Immunomodulatory drugs target IKZF1-IRF4-MYC axis in primary effusion lymphoma in a cereblon-dependent manner and display synergistic cytotoxicity with BRD4 inhibitors.

Primary effusion lymphoma (PEL) is an aggressive type of non-Hodgkin lymphoma localized predominantly in body cavities. Kaposi's sarcoma-associated herpes virus (KSHV) is the causative agent of PEL. PEL is an incurable malignancy and has extremely poor prognosis when treated with conventional chemotherapy. Immunomodulatory drugs (IMiDs) lenalidomide and pomalidomide are Food and Drug Administration-approved drugs for the treatment of various ailments. IMiDs display pronounced antiproliferative effect against majority of PEL cell lines within their clinically achievable concentrations, by arresting cells at G0/G1 phase of cell cycle and without any induction of KSHV lytic cycle reactivation. Although microarray examination of PEL cells treated with lenalidomide revealed activation of interferon (IFN) signaling, blocking the IFN pathway did not block the anti-PEL activity of IMiDs. The anti-PEL effects of IMiDs involved cereblon-dependent suppression of IRF4 and rapid degradation of IKZF1, but not IKZF3. Small hairpin RNA-mediated knockdown of MYC enhanced the cytotoxicity of IMiDs. Bromodomain (BRD) and extra-terminal domain (BET) proteins are epigenetic readers, which perform a vital role in chromatin remodeling and transcriptional regulation. BRD4, a widely expressed transcriptional coactivator, belongs to the BET family of proteins, which has been shown to co-occupy the super enhancers associated with MYC. Specific BRD4 inhibitors were developed, which suppress MYC transcriptionally. Lenalidomide displayed synergistic cytotoxicity with several structurally distinct BRD4 inhibitors (JQ-1, IBET151 and PFI-1). Furthermore, combined administration of lenalidomide and BRD4 inhibitor JQ-1 significantly increased the survival of PEL bearing NOD-SCID mice in an orthotopic xenograft model as compared with either agent alone. These results provide compelling evidence for clinical testing of IMiDs alone and in combination with BRD4 inhibitors for PEL.

Introduction

Primary effusion lymphoma (PEL) is an aggressive type of non-Hodgkin lymphoma localized predominantly in body cavities that is observed primarily in patients with human immunodeficiency virus (HIV) infection[1] and is associated with infection by Kaposi’s sarcoma associated herpesvirus (KSHV)[2]. The prognosis of PEL is extremely poor with median survival of 4 and 6 months in HIV-positive and -negative patients, respectively.[3] Thus, there is an urgent need to develop new treatment regimens for PEL.

Thalidomide and its analogues, lenalidomide and pomalidomide, are collectively known as immunomodulatory drugs (IMiDs).[4] Thalidomide was originally introduced as a sedative but was later withdrawn from the market due to birth defects and neuropathy.[5] Subsequently, thalidomide was found to significantly improve the response rate and survival of patients with multiple myeloma (MM).[6] The second generation IMiDs, lenalidomide and pomalidomide, possess more potent anti-myeloma, anti-inflammatory and immunomodulatory activities than thalidomide.[5] Thalidomide directly binds to and inhibits the cereblon (CRBN) ubiquitin ligase,[7] and CRBN has been shown to be required for the anti-myeloma activity of IMiDs.[8, 9] Lenalidomide-bound CRBN acquires the ability to degrade Ikaros family zinc finger proteins 1 and 3 (IKZF1 and IKZF3), two specific B-cell transcription factors. The loss of IKZF1 and IKZF3 was shown to be both necessary and sufficient for the anti-myeloma effect of lenalidomide.[10, 11]

Bromodomain (BRD)-containing proteins regulate lysine acetylation[12], an important mechanism to regulate chromatin structure. Bromodomain and extra-terminal (BET) subfamily has four members, BRD2, BRD3, BRD4 and BRDT, all of them share common domain structure. Recently, potent and highly specific BRD4 inhibitors were developed[13-15]. These inhibitors were shown to suppress MYC transcriptionally and demonstrate promising preclinical activity against MYC-driven cancers[16-18].

Here, we report that PEL cells are highly sensitive to IMiDs, lenalidomide and pomalidomide within their physiologically achievable concentrations. Furthermore, we discovered that low-dose combinations of IMiDs with BRD4 inhibitors, displayed synergistic anti-proliferative activity against PEL.

Results

IMiDs show selective cytotoxicity towards PEL

To examine the cytotoxicity of IMiDs against PEL, we treated a panel of 35 logarithmically growing hematopoietic cell lines (Supplementary Table 1) for 5 days with increasing concentrations of IMiDs. At concentrations that are achievable clinically [2.2 μM for lenalidomide[19] and 179 nM for pomalidomide[20]], 6 out of the 9 PEL cell lines (BC-3, BCBL-1, JSC-1, VG-1, UMPEL-1, and UMPEL-3) were sensitive to IMiDs with IC50 ranging from 0.2–1.2 μM and 32–111 nM for lenalidomide and pomalidomide, respectively (Figure 1A and Table1). Whereas BC-1, BCP-1 and APK-1 were sensitive to only higher doses of IMiDs with IC50 ranging from 2.6–10 μM and 226–744 nM for lenalidomide and pomalidomide, respectively (Figure 1A and Table 1). MM.1S (Myeloma), Daudi (Burkitt’s lymphoma) and TMD8 (Activated B-Cell Diffuse Large B-Cell Lymphoma; ABC-DLBCL) were also sensitive to both IMiDs with IC50 ranging from 0.2 – 2.1 μM and 38–113 nM for lenalidomide and pomalidomide, respectively (Supplementary Figure S1). All the remaining cell lines were either resistant to IMiDs or required higher doses for a moderate effect (Supplementary Figure S1). Consistent with its known requirement for in vivo metabolism,[8] thalidomide did not have any major effect on the growth of any of the cell lines tested or required a high dose for moderate effect (Figure 1A and Supplementary Figure S1). Treatment of PEL cells with IMiDs resulted in G1 cell-cyle arrest (Figure 1B and Supplementary Figure S2A). In contrast, IMiDs had no major effect on cell-cycle progression in DG-75 (Burkitt lymphoma) and OCILY-8 (Germinal Center B-cell Diffuse Large B-Cell Lymphoma; GCB-DLBCL) cells that were resistant to their anti-proliferative effect (Figure 1B and Supplementary Figure S2A).

