The Novel Bromodomain and Extraterminal Domain Inhibitor INCB054329 Induces Vulnerabilities in Myeloma Cells That Inform Rational Combination Strategies

Bromodomain and extraterminal domain (BET) proteins comprise a family of four related proteins (BRD2, BRD3, BRD4, and BRDT), each harboring two tandem amino-terminal bromodomains (BD1 and BD2) that bind selectively to acetylated lysine residues. BET proteins regulate the expression of an array of genes by binding to highly acetylated histone tails at the promoters and enhancers of target genes and by recruiting transcriptional complexes including the super-elongation complex and the protein transcriptional elongation complex B (1, 2). Association of BRD4 with chromatin is strongly enriched at large enhancer elements, called super-enhancers, which are frequently co-opted in tumor cells to drive expression of genes involved in cell proliferation, cell fate, and survival (3, 4). Correspondingly, reduction of BRD4 by siRNA in transformed cells inhibits cell-cycle progression, due to G₁ arrest (5, 6).

The small-molecule inhibitors, JQ-1 and iBET, that block BET binding to chromatin showed therapeutic potential against both solid and hematologic tumor models, in part, through suppression of oncogenes such as c-MYC (7–9). In addition, BRD2, BRD3, and BRD4 have been shown to control the inflammatory response in models of acute inflammation by regulating the expression of proinflammatory modulators such as interleukin (IL)-6 and tumor necrosis factor (TNF)α (8, 10). Tumor-associated inflammation is a hallmark of cancer, and elevated levels of proinflammatory proteins, including IL6, have been shown to promote multiple aspects of tumorigenesis. Thus, antitumor activity of BET inhibitors may result from modulation of inflammation in addition to direct effects on the tumor cell.

Multiple myeloma is a progressive cancer of malignant plasma cells that initiates from genetic lesions including copy-number alterations and chromosomal translocations resulting from immunoglobulin switch recombination (11, 12). Juxtaposition of the immunoglobulin heavy-chain (IgH) enhancers to oncogenes, such as FGFR3/MMSET or CCND1 in t(4;14) and t(11;14), respectively, occurs as early initiating events in approximately 25% of cases (13). Progression corresponds to the accumulation of secondary genetic and epigenetic alterations including the deregulation of c-MYC by various chromosomal rearrangements (14, 15). The survival and proliferation of both normal and malignant plasma cells are critically dependent upon the interaction with a variety of cells in the bone marrow microenvironment, including stromal, vascular, bone, and immune cells (16). Interactions among these cells are mediated by cytokine receptors, which include IL6, insulin-like growth factor-1, basic fibroblast growth factor, and various other chemokines and cytokines.

We present the preclinical characterization of a novel, phase I BET inhibitor, INCB054329. Our data demonstrate that, in addition to c-MYC, other putative oncogenes such as FGFR3 and pathways including JAK–STAT are strongly affected by BET inhibition in models of myeloma, thereby creating vulnerabilities that can be exploited with targeted agents. These results provide a strategy for potential selection of rational combinations with BET inhibitors.

All cell lines except INA-6 were purchased from ATCC, DSMZ, or the Japanese Collection of Research Bioresources Cell Bank and were acquired between 2003 and 2013. INA-6 cells were a gift from Dr. Renate Burger (University Hospital Schleswig-Holstein, Kiel, Germany). All cell lines were confirmed to be negative for mycoplasma (Bionique Testing Laboratories, Inc.), and cell lines used for in vivo testing were authenticated by short tandem repeat analysis between 2013 and 2017. INCB054329, INCB054828, ruxolitinib, itacitinib, and JQ-1 were synthesized at Incyte Corporation.

The binding of INCB054329 to BET bromodomains BRD4-BD1 and BRD4-BD2, BRD3-BD1 and BRD3-BD2, BRD2-BD1 and BRD2-BD2, and BRDT-BD1 was assessed using the AlphaScreen assay (PerkinElmer). A fluorescence polarization assay was used to measure binding to the BRDT-BD2. Binding to non-BET bromodomains was assessed using the BROMOscan assay at a fixed concentration of 3 μmol/L of INCB054329 (DiscoverX).

