Neuroblastoma (NB) originates from progenitors of mature sympathoadrenal cells and is characterized by cellular heterogeneity.1, 2 This is reflected in diverse clinical behaviors with at least three different courses including spontaneous regression, differentiation into mature ganglioneuroma and progression despite intensive multimodal therapy.2 Amplified MYCN signals poor prognosis in both localized and advanced NB and is an international criterion for risk group assignment.2 Additional poor-prognosis factors include high levels of E2F-1, Skp2, Mad2 and survivin proteins.3–7
Conventional therapeutic agents act by causing DNA damage, by interfering with the mitotic spindle organization or by inhibiting ribonucleotide synthesis or DNA replication. All these not only cause acute toxicity but also result in long-term complications that affect as many as 40% of the survivors.8 The limited efficacy and the frequent and serious side effects necessitate the development of new, more efficacious and less toxic strategies.
Histone deacetylases (HDACs) are promising targets for such a novel NB therapy. HDACs catalyze the removal of acetyl groups from the lysine residues of proteins including the core nucleosomal histones. They are involved in the remodeling of chromatin, and consequently, in the regulation of gene expression.9 There is growing evidence that perturbation of epigenetic balances promotes the initiation and progression of cancer.10 An overall loss of monoacetylation at H4-Lys16 and trimethylation at H4-Lys20 located at hypomethylated DNA repetitive loci was identified in both primary tumors and a panel of human tumor cell lines including the NB cell line SK-N-SH.11 Thus, epigenetic deregulation during differentiation of the sympathoadrenal lineage may contribute to the onset and growth of NB.
Aberrant epigenetic patterns in tumor cells can be pharmacologically modified.9, 12, 13 Small-molecule histone deacetylase inhibitors (HDACIs) with diverse chemical structures and pharmacologic properties interfere with HDAC-activity in tumor cells in vitro initiating antiproliferative, proapoptotic and prodifferentiating effects.12, 14 Phase I/II clinical trials elucidating HDACI-action against refractory solid tumors, leukemias and lymphomas show an encouraging efficacy and safety profile.9, 13
Here, we analyzed the potential of Helminthosporium carbonum (HC)-toxin for NB therapy. HC-toxin is a cyclic tetrapeptide purified from the plant pathogenic fungus Cochliobolus (Helminthosporium) carbonum,15 which was identified as an inhibitor of HDACs in yeast, plant and mammalian cells.16 In particular, it inhibits human class I HDAC members (HDAC1, -2, -3 and -8),15, 17 which are defined by their homology to the yeast HDAC Rpd3.9 Class I target specific DNA regions after binding to transcription- and chromatin-related factors including the retinoblastoma tumor suppressor protein (RB).18 This study demonstrates (i) that HC-toxin causes a shift to a benign and differentiated phenotype of NB cells and (ii) that this phenotype shift is associated with an activation of the RB tumor suppressor network.
Human NB cell lines with (BE(2)-C, Kelly, LA-N-1, NGP) and without (SH-EP, SH-SY5Y) amplified MYCN were cultured at 37°C, 5% CO2.19 BE(2)-C, Kelly and SH-SY5Y were grown in DMEM medium (Cambrex) with 10% FCS (Sigma) and 1% nonessential amino acids (NEAA; Cambrex), LA-N-1, NGP and SH-EP in RPMI 1640 medium (Cambrex) with 10% FCS. WAC-2 containing an ectopic CMV-MYCN governing enhanced expression was cultured in RPMI 1640 with 10% FCS and 200 μg/ml geneticin.20 Fresh NB cells were isolated from the bone marrow aspirate of a patient with metastasized relapsed INSS stage 4 NB following informed consent. Cells (MYCN single copy) were grown in DMEM (10% FCS, 1% NEAA) at 37°C, 5% CO2. Primary human skin fibroblasts (PHFs) derived from a healthy donor were cultured likewise.
Cells were treated with nanomolar doses of HC-toxin (Lot no. 054K4121; Sigma) dissolved in methanol p.a. (MeOH; Roth). For comparison, cells were treated with 1 mM valproic acid (VPA; Sigma) dissolved in DPBS (Cambrex).
