, supernatants were removed and set aside. Cells were treated with 50 μl trypsin-EDTA 0.05 % (GIBCO) for 15 min (37 °C, 5 % CO2), and supernatant was added back to the wells. The resulting final volume of 250 μl was sup- plemented with 100 μl staining buffer containing 71.5 % phosphate-buffered saline (PBS), 1.5 % Igepal, 12 % propidium iodide, and 15 % RNase-A (all Sigma). After incubation at room temperature for 15 min followed by stor- age on ice, flow cytometry was conducted using a BD FACSCanto II (BD Biosciences). Flow cytometric cell counting was performed by quantification of viable cells with DNA content between 2n (single diploid) and 4n (double diploid) of samples that had been acquired in a fixed amount of 150 s with an upper limit of 10,000 viable cells. Data were analyzed using FACS Diva, Flowjo, and Microsoft Excel.
Quantification of apoptosis
Apoptosis was quantified with flow cytometry using Annexin V-FITC (BD Biosciences) in combination with propidium iodide (Sigma) staining according to the protocol of BD Biosciences. Cells were washed in PBS twice and resuspend- ed in 100 μl 1× binding buffer (10 mM HEPES, pH 7.4; 140 mM NaCl; 2.5 mM CaCl2). After addition of 5 μl Annexin Vand 2 μl propidium iodide, samples were incubat- ed for 15 min at room temperature in the dark. Samples were then diluted to a final volume of 500 μl with 1× binding buffer and immediately analyzed by flow cytometry using a BD FACS Canto II (BD Biosciences).
Cell viability assay
Cell viability was assessed using a 3-(4,5-dimethylthiazol-2- yl)-2,5-diphenyltetrazolium bromide-based MTT test (Sig- ma). Supernatant of cells was replaced with 100 μl of new
medium, and 20 μl MTT reagent (5 mg/ml) was added per 96- well. After incubation for 3 h (37 °C, 5 % CO2), supernatant was removed and cells were lysed using 120 μl MTT solvent containing isopropanol with 4 mM HCl and 0.1 % Igepal (Roth, EMSURE, Sigma). After 5 min, absorption was mea- sured at a wavelength of 584 nm using a 96-well ELISA reader (FLUOstar OPTIMA).
Methylation analysis with mass spectrometry
Preparation of genomic DNA and UPLC-MS/MS analysis of global DNA methylation were performed as described by Schiesser et al., 2013 [14]
Results
A flow cytometry-based assay in a single-well format
Flow cytometry is a powerful tool for characterizing samples at a single-cell level. In combination with propidium iodide (PI) staining, it allows simultaneous analysis of cell prolifer- ation, cell cycle progression, and cell death. However, most PI staining protocols are generally labor intensive and too time- consuming for screening purposes [15]. We therefore devel- oped a screening procedure in which cell culture, drug expo- sure, shRNA-mediated knockdown of potential therapeutic targets and PI staining are subsequent steps that are performed in the same well of a 96-well plate (Fig. 1a, b). In order to study synthetic cytotoxicity in combination with doxorubicin, we decided to employ SK-N-AS and SH-EP (N-Myc single copy), as well as LAN-1 and SK-N-BE(2) (N-Myc amplification) as neuroblastoma reporter cell lines. While SK-N-AS cells have previously been used to study synthetic lethality of azacytidine and doxorubicin, SH-EP is a standard model for studies of induced cell death in neuroblastoma cell lines [2, 16]. LAN-1 and SK-N-BE(2) are widely used to model drug resistance in N-Myc-amplified neuroblastoma [17].
First, we set out to verify that the established protocol was able to detect alterations of proliferation and cell cycle pro- gression accurately and that it permitted quantitative analysis of cell death. Inhibition of PLK1 is known to induce an M- phase arrest, which correlates with an accumulation of cells with 4n (double diploid) DNA content [18]. It also increases cell death, as arrested cells are sensitized to apoptosis com- pared to untreated controls [19]. Therefore, we decided to inactivate PLK1 by shRNA-mir-mediated knockdown in a proof-of-concept experiment.
