Abstract
Background/Aim: Retinoic acid (RA) induces tumor cell differentiation in diseases like acute promyelocytic leukemia or high-risk neuroblastoma. However, the formation of resistant cells, which results from dysregulation of different signaling pathways, limits therapy success. The present study aimed to characterize basic regulatory processes induced by the application of RA in human neuroblastoma cells, to identify therapeutic targets independent of the often amplified oncogene MYCN. Materials and Methods: In MYCN-amplified Kelly and MYCN non-amplified SH-SY5Y cells, different assays were employed to quantify the viability and cytotoxicity, while RA-mediated expression changes were examined using genome-wide gene expression analysis followed by quantitative PCR. Enzyme-linked immunoabsorbent assays (ELISA) and western blots were used to determine the levels or activation of the examined proteins. Results: In Kelly cells, treatment with 5 μM RA for 3 days significantly reduced the cell number due to attenuated proliferation, while SH-SY5Y cells were less responsive. An up-regulation of the RA-metabolizing enzymes CYP26A1 and CYP26B1 was observed in both cell lines, and co-treatment with the selective CYP26 inhibitor talarozole markedly decreased cell viability. When RA and ketoconazole, which inhibits CYP26 as well as RA-degrading CYP3A enzymes, were co-administered, not only cell survival was impaired in both cell lines, but also the release of hepatocyte growth factor (HGF). Accordingly, co-application of the c-Met inhibitor tepotinib and RA or ketoconazole substantially decreased cell viability. Conclusion: Independent of MYCN amplification, inhibitors of RA metabolism or HGF signaling might prevent the emergence of RA-resistant neuroblastoma cells when co-applied with RA.
Neuroblastoma is the most common extracranial solid tumor in children and accounts for approximately 15% of pediatric cancer-related mortality (1, 2). Dysregulated differentiation and cell death of sympathoadrenal neural crest cells during embryonal development drives oncogenesis and promotes transformation to malignant neuro-ectodermal precursor cells, which can metastasize in advanced stages (1, 2). Clinical manifestations range from localized tumors with spontaneous regression to aggressively metastatic and often relapsing disease (1, 2). Especially the treatment of high-risk patients requires a multimodal approach, which includes a maintenance therapy for the removal of residual neuroblastoma cells with the GD2 antibody dinutuximab or retinoids. Still, the development of drug resistance limits overall survival rates to 50-60% (3, 4).
Retinoids are vitamin A derivatives and important mediators of cellular differentiation, growth or apoptosis, which predisposes them for the therapeutic use in acute promyelocytic leukemia or high-risk neuroblastoma. Still, the emergence of resistant tumor cells often leads to treatment failure. Numerous studies have examined the mechanisms of retinoid resistance in neuroblastoma and described a plethora of regulations, which account for different resistance phenotypes. So far, a set of biomarkers or super-enhancers were characterized to predict sensitivity or resistance to retinoids (5, 6). A different approach to identify potential drug targets is to focus on signaling molecules, which are consistently induced by treatment with RA. Often, mediators involved in intracellular binding or degradation of RA are up-regulated (7-9) or BDNF signaling is stimulated as observed by induction of neurotrophic tyrosine kinase 2 (NTRK2) (10, 11).
Ketoconazole is primarily known as a broad-spectrum antifungal drug. Additionally, by inhibiting the activity of relevant drug metabolizing cytochrome P450 enzymes, e.g., CYP3A4/A5, CYP1A2, CYP2C9/C8, CYP2C19, CYP2B6, CYP26A1 and CYP26B1 (12-15), ketoconazole has been used as a probe drug inhibitor in clinical development and as a pharmacokinetic booster to elevate plasma levels of drugs with low bioavailability (16-18). However, due to interfering with additional intracellular targets, it has also been of interest as an anticancer substance (19-21).
The aim of the study was to identify signaling molecules, which contribute to RA resistance in neuroblastoma and can be used as therapeutic targets to maintain RA-induced tumor cell differentiation independent of MYCN amplification.
Materials and Methods
Cell culture. Human Kelly and SH-SY5Y neuroblastoma cells were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ; Braunschweig, Germany; RRID: CVCL_2092 and CVCL_0019) and used from passage 1 to 20 for the experiments. Kelly cells were cultured on 10-cm plates in RPMI 1640 medium (Thermo Fisher Scientific, Darmstadt, Germany) supplemented with 10% fetal bovine serum (Bio&Sell, Nürnberg, Germany) at 37°C and 5% CO2. SH-SY5Y cells were kept under the same conditions in Dulbecco’s modified Eagle’s medium (Thermo Fisher Scientific) supplemented with 20% fetal bovine serum (Bio&Sell, Nürnberg, Germany). Cells were mycoplasma-free as monthly tested with the Venor®GeM qOneStep kit (Minerva Biolabs, Berlin, Germany). For the experiments, both cell lines were plated at low density (0.05×105 cells per well on 96-well plates; 0.1×106 cells per well on 6-well plates; 1.2-1.8×106 cells per 10-cm plate) and incubated with the following substances: all-trans retinoic acid (5 μM), cobicistat (10 μM, Santa Cruz, Heidelberg, Germany), ketoconazole (10 μM, Santa Cruz), talarozole (10 μM, Santa Cruz) or tepotinib (5 μM, Absource Diagnostics, Munich, Germany). The substances were applied to the cells for the indicated duration. Cells termed as controls were routinely incubated with a comparable amount of DMSO if used as a solvent.
