Human high-risk neuroblastoma cell lines UKF-NB-3 and the derived CDDP-resistant line UKF-NB-3^CDDP were provided by Prof. Jindrich Cinatl, Goethe University, Frankfurt am Main, Germany. IMR-32 was purchased from MillporeSigma and SK-N-F1 by American Type Culture Collection. The CDDP-chemoresistant cell lines IMR-32^CDDP and SK-N-F1^CDDP were derived in our laboratory from their chemosensitive parental cell lines (IMR-32 and SK-N-F1) after long-term cultivation with increasing CDDP (Sandoz Group AG) concentration (57,58). All tested cell lines were of male origin. Cells were cultured at 37°C and 5% CO₂ in Iscove's Modified Dulbecco's Medium (IMDM) supplemented with 10% (v/v) fetal bovine serum (both Thermo Fisher Scientific, Inc.). KDOAM-25 citrate (400 nM; HY-102047B; MedChemExpress) was used to inhibit KDM5s, where it was added to cells and incubated for 48 h at 37°C.

For the inhibition of KDM5 demethylases, cells were treated with KDOAM-25 citrate for 24–48 h at 37°C and 5% CO₂ (cat. no. HY-102047B; MedChemExpress) at a final concentration of 400 nM. This concentration was selected based on preliminary optimization experiments demonstrating efficient reduction of H3K4 trimethylation without affecting cell viability (data not shown). For supplementary validation experiments, KDOAM-25 was added 24 h before CDDP treatment, whereas for all main experiments presented in the manuscript, KDOAM-25 was added 48 h before CDDP administration and maintained throughout the treatment period.

To determine cell proliferation, cells were placed in 24-well plates (1×10⁵ cells per well) or 96-well plates (1×10⁴ cells per well) and seeded for 24 h at 37°C.and then cells were treated with CDDP at a final concentration of 0.6-300 µM for 48 h at 37°C. Cells were then incubated with PrestoBlue^® Cell Viability Reagent (Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. Fluorescence was measured at an excitation wavelength of 560 nm and an emission wavelength of 590 nm using the SpectraMax^® i3× Multi-Mode Microplate Reader (Molecular Devices, LLC). Each sample was analyzed in triplicate. The optical density of the medium was read as background, and the value of the optical density of the control cells was taken as 100%. IC₅₀ values were calculated using SOFTmax^® Pro 7.2 GxP software (Agilent Technologies, Inc.).

On-Targetplus small interfering (si)RNA (Revvity Discovery Limited), a smart pool of three siRNAs [12.5 or 25 nM; cat. no. L-007948-00-0005; National Center for Biotechnology Information (NCBI) accession nos. NM_001146705.2, NM_001146706.2 and NM_004653.5] was used for silencing of KDM5D and On-Targetplus Non-Targeting control siRNA (12.5 or 25 nM; cat. no. D-001320-01-20), a smart pool of four siRNAs, was used as a negative control (Revvity Discovery Limited). For initial validation experiments, siRNA was tested at 12.5 and 25 nM. Based on these results, a final concentration of 25 nM was used for all subsequent functional experiments. siRNAs were transfected with Dharmafect transfection reagent (cat. no. T-2001-03; Revvity Discovery Limited) for 48 h according to the manufacturer's protocol. Following 48 h of incubation at 37°C, cells were harvested and used for further analysis.

To ectopically express KDM5D, UKF-NB-3^CDDP cells were transfected for 48 h with the GenEZ^TM ORF clone plasmid KDM5D pCMV-3Tag1a (OHu18895C; Genscript) designed for transcript variant 1 of KDM5D (8 and 16 ng/µl; NCBI accession number NM_001146705.1; GenScript Biotech Corporation) and pCMV6-AC-GFP, mammalian vector with C-terminal tGFP tag (20 nM; OriGene Technologies, Inc.) was used as a negative control. For initial validation experiments, siRNA was tested at 8 and 16 ng/µl. Based on these results, a final concentration of 16 ng/µl was used for all subsequent functional experiments. The KDM5D ORF clone plasmid or control pCMV6-AC-GFP was transfected using Dharmafect transfection reagent (T-2001-03; Revvity Discovery Limited). The transfected cells were selected by Gibco geneticin™ Selective Antibiotic (cat. no. G418 Sulfate; 50 mg/ml) (Thermo Fisher Scientific, Inc.) in a final concentration 400 µg/ml for 72 h at 37°C. Overexpression efficiency was determined by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and flow cytometry respectively.

