Oncolytic Reactivation of KSHV as a Therapeutic Approach for Primary Effusion Lymphoma Free

Gamma herpesviruses are responsible for a substantial proportion of virus-associated human cancers, particularly in immunocompromised individuals (1, 2). Kaposi's sarcoma-associated herpesvirus (KSHV), also known as human herpesvirus-8, is one of the tumorigenic viruses with a large double-stranded DNA genome. KSHV has been linked to Kaposi's sarcoma (KS) as well as primary effusion lymphoma (PEL), or body-cavity B-lymphoma (BCBL), and a subset of multicentric Castleman's disease.

The aggressive nature of PEL is evident by the poor median overall survival for this cancer, 10.2 months in one recent study in which patients received a multidrug cytotoxic regimen (3). Given the limited treatment options for this Non-Hodgkin's lymphoma, novel, rationally designed therapeutic approaches are urgently needed. Although the introduction of antiretroviral therapy (i.e., highly active antiretroviral therapy, HAART) has reduced the incidence of KS lesions in HIV patients, KSHV-associated complications remains a significant problem. The absence of HAART before PEL diagnosis is also associated with poor outcome in a multivariate analysis (4), indicating a strong correlation of cancer progression with HIV replication. This is supported by the finding that complete remissions have been reported after treatment of PEL patients with HAART therapy alone (5–7). PEL may also be responsive to Zidovudine (AZT) alone or AZT in combination with IFNα (8, 9). Accordingly, using HAART to target HIV concomitantly with targeted therapy to inhibit PEL growth should be clinically beneficial, as has been suggested elsewhere (6, 7, 10).

Ingenol-3-angelate (PEP005), an FDA-approved drug for topical treatment of actinic keratosis, is a natural extract from the plant, Euphorbia peplus, and has also been found to suppress the growth of some cancer cell types, including melanoma, acute myeloid leukemia, and non-Hodgkin B-cell lymphoma (NHL; refs. 11–13). Mechanistically, PEP005 is an agonist of protein kinase C and activates NF-κB. Treatment of cancer cell lines with PEP005 has been shown to induce apoptosis (13). Furthermore, we recently found that PEP005 robustly reactivated latently-infected HIV in cells purified from patients undergoing HAART (14), and that combined treatment of PEP005 with an epigenetic drug, the bromodomain and extra-terminal (BET) family protein inhibitor, JQ1, synergistically increased reactivation of HIV (14). Accordingly, PEP005 represents a new group of lead compounds for combating HIV latency (14).

JQ1 is a small molecule, which was developed as a mimetic to an acetylated histone tail (15). This then competes with, and antagonizes, bromo-domain containing proteins (mainly BRD2 and 4) for binding to acetylated histone tails, which consequently mutes the expression of pro-inflammatory genes in particular (16–18). BRD2/4 binds to the C-terminal positive transcription elongation factor b (pTEFb), which is a heterodimer composed of cyclin T and CDK9 (19, 20). The pTEFb factor phosphorylates Ser-2 on the heptad repeats of the C-terminal domain (CTD) in the stalled RNA polymerase II, thereby stimulating transcription elongation and coupling the histone acetylation mark to the transcription activity (19–21). Accordingly, significant downregulation of cellular gene expression was observed by treatment with JQ1, especially of those genes that are regulated transiently by strong and inducible transcriptional factors, like Myc and NF-κB (17, 22). In addition, NF-κB (p65) undergoes protein acetylation once it becomes activated. The acetylated domain holds structural homology with the histone tail, thus the posttranslational modification allows BRD2/4 to directly bind to p65 (23). This mechanism makes JQ1 (BET inhibitor) particularly effective for inhibition of the NF-κB signaling pathway. BET inhibitors have also been proposed as a Myc pathway targeted therapeutic, with preclinical activity demonstrated in multiple lymphomas including PEL as it downregulates Myc expression and its downstream target genes (22, 24–26).

Similar to the “shock and kill” strategy for elimination of the latent HIV reservoir, methods that employ lytic reactivation of viruses from tumors latently infected with an oncogenic herpesvirus represent a unique strategy of antineoplastic therapy, as it may increase the specificity of cytotoxic cancer therapeutics (27–29). This is conceptually similar to the mechanism of an oncolytic virus-based therapy like the FDA-approved Talimogene Laherparepvec, which is designed to: (i) replicate specifically in cancer cells, but not in normal cells; (ii) stimulate the host immune system to the tumor microenvironment; and (iii) have their replication manageable by replication-suppressive drugs for patient safety. Accordingly, lytic reactivation of KSHV from latently infected cancer cells would theoretically be beneficial by providing additional highly specific direct tumor lytic effects as well as by evoking an increased cytotoxic immune response against cancer cells expressing lytic viral antigens.