Gene-expression analysis was performed using Affymetrix Human Genome U133 PLUS 2.0 microarrays with data being analyzed using OmicSoft's Array Studio software (OmicSoft Corporation) and gene set enrichment analysis (GSEA; refs. 17, 18). Chromatin immunoprecipitation (ChIP) sequencing was performed as described (19), with mapped reads being visualized in Array Studio.

Chromatin was isolated from cell lines using the SimpleChIP Enzymatic ChIP kit (Cell Signaling Technology). The antibodies used for ChIP were anti-BRD4 (Bethyl Laboratories; A301-985) and anti-Histone H3-acetyl K27 (Abcam; ab4729). Analyses of the IgH enhancers and the c-MYC transcriptional start site (TSS) were performed using published primer-probe sets (19). The forward and reverse primers for the negative control region of FGFR3 are 5′-GCGCTAACACCACCGACAA and 5′-GAGTGATGAGAAAACCCAATAGAATTG with the probe sequence 5′-AGCTAGAGGTTCTCTCCTTGCACAACGTCA. The primer and probe sequence for IL6 receptor alpha (IL6Rα) are 5′-GCGCGAGTTCCTCAAATGTT and 5′-TCCCACTCGCACATGACTCA with the probe sequence CCTGCGTTGCCAGGACCGTCC. These data are expressed as fold enrichment (2E[-ΔΔCt[ChIP/nonimmune control]) and normalized to the input (2% of the dimethyl sulfoxide [DMSO] sample).

All antibodies were purchased from commercial sources as follows: antibodies against phosphorylated signal transducers and activators of transcription 3 (pSTAT3; pY705), STAT3, c-MYC, PIM2, phosphorylated fibroblast growth factor substrate 2 (FRS2; pY436), GAPDH, and Actin were from Cell Signaling Technology; antibodies against WHSC1/MMSET were from Novus Biologicals; antibodies against p21 were from BD Biosciences, and antibodies against fibroblast growth factor receptor 3 (FGFR3), pFGFR3 (pY724), and FRS2 were from Abcam. The c-MYC ELISA kit was purchased from Cell Signaling Technology, and ELISA was performed per the manufacturer's protocol.

Cell viability assays were performed after 72 hours of incubation with a serial dilution of INCB054329 or 0.1% DMSO as control using the CellTiter-Glo ATP assay (Promega). All cell lines were tested with a minimum of three independent experiments, and data are reported as the mean ± standard deviation. To test dependence of the OPM-2 cells on FGFR3, gene editing by CRISPR-Cas9 was performed using 3 gRNAs targeted to FGFR3 and a control gRNA targeting Green Fluorescent Protein (Cellecta Inc.). OPM-2-Cas9 cells were generated by infection of lentiviral Cas9, followed by selection with 5 μg/mL blasticidin for 1 week. OPM-2-Cas9 cells were infected with gRNAs for 2 days and then selected using 1 μg/mL puromycin for 3 days to eliminate uninfected cells. Edited cells were counted and seeded into 96 wells (3,000/well), and growth of the cells was measured over time as described. FGFR3 knockdown was confirmed by Western blotting. The gRNA sequences used for targeting GFP and FGFR3 are as follows: gGFP: AAGATCGAGTGCCGCATCAC; gFGFR3-1: GGGGACGGAGCAGCGCGTCG; gFGFR3-1: GGGAGATGACGAAGACGGGG; gFGFR3-1: GCTGCCGGCCAACCAGACGG.

T cells were separated from human peripheral blood mononuclear cells of healthy adults by elutriation and treated with phytohemagglutinin (PHA) for 3 days to induce cell-surface expression of IL2 receptors. The PHA-activated and phosphate-buffered saline-washed T cells were grown in the presence of 100 U/mL of IL2 and different concentrations of INCB054329 for 3 days. Proliferation analysis was performed using CellTiter-Glo Reagent as described above.

Cell-cycle distribution was determined using propidium iodide staining and analyzed by flow cytometry according to established protocols. The kinetic parameters of the cell cycle (G₁, S, G₂+M) and the coefficient of variance of G₁ peak from 50,000 events were computed using the Cflow Plus software (BD Biosciences).

