Type: Oral
Session: 621. Lymphomas: Translational – Molecular and Genetic: Molecular Profiling and Prognostic Biomarkers in Hodgkin and Non-Hodgkin Lymphomas
Hematology Disease Topics & Pathways:
Research, Fundamental Science, Translational Research
Methods: Single-cell RNA sequencing and TCR sequencing data from 79 CTCL (MF=57, SS=22) cases were collected from 9 public databases and integrated using the concatenate function in anndata. Scrublet was applied to each database to obtain per-cell scrublet scores and used a doublet exclusion threshold of median plus four median absolute deviations of the doublet score. Cells with greater than 30% mitochondrial gene expression or expression of fewer than 200 detected genes were excluded. Genes that were expressed in fewer than 3 cells were also removed. Data normalization and highly variable genes were processed by Scanpy. Batch correction and integration were performed using harmonypy. Human skin cell atlas datasets were also integrated to identify malignant T cells by inferring copy number variation (CNV) based on scRNA-seq data using InferCNV. Annotation was performed based on well-established and lineage-specific cell markers and the Human Cell Atlas. Non-negative matrix factorization (NMF) was performed to generate gene modules with each sample separately. NMF modules were clustered according to Jaccard similarity and meta-programs (MPs) were defined as shared NMF modules in the cluster.
Results: We generated a single cell atlas of CTCL by collecting 581,024 cells after strict quality controls from 79 CTCL cases. We firstly isolated total T cells to identify 227,064 malignant T cells by TCR clonotype and inferred CNV. By NMF, we identified over 3000 intra-tumor gene modules and 12 MPs shared by multiple tumors. Among them, cell cycle (MP1), antigen-presenting pathways (MP2), apoptosis (MP3), TNF signaling pathways (MP4), rRNA processing (MP5), and extracellular matrix organization (MP6) were enriched in more than 10 CTCL cases. 18% of genes in these MPs regulating translation are dependent on translation initiation factor eIF4E and nuclear export receptor XPO1. We validated this by performing eIF4e-RIP and RNA-seq in CTCL cell lines, and observed that at least 13.3% of genes in each MP were eIF4E clients, except MP4. Among them, 35% of genes in MP3 and 47% of genes in MP5, including the oncogene GATA-3, were eIF4E clients. To further examine this, we applied our eIF4E dependent transcripts and our previously identified GATA-3 target genes (Geng et al, Blood Cancer Journal, 2022) to this single cell atlas and found that eIF4E and GATA-3 dependent gene signatures were significantly enriched in malignant cells (p = 3.2e-125). We stratified cases by large cell transformation (LCT) status, and further observed a significant increase in these gene signatures in LCT cases compared to non-LCT (p < 0.0001). We similarly stratified patients by stage and observed a significant enrichment in eIF4E and GATA-3 dependent gene signatures in patients with late-stage (stages ≥ IIB) disease (p < 0.0001). These findings implicate eIF4E/XPO1 in GATA-3 driven CTCL. We then adopted a pharmacologic strategy using the selective XPO1 antagonist selinexor to determine the extent to which XPO1 is a therapeutic vulnerability in T-cell lymphomas. With only a few exceptions, the IC50 for most primary specimens and patient-derived cell lines examined was ≤250 nM.
Conclusion: Collectively then, our single cell atlas demonstrates that oncogenic transcripts in CTCL, including GATA-3, are dependent upon eIF4E/XPO1 for nuclear export. Patients with high-risk disease (LCT, advanced-stage) harbor malignant T cells that are highly enriched in eIF4E/XPO1 dependent transcripts. Therefore, the eIF4E/XPO1 axis is a novel dependency and therapeutic target in CTCL.
Disclosures: No relevant conflicts of interest to declare.
See more of: Oral and Poster Abstracts