AUTHORS
Mariam Khanfar, MS, Charlz Nithin Jerold, MS2, Sivasankar Annamalai Sharavanan, MS3, Amanda Cashen, MD4, Neha Mehta-Shah, MD, MSCI5, Brad S. Kahl, MD6, Nancy L. Bartlett, MD7, Malachi Griffith, PhD8, Todd Fehniger, MD, PhD9, Obi Li Griffith, PhD10, Felicia Gomez, PhD11
AFFILIATIONS
1 Washington University in St. Louis, St. Louis, Missouri, United States
2 Washington University in St. Louis, St. Louis, Missouri, United States
3 Washington University in St. Louis, St. Louis, Missouri, United States
4 Washington University in St. Louis, St. Louis, Missouri, United States
5 Washington University in St. Louis, St. Louis, Missouri, United States
6 Washington University in St. Louis, St. Louis, Missouri, United States
7 Washington University in St. Louis, St. Louis, Missouri, United States
8 Washington University in St. Louis, St. Louis, Missouri, United States
9 Washington University in St. Louis, St. Louis, Missouri, United States
10 Washington University in St. Louis, St. Louis, Missouri, United States
11 Washington University in St. Louis, St. Louis, Missouri, United States
ABSTRACT
Introduction
Classical Hodgkin lymphoma (cHL) is defined by rare malignant Hodgkin/Reed–Sternberg (HRS) cells surrounded by a rich immune and stromal microenvironment that supports tumor survival and immune evasion. Because HRS cells are scarce and spatially dispersed, their transcriptional states and interactions with surrounding cells have been difficult to characterize. Here, we integrated single nucleus RNA sequencing (snRNA-seq) and whole-exome sequencing (WES) to define HRS-cell programs and tumor-microenvironment composition relative to previously published reactive lymph nodes (RLN).
Methods
Eight fresh frozen cHL tissues were profiled by snRNA-seq and WES. Seven RLN samples served as nonmalignant controls for comparison. Key transcriptional findings were validated using publicly available microarray datasets and an independent published snRNA-seq cohort.
Results
Integration of snRNA-seq, WES-detected somatic variants and canonical HRS markers (CD30, CD274, IRF4) identified an HRS population comprising 5,983 nuclei. HRS abundance varied across patients [0.3% to 13.5%]. Relative to nonmalignant B cells, HRS cells showed enrichment of established oncogenic and inflammatory programs, including mTORC1, MYC targets, JAK–STAT, and NF-κB pathways associated signaling. Among the most highly enriched genes in the HRS cluster were the canonical HRS markers and notably, a less appreciated neurodevelopmental program involving axon guidance, cell–cell adhesion, and ion-channel signaling, marked by the upregulation of RYR2, NRXN3, SOX5, KCNMA1, TENM2, KCND2, TENM4, ADAM22, NAV3, ROBO2, and LSAMP. This program recapitulated a recently described neuroglial-like state in EBV-negative HRS cells and was validated in external microarray datasets. The biological significance of this pathway in HRS cells remains unresolved.
In cell-cell communication analysis, 40% of the top 10 ranked predicted interactions involving HRS cells as either the sender or receiver included ROBO2 as the HRS-expressed receptor. ROBO2 was predicted to interact with the tumor microenvironment (TME) via the SLIT family of ligands (SLIT1/2/3). The most significant of these interactions was between ROBO2 and SLIT3 on tumor-associated fibroblasts. These results suggest that HRS cells may engage a SLIT–ROBO signaling program with the surrounding TME. Although poorly characterized in lymphoma, previous studies have reported SLIT–ROBO signaling in association with tumor-cell migration, immune regulation, angiogenesis, and microenvironmental remodeling in glioblastoma and other solid tumors. These findings nominate SLIT3–ROBO2 as a potential link between the stromal microenvironment and the neurodevelopmental HRS-cell program.
Overall, the most abundant cell type in the TME was T cells followed by fibroblasts and B cells. To identify cHL-associated immune remodeling we compared the cellular composition of cHL TME and RLNs within matched immune compartments. Within the pooled T/NK nuclei, cycling T cells were markedly more abundant in cHL (12.1% vs. 0.25%). These cells expressed CTLA4, consistent with a proliferative CTLA4⁺ T-cell state previously described. CD4 regulatory T cells (13.1% vs. 19.8%) and CD8+ T cells (18.9% vs. 30.6%) were less represented in the cHL T/NK cell compartment, whereas Naïve/memory-like CD4 T cells (39.5% vs. 36.5%) and activated CD4/Tfh-like cells (13.3% vs. 11.8%) showed similar prevalences. These cells might contribute to both lymphoid niche organization and suppression of antitumor immune responses. Within the pooled myeloid compartment, macrophages were the predominant population in cHL, comprising 61.4% of nuclei compared with 36.5% in RLNs. TIMD4-enriched resident/M2-like macrophages comprised the largest subset in the myeloid compartment (42.6%). These macrophages are associated with T cell sequestration and immune evasion. Furthermore, we identified a population of LAMP3⁺ dendritic cells (DC) that were 5.2% of the myeloid compartment. LAMP3⁺ DC have been implicated in immune regulation through recruitment of CCR4⁺ regulatory T cells and engagement of inhibitory checkpoint pathways.
In summary, these findings further describe the transcriptional profile of HRS, including a reproducible neurodevelopmental program and a candidate stromal SLIT3–ROBO2 signaling axis. Within the TME we show a previously identified population of cycling T-cells, a reduction of CD4+ and CD8+ T cells and a presence of TIMD4+ resident/M2-like macrophages.
DISCLOSURES
Mariam Khanfar, MS: No relevant conflicts of interest to declare.