Apoptosis was determined using Annexin V–propidium iodide staining of cells treated with INCB054329 or DMSO vehicle according to standard procedures, and data from 10,000 events were analyzed by CflowPlus Software using an Accuri C6 cytometer (BD Biosciences).

All experiments involving mice were performed under the guidelines and regulations of the Institutional Animal Care and Use Committee of Incyte Corporation. CB17/Icr-Prkdc scid/IcrIco Crl (severe combined immunodeficiency [SCID]) and Nu/Nu (nude) mice were purchased from Charles River Laboratories.

For the administration by oral gavage, all Incyte compounds were first reconstituted in N,N-dimethylacetamide (DMAC; Sigma-Aldrich) and then diluted in a 0.5% methylcellulose (Sigma-Aldrich) solution for a final concentration of 5% DMAC.

For xenograft studies, 10⁷ cells were resuspended in Matrigel (BD Biosciences; KMS-12-BM and MM1.S) or resuspended as brei (INA-6) and injected subcutaneously into the dorsal flanks of mice. Subcutaneous tumors were allowed to grow until they reached approximately 200 mm³, at which point the mice were randomized into dosing groups. Tumors were measured at least twice weekly with electronic calipers. Tumor volumes were calculated using the following formula: volume = (length × width²)/2, where width was the smaller dimension. Body weights were also monitored.

A fragment to pharmacophore strategy was used to identify a novel fused tricyclic scaffold and subsequently led to the discovery of the candidate INCB054329 (Fig. 1A). The isoxazole moiety mimics the acetyl lysine group in the BRD4-BD1 pocket, and the pyridyl ring binds on the WPF shelf (Fig. 1B). In biochemical binding assays, INCB054329 potently inhibited the binding of tetra-acetylated histone H4 peptide to BRD2, BRD3, and BRD4, with modest selectivity for BRDT-BD1 and BRDT-BD2 (Fig. 1C). A screen against 16 non-BET bromodomains showed no significant inhibitory activity at 3 μmol/L (Fig. 1D).

The antiproliferative activity of INCB054329 was profiled against a panel of 32 hematologic cancer cell lines derived from acute myeloid leukemia, non-Hodgkin lymphoma, and multiple myeloma (Fig. 2A). The median 50% growth inhibition (GI₅₀) value was 152 nmol/L (range, 26–5,000 nmol/L), and inhibitory activity was observed across all histologies (Supplementary Table S1). In contrast to tumor cell lines, the GI₅₀ value against T cells isolated from non-diseased donors stimulated ex vivo with IL2 was 2.435 μmol/L. Growth inhibition correlated with a concentration-dependent accumulation of cells in the G₁ phase of the cell cycle (Fig. 2B). An increase in apoptosis was also observed in four of five myeloma cell lines at 1,000 nmol/L, and significant increases were observed from 100 nmol/L in INA-6 cells (Fig. 2C). Consistent with published experiments (19), ChIP of BRD4 showed enriched binding at the IgH enhancers in MM1.S cells (Fig. 2D). Treatment with INCB054329 displaced BRD4 binding to the E2 3′ enhancer, similar to findings with the BET inhibitor JQ-1 (Supplementary Fig. S1; ref. 19). Concomitant with the abrogation of BRD4 binding, levels of c-MYC mRNA were rapidly suppressed, leading to a concentration-dependent decrease in c-MYC protein in multiple cell lines (Fig. 2E and F). Using c-MYC as a pharmacodynamic marker in tumor cells, we evaluated the pharmacokinetic–pharmacodynamic relationship after a single oral administration of INCB054329 in mice bearing subcutaneous xenografts of KMS-12-BM tumors. The KMS-12-BM cell line has been shown to have very high c-MYC protein levels (20) and is therefore amenable for pharmacokinetic/pharmacodynamic study. A concentration-dependent decrease was observed 3 hours after dose with an in vivo half-maximal inhibitory concentration (IC₅₀) of 66 nmol/L and a 90% inhibitory concentration (IC₉₀) of approximately 400 nmol/L (Fig. 2G). INCB054329 exhibited high clearance in mice resulting in a short half-life (Supplementary Fig. S2A); therefore, twice-daily (b.i.d.) dosing was used in efficacy studies. At exposures that effectively suppressed c-MYC, INCB054329 was found to be efficacious and well tolerated in both the KMS-12-BM and MM1.S xenograft models (Supplementary Fig. S2B and S2C).