Figure 1c demonstrates the anti-proliferative effects of PLK1 inhibition using our single well-based experimental setup. SH-EP cells transduced with shRNA-mir against
Fig. 1 A flow cytometric procedure to evaluate modulators of doxorubicin resistance in a single-well approach. a Schematic overview of the assay used for inhibitors of DNA methyltransferases: seeding of cells on day 0, growth under exposure to inhibitor for 3 days with daily replacement of inhibitor, addition of doxorubicin on day 3, and flow cytometric analysis on day 4. b Schematic overview of the assay employed for RNA-interference: transduction of cells on day 0; growth under selection with puromycin from day 1 to 4; splitting of cells at a ratio of 1:4 and replacement of media on day 4; if applicable, addition of doxorubicin on day 4; and analysis of cells on day 7 by flow cytometry or cell viability assay. c The feasibility of the approach is validated by
transduction of SH-EP cells with shRNA-mir against PLK1. M-phase arrest in SH-EP cells due to PLK1 inhibition is reflected by an accumu- lation of cells in G2/M phase with 4n (double diploid) DNA content and a corresponding depletion of cells in G1/G0 phase with 2n (single diploid) DNA content. Furthermore, the sub-G1 fraction, indicating apoptotic cells, increased upon PLK1-inhibition. d Increased cell death of SH-EP cells due to PLK1 inhibition after growth in the presence of doxorubicin at different concentrations 10–1000 nM as compared to control cells transduced with scrambled shRNA. e Increased accumulation of SH-EP cells in G2/M phase due to PLK1 inhibition after growth in the presence of doxorubicin as compared to control cells
PLK1 accumulated in both G2/M and sub-G1 phase of the cell cycle. We were also able to accurately quantify cell death caused by PLK1 inhibition alone as well as in combination with doxorubicin. Doxorubicin-induced cell death increased almost linearly as a consequence of escalating concentrations of doxorubicin. However, PLK1 knockdown clearly further added to the cytotoxic effect of doxorubicin on SH-EP cells. This is demonstrated by an increase of sub-G1 fractions in PLK1 knockdown cells in comparison to cells transduced with scrambled shRNA (Fig. 1d). It is noteworthy that in- creasing doses of doxorubicin alone induced an accumulation of cells in G2/M phase, which was further enhanced by knockdown of PLK1 (Fig. 1e).
Synergistic growth inhibition by doxorubicin and DNMT inhibitors
The applicability of our single-well setup prompted us to assess synthetic cytotoxicity of inhibitors of DNA methyltransferases (DNMT) in combination with doxorubicin. RG108, SGI-1027,
and nanaomycin A are DNMT inhibitors that—in contrast to traditional azacytidine—so far displayed low levels of innate cytotoxicity in most cell systems tested [10–12]. Therefore, they might provide an alternative option for epigenetic modifi- cations to counter proliferation of cancer cells with fewer adverse side effects, considering that a clinical phase I trial with azacytidine in patients with pediatric malignancies including neuroblastoma had to be discontinued due to intolerable toxic- ity at clinically ineffective doses [9]. To the best of our knowl- edge, RG108, SGI-1027, and nanaomycin A have not been characterized in neuroblastoma cells so far. In order to initiate an analysis for their therapeutic potential, we tested the ability of these compounds to inhibit cell proliferation of neuroblasto- ma cells either alone or in the presence of doxorubicin. We aimed to employ the DNMT inhibitors at maximum tolerated doses that still allowed to discern additional effects of doxoru- bicin. Therefore, the inhibitors were first applied at concentra- tions that in other systems had shown to be significantly below reported half maximal inhibitory concentrations, while still inducing effective DNA demethylation [2, 10–12].
As expected, exposure to 2 μM azacytidine only moder- ately affected total cell numbers of SK-N-AS and SH-EP cells but reduced total cell numbers of N-Myc-amplified LAN-1 and SK-N-BE(2) cells by almost 100 %, while 10 μM RG108 did not affect cell proliferation at all. In contrast, 3 μM SGI- 1027 reduced total cell numbers of all cell lines by >80 %. Nanaomycin A at a concentration of 100 nM also reduced cell numbers of SK-N-AS and SH-EP cells by >70 % and cell numbers of LAN-1 and SK-N-BE(2) by almost 100 % (Fig. 2a). The unexpected sensitivity of all four neuroblastoma cell lines allowed us to reduce inhibitor concentrations down to 1 μM (SGI-1027) and 30 nM (nanaomycin A) in all subsequent experiments. Even at these lower concentrations, enhanced cell death was observed in all four cell lines as assessed by quantitation of sub-G1 fractions of the flow cytometric data sets (Fig. 3c, d and Online Resource 1). In order to correlate the anti-proliferative effects of SGI-1027 and nanaomycin A with global DNA methylation, 5- methylcytosine levels in SK-N-AS cells were determined by HPLC/MS after 3 days of growth in the presence of RG108, SGI-1027, nanaomycin A, and azacytidine (Fig. 2c). Expo- sure to azacytidine reduced 5-methylcytosine levels by almost 50 % in a statistically highly significant manner, whereas none
of the other proposed DNMT inhibitors were able to achieve any noticeable reduction of 5-methylcytosine levels in SK-N- AS cells.