Phase contrast images were acquired with an inverted phase contrast microscope (Olympus; Model CKX41) using Cell B software (Olympus), fluorescent images with a confocal microscope (Leica Microsystems; Model TCS SP2 with AOBS) using Leica confocal software.
Cell viability assay
Cells were plated, treated with HC-toxin, VPA or solvents, harvested and evaluated for number and viability by trypan blue exclusion (Sigma). Trypan blue is a diazo dye used to selectively color dead cells with disrupted membranes. Viable cells with intact cell membranes do not absorb trypan blue.
For cell cycle distribution, cells were collected in 2.1% citric acid/0.5% Tween 20 and stained with 50 μg/ml DAPI dissolved in phosphate buffer (0.4 M Na2HPO4, pH 8.0). About 50,000–100,000 cells were analyzed with a Galaxy Pro Flow Cytometer (Partec) using the Multicycle software (Phoenix Flow Systems). For evaluation of apoptosis, cells were labelled with Nicoletti stain. About 5,000–10,000 cells were analyzed with a FACS Calibur (Becton Dickinson) using the Cell Quest software (Becton Dickinson). Measurements were acquired in Fl-2 in logarithmic mode. The number of apoptotic cells was calculated by setting the gate over the first three decades.
Cells were grown on poly-D-lysine coated coverslips (Menzel). Primary antibodies were anti-Nef3 150 kDa rabbit polyclonal (Chemicon), anti-Map2 mouse monoclonal (clone AP20, Chemicon), anti-Syn1 mouse monoclonal (Synaptic Systems), anti-Syp1 mouse monoclonal (Synaptic Systems) and anti-Th rabbit polyclonal (Chemicon). Secondary antibodies were Cy3-conjugated goat anti-mouse (Jackson ImmunoResearch) and goat anti-rabbit (Rockland).
The following antibodies were used: anti-β-actin mouse monoclonal (clone AC-15, Sigma-Aldrich), anti-E2F-1 mouse monoclonal (clone KH95, Santa Cruz), anti-acetyl-histone H4 rabbit polyclonal (Upstate), anti-Mad2 mouse monoclonal (clone 48, BD Biosciences), anti-N-myc mouse monoclonal (BD Biosciences), anti-p15INK4b mouse monoclonal (clone 15P06, Dianova), anti-p16INK4a mouse monoclonal (clone ZJ11, Dianova), anti-p21cip1/waf-1 mouse monoclonal (Upstate), anti-p27kip1 mouse monoclonal (clone DCS-72.F6, Dianova), anti-Skp2 mouse monoclonal (clone 2C8D9, Zymed), anti-survivin mouse monoclonal (clone D-8, Santa Cruz) and anti-RB mouse monoclonal (clone G3-245, BD Biosciences). Protein concentrations of the cell lysates were determined by Bradford assay (Bio-Rad).
Quantitative real time RT-PCR
Quantification of mRNA was carried out by real time reverse transcription (RT)-PCR with SYBR Green Dye (Eurogentec) using an ABI Prism 7700 thermal cycler (Perkin-Elmer Applied Biosystems).21 Target mRNA levels were normalized to the average mRNA levels of ACTB and GAPD. The following primers were selected from Qiagen's QuantiTect Primer Assay program: GAPD, no. QT00079247; NEF3, no. QT00073885; NES, no. QT01015301; MAP2, no. QT00057358; SYN1, no. QT00045913; SYP, no. QT00013062 and TH, no. QT00067221. The following exon-spanning primers (5′–3′ orientation) were purchased from Thermohybaid: ACTB (cDNA amplicon length 151 bp), GC ATCCCCCAAAGTTCACAA (forward) and AGGACTGGGC CATTCTCCTT (reverse).
Colony formation assay
About 4–20 cells/cm2 were plated and treated. After 5 days, colonies were fixed with MeOH/ acetic acid (3:1), stained with crystal violet and counted.
Soft agar assay
Cells were pretreated with HC-toxin or solvent for 48 hr and harvested. Cells were evaluated for number and viability by trypan blue exclusion and viable cells were seeded into soft agar at equal density according to the instructions of Chemicon's Cell Transformation Detection Assay. Treatment was continued for 10 (WAC-2) and 14 days (BE(2)-C). Viable colonies were labeled with cell stain solution (Chemicon) and counted.
