Gene Expression and Early Radiation Response of Two Distinct Neuroblastoma Cell Lines

Introduction: Neuroblastoma is one of the most common childhood cancers with one of the lowest survival rates, accounting for 15% of childhood cancer mortality. Approximately half of children treated for high-risk neuroblastoma will relapse following remission, while another 15% of patients do not respond to initial treatment. External beam radiation is infrequently used for treatment of pediatric cancer such as neuroblastoma, typically reserved for palliative care in patients with aggressive metastatic disease who fail to respond to alternative treatments. Understanding effects of radiation on neuroblastoma cells could improve efficacy of this final means of therapy to decrease tumor burden and stabilize the disease. Methods: In this study, we found that two microRNAs with opposite functions were expressed in two neuroblastoma cell lines with marked differences in radiosensitivity. Clonogenic assays were used to evaluate the radiation responses for these 2 cell lines, designated SK-N-AS and SK-N-DZ; cells were then irradiated at doses that cause 90% cell killing based on clonogenic assay and their RNA isolated and subjected to microarray analysis. In addition, cells were transfected with pre-miRNA constructs that led to overexpression of microRNAs miR-34a and miR-1228 to determine possible microRNA regulation of radiation response. Results: Statistically significant differences were detected for expression of several thousand genes when the 2 cell lines were compared with each other. In comparison, radiation exposure resulted in only minor gene expression differences of less than 2-fold at the 1 h postirradiation timepoint in both cell lines. Overexpression of miR-34a and miR-1228 in either cell line did not alter this outcome. Discussion: While these two neuroblastoma cell lines are phenotypically diverse and gene expression differences between them are extensive, we observed that the regulation of gene expression in both cell lines is in a stable equilibrium at early timepoints after exposure to ionizing radiation.

into low-or high-risk categories based on MYCN status, patient age, tumor histopathology, and other genetic signatures or chromosome structural changes such as tumor ploidy, 1p, and 11q status [2][3][4]. Low-risk patients are treated by gross complete surgical resection or subjected to observation without surgery, resulting in 90-95% 5-year overall survival rate [5]. However, nearly 70% of patients present with advanced stage disease at time of diagnosis, with a large percentage of these patients facing therapy resistance (15%) or disease relapse (40-60%) [4][5][6]. Current treatments for high-risk patients include chemotherapy, radiation, surgery, bone marrow stem cell transplant, and immunotherapy; despite such an aggressive regimen, high-risk patients demonstrate approximately 50-60% overall survival at 5 years [7]. Radiation is a significant component of high-risk characterization and treatment; radiolabeled metaiodobenzylguanidine is used for imaging and therapy in NBL [8,9], while external-beam radiation is used to directly target tumor cells, or as part of a treatment regimen prior to stem cell transplant [10]. While effective at treating the tumors, especially in reducing risk of local recurrence, radiation therapy is harsh on young patients and relapses are frequently radiation resistant [11].
Heterogeneity of NBL as a disease is echoed by heterogeneities between cell lines derived from individual NBL cancers [12,13]. Such variety is reflected in features such as disparate levels of gene amplification or deletion, morphologic diversity, and tumorigenicity; this led to NBL cell lines being grouped into two major subtypes, neuroblastic (N-type) and substrate-adherent (S-type), with a third intermediate (I-type) grouping bridging them [14,15]. Our study is focused on cell lines SK-N-AS (S-type) and SK-N-DZ (N-type), both isolated from bone marrow metastatic sites but showing diverse genetic backgrounds and disparate responses to potential cancer treatments, including ionizing radiation (online suppl. Table S1; for all online suppl. material, see https://doi.org/ 10.1159/000530902). Over the years, these 2 cell lines were sometimes used as a source for investigation of commonalities between different NBLs, with regard to their proteomic profiles and secretion [16], cellular signaling [17], and response to therapy [18,19]; and sometimes as models for differences between different NBL types.
