The Super Enhancer-Driven Long Noncoding RNA PRKCQ-AS1 Promotes Neuroblastoma Tumorigenesis by Interacting With MSI2 Protein and Is Targetable by Small Molecule Compounds

Abstract

Tumorigenic drivers of MYCN gene nonamplified neuroblastoma remain largely uncharacterized. Long noncoding RNAs (lncRNAs) regulate tumorigenesis, however, there is little literature on therapeutic targeting of lncRNAs with small molecule compounds. Here PRKCQ-AS1 is identified as the lncRNA most overexpressed in MYCN nonamplified, compared with MYCN-amplified, neuroblastoma cell lines. PRKCQ-AS1 expression is controlled by super-enhancers, and PRKCQ-AS1 RNA bound to MSI2 protein. RNA immunoprecipitation and sequencing identified BMX mRNA as the transcript most significantly disrupted from binding to MSI2 protein, after PRKCQ-AS1 knockdown. PRKCQ-AS1 or MSI2 knockdown reduces, while its overexpression enhances, BMX mRNA stability and expression, ERK protein phosphorylation and MYCN nonamplified neuroblastoma cell proliferation. PRKCQ-AS1 knockdown significantly suppresses neuroblastoma progression in mice. In human neuroblastoma tissues, high levels of PRKCQ-AS1 and MSI2 expression correlate with poor patient outcomes, independent of current prognostic markers. AlphaScreen of a compound library identifies NSC617570 as an efficient inhibitor of PRKCQ-AS1 RNA and MSI2 protein interaction, and NSC617570 reduces BMX expression, ERK protein phosphorylation, neuroblastoma cell proliferation in vitro and tumor progression in mice. The study demonstrates that PRKCQ-AS1 RNA interacts with MSI2 protein to induce neuroblastoma tumorigenesis, and that targeting PRKCQ-AS1 and MSI2 interaction with small molecule compounds is an effective anticancer strategy.

Keywords: Neuroblastoma; RNA binding protein; long noncoding RNA; long noncoding RNA inhibitor; small molecule compounds; tumorigenesis.

Figure 1

The lncRNA PRKCQ‐AS1 is overexpressed in MYCN nonamplified human neuroblastoma cells. A‐B) RNA was extracted from two MYCN‐nonamplified (SY5Y and SK‐N‐AS) and four MYCN‐amplified (BE(2)‐C, Kelly, CHP134 and SK‐N‐DZ) neuroblastoma cell lines for RNA sequencing analysis of differential gene expression. Volcano plot A) and heatmap B) revealed the top lncRNAs most significantly differentially expressed between MYCN‐nonamplified and MYCN‐amplified cell lines, with the highest false discovery rate (FDR). C) PRKCQ‐AS1 RNA expression was examined by RT‐PCR with RNA extracted from MYCN‐nonamplified and MYCN‐amplified human neuroblastoma cell lines. Data were shown as the mean ± standard error of three independent experiments. D) PRKCQ‐AS1 lncRNA expression was extracted from the publicly available Maris – 41 – FPKM – rsg001 RNA sequencing dataset of 39 human neuroblastoma cell lines downloaded from the R2 genomics analysis and visualization platform (http://r2.amc.nl). Differential expression of PRKCQ‐AS1 between MYCN‐nonamplified and MYCN‐amplified neuroblastoma cell lines was examined by two‐sided unpaired Student's t‐test, * indicates p < 0.05.

