Exportin-1 Connects Dynamic Transcription and Translation of Genotoxic Stress Genes in Diffuse Large B-Cell Lymphoma Patients

Exportin-1 (XPO1), that exports a subset of proteins from the nucleus to the cytoplasm, is amplified in 30% of DLBCL. Regardless of XPO1 gene status, DLBCLs require XPO1 to proliferate and survive as demonstrated in pre-clinical (Marullo et al. AACR 2015) and clinical (Kuruvilla at al. Blood 2017) studies with the covalent XPO1 inhibitor selinexor. However, the mechanism underlying such dependency remains unknown.

We analyzed the expression of XPO1 protein in a cohort of 57 DLBCL patients by IHC and correlated with XPO1 gene status and clinical outcome. XPO1 increased copy number is associated with elevated protein expression (p = 0.03) and, importantly, patients with elevated XPO1 levels (XPO1 ^high) are less likely to respond to RCHOP as refractory/early relapse diseases (p = 0.005). This suggests that XPO1 ^high DLBCL could be biologically fitter to tolerate genotoxic stress. To explore this hypothesis, we compared gene expression profiles (RNA-sequencing) of XPO1 ^high (n = 15) vs. XPO1 ^low (n = 16) DLBCLs by DESeq2. XPO1 ^high DLBCLs were enriched (q<0.005) for genes regulating "cell proliferation", "DNA damage response and repair", "mRNA processing" and "RNA export"; whereas depleted (q<0.05) for genes regulating "immune response". Gene Set Enrichment Analysis also revealed that XPO1 ^high were enriched for transcripts up-regulated in highly aneuploid tumors (FDR = 0, NES = 2.19). All these findings were validated in an independent cohort of 72 DLBCL patients.

To gain insight into the mechanism by which XPO1 enhances genotoxic stress tolerance, we exposed a panel of 5 DLBCL cell lines to DNA damaging agents (etoposide, doxorubicin, mechloretamine) with or without XPO1 inhibition (selinexor). Comet assay experiments revealed that selinexor did not affect the amount of DNA damage induced by chemotherapy, but delayed cell's ability to repair the DNA damage (P>0.005). This translated into a significant increase of sensitivity toward otherwise sub-lethal effect of chemotherapy, as all the combinations tested showed synergistic effect on growth inhibition. To determine the existence of a therapeutic window to capitalize on this effect, we evaluated the combination of CHOP +/- selinexor in an XPO1 ^high DLBCL patient-derived xenograft mice model (n = 40). Combination treatment achieved higher anti-lymphoma effect compared to either selinexor or CHOP alone (p>0.005 and p= 0.0015, respectively). Importantly, no increase in toxicity was observed by biochemistry or pathological examination. Thus, proliferating DLBCL cells but not normal cells rely on XPO1 to survive exogenous genotoxic stress.

To characterize the molecular consequences of XPO1 inhibition, we quantified the expression of proteins involved in DNA damage response and repair (immunoblotting). Exposure to selinexor decreased the expression of a subset of these proteins including CHEK1, BRCA1, BCL6 and RAD51. This reduction was neither due to decreased gene transcription (qRT-PCR) nor to increased protein degradation (cycloheximide chase). Instead, analysis of nascent proteins by L-azido-homo-alanine labeling revealed that XPO1 inhibition decreases the synthesis of specific DNA damage repair proteins whose increased turnover is required to survive genotoxic stress. Since XPO1 inhibition affected only a subset of proteins, the mechanism may involve selective decrease in nuclear export of mRNA templates. Thus, we quantified the nuclear/cytoplasmic ratio of mRNAs encoding for DNA damage repair proteins in DLBCL cells treated with selinexor. XPO1 inhibition resulted in nuclear retention of BRCA1, RAD51, BCL6 and CHEK1 mRNAs but not of β-actin. Since XPO1 is unable to directly bind mRNAs, we reasoned that its inhibition may impair the export of proteins carrying mRNA (ribonucleoproteins). Indeed, we identified eIF4E as one of these cargoes. Immunofluorescence experiments revealed that eIF4E is retained in the nucleus of DLBCL cells upon XPO1 inhibition and its nuclear retention is enhanced in etoposide-treated cells. Moreover, RIP-qPCR experiments revealed that eIF4E binds several DNA damage repair mRNAs such as CHEK1, BCL6 and BRCA1 in the nucleus of DLBCL cells and this binding increases significantly in response to DNA damage.

Overall, our data demonstrate that XPO1, by regulating the selective export of mRNA-binding proteins, promotes genotoxic stress tolerance thus conferring a survival advantage to XPO1^high DLBCLs.