Activating RARβ to overcome immunomodulatory drug resistance in t(4;14) human myeloma

Abstract

The t(4;14) translocation is a high-risk cytogenetic abnormality in multiple myeloma (MM) that results in overexpression of fibroblast growth factor receptor 3 (FGFR3) and enhanced MM proliferation, leading to poor prognosis. Herein, we carried out a high-throughput screen on 1855 Food and Drug Administration (FDA)-approved pharmaceuticals and identified all-trans retinoic acid (ATRA), which alone has no anti-MM effect, as a potent drug that enhances the cytotoxic effects of immunomodulatory drugs (IMiDs) in t(4;14) MM cells. Mechanistically, ATRA activates retinoic acid receptor β (RARβ), which then binds to retinoic acid response elements in the FGFR3 promoter. IMiDs enhanced nuclear translocation of histone deacetylase (HDAC)-5 and 3 by reducing HDAC5 Ser498 phosphorylation levels. RARβ, HDAC5 and HDAC3 formed a co-repressor complex that reduced chromatin accessibility and H3K27 acetylation in FGFR3 promoter, FGFR3 expression, and suppressed phosphoinositide 3-kinase/AKT signaling pathways, leading to more MM cell death. Similarly, CD2314, a selective RARβ agonist, sensitized and resensitized t(4;14) MM cells to IMiDs in vitro and in vivo. Thus, these findings underscore the therapeutic potential of ATRA and RARβ agonists in enhancing the efficacy of IMiD-based treatments for t(4;14) MM and offer a promising strategy to overcome IMiD resistance and improve outcomes in this high-risk subgroup.

Introduction

Multiple myeloma (MM) is the second most common hematological malignancy with a steadily rising global incidence [1]. Over the past two decades, the introduction of novel therapeutic strategies has significantly extended the survival of MM patients. However, high-risk MM continues to show limited benefit from current treatment approaches [2, 3]. Since its recognition in major clinical guidelines, including the International Myeloma Working Group (IMWG) guidelines (2016), the Mayo mSMART 3.0 (2018), the R2-ISS by the European Myeloma Network (2022), and the National Comprehensive Cancer Network guidelines V1 version (2024), the t(4;14) translocation has been consistently identified as a high-risk cytogenetic abnormality and an independent poor prognostic factor [4, 5]. The t(4;14) translocation is one of the most frequent translocations in MM, occurring as a primary clonal event in 11% to 15% of newly diagnosed MM patients [2]. The t(4;14)(p16;q32) translocation results in the relocation of fibroblast growth factor receptor 3 (FGFR3) under the control of the IgH super-enhancer elements, leading to overexpression of FGFR3 [6, 7]. FGFR3, a tyrosine kinase receptor, plays a crucial role in various cell signaling pathways, including RAS/mitogen-activated protein kinase (MAPK), phosphoinositide 3-kinase (PI3K)/AKT, and Janus kinase/Signal transducer and activator of transcription (JAK/STAT), which are essential for regulating cell growth, survival, and differentiation [6]. Currently, small-molecule FGFR3 inhibitors, such as erdafitinib, and monoclonal antibodies targeting FGFR3 are being evaluated in clinical trials. However, their efficacy remains unclear. Additionally, resistance to these therapies, often due to mutations in the FGFR3 target, presents a significant challenge [8–10]. Therefore, there remains an unmet medical need for patients with t(4;14) MM, and there is a pressing demand for the development of novel, safe, and cost-effective therapeutic agents that can sensitize or even resensitize t(4;14) MM cells to treatment.

Immunomodulatory drugs (IMiDs) such as thalidomide, lenalidomide (LEN), and pomalidomide (POM) are a cornerstone of treatment for patients with MM. However, patients become resistant to ongoing therapy over time and inevitably relapse. IMiD resistance can originate from genetic alterations of cereblon (CRBN), including mutations, copy number changes and differential splicing. These alternate to the E2 ubiquitin conjugating enzymes and the COP9 signalosome, and transcriptional plasticity [11, 12]. To address this challenge, combination therapies involving agents with different mechanisms of action are often employed to overcome IMiD resistance.

To overcome IMiD resistance in human MM, we performed a high-throughput screening of 1855 Food and Drug Administration (FDA)-approved drugs [13] to identify repurposed agents capable of enhancing or sensitizing MM cells to IMiDs. Interestingly, we observed that all-trans retinoic acid (ATRA), which is a vitamin A derivative and is primarily used to treat acute promyelocytic leukemia (APL) [14–16], significantly enhanced the efficacy of IMiDs in t(4;14) MM cells and (re)sensitized IMiD-resistant t(4;14) MM cells to these agents. To elucidate the underlying mechanisms and explore the potential of ATRA in sensitizing t(4;14) MM cells to IMiD-based regimens, we conducted a series of in vitro and in vivo studies using human MM cell lines, primary MM cells, and murine models. Our study underscores the potential of ATRA as a novel therapeutic agent to improve treatment outcomes in t(4;14) MM patients.

Materials and methods

Primary MM cells and cell lines

This study was approved by the Houston Methodist Research Institute, Houston Methodist Hospital, and all patients provided written informed consent. Bone marrow (BM) aspirates were obtained from newly diagnosed t(4;14) MM patients as part of their routine clinical work-up. The detailed characteristics of patients were shown in Supplemental Table 1. BM CD138⁺ MM cells were isolated by EasySep™ Human Whole Blood and CD138 Positive Selection Kit II (STEMCELL Technologies, Vancouver, British Columbia, Canada). MM cell line OPM-2 was kindly provided by Dr Keith Stewart [17]. KMS-26 was purchased from JCRB Cell Bank (JCRB 1187, Ibaraki-shi, Osaka, Japan). POM-resistant OPM-2 and POM-resistant H929 were established according to the previous report [18]. Other cell lines were purchased from the American Type Culture Collection (Rockville, MD, United States). STR profiles of JJN3, ARP1, KMS-12-PE, MOLP-8, U266, RPMI 8226, MM.1S, MM.1 R, OPM-2, and H929 were performed by the Cytogenetics and Cell Authentication Core of MD Anderson Cancer Center. All MM cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA, United States).