Figure 2

PRKCQ‐AS1 is overexpressed in MYCN‐nonamplified neuroblastoma cells due to transcriptional super‐enhancers. A‐B) Publicly available paired H3K27ac ChIP‐Seq and RNA‐Seq data from 36 neuroblastoma cell lines were analyzed. Scatterplot revealed strong positive correlation between cumulative ChIP‐Seq H3K27ac signal at the SE_513 super‐enhancer locus and PRKCQ‐AS1 RNA expression across the 36 neuroblastoma cell lines (A). Track plots showed the SE_513 super‐enhancer in MYCN nonamplified SK‐N‐AS, SK‐N‐SH, SH‐EP and GI‐MEN but not MYCN‐amplified CHP134, Kelly, IMR32 and SK‐N‐DZ neuroblastoma cell lines at the PRKCQ‐AS1 gene locus. The x‐axis indicated genomic position and the y‐axis depicted ChIP‐seq H3K27ac signal, in arbitrary units normalized against input control, and super‐enhancer regions (SE_513) were boxed in red (B). C‐E) SK‐N‐AS and SK‐N‐SH cells were transfected with control siRNA, BRD4 siRNA‐1, BRD4 siRNA‐2 (C‐D), CDK7 siRNA‐1 or CDK7 siRNA‐2 (E) for 48 h. The expression of BRD4, CDK7, and PRKCQ‐AS1 RNA was analyzed by immunoblot and/or RT‐PCR. F‐G) SK‐N‐AS and SK‐N‐SH cells were treated with vehicle control, 250 × 10⁻⁹ m or 1000 × 10⁻⁹ m AZD5153 (F), 32 × 10⁻⁹ m or 64 × 10⁻⁹ m THZ1 G) for 24 h. RNA was extracted from the cells for RT‐PCR analyses of PRKCQ‐AS1 RNA expression. Data were shown as the mean ± standard error of three independent experiments and evaluated by one‐way ANOVA. *, ** and *** indicates p < 0.05, 0.01 and 0.001 respectively.

Figure 3

PRKCQ‐AS1 is required for MYCN‐nonamplified neuroblastoma cell proliferation. A‐C) SK‐N‐AS and SK‐N‐SH cells were transfected with control siRNA, PRKCQ‐AS1 siRNA‐1 or PRKCQ‐AS1 siRNA‐2. RT‐PCR analysis of PRKCQ‐AS1 (A) and PRKCQ (B) RNA expression was performed 48 h after transfections, and Alamar Blue assays were performed 96 h (SK‐N‐AS cells) or 144 h (SK‐N‐SH cells) after transfections (C). D‐E) DOX‐inducible control shRNA, PRKCQ‐AS1 shRNA‐1 or PRKCQ‐AS1 shRNA‐2 SK‐N‐AS and SK‐N‐SH cells were treated with vehicle control or DOX for 48 h, followed by RT‐PCR analysis of PRKCQ‐AS1 D) or PRKCQ (E) RNA expression. F‐G) DOX‐inducible control shRNA, PRKCQ‐AS1 shRNA‐1 or PRKCQ‐AS1 shRNA‐2 SK‐N‐AS and SK‐N‐SH cells were treated with vehicle control or DOX for 96 h (SK‐N‐AS cells) or 144 h (SK‐N‐SH cells), followed by Alamar blue assays (F) or BrdU incorporation assays (G). H‐I) NB69 MYCN‐nonamplified neuroblastoma cells were transfected with an empty vector or PRKCQ‐AS1 expression construct, followed by RT‐PCR analysis of PRKCQ‐AS1 expression (H) and Alamar blue assays (I). Data were shown as the mean ± standard deviation of three independent experiments and evaluated by two‐tailed unpaired Student's t‐test for two groups or one‐way ANOVA for more than two groups. * and ** indicated p < 0.05 and 0.01 respectively.

Figure 4

PRKCQ‐AS1 is required for MYCN‐nonamplified neuroblastoma cell clonogenicity in vitro and tumor progression in mice. A‐B) DOX‐inducible control shRNA, PRKCQ‐AS1 shRNA‐1 or PRKCQ‐AS1 shRNA‐2 SK‐N‐AS and SK‐N‐SH cells were treated with vehicle control or DOX for 14 days. Cells were then fixed with crystal violet (A) and the numbers of colonies were quantified (B). Data were shown as the mean ± standard error of three independent experiments and evaluated by two‐tailed unpaired Student's t‐test. *** indicated p < 0.001. C‐D) DOX‐inducible PRKCQ‐AS1 shRNA‐1 SK‐N‐AS and PRKCQ‐AS1 shRNA‐2 SK‐N‐SH cells were xenografted into mice. When tumors reached 0.05cm³, the mice were divided into two sub‐groups and fed with food with or without DOX, and tumor growth was monitored. When tumors reached 1.0cm³, the mice were culled (C). The probability of overall survival of the mice was analyzed by Kaplan‐Meier survival curves (D). P value was obtained from two‐sided log‐rank test.