We next tested the ability of SGI-1027 and nanaomycin A to mediate synthetic cytotoxicity with doxorubicin. Cells were incubated with the indicated concentrations of inhibitors (Fig. 2c) 3 days prior to doxorubicin treatment (treatment outlined in Fig. 1a) as previously described [2]. After 24 h of doxorubicin exposure, cells were analyzed by flow cytometry. Low-dose doxorubicin treatment was performed at 30 nM (SK-N-AS, LAN-1, and SK-N-BE(2)) and 300 nM (SH-EP), which had almost no effect on total cell numbers (Fig. 2b) and did not affect cell death, as indicated by comparable levels of cells in sub-G1 (Fig. 3c, d). Azacytidine, which is known to enhance the effects of doxorubicin on neuroblastoma cells, was used as a positive control for DNA demethylation-dependent sensitization [2]. As expected, azacytidine clearly reduced cell numbers of cells treated with low-dose doxorubicin. Consis- tent with the preceding experiments, RG108 had no effect on doxorubicin-induced cytotoxicity (Figs. 2b and 3c, d). Com- binatorial treatment with SGI-1027 or nanaomycin A and doxorubicin significantly decreased cell numbers of cells in comparison to single treatment with doxorubicin (Fig. 2b).
Fig. 2 Effects of azacytidine, RG108, SGI-1027, and nanaomycin A ± doxorubicin on cell growth and global 5-methylcytosine levels. a Cell counts of SK-N-AS and SH-EP (N-Myc single copy) as well as LAN-1 and SK-N-BE(2) (N-Myc-amplified) cells demonstrate the maximum tolerated concentrations of SGI-1027 and nanaomycin A, as well as the inefficacy of even high doses of RG108. b Synergistic inhibition of cell growth due to exposure to azacytidine, SGI-1027, and nanaomycin A in
combination with doxorubicin. c Reduction of 5-methylcytosine levels in SK-N-AS cells after exposure to azacytidine and absence of changes to 5- methylcytosine levels in SK-N-AS cells after treatment with RG108, SGI-1027, and nanaomycin A. Student’s t test was performed for detec- tion of statistically significant differences between samples (*p <0.05; **p<0.01; ***p <0.005)
Fig. 3 Effects of azacytidine, RG108, SGI-1027, and nanaomycin A ± doxorubicin on cell cycle distribution, cell death, and apoptosis in SK-N- AS and SH-EP (N-Myc single copy) as well as LAN-1 and SK-N-BE(2) (N-Myc-amplified) cells. a, b Quantification of cell cycle distribution with flow cytometry using propidium iodide fluorescence labeling. c, d Enhanced doxorubicin-induced cytotoxicity due to combinatorial treat- ment with azacytidine, SGI-1027, and nanaomycin A is demonstrated by quantification of cell death after exposure to inhibitors ± doxorubicin with
flow cytometry and shown as sub-G1 fraction. e, f Enhanced doxorubicin-induced cytotoxicity through combination with azacytidine, SGI-1027, and nanaomycin A is validated by quantification of apoptosis after exposure to inhibitors ± doxorubicin with flow cytometry using Annexin V and propidium iodide fluorescence labeling. Student’s t test was performed for detection of statistically significant differences be- tween samples (*p <0.05; **p<0.01; ***p <0.005)
Detailed analysis of the corresponding flow cytometric datasets (Online Resource 1) revealed that combinatorial treat- ment of azacytidine, SGI-1027, and nanaomycin Awith doxo- rubicin induces either a decrease of cells in G2 (SH-EP, LAN-1, SK-N-BE(2)) or in S-phase (SK-N-AS) of the cell cycle (Fig. 3a, b), as compared to exposure to doxorubicin alone. It also enhanced doxorubicin-induced cell death as measured by an increase of sub-G1 in an at least additive manner (Fig. 3c, d). RG108 on the other hand did not have any significant effect on cell cycle distribution or cell death.
In order to validate the observed effects, we repeated the combinatorial treatments in 12-well plates and quantified apoptosis using Annexin V staining in combination with propidium iodide fluorescence labeling (see Online Resource 2 for flow cytometric datasets). In concordance with the preceding experiments, azacytidine increased apoptosis induced by low-dose doxorubicin in all four cell lines (Fig. 3e, f). Likewise, SGI-1027 and nanaomycin Awere able
to enhance doxorubicin-induced apoptosis in all four cell lines. Azacytidine, SGI-1027, and nanaomycin A also caused marked induction of apoptosis when applied alone. None of the observed effects were dependent on N-Myc status. Taken together, these results confirm the earlier findings. Contrary to preceding results, RG108 showed increased apoptosis in SH- EP and LAN-1 cells alone and in combination with doxoru- bicin but remained largely without effect in SK-N-AS and SK- N-BE(2) cells (Fig. 3e, f).