MM mouse models

NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice were purchased from the Jackson Laboratory (Bar harbor, ME, United States). All mice were maintained in American Association of Laboratory Animal Care-accredited facilities, and the studies were approved by the Institutional Animal Care and Use Committee of the Houston Methodist Research Institute. Our study examined male and female animals aged 8–10 weeks, and similar findings were reported for both sexes.

Statistical analysis

Data were presented as mean ± standard deviation (SD). Student’s two-tailed t-test was used for comparisons between two groups. A P value < 0.05 was considered statistically significant. Statistical analyses were performed using GraphPad Prism 8 software. For analyses involving more than two groups, one-way analysis of variance (ANOVA) was used for multiple comparisons, followed by Bonferroni correction to control for type I error. 95% confidence intervals were calculated as estimates of effect size for all key quantitative comparisons. For survival analyses, survival curves were estimated using the Kaplan-Meier method and compared using the log-rank test.

For high-throughput sequencing data including RNA-seq and ATAC-seq, the false discovery rate (FDR) was controlled using the Benjamini-Hochberg procedure to correct for multiple testing. Significantly differentially expressed genes or differentially accessible regions were defined using a threshold of FDR < 0.05.

Combination index (CI) values were calculated based on the cell viability data of MM cell lines treated with DMSO, ATRA, POM, and POM (1 µM) with a range of ATRA (1 µM, 2 µM, 4 µM, or 8 µM) according to the Chou-Talalay method [19] to evaluate drug synergy. CI < 1 indicates synergism, CI = 1 indicates an additive effect, and CI > 1 indicates antagonism. CI was calculated using the CompuSyn software. Each point represents the CI value plotted against the corresponding mean fraction affected (Fa) by the combination treatment relative to DMSO control in 3 independent experiments.

Results

ATRA enhances IMiD direct killing effect on MM cells with t(4;14) translocation

To identify potential candidates that can promote the cytotoxicity of POM on human MM cells, we performed a high-throughput screen of 1855 FDA-approved drugs using a cell viability assay on OPM-2 cells (Supplementary Fig. S1A, B). ATRA emerged as one of the top candidate drugs that elicited a remarkable inhibitory effect on cell viability of POM-treated OPM-2 cells compared to POM or ATRA alone (Supplementary Fig. S1C). To confirm the results, we tested a panel of 11 human MM cell lines and found that, interestingly, ATRA, while having no direct effect on MM cells, significantly enhanced the tumor cell autonomous pro-apoptotic effect of POM on OPM-2, KMS-26, and H929 cells. No such effect was observed in the other 8 cell lines (Fig. 1A, B). As these three MM cell lines harbor the t(4;14) translocation, this result indicates that ATRA may be able to enhance the therapeutic effect of IMiDs on t(4;14) translocated human MM cells. To confirm the result, we performed MTS assays and showed that ATRA significantly enhanced the effect of POM in suppressing the growth of MM cells with, but not without t(4;14), and the effect could be observed with POM at a low concentration of 1 µM (Supplementary Fig. S1D). The combination indices for ATRA and POM were less than 1 in the t(4;14) MM cell lines [19], indicating a synergistic effect with the two agents (Fig. 1C). ATRA not only enhanced POM-induced apoptosis in t(4;14)-positive MM cell lines but also in primary MM cells derived from patients harboring the t(4;14) translocation (Fig. 1D). Moreover, ATRA enhanced POM-induced cell death in POM-sensitive t(4;14) MM cells and restored, at least partially, POM intrinsic sensitivity to t(4;14) MM cells resistant to POM (Fig. 1E, Supplementary Fig. S1E–G). Furthermore, ATRA significantly enhanced LEN-induced cell death (Supplementary Fig. S2A, B) and growth inhibition (Supplementary Fig. S2C) in t(4;14) positive, but not negative, MM cells. Again, the anti-MM effect of ATRA and LEN was synergistic as the combination index for ATRA and LEN was less than 1 (Supplementary Fig. S2D). Similarly, ATRA and LEN displayed synergistic killing effects on primary MM cells from t(4;14) translocated patients (Supplementary Fig. S2E, F). Taken together, these findings suggest that ATRA may be a promising therapeutic agent to sensitize and re-sensitize t(4;14)-positive MM cells to IMiD treatment.

Fig. 1. ATRA enhances IMiD direct killing effect on t(4;14)-positive MM cells.