In this study, we evaluated the effects of the "shock and kill" strategy on latently KSHV-infected tumor cells, which was originally designed to target latently HIV-infected cells. We demonstrated that the same drug combination with the same dosage, which reactivates HIV from latently infected T cells, could strongly reactivate KSHV in PEL cell lines and inhibit PEL cell growth in both in vitro tissue culture and in vivo xenograft tumor models.

HBL-6, JSC-1, and BC2 cell lines, obtained from Dr. Masahiro Fujimuro (Kyoto Pharmaceutical University, Kyoto, Japan) in 2015. BCBL-1 cell line was obtained from Dr. Ganem (University of California, San Francisco, CA) in 2001. These cell lines were cultured in RPMI1640 medium supplemented with 15% FBS. The Flag-HA tagged-K-Rta-inducible, TREx-K-Rta BCBL-1 cell line was generated according to methods previously described (30). No testing for the cell authentication for HBL-6, JSC-1, BC2, and BCBL-1 cell lines was performed. BC3 cell line was obtained from ATCC, expanded and stored to obtain early passage stocks. The BC3 cell was used for mouse xenograft studies with early passages of cells within a month. Mycoplasma contamination was tested by PCR.

Anti-K-Rta and anti-K-bZIP antibodies were previously described (31). Anti-LANA antibody (Advanced Biotechnologies), anti-p-IκBα (Ser32/36), anti-IκBα, anti-p-NF-κB p65 (Ser536), and anti-NF-κB p65 antibody (Cell Signaling Technologies), anti-phospho-S2 RNA Polymerase II, anti-phospho-S5 RNA Polymerase II (Abcam), anti-RNA Polymerase II antibody (Active Motif), anti-BRD4 antibody (Bethyl Laboratories), and anti-Actin, anti-GAPDH, and normal mouse and rat IgG (Santa Cruz Biotechnologies) were commercially obtained.

BET inhibitor (+)-JQ1 (ApexBio Technology), PEP005 (Tocris Bioscience), suberoylanilide hydroxamine (SAHA; also known as Vorinostat; Santa Cruz Biotechnologies), and GSK343 (ref. 32; Sigma) were obtained from commercial sources. Drugs were added into culture media alone or in combination, and the effects on KSHV reactivation as well as PEL cell growth were monitored.

Cells were fixed with 4% formaldehyde and permeabilized with successive treatments of 1% SDS and 1% Triton-X 100 in PBS for 15 minutes each at room temperature. Primary antibody was incubated overnight in 2% BSA/PBS at 4°C. Secondary antibody (Alexa Fluor 488 or Alexa Fluor 555-conjugated antibodies; Invitrogen) was incubated for 1 hour at room temperature. Slides were mounted with Anti-fade Gold containing DAPI (Invitrogen).

Total RNA was isolated from cells using the RNeasy Kit (Qiagen) followed by digestion with DNase I (Invitrogen). First-strand cDNA was synthesized using Superscript II Reverse Transcriptase (Invitrogen). All cDNAs were analyzed by SYBR green-based qPCR (Bio-Rad) with primers described previously (33). qPCR was performed in triplicate.

Whole transcriptome profiling was performed using a directional, strand-specific mRNA-Seq approach. Briefly, total RNA samples were submitted to the UC Davis Comprehensive Cancer Center's Genomics Shared Resource (GSR), and indexed RNA-Seq libraries were prepared from total RNA (200 ng) using the KAPA Stranded mRNA-Seq Kit (Kapa Biosystems, Inc.) according to the manufacturer's protocol. Briefly, poly-adenylated mRNA was purified by binding to oligo(dT) beads, which was followed by fragmentation by incubation at 94°C in the presence of magnesium. Double-stranded cDNA was then generated by random-primed first-strand synthesis and second strand synthesis in the presence of dUTP for strand marking. Subsequently, the cDNA was 3′-A tailed and indexed, Illumina-compatible adapters were ligated. The libraries were then enriched by high-fidelity PCR amplification (13 cycles) with KAPA HiFi HotStart DNA Polymerase and adapter-specific primers. The libraries were combined for multiplex sequencing on an Illumina HiSeq 4000 System (50-bp, single read; ∼30 million reads/sample).