Figure 1

IMiDs are effective against PEL. A, Indicated PEL cell lines were treated with increasing concentrations of lenalidomide, pomalidomide and thalidomide for 5 days, and cell viability was measured using an MTS assay. The values shown are mean±SE (n=3) of a representative experiment performed in triplicate for 3 times. B, Cell cycle analysis of BC-3, BCBL-1, JSC-1 and DG-75 cells treated with indicated doses of lenalidomide (Len) and pomalidomide (Pom) for 48 h. Cells were stained with propidium iodide and analyzed by flow cytometry. Data is representative of more than 3 individual experiments. C, Heat map representation of 992 genes that are up- or down-regulated (p<0.05) in BC-3 and BCBL-1 cells following 24 h treatment with lenalidomide (5 μM). D, Gene set enrichment analysis showing enrichment of gene sets which are involved in interferon signaling among genes affected by lenalidomide treatment in PEL. NES, normalized enrichment score; q, false discovery rate.

Table 1

List of PEL cell lines, their characteristics, and 50% inhibitory concentration (IC50)a for IMiDs

Cell line	Year established	Associated Virus	Source	HIV status	Lenalidomide (IC50, μM)	Pomalidomide (IC50, nM)
BC-3	19951	KSHV	Pleural effusion	Negative	0.96	107
BCBL-1	19962	KSHV	N/A	Positive	0.20	74
JSC-1	20003	KSHV and EBV	Peritoneal effusion	Positive	0.28	34
VG-1	19984	KSHV	Pleural effusion	Negative	0.87	101
UMPEL-1	20105	KSHV and EBV	Pleural effusion	Negative	0.36	32
UMPEL-3	20136	KSHV and EBV	Peritoneal effusion	Positive	1.2	111
BC-1	19927	KSHV and EBV	Peritoneal effusion	Positive	2.6	744
BCP-1	19958	KSHV	PBMC	Negative	10	396
APK-1	Pre-20039	KSHV	N/A	N/A	2.8	226

IC50 values were calculated using Graphpad Prism 5 software. N/A, not available

References

Arvanitakis L, Mesri EA, Nador RG, Said JW, Asch AS, Knowles DM et al. Establishment and characterization of a primary effusion (body cavity- based) lymphoma cell line (BC-3) harboring kaposi’s sarcoma-associated herpesvirus (KSHV/HHV-8) in the absence of Epstein-Barr virus. Blood 1996; 88: 2648–2654.

Renne R, Zhong W, Herndier B, McGrath M, Abbey N, Kedes D et al. Lytic growth of Kaposi’s sarcoma-associated herpesvirus (human herpesvirus 8) in culture. Nat Med 1996; 2: 342–346.

Cannon JS, Ciufo D, Hawkins AL, Griffin CA, Borowitz MJ, Hayward GS et al. A new primary effusion lymphoma-derived cell line yields a highly infectious Kaposi’s sarcoma herpesvirus-containing supernatant. J Virol 2000; 74: 10187–10193.

Jones D, Ballestas ME, Kaye KM, Gulizia JM, Winters GL, Fletcher J et al. Primary-effusion lymphoma and Kaposi’s sarcoma in a cardiac-transplant recipient. N Engl J Med 1998; 339: 444–449.

Sarosiek KA, Cavallin LE, Bhatt S, Toomey NL, Natkunam Y, Blasini W et al. Efficacy of bortezomib in a direct xenograft model of primary effusion lymphoma. Proc Natl Acad Sci U S A 2010; 107: 13069–13074.

Bhatt S, Ashlock BM, Natkunam Y, Sujoy V, Chapman JR, Ramos JC et al. CD30 targeting with brentuximab vedotin: a novel therapeutic approach to primary effusion lymphoma. Blood 2013; 122: 1233–1242.

Cesarman E, Moore PS, Rao PH, Inghirami G, Knowles DM, Chang Y. In vitro establishment and characterization of two acquired immunodeficiency syndrome-related lymphoma cell lines (BC-1 and BC-2) containing Kaposi’s sarcoma-associated herpesvirus-like (KSHV) DNA sequences. Blood 1995; 86: 2708–2714.

Gao SJ, Kingsley L, Li M, Zheng W, Parravicini C, Ziegler J et al. KSHV antibodies among Americans, Italians and Ugandans with and without Kaposi’s sarcoma. Nat Med 1996; 2: 925–928.

Lee BS, Connole M, Tang Z, Harris NL, Jung JU. Structural analysis of the Kaposi’s sarcoma-associated herpesvirus K1 protein. J Virol 2003; 77: 8072–8086.

GSEA analysis identifies activation of interferon signaling in PEL by lenalidomide

To delineate the gene network affected by lenalidomide, BC-3 and BCBL-1 cells were treated with lenalidomide (5 μM) for 24 hours (h) followed by genome-wide microarray analysis. Unsupervised hierarchical clustering separated samples according to their treatment group, indicating a common transcriptional response to treatment with lenalidomide in PEL (Figure 1C). Rather than inducing non-specific changes in gene expression, lenalidomide induced significant changes in a limited number of genes. Thus, there were 992 genes (390 down- and 602 up-regulated genes) whose expression were changed significantly (p<0.05) in both the cell lines. We used a Gene Set Enrichment Analysis (GSEA) program to identify functional gene sets that were enriched in PEL cells upon treatment with lenalidomide.[21] Among the gene signatures identified by this analysis were gene sets containing genes that are known targets of interferon (IFN) and MYC signaling pathways (Figure 1D and Supplementary Figure S2B). We used qRT-PCR to confirm up-regulation of IFNs and interferon specific genes (ISGs) by lenalidomide in PEL (Supplementary Figure S2C).

