NFIB Mediates BRN2 Driven Melanoma Cell Migration and Invasion Through Regulation of EZH2 and MITF

Abstract

While invasion and metastasis of tumour cells are the principle factor responsible for cancer related deaths, the mechanisms governing the process remain poorly defined. Moreover, phenotypic divergence of sub-populations of tumour cells is known to underpin alternative behaviors linked to tumour progression such as proliferation, survival and invasion. In the context of melanoma, heterogeneity between two transcription factors, BRN2 and MITF, has been associated with phenotypic switching between predominantly invasive and proliferative behaviors respectively. Epigenetic changes, in response to external cues, have been proposed to underpin this process, however the mechanism by which the phenotypic switch occurs is unclear. Here we report the identification of the NFIB transcription factor as a novel downstream effector of BRN2 function in melanoma cells linked to the migratory and invasive characteristics of these cells. Furthermore, the function of NFIB appears to drive an invasive phenotype through an epigenetic mechanism achieved via the upregulation of the polycomb group protein EZH2. A notable target of NFIB mediated up-regulation of EZH2 is decreased MITF expression, which further promotes a less proliferative, more invasive phenotype. Together our data reveal that NFIB has the ability to promote dynamic changes in the chromatin state of melanoma cells to facilitate migration, invasion and metastasis.

Keywords: BRN2; Epigenetic; Invasion; Melanoma; Metastasis; NFIB.

Fig. 1

NFIB expression correlates with BRN2 in melanocytic and melanoma cells. (A–D) QPCR analysis on A2058 melanoma cells following lentiviral transduction to produce stable over-expression of MITF or BRN2, investigating NFIA, NFIB, NFIC, and NFIX expression. Data represented as fold change relative to the empty vector control and normalized to B2M gene. (E–F) Western Blot analysis on neonatal foreskin-derived QF1236 and QF1566 primary human melanoblast cells induced to differentiate into pigmented melanocytes over a 5-day period. Antibodies were used against BRN2, NFIB, MITF, and GAPDH. (G) Whole cell lysates from six human melanoma cell lines immunoblotted for NFIB, BRN2, and MITF. (H) ChIP-ChIP analysis data (Kobi et al., 2010) in 501 Mel human melanoma cells investigating BRN2 binding to chromatin regions, reveals BRN2 binds to a 2Kb intronic region located upstream of the NFIB promoter. *: P < 0.05, **: P < 0.01, ***: P < 0.001. Data representative of three independent experiments. Band expression intensity of Western Blots was normalized to the first lane (GAPDH used as a loading control) using ImageJ software and indicated below each blot. Data from (A–D) is represented as the mean ± SEM and analysed using a one-way ANOVA with Dunnett's multiple comparisons test. See also Fig. S3.

Fig. 2

BRN2 positively regulates NFIB and EZH2 expression. (A) Cell lysates from A2058, MM96L, and HT144 human melanoma cells treated with three different siRNA directed against BRN2 were analyzed by Western Blot with antibodies against BRN2, NFIB, EZH2, and GAPDH (B) Cell lysates from A2058, MM96L, and HT144 human melanoma cells treated with lentivirus to create stable overexpression of BRN2 were analyzed by Western Blot with antibodies against BRN2, NFIB, EZH2, and GAPDH (C) Cell lysates from doxocycline-off inducible BRN2 expressing A2058, MM96L, and HT144 cells treated with and without dox for 48 h were analyzed by Western Blot with antibodies against BRN2, NFIB, EZH2, and GAPDH. (D) NFIB immunofluorescence (red) on A2058 parental cells grown on coverslips and treated with siRNA directed against BRN2. DAPI used to stain cell nuclei. (E) NFIB immunofluorescence (red) on A2058 BRN2 over-expressing cells. DAPI used to stain cell nuclei. (F) Cell lysates from A2058 and MM96L melanoma cells stably over-expressing BRN2 were analyzed by Western Blot for H3K27 tri methylation status (EZH2 global methylation marker). Band expression intensity of Western Blots was normalized to the first lane (GAPDH used as a loading control) using ImageJ software and indicated below each blot. All data representative of three independent experiments.

