Histone demethylase KDM6B has an anti-tumorigenic function in neuroblastoma by promoting differentiation

Abstract

Induction of differentiation is a therapeutic strategy in high-risk neuroblastoma, a childhood cancer of the sympathetic nervous system. Neuroblastoma differentiation requires transcriptional upregulation of neuronal genes. How this process is regulated at epigenetic levels is not well understood. Here we report that the histone H3 lysine 27 demethylase KDM6B is an epigenetic activator of neuroblastoma cell differentiation. KDM6B mRNA expression is downregulated in poorly differentiated high-risk neuroblastomas and upregulated in differentiated tumors, and high KDM6B expression is prognostic for better survival in neuroblastoma patients. In neuroblastoma cell lines, KDM6B depletion promotes cell proliferation, whereas KDM6B overexpression induces neuronal differentiation and inhibits cell proliferation and tumorgenicity. Mechanistically, KDM6B epigenetically activates the transcription of neuronal genes by removing the repressive chromatin marker histone H3 lysine 27 trimethylation. In addition, we show that KDM6B functions downstream of the retinoic acid-HOXC9 axis in inducing neuroblastoma cell differentiation: KDM6B expression is upregulated by retinoic acid via HOXC9, and KDM6B is required for HOXC9-induced neuroblastoma cell differentiation. Finally, we present evidence that KDM6B interacts with HOXC9 to target neuronal genes for epigenetic activation. These findings identify a KDM6B-dependent epigenetic mechanism in the control of neuroblastoma cell differentiation, providing a rationale for reducing histone H3 lysine 27 trimethylation as a strategy for enhancing differentiation-based therapy in high-risk neuroblastoma.

Fig. 1. Low KDM6B expression is associated with poor prognosis in neuroblastoma patients.

a–c Kdm6b expression is downregulated in mouse neuroblastoma sphere-forming cells as measured by microarray (a), qRT-PCR (b, error bars, s.d., n = 3), and immunoblotting (c), in comparison with primary tumor cells. α-tubulin levels are shown as loading control (c). Data were analyzed using two-tailed Student’s t-test. d–e Low KDM6B expression is associated with reduced event-free survival in neuroblastoma patients (d), high-risk neuroblastoma (e, left panel), and advanced stages of the tumors (e, right panel). Patient data analyses (d–e) were conducted online (R2 Genomics Analysis and Visualization Platform), and the resulting figures and log-rank test (d) and Student’s t-test (e) p values were downloaded. **p < 0.01; ***p < 0.001

Fig. 2. KDM6B inhibits the proliferation and tumorigenicity of neuroblastoma cells.

a Immunoblot analysis of overexpression of HA-tagged KDM6B in BE(2)-C cells. α-tubulin levels are shown as loading control. b–c Growth curves of human neuroblastoma cell lines (b) and mouse neuroblastoma sphere-forming cell lines (c) with or without overexpression of HA-tagged human KDM6B obtained by trypan blue exclusion assay. d Xenograft assay of BE(2)-C and SK-N-AS cells without (vector control) or with overexpression of HA-tagged KDM6B. Tumor weight was analyzed by scatter plot with horizontal lines indicating the mean (n = 6). Data were analyzed using two-tailed Student’s t-test with p values indicated. e–f qRT-PCR (e) and immunoblot (f) analyses of KDM6B mRNA and protein expression in BE(2)-C cells 7 days after infection with control (shGFP) or shKDM6B-expressing lentiviruses. Values (e) correspond to the mean of three technical replicates ± s.d. and are representative of two independent experiments. α-tubulin levels are shown as loading control (f). g Growth curves of the indicated human neuroblastoma cell lines infected with lentiviruses expressing shGFP (control) or shKDM6B. All cell growth data (b, c, g) correspond to the mean of four technical replicates ± s.d. and are representative of at least two independent experiments. Data were analyzed by two-way ANOVA with p values indicated

Fig. 3. KDM6B induces neuronal differentiation of neuroblastoma cells.