Interferons α, β & γ are cytotoxic to PEL but are not essential for the anti-proliferative effect of IMiDs

In the case IMiDs block the proliferation of PEL by activating the IFN pathway, then treatment with recombinant IFNs (rIFNs) should mimic the effect of IMiDs. To test this hypothesis, we treated a panel of cell lines with increasing concentrations of rIFNs α, β and γ. All the PEL cell lines were sensitive to recombinant IFNs α, β or γ (Figure 2A). In particular, BC-3, BCBL-1 and JSC-1 were highly sensitivity to IFNs α and β. Although BC-1 and VG-1 cells were relatively resistant to IFNs α and β, they were sensitive to IFN-γ. In contrast, DG-75 and BJAB, the two IMiD-resistant cell lines, showed little or no inhibitory effect upon treatment with any IFN (Figure 2A).

Figure 2

PEL cells are sensitive to interferons (IFNs) α, β and γ. A, BC-3, BCBL-1, JSC-1, BC-1, VG-1, BJAB and DG-75 cells were treated with indicated concentrations of recombinant IFNs for 5 days, and cell viability was measured using an MTS assay. The values shown are mean±SE (n=3). B, Blocking of interferons α, β, and γ (IFNs αβγ) together did not block the anti-proliferative activity of IMiDs in PEL. BC-3 and BCBL-1 were treated with indicated concentrations of IMiDs, IFNs αβγ and IFNs αβγ blocking antibodies combined (Block Abs Combi) for 4 days. IFN-α blocking antibody was used at a concentration which blocks 450 U/ml of IFN- α by 50%, IFN-β blocking Antibody was used at a concentration which blocks 350 U/ml of IFN- α by 50% and IFN-γ blocking antibody was used at a concentration which blocks 1090 U/ml of IFN-γ by 50%. Isotype antibodies (Iso Ab) corresponding to same species was as used as control. The values shown are mean±SE (n=3). C, Interferons are not secreted into the supernatants of PEL cells upon treatment with IMiDs. BC-3 and BCBL-1 cells were treated with dimethyl-sulfoxide (DMSO, vehicle control), lenalidomide 5 μM (Len), and pomalidomide 500 nM (Pom) for 48 hours. Recombinant Interferons (rIFN) – α, β, and γ was used at a concentration of 100, 200, and 1000 picograms/mL, respectively as positive controls.

Interestingly, the neutralizing antibodies against IFNs α, β and γ, when used singly (Supplementary Figure S3) or in combination (Figure 2B), did not block the anti-proliferative effect of IMiDs against PEL although they effectively blocked the anti-proliferative effect of their respective rIFNs. Further none of the interferons were secreted into the supernatants of the PEL cells treated with IMiDs (Figure 2C). Collectively, these results suggest that activation of the interferon pathway is not solely responsible for the anti-proliferative effect of IMiDs against PEL.

IMiDs have no effect on KSHV lytic replication in PEL

It was conceivable that induction of KSHV lytic replication contributed to the cell death and activation of IFN signaling observed following treatment with IMiDs. KSHV Replication and Transcription Activator (RTA), is a master regulator and marker for lytic reactivation.[22] Treatment of PEL cell lines with IMiDs failed to induce RTA expression, as determined by immunoblotting (Supplementary Figure S4A). Additionally, we failed to detect infectious virions in the supernatant from IMiDs-treated PEL cells when assayed on 293-PAN-Luc reporter cell line (Supplementary Figure S4B).[23] Thus, IMiDs do not induce lytic reactivation of KSHV.

IMiDs down-regulate IRF4 expression in PEL

MM cells are addicted to IRF4[24] and the anti-proliferative activity of lenalidomide and pomalidomide in myeloma and ABC-DLBCL is associated with down-regulation of IRF4.[25, 26] To delineate the role of IRF4 in the survival of PEL and in their response to IMiDs, we used western blotting to compare its expression in a panel of 35 cell lines comprising 11 hematologic malignancies, including 9 PEL cell lines. Expression of IRF4 was robust in all cell lines derived from PEL, MM, ABC-DLBCL, Waldenstrom macroglobulinemia and Hodgkin lymphoma, but was weak to absent in the cell lines derived from other hematologic malignancies (Figure 3A). Treatment of BC-3, BCBL-1 and JSC-1 with IMiDs resulted in significant decrease in the expression of IRF4 and its downstream target MYC, thus suggesting that IMiDs exert their cytotoxicity towards PEL by down-regulating IRF4 (Figure 3B). In contrast IMiDs had no significant effect on the levels of IRF4 and MYC in DG-75 cells (Figure 3B).

Figure 3

Uniform expression of IRF4 in PEL. A, Expression of IRF4 in a panel of 35 hematological cancer cell lines. Cell lysates were prepared from logarithmically growing cell lines and blotted for IRF4 and GAPDH. Blots are representative of 3 individual experiments. PEL: Primary Effusion Lymphoma; CML: Chronic Myelogenous Leukemia; T-ALL: T-cell Acute Lymphoblastic Leukemia; AML: Acute Myelogenous Leukemia; ABC-DLBCL: Activated B-Cell Diffuse Large B-Cell Lymphoma; GCB-DLBCL: Germinal Center B-cell Diffuse Large B-Cell Lymphoma; MCL: Mantle Cell Lymphoma; WM: Waldenstrom Macroglobulinemia; MW: Molecular Weight; kDa: Kilodalton. B, Immunoblot analysis showing the effect of lenalidomide (Len) and pomalidomide (Pom) at the indicated doses for 48 h on the expression of IRF4, MYC and TUBA (Tubulin, loading control) in BC-3, BCBL-1, JSC-1 and DG-75 cells. Blots are representative of 3 individual experiments.

