BET inhibitors downregulate the expression of the essential lncRNA SMILO in multiple myeloma through regulation of the transcription factor FLI1

Bromodomain and extraterminal domain (BET) proteins are epigenetic readers that bind to acetylated histones in active enhancer and promoter regions participating in their transcriptional regulation.^1-3 Several lines of evidence indicate the role of these BET proteins in oncogenic enhancers of human cancers providing the rationale to target BET proteins as a therapeutic strategy in different tumors. BET inhibitors (BETi) are small molecules that displace BET from the chromatin disrupting the transcriptional complex, including transcription factors (TF) and co-activators, leading to the inhibition of gene expression.^1,4

We have recently described that patients with multiple myeloma (MM) are characterized by the presence of an aberrant de novo activation of chromatin regions located mainly at promoters and enhancers that regulate genes involved in the pathogenesis of MM, including long non-coding RNA (lncRNA).^5,6 Based on these findings we hypothesized that the use of BETi in MM could modulate these de novo active regions contributing to the elimination of abnormal myeloma plasma cells.

To assess this hypothesis, we conducted a RNA-sequencing (RNA-seq) analysis following a 72-hour treatment of the MM-derived cell lines KMS-11, RPMI 8226, and KMS-12-BM using a non-cytotoxic concentration (1 μM) of the BETi JQ1 (dose that inhibits the growth of cells by 50% [GI⁵⁰]: 43.3 µM for KMS-11, 56.2 μM for RPMI 8226 and 19.6 μM KMS-12-BM), a pan-BETi with a well-described mechanism of MYC inhibition in hematologic malignancies^1,2,4 (our research complies with all relevant ethical regulations and has been approved by the committee of ethics of research of the University of Navarra). This analysis revealed a significant downregulation of more than 10% of genes regulated by de novo active regions in MM cell lines (Figure 1A). Notably, previously identified BETi targets such as MYC, IRF4, POU2AF1 and DUSP22⁷ were also downregulated (Figure 1B). These results suggest that JQ1 is able to negatively modify the expression of genes regulated by de novo active regions in MM. Interestingly, we observed that one of the genes with greatest inhibition induced by JQ1 was the lncRNA SMILO (LINC00582) (Figure 1B). The inhibition of the expression of SMILO and MYC were validated by real-time quantitative polymerase chain reaction (RT-qPCR) in the three MM cell lines used before (Figure 1C). The fact that pharmacological regulation of the lncRNA has been proposed as a therapeutic approach^8,9 and that we have previously described that SMILO is overexpressed in 64% of MM patients and is an essential lncRNA for the proliferation of MM⁶ suggested that JQ1 could be used not only to explore the mechanism of regulation of this lncRNA but also as a therapeutic approach to target lncRNA in MM. In order to determine whether JQ1 was indeed regulating gene expression by modulation of de novo active chromatin regions, we performed the assay for transposase-accessible chromatin using sequencing (ATAC-seq) and chromatin immunoprecipitation sequencing (ChIP-seq) for H3K27ac in KMS-11, RPMI 8226 and KMS-12-BM cells after 72 hours of JQ1 treatment. We observed a modest change in the percentage of chromatin activation marks or chromatin accessibility in de novo chromatin active regions (Figure 1D), even when we considered regulatory regions of genes downregulated by JQ1, with only 1.5% and 2.9% exhibiting a reduction in chromatin activation and accessibility, respectively. This is exemplified by the lack of changes in chromatin regions regulating the expression of SMILO (Figure 1E). These results indicate that JQ1 may displace the transcriptional machinery, such as TF, from de novo active regions and specifically from those related to SMILO.