Invasive growth was evaluated with Matrigel-coated Boyden chambers (Becton Dickinson).22 For comparison, uncoated Boyden chambers were used. Cells growing in medium with 10% FCS were pretreated with HC-toxin or solvent for 48 hr, harvested and resuspended in DMEM medium without FCS at equal density (5 × 104 cells per milliliter). About 0.5 ml of this cell suspension was placed in the upper Boyden chamber and treated with HC-toxin or solvent. DMEM medium with 10% FCS was used as chemoattractant. The Boyden chambers were incubated for 14 hr (WAC-2) and 22 hr (BE(2)-C). Thereafter, cells on the top side of the membrane were removed. Cells that had invaded the Matrigel had passed across the pored membrane and had attached to the bottom side of the membrane were fixed with MeOH/acetic acid (3:1) and stained with hematoxylin/eosin (Roth). Cell invasion (%) was expressed as the number of cells invading through the Matrigel-coated membrane compared to the number of cells migrating through the uncoated control membrane. Ten randomly chosen fields per membrane were analyzed.
Results are expressed as mean values with standard deviations (SDs). All experiments were independently carried out at least 3 times in triplicates. Comparisons of treatments were performed by unpaired t-test (SigmaPlot 9.0). p values below 0.05 were considered statistically significant.
Inhibition of NB cell growth and increase in histone H4 acetylation
To determine if HC-toxin influences the growth behavior of NB cells, SH-EP, WAC-2, SH-SY5Y, BE(2)-C, NGP, LA-N-1 and Kelly were exposed to different concentrations of HC-toxin for varying time periods. For comparison, primary human fibroblasts were used. Three growth parameters were evaluated, number of viable cells, cell cycle distribution and number of apoptotic cells.
The number of viable cells was assessed by trypan blue exclusion. HC-toxin treatment of all NB cell lines resulted in a time- and dose-dependent decrease in the number of viable cells (Fig. 1a, shown for 20 nM HC-toxin). The number of viable WAC-2 cells expressing high MYCN and that of parental SH-EP with barely detectable MYCN were similar (Fig. 1a). HC-toxin treatment did not obviously alter the number of viable primary human fibroblasts (Fig. 1a).
To estimate the concentration of HC-toxin required for a 50% reduction of viable NB cells (IC50) at 72 hr, dose-response curves were plotted with HC-toxin concentration against percentage of dead cells. IC50-values were obtained by interpolation from the dose-response curves and ranged for SH-EP, WAC-2, SH-SY5Y, BE(2)-C, NGP, LA-N-1 and Kelly from 16 nM (NGP) to 19.5 nM (SH-SY5Y) (Fig. 1b). At those low nanomolar concentrations, HC-toxin increased the level of acetylated histone H4 in both NB cell lines and primary human fibroblasts (Fig. 1c shown for SH-EP, BE(2)-C and primary human fibroblasts).
Changes in cell cycle distribution and apoptosis, exemplarily shown for SH-EP and BE(2)-C, were assessed by cytometry. For comparison, changes in cell cycle distribution and apoptosis by 1 mM VPA were evaluated. Both HC-toxin and VPA were used in concentrations that caused a 50% inhibition of HDAC activity in vitro in non cell-based assays and that similarly increased the level of acetylated histone H4 in NB cells (data not shown).17
HC-toxin induced in both SH-EP and BE(2)-C at all time periods investigated a G0/G1-cell cycle arrest (Fig. 2a) that was accompanied by an ∼2- to 3-fold decrease in the number of proliferating cells. Similar results were obtained for all other NB cell lines investigated (n = 5, data not shown). In contrast, VPA did not induce a G0/G1-cell cycle arrest in SH-EP but a slight, ∼1.3-fold increase in the number of cells in G2/M (Fig. 2a). In BE(2)-C, VPA did not induce obvious changes in the cell cycle distribution except at 72 hr, where a G0/G1-cell cycle arrest accompanied by an 1.5-fold decrease in the number of proliferating cells was observed (Fig. 2a).