PEL cells have constitutive NF-κB activity due to the presence of KSHV viral proteins[27-29] and aberrant NF-κB activity has been shown to up regulate the expression of IRF4[30]. To test whether IMiDs represses IRF4 expression by inhibiting NF-κB pathway, BC-3 and BCBL-1 cells stably expressing an NF-κB promoter-driven luciferase reporter construct were treated with increasing concentrations of IMiDs. As shown in Supplementary Figure S5A, IMiDs failed to block NF-κB promoter-driven luciferase activity. Further, no change in the secretion of IL-6, a known target of classical NF-κB pathway[31], processing of p100 into p52, and expression of NF-κB pathway proteins, were observed in IMiDs-treated PEL cells (Supplementary Figure S5B–C). Taken collectively, these results demonstrate that IMiDs have no significant effect on the constitutive NF-κB activity present in PEL cells.

PEL cells are addicted to IRF4 for survival

To provide genetic evidence in support of the hypothesis that IMiDs exert their cytotoxic effect against PEL through down-regulation of IRF4, we studied the effect of IRF4 knockdown in BC-3 cells. For this purpose, we generated a polyclonal population of BC-3 cells expressing a tetracycline-inducible-H1 (TO/H1) promoter-driven shRNA targeting IRF4 (shIRF4) (Supplementary Figure S6A–B) followed by generation of single cell clones (Supplementary Figure S6C). Upon treatment with doxycycline (Dox) significant down-regulation of IRF4 was observed in a number of clones (Supplementary Figure S6C). Down-regulation of MYC, a target of IRF4,[24] and cleavage of PARP upon Dox-treatment was observed only in those clones where IRF4 was down-regulated (Figure 4A and Supplementary Figure S6C). Strikingly, Cellular proliferation was decreased rapidly only in clones were IRF4 is down-regulated upon treatment with Dox (Figure 4B and Supplementary Figure S6D). Treatment of a clone, BC-3-shIRF4-F11 with Dox resulted in G1 cell-cycle arrest (Figure 4C) and appearance of cells with condensed and fragmented nuclei suggestive of apoptosis (Figure 4D), a finding further confirmed by staining with annexinV/propidium iodide (Figure 4E). In contrast, Dox treatment had no significant effects on cell-cycle progression and apoptosis in BC-3-shSCR cells (Figure 4C–E). Collectively, the above results suggest that down-regulation of IRF4 is toxic to BC-3 cells by inhibiting cell-cycle progression and through induction of apoptosis.

Figure 4

PEL cells are addicted to IRF4. A, BC-3 cells stably expressing tetracycline-inducible H1 promoter (H1/TO)-driven shRNA targeting IRF4, clone F11 (shIRF4-F11) and shRNA targeting scrambled sequence (shSCR) were treated with doxycycline (Dox, 500 ng/ml) for 3 days and immunoblotted for the expression of IRF4, MYC, PARP and TUBA. B, BC-3 cells stably expressing shSCR and shIRF4-F11 were treated with Dox for indicated time points and cell viability was measured by MTS assay. The values shown are mean±SE of a representative experiment performed in triplicate for 3 times. C, Cell cycle analysis of BC-3 cells stably expressing shSCR and shIRF4-F11 treated with and without Dox for 48 h. Cells were stained with propidium iodide (PI) and analyzed by flow cytometry. Data is representative of 2 individual experiments. D, BC-3 cells stably expressing shSCR and shIRF4-F11 were treated with Dox for 72 h. Cells were then stained with Hoescht 33342 (50 μg/ml) and photographed. E, BC-3 cells stably expressing shSCR and shIRF4-F11 were treated with Dox for 48h, stained with annexinV-FITC/PI, and analyzed for apoptosis by flow cytometry. Data is representative of 2 individual experiments.

IMiDs rapidly down-regulate IKZF1 and silencing of IKZF1 is toxic to PEL

Ikaros family proteins IKZF1 and IKZF3 are B cell transcription factors that play crucial roles in immunity and cell-fate decisions.[32] Recently, it was shown that IMiDs selectively degrade these transcription factors in MM cells.[10, 11] In PEL, both IMiDs led to significant and near complete down-regulation of IKZF1 in all the three PEL cell lines even at the lowest concentration (i.e. 0.5 μM lenalidomide and 50 nM pomalidomide) tested, but had only a modest effect in the DG-75 cell line (Figure 5A). In contrast, the effect of IMiDs on the level of expression of IKZF3 was modest at best and, in general, required higher doses of the drugs (Figure 5A). Consistent with the results seen with IMiDs, silencing of IKZF1 by two different shRNAs were selectively toxic to PEL cells (Figure 5B and Supplementary Figure S7A), and was accompanied by partially reduced expressions of IRF4 and MYC (Figure 5C). Additional studies revealed that IMiDs down-regulate IKZF1 expression at the post-translational level (Supplementary Figure S7B–C). Furthermore, time-course experiments revealed rapid and near complete down-regulation of IKZF1 expression as early as 12 h post-treatment even at the lowest concentrations of both IMiDs (Figure 5D). In contrast, the levels of IRF4 and MYC were less sensitive to down-regulation by IMiDs (Figure 5D). Thus, near complete down-regulation of these proteins was either not observed or required treatment with longer duration (i.e. 48 h) and higher concentrations of the drugs (Figure 5D). Collectively, these results support the hypothesis that IKZF1 is an upstream target of IMiDs in PEL.