A Representative flow histograms showing apoptosis of OPM-2 and JJN3 MM cells and (B) summarized results of three repeated experiments of apoptosis of t(4;14)-positive and -negative MM cell lines treated with DMSO (dimethyl sulfoxide), ATRA (all-trans retinoic acid; 1 µM) and/or POM (pomalidomide; 1 µM) for 72 h. C Combination index (CI) plot of cell viability showing a synergistic effect of POM at 1 µM with a range of ATRA (1 µM, 2 µM, 4 µM, or 8 µM) on t(4;14)-positive MM cell lines. CI was calculated using the CompuSyn software. Each point represents the CI value plotted against the corresponding mean fraction affected (Fa) by the combination treatment relative to DMSO control in 3 independent experiments. CI < 1 represents synergism, CI = 1 represents additive effect, and CI > 1 represents antagonism. D Summarized results of apoptosis in primary MM cells from t(4;14)-positive and -negative patients (n = 5 for each group). E Representative flow histogram showing the induction of apoptosis in POM-sensitive (POM-S) and POM-resistant (POM-R) OPM-2 MM cell line in culture with DMSO, 1 µM POM and/or 1 µM ATRA for 72 h. Data were shown as mean ± SD. Student’s two-tailed t-test or one-way ANOVA with Bonferroni correction was used as indicated. 95% confidence intervals were calculated for effect size estimates. ns, no significance, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

ATRA enhances the therapeutic effects of IMiDs on t(4;14) MM cells by activating RARβ

Given that ATRA is a pan-RAR agonist [20], we investigated which RAR isoforms are critical for its ability to enhance IMiD-induced apoptosis in t(4;14)-positive MM cells. First, we evaluated RAR expression profiles in MM cell lines. RARα, RARβ, and RARγ were detectable in all tested cell lines (Supplementary Fig. S3A). Second, by employing selective RAR agonists, we observed that CD2314, a selective RARβ agonist, significantly enhanced POM-induced apoptosis in t(4;14)-positive MM cells, while RARα (AM80) or RARγ (BMS961) agonists showed no such effect (Fig. 2A, B and Supplementary Fig. S3B). Combinations of CD2314 and POM also significantly reduced the proliferation of t(4;14)-positive (Fig. 2C). To confirm the results, we generated RARβ-knockout (KO) t(4;14)-positive MM cell lines using CRISPR/Cas9 (Fig. 2D). RARβ-KO MM cells showed a marked decrease in apoptosis when treated with ATRA plus POM compared to that of control-KO MM cells (Fig. 2E, F). The synergistic cytotoxic effect of ATRA-POM combination on OPM2 and H929 cells remained unchanged in RARα-KO and RARγ-KO cells (Supplementary Fig. S3C-F).

Fig. 2. ATRA enhances the therapeutic effects of IMiDs on t(4;14) MM cells by activating RARβ.

A Representative flow histogram and (B) summarized results of three repeated experiments showing apoptosis and (C) cell viability of t(4;14)-positive MM cells in culture with DMSO, 1 µM CD2314 and/or 1 µM POM for 72 h. D Western blot showing protein expression of RARβ in RARβ knockout (RARβ-KO) or control (Ctrl-KO) t(4;14) MM cell lines. E Summarized results of three repeated experiments and (F) representative flow histogram showing apoptosis of RARβ-KO or Ctrl-KO MM cells in culture with DMSO, 1 µM CD2314 and/or 1 µM POM for 72 h. Data were shown as mean ± SD. Student’s two-tailed t-test or one-way ANOVA with Bonferroni correction was used as indicated. ns, no significance, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Unlike t(4;14) MM cells, CD2314 and POM failed to demonstrate synergy in t(4;14)-negative MM cells (Supplementary Fig. S3G, H). Following RARβ KO, the apoptotic rate of the negative cells showed no significant changes with either POM monotherapy or the combination treatment (Supplementary Fig. S3I, J). Thus, these results underscore the importance of RARβ activation in ATRA-induced sensitization of t(4;14) MM cells to IMiD treatment.

ATRA and IMiDs significantly downregulate the expression of FGFR3 in t(4;14) MM cells

We elucidated the molecular mechanism underlying the synergistic anti-MM effects of ATRA and IMiDs by focusing on oncoproteins affected by t(4;14) in MM cells. We observed that ATRA and IMiDs significantly reduced both the mRNA and protein expression levels of FGFR3 in t(4;14) positive (Fig. 3A–C and Supplementary Fig. S4A). This effect was abolished in RARβ-KO OPM2 and H929 MM cells (Fig. 3D, E). Relative to POM monotherapy, the combined ATRA-POM treatment induced a significant downregulation of FGFR3 expression in both RARα-KO and RARγ-KO cells (Supplementary Fig. S4B, C). Interestingly, ATRA and POM also downregulated FGFR3 protein expression in POM-resistant MM cell lines (Fig. 3F). Since MMSET (multiple myeloma SET domain) expression, which is another key oncoprotein driven by t(4;14) translocation, was unaffected by ATRA and POM treatment (Supplementary Fig. S4D), we focused on ATRA and IMiD-induced FGFR3 suppression in this study.

Fig. 3. ATRA and IMiDs significantly downregulate the expression of FGFR3 in t(4;14) MM cells.

Expression of FGFR3 detected by (A) Western blot showing representative Western blots and (B) summarized results of three repeated experiments or (C) RT-PCR of three repeated experiments in MM cells after treatment with DMSO or 1 μM ATRA, POM or their combination for 72 h. Expression of FGFR3 in Ctrl-KO or RARβ-KO MM cells detected by (D) Western blot or (E) RT-PCR of three repeated experiments after treatment with DMSO or 1 μM ATRA, POM or their combination for 72 h. F Expression of FGFR3 in POM-resistant OPM-2 (OPM-2/POM) and H929 (H929/POM) MM cells after treatment with DMSO or 1 μM ATRA, POM or their combination for 72 h. Data were shown as mean ± SD. Student’s two-tailed t-test or one-way ANOVA with Bonferroni correction was used as indicated. ns, no significance, *P < 0.05.

In t(4;14)-negative MM cell lines, FGFR3 expression was extremely low, and the treatments exerted no effect on FGFR3 expression levels (Supplementary Fig. S5A, B). Following RARβ KO, subsequent administration of the treatments resulted in no change in FGFR3 expression (Supplementary Fig. S5C).