Different MYCN amplification status in these cells led to their use in studies focused on MYC-related treatments [20], with SK-N-AS having only a single copy while SK-N-DZ carries amplified MYCN. MYCN and its related oncoprotein c-MYC are transcription factors that can lead to increased cell proliferation and deregulated growth when overexpressed. MYCN amplification (>10 copies per diploid cell) is detected in 16-25% of all NBL cases [21], and 40-50% of high-risk cases [22]. High-risk NBL tumors show few recurrent somatic mutations, except the pathognomonic MYCN amplification, making subgroup stratification and development of targeted therapies difficult [23]. Recent focus has turned to correlative effects, demonstrating chromosomal alterations such as 1p36 deletion that act in conjunction with MYCN amplification to circumvent cell death [24]. As researchers seek to find new therapeutic targets, traditional gene alterations are set aside in favor of novel targets such as microRNAs (miRs) in cancer genomics. As small, noncoding RNAs, miRs can bind multiple diverse genes and suppress expression; leveraging this functionality as a means of therapy enables regulating of a larger panel of genes responsible for tumor progression and survival.
In this work, we examined early gene expression changes in SK-N-AS and SK-N-DZ cells in response to radiation exposure as a measure of cell-specific "strategies" to recover from insult. Ionizing radiation exposure has different effects on cells in different stages of the cell cycle, which causes different responses on the cell lines due to inherent differences in cell cycle arrest. Viability of these cells after exposure to ionizing radiation differs significantly, as previously demonstrated in preliminary work by MTS assay using a cesium-137 (Cs-137) gamma irradiator on SK-N-AS and SK-N-DZ cells [25]. With dramatic differences in radiation response, gene expression, and pathophysiologies between the cell lines, we expected to see marked differences between SK-N-AS and SK-N-DZ cells potentially even at very short time periods. In addition, we explored possible modulation of radiation response in the 2 cell lines associated with overexpression of hsa-miR-34a-5p (miR-34a) and hsa-miR-1228-3p (miR-1228) by transfection of synthetic precursor micro-RNAs (pre-miRNAs). Previous research identified miR-34a as a tumor suppressor in several cancers including NBL; located on chromosome 1p36, a region commonly deleted in NBL, miR-34a loss is correlated with MYCN amplification in high-risk NBL and contributes to therapy resistance [24,[26][27][28]. Additionally, miR-1228 may act as a putative oncogene in cancers including NBL. Located at chr12q13.3, a chromosomal region amplified in several cancers including NBL [29][30][31][32], prior research has shown that miR-1228 may play a role in disease progression by suppressing cellular apoptosis and promoting proliferation and metastasis through targeting genes including p53 [28]. Current research into miR-1228 functionality is limited and its effects following irradiation have not been explored. These two miRNAs are especially intriguing because of their opposite patterns of baseline expression in untreated SK-N-AS and SK-N-DZ cells, implying possible involvement in cell-specific repair following radiation exposure.

Materials and Methods
Cell Culture, Transfections, and Irradiation Treatments Human NBL cell lines SK-N-AS and SK-N-DZ (catalog numbers CRL-2137 and CRL-2149, respectively, ATCC, Manassas, VA) were maintained in high-glucose Dulbecco's modified Eagle medium (Corning, Glendale, AZ) supplemented with 10% fetal bovine serum (Gibco TM , Waltham, MA). All cells were cultured in the presence of 1x antibiotic antimycotic solution (Corning, Glendale, AZ) and 1x Essential amino acid solution (Corning, Glendale, AZ), at 5% CO 2 at 37°C. For all transfection experiments, 6 × 10 5 cells per well were seeded in 6-well plates 1 day prior to transfection. SK-N-DZ cells were plated in wells coated with Geltrex (Invitrogen, Waltham, MA, USA) before use. A schematic of experimental treatments, cell collection, and assays performed is presented in online supplemental Figure S1.
For clonogenic assays (online suppl. Fig. S1A), cells were cultured to 75% confluence in T25 flasks before being irradiated with 160 kVp X-rays at a dose rate of 3.12 Gy/min (RS200 Irradiator, RadSource, Buford, GA) with doses of 1, 2, 4, 6, or 7 Gy. Cells were provided with fresh medium immediately prior to irradiation and were trypsinized 15 min postirradiation to be seeded at different cell densities in 12-, 24-, and 48-well plates. Three to 4 weeks after plating, media were removed, and cell colonies were fixed and stained overnight with 0.001% crystal violet dissolved in 10% neutral buffered formalin. The following day, plates were washed and dried prior to colony counting. This approach avoided extensive washes prior to fixation that could remove loosely attached SK-N-DZ cell colonies. Colonies with more than 50 cells were counted. Colonies that grew in six wells of a multi-well plate (technical replicates) were counted for each separate experiment (biological replicate). The proportion of untreated and nonirradiated cells that successfully grew into colonies varied between experiments. This cell number was considered as 100% growth, and colony growth numbers for all treated cell samples were presented as percentages of this value.