The exact siRNA/shRNA sequences used in the present study are proprietary to Revvity Discovery Limited and are not publicly available. They can be obtained for manufacturer upon reasonable request (www.dharmacon.com).

mRNA from all NBL cell lines was isolated using the PureLink™ RNA Mini Kit (Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. The quality of the extracted mRNA was measured using the NanoDrop One spectrophotometer (260/280 nm ratio) (Thermo Fisher Scientific, Inc.). Complementary DNA was synthesized from 1.0 µg of mRNA using the Generi Biotech Reverse Transcription Kit according to the manufacturer's instructions (GENERI BIOTECH). RT-qPCR was performed using the gb Easy PCR Master Mix (cat. no. 3006; GENERI BIOTECH) according to the manufacturer's instructions, with custom primers (Generi custom oligo synthesis) and hydrolysis probes. The RT-qPCR assays are identified by Generi Biotech IDs. These assays can be ordered at ID (ID: ‘hKDM5B_Q1’ and ‘hKDM5D_Q1’) from GENERI BIOTECH. Expression levels of target genes and the internal control POLR2A (ID, hPOLR2A ‘hPOLR2A_Q1’), which is homogeneously and uniformly expressed in neuroblastoma cell lines (59), were analyzed by RT-qPCR on a QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific, Inc.). Each sample was analyzed in triplicate. The thermocycling conditions were: 95°C 3 min, 50 cycles of 95°C for 10 sec, 60°C 20 sec. The relative differences in gene expression were expressed as fold change and were obtained with the 2^−ΔΔCq method (Relative Expression Software Tool REST 2009 software, v2.0.11; QIAGEN) (60). The exact primer sequences used in the present study are proprietary to GENERI BIOTECH and are not publicly available. They can be obtained for manufacturer upon reasonable request (www.generi-biotech.com).

Cell cycle analysis was carried out using FxCycle^TM Violet Ready Flow^TM reagent (Thermo Fisher Scientific, Inc.). After treatment with CDDP and/or transfection, harvested neuroblastoma cells with 0.25% trypsin (Thermo Fisher Scientific, Inc.) were pelleted by centrifugation (300 × g) for 3 min at room temperature, resuspended in 100 µl of 3.6% paraformaldehyde (Biogen) and incubated at 20°C for 15 min. After this incubation, the suspension was then centrifuged (300 × g) for 3 min at room temperature, and the pellet was washed twice with phosphate-buffered saline (PBS; Thermo Fisher Scientific, Inc.). Permeabilization was performed with 90% methanol (PENTA) for 1 h at −20°C. Pellets were washed with PBS, resuspended in 500 µl of PBS and a drop of FxCycle^TM Violet Ready Flow^TM reagent was added. After 30 min of incubation at room temperature, cell cycle analysis was performed using a BD FACSCelesta (BD Biosciences) and data were analyzed using Flowlogic software, version 8 (Inivai Technologies).

After treatment with CDDP and/or transfection, harvested neuroblastoma cells (UKF-NB-3; UKF-NB-3^CDDP) were washed with cold PBS (Thermo Fisher Scientific, Inc.), trypsinized with 0.25% trypsin (Thermo Fisher Scientific, Inc.) and collected by centrifugation (300 × g) for 3 min at room temperature. Cell pellets were washed with PBS and fixed in 3.6% paraformaldehyde for 15 min at room temperature. The cell pellets were then washed with PBS and permeabilized with 90% methanol for 1 h at −20°C. The pellets were then washed three times with 0.5% bovine serum albumin (BSA; Roth) in PBS. After washing, the cell pellets were blocked with 5% BSA in PBS for 1 hour at room temperature After blocking, the pellets were incubated with the primary antibody anti-JARID1D rabbit mAB at a dilution of 1:100 (cat. no. PA5-100844; Invitrogen, Thermo Fisher Scientific, Inc.), anti-JARID1B rabbit mAB at dilution of 1:400 (cat. no. 15327S; Cell Signaling Technology, Inc.), anti-trimethyl histone H3 (Lys4) rabbit (cat. no. 07-473; MilliporeSigma) at a dilution of 1:400, Cleaved-Caspase 3 (Asp175) Rabbit mAB at a dilution of 1:50 (cat. no. 9602S; Cell Signaling Technology, Inc.) or CUL4A Rabbit mAB at a dilution of 1:50 (cat. no 2699; Cell Signaling Technology, Inc.) for 1 h at laboratory temperature. Cell pellets were then washed with 0.5% BSA (Roth) and incubated in fluorochrome-conjugated secondary antibody Anti-Rabbit IgG (H+L) Alexa Fluor^® 647 Conjugate (cat. no. A21245; Thermo Fisher Scientific, Inc.) diluted 1:500 and incubated for 30 min at room temperature in the dark. Cell pellets incubated with secondary antibody (1:500) only were used as a control. Washed and resuspended cells were measured using a BD FACSCelesta (BD Biosciences), and data were analyzed using Flowlogic software, version 8 (Inivai Technologies).