In order to identify potential TF bound to the active chromatin regions of SMILO, thereby responsible for regulation of this lncRNA, we performed a Reverse-ChIP¹⁰ in three accessible and de novo active chromatin regions of SMILO (Online Supplementary Figure S1A). The Reverse-ChIP was carried out in three MM cell lines, two with high SMILO expression (KMS-11 and MM.1R), and one exhibiting low levels of the lncRNA (KMS-12-BM) (Online Supplementary Figure S1B). We focused our analysis on proteins shared by KMS-11 and MM.1R, excluding those proteins that were present in KMS-12-BM. The analysis of the Reverse-ChIP identified 16 proteins associated with the three SMILO regions analyzed in KMS-11 and MM.1R cell lines (Figure 2A). When we analyzed the mRNA expression level of these 16 genes in B cells at different stages of differentiation (from naïve B cells to plasma cells), and MM patients we observed that four of them, FLI1, SUB1, PKP2, and MAFG, were overexpressed in MM (Figure 2B; Online Supplementary Figure S2A). Among these candidates, FLI1 showed the highest positive correlation with SMILO expression (Figure 2C; Online Supplementary Figure S2B). Overexpression of FLI1 has been linked to different genomic alterations and mutations not only in solid tumors but also in several hematological malignancies.^11,12 The analysis of a large cohort of MM patients indicated a very low frequency of mutations of FLI1 (3 of 604 patients from the CoMMpass MM patient’s cohort). Furthermore, the expression of FLI1 along with the expression of SMILO was not associated with classical genetic alterations in MM (Online Supplementary Figure S3A). However, FLI1 presented de novo chromatin active marks in MM (Online Supplementary Figure S3B), suggesting that overexpression of FLI1 could be epigenetically regulated. High expression of FLI1 was associated with a significant decrease in progression-free survival (PFS) and overall survival (OS) of patients with MM using the CoMMpass (IA18) database (Online Supplementary Figure S3C, D). On the multivariate analysis, when considering classic high risk genetic alterations in MM as described before,¹³ high expression of FLI1, together with the del13q, amp1q and International Staging System (ISS) stage resulted in a significantly lower PFS (global P<0.0001). According to the number of events, patients were stratified into five risk groups (Online Supplementary Figure S3E, F). Similarly, high expression of FLI1 together with del13q, del17p, amp1q or ISS stage identified six groups with significant differences in their OS (P<0.0001) (Online Supplementary Figure S3G, H). These results suggest that FLI1 is an important TF with a potential therapeutic as well as prognostic impact in patients with MM.

Figure 1.The expression of de novo active long non-coding RNA SMILO decrease after JQ1 treatment. (A) Graphical representation of the number of genes associated with de novo active chromatin regions that present a significant downregulation (P<0.05 and a log2 fold change [log2FC]<-1) of their expression after JQ1 treatment in each multiple myeloma (MM) cell line. In this graph the expression of 1,059 coding genes and 8 annotated long non-coding RNA (lncRNA) from de novo active regions are shown. Two biological replicates were carried out in the 3 cell lines for the RNA-sequencing analysis. (B) Gene expression of SMILO and the former targets of JQ1 MYC, IRF4, POU2AF1, DUSP22 after JQ1 treatment in MM cell lines. ND: not detected. (C) Validation by real-time quantitative polymerase chain reaction (RT-qPCR) of SMILO and MYC expression after JQ1 in the 3 MM cell lines used before (experiment performed in 3 biological replicates); (SMILO forward [Fw] primer CACCTTGGGTCAGATGTGTG and reverse [Rv] primer TCGAGGTCAAGCAGAAAAGAA; MYC Fw primer TGCTCCATGAGGAGACACC and Rv primer CTCTGACCTTTTGCCAGGAG; and the house-keeping gene GUSβ Fw primer GAAAATATGTGGTTGGAGAGCTCATT and Rv primer CCGAGTGAAGATCCCCTTTTA). (D) Percentage of genes associated with de novo active regions presenting a decrease in chromatin acetylation (H3K27ac chromatin immunoprecipitation-sequencing [ChIP-seq] and accessibility (assay for transposase-accessible chromatin using sequencing [ATAC-seq]) after JQ1 treatment in MM cell lines. One replicate of the ChIP-seq for the H3K27ac and ATAC-seq upon JQ1 treatment experiments was carried out in each cell line. The chromatin regions downregulated for both H3K27ac and chromatin accessibility were determined as those presenting a log2FC<-0.5 in 2 cell lines, having the same tendency in the third cell line (log2FC<0). (E) ChIP-seq for H3K27ac and ATACseq of SMILO chromatin region without (C-) and after JQ1 treatment in MM cell lines. RNA sequencing, ChIP-seq for the H3K27ac and the ATAC-seq were conducted as previously described.⁵ The normal distribution of the experimental data was assessed though Shapiro-Wilk test and the statistical differences were determined by one-tailed parametric t test or one-tailed non-parametric Mann-Whitney test (GraphPad Prism 8.0.1). The significance levels were denoted as follows: *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. NB: naïve B cells; GCBC: germinal center B cells; MBC: memory B cells; PC: plasma cell; MM: multiple myeloma.