ATRA and IMiDs regulate chromatin accessibility and H3K27 acetylation of FGFR3 via HDACs in t(4;14) MM cells

To examine the mechanisms underlying ATRA and IMiD-induced transcriptional regulation of FGFR3 expression, we first analyzed the expression of CRBN, a direct target of IMiDs. ATRA and POM treatment did not change CRBN expression levels in t(4;14)-positive MM cells (Supplementary Fig. S6A), and CRBN knockdown (KD) did not prevent the combination treatment from significantly reducing FGFR3 expression (Supplementary Fig. S6B, C). These results suggest that CRBN may not be responsible for the observed synergy of the two drugs.

We then investigated how ATRA exerts its regulatory effects in the cells. ATRA regulates gene expression by binding to nuclear retinoic acid receptors (RARs), which subsequently interact with retinoic acid response elements (RAREs) [21]. RARE, a specific DNA sequence, canonically consists of two hexameric repeat sequences (AGGTCA) separated by a specific spacer (typically 1–5 bp), forming a direct repeat [21]. The presence of a conserved RARE in the FGFR3 promoter suggests potential regulation by retinoid signaling. ChIP-qPCR assay showed that CD2314 treatment promoted RARβ recruitment or binding to this element in t(4;14)-positive MM cells (Fig. 4A). RARβ orchestrates bidirectional transcriptional regulation by recruiting either repressive or activating complexes to modulate chromatin accessibility at target loci [21]. Therefore, we employed ATAC-seq to investigate whether the combination therapy alters chromatin accessibility of FGFR3. In ATAC-seq, ATRA and POM treatment exerted a significantly lower signal intensity of chromatin accessibility at the FGFR3 promoter region, including a 2.16-fold reduction at the FGFR3 locus (chr4:1,793,213-1,794,396, hg38) (Fig. 4B). As the histone modification H3K27ac (acetylation of lysine 27 on histone H3) serves as a key epigenetic marker that promotes chromatin accessibility [10], we performed H3K27ac ChIP-qPCR in t(4;14) MM cells. The analysis revealed that ATRA and POM treatment significantly reduced H3K27ac signals at FGFR3 promoter regions, indicating a transcriptional repression through chromatin silencing (Fig. 4C). In t(4;14)-negative MM cells, ChIP assays using anti-RARβ antibody, followed by qPCR amplification of the FGFR3 promoter region, revealed that ATRA combined with IMiDs did not significantly enhance the binding of RARβ to RARE within the FGFR3 promoter (Supplementary Fig. S7A). ATAC-seq analysis in t(4;14)-negative MM cells revealed that chromatin accessibility at the FGFR3 promoter remained low and showed no significant differences across the treatment groups (Supplementary Fig. S7B). Consistent with this finding, H3K27ac ChIP-qPCR targeting the FGFR3 promoter demonstrated that H3K27ac enrichment in the t(4;14)-negative cells was extremely low and did not differ significantly among the treatment conditions (Supplementary Fig. S7C).

Fig. 4. ATRA and IMiDs regulate chromatin accessibility and H3K27 acetylation of FGFR3 via HDACs in t(4;14) MM cells.

A Chromatin immunoprecipitation (ChIP) analysis showing RARβ recruitment or binding to the RARE (Retinoic Acid Response Element) motif of FGFR3 transcriptional start site (TSS) (TSS-1658 to TSS-1644) in t(4;14)-positive MM cells after treatment with DMSO or 1 μM CD2314, POM or their combination for 72 h. The upper panel showed the sequence logo of the identified RARE motif. The vertical axes (Bits) indicated the information content of the base frequency at that position, and the horizontal axes referred to the consensus site position. Lower panel showed summarized results of three repeated experiments in OPM-2, KMS-26, and H929 cells. B IGV snapshot of ATAC-seq signal tracks showing chromatin accessibility at the FGFR3 TSS in treatment groups and relative signal intensity in the FGFR3 promoter region in OPM-2 after treatment with DMSO or 1 μM ATRA, POM or their combination for 72 h. C ChIP-qPCR analysis of three repeated experiments showing H3K27ac (Histone H3 lysine 27 acetylation) enrichment at the FGFR3 promoter in t(4;14) MM cells after treatment with DMSO or 1 μM ATRA, POM or their combination for 72 h. ChIP-qPCR showing the binding of (D) HDAC3 or (E) HDAC5 to FGFR3’s promoter in t(4;14) MM cells after treatment with DMSO or 1 μM ATRA, POM or their combination for 72 h. Representative results of three independent experiments were shown. Data were shown as mean ± SD. Student’s two-tailed t-test or one-way ANOVA with Bonferroni correction was used as indicated. ns no significance, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

RARs can recruit HDAC to decrease chromatin accessibility, thus suppressing transcription [21]. RNA-seq analysis demonstrated that HDAC1, HDAC2, HDAC3, HDAC5, and HDAC7 were the predominant HDAC isoforms expressed by OPM-2 and H929 cells (Supplementary Fig. S8A). Among the five HDACs analyzed by ChIP-qPCR, only HDAC3 and HDAC5 were significantly enriched at the FGFR3 promoter in t(4;14) MM cells following combination drug treatment, implicating their specific involvement in the observed transcriptional regulation (Fig. 4D, E and Supplementary Fig. S8B). In t(4;14) negative MM cells, binding of HDAC3/HDAC5 to the FGFR3 promoter was extremely low and showed no significant inter-group differences, and no ATRA + POM-mediated enrichment effect was observed (Supplementary Fig. S8C). Compared with the Ctrl-KD cells, KD of HDAC3 (Supplementary Fig. S9A) not only blocked drug-induced apoptosis (Supplementary Fig. S9B) but also restored FGFR3 expression (Supplementary Fig. S9C) and H3K27ac occupancy (Supplementary Fig. S9D). Similar changes were also observed by knocking down HDAC5 (Supplementary Fig. S9E–H). Together, these findings revealed that the combination therapy utilizes HDAC3 and HDAC5 to exert its epigenetic and cytotoxic effects in t(4;14) MM cells.

