Dissecting the impact of bromodomain inhibitors on the Interferon Regulatory Factor 4‐MYC oncogenic axis in multiple myeloma

Abstract B‐cell progenitor fate determinant interferon regulatory factor 4 (IRF4) exerts key roles in the pathogenesis and progression of multiple myeloma (MM), a currently incurable plasma cell malignancy. Aberrant expression of IRF4 and the establishment of a positive auto‐regulatory loop with oncogene MYC, drives a MM specific gene‐expression program leading to the abnormal expansion of malignant immature plasma cells. Targeting the IRF4‐MYC oncogenic loop has the potential to provide a selective and effective therapy for MM. Here we evaluate the use of bromodomain inhibitors to target the IRF4‐MYC axis through combined inhibition of their known epigenetic regulators, BRD4 and CBP/EP300. Although all inhibitors induced cell death, we found no synergistic effect of targeting both of these regulators on the viability of MM cell‐lines. Importantly, for all inhibitors over a time period up to 72 h, we detected reduced IRF4 mRNA, but a limited decrease in IRF4 protein expression or mRNA levels of downstream target genes. This indicates that inhibitor‐induced loss of cell viability is not mediated through reduced IRF4 protein expression, as previously proposed. Further analysis revealed a long half‐life of IRF4 protein in MM cells. In support of our experimental observations, gene network modeling of MM suggests that bromodomain inhibition is exerted primarily through MYC and not IRF4. These findings suggest that despite the autofeedback positive regulatory loop between IRF4 and MYC, bromodomain inhibitors are not effective at targeting IRF4 in MM and that novel therapeutic strategies should focus on the direct inhibition or degradation of IRF4.


| INTRODUCTION
Transcription factor interferon regulatory factor 4 (IRF4) is a key activator of lymphocyte development, affinity maturation and terminal differentiation into immunoglobulin-secreting plasma cells. 1,2 Faulty regulation of IRF4 expression is associated with numerous lymphoid malignancies, including multiple myeloma (MM), an aggressive and incurable hematologic cancer characterized by the abnormal proliferation of bone marrow plasma cells. 2,3 At the molecular level MM is an heterogenous disease with several subgroups defined by specific gene-expression profiles and recurrent chromosomal rearrangements. In a minority of MM cases, chromosomal translocation t(6; 14) (p25; q32) brings the IRF4 gene under the control of immunoglobulin heavy-chain regulatory regions. 4,5 Interestingly while IRF4 is not always genetically altered in MM, 6 its expression levels are always higher than in plasma cells. 7 Overexpression of IRF4 leads to an aberrant gene-expression program and to the mis-regulated transcription of a wide network of target genes. Interferon regulatory factor 4 loss-of-function in RNAinterference-based experiments have shown that MM cells are "addicted" to this abnormal gene-expression program since reduced IRF4 expression causes rapid and extended non-apoptotic cell death, irrespective of genetic etiology. 6 Similarly, targeting the 3 0 UTR of IRF4 mRNA for degradation by overexpression of miR-125-b, leads to MM cell death. 8 MM accounts for 2% of all cancers and 10% of all hematologic malignancies. 9 In the UK around 5800 MM cases are diagnosed every year (2015-2017) and incidence rates are projected to rise by 11% by 2035. The past decade has seen a revolution in the management of MM with the availability of novel therapies which are both more effective and less toxic. Despite the ensuing improvement of clinical outcomes, nearly every patient becomes refractory to therapies and overall 5-year survival rates are 52%. 10 Considering that existing treatments are not curative, there is a need for new therapeutic approaches. Targeting IRF4 has potential to be a powerful therapeutic strategy in MM. Firstly, IRF4 inhibition likely presents manageable side effects as phenotypes in IRF4-deficient mice are restricted to lymphoid and myeloid lineages and mice lacking one allele of IRF4 are phenotypically normal. 6 Additionally, MM cells' "addiction" to IRF4 renders them fairly sensitive to even small decreases in IRF4 levels leading to cell death. Finally, IRF4 inhibition is lethal to all MM cells regardless of their underlying transforming oncogenic mechanism. 6 An attractive approach to inhibit IRF4 might be targeting a known regulator of IRF4 expression in MM, MYC. Constitutive activation of MYC signaling is detected in more than 60% of patient-derived cells and one of the most common somatic genomic aberrations in MM is rearrangement or translocation of MYC. 11 MYC transactivates IRF4 by binding to a conserved intronic region whilst IRF4 binds to the MYC promoter region in MM cells and transactivates its expression, creating a positive autoregulatory feedback loop. 6 The expression of MYC in MM cells is abnormal since normal plasma cells do not express MYC as a result of repression by PR domain zinc finger protein 1 (PRDM1). 12 Moreover, IRF4 binds to its own promoter region, creating a second positive autoregulatory loop which would potentiate any therapeutic effect of targeting the MYC-IRF4 loop. 6 The IRF4-MYC axis is thus considered to be a promising therapeutic target in MM, however the complex regulatory feedbacks make predictable targeting of this axis challenging.
One way to target the IRF4-MYC axis is through upstream epigenetic regulators. Bromodomain and extra-terminal (BET) proteins inhibitors have emerged as potential therapeutic agents for the treatment of hematologic malignancies. 13 BET protein BRD4 is specifically enriched at immunoglobulin heavy chain (IgH) enhancers in MM cells bearing IgH rearrangement at the MYC locus, causing their aberrant proliferation. 14 BET inhibitors such as JQ1, which displace BRD4 from chromatin by competitively binding to its bromodomain acetyl-lysine recognition pocket, trigger inhibition of MYC transcription. 14,15 CREB binding protein (CBP) and EP300 are bromodomaincontaining histone acetyltransferases. 16