In order to determine the mechanism of FLI1-mediated downregulation of SMILO, we analyzed ChIP-seq data after treatment with JQ1. We observed a significant decrease in H3K27ac at the promoter region of FLI1 (Figure 3A). Additionally, expression of protein levels of FLI1 were significantly decreased after JQ1 treatment in MM cell lines (Figure 3B), as previously described in acute myeloid leukemia (AML).¹⁴ To support the hypothesis that FLI1 is causally involved in the expression of SMILO, FLI1 was silenced using CRISPR/Cas9 technology as previously explained.⁵ The inhibition of FLI1 by two guide RNA led to a substantial reduction in protein levels after 7 days (Figure 3C) that was correlated with a significant reduction of the expression of SMILO further exacerbated by day 10 (Figure 3D). Given that JQ1 treatment induces a greater reduction in SMILO expression (Figure 1C) compared to the direct inhibition of FLI1 (Figure 3D), it is plausible that JQ1 may regulate SMILO through other mechanisms. All these results suggest a model where JQ1 leads to a substantial drop of SMILO expression, in part through the downregulation of the TF FLI1, displacing it from the chromatin of the lncRNA SMILO (Figure 3E). Altogether, our data posit FLI1 as one of the TF that directly regulate SMILO expression in MM cells and unlocks the opportunity of modulating an essential lncRNA through the inhibition of its associated TF. This is particularly promising provided the therapeutic potential of targeting FLI1.^11,15

In conclusion, we have demonstrated the potential of BETi to modify gene expression through regulation of de novo active chromatin regions and highlight the possibility of modulating the expression of lncRNA by affecting the TF that govern their regulation. These findings support the development of novel therapeutic strategies against RNA with a broad application in the treatment of MM and in general of cancer.

Figure 2.FLI1 bound to the active chromatin regions of SMILO. (A) Reverse chromatin immuoprecipitation (Reverse-ChIP) results represented as a Venn diagram for each assay for transposase-accessible chromatin using sequencing (ATAC) peak, resulting in 16 candidate proteins in common to KMS-11 and MM.1R cell lines. (B) Gene expression of the 4 candidate genes, FLI1, SUB1, PKP2 and MAFG, in different subpopulations of healthy B cells and multiple myeloma (MM) samples. (C) Expression correlation of FLI1 with SMILO, using CoMMpass (IA18) cohort from MMRF (Multiple Myeloma Research Foundation). NB: naïve B cells; CB: centroblast B cells; CC: centrocyte B cells; MBC: memory B cells; TPC: tonsillar plasma cells; BMPC: bone marrow plasma cells; MM: multiple myeloma; FPKM: fragments per kilobase million.

Figure 3.SMILO expression is regulated in part by the transcription factor FLI1. (A) Chromatin states of FLI1 gene in multiple myeloma (MM) samples and different subpopulations of healthy B cells together with the chromatin acetylation analyzed by chromatin immunoprecipitation sequencing (ChIP-seq) for the H3K27ac in 2 MM cell lines (KMS-11 and RPMI 8226) without (C-) and with JQ1 treatment. (B) Quantification of the western blot of FLI1 after JQ1 treatment in the MM cell lines used for the Reverse-ChIP. (C) Western blot of FLI1 in KMS-11 and MM.1R after FLI1 knockdown by CRISPR/CAS9 (used primers to clone the guide RNA in the CRISPseq-BFP-backbone [AddGene #85707] vector: FLI1-g1 Fw primer CACCGCACAGGTCCTCCCCTTGGAG and Rv primer AAACCTCCAAGGGGAGGACCTGTGC; FLI1-g2 Fw primer CACCGGACGATGACTCTCCTCTCGT and Rv primer AAACACGAGAGGAGAGTCATCGTCC). (D) Expression of SMILO in KMS-11 and MM.1R cell lines after FLI1 knockdown by CRISPR/CAS9. (E) Graphical representation of the regulation mechanism of SMILO by JQ1 treatment: JQ1 degrades the FLI1 protein impairing its union to the chromatin of SMILO, giving rise to the downregulation of this essential long non-coding RNA (lncRNA) for MM cells. The used antibodies and conditions for western blots are: FLI1 (35980 Cell Signaling, 1:1000) and as loading control β-Actin (A5441 Sigma-Aldrich, 1:4000). The normal distribution of the experimental data was assessed though Shapiro-Wilk test and the statistical differences were determined by one-tailed parametric t test or one-tailed non-parametric Mann-Whitney test (GraphPad Prism 8.0.1). The significance levels were denoted as follows: *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. NB: naïve B cells; GCBC: germinal center B cells; MBC: memory B cells; PC: plasma cells; Chr.: chromatin.

Footnotes

• Received May 29, 2024
• Accepted September 16, 2024

Correspondence

^*NG-E and ESJ-E contributed equally as first authors.

^#FP and XA contributed equally as senior authors

F. PROSPER - fprosper@unav.es

X. AGIRRE - xaguirre@unav.es

Disclosures

No conflicts of interest to disclose.