Real-time monitoring of cell migration of sensitive cells (UKF-NB-3) and their derived chemo-resistant cells (UKF-NB-3^CDDP) was carried out using the xCELLigence RTCA DP instrument (Agilent Technologies, Inc.) in a humidified incubator at 37°C and 5% CO₂. For sensitive cells, KDM5D silencing (siKDM5D) and for CDDP-resistant UKF-NB-3^CDDP ORD cDNA transfection were performed. Cells were serum-starved for 2 h in IMDM without FBS at 37°C and 5% CO₂. After the starvation period cells were trypsinized and seeded at a density of 1×10⁴ cells/well into upper chamber of the 16-well electronically integrated Boyden chamber for invasion/migration assays-CIM-plate 16 (RTCA, Agilent) containing starvation media. The wells of the lower chamber were loaded with IMDM with 10% FBS. As the cells migrated towards the chemoattractant across microelectronics sensors integrated at the bottom side of upper chamber, the impedance was measured every 30 min for 168 h at 37°C. The measurements were recorded and analyzed using Real-Time Cell Analysis Software 1.2 (Agilent Technologies, Inc.).

Real-time monitoring of cell proliferation was carried out using the xCELLigence RTCA DP instrument (Agilent Technologies, Inc.) in a humidified incubator at 37°C with 5% CO₂. Sensitive cells and their derived chemoresistant cells (UKF-NB-3, UKF-NB-3^CDDP, IMR-32, IMR-32^CDDP, SK-N-F1 or SK-N-F1^CDDP) were seeded at a concentration 8,000 cells per well in wells of 16-well E plates for impedance-based detection. For sensitive cells KDM5D silencing (siKDM5D) was carried out. After 48 h of transfection (siKDM5D), cells were seeded at a concentration 8,000 cells per well in wells of 16-well E plates for impedance-based detection. The cell index was monitored every 30 min for 96 h at 37°C and data were recorded using the xCELLigence RTCA software Pro version 2.8 (Agilent Technologies Inc.) provided.

The R2 Genomics Platform (https://r2.amc.nl) is an open-access online genomic analysis and visualization platform that is publicly available to analyze and interpret clinical and genomic data. Several neuroblastoma datasets are available for survival analysis. The present study used three different datasets: i) Tumor neuroblastoma from Kocak (649; custom; ID: Agilent 44K microarray, Kocak dataset, Wolf normalization-ag44kcwolf;), which contains gene expression profiles from 649 patient-derived neuroblastoma tumors (https://hgserver1.amc.nl/cgi-bin/r2/main.cgi R2 internal identifier: ps_avgpres_gse45547geo649_ag44kcwolf) (61); ii) tumor neuroblastoma from SEQC (498; RPM; ID: SEQC/MAQC-III neuroblastoma dataset - seqcnb1), which contains expression data from gene expression microarrays for 498 patient-derived neuroblastoma tumors (https://hgserver1.amc.nl/cgi-bin/r2/main.cgi R2 internal identifier: ps_avgpres_gse62564geo498_seqcnb1) (62) and iii) tumor neuroblastoma from Oberthuer (251; user-defined; ID: Oberthuer neuroblastoma gene-expression dataset - amexp255), which consists of gene expression profiles from 251 patient-derived neuroblastoma tumors (https://hgserver1.amc.nl/cgi-bin/r2/main.cgi R2 internal identifier: ps_avgpres_nb251_amexp255) (63), (Table SI). To analyze the prognostic significance of the KDM5D and CUL4A genes Kaplan-Meier curves were generated comparing overall survival of patients with low and high expression of KDM5D or CUL4A using the integrated plotting function with default settings on R2: Genomics Analysis and Visualization Platform (Department of Oncogenomics, Amsterdam University Medical Centers (AMC), University of Amsterdam; accessed in 2024). Median expression cutoff mode and Bon-Ferroni correction for multiple testing was performed for all analyses. Statistical significance of survival differences was determined using the log-rank (Mantel-Cox) test.