Figure 5

IMiDs rapidly down-regulate IKZF1 and silencing of IKZF1 is toxic to PEL. A, Immunoblot analysis showing the effect of treatment with lenalidomide (Len) and pomalidomide (Pom) at the indicated doses for 48 h on the expression of IKZF1, IKZF3 and GAPDH (loading control) in BC-3, BCBL-1, JSC-1 and DG-75 cells. Blots are representative of 2 individual experiments. B, Change in % red fluorescent protein (RFP) positivity over time in BC-3 and BCBL-1 cells infected with viruses encoding RFP and the indicated shRNAs. The day 2 %RFP for each virus was normalized to 1, and subsequent values are expressed relative to cells infected with a virus encoding RFP and a control shRNA. Data is representative of 2 individual experiments. C, Immunoblot analysis of BC-3 and BCBL-1 cells transiently infected with lentiviruses expressing the indicated shRNAs for 72 hours. Immunoblots were quantified (normalized to the expression of GAPDH) using image studio version 5.0 from LI—COR biosciences. Blots are representative of 2 individual experiments. D, Immunoblot analysis showing the expression IKZF1, IRF4, MYC, TUBA and HSP90 (loading controls) in BC-3 and BCBL-1 cells treated with indicated concentrations of IMiDs for 12, 24, 48, and 72 h. Blots are representative of 2 individual experiments.

We also checked the hypothesis that IKZF1 may be responsible for the high level expression of IRF4 observed in PEL cells. We found that IRF4 and IKZF1 are both consistently expressed in PEL, myeloma, Waldenstrom macroglobulinemia, ABC-DLBCL and Hodgkin’s lymphoma cell lines. However, there was little correlation between IRF4 and IKZF1 expression in cell lines derived from other hematologic malignancies (Supplementary Figure S8). Therefore, while it is possible that IKZF1 may contribute to the overexpression of IRF4 (and MYC) in PEL, it is unlikely to be the sole regulator of their expression.

CRBN is dispensable for the survival of PEL

IMiDs exert their anti-proliferative effect by binding to their cellular protein target CRBN.[7-9] However, we failed to observe a significant and consistent effect of IMiDs on the expression of CRBN in BC-3 and BCBL-1 cells (Figure 6A). It has been shown that silencing of CRBN by shRNA significantly decreases the proliferation of MM[8] and ABC-DLBCL cells.[25] We generated polyclonal populations of BC-3 and BCBL-1 cells stably expressing a TO/U6 promoter-driven shRNA targeting CRBN (shCRBN).[10] Treatment of shCRBN-expressing cells with Dox for 4 days significantly down-regulated the expression of CRBN (Figure 6B, Upper panel), while Dox-treatment was without effect in control shRNA-expressing cells (Figure 6B, Upper panel). Interestingly, silencing of CRBN expression did not have any significant effect on the proliferation of BC-3 and BCBL-1 cells (Figure 6B, lower panel). Thus, in contrast to myeloma and ABC-DLBCL cells, CRBN is dispensable for the survival of PEL.

Figure 6

CRBN is dispensable for the survival of PEL cells but is essential for the anti-proliferative activity of IMiDs in PEL cells. A, Immunoblot analysis showing the effect of lenalidomide (Len) and pomalidomide (Pom) at the indicated doses for 48 h on the expression of cereblon (CRBN) and GAPDH in BC-3 and BCBL-1 cells. The band corresponding to CRBN is marked with an asterisk. Blots are representative of 2 individual experiments. B, Upper panel: BC-3 and BCBL-1 cells stably expressing tetracycline-inducible shRNA targeting CRBN (shCRBN) and shRNA targeting scrambled sequence (shCON) were treated with doxycycline (Dox, 500 ng/ml) for 4 days and immunoblotted for the expression of CRBN, GAPDH and TUBA. Blots are representative of 2 individual experiments. Lower panel: BC-3 and BCBL-1 cells stably expressing shCON and shCRBN were treated with Dox for indicated time points and cell viability was measured by MTS assay. The values shown are mean±SE of a representative experiment performed in triplicate for 2 times. C, BC-3 and BCBL-1 cells stably expressing shCON and shCRBN were pre-treated with Dox for 3 days followed by treatment with vehicle and IMiDs at indicated concentrations for 6 days in the presence of Dox and cell viability was measured by MTS assay. The values shown are mean±SE of a representative experiment performed in triplicate for 3 times. D, BC-3 and BCBL-1 cells stably expressing shCON and shCRBN were pre-treated with Dox for 3 days followed by treatment with vehicle and IMiDs at indicated concentrations for 48 h in the presence of Dox and cell lysates were collected and immunoblotted for indicated proteins. Blots are representative of 2 individual experiments. E, BC-3 and BCBL-1 cells stably expressing shCON and shCRBN were pre-treated with Dox for 3 days followed by treatment with vehicle or IMiDs along with Dox in the presence of 100 μg/ml of cycloheximide (CHX) for 0, 1, 2, and 3h respectively. Whole cell lysates were immunoblotted for IKZF1, CRBN and GAPDH. Blots are representative of 2 individual experiments. Note: The CRBN antibody gives a non-specific band when CRBN is probed as first antigen but when the blot is probed for some other antigen then stripped and probed for CRBN then the intensity of the non-specific band is decreased or gone.

CRBN is essential for the anti-proliferative effect of IMiDs in PEL

We next asked the question if CRBN is essential for the activity of IMiDs in PEL. Although IMiDs significantly inhibited the proliferation of shCON-expressing PEL cells, the anti-proliferative activity of IMiDs was almost completely blocked in shCRBN-expressing PEL cells (Figure 6C). Furthermore, treatment with IMiDs failed to induce G1 cell-cycle arrest in shCRBN-expressing PEL cells, but successfully did so in shCON-expressing PEL cells (Supplementary Figure S9). Further, IMiDs treatment resulted in near complete abrogation of IKZF1 expression in shCON-expressing BC-3 and BCBL-1, which was accompanied by a significant decrease in the expressions of IRF4 and MYC but was without any effect on the expression of CRBN (Figure 6D). Remarkably, IMiDs had no significant effect on the expression levels of IKZF1, IRF4 and MYC in the shCRBN- expressing BC-3 and BCBL-1 cells. (Figure 6D). In addition, CRBN is essential for the post-translational degradation of IKZF1 by IMiDs, as observed by a complete block in the degradation of IKZF1 by IMiDs in shCRBN-expressing cells (Figure 6E). Whereas, IKZF1 was degraded within 1 h by IMiDs in shCON-expressing cells (Figure 6E). These results clearly suggest that CRBN is essential for the anti-proliferative potential of IMiDs in PEL. However, we did not observe a significant difference in the level of expression of CRBN between IMiDs-sensitive and-resistant cell lines (Supplementary Figure S10), suggesting that the resistance to IMiDs in these cells is not linked to CRBN expression.