At 72 hr, the number of SH-EP cells in sub-G1, which consists of apoptotic cells, was increased by HC-toxin approximately up to 14-fold from 3% (±0.4%) to 41% (±0.4%) (Fig. 2b), the number of BE(2)-C up to 4-fold from 6.2% (±0.9%) to 21% (±2.2%). Similar results were obtained for all other NB cell lines investigated (n = 5, data not shown). In comparison, the changes induced by VPA were less pronounced. At 72 hr, the number of SH-EP in sub-G1 increased ∼2.5-fold from 3% (±0.8%) to 5.9% (±0.6%) (Fig. 2b), the number of BE(2)-C 3-fold from 6.2% (±0.9%) to 14% (±1.9%). Primary human fibroblasts did not show obvious changes in both cell cycle distribution and apoptosis (Fig. 2a,b).
To evaluate if HC-toxin induces comparable growth responses in primary NB cell cultures, cells were isolated from the bone marrow metastases of a relapsed INSS stage 4 NB (Fig. 3a) and treated with 20 nM HC-toxin for 36 hr. In line with the growth responses of established cell lines, HC-toxin induced a G0/G1-cell cycle arrest, a 2-fold decrease in proliferating cells and a 4-fold increase in apoptotic cells (Fig. 3b,c). We conclude that nanomolar doses of HC-toxin (IC50 < 20 nM) induce G0/G1-cell cycle arrest and apoptosis in both established NB cell lines and primary NB cell cultures but not in primary human fibroblasts. This effect is superior to that of VPA when both compounds are used at their in vitro HDAC activity IC50s.
Induction of neuronal and neuroendocrine differentiation
The therapeutic effect of most conventional anticancer drugs has been attributed for years to their ability to induce apoptotic cell death. More recently, it has been recognized that induction of differentiation of those tumor cells that survive the treatment contributes to treatment outcome in vivo.23 To elucidate if HC-toxin induces neuronal and neuroendocrine differentiation in the nonapoptotic NB cell fractions, SH-EP, WAC-2, SH-SY5Y, BE(2)-C, NGP, LA-N-1 and Kelly were exposed to 20 nM HC-toxin for varying time periods. In this panel of NB cell lines containing neuroblastic (N-type), substrate-adherent (S-type) and intermediate (I-type) phenotypic variants,24 3 differentiation parameters were evaluated, cellular morphology and protein and mRNA levels of neuronal and neuroendocrine markers.
Microscopic inspection revealed morphological differentiation with neurite outgrowth and interconnections among neurites in SH-EP, SH-SY5Y, BE(2)-C, NGP, LA-N-1 and Kelly (Fig. 4a, shown for SH-EP and BE(2)-C), in WAC-2 these changes were less pronounced. Levels of neuronal and neuroendocrine differentiation proteins were evaluated by immunofluorescence (Fig. 4b, shown for SH-EP). Protein expression of Nef3, Map2, Syp1, Syn1 and Th was higher compared to controls. Levels of mRNA were quantified by real-time RT-PCR (Fig. 4c, shown for SH-EP and BE(2)-C). In SH-EP, morphological differentiation was associated with a 2- to 10-fold higher expression of NEF3, MAP2, SYP and SYN1. These genes are involved in axonal (NEF3) and dendritic formation (MAP2) as well as in synaptic function and transmitter release (SYP, SYN1). TH, a marker for neuroendocrine chromaffin differentiation, was up-regulated threefold. In BE(2)-C, 3- to 11-fold elevated mRNA levels of MAP2, SYP and SYN1 were seen. Up-regulation of NEF3 and TH was not observed. mRNA of the intermediate filament nestin (NES), which is highly expressed in both multipotent neuroectodermal precursor cells and the infiltrating cell fractions of malignant brain tumors,25, 26 was down up to 3-fold (Fig. 4c). NES mRNA in SH-EP is low and was not analyzed. We conclude that HC-toxin induces neuronal and neuroendocrine differentiation of NB cells.
