Twist-1 is upregulated by NSD2 and contributes to tumour dissemination and an epithelial-mesenchymal transition-like gene expression signature in t(4;14)-positive multiple myeloma

Abstract

Approximately 15% of patients with multiple myeloma (MM) harbour the t(4;14) chromosomal translocation, leading to the overexpression of the histone methyltransferase NSD2. Patients with this translocation display increased tumour dissemination, accelerated disease progression and rapid relapse. Using publicly available gene expression profile data from NSD2^high (n = 135) and NSD2^low (n = 878) MM patients, we identified 39 epithelial-mesenchymal transition (EMT)-associated genes which are overexpressed in NSD2^high MM plasma cells. In addition, our analyses identified Twist-1 as a key transcription factor upregulated in NSD2^high MM patients and t(4;14)-positive cell lines. Overexpression and knockdown studies confirmed that Twist-1 is involved in driving the expression of EMT-associated genes in the human MM cell line KMS11 and promoted the migration of myeloma cell lines in vitro. Notably, Twist-1 overexpression in the mouse MM cell line 5TGM1 significantly increased tumour dissemination in an intratibial tumour model. These findings demonstrate that Twist-1, downstream of NSD2, contributes to the induction of an EMT-like signature in t(4;14)-positive MM and enhances the dissemination of MM plasma cells in vivo, which may, in part, explain the aggressive disease features associated with t(4;14)-positive MM.

Keywords: Dissemination; Epithelial-mesenchymal transition; Multiple myeloma; NSD2; Twist-1.

Figure 1.. Identification of an EMT-like gene expression signature in t(4;14)-positive MM.

In silico analyses of gene expression in CD138⁺ BM MM PCs from NSD2^high and NSD2^low MM patients in publicly available microarray datasets (GSE19784 [NSD2^low, n = 288; NSD2^high, n = 40], GSE26863 [NSD2^low, n = 270; NSD2^high, n = 34], E-MTAB-317 [NSD2^low, n = 192; NSD2^high, n = 34], E-MTAB-363 [NSD2^low, n = 128; NSD2^high, n = 27]). Gene set enrichment analysis (GSEA) for MSigDB Hallmark gene sets was performed on genes that were significantly upregulated (Fishers’ combined p-value < 0.05) across the four datasets demonstrated enrichment for EMT-associated genes. The 20 most significantly enriched gene sets are shown (A). Heat map shows the average z-score for NSD2^high and NSD2^low patients for the 39 significantly upregulated genes in the MSigDB Hallmark EMT gene set (B). Venn diagram shows the overlap between the upregulated genes in MM PC from NSD2^high MM patients, genes that were significantly upregulated by NSD2 overexpression in KMS11-TKO cells and the 200 genes in the MSigDB Hallmark EMT gene set (C). Expression of the 13 EMT-associated genes that were significantly upregulated by NSD2 overexpression in KMS11-TKO cells (TKO-NSD2) and in NSD2^high MM patients is shown for control (TKO-EV) and TKO-NSD2 cells. Graphs depict mean + SEM; * p < 0.05 (LIMMA) (D).

Figure 2.. TWIST1 expression is upregulated by NSD2 in MM.

TWIST1 expression, as determined by microarray, was assessed in CD138-selected BM PCs from NSD2^low and NSD2^high MM patients (A). Upregulation of TWIST1 expression with NSD2 overexpression is shown in KMS11-TKO-NSD2 cells compared with control cells (KMS11-TKO-EV), as assessed by microarray (B). Expression of TWIST1 in t(4;14)-positive and non-t(4;14) MM cell lines was assessed in a publicly available RNAseq dataset (C) and by qRT-PCR in a panel of human B-cell and MM cell lines (D). The non-t(4;14) MM cell line RPMI-8226 was retrovirally transduced to constitutively overexpress NSD2 and NSD2 expression, normalised to GAPDH, was assessed by qRT-PCR (E). Upregulation of NSD2 protein in RPMI-8226-NSD2 cells, relative to RPMI-8226-EV cells, was confirmed by Western blotting on nuclear lysates, run on 10% SDS-PAGE gels, using histone H3 as a loading control. Numerical values indicate NSD2 protein levels relative to histone H3 protein levels (F). Relative TWIST1 mRNA expression, normalised to GAPDH, was measured by qRT-PCR in RPMI-8226-EV and RPMI-8226-NSD2 cell lines (G). Graphs depict the median and interquartile ranges (GSE19784 [NSD2^low, n = 288; NSD2^high, n = 40], GSE26863 [NSD2^low, n = 270; NSD2^high, n = 34], E-MTAB-317 [NSD2^low, n = 192; NSD2^high, n = 34], E-MTAB-363 [NSD2^low, n = 128; NSD2^high, n = 27], A; t(4;14), n = 18; non-t(4;14), n = 48; C) or mean + SEM of three independent biological replicates (B, D, E, G). * p < 0.05 (LIMMA [A, C] or unpaired t-test [B, E, G]).