Proteins were extracted from cultured neuroblastoma cells. Extraction of proteins was conducted using the ReadyPrep™ Protein Extraction Kit (Bio-Rad Laboratories, Inc.), with the addition of a complete protease inhibitor cocktail (Roche Applied Science). The concentration of proteins was subsequently measured using a BCA Protein Assay kit (Thermo Fisher Scientific Inc.). Samples (40 µg) were resolved on 4–20% precast gradient SDS-polyacrylamide gels (Mini-PROTEAN^® TGX™, Bio-Rad Laboratories, Inc.), transferred onto PVDF membranes (Bio-Rad Laboratories, Inc.) and blotted on PVDF membranes (Bio-Rad Laboratories, Inc.). Membranes were blocked in 5% BSA in TBS containing 0.1% Tween-20 for 1 h at room temperature, followed by incubation with primary antibodies overnight at 4°C. The following primary antibodies were used: KDM5D Rabbit pAb (cat. no. PA5-100844; Thermo Fisher Scientific, Inc.), was diluted to a concentration of 1:500, while CUL4A Rabbit pAb (cat. no. 2699S; Cell Signaling Technology Inc.) was diluted to 1:1,000. The PARP Rabbit pAb (cat. no. 9542S; Cell Signaling Technology Inc.) and the Caspase-3 Rabbit pAb (cat. no. 9622S; Cell Signaling Technology Inc.) were both diluted to 1:1,000. β-tubulin Mouse mAb (cat. no. 86298S, Cell Signaling Technology Inc.) was diluted to 1:1,000 and used as a loading control. Secondary antibodies, AlexaFluor 488 Mouse, Rabbit (cat. nos. A-11008 and A28175; Thermo Fisher Scientific Inc.) were diluted 1:2,000 and the incubation conditions were 1 h at room temperature. The membranes were then subjected to visualization via the ChemiDoc MP imaging system (Bio-Rad Laboratories, Inc.). The analysis was conducted utilizing ImageJ 1.52a software (National Institutes of Health).

For the verification of sustained KDM5D silencing, UKF-NB-3 cells were transfected with ON-TARGETplus KDM5D siRNA as described above and lysed at multiple time points up to 7 days after transfection. KDM5D protein levels were analyzed by western blotting using the KDM5D antibody and β-tubulin as a loading control.

All experiments were repeated independently at least three times, and data are expressed as mean ± standard error of the mean. ANOVA with post hoc Tukey honestly significant difference was used to compare different groups. One-way ANOVA with Tukey's post hoc test was used when a single independent variable (such as treatment, transfection or time) was analyzed. Two-way ANOVA with Tukey's post hoc test was applied when two independent variables (such as cell line and treatment) were analyzed simultaneously. Differences were considered statistically significant when. The exact statistical tests used for each analysis are indicated in the respective figure legends.

Chemotherapy is one of the main treatment modalities for high-risk neuroblastoma, and CDDP is part of the majority of treatment protocols. The occurrence of chemoresistance is a considerable negative sign and contributes fundamentally to treatment failure. In this context, cell lines with intrinsic induced resistance to CDDP were analyzed. The IC₅₀ of the CDDP-resistant cell lines (UKF-NB-3^CDDP, IMR-32^CDDP and SK-N-F1^CDDP) was significantly higher compared with the drug-sensitive parental lines UKF-NB-3, IMR-32 and SK-N-F1 (Figs. 1A and S1). In this group of cell lines, the mRNA and protein levels of KDM5D were examined in relation to CDDP resistance and CDDP treatment. The RT-qPCR results showed a loss of KDM5D expression in all CDDP-resistant cell lines. In addition, CDDP treatment decreased the expression of KDM5D in sensitive cells and this decrease was dependent on the duration of incubation with CDDP. The present study observed that no expression was present in drug-resistant cell lines even after treatment with CDDP (Figs. 1B and S1). The same results were observed at the protein level, as determined by means of flow cytometry (Figs. 1C and S1, S2, S3, S4, S5, S6, S7). For further experiments, UKF-NB-3 and UKF-NB-3^CDDP cell lines were selected, as they were used previously to conduct a study with this cell line on the importance of KDM5B for CDDP-resistance (18).