The t(4;14) translocation involves a balanced rearrangement of the IgH locus on chromosome 14, with chromosome 4 leading to deregulation of two potential oncogenes FGFR3 (encoding a receptor tyrosine kinase) and NSD2/MMSET/WHSC1 (encoding a DNA methyl transferase; Fig. 3A; ref. 13). To assess whether both translocation partners are regulated by BRD4, ChIP-PCR was performed on both the Eμ and E2 enhancers associated with the IgH locus. BRD4 binds to both elements in two t(4;14)-positive cell lines OPM-2 and KMS-11, with higher normalized levels at the E2 enhancer. Treatment with 500 nmol/L INCB054329 reduced BRD4 association with the E2 and Eμ enhancers to 8% and 29% of the DMSO control samples, respectively, in OPM-2 cells, while only BRD4 binding to E2 was reduced (18% of control) in the KMS-11 cells (Fig. 3B; Supplementary Fig. S1). Steady-state mRNA levels of FGFR3 were reduced to nearly background in the OPM-2 cells similar to c-MYC, while mRNA levels of NSD2 were decreased approximately 50% (Fig. 3C). FGFR3 protein levels were also suppressed to near background levels (1% of DMSO) at a concentration of 250 nmol/L (approximately twice the IC₅₀), whereas NSD2/MMSET protein was maximally reduced by 50% at 24 hours (Fig. 3D).

In a subcutaneous xenograft model of OPM-2 tumors, oral administration of INCB054329 resulted in a dose-dependent suppression of tumor growth and was well tolerated as assessed by body-weight measurements (Fig. 3E). The strong suppression of FGFR3 suggested that greater inhibition of tumor growth could be achieved in combination with FGFR targeting. INCB054828 is a selective kinase inhibitor of the FGFR 1, 2, and 3 and is currently being tested in phase I/II trials of solid and hematologic cancers (NCT02393248 and NCT03011372). INCB054828 inhibited OPM-2 cell proliferation (GI₅₀, 97 nmol/L) similar to deletion of FGFR3, indicating at least partial dependence of these cells on FGFR signaling (Supplementary Fig. S3). In vivo, partial antitumor efficacy by INCB054828 was observed against OPM-2 xenografted tumors at its optimal dose of 0.3 mg/kg once daily (Fig. 3E). When combined with INCB054329, however, a significantly greater decrease in terminal tumor volume was achieved than by either compound alone (P < 0.001 by two-way ANOVA for combination versus each single agent). Strikingly, FGFR inhibition alone resulted in a feedback upregulation of FGFR3 that was completely abrogated in the combination group (Fig. 3F). Despite the increased FGFR3 levels in tumors from INCB054828-treated mice, signaling through FGFR3 is blunted by INCB054828 (as determined by pFGFR3 and pFRS2 levels), and slightly more so by the combination of INCB054828 and INCB054329. The combination also decreased levels of c-MYC and pSTAT3 and increased levels of p21 when compared with levels from single-agent–treated tumor samples.

To characterize the broader effects of BET inhibition on gene expression, two myeloma cell lines MM1.S and INA-6 were treated for 4 and 24 hours with INCB054329 and analyzed by gene-expression profiling (Supplementary Fig. S4A). Pathway analysis of differentially expressed genes was evaluated using two methods: MetaCore and GSEA. By MetaCore, a number of pathways related to growth, development, and inflammation were enriched (Fig. 4A). To explore molecular signatures in greater detail, the data were also analyzed by GSEA. c-MYC–dependent gene sets emerged among the top downregulated gene sets in both MM1.S and INA-6 cell lines (Fig. 4B; Supplementary Fig. S4B), and c-MYC was identified as the most significantly downregulated transcription factor network in MM1.S cells, recapitulating results with JQ-1 and other BET inhibitors (19, 21). In addition, the IL6 signature in myeloma was among the top pathways in both cell lines. In monocytes, BET inhibitors repressed production of IL6 (10), and INCB054329 also suppressed secretion of IL6 by immortalized, human stromal cell lines (Supplementary Fig. S5); however, neither myeloma cell line expressed high levels of IL6 (data not shown). In contrast, the mRNA levels of IL6R encoding CD126 or the IL6Rα subunit was significantly downregulated in MM1.S cells treated with INCB054329 at 4 hours [1.6-fold, false discovery rate (FDR) <0.002; Fig. 4C). A 2.4- to 3.5-fold reduction in IL6R mRNA within 4 hours was also observed in additional cell lines, including U-266, OPM-2, and KMS-11. Analysis of the gene track for BRD4 association with the IL6R gene locus showed strong enrichment near the TSS that corresponded to a peak of histone H3K27 acetylation in CD14-positive myeloid cells from project ENCODE data (Fig. 4D; ref. 22). This activation histone mark at the TSS was confirmed in MM1.S cells by ChIP-PCR for H3K27Ac (data not shown) and correlated with BRD4 association in control cells. Treatment with JQ-1 abrogated the BRD4 peaks at this site. Similarly, treatment with INCB054329 resulted in depletion of BRD4 from the IL6Rα TSS in OPM-2 cells (Fig. 4E).