If not indicated otherwise, chemicals were purchased from Sigma-Aldrich (Taufkirchen, Germany) or Carl Roth (Karlsruhe, Germany).
Determining viability, proliferation and cytotoxicity of Kelly and SH-SY5Y cells. Cells were grown and stimulated on a 96-well-plate (0.05×105 cells per well). To determine the amount of metabolically active cells within three days, 50 μl of the ATP-detecting CellTiter-Glo® reagent (Promega, Mannheim, Germany) were added to each well. The contents were mixed for 2 min, and the plate was incubated for 10 min at room temperature. The plate was placed into an Infinite M200 reader (Tecan, Crailsheim, Germany), and luminescence was recorded.
To determine the number of viable cells after three days, trypan blue exclusion assays were performed. The cells were detached from the plates, centrifuged (1,000×g for 10 min at room temperature) and resuspended in 1 ml PBS. Twenty microliters of cell suspension were added to 20 μl trypan blue staining solution (Sigma Aldrich) and the living cells were counted. For the analysis of cell viability after ten days, the cells were grown on a 6-well-plate (0.1×106 cells per well) and incubated with ketoconazole, talarozole, retinoic acid or a combination of retinoic acid with one of the inhibitors for ten days. The cells were provided with fresh medium containing the different substances every 2-3 days. The percentage of viable cells was manually counted in at least three fields of view per sample (bright field, 20× magnification).
Cell proliferation was measured using a Bromo-2-deoxyuridine (BrdU) incorporation assay (Roche Applied Sciences, Mannheim, Germany) according to the manufacturer’s instructions. Cells grown on a 96-well-plate (5×105 cells per well) were incubated with BrdU for 16 h and fixed. After incubation with an anti-BrdU-POD antibody and the subsequent substrate reaction, absorbance was measured at 370 nm (reference wavelength 492 nm) with an Infinite M200 reader (Tecan).
To determine cytotoxicity, CytoTox-Glo™ assays (Promega) were used. All reagents were prepared directly before use as recommended by the manufacturer. The cells were grown and stimulated on a 96-well-plate (0.05×105 cells per well) and 50 μl CytoTox-Glo™ reagent were added to each well. The contents were briefly mixed, and the plate was incubated for 15 min at room temperature. Luminescence was recorded using an Infinite M200 reader (Tecan). Next, 50 μl of Lysis Reagent were added to all wells. After briefly mixing the contents, the plate was incubated for 15 min at room temperature before luminescence was measured again. By adding the lysis step, it was possible to distinguish between the luminescent signal of viable cells and of the experimentally dead cells.
Whole cell extracts. Whole cell lysates were generated from subconfluent cells. Before harvesting, cells were washed with PBS. The cells were resuspended in denaturing lysis buffer [20 mM Tris, pH 7.4; 2% sodium dodecyl sulfate (SDS); 1% phosphatase inhibitor and 1% protease inhibitor (both from Roche Applied Sciences)], incubated at 95°C for 5 min, briefly sonicated, and centrifuged to remove insoluble material (15,000×g for 15 min; 4°C).
Western blots. Twenty μg of protein extract were separated on 10–15% SDS-polyacrylamide gels and transferred to nitrocellulose transfer membranes (LI-COR Biosciences, Bad Homburg, Germany). The membranes were blocked with Odyssey Blocking Buffer (TBS) (LI-COR Biosciences) and incubated with the primary antibodies according to the manufacturer’s recommendations overnight at 4°C. After four washing steps with TBST (TBS plus 0.1% Tween-20), the membranes were incubated with the corresponding IRDye 800CW or IRDye 680RD secondary antibody (LI-COR Biosciences) for 1 h. After four more washing steps with TBST, the membranes were kept in TBS before scanning with the AutoScan function of an Odyssey CLx reader (LI-COR Biosciences). To normalize protein activation or cleavage, two-colour detection was used. Densitometrical and statistical analysis was performed with the Image Studio™ Lite Software (LI-COR Biosciences). Antibodies against the following targets were used: Akt (Santa Cruz, sc-5298, 1:1,000, monoclonal, mouse anti-human), β-actin (A5441, 1:2,000, monoclonal, mouse anti-human); CYP26A1 (Biosource, San Diego, USA, MBS822084, 1:750, polyclonal, rabbit anti-human), CYP26B1 (Santa Cruz, sc-293493, 1:1,000, monoclonal, mouse anti-human), JNK (Santa Cruz, sc-7345, 1:1,000, monoclonal, mouse anti-human), phospho-Akt (Cell Signaling Technology, Frankfurt, Germany, 4060, 1:1,000, polyclonal, rabbit anti-human), phospho-JNK (Promega, Mannheim, Germany, V7938, 1:1,000, polyclonal, rabbit anti-human).
ELISA. To measure HGF and IGF2 levels in cell culture supernatants, Quantikine® solid-phase ELISAs (R&D Systems/Bio-Techne, Wiesbaden, Germany) were used according to the manufacturer’s instructions. Cells were grown and stimulated on a 96-well-plate (0.05×105 cells per well). After standards or samples were added to each well, the plates were incubated at room temperature followed by washing (three times) and an incubation with HGF (1 h) conjugate at room temperature. After further washing steps, Streptavidin-HRP was added to the wells before proceeding to the substrate reaction. When the stop solution was added, the absorbance was measured within 30 min at 450 nm (reference wavelength 540 nm) with an Infinite M200 reader (Tecan).