Interestingly, the effects of SGI-1027, nanaomycin A, and azacytidine on cell death in doxorubicin-treated and non- treated control cells appear to be almost equal in amplitude (Fig. 3c–f), even though only azacytidine induces global DNA demethylation (see above). This suggests SGI-1027 and nanaomycin A may both be able to render neuroblastoma cells more sensitive to doxorubicin-induced cytotoxicity, albeit through a global DNA demethylation-independent mechanism.
Inhibition of Fbxo5/Emi1 disrupts proliferation in neuroblastoma cells
Next, we tested the versatility of our approach by attempting to identify potential therapeutic targets in SH-EP neuroblas- toma cells through a shRNA library-based functional screen. Reporter cells were transduced with 275 pGIPZ shRNA-mir expression vectors in parallel, targeting 59 F-box proteins [20]. In a first step, we wanted to identify F-box proteins whose downregulation leads to inhibition of cell proliferation of neuroblastoma cells, a phenotype that previously has been described for the F-box protein Fbxo5/Emi1 [21].
After 4 days of puromycin selection, cells were split and allowed to grow for three additional days prior to the analysis of cell viability and flow cytometric quantitation of DNA content (Fig. 1b). Two shRNA-mirs targeting PLK1 that had been used to validate the experimental setup (see above) were included as positive controls. Genes represented by two inde- pendent shRNA-mirs causing a significant change in cell cycle distribution or DNA content of more than two standard deviations away from the median were defined as hits. In
addition to the two positive controls (shRNA-mirs against PLK1), two shRNA-mirs shifted cell cycle distribution of SH-EP cells toward an accumulation in G2/M phase by more than four standard deviations from the median (Fig. 4a). It turned out that both of these shRNA-mirs were directed against Fbxo5/Emi1, thereby convincingly demonstrating the feasibility of our approach. In the analysis of total cellular DNA content, the same shRNA-mirs against Fbxo5/Emi1 induced an increase of cells with DNA content >4n by more than fourfold (Fig. 4b). DNA profiles of SH-EP cells trans- duced with either shRNA-mirs against Fbxo5/Emi1 or shRNA-mirs against PLK1 are compared in Fig. 4c.
Inhibition of Fbxo5/Emi1 and Fbxw11/β-TrCP2 alter doxorubicin sensitivity of neuroblastoma cells
Besides inhibition of proliferation, sensitization of tumor cells toward established chemotherapeutic agents like doxorubicin is another key element in targeted cancer therapies. We ex- tended our screening to identify F-box proteins that modify sensitivity of SH-EP cells toward doxorubicin. Fbxo5/Emi1
Fig. 4 Identification of Fbxo5/Emi1 in a lentiviral shRNA screen with 275 shRNA-mirs against 59 F-box proteins in SH-EP cells. a shRNA- mirs against PLK1 and Fbxo5/Emi1 lead to an accumulation of SH-EP cells with 4n (double diploid) and a depletion of SH-EP cells with 2n (single diploid) DNA content. Analysis of DNA content after 3 days of growth without doxorubicin measured by flow cytometry quantified in
standard deviations from the mean. b shRNA-mirs against PLK1 and Fbxo5/Emi1 lead to an accumulation of SH-EP cells with >4n DNA content. Analysis of DNA content after 3 days of growth without doxo- rubicin measured by flow cytometry quantified in percentage of sample. c Accumulation of SH-EP cells with >4n DNA content due to transduction with shRNA-mir against Fbxo5/Emi1 and PLK1
and Fbxw11/β-TrCP2 are known to play a role in this context [22, 23].
To begin with, we screened for factors whose loss of function increases doxorubicin resistance, following a negative selection strategy. Exposure to a lethal concen- tration of doxorubicin completely effaces cell viability of SH-EP cells as measured by MTT assay. shRNAs that increase resistance against doxorubicin can be iden- tified by screening for viable transduced cells against a very low background. Two shRNAs targeting p53 served as positive controls for shRNA-mediated increase of resistance against doxorubicin. As demonstrated in Fig. 5a, transduction of SH-EP cells with shRNAs targeting p53 protects them against otherwise lethal concentrations of doxorubicin. The negative selection screen was not able to identify F-box proteins whose inhibition augments resistance toward doxorubicin in SH-EP cells. Figure 5b shows that cell viability of only nine samples representing shRNA-mirs against nine dif- ferent genes was two standard deviations above median. Since the read-outs of the positive controls were >15 standard deviations away from the median, experimental
conditions allowed for a very good discrimination of shRNA-mirs with an effect on doxorubicin resistance. The probability of this result being false-negative is therefore very low.