To evaluate whether the reduction of IL6Rα would translate into a vulnerability that could be therapeutically exploited, cellular analyses were performed to assess the effect on IL6Rα protein levels and signaling in myeloma cell lines. INCB054329 treatment induced a dose-dependent decrease in cell-surface IL6Rα staining as measured by flow cytometry in MM1.S and U-266 cells and in IL6Rα protein levels by Western blotting in MM1.S and U-266 cells that corresponded to decreased levels of IL6Rα protein (Fig. 5A). The consequence of IL6Rα reduction on IL6 signaling was measured in U-266 cells because they express high levels of IL6Rα but are not dependent on IL6 for survival, in contrast to INA-6 cells that rapidly lose viability upon loss of IL6 signaling (23). Basal levels of pSTAT3 were reduced by nearly eight fold with INCB054329 (Fig. 5B, compare lanes 1 to 4). Upon stimulation with IL6, a concentration-dependent increase in the pSTAT3 level was observed in both DMSO and INCB054329-treated groups; however, in cells pretreated with INCB054329, pSTAT3 was induced to an extent that only slightly exceeded the unstimulated DMSO control (Fig. 5B, compare lanes 1 and 6). Therefore, reducing cell-surface IL6Rα levels diminished the response to physiologically relevant levels of IL6. Consistent with this effect, transcriptional profiling showed decreased levels in expression of several genes also known to be at least in part transcriptionally regulated by STAT3, including BCL-xL, and cyclins D1 and D2 (Fig. 4C), indicating BET inhibition of IL6Rα levels results in decreased levels of several downstream pathway targets. As a possible compensatory response to suppressed IL6 signaling, genes in the IL6–JAK2–STAT3 pathway were upregulated, including time-dependent increases in JAK1, JAK2, STAT6, and STAT3 transcripts.

We reasoned that targeting of BET in combination with a selective JAK inhibitor could have combinatorial effects on myeloma cell viability, especially in view of potential compensatory increases in JAK–STAT pathway genes. Studies were undertaken using the JAK1/2 inhibitor ruxolitinib (24) or the JAK1-selective inhibitor itacitinib (Supplementary Table S2). INA-6 cells were sensitive to both JAK and BET inhibitors, and combining either JAK inhibitor with INCB054329 potentiated the effect on viability. Using Bliss independence calculation (25), the average fractional product (Fua) across a 5 × 5 combination grid was 7 and 12 for combinations with ruxolitinib and itacitinib, respectively, indicating a synergistic interaction. In contrast, a control grid of INCB054329 versus itself yielded an additive Fua of 1. The combination of either JAK inhibitor with INCB054329 also induced a greater degree of apoptosis (Fig. 5C). Analysis of downstream targets demonstrated that the combination of JAK1 and BET inhibitors had markedly enhanced inhibition of pSTAT3 and modestly greater suppression of c-MYC than either agent alone. Interestingly, JAK1 inhibition resulted in decreased protein levels of the STAT3 transcriptional target PIM2, whereas PIM2 levels remained unchanged by BET inhibition, suggesting that the BET and JAK1 inhibitors can differentially influence the outcome of STAT-mediated transcription (Fig. 5D). The MM1.S cell line that is independent of IL6 did not show sensitivity to any JAK inhibitor in viability assays nor did the addition of JAK inhibitor increase the activity of INCB054329 under standard culture conditions (data not shown).