RNA extraction and quantitative PCR. Total RNA was isolated using the E.Z.N.A Total RNA Kit II (Omega Bio-Tek, Norcross, GA, USA) according to the manufacturer’s instructions. For mRNA analyses, 1,000 ng of total RNA was reversely transcribed using random hexamer primers and the Transcriptor High Fidelity cDNA Synthesis Kit (Thermo Fisher Scientific) on the GeneAmp PCR System 9700 device (Applied Biosystems/Thermo Fisher Scientific, Carlsbad, CA, USA) according to the manufacturer’s recommendations. qRT-PCR of CYP26A1 mRNA (Hs00175627_m1), CYP26B1 mRNA (Hs01011223_m1), CRABP2 mRNA (Hs00275636_m1) was performed in triplicates using YWHAZ (tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein zeta, Hs03044281_g1) as an internal control, with Universal Master Mix II, without UNG (Thermo Fisher Scientific) on the QuantStudio 7 device (Thermo Fisher Scientific) under default cycling conditions. Data were analyzed using the ΔΔCt-method as previously described (22).
Genome-wide gene expression analysis. Genome-wide gene expression on untreated and retinoic acid-treated Kelly and SH-SY cells was investigated using Clariom S arrays (Thermo Fisher Scientific) according to the manufacturer’s protocol [as previously described in (22)]. For this purpose, cells were treated 3 d with retinoic acid (5 μM) followed by RNA isolation as described above. For the arrays, 100 ng RNA/sample and four technical replicates were used. Data were analyzed with the Transcriptome Analysis Software v.4.0.3 (Thermo Fisher). Genes with a fold change ±2 and a false discovery rate (FDR)-corrected p-value of padj≤0.05 were considered as differentially expressed between the respective samples. Venn diagrams were obtained using the TAC software, Wikipathways with the DAVID annotation tool [DAVID Bioinformatics Resources (23, 24)], and STRING analysis using string-db.org v.12.0 with high confidence settings.
Statistical analyses. All data were obtained from independent sets of experiments. The numbers of the repetitions are included in the figure legends of the respective experiments, whereby biological replicate is equivalent to one repetition. For assays performed in 96-well plates, generally two technical replicates were used for each sample. All quantitative data represent the mean±standard deviation of these independent experimental replications. Statistical analyses were performed with GraphPad prism software (Version 8 for Windows, La Jolla, CA, USA) using one-way ANOVA with Bonferroni post hoc test.
Results
An overview of the experimental strategy of the present study is provided in Figure 1.
Flow chart of the experimental procedures. (A) General experimental setups to evaluate the cellular effects of retinoic acid (RA) and for the identification and verification of target structures. (B) Scheme of the inhibitor experiments to examine the relevance of different intracellular mediators for enhancing the effects of RA.
Cellular effects of retinoic acid on Kelly and SH-SY5Y cells. First, the cellular effects of RA on Kelly and SH-SY5Y cells were examined. For the analysis of cellular viability (intracellular ATP levels), proliferation (BrdU incorporation) and cytotoxicity, 5 μM RA were applied to the cells for 3 days. In Kelly cells, 5 μM RA significantly decreased intracellular ATP levels and BrdU incorporation by 31% or 34%, respectively (p<0.001), while cytotoxicity was comparable to control cells (Figure 2A). When SH-SY5Y cells were incubated with 5 μM RA, ATP levels and BrdU incorporation were markedly increased to 122% (p<0.001) or 111% (p<0.01), respectively. In contrast, cytotoxicity was reduced to 82% (p<0.05).
Viability of Kelly and SH-SY5Y cells in response to treatment with retinoic acid (RA). (A) Effect of 5 μM retinoic acid (RA) on cellular viability (intracellular ATP levels, CellTiter-Glo™ assays), proliferation [bromo-2-deoxyuridine (BrdU) incorporation] and cytotoxicity (CytoTox-Glo™ assays) in Kelly or SH-SY5Y cells after 3 days normalized to control cells. (B) Number of Kelly or SH-SY5Y cells after an incubation with 5 μM RA for 3 or 10 days compared to control cells. Control=100; ***p<0.001, **p<0.01, *p<0.05, for control vs. RA-treated cells. Data represent the mean±standard deviation of five independent experiments (five biological replicates for all experiments; five biological replicates with two technical replicates each for ATP levels, BrdU incorporation and cytotoxicity).
Additionally, the number of viable cells was determined after incubation with RA for 3 or 10 days. Similar to ATP levels, treatment with RA substantially reduced the number of Kelly cells to 73% after 3 days and to 79% after 10 days (p<0.001). In SH-SY5Y cells, incubation with 5 μM RA significantly increased the cell number to 128% (p<0.001) after 3 days, while it was comparable to control levels after 10 days (Figure 2B).