a Phase contrast images of BE(2)-C and SH-SY5Y cells without (vector control) or with overexpression of HA-tagged KDM6B or KDM6B-H1390A. b qRT-PCR analysis of GFRA3, NEFM, and RET mRNA expression in BE(2)-C and SH-SY5Y cells without (vector control) or with overexpression of HA-tagged KDM6B or KDM6B-H1390A. Error bars represent s.d. (n = 3). Data were analyzed by two-tailed Student’s t-test. c Immunoblot analysis of NEFM levels in BE(2)-C and IMR32 cells without (vector control) or with overexpression of HA-tagged KDM6B or KDM6B-H1390A. NEFM levels were quantified against α-tubulin and are presented as the fraction of the NEFM levels in vector control cells. d–e ChIP-qPCR analysis of KDM6B (d) and H3K27me3 (e) levels in the promoter region of NEFM in BE(2)-C cells without (vector control) or with overexpression of HA-tagged KDM6B. Data correspond to the mean of three technical replicates±s.d. and are representative of two independent ChIP experiments. Data were analyzed by two-tailed Student’s t-test. ***p < 0.001

Fig. 4. Retinoic acid induces KDM6B via HOXC9.

a–b qRT-PCR (a) and immunoblot (b) analysis of mRNA and protein expression of the indicated genes in BE(2)-C cells treated with vehicle (DMSO) or 5 µM RA for the indicted times. β-actin levels are shown as loading control (b). c qRT-PCR analysis of HOXC9 and KDM6B mRNA expression in BE(2)-C cells infected with lentiviruses expressing shGFP or shHOXC9 and treated with DMSO or 5 µM RA for 7 days. d–e qRT-PCR analysis of KDM6A and KDM6B mRNA expression (d) and immunoblot analysis of KDM6B and HOXC9 protein expression (e) in BE(2)-C cells with inducible HOXC9 expression in the absence of doxycycline (Doxy). All qRT-PCR data are presented as mean ± s.d. (n = 3). α-tubulin levels are shown as loading control (e). f ChIP-qPCR analysis of HOXC9 levels at the KDM6B locus in BE(2)-C cells with inducible expression of Myc-tagged HOXC9 in the absence of Doxy. Data correspond to the mean of three technical replicates±s.d. and are representative of two independent ChIP experiments. All data were analyzed by two-tailed Student’s t-test. ***p < 0.001

Fig. 5. KDM6B is required for HOXC9-induced neuronal differentiation.

a–b qRT-PCR (a) and immunoblot (b) analyses of KDM6B mRNA and protein expression in BE(2)-C_ and SK-N-DZ_tetoff-myc-HOXC9 cells infected with lentiviruses expressing shGFP or shKDM6B. Error bars represent s.d. (n = 3). α-tubulin levels are shown as loading control (b). c Growth curves of BE(2)-C_ and SK-N-DZ_tetoff-myc-HOXC9 cells expressing either shGFP or shKDM6B-36 that were cultured in the presence or absence of Doxy for the indicted times. Data correspond to the mean of four technical replicates ± s.d. and are representative of at least two independent experiments. Data were analyzed by two-way ANOVA with p values indicated. d qRT-PCR analysis of GFRA3, NEFM, and RET mRNA expression in BE(2)-C_tetoff-myc-HOXC9 cells expressing either shGFP or shKDM6B-36 that were cultured in the presence or absence of doxycycline (Doxy) for 9 days. Data were analyzed by two-tailed Student’s t-test. ***p < 0.001

Fig. 6. KDM6B interacts with HOXC9 for targeting neuronal genes for epigenetic activation.

a ChIP-qPCR analysis of HOXC9, KDM6B, and H3K27me3 levels in the promoter region of NEFM in BE(2)-C_tetoff-myc-HOXC9 cells that were cultured in the presence of absence of Doxy for 6 days. b ChIP-qPCR analysis of KDM6B levels in the promoter region of NEFM in BE(2)-C cells with overexpression of HA-tagged KDM6B without (shGFP) or with HOXC9 knockdown by shHOXC9-9. All ChIP-qPCR data correspond to the mean of three technical replicates ±s.d. and are representative of two independent ChIP experiments. Data were analyzed by two-tailed Student’s t-test. ***p < 0.001. c Co-IP showing HOXC9 interaction with endogenous KDM6B in BE(2)-C_tetoff-myc-HOXC9 cells following HOXC9 induction in the absence of Doxy for 6 days. IgH immunoglobulin heavy chain

Fig. 7. Model for HOXC9-KDM6B cooperation in RA induction of neuronal differentiation in neuroblastoma cells.

RA induces HOXD8, which, in turn, transcriptionally activates HOXC9 expression. HOXC9 then activates KDM6B transcription and interacts with KDM6B for targeting neuronal genes for epigenetic activation by removing the repressive chromatin marker H3K27me3