Xpo-1 Antagonism Impairs CSF-1R Expression and Depletes Lymphoma-Associated Macrophages in T-Cell Lymphomas

Background: Lymphoma-associated macrophages (LAM) provide ligands for antigen, costimulatory, and cytokine receptors expressed by malignant T cells and directly promote T-cell lymphomagenesis. LAM also play an important role in suppressing host anti-tumor immunity and indirectly promote T-cell lymphomagenesis. Therefore, LAM are a therapeutic vulnerability in T-cell lymphomas (TCL), and their depletion from the tumor microenvironment (TME) is a rational therapeutic goal.

Methods: XPO1 and CD163 expression were examined by immunohistochemical staining. In order to determine the extent to which XPO1 antagonism depletes LAM, a pharmacologic strategy was adopted using selinexor. Viability of human monocyte-derived and TCL conditioned macrophages upon selinexor treatment was examined in vitro using standard methods. RNA binding to eIF4e was determined by RNA immunoprecipitation (RIP) using eIF4e and IgG antibodies and quantification performed by qPCR. Nuclear export of eIF4e client mRNA upon XPO1 antagonism with selinexor was examined by qRT-PCR in nuclear and cytoplasmic pools of RNA. LAM depletion upon selinexor treatment was examined in vivo using a previously described GEM (Smarcb1-deficient) model and CD68-GFP reporter mice.

Results: As we had previously demonstrated that CD163 identifies LAM that are alternatively polarized and proliferative, co-staining was performed with CD163 and XPO1, and CD163+XPO-1+ LAM identified in 69% of PTCL cases (n=26) examined. We previously demonstrated that LAM within a PTCL context are resistant to inhibition by selective CSF-1R antagonists. Therefore, we performed an unbiased, high-throughput screen using monocytes cultured in the presence of cell-free conditioned media (CFCM) obtained from T-cell lymphoma cell lines and primary specimens. This screen identified selinexor as a potential therapeutic vulnerability. Therefore, we cultured normal-donor monocytes in RPMI media alone, or in media supplemented with cell free conditioned media (CFCM) obtained from PTCL/CTCL cell lines (n=2) or primary specimens (n=5) and determined cell viability. Consistent with our prior findings, CFCM from PTCL/CTCL cells promoted the growth and survival of monocyte-derived macrophages (MDM), but this effect was abrogated upon selinexor treatment. These findings converged with the observation that CSF-1R transcripts are eIF4e clients in ALCL cells. Therefore, the extent to which transcripts required for macrophage survival (i.e. CSF-1R), proliferation (i.e. PU.1), and polarization (i.e. STAT3) are eIF4e clients was examined by eIF4e-RIP in MDM. These experiments demonstrated that CSF-1R, PU.1 and STAT3 are eIF4e clients, and selinexor treatment significantly decreased their nuclear export in MDM. Given the obvious clinical relevance of CSF-1R as a therapeutic target, CSF-1R expression in MDM following selinexor treatment was examined, and compatible with these results, a significant decrease in CSF-1R expression observed upon selinexor treatment. We previously employed CD68-GFP reporter mice in a Smarcb1-deficient GEM model, demonstrating that GFP+ macrophages in this TCL model are transcriptionally reprogrammed and proliferative. Selective CSF-1R antagonists minimally impair LAM survival in this model. In contrast, selinexor significantly depleted LAM ex vivo in this model and impaired the viability of bone marrow derived macrophages. In order to extend these findings, lymphoma-bearing GEM mice were treated with selinexor and the abundance of CD11b+Ly6G- LAM determined by flow cytometry, whereby a significant reduction in LAM abundance and TCL progression were observed, compatible with our in vitro/ex vivo findings.

Conclusions: Therefore, we concluded that XPO1 inhibition indirectly impairs T-cell lymphoma growth and survival by depleting LAM, which are a dependency in these lymphomas.