RNA-seq data were analyzed using a HISAT-Cufflinks workflow. Raw sequence reads (FASTQ format) were aligned to the reference human genome assembly (December 2013, GRCh38) using HISAT2 v.2.0.3-beta (hierarchical indexing for spliced alignment of transcripts) software. Gene- and transcript-level expression were comprehensively quantified with Cufflinks v.2.2.1 software, which performed (i) transcript assembly, (ii) identification of splice variants, (iii) quantification of expression as (fragments per kilobase of exon per million mapped sequence reads (FPKM) values, and (iv) normalization. Normalized FPKM values (Cuffnorm output) were utilized for downstream analysis steps. Statistical analysis for treatment-specific DEGs (ANOVA, P < 0.05), comparison analyses (e.g., treatment vs. vehicle control), and hierarchical clustering of the data were performed with GeneSpring GX software (Agilent Technologies Inc.). Gene set enrichment analysis (GSEA) for each drug treatment was performed on GSEA web site (34) after conversion of gene ID to Affimetrix Human genome U133 Plus 2.0 Array ID. The raw sequence and processed expression data files are publicly available through NCBI GEO (https://www.ncbi.nlm.nih.gov/geo/) with accession number GSE89478.

ChIP assays were performed as described previously (35). Briefly, BCBL-1 cells were pretreated with JQ1 for 24 hours, followed by treatment with PEP005 or vehicle control. Cells were fixed in 1% formaldehyde for 10 minutes at room temperature and quenched with glycine. DNA was sonicated to an average size of ∼300 bp using a Diagenode Bioruptor. After precleaning with BSA-blocked Dynabeads Protein A/G (Invitrogen), chromatin was incubated with antibodies at 4°C overnight. The chromatin immunocomplexes were collected with BSA-blocked Dynabeads. Immunoprecipitated chromatin was eluted from beads by heating for 30 minutes at 65°C in elution buffer (50 mmol/L Tris-HCl, pH 8.0, 10 mm EDTA, 1% SDS). DNA was reverse crosslinked at 65°C overnight and purified with a PCR Purification Kit (Qiagen). All immunoprecipitated chromatin DNA samples were analyzed by SYBR green-based qPCR (Bio-Rad) with specific primer pairs. The primer sequences are as follows: PAN promoter, 5′-GGT GGC TAA CCT GTC CAA AA-3′ (forward) and 5′-CAG CGA GCA CAA AAT CCA TA-3′ (reverse); K12 promoter, 5′-AGT GCT TTA ATG CGG GGA GG-3′ (forward) and 5′-GTT GCA CCA AGC ACA ACA TT-3′ (reverse).

Briefly, chromatin DNA from 1 × 10⁸ BCBL-1 cells was used per each immunoprecipitation assay with 10 μg of BRD4 antibody. ChIP-enriched and input DNA samples (1 ng) were used to generate Illumina-compatible libraries with the KAPA LTP Library Preparation Kit (Kapa Biosystems, KR0453) according to the manufacturer's recommendations. Libraries were submitted for sequencing (50-bp, single read) on an Illumina HiSeq 3000 sequencing system. The ChIP-Seq data (FASTQ sequence reads) was aligned to the human hg19 reference genome and reference KSHV genome sequence (Human herpesvirus 8 strain JSC-1 clone BAC16, GenBank: GQ994935.1) with Bowtie 2. Peak finding was performed with the MACS2 program utilizing the parameters and commands described following the developer's manual. We used the default settings with a minimum FDR (q-value) cut-off of 0.05. The peaks and read alignments were visualized using the Integrative Genomics Viewer (IGV) browser from the Broad Institute. The raw and processed sequence data files are available through NCBI GEO with accession number GSE103395.

Reporter assays were performed as described previously (35). A dual luciferase reporter system (Promega) was used to quantify NF-κB activity. Briefly, 500 ng of pNF-κB-Luc reporter plasmid containing three direct repeats of the NF-κB responsive element was transfected into BCBL-1 cells in each well of 12-well plates. Two days after transfection, the cells were washed with PBS, and incubated with 250 μL passive lysis buffer provided by the manufacturer. A RenillaTK expression vector (10 ng/well) served as an internal control. At least three independent assays were carried out for each experiment.

WST-8 and MTT cell proliferation assays were performed using the Cell Counting Kit-8 (Sigma) or MTT reagents (Thermo Fisher) according to the manufacturer's instructions. Assays were performed in triplicated wells and three independent assays were carried out.