Finally, we turned to a screen for F-box proteins whose inhibition leads to sensitization toward doxorubicin. shRNA-mirs against PLK1 were employed as positive controls. In this experiment, shRNA-mir transduced cells were grown in the presence (300 nM) or absence of doxorubicin for 3 days prior to analysis of cell viability and DNA content. Only a small fraction of shRNA-mirs was able to increase doxorubicin sensitivity by more than two standard deviations above median (Fig. 5c), including both shRNA-mirs targeting PLK1. Due to the low and inhomogeneous coverage of the library with F-box protein genes being represented by one to seven shRNA-mirs (see Online Resource 3), we decided to repeat the assay with a subset of those shRNA-mirs that had mediated increased doxorubicin-induced cell death by more than 0.5 standard deviations (Fig. 5d). In this setting, one shRNA-mir against Fbxw11/β-TrCP2 scored as best hit with a doxorubicin-induced cell death more than three standard
Fig. 5 Screens for modulators of doxorubicin resistance with 275 shRNA-mirs against 59 F-box proteins. a shRNA-mir (TP53-1) and a pLentiLox3.7 shRNA (PLL p53) against p53 lead to increase of doxoru- bicin resistance of SH-EP cells compared to SH-EP cells transduced with scrambled shRNA (NSC-1 and NSC-2) in the presence of a cytotoxic concentration of 1000 nM doxorubicin measured by cell viability for 3 days. Student’s t test was performed for detection of statistically significant differences between samples (*p <0.05). b A screen for shRNA-mirs that increase doxorubicin resistance identifies no candidate besides shRNA against p53. Analysis of cell viability visualized in
standard deviations of the mean. c Primary screen for shRNA-mirs that sensitize SH-EP cells toward doxorubicin with analysis of sub-G1 frac- tions of samples after 3 days of growth in the presence of a cytostatic concentration of 300 nM doxorubicin measured by flow cytometry and quantified in standard deviations from the mean. d shRNA-mirs against Fbxw11/β-TrCP2 and PLK1 sensitize SH-EP cells toward doxorubicin in a validation screen. Analysis of sub-G1 fraction of samples performed after growth in the presence of 300 nM doxorubicin and cell viability of samples after growth in the absence of doxorubicin
deviations above median. This F-box protein is known to be a modifier of the anti-proliferative effects of anti- cancer drugs such as doxorubicin, tamoxifen, and pacli- taxel in human breast cancer cells [23]. Notably, one shRNA-mir against Fbxo5/Emi1 that had induced a strong disruption of cell proliferation in the first screen (Fig. 4c) also increased cell death moderately.
Discussion
Patients with high-risk neuroblastoma still face an unsatisfac- tory 5-year survival rate of approximately 50 % [1]. Current efforts to overcome therapeutic limitations focus on specific modulations of tumor-promoting cell properties [6]. The bot- tleneck for this approach is the identification of suitable mo- lecular targets. In order to address this issue, we established a flexible protocol for rapid assessment of potential therapeutic compounds in four neuroblastoma cell lines. Furthermore, this system allows unbiased screens for factors that potentially play a key role in the pathogenesis of neuroblastoma.
DNA methyltransferase inhibitors in combination with doxorubicin
In the first part, we assessed the potential of three inhibitors of DNA methyltransferases (DNMT), RG108, SGI-1027, and nanaomycin A, to replace azacytidine as less toxic therapeutic agents to sensitize neuroblastoma cells to doxorubicin through induction of DNA demethylation [2]. This was motivated by a phase I trial probing azacytidine in various pediatric malig- nancies that had to be discontinued due to adverse side effects, which impeded application of azacytidine at clinically effec- tive doses [9].