Fig. 3

NFIB manipulation increases migration and EZH2 expression and decreases MITF expression. (A) Cell lysates from A2058, MM96L, and HT144 human melanoma cells treated with three different siRNA directed against NFIB were analyzed by Western Blot with antibodies against BRN2, NFIB, EZH2, MITF, and GAPDH. (B) Cell lysates from A2058, MM96L, and HT144 human melanoma cells treated with lentivirus to create stable overexpression of NFIB were analyzed by Western Blot with antibodies against BRN2, NFIB, EZH2, MITF and GAPDH. (C) A2058 stables for MITF, BRN2, NFIB, and empty control cells transfected with two luciferase reporter constructs; A wild type construct containing a region of the EZH2 promoter containing the NFIB putative binding site (WT) and a mutant construct with the NFI binding site mutated out (Mut). Data represented as relative luciferase fold activity following normalisation against the empty-Mut. (D) Cell lysates from stable NFIB overexpressing A2058 and MM96L melanoma cells were analyzed by Western Blot for H3K27 tri methylation status (EZH2 global methylation marker). (E) Quantification of wound healing assay performed on A2058 human melanoma cells treated with siRNA against BRN2, MITF or a scrambled control 24 h prior to wound initiation. (F) Quantification of wound healing assay performed in A2058 stable BRN2, MITF and empty control cells. (G) Quantification of wound healing assay performed in A2058 human melanoma cells treated with siRNA against BRN2, MITF or a scrambled control 24 h prior to commencement of the experiment. (H) Quantification of wound healing assay performed in A2058 stable NFIB and empty control cells. Data represented as the mean ± SEM. *: P < 0.05, ***: P < 0.001, ****: P < 0.0001. A two-way ANOVA with a Tukey's post-hoc test was performed in (C) and (E–H). All data representative of three independent experiments. See also Figs. S1, S2, S3 and S5.

Fig. 4

NFIB drives migration downstream of BRN2 through interactions with EZH2. (A–B) Lightphase images and quantification of wound healing assays performed in A2058 stable BRN2 cells treated with siRNA against BRN2, NFIB or a scrambled control 24 h prior to wound initiation (0 h), with images taken at 0, 24, and 48 h. (C) Cell lysates from A2058 BRN2 stable melanoma cells treated with siRNA against BRN2, NFIB or a scrambled control were analyzed by Western Blot with antibodies against BRN2, NFIB, EZH2, and GAPDH. (D-E) Lightphase images and quantification of wound healing assays performed in A2058 stable NFIB cells treated with EZH2 inhibitor GSK343 at 1 μM or a vehicle control (DMSO) 24 h prior to wound inititation (0 h), with images taken at 0, 24, and 48 h. (F) Cell lysates from A2058 NFIB stable melanoma cells treated with GSK343 at 0.1 μM, 1 μM or a vehicle control (DMSO) were analyzed by Western Blot with antibodies against BRN2, NFIB, EZH2, MITF and GAPDH. (G-H) Lightphase images and quantification of wound healing assays performed in A2058 stable NFIB melanoma cells treated with MITF or empty control lentivirus 24 h prior to commencement of the experiment, with images taken at 0, 24, and 48 h. Data represented as the mean ± SEM and analysed with two-way ANOVA with a Tukey's post hoc test. **: P < 0.01, ***: P < 0.001, ****: P < 0.0001. All data representative of three independent experiments. Band expression intensity of Western Blots was normalized to the first lane (GAPDH used as a loading control) using ImageJ software and indicated below each blot. See also Figs. S2 and S6.