Knocking down MYC by shRNA enhances the sensitivity of IMiDs to PEL

To test if the loss of MYC could synergize with IMiDs, we generated polyclonal population of BC-3 cells stably expressing TO/H1-driven shRNAs targeting MYC (shMYC) and a scrambled sequence (shSCR). Consistent with our published results[18], treatment of BC3-shMYC cells with Dox resulted in a significant down-regulation of MYC (Figure 7A), which was accompanied by a decrease in cell proliferation (Figure 7B), while without any effect on BC-3-shSCR cells (Figure 7A–B). Interestingly, knockdown of MYC significantly enhanced the anti-proliferative effect of IMiDs (Figure 7C), which was accompanied by cell-cycle arrest and apoptosis (Figure 7D–E). In contrast, no significant difference in cell proliferation, cell-cycle progression and apoptosis was observed in BC-3 shSCR cells treated with IMiDs in the presence or absence of Dox (Figure 7C–E). These results pointed to the existence of a potential synergism between IMiDs and inhibition of MYC.

Figure 7

Knocking down MYC enhances the anti-proliferative effect of IMiDs in PEL, BRD4 inhibitors and IMiDs display synergistic anti-proliferative activity against PEL. A, BC-3 cells stably expressing tetracycline-inducible H1 promoter (H1/TO)-driven shRNA targeting MYC (shMYC) and shRNA targeting scrambled sequence (shSCR) were treated with doxycycline (Dox, 500 ng/ml) for 4 days and immunoblotted for the expression of MYC and GAPDH. Blots are representative of 2 individual experiments. B, BC-3 cells stably expressing shSCR and shMYC were treated with Dox for indicated time points and cell viability was measured by MTS assay. The values shown are mean±SE of a representative experiment performed in triplicate for 3 times. C, BC-3 cells stably expressing shSCR and shMYC were treated in the presence/absence of Dox with indicated concentrations of IMiDs or vehicle for 72 h and cell viability was measured by MTS assay. Asterisks (***) indicate significance at the level of p≤0.001. The values shown are mean±SE of a representative experiment performed in triplicate for 3 times. D, Cell cycle analysis of BC-3 cells stably expressing shSCR and shMYC that were treated in the presence/absence Dox with indicated concentrations of IMiDs or vehicle for 72 h. Data is representative of 2 individual experiments. E, Apoptosis analysis of BC-3 cells stably expressing shSCR and shMYC that were treated in the presence/absence Dox with indicated concentrations of IMiDs or vehicle for 48 h. Data is representative of 2 individual experiments. F, BC-3 and BCBL-1 cells were treated with low doses of lenalidomide (Len) in combination with low doses of three structurally different BRD4 inhibitors (JQ-1, IBET151 and PFI-1) for 5 days and then assessed for viability using MTS assay. Combination index (CI) was calculated using the calcusyn software which is based on the method of Chou and Talalay.[33] Each BRD4 inhibitor was tested in combination with lenalidomide at 12 different combinations (please see supplementary Tables 2–8 for details). CI values of <1 denotes synergism and CI values >1 denotes antagonism. Data presented is representative of 3 individual experiments performed in triplicate.

BRD4 inhibitors JQ-1, IBET151 and PFI-1 are synergistic with IMiDs in PEL

Recently, it has been shown that MYC transcription can be targeted using BRD4 inhibitors.[16-18] JQ-1, IBET151 and PFI-1 are three structurally distinct BRD4 inhibitors.[13-15] To test whether IMiDs show synergistic anti-proliferative activity when combined with BRD4 inhibitors, BC-3 and BCBL-1 cells were treated with low doses of lenalidomide in combination with low doses of JQ-1, IBET151 and PFI-1, respectively. The combination index (CI) was calculated using the calcusyn software, which is based on the method of Chou and Talalay.[33] Lenalidomide was highly synergistic with all BRD4 inhibitors at all the combination doses tested in both BC-3 and BCBL-1 cells (Figure 7F and Supplementary Tables 2–8). In contrast, lenalidomide is not synergistic with the inactive isomer of JQ-1 in either cell line (Supplementary Tables 9–10). Furthermore, combined treatment with lenalidomide and JQ-1 in BC-3 and BCBL-1 cells significantly decreased the expression of MYC and IRF4 at both protein (Figure 8A) and mRNA (Supplementary Figure S11A) levels as compared to treatment with either drug alone. The combination of lenalidomide with JQ-1 also resulted in G1 cell-cycle arrest, cleavage of PARP, and appearance of apoptotic cells as compared to treatment with either drug alone (Supplementary Figure S11B and Figures 8A–B, respectively).

