Elucidation of Biological Role and Regulatory Landscape of PD-L2 in B-Cell Lymphoma

PD-L2 is a ligand for PD-1 immune checkpoint. In contrast to another ligand PD-L1, little is known about the biological relevance and regulatory mechanism of PD-L2 in cancer. Here we found by pan-cancer transcriptome analysis that PD-L2 was highly expressed in limited cancer types, especially in diffuse large B-cell lymphoma (DLBCL). In particular, PD-L2 expression was elevated in patients with PD-L2 genetic alterations, such as copy number amplifications and rearrangements causing promoter replacement or 3′-untranslated region (UTR) disruption.

To evaluate the effect of PD-L1 and PD-L2 on the tumor microenvironment and clarify their similarities and differences at a similar expression level, we generated a A20-ovalbumin (OVA) murine B-cell lymphoma cell line lacking Pd-l1 and introduced exogenous Pd-l1 or Pd-l2 expression. Analysis of A20-OVA model showed that Pd-l2 expression accelerated tumor growth and attenuated CD8 ⁺ T cell infiltration in vivo, similar to Pd-l1 expression. Then, we performed multi-omics single-cell analysis in this model, constructing transcriptomic, surface phenotypic, and immune repertoire maps of > 20,000 cells from mock-, Pd-l1-, and Pd-l2-expressing A20-OVA tumors. Importantly, Pd-l1- and Pd-l2-expressing tumors exhibited similar cellular dynamics as well as transcriptomic and surface phenotypic changes in the tumor microenvironment. Specifically, a significant decrease of CD8 ⁺ T cells, particularly effector/memory cells showing high clonality, and regulatory T cells as well as a significant increase of myeloid-derived cells, including monocytes/macrophages and plasmacytoid dendritic cells (DCs), were observed in Pd-l1- and Pd-l2-expressing tumors. Differentially expressed gene analysis demonstrated the downregulation of response to bacterial molecules, including lipopolysaccharide, and antigen processing and presentation pathways in monocytes/macrophages and conventional and plasmacytoid DCs, respectively, in Pd-l1- and Pd-l2-expressing tumors. In line with this, pro-inflammatory cytokine‒inducible markers, such as Ly6A/E and I-A/I-E, were down-regulated in various cell types in Pd-l1- and Pd-l2-expressing tumors. These results suggest that delineates pleiotropic effects shared by PD-L1 and PD-L2, mainly enhancing anti-inflammatory, pro-tumorigenic responses in the tumor microenvironment.

Given similar functions of PD-L1 and PD-L2, we hypothesized that the expression level of PD-1 ligands determines their biological relevance. Therefore, we aimed to dissect PD-L2 regulatory landscape by performing CRISPR tiling screening targeting 51 candidate regulatory elements predicted from Hi-C and DNase-seq data of a human transformed B cell line (GM12878). In addition to known cis-regulatory elements including the canonical transcription start site (TSS) and 3′-UTR, we identified a novel TSS, which was validated by cap analysis of gene expression (CAGE) with sequencing (CAGE-seq). Pan-cancer and -tissue expression analyses revealed that this novel element was expressed in 13% of DLBCL, but not in normal tissues nor other cancer types, suggestive of a unique PD-L2 regulatory mechanism in DLBCL. In addition, we identified an element located in the PD-L1 promoter which function as a distal silencer, suggesting functional complexity of this regulatory element. CRISPR-mediated knockout of other PD-L1 exons did not affect PD-L2 expression, suggesting that a silencer function in the PD-L1 promoter is independent of PD-L1 expression.

The identified PD-L2 regulatory elements can be occupied by an array of trans-regulatory factors. Indeed, ENCODE ChIP-seq of GM12878 revealed that many chromatin-associated proteins (CAPs) were bound within the PD-L1/PD-L2 topology associating domain. Therefore, to determine key regulators, we performed loss-of-function CRISPR screening for 103 CAPs. This CRISPR screening identified seven negative (such as IRF4 and BATF) and two positive regulators of PD-L2 expression. CRISPR/Cas9-based inhibition exhibited differential usage of canonical and novel TSSs among these factors.

Taken together, our findings reveal lineage-specific complex network of cis-regulatory elements and CAPs in regulating PD-L2 expression. These data provide insights into the molecular mechanisms underlying immune evasion and help refining immune-based therapeutic strategy in DLBCL.