For clonogenic assays with miRNA overexpression, cells were cultured in six-well plates and left either non-transfected or transfected with pre-miRNA molecules for 20-24 h prior to being irradiated with the same X-ray source, with doses of 4 Gy or 6 Gy. Again, cells were harvested by trypsinization and seeded at different cell densities; after several weeks of growth in the incubator, colonies were stained, washed, dried, and counted as described above.
For RNA and protein isolation, cells were irradiated 20 h after transfection (online suppl. Fig. S1B). For each experiment, transfections were done simultaneously with the same cell density and incubation times. Doses used for RNA and protein isolation from SK-N-AS cells were 4 Gy and 6 Gy, and for SK-N-DZ cells 2 Gy and 4 Gy. Based on clonogenic assay data, the radiation doses chosen gave roughly equivalent cell survival rates between the 2 cell lines. Cells were incubated 1 h (online suppl. Fig. S1B a-c) or 24 h (online suppl. Fig. S1B.d,e) after irradiation prior to RNA isolation; cells used for preparation of nuclear extracts were harvested 24 h (online suppl. Fig. S1B.f) after irradiation. In each case, experiments were done as biological triplicates for each transfection, irradiation dose, and post-irradiation timepoint.

RNA Isolation for Quantitative Real-Time Polymerase Chain Reaction
Gene and microRNA expression levels were analyzed in SK-N-AS and SK-N-DZ cell lines, with and without radiation exposure. Total RNA and miRNA fractions were isolated with the mirVana TM miRNA Isolation kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's protocol. miRNA overexpression was induced by transfecting SK-N-AS and SK-N-DZ cells with control miR, miR-34a, or miR-1228 constructs 24 h prior to radiation treatment. Nonirradiated cultures were used as baseline controls; SK-N-AS cells were irradiated with 4 Gy and 6 Gy, while SK-N-DZ were irradiated with 2 Gy and 4 Gy, consistent with our other experiments. To confirm miRNA overexpression, cDNA was generated using specific primers for miR-34a, miR-1228, and endogenous control U6 snRNA (Thermo Fisher Scientific) and TaqMan MicroRNA Reverse Transcription kit (#4366496, Applied Biosystems). Overexpression of selected miR-NA was confirmed using TaqMan assay for corresponding miR-NAs (Applied Biosystems) 24 h after transfection. Relative miRNA expression level was determined using a control, nonirradiated sample ( Fig. 2). Although the pre-miR concentration used was the same for each cell line, a range of overexpression was detected, which may be correlated with factors that regulate different endogenous expression of these miRNAs in the 2 cell lines.
Quantitative real-time polymerase chain reaction (qPCR) for target genes of interest was performed with the Fast SYBR Green Master Mix (Applied Biosystems) using the Applied Biosystems thermal cycler Model 7300. Expression of mRNA was normalized against GAPDH expression as before [33], and relative quantification was calculated using the ΔΔCt method.
Results are presented as fold increase in the test samples compared with the sample transfected with control pre-miRNA. For qPCR, all samples were analyzed in three technical and biological replicates. The sequences of oligonucleotides used as qPCR primers are listed in online supplemental Table S2.
Gene Expression Array Analysis cDNA for gene expression analysis was generated from 1 μg of DNase-treated total RNA using SuperScript ® III First-Strand Synthesis SuperMix (Thermo Fisher Scientific). RNA samples were diluted with DEPC-treated water to a final concentration of 50 ng/μL and submitted to Northwestern University's NUSeq Core Facility for gene expression analysis.
The Human Clariom™ D GeneChip (Applied Biosystems, Thermo Fisher Scientific), which allows analysis of the expression of over 540,000 transcripts was used for the microarray study. The GeneChip WT PLUS Reagent Kit was used for RNA sample preparation for array hybridization. For each sample, 50 ng total RNA was used for cRNA target preparation, array hybridization, washing, staining, and image scanning. The washing and staining steps were performed on a GeneChip Fluidics Station 450 and the scanning of hybridized arrays was conducted on a GeneChip Scanner 3000 7G. After generation, the array data were first examined for quality using QC criteria set by Affymetrix. All hybridizations were done in triplicate for biological replicates for each transfection, irradiation dose, and postirradiation timepoint. Sample and hybridization quality controls met the criteria for this array type.