For comparison, the influence of 1 mM VPA on mRNA levels of neuronal and neuroendocrine markers was assessed. In SH-EP, VPA induced an ∼1.3- to 10-fold higher expression of NEF3, MAP2, SYP, SYN1 and TH (Fig. 4d). In BE(2)-C, NEF3 was down 2-fold (Fig. 4d), up-regulation of MAP2, SYP, SYN1 and TH was not observed. NES mRNA was elevated approximately up to 2-fold (Fig. 4d). We conclude that the influence of HC-toxin and VPA on NB cell differentiation qualitatively differs.
Suppression of colony formation and invasive growth
To investigate if HC-toxin influences colony formation and invasive growth, we used BE(2)-C and WAC-2. In contrast to SH-EP and SH-SY5Y, both cell lines were found to be well suited for these assays due to both their high colony forming efficiency in soft agar and their ability to digest and migrate through the extracellular matrix in the Matrigel invasion assay (data not shown). For anchorage-dependent colony formation, cells were seeded at low density, colonies were counted after 5 days. HC-toxin decreased anchorage-dependent colony formation dose-dependently 3- to 10-fold (Fig. 5a). For anchorage-independent colony formation, cells were treated with 20 nM HC-toxin or solvent for 48 hr. The same number of viable cells (as judged by trypan blue exclusion) from HC-toxin treated and control cultures was then seeded at low density into soft agar, colonies were counted after 10 (WAC-2) and 14 days (BE(2)-C). HC-toxin reduced the number of colonies dose-dependently 5- to 12-fold (Fig. 5b).
For invasive growth, cells were seeded at equal density into Matrigel-coated Boyden chambers. The number of invasive cells was evaluated after 14 hr (WAC-2) and 22 hr (BE(2)-C) and normalized to the number of cells migrating through the uncoated control Boyden chambers. HC-toxin diminished invasive growth 5- to 8-fold (Fig. 5c). We conclude that HC-toxin suppresses anchorage-dependent and -independent colony formation as well as invasive growth of NB cells.
Activation of the RB tumor suppressor network
High levels of E2F-1, N-myc, Skp2, Mad2 and survivin proteins in NB tumors signal poor prognosis.2–7 We have investigated if HC-toxin alters the protein expression of E2F-1 and its targets. Because E2F-1 activity is controlled by RB, the effect of HC-toxin on RB phosphorylation was analyzed. In addition, the influence of HC-toxin on the expression of the cyclin-dependent kinase (Cdk)-inhibitors p21cip1/waf-1, p27kip1, p15INK4b and p16INK4a within the RB network was assessed. BE(2)-C, NGP, LA-N-1 and Kelly were exposed to different concentrations of HC-toxin for varying time periods. Expression of E2F-1 and its targets, phosphorylation of RB and Cdk-inhibitor levels were assessed by Western analysis.
HC-toxin induced similar changes in all 4 NB cell lines investigated and results are exemplarily shown for BE(2)-C. E2F-1, N-myc, Skp2, Mad2 and survivin protein levels were reduced by HC-toxin (Fig. 6a). The level of the hypophosphorylated active form of RB (pRB) was increased (Fig. 6a). p21cip1/waf-1 and p27kip1 levels were up-regulated (Fig. 6b). p15INK4b and p16INK4a levels appeared unaltered at 24–48 hr (data not shown). At 120 hr, both were increased over controls (Fig. 6c). Taken together, these results demonstrate an activation of the RB tumor suppressor network in NB cells by HC-toxin.
Better risk stratification, dose intensification of chemotherapeutic regimens followed by autologous bone marrow transplantation and minimal residual disease (MRD) therapy have improved the survival rates of children with high-risk NB. MRD therapy is based on 13-cis retinoic acid, which was shown both to induce differentiation of NB cells in vitro and to improve survival rates when administered orally at a high-dose intermittent schedule after bone marrow transplantation.27–30 However, all these approaches have not principally altered the scenario of early and late relapses due to MRD composed of resistant clones. Therefore, new, less toxic therapeutics are warranted.