Figure 8

A, BC-3 and BCBL-1 cells were treated with vehicle, lenalidomide 1 μM (L1), pomalidomide 100 nM (P100), JQ-1 50 nM (J50), JQ-1 100 nM (J100) and the combinations for 48 h. Whole cell lysates were immunoblotted for MYC, IRF4, PARP and GAPDH. Blots are representative of 3 independent experiments. B, Apoptosis analysis of BC-3 and BCBL-1 cells treated with vehicle or indicated concentrations of IMiDs and JQ-1 or the combination for 72 h. Data presented is representative of 2 individual experiments. C, BC-3 cells stably expressing tetracycline-inducible H1 promoter (H1/TO)-driven shRNA targeting BRD4 (shBRD4) and shRNA targeting scrambled sequence (shSCR) were treated with doxycycline (Dox, 500 ng/ml) for 4 days and immunoblotted for the expression of BRD4 and GAPDH. The band corresponding to BRD4 is marked with an asterisk. Blots are representative of 2 individual experiments. D, BC-3 cells stably expressing shSCR and shBRD4 were treated in the presence/absence Dox with indicated concentrations of IMiDs or vehicle for 4 days and cell viability was measured by MTS assay. The values shown are mean±SE of a representative experiment performed in triplicate for 2 times. E, Body weight gain of mice injected with BC-3 cells followed by indicated treatments (n=7 in each group) over the period of experiment. Statistically significant differences (on day 27 of the treatment) are shown by asterisks (*) and (**) at the levels of P≤0.05 and 0.01, respectively. The black arrows in the X-axis mark the start (day1) and end (day28) of the treatment. F, Survival curves (Kaplan-Meier) of mice injected with BC-3 cells followed by indicated treatments (n=7 in each group). The survival curve was generated in GraphPad Prism 5 software and statistical values for the curves are calculated by log rank (Mantel-Cox) test. Asterisks (*) and (**) indicate significance at the level of p≤0.05 and 0.01, respectively.

Knocking down BRD4 by shRNA enhances the sensitivity of IMiDs to PEL

To confirm whether the observed synergism between lenalidomide and the BRD4 inhibitors is due to inhibition of BRD4, we generated stable clones of BC-3 and BCBL-1 cells expressing TO/H1 promoter-driven shRNA targeting BRD4 (shBRD4). Treatment of BC-3 and BCBL-1 cells expressing shBRD4 with Dox resulted in a significant down-regulation of BRD4 (Figure 8C and Supplementary Figure S11C) and decrease in cellular proliferation (Figure 8D and Supplementary Figure S11D), while Dox treatment had no effect on BRD4 expression or cellular proliferation in BC-3 and BCBL-1 cells expressing a scrambled shRNA sequence (shSCR) as control (Figure 8C–D and Supplementary Figure S11C–D). More importantly, Dox enhanced the anti-proliferative activity of IMiDs in the shBRD4-expressing BC-3 and BCBL-1 cells but was without effect in shSCR-expressing cells (Figure 8D and Supplementary Figure S11D).

Lenalidomide and JQ-1 are synergistic against PEL in vivo

To check the in vivo efficacy of lenalidomide, alone and in combination with JQ-1, BC-3 cells were injected into the intra-peritoneal cavity of NOD.SCID mice. Five days after the injection, animals were randomly assigned to vehicle control, lenalidomide (50 mg/kg once daily for 28 days), JQ-1 (50 mg/kg once daily for 28 days), and the combination. Intraperitoneal inoculation of BC-3 cells resulted in rapid tumor growth and massive ascites, which resulted in weight gain (Figure 8E). There was a significant reduction (p≤0.01) in body weight gain (a measure of ascites)[34] of animals treated with lenalidomide and JQ-1 when compared to vehicle control (Figure 8E). Additionally, the combination of lenalidomide and JQ-1 showed a further reduction in body weight gain over time when compared with mice treated with either agent alone (Figure 8E). Furthermore, the median survival of mice that received combination treatment (51 days) was significantly (p≤0.01) increased as compared with the median survival of mice treated with lenalidomide (35 days) or JQ-1(42 days) alone (Figure 8F).

Discussion

In this report, we demonstrate that majority of the PEL cells are highly sensitive to lenalidomide and pomalidomide, two FDA-approved drugs for the treatment of MM. Both drugs have predictable and manageable safety profiles and limited cumulative long-term toxicity,[35, 36] making them attractive treatment options for PEL.

The anti-proliferative action of lenalidomide in PEL cell lines was associated with the activation of the IFN signaling pathway. However, we did not detect IFN in the supernatant of IMiDs-treated cells and neutralizing antibodies against IFNs failed to block the activity of IMiDs against PEL. Since IRF4 has been recently shown to modulate IFN signaling,[25] these results prompted us to explore the role of IRF4 in the anti-proliferative effects of IMiDs in PEL. We observed that IRF4 is not only uniformly expressed in PEL cell lines but is significantly down-regulated following treatment with IMiDs. Furthermore, shRNA-mediated silencing of IRF4 was toxic to PEL cells, thereby supporting the argument that down-regulation of IRF4 contributes to the anti-proliferative effect of IMiDs in PEL. Expression of IRF4 in myeloma cells has been attributed to their plasmacytic differentiation.[24] As PEL cells resemble plasma cells in the gene expression profile,[37] the uniform expression of IRF4 in these cells may also reflect their plasma cell lineage. IRF4 expression is also a feature of ABC-DLBCL.[25, 38] In these cells oncogenic mutations affecting the B-cell receptor (BCR) and MYD88 signaling pathways induce NF-κB,[39, 40] which is a strong inducer of cytotoxic IFNβ.[41] IRF4, however, places a brake on IFNβ production by repressing IRF7, thereby allowing ABC-DLBCL to survive and proliferate.[25] Furthermore, IRF4 is believed to promote ABC-DLBCL survival by transactivating CARD11 and potentiating NF-κB signaling.[25] Although oncogenic mutations affecting the BCR and MYD88 signaling pathways have not been reported in PEL, they do possess constitutively active NF-κB signaling pathway due to the activity of KSHV-encoded viral FLICE Inhibitory Protein (vFLIP) K13.[27, 28, 42] Therefore, similar to ABC-DLBCL, IRF4 may be up-regulated in PEL cells to augment the pro-survival aspect of NF-κB signaling while simultaneously protecting against the deleterious effects (e.g. IFNβ production) of uncontrolled NF-κB activation. IRF4 is also known to bind to MYC promoter and stimulate MYC gene expression.[24] Even though PEL cells lack structural alterations in the MYC gene,[43] they nevertheless demonstrate elevated MYC expression, which has been shown to be essential for their survival and proliferation.[44] The over-expression of IRF4 in PEL might contribute to the elevated MYC expression observed in these cells.