Gene expression pattern after treatment with retinoic acid in Kelly and SH-SY5Y cells identified common changes in CYP enzyme expression. To identify the molecular changes contributing to the effects of RA on Kelly and SH-SY5Y cells, genome-wide gene expression was analyzed comparing cells after 3 days treatment with RA to unstimulated cells. As expected, Kelly and SH-SY5Y cells showed diverse gene expression patterns in their untreated state, but also after RA treatment (Figure 3A). After incubation with 5 μM RA, the number of differentially expressed genes (DEGs) varied between 2579 (Kelly) and 4626 (SH-SY5Y) (Table I). To identify common DEGs, deregulated genes in both cell lines were compared obtaining 651 overlapping genes (Figure 3B). Subsequent STRING analysis identified several nodes altered in both cell lines (Figure 3C). Further, pathway prediction revealed Vitamin A and carotenoid metabolism being significantly enriched after RA treatment in both cell lines. The main genes enriched in this pathway were CYP26A1, CYP26B1, CRABP2, RARB, BCO2, DHRS3 (Table I). Among these genes, CYP26A1 had the highest fold change in both cell lines (Kelly: 2503, padj=8.45×10−20; SH-SY5Y: 4647, padj=1.15×10−20).
Changes of gene expression patterns in Kelly and SH-SY5Y cells after treatment with retinoic acid (RA). (A) Hierarchical clustering of genes altered by retinoic acid (RA, 5 μM) treatment for 3 days in Kelly and SH-SY5Y cells with a fold change ±2 and false discovery rate (FDR)-corrected p-value <0.05. Red – high expression, blue – low expression. Numbers indicate the four replicates of the experiments. (B) Venn diagram of genes altered by RA comparing Kelly and SH-SY5Y cells. (C) STRING analysis of genes differentially altered by RA treatment in both Kelly and SH-SY5Y cells. For each cell line, the experiments were performed with two biological replicates with two technical replicates each.
Differentially expressed genes and enriched pathways in Kelly and SH-SY5Y cells after treatment with RA.
Expression and protein levels of CYP26A1 and CYP26B1 in Kelly and SH-SY5Y cells and the cellular effects of their inhibition. Next, expression levels of CYP26A1 and CYP26B1 were validated by quantitative PCR after incubation with RA. Treatment with 5 μM RA for 3 days significantly induced the expression of CYP26A1 (13.1-fold, p<0.001) and CYP26B1 (65.7-fold, p<0.001) in Kelly cells (Figure 4A). Similarly, protein levels were increased compared to untreated cells for CYP26A1 (9-fold, ±19%, p<0.001) and CYP26B1(13-fold, ±17%, p<0.001) (Figure 4B).
CYP26A1 and CYP26B1 mRNA and protein levels after treatment of Kelly and SH-SY5Y cells with 5 μM retinoic acid (RA) for 3 days. Expression of CYP26A1 or CYP26B1 was analyzed using RT-qPCR in Kelly (A) or SH-SY5Y cells (C). Protein levels of CYP26A1 or CYP26B1 were determined in response to RA (5 μM) by Western blots in Kelly (B) or SH-SY5Y cells (D). Control=1; ***p<0.001, **p<0.01 for control vs. RA-treated cells. Data represent the mean±standard deviation of five independent experiments (five biological replicates for all experiments; five biological replicates with two technical replicates each for qRT-PCR). c, Control.
When SH-SY5Y cells were incubated with 5 μM RA for 3 days, expression of CYP26A1 (8.1-fold, p<0.001) and CYP26B1 (8.5-fold, p<0.001) was enhanced (Figure 4C). Likewise, protein levels of CYP26A1 (3-fold, ±0.73%, p<0.01) and CYP26B1 (80-fold, ±8%, p<0.001) were increased.
In order to analyze if CYP26 inhibition could improve RA-mediated effects, the CYP26 inhibitor talarozole (10 μM) (25, 26) was simultaneously applied with RA. After Kelly cells were incubated with RA or the combination of RA and talarozole for 10 days, the number of cells was reduced to 80% (p<0.01) in response to 5 μM RA, respectively (Figure 5A). Concomitant application of talarozole and RA decreased the number of Kelly cells to 31.8% (p<0.001) compared to control cells (Figure 5A). When SH-SY5Y cells were incubated with RA or a combination of RA and talarozole for 10 days, RA did not lead to any changes compared to control cells. Concomitant stimulation with talarozole and RA decreased the cell number to 65.6% (p<0.001) (Figure 5A).
Inhibition of CYP26 by talarozole. (A) The number of Kelly or SH-SY5Y cells in response to 5 μM retinoic acid (RA) or a combination of 5 μM RA and 10 μM talarozole (tala) after 10 days of exposure. (B) Kelly or SH-SY5Y cells were treated with 5 μM RA or a combination of RA and talarozole for 3 days, before expression of CRABP2 was analyzed using RT-qPCR. (C, D) mRNA levels of CYP26A1 and CYP26B1 after treatment with RA or a combination of RA and talarozole for 3 days in Kelly (C) or SH-SY5Y cells (D). Control=100%; ***p<0.001, **p<0.01, for control vs. cells that were incubated with RA and/or talarozole; ###p<0.001, ##p<0.01, #p<0.05, for RA-treated cells vs. cells incubated with RA and talarozole. Data represent the mean±standard deviation of four independent experiments (four biological replicates for all experiments; four biological replicates with two technical replicates each for qRT-PCR).