H3K4me3 is frequently associated with transcriptional activation of neighboring genes and therefore could activate oncogenes (64,65). We observed the change in trimethylation of H3K4 after CDDP treatment, silencing or overexpression of KDM5D in neuroblastoma cell lines (Fig. S8). To demonstrate the importance of KDM5D for methylation, the present study monitored H3K4 trimethylation after KDM5D silencing and overexpression. siRNA against KDM5D in the cell line UKF-NB-3 decreased expression levels of KDM5D mRNA (measured by RT-PqCR; Fig. 2A) and protein (measured by flow cytometry; Figs. 2B and S9) compared with non-coding siRNA-transfected control cells. Treatment of UKF-NB-3 with CDDP reduced KDM5D expression in controls, but did not further reduce expression in siRNA-transfected cells (Figs. 2C, D and S10). In addition, trimethylation of histone H3K4 was significantly increased in the KDM5D knockdown cells compared with the UKF-NB-3 control cells as shown by flow cytometry results, whereas the treatment with CDDP increased trimethylation only in the transfected cells, but not in control (Figs. 2E, F and S11).

The CDDP-resistant cell line UKF-NB-3^CDDP, which does not express KDM5D, was transfected with the KDM5D ORF clone plasmid pCMV-3Tag1 carrying the KDM5D gene for 48 h, resulting in overexpression of KDM5D at the protein level compared with cells transfected with the control plasmid (Figs. 3A and S12). In addition, treatment with CDDP decreased the expression of KDM5D in transfected cells (Figs. 3B, C and S13). Trimethylation of histone H3K4 was significantly decreased in cells transfected with KDM5D compared with control, while treatment with CDDP increased H3K4 trimethylation in both control and transfected cells (Figs. 3D, E and S14).

To determine the effect of KDM5D expression on sensitivity to CDDP in neuroblastoma, the percentage of viable cells following treatment with CDDP at varying concentrations and time intervals was monitored. KDM5D knockdown did not affect the viability of cells in UKF-NB-3 after 24 h of transfection (Fig. S15) or after 48 h of transfection (Fig. 4A). The viability of KDM5D knockdown UKF-NB-3 cells decreased after incubation with CDDP depending on incubation time and concentration, but the decrease in viability of NC transfected cells after treatment with CDDP was more substantial (Figs. 4A and S16).

To compare the effect of inhibition of KDM5s demethylases on the viability and sensitivity of UKF-NB-3 and UKF-NB3^CDDP cells to CDDP, KDOAM-25 citrate was used as an inhibitor of the KDM5 demethylases subfamily (66), and therefore a concentration that effectively reduced H3K4 trimethylation without influencing cell survival (400 nM) was selected (data not shown). Inhibition of KDM5 by KDOAM-25 protected both sensitive (UKF-NB-3) and resistant (UKF-NB-3^CDDP) cells from the effects of CDDP (Fig. S16).

Overexpression of KDM5D did not alter cell viability in UKF-NB-3^CDDP after 24 h, even after incubation with CDDP (Fig. S17). However, after 48 h of incubation with CDDP, cell viability decreased with increasing CDDP concentration compared with mock-transfected controls, whose viability did not change after treatment with CDDP (Fig. 4B).

The decreased sensitivity to CDDP after incubation with the inhibitor was consistent with the downregulation of KDM5D by siRNA, and conversely, the increase in sensitivity to CDDP in the resistant line with induced KDM5D expression suggests the importance of KDM5D in the anticancer efficacy of CDDP.

To determine whether overexpression of KDM5D in resistant cells and CDDP treatment induces apoptosis, activated caspase-3 and cleavage of PARP was examined to determine the relationship between the expression of KDM5D and CUL4A. Control cells had increased levels of activated caspase-3 after treatment with CDDP. Silencing of KDM5D in the sensitive cell line UKF-NB-3 did not change the amount of activated caspase-3 compared with control cells. Treatment with CDDP increased activated caspase-3, but the level of cleaved caspase-3 was significantly lower in cells with KDM5D knockdown compared with controls after CDDP treatment (Figs. 5A, S18 and S19). The KDM5 pan-inhibitor KDOAM-25 citrate had a similar effect as KDM5D silencing (Fig. S20, Fig. S21, Fig. S22). Overexpression of KDM5D resulted in an increase in activated caspase-3 level after treatment with CDDP in contrast with control cells (Figs. 5B and S23). Taken together, KDM5D is involved in the sensitivity to CDDP induced apoptosis in neuroblastoma cells.