Cytokines, IL6, vascular endothelial growth factor (VEGF), and IL10, in the culture supernatant were measured in triplicates with a combination of magnetic-beads set for each cytokine, an appropriate cytokine standard and a reagent kit from Bio-Rad according to the company's protocol. Each cytokine concentration was determined by a Bio-Plex Manager software (Bio-Rad) based on the standard.

All animal studies were conducted according to a UC Davis Institutional Animal Care and Use Committee (IACUC)-approved protocol. Five-week-old female NOD/SCID mice were injected intraperitoneally with 2 × 10⁷ BC3 cells in 200 μL PBS. On day 3, mice were randomly assigned to Vehicle (5% DMSO in 2% Dextran PBS), PEP005 (10 μg/kg; daily), JQ1 (100 mg/kg; daily), and JQ1 in combination with PEP005 treatment groups. Treatments were administered by intraperitoneally injection for 10 days. Mice were monitored daily and sacrificed when exhibiting signs of discomfort.

SPSS version 18.0 was used for the statistical analysis of experimental data, with data presented as mean ± SEM, calculated for all points from at least three independent experiments in triplicates. Statistical significance was determined using the ANOVA and Student t test. Survival curves were generated using Kaplan–Meier analysis, and statistical significance was determined using Gehan–Breslow–Wilcoxon test with Graphpad Prism software (San Diego, CA). P values <0.05 were considered significant.

KSHV-associated malignancies are most frequently seen in HIV-infected patients. PEP005 has been identified as an effective drug to reactivate HIV within T cells of HIV-infected patients (14). Thus, we sought to characterize the ability of PEP005 to induce KSHV reactivation from latently infected PEL cells, because PEL is most frequently seen in HIV-infected individuals. TREx K-Rta BCBL-1 cells were treated with different concentrations of PEP005 (10–1,000 nmol/L) or doxycycline (Dox: 1 μg/mL). Dox was used for a positive control, as it induces K-Rta protein expression (under the control of a tetracycline/doxycycline-responsive promoter), which leads to KSHV lytic replication. Similar to its ability to reactivate HIV, PEP005 triggers KSHV lytic replication in a dose-dependent manner (Fig. 1A,a), at a concentration as low as 10 nmol/L, which is the same concentration known to reactivate HIV in an ex vivo model (14). In addition, PEP005 induced transient expression of both K-Rta and K-bZIP, with a peak expression occurring at 48 hours [Fig. 1A(b), B, and C]. High concentrations of PEP005 (500–1,000 nmol/L) demonstrated apparent cell toxicity at 48 hours posttreatment (Fig. 1D,a). PEP005-mediated KSHV reactivation was also seen in other PEL cell lines, HBL-6, JSC-1, BC2, and BC3 (Supplementary Fig. S1A, a). Similar to BCBL-1 cells, lower concentrations of PEP005 (i.e., ≤100 nmol/L) did not significantly induce cell death in other four PEL cell lines tested (Supplementary Fig. S1B). These results demonstrated that PEP005 can trigger KSHV reactivation under the same conditions in which it reactivates HIV in vitro from latently infected T cells isolated from patients. Although it is difficult to estimate clinical dosage for the systemic application of PEP005, we anticipate that greater than 100 nmol/L would have significant side effects due to the molecular action of this agent (PKC agonist). In preclinical studies, the maximum-tolerated dose in a rat model was determined as 30 μg/kg.

Histone modification regulates the accessibility of transcriptional factors to the genome; thus, it has significant impacts on the gene regulation. PEP005-mediated KSHV reactivation with epigenetic drugs was next examined. We tested several agents in this group: the BET inhibitor, JQ1 (100 nmol/L); the HDAC inhibitor, SAHA (100 nmol/L); and the EZH2 inhibitor, GSK343 (2 μmol/L). Cells were treated with each agent in both the presence and absence of PEP005, and KSHV reactivation was subsequently examined by immunoblotting for the expression of viral lytic antigens. Although JQ1, SAHA, or GSK343 alone did not reactivate KSHV significantly at clinically-relevant dosages, all of three drugs clearly enhanced PEP005-mediated KSHV reactivation (Fig. 2A). The combination of JQ1 with PEP005 has also been examined in detail for HIV reactivation (14). Thus, we decided to further study this combination. The kinetics of KSHV gene expression showed that PEP005 increased transcripts of immediate-early (a), early (b), and late (c) genes, and the combination with JQ1 increased viral gene expression of all kinetics class of the lytic genes (Fig. 2B). The protein expression of K-Rta and K-bZIP confirmed the qRT-PCR results (Fig. 2C; Supplementary Fig. S1A,b), although timing of lytic protein expression was slightly different among cell lines.