To evaluate whether the combination of JAK and BET inhibitors would translate to improved efficacy, INA-6 xenografts were established in female SCID mice. At a dose of 50 mg/kg b.i.d., INCB054329 reduced tumor growth by 51% (P < 0.0001, two-way ANOVA) compared with the vehicle control group after 8 days of oral dosing (Fig. 6A). A dose of ruxolitinib suitable for combination studies (30 mg/kg b.i.d.) also suppressed tumor growth by 47% (P < 0.0001). The combination of ruxolitinib and INCB054329 significantly enhanced the efficacy (89% tumor growth inhibition) and resulted in partial tumor regressions in five of eight animals. The combinations with itacitinib at a dose of 60 mg/kg b.i.d. exhibited similar results, with each combination significantly reducing tumor size when compared with its single-agent arms (P < 0.0001, two-way ANOVA for each). All combinations were well tolerated based on body-weight measurements.

The pharmacodynamic effects of the BET and JAK inhibitors were also examined in the INA-6 xenograft model. Levels of c-MYC protein were lower in tumors from INCB054329-treated mice than vehicle controls (Fig. 6B). Levels of pSTAT3 were decreased by ruxolitinib alone, and the combination of the BET inhibitor with either JAK inhibitor further decreased STAT3 phosphorylation (Fig. 6C), consistent with the results in vitro.

The combination of INCB054329 with ruxolitinib was also tested in the IL6-independent model MM1.S. Similar to the in vitro data, subcutaneous xenografts of the MM1.S tumor were insensitive to ruxolitinib monotherapy (Fig. 6D) at the same dose (30 mg/kg b.i.d.) used in the INA-6 study. Surprisingly, despite this lack of effect with ruxolitinib, tumor growth was suppressed to a significantly greater extent by INCB054329 in combination with ruxolitinib compared with the BET inhibitor alone (P < 0.05, two-way ANOVA for combination vs. either single agent). INCB054329 significantly decreased tumor c-MYC levels (Fig. 6E), and levels of pSTAT3 were partially reduced by either BET or JAK inhibition and showed further reduction in the combination groups (Fig. 6F).

Here, we present an extensive preclinical characterization of INCB054329, a structurally distinct and rationally designed BET inhibitor that has been evaluated in phase I clinical trials (NCT02431260). INCB054329 potently and selectively antagonizes the BET subfamily of bromodomain proteins and shows antiproliferative activity against cancer cell lines derived from hematologic malignancies of different histologies. Treatment with INCB054329 reduced c-MYC levels in cell lines as well as in subcutaneous xenograft tumors, with an in vivo IC₅₀ value of 66 nmol/L and maximally reduced c-MYC with a plasma concentration of 399 nmol/L. Using these data, the maximum efficacy of single-agent therapy was observed when the in vivo IC₅₀ was exceeded for at least 14 hours, which was achieved by twice-daily administration.

Previous studies have shown antimyeloma activity of BET inhibitors by downregulation of oncogenic signaling through c-MYC (13, 19, 26). In addition to these well-characterized effects, BET inhibitors also diminish bromodomain protein binding to the IgH enhancer in the 14q32 translocation setting, including the t(8;14) translocation (involving 14q32) linking IgH to MYC. Although the t(8;14) translocation is present only in a very small percentage of patients with multiple myeloma, several other reciprocal translocations involving 14q32 lead to highly increased levels of oncogenic proteins, most prominently MAF, CCND1, CCND3, and MMSET/FGFR3 (11, 12). Oncogene upregulation by IgH drives distinct gene-expression signatures that can be used to categorize myelomas into subgroups (27). The utility of BET inhibition in myeloma becomes more apparent due to the ability of BET proteins to target downstream growth-promoting effectors such as growth factors and cytokines that are expressed across myeloma subgroups. Likewise, BET inhibition may also be effective regardless of subgroup or genetic alteration due to its ability to block transcriptional activity from the IgH locus.

