RNA Sequencing of Hodgkin Lymphoma Reed-Sternberg Cells Uncovers a Plasma Cell Signature and Escape from NK Cell Recognition

Introduction: Previous gene expression profiling studies of classical Hodgkin lymphoma (cHL) have been confined to cell lines and microdissected HRS cells from tissue biopsies given the difficulty of isolating sparse Hodgkin and Reed-Sternberg (HRS) cells from reactive background tissue. We previously used flow sorting to separate HRS cells from fresh or viably frozen cHL biopsies, and performed the first full exome sequencing of HRS cells. Here we report use of the same cell separation approach to examine the HRS cell transcriptome using RNA sequencing.

Methods: We used flow cytometric cell sorting and low-input RNA sequencing to generate full transcriptome data from viable primary HRS cells, along with intratumor B cells. Nine primary cases of cHL and four cell lines were assessed for RNA expression, expressed mutations, cell type of origin, signaling pathways, gene fusions and pathogen identification. We used immunohistochemistry to evaluate expression of PDIA6 and CD48 in the 9 cases sequenced and a tissue microarray containing 16 additional cases of cHL. Flow cytometry for CD48 was performed in two cell lines and 5 primary cases.

Results: Clustering show that primary HRS cells have a transcriptional profile that is unique, and different from that of intratumoral B cells, as well as cHL cell lines. Comparison of HRS cells with normal cellular subsets indicated plasma cell differentiation, suggesting that the cell of origin is a B cell on its way to becoming a plasma cell. Clustering with B cells showed much lower similarity. Consistent with plasma cell differentiation, we uncovered an unfolded protein response UPR) signature, shared with plasma cell neoplasms and, to a lesser extent, activated B cell (ABC) diffuse large B cell lymphoma, but not other B cell lymphoma types, including primary mediastinal B cell lymphoma (PMBCL). Among other UPR response genes, PDIA6 showed strong downregulation at the RNA level (2.4 logFC, p=9.4E-17). This finding was validated by immunohistochemistry for PDIA6, which showed strong positivity in the HRS cells of all 25 cases examined, confirming that this is a common feature of cHL, including nodular sclerosis and mixed cellularity subtypes.

Top upregulated genes included those involved in oncogenesis (HGF/MET, NFkB/apoptosis inhibition), stem cell differentiation (homeobox genes MEIS1 and PBX1), and mitotic checkpoints, mitotic spindle formation and DNA repair, possibly explaining the unique nuclear morphology of HRS cells. Downregulation of MHC-1 and MHC-2 driven antigen processing and presentation was confirmed, and so was overexpression of PDL1 (CD274). Importantly, we detected loss of SLAM family receptors, which serve as activation signals for NK cells providing an additional mechanism for tumor immune evasion. One of these is CD48 (-2.63 logFC, p=1.56E-05), which was confirmed to be strongly downregulated. This finding was confirmed by immunohistochemistry (25 cases) and flow cytometry (2 cell lines and 5 primary cases) on the expanded sample set. Given that only some cHL cases are associated with EBV infection, it has been speculated that other viruses are involved in negative cases. However, our analysis did not reveal additional viruses in the HRS cells.

Conclusions: Our data indicate that cHL more closely resembles plasma cells than B cells, and plasma cell malignancies than other lymphomas. The salient feature of plasmacytic differentiation is a UPR, which is seen in HRS cells and multiple myeloma. In contrast, UPR is not a feature of primary mediastinal B cell lymphoma, which is thought to be the DLBCL most similar to cHL clinically, immunophenotypically and in terms of gene expression patterns. We also provide an integrated view of potential immune evasion mechanisms by HRS cells that potentially explain lack of anti-tumor T or innate response. These include lack of antigen presentation due to B2M mutations, overexpression of PDL1 and PDL2, immunosuppressive cytokine secretion and, for the first time, a demonstration of lack of NK activating receptors of the SLAM family. Lack of SLAM family receptors may explain lack of NK cells clearance of HRS cells in the face of MHC-I downregulation. It has long been suspected that cHL is a tumor where there likely exists a previously undiscovered virus in addition to EBV, but RNA sequencing failed to reveal additional infectious transcripts in the HRS cells.