The effects of the drug combination on cell growth and killing were examined next. For this, the number of dead cells was counted by staining with trypan blue, and cell proliferation was examined by the MTT assay. The results showed that neither PEP005 nor JQ1 alone induced cell death at patient-relevant drug concentrations (i.e., PEP005: 10 nmol/L, JQ1: 100 nmol/L; Fig. 3A), although JQ1 alone could decrease cell growth in vitro (Fig. 3B). Cell death rate was not increased in the drug combination treatment of PBMC B cells from healthy donors. The combined treatment of PEP005 plus JQ1 exhibited cell killing effects on three PEL cell lines tested (i.e., BCBL-1, HBL-6, and BC3). Consistent with the cell killing effects, the growth of PEL cell lines (BCBL-1, HBL-6) was further decreased by the combination of PEP005 plus JQ1 treatment, whereas BJAB (KSHV negative) cell growth was not inhibited by either drug alone, or in combination (Fig. 3B). PEP005 at 10 nmol/L alone did not show significant growth-inhibitory effects in all cell lines except HBL-6 within a 3-day culture period (Fig. 1D). In contrast, the KSHV-negative BJAB cell was relatively resistant to treatment with the drug combination (Fig. 3A and B).

The molecular mechanism underlying KSHV reactivation by the PEP005/JQ1 drug combination was examined next. In previous studies, PEP005 has been shown to activate the NF-κB pathway (14). NF-κB activation is tightly associated with the KSHV latency/reactivation switch (35–37), and its activation is transiently regulated through the induction of own inhibitor, IκB. Accordingly, the phosphorylation levels of both IκB and NF-κB p65 were examined by immunoblotting. As shown in Fig. 4A, PEP005 (12 nmol/L) increased the levels of phosphorylated IκBα and NF-κB p65. The maximum level of IκBα phosphorylation was reached at 8 hours poststimulation and decreased thereafter, whereas phosphorylation of NF-κB p65 exhibited a gradual and persistent increase during time course examined. Consistent with phosphorylation of IκBα, incubation with PEP005 promoted nuclear translocation of NF-κB p65 (Fig. 4B and C) and lead to NF-κB p65 mediated transcriptional activity in BCBL-1 cells, which was measured with an NF-κB-specific luciferase reporter (Fig. 4D).

Next, the effects of JQ1 on NF-κB recruitment to the KSHV genome were then examined by chromatin immunoprecipitation (ChIP) assays. First, the occupancies of BRD4, one of the major JQ1 targets, were determined in an unbiased manner by ChIP-seq analysis. The results showed that 24 hours after KSHV reactivation, BRD4 mainly bound to the KSHV IE and E gene cluster regions (Fig. 4E). Based upon these results, we next designed specific primer pairs to regions that contained BRD4 recruitment sites that also overlapped with putative NF-κB target sequences on the KSHV genome (e.g., PAN promoter region), and examined the effects of the drug treatment on the recruitment of NF-κB. In addition, the K12 promoter region, where NF-κB directly binds was also included (35). Consistent with the observed nuclear translocation of NF-κB, the results demonstrated that PEP005 increased occupancies of NF-κB on the KSHV genome at the PAN RNA promoter region. We also found that PEP005 treatment alone induced dissociation of BRD4 weakly. As expected, implementation of JQ1 inhibited the binding of BRD4 to the KSHV promoter (Fig. 4F). The dissociation of BRD4 enhanced recruitment of NF-κB and RNA polymerase II, and consequently increased PAN RNA expression (Fig. 2B). Similar effects were also seen in other KSHV promoters (Supplementary Fig. S2). Taken together, these results indicated that further activation of NF-κB by PEP005 may contribute to KSHV reactivation, and that JQ1 enhanced RNA polymerase II recruitment.