The non-nucleoside DNMT inhibitors RG108, SGI-1027, and nanaomycin A presented low toxicity in several test systems [10–12]. This probably results from a lack of DNA incorporation that has been attributed as the main cause of azacytidine toxicity, which is even observed in complete absence of DNMT activity [11, 24]. All three DNMT inhibi- tors have previously been described to be able to induce global DNA demethylation and reexpression of silenced tumor sup- pressor genes. Nanaomycin A, a selective inhibitor of DNMT3b, has been shown to induce global DNA demethyl- ation and reexpression of the tumor suppressor gene RASSF1A in HCT116 (colon cancer), A549 (lung cancer) and HL60 cells (human promyelocytic leukemia) [12]. SGI- 1027 induces degradation of the maintenance DNA methylase DNMT1 at low (2.5 μM) and inhibits DNMT3a and DNMT3b at higher concentrations (7.5–12.5 μM). It can induce global DNA demethylation and was shown to reacti- vate tumor suppressor genes like TIMP3, p16, and MLH1 in
human colon cancer cells [11]. Exposure to SGI-1027 at concentrations >10 μM led to a moderate inhibition of cell proliferation of rat hepatoma cells [11]. Cytotoxicity induced by SGI-1027 in U937 leukemia cells was only observed at doses >5 μM [25]. RG108 induces DNA demethylation and reactivation of tumor suppressor genes in HCT116 and Nalm6 cells [10]. Unexpectedly, exposure of neuroblastoma cells to 10 μM RG10 did neither affect global DNA methylation nor cell death nor cell proliferation in the absence or presence of doxorubicin in most experiments (Figs. 2 and 3).
Combinatorial treatment of SK-N-AS, SH-EP, LAN-1, and SK-N-BE(2) cells with doxorubicin and either SGI-1027 or nanaomycin A enhanced cell death in comparison to exposure to doxorubicin alone (Fig. 3c–f), which phenocopies the sen- sitization of neuroblastoma cells toward doxorubicin by azacytidine [2]. N-Myc status did not influence this effect. Unfortunately, in all neuroblastoma cell lines tested, cytotox- icity of SGI-1027 and nanaomycin A equaled or even surpassed cytotoxicity of azacytidine as measured by sub-G1 fractions (Fig. 3c, d). Validation experiments using Annexin V staining also showed strong induction of apoptosis that was superior to azacytidine-mediated apoptosis in three out of four cell lines (Fig. 3e, f). Interestingly, this high sensitivity to azacytidine, SGI-1027, and nanaomycin A seemed to be even more prominent in N-Myc-amplified neuroblastoma cells than in cell lines with a single copy of N-Myc (Fig. 2a). Especially, in case of nanaomycin A, the effect was pronounced, as the tested neuroblastoma cell lines are at least 20-fold more sen- sitive to nanaomycin A than to azacytidine exposure (Fig. 2a).
Taken together, these results seem to disqualify RG108, SGI-1027, and nanaomycin A as less harmful azacytidine replacements in the treatment of high-risk neuroblastoma.
Interestingly, both SGI-1027 and nanaomycin A dis- play cytotoxicity in the absence of detectable effects on global DNA methylation in SK-N-AS cells (Fig. 2c). This is reminiscent of recent RNAi-mediated DNMT3b depletion experiments that demonstrated an anti- proliferative effect of DNMT3b knockdown even in the absence of DNA demethylation [26]. In murine neuroblastoma cells, drug resistance correlates with Dnmt1 and Dnmt3b overexpression and azacytidine re- verses cisplatin resistance induced by stable transfection of Dnmt3a and Dnmt3b expression vectors [27, 28]. Combined, these observations suggest that DNMT3b may play a critical role in neuroblastoma cell prolifera- tion and drug resistance. In this context, it is promising that the tested neuroblastoma cell lines showed high sensitivity to the DNMT3b inhibitor nanaomycin A— especially in the context of N-Myc amplification. Con- sidering the reported low cytotoxicity in other systems, nanaomycin A might thus harbor a therapeutic potential for patients with high-risk neuroblastoma that seems worth to be evaluated in further studies [10–12].
A screen of F-box proteins for potential therapeutic targets for neuroblastoma patients
In addition to efforts to reverse aberrant DNA methylation patterns in order to restore susceptibility to chemotherapeutic drugs like doxorubicin by epigenetics-based therapies, it is also essential to continue the quest for additional molecular targets in neuroblastoma cells for therapeutic inhibition. Along these lines, we applied the developed system also to attempt to identify novel therapeutic targets by screening for factors that modulate tumor cell proliferation or doxorubicin sensitivity, thereby focusing on members of the family of F- box proteins. In our screen, Fbxo5/Emi1 was identified as an F-box protein whose inhibition disrupts normal mitosis by induction of endoreplication and also causes a moderate in- crease of doxorubicin sensitivity (Figs. 4 and 5c, d). These results are in accordance with results from previous screens, which identified Fbxo5/Emi1 as a main regulator of cell proliferation [21, 29] and demonstrated that inhibition of Fbxo5/Emi1 can sensitize cancer cell lines to doxorubicin and X-ray irradiation-induced cell death [22]. This effect was cell type-dependent and ranged from a modest, as ob- served in SH-EP cells in Fig. 5d, to a very pronounced extent [22]. The shRNA-mir against Fbxo5/Emi1 with the most prominent impact on cell proliferation in our screen was the same that also scored prominently in the screen published by Schlabach et al. 2008, albeit in different reporter cell lines [29].