Probeset data from the raw CEL files were summarized using the "affy" package (v. 1.68) [34] of Bioconductor (v. 3.12) [35] in R (v. 4.0.4) [36]. Genes were annotated with a custom CDF (v. 25) from the Molecular and Behavioral Neuroscience Institute (Brainarray) at the University of Michigan [37]. Signals were then standardized across arrays using the supervised normalization of microarrays method (v. 1.38) [38] in R/Bioconductor. The supervised normalization of microarrays model was fit with three biological variables: cell line, radiation dose level, and miRNA treatment. Variance due to batch processing was removed by modeling the microarray scan date as the adjustment variable and setting the "Rm = True" option.
Differential expression analysis was performed using the R package Limma (v. 3.46) [39]. Normalized intensities were fit to a linear model with coefficients for each of the 16 factor combinations of cell line (SK-N-AS or SK-N-DZ), radiation dose (0 Gy or high dose), and miRNA treatment (miR-34a, miR-1228, control miRNA, or no miRNA). All pairwise comparisons of interest were extracted from this model as contrasts using empirical Bayes smoothing [40]. The false discovery rate was controlled using the Benjamini and Hochberg correction. Probes with adjusted p value <0.05 and fold-change (FC) greater than 1.5 (that is, |log2FC| > 0.58) were judged to be differentially expressed.
For cellular pathway analyses, the gene expression lists were ordered based on differential gene expression in SK-N-AS versus SK-N-DZ cells or vice versa under different conditions, and ENSEMBL codes for genes were submitted to g-profiler (https:// biit.cs.ut.ee/gprofiler_beta/gost) [41]. Query conditions were as follows: ordered gene list, TRUE; sources of pathways GO:MF, GO: CC, GO:BP, KEGG, REAC, TF, MIRNA, HPA, CORUM, HP, WP; significance threshold method g_SCS, threshold 0.05. Versions of the g-profiler used work with Ensembl 106 and Ensembl Genomes 53 as sources of ENSEMBL codes.

Isolation of Nuclear Extracts and ELISA Assays for c-Myc and NF-kB
Content of c-Myc and NF-kB in nuclear extracts was detected using TransAM™ (Active Motif, Carlsbad, CA, USA) ELISAbased kits. Nuclear extracts were isolated from non-transfected and transfected, irradiated and nonirradiated SK-N-AS and SK-N-DZ cells using Nuclear Extract Kit (Active motif). Transfections were performed as described previously, cells were irradiated 24 h after transfection and extracts isolated after a 24-h postirradiation incubation. Ten to 17 µg per well were used for nuclear protein extract ELISAs for detection of NF-kB proteins p65 and c-Rel, and c-Myc. Detection was performed following the manufacturer's instructions using a ClarioStar plate reader (BMG Labtech, Cary, NC 27513). All experiments were done in triplicate, with cell nuclear extracts isolated from three independently performed transfection-irradiation experiments.

Clonogenic Assays Show Marked Differences in Radiation Response between NBL Cell Lines SK-N-AS and SK-N-DZ
Neuroblastic SK-N-DZ cells [12] have low adhesion, which hampers manipulation required for irradiation of already plated cell culture dishes with single cells. In previous ionizing radiation experiments using Cs-137 gamma rays, we have used MTS assays in order to circumvent this problem [25]. In this work, we utilized a different approach by irradiating 70-80% confluent cell culture dishes, followed by trypsinization and seeding after a 15-min postirradiation incubation (online suppl .   Fig. S1A). The same approach was used for SK-N-AS cells, which is more substrate adherent. Single cells generated by trypsinization of irradiated or sham irradiated cells at 70% confluence were permitted to attach to the plate and grow into colonies over a period of 3-6 weeks depending on the cell line. Figure 3 shows the survival curves for SK-N-AS and SK-N-DZ cell lines, demonstrating that a dose of 6 Gy causes 90% cell killing of SK-N-AS cells while equivalent cell killing of SK-N-DZ requires only 4 Gy.

NBL Cell Lines SK-N-AS and SK-N-DZ Show Opposite Pattern of Expression for microRNAs miR-1228 and miR-34a
A preliminary gene expression evaluation of cell lines SK-N-AS and SK-N-DZ using Cs-137 gamma rays (not published) showed numerous mRNA expression differences as well as opposite expression patterns for micro-RNAs miR-1228 and miR-34a. We repeated evaluation of expression for the two miRs in the 2 cell lines by qPCR ( Fig. 1) and found robust and reproducible expression of miR-1228 in SK-N-AS cells while expression of miR-34a was several fold lower in the same cell line. MiR-1228 is one of the handful of miRs known to be increased in response to ionizing radiation in human embryonic stem cell line H1 [42], and is a so-called mirtrona microRNA released from an intron of low density lipoprotein receptor-related protein 1 precursor [43].