Inflammatory cytokine secretion from KSHV-infected cells plays an important role in the pathogenesis of KS and PEL (38). To explore the effects of the drug combination on cytokine secretion, Luminex assays were used to measure the amount of cytokine secreted into tissue culture supernatants in response to drug treatments. Culture supernatants were harvested 24 hours after drug treatment, and the concentrations of IL6, VEGF, and IL10 were measured. Different drug combinations did not significantly affect cell viability at this time point (Fig. 3B), thus bias in cytokine production caused by cell numbers should be minimal. The results demonstrated that overall a therapeutic dosage of JQ1 (100 nmol/L) inhibited secretion of all three cytokines from KSHV-positive cell lines (Fig. 5). Importantly, JQ1 and the combination with 10 nmol/L PEP005 reduced IL6, an essential growth factor for PEL cells, in all three KSHV-positive cell lines tested. As expected, PEP005, an activator of NF-κB, showed a tendency of increasing the production of NF-κB-targets VEGF and IL10 when used alone or in combination with JQ1, GSK343, or SAHA. Despite the latter, addition of PEP005 did not completely abolish the beneficial regulatory effect of JQ1 in terms of reducing inflammatory cytokine production when compared with the DMSO control, resulting in an overall favorable effect in the regulation of inflammatory cytokines. In contrast, treatment with JQ1 and PEP005, individually or as a combination, showed completely different modulatory effects on VEGF and IL10 production in KSHV-negative BJAB cells, which also express barely measurable amounts of IL6.

Because the mechanisms of both PEP005 and JQ1 are known to impact gene expression via transcription factor activation (Fig. 4) and/or epigenetic processes (e.g., NF-κB activation, BET inhibition, etc.), RNA-seq analysis was performed in order to better define their impact, alone and in combination, on the transcriptome in three PEL cell lines. First, unsupervised clustering of the 12 datasets (i.e., three cell lines x four treatments each) demonstrated that the data expectedly organized the samples into major clusters based on cell line, rather than treatment, thereby indicating the prominent heterogeneity of the PEL cell lines (Fig. 6A). Next, comparison and statistical analyses were conducted for the identification of differentially-expressed genes (DEG; treatment group relative to vehicle control). As shown in Fig. 6B, numerous gene expression alterations were induced by the different treatments in each cell line; for instance, 2,650, 2,757, and 2,820 genes were altered in BC3 cells by PEP005, JQ1, and the combination, respectively. Notably, approximately 28.4% to 35.4% of the DEGs (389–435 genes) observed for each treatment were conserved among the three cell lines (Supplementary Fig. S3). Of these, 226 genes exhibited statistically-significant expression changes (ANOVA, P < 0.05) in response to one or more treatments (i.e., relative to the control), and consistent with this result, hierarchical clustering of this DEG expression data organized the samples into major clusters based on treatment (Fig. 6C). The significance of the PEP005+JQ1 combination treatment was then determined by evaluating its ability to further enhance or suppress the expression of genes that were differentially regulated by the individual treatments. The fold-change values for DEGs from each treatment (i.e., relative to control) of each cell line were visualized with heatmaps (Supplementary Fig. S3B), and these showed that in addition to there being subsets of DEGs that were further up- or downregulated by the addition of the second agent (i.e., combination treatment), there were also subsets that were counter-regulated by the combination.

Next, GSEA was performed on the DEGs that were induced by either treatment with PEP005 or JQ1 alone, and also by the PEP005+JQ1 combination. The results demonstrated that E2F target genes were significantly enriched in both JQ1 [normalized enrichment score (NES) 1.640] and PEP005 treatment (NES 2.042) in BCBL-1, respectively. Importantly, enrichment score was further increased in the combined treatment (NES 2.362; Fig. 6D). Consistent with previous studies, treatment with JQ1 also displayed significant enrichment of MYC target genes in both JQ1 alone (NES 1.485) and in combination with PEP005 (NES 1.730; Supplementary Fig. S3C; refs. 22, 39). Again, the combination further increased NES score of MYC target genes (Supplementary Fig. S3C). The results indicated that the combination with PEP005 arguments the effects of JQ1 in PEL cells.

Finally, the effects of the drug combination on in vivo tumor growth were evaluated in a PEL xenograft model. We inoculated female NOD/SCID mice with 2 × 10⁷ BC-3 cells via intraperitoneal injection followed by intraperitoneal injections of JQ1 (100 mg/kg) alone or in combination with PEP005 (10 μg/kg), daily for 10 days starting from day 3 of tumor inoculation. At the time point, tumors and ascites were not visible. Nonetheless, all mice that received injection of BC-3 cells developed PEL with ascites in the peritoneal cavity. Several mice also developed solid tumors subcutaneously near the site of injection. All mice in the control vehicle-treated groups died within 17 days; however, in mice that received JQ1 treatment, tumor growth was completely inhibited as evidenced by the absence of weight gain during the course of therapy on days 3 to 12 (Fig. 7A), and extending the median survival from 12 to 23 days, compared with the vehicle-treated group (P = 0.0017; Gehan–Breslow–Wilcoxson test; Fig. 7B). Combination JQ1+PEP005 treatment further delayed tumor development (P = 0.0049 compared to JQ-1 group) and extended the overall survival of the mice (median survival 23 and 30.5 for JQ1 and JQ1 + PEP005 group, respectively). Even though the both therapies effectively managed PEL tumor growth during the treatment period and extended mouse survival, they could not prevent tumor recurrence as evidenced by rapid weight gain as soon as the withdrawal of the therapies after day 13 (Fig. 7A).