Besides Fbxo5/Emi1, Fbxw11/β-TrCP2 was also iden- tified as an F-box protein that provides a potential ther- apeutic target for neuroblastoma patients, as its inhibition markedly sensitizes SH-EP cells to doxorubicin-induced cell death (Fig. 5c, d). This is in accordance with sensi- tization of breast cancer cells toward doxorubicin- induced cell death through inhibition of Fbxw1/β-TrCP, a functional analog of Fbxw11/β-TrCP2 [23]. Further- more, it has been shown that inhibition of Fbxw1/β- TrCP and Fbxw11/β-TrCP2 is able to inhibit growth of prostate cancer cells by upregulation of the aryl hydro- carbon receptor [30].
Although our screen did not identify F-box proteins whose inhibition conveys an increase in doxorubicin resistance, an increase of doxorubicin resistance was clearly detected due to knockdown of p53 mediated by both positive control shRNAs (Fig. 5a, b).
The detection of the remarkable sensitivity of SH-EP and SK-N-AS (N-Myc single copy) as well as LAN-1 and SK-N-BE(2) (N-Myc-amplified) neuroblastoma cell lines to the DNA methyltransferase inhibitor nanaomycin A and the identification of Fbxw1/β-TrCP as well as Fbxo5/Emi1 as potential therapeutic targets in neuroblastoma convincingly demonstrates the ability our flow cytometry-based protocol to identify potential drug
targets and to validate candidate inhibitors in high-risk neuroblastoma model cell lines.
Conflict of interest The authors declare that they have no conflict of interest.
References
1.Maris J (2010) Recent advances in neuroblastoma. N Engl J Med 362(23):2202–11
2.Charlet J, Schnekenburger M, Brown KW, Diederich M (2012) DNA demethylation increases sensitivity of neuroblastoma cells to chemo- therapeutic drugs. Biochem Pharmacol 83(7):858–65
3.Chen ZC, Chen LJ, Cheng JT (2013) Doxorubicin-induced cardiac toxicity is mediated by lowering of peroxisome proliferator-activated receptor δ expression in rats. PPAR Res 2013:456042
4.Takemura G, Fujiwara H (2007) Doxorubicin-induced cardiomyop- athy from the cardiotoxic mechanisms to management. Prog Cardiovasc Dis 49(5):330–52
5.Druker BJ, Guilhot F, O’Brien SG, Gathmann I, Kantarjian H, Gattermann N et al (2006) Five-year follow-up of patients receiving imatinib for chronic myeloid leukemia. N Engl J Med 355(23):2408– 17
6.Cheung NK, Dyer MA (2013) Neuroblastoma: developmental biol- ogy, cancer genomics and immunotherapy. Nat Rev Cancer 13(6): 397–411
7.Esteller M (2008) Epigenetics in cancer. N Engl J Med 358(11): 1148–59
8.Fenaux P, Mufti GJ, Hellstrom-Lindberg E, Santini V, Finelli C, Giagounidis A et al (2009) Efficacy of azacitidine compared with that of conventional care regimens in the treatment of higher-risk myelodysplastic syndromes: a randomised, open-label, phase III study. Lancet Oncol 10(3):223–32
9.George RE, Lahti JM, Adamson PC, Zhu K, Finkelstein D, Ingle AM et al (2010) Phase I study of decitabine with doxorubicin and cyclo- phosphamide in children with neuroblastoma and other solid tumors: a Children’s Oncology Group study. Pediatr Blood Cancer 55(4): 629–38
10.Brueckner B, Garcia Boy R, Siedlecki P, Musch T, Kliem HC, Zielenkiewicz P et al (2005) Epigenetic reactivation of tumor sup- pressor genes by a novel small-molecule inhibitor of human DNA methyltransferases. Cancer Res 65(14):6305–11
11.Datta J, Ghoshal K, Denny WA, Gamage SA, Brooke DG, Phiasivongsa P et al (2009) A new class of quinoline-based DNA hypomethylating agents reactivates tumor suppressor genes by blocking DNA methyltransferase 1 activity and inducing its degra- dation. Cancer Res 69(10):4277–85