Fig. 5

Overexpression of BRN2 decreases melanoma cell tumourigenicity but increases invasion. (A–B) 2 × 10⁵ A2058 human melanoma cells with stable over-expression of empty control or NFIB were injected subcutaneously into the hind flanks of five 5-week old male immunocompromised BALB/c Foxn1^nu mice. Three-dimensional measurement was performed two times per week, with tumour volume expressed as mm³. (C) Analysis of 471 melanoma samples in the TCGA dataset comparing NFIB expression and its correlation with a previously reported invasive gene signature (Verfaillie et al., 2015). Grey bars represent average expression of NFIB in each individual tumour, while the black line is a rank based on the average expression of the invasive gene signature used for initial sorting of the samples from low invasive to high invasive phenotypes. The purple line represents a moving average of NFIB expression per 20 tumours. Linear regression analysis reveals a Spearman P-value = 3.553e − 15 indicating a positive correlation between NFIB and invasiveness. (D) MM96L stable BRN2, MITF, NFIB, or empty human melanoma cells grown on agarose to generate 3D non-adherent melanoma spheroids. Spheroids were embedded in a collagen-media mixture and left to grow over a 72 h time-frame, with light phase photographs taken every 24 h. (E) Spheroid invasion was calculated from (D) by determining the change in the area of the invading cells disseminating away from the spheroid at 24 h time intervals relative to the 0 h timepoint. (F) The change in spheroid size was determined in (D) by measuring the change in area occupied by the spheroid alone (not the invasive populations) at 0 h vs. 72 h. *: P < 0.001, **: P < 0.001 ****: P < 0.0001. A two-way ANOVA with Tukey's post hoc test was performed on E–F. Changes in area occupied by invading cells and spheroid growth were calculated using ImageJ software. Data from (D–F) is representative of three independent experiments and is represented as the mean ± SEM. See also Fig. S3.

Fig. 6

NFIB shows colocalisation with BRN2 in melanoma tumours and shows increased expression in aggressive/metastatic melanoma models. (A) Immunofluorescence microscopy on A2058 xenograft tumours surgically excised, formalin-fixed, and embedded in paraffin. Tumours were sectioned at 5 μm thickness and antigen-retrieved before labeling with BRN2 (red) and NFIB (green) antibody. DAPI was used to stain nuclei. (B-C) Immunofluorescence microscopy as described above on patient derived subcutaneous primary melanoma tumours and Lymph node metastatic melanoma tumours. (D) Microarray analysis of melanoma clinical samples representing 31 primary melanomas and 52 melanoma metastases from a previously published data set (Xu et al., 2008). Relative RNA expression was plotted and linear regression analysis was performed investigating the relationship between BRN2 and NFIB expression in metastatic samples. (E) Regression analysis on the above dataset looking at a correlation between MITF and NFIB expression in metastatic tumours. (F) Analysis of relative NFIB expression (log2 transformed) in 102 primary and 368 metastatic tumours from the TCGA dataset. Data represented as a violin plot and analyzed using the Mann-Whitney rank test. (G) Microarray analysis of subcutaneous tumours or lung metastases from immunodeficient mice injected subcutaneously or intravenously with a poorly-metastatic A375 melanoma cell line or with highly-metastatic derivative cell lines from a previously published dataset (Xu et al., 2008). Relative NFIB RNA expression was investigated in three specific groups; Poorly metastatic (PM), Subcutaneous tumours (SC), and the resultant lung metastases from the aforementioned subcutaneous tumours (LM). (H) Microarray analysis of relative NFIB expression (log2 transformed) in primary cutaneous melanomas derived from iMet (metastasis-capable) and iHRAS (non-metastatic) models from a previously published dataset (Scott et al., 2011). *: P < 0.05, ***: P < 0.001. A one-way ANOVA with a Tukey's post-hoc test was performed on (H). Data is represented as the mean ± SEM. Scale bars in white represent 200 μm. See also Figs. S3 and S6.

Fig. 7

BRN2 and NFIB colocalise with EZH2 populations in vivo. (A) Immunofluorescence microscopy on A2058 xenograft tumours surgically excised, formalin-fixed, and embedded in paraffin. Tumours were sectioned at 5 μm thickness and antigen-retrieved before labeling with BRN2 (red) and EZH2 (green) antibody. DAPI was used to stain nuclei. (B–C) Immunofluorescence microscopy as described above on patient derived subcutaneous primary melanoma tumours and Lymph node metastatic melanoma tumours, labeled with BRN2 (red) and EZH2 (green). (D) Immunofluorescence microscopy on A2058 xenograft tumours as described above, labeled with NFIB (red) and EZH2 (green) antibody. (E–F) Immunofluorescence microscopy as described above on patient derived subcutaneous primary melanoma tumours and Lymph node metastatic melanoma tumours, labeled with NFIB (red) and EZH2 (green).