ATRA and IMiDs promote nuclear translocation of HDAC5 and HDAC3 to form a complex with RARβ for suppressing FGFR3 transcription in t(4;14) MM cells

Further analysis revealed that neither single agent nor their combination treatment significantly affected the total intracellular levels of HDAC3 and HDAC5 (Supplementary Fig. S10A). HDAC3 and HDAC5 dynamically shuttle between nuclear and cytoplasmic compartments to mediate transcriptional repression through co-repressor complexes [22–24]. While cells treated with ATRA alone or untreated showed cytoplasmic retention, combined ATRA and POM treatment triggered robust nuclear accumulation of HDAC3 and HDAC5 in treated t(4;14) MM cells (Fig. 5A). Based on these findings, we speculated that the combination therapy promoted nuclear translocation of HDAC3 and HDAC5. HDAC5 targets HDAC3 to a specific nuclear location through its C-terminal deacetylase domain, and HDAC5 nuclear import is regulated by diminished Ser498 phosphorylation [25]. We therefore assessed this modification and found that combination therapy significantly lowered the phosphorylation level of Ser498 (Fig. 5B). ATRA combined with POM-induced nuclear translocation of HDAC3/HDAC5 and reduced the phosphorylation level of HDAC5 at Ser498 in t(4;14)-negative MM cells (Supplementary Fig. S10B–D). Notably, even after CRBN KO, POM remained capable of inducing HDAC3 and HDAC5 translocation into the nucleus and decreasing the phosphorylation level of HDAC5 (Supplementary Fig. S10E–G).

Fig. 5. ATRA and IMiDs promote nuclear translocation of HDAC5 and HDAC3 to form a complex with RARβ for suppressing FGFR3 transcription in t(4;14) MM cells.

A Western blot analysis showing HDAC5 and HDAC3 localization in nuclear and cytoplasmic fractions in the cells after culture with DMSO or 1 μM ATRA, POM or their combination for 72 h. Lamin B1 and β-tubulin served as nuclear and cytoplasmic markers, respectively. B Representative western blot (left panel) and summarized results of three independent experiments (right panel) showing changes in the expression levels of pHDAC5 (Ser498) and total HDAC5 in OPM-2 after culture with DMSO or 1 μM ATRA, POM or their combination for 72 h. C Microscale Thermophoresis (MST) analysis showing the interaction of POM and HDAC5. Wild-type HDAC5 Ser498 exhibited a higher binding affinity for POM (K_d = 0.77 μM, top panel) compared to the HDAC5 S498A mutant (K_d= 1.63 μM, lower panel). D Confocal microscopy images demonstrating the colocalization of RARβ (red) and HDAC3 (green) in the nuclei in OPM-2 after culture with DMSO or 1 μM ATRA, POM or their combination for 72 h. E Schematic structure of HDAC3 (top panel) and computational docking analysis of RARβ-HDAC3 interaction (lower panel). The nuclear export signal (NES), nuclear localization signal (NLS) domain, and deacetylation sites were indicated. Key interacting residues were identified (RARβ in purple and HDAC3 in green). The dashed line represented the hydrogen bonds with distances (Å) labeled. F Co-IP assay of HEK 293 T cells transfected with His-tagged HDAC3 truncations (1, 1–428; 2, 1–401; 3, 1–373; 4, 1-313; 5, 313–428; 6, 180–313; 7, 1–180; 8, 180–428; 9, 1–122; 10, 122–428; 11, 1–265) and Flag-tagged RARβ 24 h using anti-His beads for pulldown. Representative results of at least three independent experiments were shown. Data were shown as mean ± SD. Student’s two-tailed t-test or one-way ANOVA with Bonferroni correction was used as indicated. *P < 0.05.

By mutating Ser498 to Alanine (S498A) and performing MST binding assays with POM, we found that the wild-type HDAC5 exhibited significantly higher binding affinity for POM (Kd = 0.77 μM) compared to the S498A mutant (Kd = 1.63 μM), suggesting that POM interacts with Ser498 to promote HDAC5 nuclear translocation (Fig. 5C). To further validate the essential role of the Ser498 site in mediating POM binding and the downstream functional response of HDAC5, we conducted genetic KO and rescue experiments in t(4;14) MM cells. We first knocked out HDAC5 in t(4;14) MM cells (OPM2, Supplementary Fig. S11A). To validate Ser498 as the specific POM-binding site, rescue assays were performed in HDAC5-KO cells. Reconstitution with the HDAC5 S498A mutant abrogated phosphorylation at this residue, resulting in undetectable or significantly reduced p-Ser498 levels across all four treatment conditions (Supplementary Fig. S11B). HDAC5 S498A mutant was also exclusively enriched in the nucleus without cytoplasmic translocation in MM cells regardless of ATRA or POM treatment (Supplementary Fig. S11C). As a result of the constitutive dephosphorylation of HDAC5 conferred by the S498A mutation, ATRA alone or in combination with POM led to further downregulation of FGFR3, suggesting that the contribution of POM was largely attenuated in the presence of this HDAC5 mutant (Supplementary Fig. S11D). By contrast, wild-type HDAC5 rescued cellular responsiveness to the combination. POM or dual treatment decreased HDAC5 Ser498 phosphorylation, promoted HDAC5 nuclear translocation, and suppressed FGFR3 expression in t(4;14) MM cells (Supplementary Fig. S11B–D).