Among the most prevalent alteration is the t(4;14)-balanced translocation, in which the intronic Eμ enhancer is joined to NSD2 on der4 while the 3′ enhancers segregate with FGFR3 on der14 (13). BRD4 binding was detected at both enhancer elements in t(4;14)-positive cell lines, but the dependence upon BRD4 was not equivalent. Inhibition by INCB054329 resulted in a dramatic decrease in FGFR3 expression and protein, while levels of NSD2 were only modestly affected. In cell lines that have a partial dependency on FGFR3 signaling for survival, the depletion of FGFR3 created a vulnerability that could be exploited by combination with selective FGFR-targeted inhibitors. In vivo, combination of INCB054329 with the selective FGFR1/2/3 inhibitor INCB054828 showed significantly greater efficacy than either mechanism alone. A strong feedback signal resulted from FGFR targeting that may have limited the efficacy of FGFR inhibitor monotherapy. Importantly, this adaptive response was completely abrogated by combination with INCB054329. The idea of combining signaling pathway inhibitors with BET inhibitors to decrease expression of genes within the pathway and to counter feedback activation is a potentially powerful strategy. Indeed, this concept has been previously validated for combinations of BET inhibitors with the Bruton tyrosine kinase inhibitor ibrutinib in models of non-Hodgkin lymphoma (28, 29), and with phosphoinositide 3-kinase or epidermal growth factor receptor 1/2 kinase inhibitors in models of breast cancer and other solid tumors (30, 31).

In addition to direct regulation of oncogenes, BET inhibition can regulate levels of effectors downstream of the oncogenic signal. For example, BET proteins have been shown to regulate expression of proinflammatory genes including cytokines, chemokines, and their receptors in models of acute inflammation (8, 10). Gene-expression profiling revealed that immune and inflammatory pathways were strongly enriched by treatment of myeloma cell lines with INCB054329, even in the absence of acute challenge. Notably, IL6R has been found to be strongly overexpressed across several multiple myeloma subgroups (27). IL6R is a direct target of BRD4, and inhibition by INCB054329 or JQ-1 displaced it from the IL6R TSS suppressing the expression of IL6Rα and resulting in diminished signaling through STAT3 even upon IL6 stimulation. An apparent compensatory response to decreased IL6–JAK–STAT signaling was evidenced by upregulation of components of the JAK–STAT signaling pathway. Consistent with this finding, combinations with the clinical JAK inhibitors ruxolitinib or itacitinib further reduced JAK–STAT signaling in the IL6-dependent INA-6 model, resulting in synergistic inhibition of proliferation and enhanced apoptosis in cells. Combining BET and JAK inhibitors resulted in superior efficacy at well-tolerated doses and together suppressed levels of pSTAT3 more effectively than either agent alone. The importance of IL6 as a growth and survival factor for myeloma cells and as a modulator of cellular interactions in the myeloma tumor microenvironment is well documented (11, 32). Gene-expression analysis of human myeloma has revealed a significant subset of tumors with an activated STAT signature (33). In murine models, activated IL6–JAK–STAT signaling and c-MYC have been shown to collaborate in the development of myeloma. An intercross of mice harboring a human IL6 transgene with Eμ-c-MYC mice resulted in the acceleration of plasma cell tumor formation with full penetrance (34). Separately, a murine transduction–transplantation model of myeloma driven by a constitutively active allele of Gp130 encoding the common coreceptor subunit for cytokine receptors, including IL6Rα, also showed enhanced tumorigenesis in cooperation with c-MYC. In plasmacytomas arising from Gp130 activation alone, activating genetic alterations in the c-MYC locus were recovered in three of ten independently arising tumors (33). Combining JAK and BET inhibitors is a potentially promising strategy for other hematologic malignancies, including JAK2V617-dependent myeloproliferative neoplasms and acute lymphocytic leukemia (ALL) characterized by IL7R activation (35–38). In the ALL models, JQ-1 suppressed STAT5-dependent signaling and improved survival; significantly at relapse, these tumors showed reactivation of STAT5 signaling indicating a feedback upregulation of JAK–STAT signaling.