IMiDs were shown to degrade both IKZF1 and IKZF3 in MM.[10, 11] In contrast, we observed that IKZF1 was the primary target of IMiDs in PEL cells. The expression of IKZF1 was down-regulated earlier than IRF4 and MYC. Additionally, IKZF1-specific shRNAs was not only toxic to PEL cells but also partially down-regulated the expression of IRF4 and MYC. IKZF1 has been previously shown to bind to IRF4 promoter and regulate its expression at the transcriptional level.[10] Furthermore, MYC is a known transcriptional target of IRF4.[24] Taken collectively with prior studies, our results suggest that degradation of IKZF1 by IMiDs down-regulates IRF4 expression at the transcriptional level, which in turn, downregulates MYC expression.

CRBN is the direct cellular binding-target of IMiDs[7] and essential for their immunomodulatory and antiproliferative activities.[8, 9] We observed that CRBN is dispensable for the survival of PEL, which is in contrast to the situtation in MM and ABC-DLBCL cells where shRNA-mediated knock down of CRBN has been reported to be toxic.[8, 25] However, while CRBN is not essential for the survival of PEL, it is essential for the anti-proliferative activity of IMiDs in PEL since all the IMiDs induced anti-PEL effects are blunted in cells expressing an shRNA targeting CRBN.

Our study along with work of others[26] suggest that MYC is one of the down-stream target of IMiDs in PEL. We found that shRNA-mediated knockdown of MYC enhanced the anti-proliferative effect of IMiDs on PEL, thus suggesting a potential synergism between IMiDs and inhibition of MYC. BRD4 inhibitors have been shown to block MYC expression.[16-18] In support of this premise, we observed striking synergy between low doses of lenalidomide and BRD4 inhibitors (JQ-1, IBET151 and PFI-1) against PEL. Furthermore, shRNA-mediated BRD4 knockdown also enhanced the cytotoxicity of IMiDs towards PEL suggesting that the synergism observed between IMiDs and BRD4 inhibition may not be limited only to the BRD4 inhibitors used in our study. There are several potential explanations for the observed synergism between IMiDs and BRD4 inhibitors. First, since inhibition of MYC is not complete upon treatment with lower doses of IMiDs (Figure 5D); addition of low doses of BRD4 inhibitors may eliminate any residual MYC expression seen following IMiDs treatment. Second, apart from MYC, BRD4 inhibitors are known to modulate the expression of other genes[45], which may have synergistic cytotoxicity when combined with IMiDs. Finally, in addition to degrading IKZF1/IKZF3 via CRBN, IMiDs may also degrade other proteins, which may result in synergistic cytotoxicity when combined with BRD4 inhibitors.

In summary, we provide strong in vitro and in vivo data showing that IMiDs are effective against PEL and combined treatment of IMiDs with BRD4 inhibitors have synergistic activity against this deadly incurable cancer. BRD4 inhibitors have shown promising activity against multiple cancers in pre-clinical studies and at present there are 5 BRD4 inhibitors in phase 1–2 clinical trials.[12] Our results suggest that IMiDs, alone and in combination with BRD4 inhibitors, deserve further testing for the treatment of PEL. While this study was in its final stage of completion, a case report was published describing the successful treatment of a PEL patient with lenalidomide,[46] which supports our pre-clinical data.

Materials and methods

Cell lines

BC-3, BCBL-1, JSC-1, BC-1, BCP-1, VG-1 and APK-1 were obtained from Dr. Jae Jung (University of Southern California, CA, USA). UMPEL-1 and UMPEL-3 were provided by Drs. Izidore Lossos and Juan Ramos, respectively (both from University of Miami, FL, USA). DG-75 was obtained from Dr. Alan Epstein (University of Southern California, CA, USA). All the cells were grown in conditions as described previously[47]. The cell lines were expanded, stored in liquid nitrogen and used within 3 months after resuscitation. The identities of the PEL cell lines were routinely authenticated by western blotting detection of KSHV LANA. No further authentication of cell lines characteristics was done. The authentication information for remaining cell lines is not available.

Cell viability, cell-cycle, apoptosis, luciferase assays, and western blotting

Cell viability, cell-cycle, apoptosis, luciferase assays and western blotting were performed as described earlier[47-51].

Lentiviral shRNA constructs

shRNA oligonucleotides (Supplementary Table 11) directed against human IRF4, MYC and BRD4 mRNAs were annealed and cloned into a modified pENT entry vector containing a TO/H1 promoter as described previously[18]. Lentiviral shRNAs for CRBN, IKZF1-1, and IKZF1-2 along with their respective controls were obtained from Dr. Willian Kaelin Jr. (Harvard University, MA, USA).[10]

Real-time RT-PCR

Real-time quantitative reverse transcript-polymerase chain reaction (qRT-PCR) was performed as described earlier[47] using gene-specific PCR primers listed in Supplementary Table 12.

PEL Orthotopic tumor model

A total of 2×107 BC-3 cells were injected intraperitoneally into female NOD.SCID mice (NCI Frederick, 6 weeks old). 5 days later the mice were randomly divided in to 4 groups (n=7 each). Investigators are not blinded. Vehicle control (10 % Hydroxypropyl-β-cyclodextrin), lenalidomide 50 mg/kg b.w. (once daily), JQ-1 50 mg/kg b.w. (once daily) and the combination were administered intraperitoneally for 28 days. Then the animals were monitored for survival. Body weight gain was measured once in 3 days as a surrogate measure of tumor progression [34]. The experiments were performed following the approval of institutional animal ethics committee of University of Southern California.

Statistical analysis

Two-tailed unpaired Student’s t test was used to test for differences between two groups. Differences with a p value ≤ 0.05 were considered statistically significant. The data’s were given as mean±SE. All the experiments were reproduced at-least twice. No inclusion/exclusion criteria are applied and none of the samples or animals was excluded from the analysis. The vehicle and drug treatments were performed at the same time in same condition. The investigators are not blinded for any of the experiments.

Detailed information about materials and methods is provided in the Supplementary Information.