Figure 5

PRKCQ‐AS1 and its binding protein MSI2 up‐regulate BMX mRNA expression and ERK protein phosphorylation. A) PRKCQ‐AS1 RNA was in vitro transcribed and incubated with protein lysates from SK‐N‐AS cells, followed by mass spectrometry analysis of PRKCQ‐AS1 RNA‐binding proteins. *Score: total score from each identified peptide for each protein; †matches/sequences: peptides/sequences for each protein (identified peptides/sequences that matched the corresponding protein); ‡ emPAI: exponentially modified protein abundance index. B‐D) RNA‐immunoprecipitation assays were performed with a control IgG or an anti‐MSI2 antibody (Ab) in SK‐N‐AS and SK‐N‐SH cells. Immunoprecipitated protein was subjected to immunoblot analysis with the anti‐MSI2 Ab (B), and immunoprecipitated RNA was subjected to PCR analysis of PRKCQ‐AS1 RNA (C) and BMX mRNA (D). E) DOX‐inducible PRKCQ‐AS1 shRNA‐1 SK‐N‐AS and PRKCQ‐AS1 shRNA‐2 SK‐N‐SH cells were treated with control or DOX for 48 h, followed by RNA immunoprecipitation assays with a control IgG or an anti‐MSI2 Ab and RT‐PCR analysis of BMX mRNA. F‐G) DOX‐inducible control shRNA, PRKCQ‐AS1 shRNA‐1 and PRKCQ‐AS1 shRNA‐2 SK‐N‐AS and SK‐N‐SH cells were treated with vehicle control or DOX for 48 h, followed by RT‐PCR and immunoblot analyses of BMX mRNA (F) and protein (G) expression, total ERK (ERK) and phosphorylated ERK (phos‐ERK) proteins. H‐I) SK‐N‐AS and SK‐N‐SH cells were transfected with control siRNA, MSI2 siRNA‐1 or MSI2 siRNA‐2 for 48 h, followed by RT‐PCR and immunoblot analyses of BMX mRNA (H) and protein expression and ERK protein phosphorylation (I). J) SK‐N‐AS and SK‐N‐SH cells were transfected with control siRNA, BMX siRNA‐1 or BMX siRNA‐2 for 48 h, followed by immunoblot analyses of BMX expression and ERK phosphorylation. K) SK‐N‐AS cells were transfected with control siRNA, PRKCQ‐AS1 siRNA‐2 or MSI2 siRNA‐2 for 48 h, followed by actinomycin treatment at 5 µg mL⁻¹ every 12 h for 36 h. RNA was extracted for RT‐PCR analysis of BMX mRNA expression and half‐life calculation. Data were shown as the mean ± standard error of three independent experiments and evaluated by two‐sided unpaired Student's t‐test for two groups or by two‐way ANOVA for more than two groups. *, ** and *** indicated p < 0.05, 0.01 and 0.001 respectively.

Figure 6

PRKCQ‐AS1 interacts with MSI2 to induce neuroblastoma cell proliferation by up‐regulating BMX expression. A‐D) SK‐N‐AS and SK‐N‐SH cells were transfected with control siRNA, MSI2 siRNA‐1, MSI2 siRNA‐2, BMX siRNA‐1 or BMX siRNA‐2 for 96 h (SK‐N‐AS cells) or 144 h (SK‐N‐SH cells), followed by Alamar blue assays (A, C) or BrdU incorporation assays (B, D). E‐F) SHEP cells were stably transfected with empty vector or BMX expression construct, followed by immunoblot analysis of total ERK (ERK) and phosphorylated ERK (phos‐ERK) proteins 48 h later (E) or Alamar blue assays 72 h later (F). G‐I) NB69 cells were co‐transfected with an empty vector or PRKCQ‐AS1 expression construct, together with control siRNA, BMX siRNA‐1, BMX siRNA‐2, MSI2 siRNA‐1 or MSI2 siRNA‐2, followed by RT‐PCR analysis of BMX mRNA (G), immunoblot analysis of BMX protein expression and ERK protein phosphorylation (H), or Alamar blue assays (I). J‐L) SY5Y cells were co‐transfected with an empty vector or MSI2 expression construct, together with control siRNA, BMX siRNA‐1, BMX siRNA‐2, PRKCQ‐AS1 siRNA‐1 or PRKCQ‐AS1 siRNA‐2, followed by RT‐PCR analysis of BMX mRNA (J), immunoblot analysis of BMX protein expression and ERK protein phosphorylation (K), or Alamar blue assays (L). Data were shown as the mean ± standard deviation of three independent experiments and evaluated by two‐sided unpaired Student's t‐test for two groups or by ANOVA for more than two groups. *, ** and *** indicated p < 0.05, 0.01 and 0.001 respectively.