Since elevated RA levels might be responsible for the reduced number in both cell lines, the expression of the retinoic acid binding protein CRABP2 was examined, which is a marker for RA-mediated differentiation in neuroblastoma cells (27). In Kelly cells, incubation with 5 μM RA for 3 days, mRNA levels of CRABP2 were significantly increased (13.2-fold, p<0.001). After co-application of talarozole (10 μM), expression was further enhanced (19.8-fold, p<0.001 vs. control cells, p<0.01 vs. RA-treated cells). When SH-SY5Y cells were stimulated with RA, mRNA levels of CRABP2 were substantially elevated (37.3-fold, p<0.001). After co-treatment with RA and talarozole, CRABP2 expression was even higher (46.2-fold, p<0.001 vs. control cells, p<0.05 vs. RA-treated cells) (Figure 5B).
Next, it was analyzed if the co-application of talarozole (10 μM) and RA (5 μM) affected the expression of CYP26A1 or CYP26B1. When Kelly cells were treated with 5 μM RA, mRNA levels of CYP26A1 were significantly increased (11.9-fold, p<0.001) after 3 days. Co-treatment with RA and talarozole further enhanced CYP26A1 induction (15.3-fold) compared to control cells (p<0.001) as well as to RA-treated cells (p<0.01) (Figure 5C). Similarly, CYP26B1 expression was induced in response to RA (64.3-fold, p<0.001) after 3 days. The simultaneous incubation with RA and talarozole further increased mRNA levels of CYP26B1 (84.7-fold, p<0.001) (Figure 5C). In SH-SY5Y cells, treatment with 5 μM RA significantly induced the expression of CYP26A1 (7.8-fold, p<0.001) after 3 days. When RA and talarozole were co-applied, mRNA levels of CYP26A1 were further increased (11.8-fold, p<0.001) (Figure 5D). Likewise, CYP26B1 expression was enhanced (8.3-fold, p<0.001) after incubation with RA for 3 days. Co-stimulation with RA and talarozole increased mRNA levels of CYP26B1 (9.8-fold,), which was significantly higher than in control cells (p<0.001) (Figure 5D).
Cellular effects of retinoic acid and ketoconazole. In the following set of experiments, it was analyzed if a more extensive inhibition of RA degradation would yield a more sustainable effect on cellular viability. Therefore, CYP3A enzymes were additionally targeted, since they are known to be part of RA metabolism (28). For the simultaneous inhibition of CYP26 and CYP3A enzymes, ketoconazole was used. To assess the effects of blocking CYP3A activity, the cells were incubated with the selective compound cobicistat (29). After 10 days, incubation with ketoconazole decreased the number of Kelly cells to 38.8% (p<0.001), whereas treatment with cobicistat did not change the cell number compared to controls. Co-application of RA and ketoconazole further lowered the cell count to 12.9% (p<0.001). The concomitant incubation with RA and cobicistat reduced the cell number to 79.8% (p<0.01), which did not differ from RA-treated cells (Figure 6A). In SH-SY5Y cells, incubation with 10 μM ketoconazole for 10 days attenuated the number of cells by 48% (p<0.001), while incubation with cobicistat did not affect the cell count. Simultaneous stimulation with RA and ketoconazole reduced the number of cells to 57% (p<0.001). Co-application of RA and cobicistat did not induce any changes (Figure 6A).
The number of viable Kelly or SH-SY5Y cells in response to retinoic acid (RA), ketoconazole and cobicistat. (A) After an incubation with ketoconazole (10 μM; ket), cobicistat (10 μM; cob), RA (5 μM) or the combination of RA and ketoconazole or cobicistat for 10 days, the number of cells was counted. (B) Western blots of Kelly or SH-SY5Y whole cell extracts of ketoconazole- (10 μM, 3 days) and RA-treated cells (5 μM, 3 days) as well as cells incubated with both substances were performed using antibodies against the phosphorylated forms of Akt or JNK. The blots were simultaneously incubated with the corresponding total kinase antibodies. (C), (D) In Kelly (C) or SH-SY5Y cells (D), the effects of 10 μM ketoconazole, 5 μM RA or a combination of both on proliferation [Bromo-2-deoxyuridine (BrdU) incorporation] or cytotoxicity (CytoTox-Glo™ assays) were determined after 3 days (normalized to controls). Control=100%; ***p<0.001, **p<0.01, for control vs. cells that were exposed to RA, ketoconazole, cobicistat or a combination of RA and cobicistat or ketoconazole; ###p<0.001, #p<0.05, for RA-treated cells vs. cells incubated with RA and ketoconazole; •••p<0.001, ••p<0.01, •p<0.05 for cells incubated with ketoconazole vs. cells treated with RA and ketoconazole. Data represent the mean±standard deviation of four independent experiments (four biological replicates for all experiments; four biological replicates with two technical replicates each for BrdU incorporation and cytotoxicity).
As the application of ketoconazole with or without RA substantially reduced the number of cells, further experiments were conducted to examine proliferation and cytotoxic reactions in both cell lines. For the analysis of survival-associated signal transduction, the phosphorylation levels of Akt and JNK were examined after incubation with 10 μM ketoconazole, 5 μM RA or the concomitant application of both substances. After 3 days, stimulation with ketoconazole significantly enhanced the activation of JNK (1.4-fold, ±19%, p<0.01), while Akt phosphorylation was decreased by 29% (±11%, p<0.05) (Figure 6B). Treatment with RA only changed the activation of Akt by increasing its phosphorylation (1.3-fold compared to controls, ±9%, p<0.05). After concomitant stimulation with RA and ketoconazole, the activity of Akt was decreased to 75% (±10%, p<0.05), while the phosphorylation of JNK was significantly increased 1.6-fold (±16%, p<0.01) compared to controls (Figure 6B). In SH-SY5Y cells, application of ketoconazole decreased Akt activity by 21% (±9%, p<0.05) and enhanced the activation of JNK (1.4-fold, ±9%, p<0.05). Incubation with RA or co-stimulation with RA and ketoconazole increased Akt phosphorylation (1.7-fold, ±15% or 1.6-fold, ±14%, respectively, compared to controls, p<0.01), while JNK was not affected (Figure 6B).