ATRA and POM treatment not only promoted nuclear accumulation of HDACs but also significantly enhanced their co-localization with RARβ in the nuclei (Fig. 5D). To investigate protein-protein interactions, we conducted a reciprocal co-immunoprecipitation (Co-IP) assay in HEK 293 T cells transiently expressing His-tagged RARβ or HDAC3 along with Flag-tagged HDAC5. These experiments demonstrated that HDAC3 interacted with both RARβ (Supplementary Fig. S12A, B) and HDAC5 (Supplementary Fig. S12B, C). Importantly, we confirmed these associations at endogenous levels following ATRA + POM treatment (Supplementary Fig. S12D, E), while no such interactions were detected in untreated cells (Supplementary Fig. S12F). Using the schematic structure of HDAC3 (Fig. 5E, top panel), computational docking was employed to predict the binding positions. Hydrogen bonds and salt bridges were identified as the key intermolecular forces driving these interactions. Three major interacting domains were detected between RARβ (purple) and HDAC3 (green) (Fig. 5E, bottom panel). Guided by these findings, a series of truncations were designed (Supplementary Fig. S12G, H). Subsequent analysis of these truncations pinpointed residues 313–428 of HDAC3 as the critical binding site for RARβ (Fig. 5F).

ATRA and IMiDs repress FGFR3 expression through inhibiting the PI3K-AKT pathway to sensitize t(4;14) MM cells to IMiD treatment

After demonstrating that the combination therapy epigenetically silenced FGFR3 expression, we explored the mechanistic link between FGFR3 downregulation and enhanced cell apoptosis. ChIP-seq analysis of differential peaks (DMSO vs ATRA + POM) revealed significant enrichment and suppression of the PI3K-AKT pathway (Fig. 6A) in MM cells treated with ATRA plus POM. Consistent with this result, KEGG pathway analysis of treated t(4;14) MM cell’s RNA-seq data also showed a pronounced downregulation of PI3K-AKT signaling by combinational treatments compared to ATRA treatment alone (Fig. 6B). Further validation via RNA-seq analysis demonstrated that ATRA + POM treatment markedly repressed the expressions of key PI3K-AKT pathway components (Fig. 6C). Western blot analysis confirmed that combined ATRA + POM treatment substantially attenuated phosphorylation of PI3K (Tyr458/Tyr199) and AKT (Thr308) and reduced 3-phosphoinositide-dependent protein kinase-1 (PDK1) expression (Fig. 6D). Moreover, the combination treatment concomitantly increased the cleavage of poly(ADP-ribose) polymerase 1 (PARP1) and caspase-3 while reducing myeloid cell leukemia 1 (Mcl-1) protein levels (Fig. 6E), which are consistent with enhanced apoptotic signaling.

Fig. 6. ATRA and IMiDs repress FGFR3 expression through inhibiting the PI3K-AKT pathway to sensitize t(4;14) MM cells to IMiD treatment.

A ChIP-seq analysis identifying pathway enrichment patterns associated with differential transcription factor-binding peaks under the combination treatment. The rich factor (x-axis) was the ratio of the number of differentially bound genes mapped to a given pathway to the total number of genes annotated in that pathway. Bubble size represented the GeneRatio, corresponding to the proportion of differentially bound genes assigned to each enriched pathway. Bubble color indicated the enrichment P value. An arrow indicates the PI3K-AKT pathway. B KEGG pathway analysis of RNA-seq data highlighting the top significantly enriched pathways, with PI3K-AKT signaling as a key target (arrow) in OPM-2 after culture with ATRA and POM or ATRA alone for 72 h. Bar length represents enrichment significance as −log10P. C Heatmap depicting expression changes of PI3K-AKT pathway-related genes in OPM-2 after culture with DMSO, 1 μM ATRA, POM or their combination for 72 h. Western blot analysis showing (D) reduced phosphorylation of PI3K-AKT pathway markers in ATRA and POM-treated MM cells and (E) Apoptosis markers for enhanced cell death in t(4;14) MM cells after culture with DMSO, 1 μM ATRA or POM or their combination for 72 h.

ATRA and RARβ agonists sensitize t(4;14) MM cells to IMiD treatment in vivo

We next assessed the in vivo therapeutic potential of ATRA or CD2314 in combination with POM using a luciferase-expressing OPM-2-luc xenograft NSG model. While neither ATRA nor CD2314 monotherapy showed therapeutic efficacy, ATRA or CD2314 strongly enhanced POM-mediated anti-MM activity. Combination treatment significantly reduced tumor burden measured by bioluminescence intensity (Fig. 7A, B) and the level of circulating human λ-chain (M-protein) (Fig. 7C), and extended mouse survival compared to single-agent treatment (Fig. 7D). Ex vivo analysis demonstrated that ATRA + POM or CD2314 + POM treatment significantly reduced CD138⁺ plasma cell populations and increased MM cell apoptosis (Fig. 7E, F and Supplementary Fig. S13). Notably, only combination therapy reduced FGFR3 expression in t(4;14)-positive MM cells. The phosphorylation level of HDAC5 Ser498 was significantly reduced in both POM monotherapy and ATRA + POM combination groups compared to the control groups, which directly confirms the physiological relevance of POM-mediated regulation of HDAC5 Ser498 phosphorylation in the physiologically relevant tumor (Fig. 7G). Finally, we determined the role of RARβ on the therapeutic effects of the combination therapy in vivo. RARβ-KO MM-bearing mice showed higher tumor burden and poor survival compared to Ctrl-KO MM-bearing mice treated with ATRA plus POM (Supplementary Fig. S14). Thus, these results clearly indicate that ATRA and RARβ agonists may be promising therapeutic agents to sensitize t(4;14) MM to IMiD treatment.