This study has analyzed the potential of HC-toxin, an HDACI with cyclic tetrapeptide structure, for NB treatment. Suppression of tumorigenicity was obtained at nanomolar dosages, which is in line with previous findings in both breast and colon tumor cell lines.31–33 Different NB cell lines were similarly sensitive independent of MYCN status, suggesting that patients with both amplified and nonamplified MYCN NB could benefit from HC-toxin treatment. In response to millimolar doses of the anti-epileptic drug VPA, one of the first known HDACIs,34 different changes in the cell cycle distribution of NB cell lines with and without amplified MYCN were observed. Nanomolar doses of HC-toxin induced both in established NB cell lines with different MYCN status and in primary NB cells derived from fresh tumor specimens a G0/G1-cell cycle arrest, which was accompanied by a decrease in the number of proliferating cells. This shows that unlike VPA, HC-toxin elicits similar growth responses in NB cells with different MYCN status. Because of the apparently unchanged growth behavior seen for primary fibroblasts despite the observed increase in histone H4 acetylation, a preferential cytotoxicity of HC-toxin to tumor cells appears in line with the low and clinically manageable toxicity of other HDACIs that are in clinical validation.
In contrast to cytotoxic agents of other activity classes, HC-toxin displayed a similar anti-tumoral activity against adherent cell mono- and multilayer colonies as well as against spheroids in both suspension and soft agar. The lower number of actively growing cells in both multilayer colonies and spheroids resembles that in solid tumors in vivo more than the high number of cells actively growing in adherent mono-layers.35 The anti-tumoral activity of HC-toxin therefore seems not restricted to the relatively low fraction of actively growing cells, which limits the antitumoral activity of many other cytotoxic agents.36
Neuronal differentiation induced by HC-toxin resembles the molecular phenotype of INSS stage 4S NB, tumors with excellent patient outcome.2 Unlike INSS stage 4 NB, this phenotype is characterized by the increased expression of genes related to both neuronal differentiation and function.37 The partly varying induction of differentiation markers by HC-toxin seen in SH-EP and BE(2)-C is likely to reflect the heterogeneity of human NB tumors. Both cell lines harbor distinct programs mirrored by specific phenotypes.
While VPA and HC-toxin similarly influenced the differentiation markers in SH-EP, a varying regulation was observed in BE(2)-C. Both compounds were used in concentrations that caused a 50% inhibition of HDAC activity in vitro in non cell-based assays and that similarly increased the level of acetylated histone H4. Thus, compared with VPA, HC-toxin had superior effects on cell proliferation and apoptosis and exhibited qualitatively different effects on cell cycle distribution and neuronal differentiation of NB cells. This highlights the varying influence of these 2 HDACIs on NB cells, possibly due to their varying affinity to the different HDAC family members. Recently, high HDAC1 mRNA expression was found to be correlated with multidrug resistance in NB cell lines and inhibition of HDAC1 expression or activity by depsipeptide, another cyclic tetrapeptide type HDACI, was shown to enhance the cytotoxicity of chemotherapeutic drugs in those cell lines.38
The cellular changes of NB cell lines were accompanied by an activation of the RB tumor suppressor network.39 This included increased levels of both pRB and several Cdk-inhibitors, which is in line with previous findings describing an up-regulation of pRB and p21cip1/waf1 in NB cells by the HDACI BL1521.40 In addition, reduced levels of E2F-1 and its targets were observed. HC-toxin thereby represents a potent modulator of a functionally critical pathway in NB cells. As E2F-1, N-myc, Skp2, Mad2 and survivin are found at high levels in NB tumors with poor prognosis,3–7 repressing these oncogenic cell cycle proteins could be a novel strategy for therapeutic intervention. The mechanistic basis of the RB network activation by HC-toxin and the causal connection between the observed transcriptional and cellular changes remain to be elucidated. In addition, future studies will both compare the antitumoral activity of HC-toxin with that of other HDACIs in NB and address the pharmacological properties of HC-toxin in mouse xenografts.14, 34, 41–43 In conclusion, nanomolar dosages of the HDACI HC-toxin induce a shift to a benign and differentiated phenotype in NB cells that is associated with an activation of the RB tumor suppressor network.
We thank Mrs. Gabriele Becker and Mrs. Monika Hoch for excellent technical assistance.