12.Kuck D, Caulfield T, Lyko F, Medina-Franco JL (2010) Nanaomycin A selectively inhibits DNMT3B and reactivates silenced tumor sup- pressor genes in human cancer cells. Mol Cancer Ther 9(11):3015–23
13.Wiebusch L, Hagemeier C (2010) p53- and p21-dependent premature APC/C-Cdh1 activation in G2 is part of the long-term response to genotoxic stress. Oncogene 29(24):3477–89
14.Schiesser S, Pfaffeneder T, Sadeghian K, Hackner B, Steigenberger B, Schröder AS et al (2013) Deamination, oxidation, and C-C bond cleavage reactivity of 5-hydroxymethylcytosine, 5-formylcytosine, and 5-carboxycytosine. J Am Chem Soc 135(39):14593–9
15.Riccardi C, Nicoletti I (2006) Analysis of apoptosis by propidium iodide staining and flow cytometry. Nat Protoc 1(3):1458–61
16.Fulda S, Lutz W, Schwab M, Debatin K (1999) MycN sensitizes neuroblastoma cells for drug-induced apoptosis. Oncogene 18(7): 1479–86
17.Goldschneider D, Horvilleur E, Plassa LF, Guillaud-Bataille M, Million K, Wittmer-Dupret E et al (2006) Expression of C-terminal deleted p53 isoforms in neuroblastoma. Nucleic Acids Res 34(19): 5603–12
18.Spänkuch-Schmitt B, Bereiter-Hahn J, Kaufmann M, Strebhardt K (2002) Effect of RNA silencing of polo-like kinase-1 (PLK1) on apoptosis and spindle formation in human cancer cells. J Natl Cancer Inst 94(24):1863–77
19.Spänkuch B, Heim S, Kurunci-Csacsko E, Lindenau C, Yuan J, Kaufmann M et al (2006) Down-regulation of Polo-like kinase 1 elevates drug sensitivity of breast cancer cells in vitro and in vivo. Cancer Res 66(11):5836–46
20.Maier B, Wendt S, Vanselow J, Wallach T, Reischl S, Oehmke S et al (2009) A large-scale functional RNAi screen reveals a role for CK2 in the mammalian circadian clock. Genes Dev 23(6):708–18
21.Machida YJ, Dutta A (2007) The APC/C inhibitor, Emi1, is essential for prevention of rereplication. Genes Dev 21(2):184–94
22.Shimizu N, Nakajima NI, Tsunematsu T, Ogawa I, Kawai H, Hirayama R, et al (2013)Selective enhancing effect of early mitotic inhibitor 1 depletion on the sensitivity of doxorubicin or X-ray treatment in human cancer cells. J Biol Chem. May
23.Tang W, Li Y, Yu D, Thomas-Tikhonenko A, Spiegelman VS, Fuchs SY (2005) Targeting beta-transducin repeat-containing protein E3
ubiquitin ligase augments the effects of antitumor drugs on breast cancer cells. Cancer Res 65(5):1904–8
24.Hegde V, McFarlane RJ, Taylor EM, Price C (1996) The genetics of the repair of 5-azacytidine-mediated DNA damage in the fission yeast Schizosaccharomyces pombe. Mol Gen Genet 251(4):483–92
25.García-Domínguez P, Dell’aversana C, Alvarez R, Altucci L, de Lera AR (2013) Synthetic approaches to DNMT inhibitor SGI-1027 and effects on the U937 leukemia cell line. Bioorg Med Chem Lett 23(6): 1631–5
26.Hagemann S, Kuck D, Stresemann, Carlo, Prinz F, Brueckner B, et al (2012) Antiproliferative effects of DNA methyltransferase 3B deple- tion are not associated with DNA demethylation. PLoS ONE; p. e36125.
27.Qiu YY, Mirkin BL, Dwivedi RS (2005) Inhibition of DNA methyl- transferase reverses cisplatin induced drug resistance in murine neu- roblastoma cells. Cancer Detect Prev 29(5):456–63
28.Qiu YY, Mirkin BL, Dwivedi RS (2002) Differential expression of DNA-methyltransferases in drug resistant murine neuroblastoma cells. Cancer Detect Prev 26(6):444–53
29.Schlabach MR, Luo J, Solimini NL, Hu G, Xu Q, Li MZ et al (2008) Cancer proliferation gene discovery through functional genomics. Science 319(5863):620–4
30.Gluschnaider U, Hidas G, Cojocaru G, Yutkin V, Ben-Neriah Y, Pikarsky E (2010) beta-TrCP inhibition reduces prostate cancer cell growth via upregulation of the aryl hydrocarbon receptor. PLoS One 5(2):e9060