Fig. 7. ATRA and RARβ agonists sensitize t(4;14) MM cells to IMiD treatment in vivo.

Tumor burden measured as (A) in vivo bioluminescent imaging and (B) quantification of bioluminescence intensity based on average bioluminescence radiance (photons/s/cm²/sr), or (C) human immunoglobulin light chain in mouse plasma by ELISA, which was normalized to control, and (D) survival curve of NSG mice injected intravenously with 5 × 10⁶ OPM-2-luc MM cells and treated intravenously with DMSO, ATRA (15 mg/kg for every 3 days), CD2314 (15 mg/kg for every 3 days), POM (3 mg/kg for two consecutive days weekly), ATRA and POM, or CD2314 and POM. Summarized results of three independent experiments showing the percentage of (E) CD138⁺ and (F) apoptotic CD138⁺ OPM-2 MM cells sorted out from bone marrow of OPM-2-bearing NSG mice that received the indicated treatments. G Western blot showing the protein levels of FGFR3, pHDAC5, and total HDAC5 in CD138⁺ OPM-2 MM cells isolated from bone marrow of OPM-2-bearing NSG mice that received the indicated treatments. Data were shown as mean ± SD. Student’s two-tailed t-test or one-way ANOVA with Bonferroni correction was used as indicated. *P < 0.05, ***P < 0.001, ****P < 0.0001.

Discussion

In this study, we observed that ATRA enhanced the direct anti-MM activity of IMiDs in both t(4;14) primary MM cells and MM cell lines. Importantly, ATRA also restored the intrinsic sensitivity of POM-resistant t(4;14) MM cells to POM, supporting its role as a therapeutic sensitizer. Mechanistically, the ATRA-IMiD combination orchestrated an epigenetic switch by enhancing HDAC5/3 nuclear translocation triggered by HDAC5-Ser498 dephosphorylation and facilitated RARβ-dependent co-repressor assembly. This complex silenced FGFR3 via chromatin compaction and H3K27ac erosion and ablated PI3K-AKT survival signaling to overcome POM resistance in t(4;14) MM cells. We further showed that CD2314, a selective RARβ agonist, achieved a similar effect to ATRA in promoting POM-induced direct killing of t(4;14) MM cells in vitro and in vivo. These findings established RARβ activation induced by ATRA or CD2314 as a novel strategy to restore IMiD efficacy and reverse resistance in high-risk t(4;14) MM patients. Nevertheless, whether ATRA and RARβ agonists can also indirectly enhance the therapeutic efficacy of IMiDs in human MM by modulating the immune system remains to be elucidated.

The t(4;14) translocation is a chromosomal abnormality commonly observed in MM patients, accounting for about 15% of newly diagnosed patients. The t(4;14) MM subtype demonstrates primary resistance to conventional chemotherapy and exhibits significantly shorter median overall survival [10]. While these patients initially show moderate sensitivity to first-line therapies, they typically experience rapid relapse characterized by short duration of response, frequent development of extramedullary disease at relapse, renal insufficiency, and pancytopenia. Following relapse, disease progression becomes markedly more aggressive with substantially diminished sensitivity to chemotherapeutic agents [26–28]. The t(4;14) translocation drives oncogene (FGFR3 and MMSET) over-expressions that sustain MM cell proliferation and therapy resistance. FGFR3 represents a promising therapeutic target, but its inhibitors are still under investigation. Target mutation-induced off-target effects continue to pose a therapeutic obstacle [29, 30]. Therefore, we propose to repurpose ATRA, a low-cost and clinically validated agent, to combine with IMiDs to address this unmet need. Indeed, our results demonstrate that this combination suppresses t(4;14) MM cell growth and ablates FGFR3 expression, offering a mechanistically grounded, immediately testable treatment strategy.

IMiDs are critical components of therapeutic combinations for all stages of MM and are also used as a single-agent maintenance therapy after autologous stem cell transplantation [11]. At diagnosis, most MM patients are sensitive to IMiD-based combination therapy; however, about 5% of patients show primary resistance [11]. In addition, patients who are initially sensitive to IMiDs eventually acquire resistance, and IMiD-refractory patients are associated with shorter progression-free survival and overall survival in response to subsequent therapies [11]. IMiD activity against MM cells results from its ability to bind to a specific tri-tryptophan pocket of CRBN, a substrate receptor of the CRL4CRBN E3 ubiquitin ligase [31, 32]. The efficacy and resistance mechanisms of IMiDs involve multiple components of the CRBN pathway, including E2 ubiquitin conjugating enzymes and the COP9 signalosome [11, 33, 34]. Beyond canonical CRBN-dependent E3 ligase activity, alternative mechanisms such as CD147/MCT1 regulation [35] and HSP90 modulation [36] also contribute to IMiD intrinsic sensitivity. Additionally, CRBN-independent pathways involving TRAF2 [37], CBP/EP300 [17], CDK6 [38], transcriptional plasticity, and ETV4 [39] have been implicated in IMiD responses. CRBN genetic alterations (point mutations, copy number alterations, structural variants, and exon-10 splice variants) emerge in ~30% of relapsed or refractory MM and confer clinical resistance to IMiDs [40].