Conversely, in SK-N-DZ cells (Fig. 1), relative quantification obtained by qPCR showed a several fold higher expression of miR-34a compared to miR-1228. In previous NBL studies, expression of miR-34a was associated with a good prognosis [44,45]. In NBL cell lines in vitro, overexpression of miR-34a was found to be cytotoxic [26,46].
To evaluate whether the presence of these two miRNAs at the time of irradiation would change cell survival or response to radiation, we transfected both cell lines with miR-34a and miR-1228 (online suppl. Fig. S1B) prior to irradiation and additional studies. Overexpression of the two miRNAs was robust in both cell lines in all subsequent experiments (Fig. 2).
Clonogenic Assays Show No Cell Survival Differences between Native and microRNA Transfected NBL Cell Line SK-N-AS SK-N-AS cells were transfected with pre-miRNAs for mature miR-1228 and miR-34a as described in the Methods section. Successful transfection was monitored by fluorescence readings of the cells used for clonogenic assay, and by qPCR of the cells transfected by pre-miRs in parallel, simultaneously, and with the same transfection mixtures as cells used for clonogenic assay. Twenty-four hours after transfection, cells were irradiated and 15 min later trypsinized and seeded as single cells to grow into colonies. Four biological replicates of each experiment were done; no effects of miRs of interest were noted (online suppl. Table S3). Across different experimental conditions, the percentage of cells growing into colonies after 4 Gy exposure varied between 0.05 and 4.03% for non-transfected cells, 0.02 and 3.8% for control miRtransfected cells; 0.15-3.56% for cells transfected with miR-1228, and 0.14-4.28% for cells transfected with miR-34a. Similarly, the percentage of cells growing into colonies after 6 Gy exposure was 0.11-1.81% for non-transfected cells, 0.16-1.63% for control miR-transfected cells; 0.13-1.73% for cells transfected with miR-1228; and 0.18-1.94% for cells transfected with miR-34a. It should be noted that these numbers are lower than what was observed for SK-N-AS cells in previous clonogenic experiment series. This may be the result of the fact that the cells in this experiment have undergone two trypsinization steps within a period of 48 h and their capacity to reattach to the plate may have been diminished. Regardless, the data are fairly consistent for each radiation dose irrespective of the pre-miR treatment, suggesting that the overexpression of miR-1228 or miR-34a has no significant effect on viability of SK-N-AS cells in combination with 4 Gy or 6 Gy radiation exposure.

Clariom™ D Assay Gene Expression Study of NBL Cell
Lines SK-N-AS and SK-N-DZ Gene expression differences between different NBL cell lines are well documented [13,47]; however, gene expression comparisons between cell lines SK-N-AS and SK-N-DZ were not done until this work. In this study, cells were irradiated with X-ray doses that cause a comparatively similar degree of cell death: 6 Gy for SK-N-AS cells and 4 Gy for SK-N-DZ cells. For each cell line, we also transfected the cells to overexpress microRNAs miR-1228 and miR-34a for 24 h prior to X-ray radiation exposure (online suppl. Fig. S1B). Three biological replicates for each treatment were done and the cells were harvested for RNA isolation for microarray work 1 h after radiation exposure. Gene expression was evaluated with Clariom D gene expression array; the data have been deposited into the GEO database (GSE197124). Gene expression normalization and differential expression were evaluated as described in the Methods section. Interestingly, despite the fact that the dose of radiation used with each cell line was cytotoxic, resulting in no more than 10% of the cells growing into colonies, at the 1 h postirradiation timepoint gene expression differences between irradiated and nonirradiated cells were almost nonexistent. For example, no differentially expressed genes were found when comparing the 0 Gy and 6 Gy conditions in either non-transfected or control miRtransfected SK-N-AS cells. In SK-N-AS cells transfected to overexpress miR-1228, only two differentially expressed genes with FC above 1.5-fold threshold were found between the irradiated and nonirradiated conditions: the-C-C motif chemokine ligand 2 (CCL2, EN-SEMBL ID ENSG00000108691) and the U5B small nuclear RNA 1 (RNU5B-1, ENSEMBL ID ENSG00000200156). In SK-N-AS cells transfected to overexpress miR-34a, a single gene, NF-kB inhibitor alpha (NFKBIA, ENSEMBL ID ENSG00000100906), was found to be overexpressed between irradiated and non-irradiated conditions. In SK-N-DZ cells, irradiation with 4 Gy X-rays did not cause any differential gene expression compared with the nonirradiated cells; this was true in all miRNA-treated and untreated cells.