To differentiate between the induction of cell death and decreased proliferation, Kelly cells were incubated with 10 μM ketoconazole for 3 days showing a reduced BrdU incorporation of 77% (p<0.001). Co-treatment with ketoconazole and RA further decreased proliferation to 59% (p<0.001). Cytotoxicity was enhanced by stimulation with ketoconazole (1.2-fold, p<0.01) or by co-treatment with ketoconazole and RA (1.4-fold, p<0.001 vs. controls or RA-treated cells, p<0.05 vs. ketoconazole-treated cells) for 3 days (Figure 6C). In SH-SY5Y cells, incubation with ketoconazole reduced BrdU incorporation to 75% (p<0.01) and significantly increased cytotoxicity (1.3-fold, p<0.001). After co-stimulation with ketoconazole and RA for 3 days, proliferation (98%) and cytotoxicity (88%) returned to control levels. However, when compared with ketoconazole-treated cells, proliferation was higher (p<0.05) by co-application of RA and ketoconazole, while cytotoxicity was lower (p<0.01) (Figure 6D).
Next, it was examined if co-application of ketoconazole (10 μM) altered the RA-mediated regulation of CYP26 enzymes. When Kelly cells were treated with 5 μM RA, mRNA levels of CYP26A1 were increased (13.2-fold, p<0.001) after 3 days. Co-stimulation of RA and ketoconazole significantly enhanced CYP26A1 induction (18.9-fold, p<0.001) (Figure 7A). In contrast, the expression of CYP26B1 was comparably induced in response to RA (64.8-fold, p<0.001) or after co-incubation with RA and ketoconazole (66.3-fold, p<0.001) (Figure 7B).
mRNA levels of CYP26A1, CYP26B1 and CRABP2 after treatment with retinoic acid (RA) or a combination of RA and ketoconazole. (A, B) Kelly cells were treated with 5 μM RA or a combination of RA and 10 μM ketoconazole (ket) for 3 days, before expression of CYP26A1 (A) or CYP26B1 (B) was analyzed using RT-qPCR. (C) SH-SY5Y cells were treated with RA or a combination of RA and ketoconazole for 3 days, before expression of CYP26A1 or CYP26B1 was analyzed using RT-qPCR. (D) Kelly or SH-SY5Y cells were treated with 5 μM RA or a combination of RA and 10 μM ketoconazole for 3 days, before expression of CRABP2 was analyzed using RT-qPCR. Control=1; ***p<0.001 for control vs. cells treated with RA or RA and ketoconazole; ###p<0.001, ##p<0.01, ##p<0.05, for RA-treated cells vs. cells incubated with RA and ketoconazole. Data represent the mean±standard deviation of five independent experiments (five biological replicates with two technical replicates each).
In SH-SY5Y cells, treatment with 5 μM RA significantly induced the expression of CYP26A1 (7.4-fold, p<0.001) after 3 days. When RA and ketoconazole were co-applied, mRNA levels of CYP26A1 were further increased (12.4-fold, p<0.001) (Figure 7C). Likewise, CYP26B1 expression was enhanced after incubation with retinoic acid 9.4-fold (p<0.001), while the simultaneous stimulation with RA and ketoconazole increased mRNA levels of CYP26B1 14.8-fold, which was significantly higher than in control (p<0.001) and RA-treated cells (p<0.01) (Figure 7C).
Again, the expression of CRABP2 was examined in both cell lines. Incubation with 5 μM RA for 3 days increased mRNA levels of CRABP2 (12.8-fold, p<0.001) in Kelly cells. After co-application of ketoconazole (10 μM), CRABP2 expression was further enhanced (18.7-fold, p<0.001 vs. control cells, p<0.01 vs. RA-treated cells). When SH-SY5Y cells were stimulated with RA for 3 days, mRNA levels of CRABP2 were substantially elevated (37.2-fold, p<0.001). After co-treatment with RA and ketoconazole, CRABP2 expression was even higher (45-fold, ±4.8%, p<0.001 vs. control cells, p<0.05 vs. RA-treated cells) (Figure 7D).
Inhibition of HGF signaling by ketoconazole. In the last sets of experiments, we examined further aspects of ketoconazole-mediated signaling. Due to the concurrent effects on proliferation and cell death, an impairment of growth factor regulation was plausible. Since genes of HGF and the IGF2 signaling pathways were differentially regulated by application of RA in both cell lines (Figure 3C), we analyzed the effects of ketoconazole stimulation on HGF or IGF2 levels in cell culture supernatants. In Kelly cells, treatment with 10 μM ketoconazole for 3 days selectively attenuated the release of HGF by 49% (p<0.001), while HGF levels in SH-SY5Y supernatants was comparable to control cells (Figure 8A). No changes were observed in IGF2 release (data not shown). Co-treatment with RA and ketoconazole for 3 days significantly reduced the amount of HGF in Kelly cells to 46% (p<0.001) and to 48% (p<0.001) in SH-SY5Y cells (Figure 8A).