Interestingly, ATRA- and POM-exerted FGFR3 suppression was independent of CRBN status because CRBN expression levels were not modulated by the combination and their effect was not abrogated by CRBN KD, which revealed a bypass mechanism of IMiD resistance in MM cells. We discovered a new target site of POM-Ser498 on HDAC5. Mutation of this site markedly diminished POM’s binding affinity to HDAC5, consequently disrupting HDAC5’s nuclear entry and its role in reducing promoter region acetylation. Therefore, we uncover a novel epigenetic axis of IMiD activity, governing chromatin dynamics, that operates independently of CRBN. This paradigm-shifting finding not only expands the mechanistic landscape of IMiD response or resistance but also provides a direct clinical solution for CRBN-driven resistance. Notably, our study focused on the cell-autonomous epigenetic mechanisms underlying ATRA-IMiD synergy in t(4;14) MM, rather than the canonical immunomodulatory functions of IMiDs. As the defined mechanistic events are intrinsic to MM cells and independent of host immune function, the NSG xenograft mouse model was a rational choice for in vivo proof-of-concept study, which enabled an accurate assessment of direct anti-tumor efficacy without immune-mediated interference. Murine effective doses of CD2314 (15 mg/kg once every 3 days) and POM (3 mg/kg 2 consecutive days weekly) are clinically translatable and safe via body surface area conversion [41], yielding a recommended starting CD2314 dose of 8–10 mg/m² (once every 3 days) and a POM dose of 3 mg/day (2 consecutive days weekly). Both are achievable and lower than established toxic thresholds [41, 42]. However, CD2314, as a selective RARβ agonist, exhibits a toxicity profile typical of retinoid-class drugs [43], including dose-dependent cutaneous mucosal toxicity (dryness, erythema), mild myelosuppression with reversible neutropenia [44], inherent teratogenic risk [45], and mild hepatic metabolic stress with transaminase elevation [46]. These toxicities are manageable via dose optimization, local symptomatic interventions, weekly hematologic and hepatic function monitoring, and strict contraception for reproductive-age patients. Pharmacokinetically, CD2314 faces unique challenges in MM patients; poor BM targeting (25%-30% plasma concentration penetration) due to the BM stromal barrier [47], and unstable oral absorption with significant food-dependent variability [48]. These issues can be addressed by administering CD2314 with a high-fat diet to improve bioavailability and developing nanoformulations to enhance BM targeting [47, 48]. Collectively, these strategies address the key translational barriers and provide a feasible roadmap for the clinical development of RARβ agonist-IMiD combination therapy for high-risk t(4;14) MM, with the potential to fill the unmet medical need for overcoming IMiD resistance in this subgroup. While the immunodeficient nature of this model limits investigation of potential crosstalk between the identified epigenetic axis and IMiD-mediated immune modulation, future studies using humanized immune NSG models or syngeneic myeloma systems will address this gap and further dissect the combined contributions of cell-autonomous epigenetic effects and canonical IMiD immunomodulation to the anti-tumor activity of the ATRA-IMiD combination in t(4;14) MM.

ATRA, a vitamin A metabolite, plays a crucial role in regulating stem cell differentiation and development. Our previous study highlights the potential of ATRA to sensitize and overcome MM resistance to Cfz treatment in MM patients [49]. As the ligand for the RAR family of transcription factors, ATRA mediates their interaction with RAREs located in proximal promoters and enhancers of target genes [50]. While the mechanisms of ATRA-mediated gene activation are well characterized, less is known about how it suppresses RA-regulated genes [21, 51]. Unlike the classical activation model, our study focuses on ATRA’s repressive function. Previous reports have shown that RARE sequence in the upstream of genes such as fibroblast growth factor 8 [52] and homeobox B1 [53] mediate transcriptional repression rather than activation. In these cases, RA-bound RAR recruits nuclear receptor corepressors (NCORs) and polycomb repressive complex 2 (PRC2), leading to H3K27 trimethylation (H3K27me3). Unlike previously reported findings [54], our work reveals a distinct mechanism that HDAC were recruited and mediated reduction of chromatin acetylation (H3K27ac) to downregulate target gene expression. We elucidated that the combination of ATRA and IMiDs reduces the acetylation level of the oncogene FGFR3 promoter region, thereby downregulating its expression and re-sensitizing POM-resistant t(4;14) MM cells to IMiD treatment. Moreover, our research provides a clinically translatable therapeutic strategy for high-risk MM patients.

Author contributions

LZ, QW, and QY initiated the study, designed the research and wrote the paper. ZL, QW, MX, JFQ, CCZ, and WW performed experiments. YLZ provided patients’ samples. QG and YFZ performed patients’ sample analyses. DR and SM planned and performed the bioinformatics analysis of the clinical RNA-seq and RNA-seq data. RZO and FHZ provided critical suggestions.

Funding

QY discloses support for research of this work from Startup Support from Houston Methodist Research Institute and Myeloma Solution Fund (Grant# MSF-006). Yi’s team was also supported by grants from Cancer Prevention & Research Institute of Texas Recruitment of Established Investigator Award (Grant# RR180044) and High-Impact/High-Risk Research Award (Grant# RP210868), and NCI RO1s (Grant# CA239255, CA282099, CA278787, CA2855029, and CA285203).

Data availability

The RNA-seq datasets (accession# GSE304473), ATAC-seq datasets (accession# GSE304474), and ChIP-seq datasets (accession# GSE304534) generated during the current study are available in the GEO repository (https://www.ncbi.nlm.nih.gov/geo/). Other original data are available from the corresponding authors upon reasonable request.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

All methods were performed in accordance with the relevant guidelines and regulations. Murine experiments were performed in accordance with the protocol (#IS00008301) approved by the Institutional Animal Care and Use Committee of the Houston Methodist Research Institute. MM patient samples were obtained from an IRB-approved tissue bank (#Pro00019970) at the Houston Methodist Research Institute. All patients provided written informed consent.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41375-026-03078-y.