Targeting the HTLV-I-Regulated BATF3/IRF4 Transcriptional Network in Adult T-Cell Leukemia/Lymphoma

After neonatal HTLV-I infection through breast feeding, approximately 5% of HTLV-I carriers eventually develop Adult T-Cell Leukemia/Lymphoma (ATLL) with a latency of ~50 years, suggesting that acquired genetic and epigenetic changes in cellular genes act in concert with HTLV-I to initiate and maintain oncogenic transformation. We and others have recently utilized next generation sequencing technology to identify mutated genes that could be pivotal in the pathogenesis of ATLL. However, due to the complexity of genomic/epigenetic alteration in the ATLL genome, the identification of indispensable genes for proliferation and/or survival of ATLL cells remains a formidable challenge.

To discover essential regulatory networks that are required for the proliferation and survival of ATLL cells, we performed a pooled shRNA screen in 8 ATLL cell lines using a library enriched for shRNAs targeting lymphoid regulatory factors and discovered that two BATF3 shRNAs and one IRF4 shRNA were highly toxic for all ATLL lines, but had little if any effect in other T cell and B cell lines. It is recently shown that a transcriptional complex of Irf4 and Batf binds to AP1-IRF composite (AICE) DNA motifs and plays key roles in the differentiation and function of certain mouse helper T cell subsets. A close paralogue of Batf, Batf3, is an indispensable transcription factor in a mouse dendritic cell subset, but also appears to play a redundant role with Batf in the differentiation of T_H2 cells and can substitute for Batf in Batf knockout T cells. Our observations from shRNA screening suggested that IRF4 and BATF3 may cooperate to drive a transcriptional program that is essential for ATLL viability.

We next used genome-wide chromatin precipitation (ChIP-seq) to identify the loci that are bound by BATF3 and IRF4. The set of binding peaks and the associated genes in IRF4 and BATF3 ChIP-seq intersected significantly. By integrating the ChIP-seq and gene expression profiling data of shBATF3- and shIRF4-ATLL cells, we defined a set of 68 BATF3-IRF4 direct target genes. Gene set enrichment analysis using gene expression profiling data from primary T cell lymphomas demonstrated that BATF3-IRF4 direct target genes were significantly enriched among genes that are more highly expressed in ATLL than in peripheral T-cell lymphoma, not otherwise specified (PTCL-NOS), suggesting that the BATF3 and IRF4 cooperatively regulate transcription in primary ATLL cells.

HBZ is unique among HTLV-I viral proteins in being maintained in expression in all ATLL cases, suggesting that it may help maintain the malignant phenotype. Given that BATF3 and IRF4 are essential regulators in ATLL, we hypothesized possible relationship between HBZ and BATF3-IRF4 complex. We defined HBZ direct target genes by integrating the ChIP-seq and gene expression profiling data of HBZ-knockout ATLL cell lines by CRISPR/Cas9. Notably we discovered that BATF3 was among these. BATF3 mRNA and protein expression decreased following HBZ inactivation.

The above considerations suggested that pharmacologic inhibition of the BATF3-IRF4 regulatory network might be a means to attack the HBZ oncogenic program therapeutically. ChIP-seq analysis of two enhancer marks, H3K27ac and BRD4, identified super-enhancers at the BATF3 locus in two ATLL cell lines. The small molecule JQ1 prevents the BET-protein BRD4 from interacting with chromatin, which is required for the function of super-enhancers. JQ1 treatment reduced BATF3 mRNA and protein levels in all ATLL lines tested, correlating with the eviction of BRD4 from the BATF3 super-enhancer. MYC mRNA and protein expression was also broadly downmodulated by JQ1. JQ1 treatment was consistently toxic for all ATLL cell lines tested at dose ranges that killed cell line models of T-ALL and DLBCL, which are known to rely on BET-proteins. In a dose-dependent manner, JQ1 also reduced the viability of primary ATLL samples and downregulated their expression BATF3 and MYC mRNA. Finally, we treated mouse xenograft models of ATLL with the BET-protein inhibitor CPI-203, a JQ1 analog with superior bioavailability in mice. In two different xenograft models, we observed significant tumor regression or growth inhibition, without evidence of systemic toxicity. Our study demonstrates that the HTLV-I virus exploits a regulatory module that can potentially be attacked therapeutically with BET protein inhibitors.