KSHV-mediated malignancies are highly associated with HIV infection. Many drugs that target cancer cells induce stress signals (e.g., cell-cycle arrest, ER stress), which in turn induce viral replication from latently infected cells. There is no clear answer if inducing latent viruses is clinically beneficial; however, having drugs (e.g., cART and ganciclovir) that are only effective to the cells in which viruses are actively replicating, an oncolytic strategy to specifically target infected cells (e.g., malignant cells) could be an option for treatment. In this study, we demonstrated that PEP005 in combination with JQ1 induces KSHV reactivation in PEL cell lines. Our studies were leveraged on previous work that demonstrated effective reactivation of HIV with the same drug combination and dosage. The main focus of this study was to determine if this “shock and kill” strategy for HIV would also be beneficial for managing KSHV associated malignancy.

KSHV reactivation was associated with transient PEP005-mediated further NF-κB activation. Interestingly, dissociation of BRD4 from the KSHV promoters appeared to enhance viral gene expression. This result was unexpected as BRD4 functions to recruit co-activators and elongation factors (17, 40). However, our results are consistent with that of HIV study, which indicates BRD2/4 is a weaker activator and physically inhibits Tat recruitment into TAR/CyclinT1/CDK9 complex, therefore suppresses HIV transcription. When BRD2/4 binding to the viral genome is inhibited by JQ1, Tat binds to TAR/CyclinT1/CDK9 complex and phosphorylates CTD of RNA pol II, leading to possessive transcription of HIV. Another explanation would be that sequestration of BRD4 from cellular genomes may increase available (free) cellular RNA polymerase II and co-activators. Nonpromoter bound RNA polymerase II could then be recruited to viral genomes (41, 42). We think that viral transcriptional factors may have higher binding capability with RNA polymerase II (43), which then recruits RNA polymerase II to form viral transcriptional factories (41).

Our ChIP-seq study identified binding sites of BRD4, and that these sites largely overlapped with those for the latent KSHV protein LANA (44), supporting previous reports that LANA and BRD4 can physically interact with each other (45). In addition, BRD4 recruitment sites also overlapped with active chromatin marks, suggesting authenticity of the BRD4 binding sites identified in this study (46).

In our Luminex studies, we observed that JQ1 substantially reduced IL6, VEGF, and IL10 production by PEL cell lines, a finding that was more prominent in KSHV-infected cells than KSHV-negative cells. This observation is supported by other studies demonstrating JQ1's ability to mute inflammation-associated gene expression. KSHV⁺ cells exhibit constitutively active NF-κB due to the presence of the latent viral proteins, and NF-κB activation is critical for both PEL cell growth and inhibition of cell apoptosis; this makes BET inhibitor attractive option as therapeutic approach for KSHV-associated malignancies (36, 37, 47). In fact, PEP005 treatment alone, which further activates NF-κB and triggers KSHV reactivation, increased VEGF and IL10 production in both KSHV⁺ (i.e., HBL-6, BC3, and BCBL-1) and KSHV⁻ (i.e., BJAB) cell lines whereas PEP005 + JQ1 treatment reduced IL6 in all KSHV-positive cell lines tested and controlled the levels of VEGF and IL10 (Fig. 5). Thus, PEP005 + JQ1 treatment had an overall favorable effect on the regulation of inflammatory cytokines.

RNA-sequencing was performed to comprehensively define each treatment's impact on the transcriptome, and to gain insight into the downstream effects on cellular functions. In each of the cell lines, these treatments had marked effects upon gene expression accounting for up to 5.11% (3,098 genes) of the annotated transcriptome, such as observed for the combination treatment of HBL6 cells (Fig. 6B). Although we could identify treatment-specific DEGs that were conserved among the three cell lines [Fig. 6C, (b)], the RNA-seq results also revealed that there was significant heterogeneity, or distinctiveness, in the responses of each cell line to these agents, since unsupervised clustering resulted in organization of the data according to cell line (Fig. 6A). Consistent with previous work (22), GSEA demonstrated that JQ1 targeted E2F and Myc target gene expression, and the combination further potentiate the targeting effects.