The mechanism for synergy merits additional study and should consider both cell-intrinsic and cell-extrinsic mechanisms given that BET inhibitors may affect expression of soluble mediators within the tumor microenvironment that may be important for supporting tumor cell viability. As an example, the MM1.S xenograft model, which does not respond to JAK inhibition alone, was unexpectedly sensitized to JAK inhibition by coadministration of INCB054329 in vivo. This sensitivity to combined JAK and BET inhibition was not observed in vitro. One potential explanation for these findings is that the levels of cytokines or growth factors in the tumor microenvironment that provide protection from JAK inhibition may not be as high as those in cell culture. It is also possible that BET inhibition could make some myelomas more reliant on cytokine signaling for survival, consistent with the upregulation of JAK–STAT pathway genes. Finally, it is possible that BET inhibitors have biological effects on other cell types within the tumor microenvironment that may support the malignant cells. Preliminary data show that BET inhibition can markedly reduce the secretion of multiple cytokines, including IL6 and IL8 (Supplementary Fig. S5), in cultured stromal cell lines at concentrations below the IC₅₀ for suppression of cell growth. Thus, in vivo BET inhibition may have pleiotropic biological impacts beyond direct effects on the malignant clone.

In summary, we present the profile of a BET inhibitor, INCB054329, that has advanced into phase I trials. Using myeloma as a model, we identified oncogenes and inflammatory pathways regulated by BET proteins that are significantly altered by treatment with the INCB054329. Preliminary clinical data from trials with BET inhibitors have shown only modest single-agent efficacy to date in hematologic cancers, including myeloma, suggesting that rational therapeutic combinations designed to target different aspects of cancer biology may be required to enhance responses in patients with advanced malignancies (39). By evaluating BET inhibitor–induced changes in growth-promoting and survival pathways, vulnerabilities are revealed that can guide rational combinations such as with FGFR inhibitors in the case of selected t(4;14)-rearranged myelomas and with JAK inhibitors in models characterized by an enriched IL6 signature. Importantly, the effects of INCB054329 on these targets were achieved using modest concentrations (approximately 2–3 × in vivo IC₅₀), suggesting that these findings can be translatable to the clinic. By targeting different aspects of tumor biology, combination therapy may overcome the tumor heterogeneity and complexity that may be contributing to limited single-agent efficacy in early trials.

M.C. Stubbs, T.C. Burn, A. Volgina, N. Zolotarjova, J. Li, R. Collins, V. Dostalik, R. Wynn, B. Ruggeri, S. Yeleswaram, R. Huber, and P.C.C. Liu have ownership interests (including patents) in Incyte Corporation. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M.C. Stubbs, T.C. Burn, T. Maduskuie, R. Wynn, W. Yao, A.P. Combs, R. Huber, G. Hollis, P. Scherle, P.C.C. Liu

Development of methodology: M.C. Stubbs, T.C. Burn, A. Volgina, N. Zolotarjova, M. Favata, X.M. Liu, J. Li, N. Falahatpisheh, P. Polam, P. Thekkat, X. He, M. Covington

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Rupar, X. Wen, N. Zolotarjova, M. Favata, X.M. Liu, J. Li, D. DiMatteo, P. Feldman, P. Thekkat, C. Gardiner, Y. Li, M. Covington, R. Wynn, P.C.C. Liu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.C. Stubbs, T.C. Burn, S. Diamond, M. Rupar, A. Volgina, N. Zolotarjova, M. Favata, H. Liu, X.M. Liu, J. Li, P. Feldman, P. Thekkat, C. Gardiner, X. He, M. Covington, R. Wynn, S. Yeleswaram, P. Scherle, P.C.C. Liu

Writing, review, and/or revision of the manuscript: M.C. Stubbs, T.C. Burn, M. Rupar, R. Collins, V. Dostalik, M. Covington, B. Ruggeri, G. Hollis, P. Scherle, P.C.C. Liu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Ruggeri

Study supervision: M.C. Stubbs, S. Diamond, B. Ruggeri, S. Yeleswaram, C.-B. Xue, R. Huber, P.C.C. Liu

Other (medicinal chemistry research): R. Sparks

Other (contributing author): P. Waeltz

Other (research: Molecular modeling): R. Jalluri

Medical writing assistance was provided by Abigail Marmont of Evidence Scientific Solutions Inc., Wilmslow, UK, and funded by Incyte Corporation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.