Conversely, statistically significant (p < 0.01) gene expression differences between the 2 cell lines were marked at the 1.5 FC threshold (Fig. 4), with more than 10% or 4,000 differentially expressed genes out of the total 38,833 genes. Figure 4a depicts an Euler plot of the number of differentially expressed genes between the SK-N-AS and SK-N-DZ cell lines in each of the miRNA conditions, without radiation. Figure 4b shows that similar ratios of differentially expressed genes between the cell lines are evident 1 h after irradiation as well.
The complete gene expression datasets are available in GEO (GSE197124); however, four tables with genes affected by radiation are provided as supplemental data (online suppl. Tables S4-S6). Each table shows the genes that are either repressed or induced by radiation and differently expressed in the 2 cell lines, for each of the four experimental conditions: no-transfection, control miR transfection, or transfection with pre-miRs for miR-34a and miR-1228. Next, ordered lists of ENSEMBL gene IDs were submitted to g-profiler and used to generate lists of GO Molecular Function (GO: MF), GO Biological Processes (GO:BP), and the lists of microRNAs likely to be involved in gene expression regulation of these pathways (Graphical online suppl. Tables S1-S4).
More specifically, comparison of gene expression in nontransfected SK-N-AS cells versus SK-N-DZ cells generated a list of genes that are not modulated by radiation (not shown) and four lists of genes either suppressed or induced by radiation in one of the cell lines compared with the other (online suppl. Table S4). In this table, "AS versus DZ w/o IR" corresponds to genes that are suppressed by radiation and that are upregulated in SK-N-AS cells compared to SK-N-DZ cells in the absence of radiation. "DZ versus AS w/o IR" is a list of genes that are suppressed by radiation and that are upregulated in SK-N-DZ cells compared to SK-N-AS cells in the absence of radiation. "AS versus DZ IR" corresponds to genes that are induced by radiation and upregulated in SK-N-AS cells compared to SK-N-DZ cells at 1 h after radiation. "DZ versus AS IR" corresponds to genes that are induced by radiation and upregulated in SK-N-DZ cells compared to SK-N-AS cells at 1 h after radiation. ENSEMBL designations for the differentially expressed genes in online supplementary Table S4 were used in g-profiler to generate Graphical online supplementary Table S1.  Table S4, with ENSEMBL designations for the differentially expressed genes from online supplementary Table S5 used in g-profiler to generate  Graphical online supplementary Table S2. Likewise, gene expression comparisons between SK-N-AS cells versus SK-N-DZ cells transfected with pre-miR-34a are compiled in online supplementary Table S6, with  ENSEMBL designations for online supplementary  Table S6 used to generate Graphical Table S3. Finally, SK-N-AS cells and SK-N-DZ cells transfected with pre-miR-1228 were used to generate lists of genes either suppressed or induced by radiation in one of the cell lines compared with the other, presented in online supplementary Table S7 with ENSEMBL designations for the differentially expressed genes used in g-profiler to generate Graphical online supplementary Table S4.

Selected Gene and Protein Expression Differences
A subset of genes were selected from the microarray gene expression analyses to confirm our findings. Cells were untransfected or transfected with control miR or pre-miRNAs that lead to overexpression of miR-34a or miR-1228 for 24 h, then irradiated and total mRNA isolated at 1 h (online suppl. Fig. S1B.b) or 24 h (online suppl. Fig. S1B.e) after irradiation and analyzed by qPCR (Fig. 5).