In a NOD/SCID xenograft mouse model, we showed that the JQ1 successfully delayed and suppressed overall PEL tumor growth and prolonged host survival, which was further improved by combining with PEP005 to some extent (Fig. 7). For treatment of skin cancer, in which high local concentration achievable by topical application, PEP005 strongly induced cancer cell death (48). In our mouse model, however, we could not apply such higher PEP005 amount systemically due to its intrinsic toxicity. In addition, with our NOD/SCID PEL-xenograft mouse model, we could not expect immune stimulatory effects by the PKC agonist. Accordingly, we did not include PEP005-alone group in our xenograft studies after consideration of minimum effects on cell growth in vitro (Fig. 1). Consistent with our in vitro observations that JQ1 did not possess strong cell killing capability (Fig. 3A), the remaining viable PEL cells resumed growth as soon as the drug treatment was terminated. Similar results have been seen in other xenograft studies with JQ1 (22, 39). Thus, the possible approach would be that administrating JQ1 to control tumor growth by targeting E2F and MYC (Fig. 6D; Supplementary Fig. S3C), and providing specific cell killing effects by inducing KSHV reactivation and/or cytolytic drugs. Although JQ1 provides other beneficial effects such as inhibition of IL6, VEGF, and IL10 production in addition to cell growth inhibition (Fig. 5), combining with other cancer drugs carrying cytolytic activities will be important to eradicate PEL cells. However, in vitro, we could readily increase the cell death rate by increasing PEP005 and/or JQ1 concentration (Fig. 1D). This indicates that we can significantly improve prognosis if we could use higher concentration of drugs without evoking unfavorable side effects to patients. In this regard, new drug delivery methods such as nanotheranotics approaches, which allow increase in local drug concentration or promote drug uptake by tumors without increasing systemic drug concentration, should be the key for one of future interventions (49). Rationally selected drugs in conjunction with novel drug delivery methods will improve prognosis of this devastating disease. Nonetheless, based on favorable effects on inflammatory cytokine production from PEL and enhancement of antigenic KSHV lytic gene expression in the cancer cells, and mild side effects with convenient oral administration (50), BET inhibitor would be considered to be included in the current combinatorial regimens.

In summary, we have examined a drug combination, which is suggested to reactivate HIV from latently infected cells. We found same drug combination robustly activated KSHV from latently infected B cells and has demonstrated some efficacy in preventing tumor growth in a preclinical PEL xenograft model.

No potential conflicts of interest were disclosed.

Conception and design: F. Zhou, E. Maverakis, S. Dandekar, Y. Izumiya

Development of methodology: F. Zhou, L. Olney, G. Jiang, R.R. Davis, E. Maverakis

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Zhou, M. Shimoda, L. Olney, Y. Lyu, K. Tran, C.G. Tepper, E. Maverakis

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Zhou, M. Shimoda, Y. Lyu, K. Nakano, R.R. Davis, C.G. Tepper, S. Dandekar, Y. Izumiya

Writing, review, and/or revision of the manuscript: F. Zhou, M. Shimoda, Y. Lyu, R.R. Davis, C.G. Tepper, E. Maverakis, M. Campbell, Y. Li, S. Dandekar, Y. Izumiya

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Zhou, G. Jiang, R.R. Davis, C.G. Tepper, M. Campbell

Study supervision: M. Shimoda, Y. Izumiya

The authors are grateful to Stephenie Y. Liu (UCDCCC Genomics Shared Resource, Department of Pathology and Laboratory Medicine) for expert technical assistance for the ChiP-Seq and RNA-Seq analysis studies. We also would like to thank Dr. Masahiro Fujimuro (Kyoto Pharmaceutical University) for providing us PEL cell lines.

This work was supported by the Jiangsu Provincial Natural Science Foundation of China (BK20141136) to F. Zhou, NIH grants (R01-DE025985) to Y. Izumiya, American Cancer Society Research Scholar Grant (RSG-13-383-MPC) to Y. Izumiya, U.S. Department of Agriculture (2014-67015-21787 and 2015-67015-23268) to Y. Izumiya, and NIH DP2 OD008752 to E. Maverakis. The UC Davis Comprehensive Cancer Center Genomics Shared Resource and the UC Davis Immune Monitoring Core are supported by the NCI Cancer Center Support Grant P30CA093373 from the NCI.

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.