Figure 3.. Twist-1 knockdown downregulates EMT-associated gene expression and inhibits cell migration in KMS11 cells.

TWIST1 expression was knocked down in t(4;14)-positive KMS11 cells by transfection with TWIST1 siRNAs (siTWIST1) or negative control siRNA (siNEG). Knockdown of Twist-1 was confirmed by qRT-PCR, normalised to GAPDH (A) and by Western blot on whole cell lysates, resolved on 12% SDS-PAGE gels, using anti-Twist-1 and HSP90 antibodies (B). Numerical values indicate Twist-1 protein levels relative to HSP90 protein levels. Expression of EMT-associated genes CDH2, SERPINE2, PLOD2 and VIM was assessed in siTWIST1 or siNEG KMS11 cells by qRT-PCR, normalised to GAPDH (C). Trans-well migration of KMS11 cells following Twist-1 siRNA knockdown was assessed in a trans-well assay in response to an FCS gradient, expressed relative to siNEG controls (D). Proliferation of KMS11 cells following Twist-1 siRNA knockdown was assessed using the BrdU assay, expressed relative to the siNEG control (E). Graphs depict mean + SEM of three (A, C, E) or four (D) independent experiments. *P < 0.05, unpaired t-test.

Figure 4.. Twist-1 overexpression in the murine MM cell line 5TGM1 increases cell migration in vitro and tumour dissemination in vivo.

5TGM1 cells were lentivirally transduced with a Twist1 expression vector or empty vector (EV) control and Twist1 expression, normalised to ActB was assessed by qRT-PCR (A). Overexpression of Twist-1 protein in 5TGM1-TWIST1 cells was validated by Western blot on nuclear protein extracts of 5TGM1-EV, 5TGM1-TWIST1 and positive control NIH-3T3 cells, resolved on 12% SDS-PAGE gels and immunoblotted with anti-Twist-1 and anti-histone H3 antibodies. Numerical values indicate Twist-1 protein levels relative to histone H3 protein levels (B). Trans-endothelial migration of 5TGM1-Twist1 and 5TGM1-EV controls cells (1×10⁵ cells/well) in response to FCS gradient was assessed in a trans-well assay (C). Proliferation of 5TGM1-Twist1 and 5TGM1-EV controls cells was assessed by WST-1, normalised to day 0 (D), or by BrdU assay, normalised to EV controls (E). C57BL/KalwRij mice were intravenously injected with 5×10⁵ 5TGM1-EV or 5TGM1-TWIST1 cells. Total body tumour burden was assessed using bioluminescence imaging at 2, 3 and 4 weeks (F). Bioluminescence images of a representative mouse from each group at week 4 are shown (G). The left tibiae of C57BL/KalwRij mice were injected with either 1×10⁵ 5TGM1-EV or 1×10⁵ 5TGM1-TWIST1 cells. After 3.5 weeks, tumour burden in the injected tibia (H) and in the contralateral tibia and femur (I) was assessed by detection of GFP+ tumour cells by flow cytometry, shown as a percentage of total mononuclear cells. Graphs depict mean ± SEM of three (A, D-E) or five (C) independent experiments, mean ± SEM of 8–10 mice/group (F) or median and interquartile range of 7–11 mice/group (H, I). * p < 0.05, unpaired t-test (A, C), two-way ANOVA with Sidak’s multiple comparison test (F) or Mann-Whitney test (H, I).