Ketoconazole-mediated regulation of hepatocyte growth factor (HGF) release and cell viability in response to ketoconazole, retinoic acid (RA) and tepotinib. (A) Release of HGF after incubation with ketoconazole (ket, 10 μM) or a combination of ketoconazole and 5 μM RA for 3 days as determined by ELISA. (B) Intracellular ATP levels were determined by CellTiter-Glo™ assays in Kelly or SH-SY5Y cells treated with tepotinib (tep, 5 μM), ketoconazole (ket, 10 μM), RA (5 μM) or a combination of two or all substances for 3 days. Control (C)=1 or 100; ***p<0.001, *p<0.05, for control vs. cells exposed to ketoconazole, RA or a combination of the different compounds. •••p<0.001, •p<0.05, for tepotinib-treated cells vs. co-stimulated cells. Data represent the mean±standard deviation of five independent experiments (five biological replicates with two technical replicates each).
To explore if the ketoconazole would sensitize the cells to the inhibition of HGF signal transduction, we treated cells with the c-Met blocker tepotinib (5 μM) alone or in combination with either ketoconazole, RA or a combination of all substances. Tepotinib treatment reduced cell viability of Kelly cells, as determined by ATP levels to 83% (p<0.05) compared to controls. When ketoconazole (10 μM) and tepotinib were simultaneously applied to the cells, viability was substantially decreased to only 3.3% (p<0.001) (Figure 8B). Co-stimulation of tepotinib and RA reduced ATP levels to 42.8% (p<0.001). After concomitant treatment with tepotinib, ketoconazole and RA for 3 days, cell viability was as low as 3.7% (p<0.001).
In SH-SY5Y cells, 5 μM tepotinib did not change cellular viability compared to control cells, while co-stimulation with 10 μM ketoconazole significantly reduced ATP levels to 65% (p<0.001). Simultaneous application of tepotinib and RA also attenuated cell viability by 20% (p<0.05). The combination of RA, ketoconazole and tepotinib decreased ATP levels to 24% (p<0.001) (Figure 8B).
These experiments show that ketoconazole alone or in combination with relevant substances can be used to attenuate survival in neuroblastoma cells, but its effect is cell-type as well as context-specific.
Discussion
RA induces differentiation in a variety of neuroblastoma cell lines (30), but altered intracellular metabolism (31) often leads to treatment failure. When investigating genome-wide expression changes after treatment of Kelly or SH-SY5Y cells with RA, a strong induction of the RA-metabolizing enzymes CYP26A1 and CYP26B1 was observed. This effect was shown to contribute to cell survival, as CYP26 inhibition by co-administered talarozole reduced viability. The application of ketoconazole proved to be more efficient in reducing the number of cells either as a single substance or in combination with RA, presumably by inhibiting retinoid metabolism. Further chemosensitizing properties of ketoconazole were linked to attenuated HGF signaling. Especially, the co-incubation of ketoconazole and the c-Met inhibitor tepotinib significantly reduced cell survival.
Due to its differentiation-promoting effects in neuroblastoma cells (30), RA was part of the post-consolidation therapy in high-risk neuroblastoma patients (4). However, overall patient survival was not improved by RA due to resistance formation, which is attributed to different compensatory mechanisms (6, 31). Treatment with RA for 3 days in Kelly cells, reduced the cell number by decreased proliferation without inducing cytotoxicity, which could be achieved with higher RA concentrations (32). As described in previous studies (33, 34), SH-SY5Y cells were less responsive.
Several pathways are considered to be responsible for the formation of RA-resistant cells or insufficient therapeutic effects (6, 35). To further address this question, we performed genome-wide gene expression analysis, which revealed the differential regulation of crucial mediators in various pathways involving proliferation, apoptosis or cell adhesion as described by several studies (36, 37). In particular, we found a distinct up-regulation of the RA-metabolizing enzymes CYP26A1 and CYP26B1 in both tested cell lines. The induction of expression of these genes has been previously shown (7, 9) and could at least be part of the potential resistance mechanism (31). In Kelly and SH-SY5Y cells, both enzymes had a very low basal expression and were significantly increased by the application of RA for 3 days. So far, only the expression of CYP26B1 had been reported in Kelly cells (38), but it is well-known that RA is an inducer of both CYP26 enzymes in other neuroblastoma cells, including SH-SY5Y cells (39-41). Moreover, in Kelly cells, mRNA levels of CYP26A1 were much lower than the expression of CYP26B1, which might have impeded detection in previous studies.
Chronic RA-mediated induction of CYP26 enzymes may, at least, be partially responsible for treatment failure (42) or a progressive clinical resistance to RA (43). By using CYP26 inhibitors, intratumoral levels of retinoic acid were increased (39) as well as retinoic acid-mediated actions (38). In Kelly and SH-SY5Y cells, co-application of the CYP26 inhibitor talarozole and RA significantly reduced cellular viability, which was accompanied by a markedly elevated expression of CYP26A1, CYP26B1 and CRABP2 implying enhanced intracellular RA levels. Although the benefit of CYP26 inhibition as co-treatment for RA application has been demonstrated in several disease models (42), no CYP26 inhibitor has yet been approved for clinical use due to adverse effects, limited selectivity or the problem of effective dose delivery (38, 44).