Figure 7

High levels of PRKCQ‐AS1 and MSI2 expression in human neuroblastoma tissues predict poor patient prognosis. A–F) PRKCQ‐AS1, MSI2 and BMX RNA expression in the total cohort of 476 human neuroblastoma tissues (A–C) and the 405 MYCN‐nonamplified human neuroblastoma tissues (D‐F) was extracted from the publicly available microarray gene expression Kocak dataset. Kaplan–Meier curves showed the probability of overall survival of patients according to the median levels of PRKCQ‐AS1 (A and D), MSI2 (B and E) or BMX (C and F) RNA expression. G‐H) Multivariable Cox regression modelling of the Kocak dataset, with PRKCQ‐AS1 (G) and MSI2 (H) expression levels in tumor tissues considered high or low relative to their median expression levels in the 476 neuroblastoma patients of the Kocak dataset. Hazard ratios (HR) were calculated as the antilogs of the regression coefficient in the proportional hazard. The prognostic factors were dichotomized by the median PRKCQ‐AS1 or MSI2 values and tumor stage was categorized as favorable (INSS stages 1, 2, and 4S) or unfavorable (INSS stages 3 and 4). p value was obtained from two‐sided log‐rank test.

Figure 8

The small molecule compound NSC617570 suppresses the interaction between PRKCQ‐AS1 RNA and MSI2 protein and shows anticancer effects in vitro and in vivo. A) PRKCQ‐AS1 RNA fragment 1 (65‐306 bp at the 5′‐ end) was in vitro transcribed, biotin‐labelled, mixed with Flag‐tagged MSI2 protein, and subjected to AlphaScreen of 2932 small molecule compounds at 10 µM for inhibitors of the interaction between PRKCQ‐AS1 RNA and MSI2 protein. Eight‐five compounds were found to reduce the interaction between PRKCQ‐AS1 RNA and MSI2 protein by > 70% (above the red line). B) Structure of the “hit” compound NSC617570. C) SK‐N‐AS cells were treated with vehicle control or 10 µM NSC617570 for 6 h, followed by RNA immunoprecipitation assays with control IgG or MSI2 antibody and RT‐PCR with primers targeting PRKCQ‐AS1 or BMX. Data were shown as the mean ± standard deviation of three independent experiments and evaluated by Student's t‐test. * indicated p < 0.05. D‐E) SK‐N‐AS and SK‐N‐SH cells were treated with vehicle control, 5 µM or 10 µM NSC617570 for 48 h, followed by RT‐PCR analysis of BMX mRNA (D) and immunoblot analysis of BMX protein, total ERK protein (ERK) and phosphorylated ERK protein (phos‐ERK) (E). F) SK‐N‐AS and SK‐N‐SH cells were treated with vehicle control or a range of doses of NSC617570 for 72 h, followed by Alamar blue assays. Data were shown as the mean ± standard deviation of three independent experiments and evaluated by ANOVA. *, ** and *** indicated p < 0.05, 0.01 and 0.001 respectively. G) SK‐N‐AS and SK‐N‐SH cells were treated with vehicle control, 5 µM or 10 µM NSC617570 for 14 (SK‐N‐AS) or 21 (SK‐N‐SH) days, followed by clonogenic assays. H‐I) SK‐N‐AS cells were xenografted into nude mice. When tumors reached 50 mm³, the mice were treated with NSC617570 at 8.75 mg/kg or vehicle control via i.p. injection, 5 times per week. Tumor growth was monitored (H) and mouse overall survival was analyzed (I). For survival analysis, P value was obtained from two‐sided log‐rank test.