A small selection of proteins was also examined by subsequent ELISA analyses in both cell lines at 24 h postirradiation (online suppl. Fig. S1B.f). In SK-N-AS cells, transfection with pre-miRNA for miR-34a led to about a 1.5-fold increase in expression of inhibitor of NF-kB alpha (NFKBIA). Nuclear concentration of some of the NF-kB proteins may thereby be impacted, and we chose p65 and c-Rel for further investigation. While p65 is one of the NF-kB proteins dependent on inhibitor of kB (IKB) expression, c-Rel is not, and we anticipated that the differences in the p65 versus c-Rel expression patterns would confirm the role of IKB in modulation of the radiation response in SK-N-AS cells. However, levels of expression of both proteins were low and no significant trend in protein quantities could be detected (online suppl. Fig. S2 and S3). Finally, we also conducted ELISA evaluation of c-Myc protein in the 2 cell lines (online suppl. Fig. S4). As expected, SK-N-AS cells (red) that do not overexpress MYCN showed higher expression of c-Myc protein. In the SK-N-DZ cell line (black), which has MYCN gene amplification, expression of c-Myc was negligible.

Discussion
Differences between the N-type and S-type NBL cell lines SK-N-DZ and SK-N-AS, respectively, are very pronounced and their responses to radiation exposure are equally heterogeneous. Studies using Chip-Seq to investigate transcription factors regulating gene expression in different NBL cell lines, including the two used here, identified 37 transcription factors with distinctive cellular levels in these cells. With such marked disparities between the cell lines, including baseline expression of miR-34a and miR-1228, we anticipated rapid differences to arise between them after irradiation that could potentially be leveraged into clinical applications and increase therapy efficacy.
In this work, we exposed the 2 cell lines to doses of radiation which caused 90% cell killing for each cell line, mimicking typical doses used in a clinical setting following a dose-fractionation approach [48,49]. In most other cancer cell lines, analogous radiation exposures lead to marked gene expression differences. In our examination of short-term effects of radiation exposure, we found only subtle differences in gene expression between nonirradiated and irradiated cells of the same type, regardless of miR-1228 or miR-34a overexpression in either cell line. At the same time, gene expression differences between the 2 cell lines included many hundreds of genes (Fig. 4). When we specifically selected radiation-associated genes from this list (online suppl. Tables S4-S7) and used them to identify GO Molecular Function and GO Biological Processes (Graphical online suppl. Tables S1-S4), none of them were found to be repeatedly associated with radiation exposure, once again suggesting that radiation exposures that cause 90% cell killing have no profound effect on early gene expression.
Minimal changes in gene expression patterns of NBL cell lines exposed to different growth and treatment conditions have been observed in the past as well. For example, SK-N-DZ cells grown as neurospheres show altered gene expression for some 11% of the registered genes only if the fold changes criterion is set to "anything above 1" [47]. In this work, not a single gene was found to be modulated above 1.5-fold in SK-N-DZ cells in the first hour after radiation. In the SK-N-AS cell line, non-transfected or controltransfected cells showed no changes in gene expression 1 h after exposure to 6 Gy. The only changes found in SK-N-AS cells were noted in cells transfected with miR-34a, which showed an increase in NFKBIA, while SK-N-AS cells overexpressing miR-1228 showed higher expression of U5B small nuclear RNA 1 (RNU5B-1) and the C-C motif chemokine ligand 2 (CCL2). Transfections of SK-N-AS cells with either pre-miRNA did not lead to any significant change in cell survival. At the same time, gene expression differences between the 2 cell lines included several thousands of genes (Fig. 4), about 1% of all the genes expressed in each cell type. Taken together, these findings suggest that gene expression regulation in NBL cell lines depends on stably engaged gene regulation patterns as suggested by genomic DNAenhancer studies [50]. It is probable that the gene expression patterns such as these found in two different types of NBL cell lines are not easily altered in response to radiation-induced stress. What the longterm effects of radiation stress may be will require evaluation of gene expression arrays for postirradiation timepoints beyond 1 h.

Author Contributions
C.R. participated in project conceptualization, investigation, and data analysis; writing of the original draft, editing, and submission. J.P. contributed to project conceptualization, investigation, and manuscript writing and editing. A.A. aided in experiments and conducted formal data analysis. Y.L. assisted in experiments and data collection. T.P. helped develop project concept and investigation, and was a major contributor to manuscript writing and editing. R.P. established methodology and assisted in funding acquisition. G.W. aided in project conceptualization and overall supervision, acquisition of funding, and writing and submission of manuscript.

Data Availability Statement
The microarray dataset presented in this study is openly available in NCBI's Gene Expression Omnibus and is accessible through GEO Series accession number GSE197124. Further inquiries can be directed to the corresponding author. A preprint version of this article is available on Research Square.