Apart from CYP26A1 and CYP26B, CYP3A4/5 contributes to degradation of RA (7, 28) and can also be induced by RA (40). Accordingly, co-incubation with CYP3A inhibitors and RA might sustain RA-induced differentiation, as shown in myeloid leukemia cells (45). However, simultaneous application of RA and the specific CYP3A4/5 inhibitor cobicistat did not increase RA-mediated effects in Kelly cells and only reduced ATP levels in SH-SY5Y cells, when compared to RA-treated cells. Thus, co-treatment with RA and inhibitors of CYP3A4/5 was not sufficient to substantially impair survival in neuroblastoma cells.
CYP26A1 and CYP26B as well as CYP3A4/5 are all sensitive to inhibition by ketoconazole (13-15, 46). When RA and ketoconazole were simultaneously applied to Kelly or SH-SY5Y cells, the number of viable cells was markedly reduced. In Kelly cells, co-treatment with ketoconazole and RA not only decreased proliferation, but also induced cytotoxic reactions. Concurrently, activity of Akt was decreased, while JNK phosphorylation was enhanced, which is characteristic of cell death initiation (47). Co-application of ketoconazole and RA also led to a markedly increased expression of CYP26A1, CYP26B1 and CRABP2 in both cell lines, which has not been described before, but would indicate enhanced intracellular RA levels, as observed for talarozole. Thus, simultaneous inhibition of CYP26A1, CYP26B1 and CYP3A4/5 by ketoconazole significantly improved RA-mediated effects in neuroblastoma cells. Previous studies also reported reduced RA metabolism by ketoconazole in cancer patients, which increased plasma levels of RA (48) and even reduced recurrence rate in patients with superficial bladder cancer (18).
Even when ketoconazole was applied to cells without addition of RA, cellular viability of Kelly and SH-SY5Y cells was significantly reduced. The findings on cell toxicity of 10 μM ketoconazole stand in line with studies that described antitumor effects of ketoconazole in different cancer cell lines, e.g. breast cancer or pancreatic carcinoma cells (19, 20). As for the changes in intracellular signaling, treatment with ketoconazole attenuated the phosphorylation of Akt and increased JNK activation. Especially when applied in higher concentrations, the JNK pathway was found to be important for ketoconazole-induced apoptosis (47). A reduction of Akt phosphorylation by ketoconazole, in contrast, had only been observed before in co-treatment of ketoconazole with substances that would normally elevate Akt activation (49, 50). However, Akt is considered crucial for the viability of neuroblastoma cells (51), which would argue for a contribution of Akt activity to the survival in Kelly and SH-SY5Y cells.
So far, our results show that ketoconazole not only potentiated the antitumor effects of RA by attenuating its degradation, but also significantly reduced the survival of Kelly and SH-SY5Y cells. As cell proliferation and Akt activity were decreased by ketoconazole, an effect on growth factor pathways seemed plausible, which had been observed for ketoconazole in VEGF signal transduction (50). In addition, genome-wide gene expression analysis revealed that application of RA altered the expression of mediators being part of the HGF or IGF2 cascades, namely HGF, NTRK2 and IGFBP2/3/5. When analyzing the release of HGF and IGF2 in cell culture supernatants of Kelly or SH-SY5Y cells, the amount of IGF2 in response to ketoconazole remained on control levels, while HGF levels were significantly reduced in Kelly cells. Co-application with RA decreased HGF release in both cell lines. So far, a direct effect of ketoconazole on HGF release had not been observed, but ketoconazole is known to alter vesicle-mediated signaling, e.g., by attenuating the secretion of MCP-1 from melanoma cells (52) or inhibiting the formation of exosomes (53) that might also contain growth factors like HGF (54). Furthermore, previous findings suggest that treatment with ketoconazole also interferes with lipid raft functionality (55). Thus, the reduction of cell viability in response to ketoconazole and the c-Met inhibitor tepotinib might be attributed to different intra- and extracellular actions in survival signaling cascades, which might even work synergistically.
Conclusion
Application of RA in neuroblastoma cells is often not sufficient to induce permanent growth inhibition or differentiation. The induction of the RA-metabolizing enzymes CYP26A1 and CYP26B1 enhances RA degradation, which sustains proliferation. Therefore, concomitant use of CYP26 inhibitors such as talarozole resulted in a significant attenuation of cellular viability ameliorating RA-mediated effects. Alternatively, combination with a pleiotropic substance like ketoconazole, which inhibits CYP26A1, CYP26B1 and CYP3A4/5 was shown to be effective. Aside from attenuating RA metabolism, ketoconazole reduced HGF release as well as the associated survival signaling and thereby improved the effects of growth factor receptor inhibitors.
Acknowledgements
The Authors would like to thank Irina Naujoks and Annika Muetze for their excellent technical assistance.
Footnotes
Authors’ Contributions
RSI and NSP conducted experiments. MK conducted experiments and reviewed the manuscript. IC reviewed and revised the manuscript. VW designed and supervised the study, designed and conducted experiments and wrote the manuscript.
Funding
This work was supported by an intramural grant of the Medical Faculty of Kiel University (Kiel, Germany).
Conflicts of Interest
The Authors declare no conflicts of interest in relation to this study.
- Received July 21, 2024.
- Revision received July 21, 2024.
- Accepted August 16, 2024.
- Copyright © 2024 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.