| Cell viability assay
Cell viability assay and statistical analysis were performed as described in the supplemental methods. In brief, cell viability after inhibitors treatment was assessed using CellTiter-Blue ® Cell Viability Assay. Each experiment was reproduced 3 times per cell line.

| Quantitative real time PCR
RNA extraction, cDNA synthesis, and quantitative real time PCR was performed as in the supplemental methods.

| Protein half-life
To measure protein half-life, cells were treated with 10 μg/ml cycloheximide for up to 72 h followed by western blotting. Detailed protocols are available in the supplemental methods.

| Gene and protein network modeling
Computational models were constructed using Ordinary Differential Equations and solved using MATLAB 2020a and ode15 s. All code, equations and parameters used in modeling are available on Github (https://github.com/SiFTW/MMModel/). Regulated reactions were modeled as described previously. 18 Detailed methods are available in the supplemental methods.  19 To test this hypothesis, we compared the effect of ISOX-DUAL treatment with a combination of JQ1+SGC-CBP30 ( Figure 1E). We found that the combination treatment had a stronger inhibitory effect on cell viability than ISOX-DUAL, with an IC 50 comparable with that of JQ1 alone. Similar results were obtained when treating the cells for 72 h ( Figure S1).

| Concomitant BRD4 and CBP/EP300 inhibition does not have a synergistic effect on MM cell viability
Taken together, our results demonstrate that ISOX-DUAL offers no advantage to treatment with a BET inhibitor alone and that combining JQ1 and SGC-CBP30 does not lead to synergistic or antagonistic cytotoxic effects.

| Bromodomain inhibitors impact IRF4 mRNA but not protein expression in MM cell-lines
We next investigated the effects of bromodomain inhibitors on the mRNA and protein expression levels of IRF4 and MYC. We treated the cells with a concentration of drugs at their IC 50 value (as in  Figure S3), although the mean reduction for MYC was more pronounced than that for IRF4. In summary, our data show that bromodomain inhibitors effectively reduce MYC and IRF4 mRNA levels and MYC protein levels, but do not show a corresponding effect on IRF4 protein levels.

| Bromodomain inhibitors affect the geneexpression levels of target genes of MYC but not IRF4
As protein levels of MYC and IRF4 were unequally affected by drug treatment, we hypothesized that expression of their downstream target genes would also be differentially affected. To test this hy-   Figure S5).
In summary, these results confirm our hypothesis that MYC, but not IRF4 downstream target genes are substantially downregulated as a result of bromodomain inhibition.

| Gene and protein network modeling are consistent with a long IRF4 protein half-life
Given the known feedback loop between MYC and IRF4 in MM cells we asked whether the reduction in IRF4 mRNA, but not protein expression could be explained by the stability of IRF4 protein.
To test this hypothesis and to assess whether the protein and

| Gene and protein network modeling suggest that bromodomain inhibitors effects on MM cell-lines are mainly exerted through MYC transcription repression and not IRF4
The initial computational modeling of the predicted drug response on   more effective if re-focused on direct inhibition or degradation of IRF4, which could be then used in synergistic combination to address relapsed or refractory cases of MM for which presently limited choices exist.

ACKNOWLEDGMENTS
We acknowledge the general laboratory support of Dr Sarah Con-