Xpo-1 Antagonism Transcriptionally Reprograms and Is a Therapeutic Vulnerability in GATA-3 Driven T-Cell Lymphomas

Background: The transcription factor GATA-3, via the oncogenic transcriptional program it directs, is a bona fide proto-oncogene, and thus therapeutic target, in a subset of T-cell lymphomas (TCL). The eukaryotic translation initiation factor eIF4E, in complex with helicase and scaffolding factors, facilitates 40S ribosome and Met-tRNA recruitment, both of which are rate-limiting steps in the initiation of translation. Pathways that converge on eIF4E, including PI3K/AKT/mTOR-dependent phosphorylation of 4E-BP1, are frequently activated in GATA-3 PTCL. Within the nucleus, eIF4E specifically binds oncogenic mRNAs, and then forms a complex with a co-factor (LRPPRC) and the nuclear export receptor XPO1. In contrast to bulk mRNA, these eIF4E-bound mRNA are dependent upon XPO1 for nuclear export. Therefore, as we have shown that GATA-3 is oncogenic in T-cell lymphomas (TCL), we hypothesized that selective XPO1 inhibition may transcriptionally reprogram these TCL and is a novel therapeutic vulnerability.

Methods: XPO1 expression was examined by immunohistochemistry or flow cytometry in TCL specimens and in TCL that emerge in multiple genetically engineered mouse (GEM) models. Transcripts that are eIF4e clients and dependent upon XPO-1 for nuclear export were identified in an unbiased manner by eIF4e RNA immunoprecipitation (eIF4e-RIP) and RNA-seq. The extent to which XPO1 antagonism transcriptionally reprograms TCL cells, and its value as a therapeutic strategy, was examined pharmacologically with selinexor in ex vivo and in vivo studies using primary TCL specimens and GEM/PDX models, respectively.

Results: Among the PTCL cases (n=53) examined, XPO1 expression was observed in 62%, and was highly expressed in CTCL, particularly in patients with large cell transformation. We sought to further understand the extent to which XPO1 expression is associated with T-cell lymphomagenesis. We utilized multiple GEM TCL models in which Smarcb1 (SNF5) or PTEN were conditionally deleted in CD4 ⁺ T cells. In both GEM models, a significant increase in XPO-1 expression was associated with the emergence of a malignant clone. A significant reduction in cell viability and increased apoptosis was observed in TCL cell lines and primary specimens treated in vitro/ex vivo with selinexor. In both GEM models, and a PDX model, selinexor treatment similarly decreased TCL progression in vivo. Given these findings, we elected to examine XPO1's potential role in transporting oncogenic, and eIF4e client, transcripts. We performed eIF4e-RIP and RNA-seq in two representative cell lines. Gene set enrichment analysis utilizing the eIF4e client RNA identified demonstrated enrichment for pathways with an established role in TCL (e.g. TCR), including GATA-3. We subsequently performed eIF4e-RIP and quantified GATA-3 transcripts by qRT-PCR in cell lines (n=4), a PDX, and in primary SS specimens (n=2). Compatible with the RNA-seq data, significant enrichment in GATA-3 transcripts was observed. These findings imply that GATA-3 transcripts are dependent upon XPO1 for nuclear export. To directly examine this, the ratio of cytoplasmic to nuclear (C:N) GATA-3 transcripts were examined by qRT-PCR in cell lines treated with selinexor. A significant decrease in the C:N ratio was observed upon XPO1 inhibition, confirming that nuclear export of GATA-3 transcripts is XPO1 dependent. We similarly examined the extent to which GATA-3 target genes are eIF4e clients and XPO1 dependent for nuclear export. We had previously described GATA-3 target genes by integrated ChIP-seq/RNA-seq. Integration of this prior dataset with our eIF4e-RIP sequencing data demonstrated that ≈21-30% of previously described GATA-3 target genes were also eIF4e clients. We selected a few such targets, including ITK, for further validation. These GATA-3 targets were eIF4e clients and dependent upon XPO1 for nuclear export. Finally, we examined expression of these proteins in a diverse panel of T-cell lymphoma/leukemia cell lines and a primary SS specimen after brief exposure to selinexor, and compatible with our findings, a significant decrease in protein expression was observed.

Conclusions: Collectively then, our findings demonstrate that XPO1 antagonism, by impairing the nuclear export of both GATA-3 and many of its target gene mRNA, is a novel strategy to transcriptionally reprogram and therapeutically target GATA-3 driven TCL.