Because inhibition of KDM5D protects cells from the effect of CDDP (Figs. 4A, 5A, S15 and S16) and artificial expression, on the contrary, increases the effect of this cytostatic agent (Figs. 4B, 5B and S17), the present study focused on the significance of KDM5D for cell cycle and proliferation. In the sensitive cells, silencing of KDM5D resulted in an increase of cells in S phase and a decrease in G2/M phase compared with the control (Figs. 6A, S24 and C). Transfection of the plasmid with KDM5D in the cell line UKF-NB-3^CDDP decreased the percentage of cells in S phase and increased G0/G1 phase compared with control (Figs. 6B, S25A and E). Treatment with CDDP resulted in cell cycle arrest at checkpoint G0/G1 and G2/M in the sensitive cell line, and therefore the cell cycle was not evaluable (Fig. S24B, D, F).

The xCELLigence system was used to monitor cell migration (Figs. 6C and S26A) and proliferation (Figs. 6D and S26B-D) in real time, which showed that the CDDP-resistant UKF-NB-3^CDDP had a higher cell index of migration and proliferation compared with the CDDP-sensitive UKF-NB-3 with KDM5D expression. In addition, KDM5D knock down increased cell proliferation even in IMR-32 and SK-N-F1 lines (Fig. S26C-D). The silencing of KDM5D in UKF-NB-3 increased cell migration and proliferation, while KDM5D overexpression in UKF-NB-3^CDDP cells decreased migration and proliferation compared with controls. KDM5D silencing was verified over a period of 7 days using western blotting (Fig. S26E).

To assess the relationship between CUL4A and KDM5D (Fig. S18), its expression in CDDP-sensitive and resistant cells was examined. The CDDP-sensitive cell line UKF-NB-3 showed reduced expression of CUL4A protein compared with the CDDP-resistant UKF-NB-3^CDDP, and the expression of CUL4A increased after treatment with CDDP in the sensitive but not in the resistant cells (Figs. 7A, B S27 and S28). In addition, knockdown of KDM5D increased CUL4A expression Figs. 7B and S27). In UKF-NB-3^CDDP cells, overexpression of KDM5D decreased CUL4A expression compared with control, while CDDP treatment of cells with overexpressed KDM5D increased CUL4A expression levels. However, in the control, CDDP treatment did not alter CUL4A expression in resistant cells (Figs. 7C and S28). Inhibition of all KDM5s members by KDOAM-25 in sensitive cells increased level of CUL4A, compared with control but did not change CUL4A expression levels in chemoresistant cells that do not express KDM5D (Figs. S11 and S29, S30, S31).

To investigate the role of KDM5D and CUL4A in neuroblastoma, patient data was screened from the R2 database: Genomics Analysis and Visualization Platform (http://r2.amc.nl). The association between KDM5D expression and overall survival was compared in male patients with two different neuroblastoma risk groups according to International Neuroblastoma Staging System (INSS) stages: non-high-risk (st1, st2, st3 and st4S) and high-risk (st4) (67). This analysis showed that low KDM5D expression was associated with poor survival rates in the group with st4 (Figs. 8B and S32). Analysis of three different datasets of patients with neuroblastoma from the R2 platform indicated a statistically significant correlation between KDM5D expression and higher survival rates in high-risk neuroblastoma patients. In the non-high-risk group, there was not a significant difference in all data sets (Figs. 8A and S32). These observations are consistent with our hypothesis that low KDM5D concentration in neuroblastoma has tumor-promoting effects and is associated with chemoresistance.

To uncover the relationship between the expression of KDM5D and CUL4A, the present study further analyzed the data from the aforementioned databases in R2: for an association between the expression of CUL4A and overall survival of neuroblastoma patients. Kaplan-Meier analysis in these datasets revealed that high relative expression of CUL4A is often associated with worse survival in two datasets of patients with stages 1, 2, 3 or 4S (Figs. 8D and S33); moreover, this association is significant in stage 4 (Figs. 8E and S32), however, this was not true in the Oberthuer dataset.