Regulation of innate lymphocyte function

National Institute Of Allergy And Infectious Diseases

Investigators

Linked publications, trials & patents

Abstract

The human B lymphoma cell line 721.221 cells was transfected with Cas9 and subjected to genome-wide CRISPR screens for reduced or augmented lysis by primary NK cells during long-term co-cultures at very low effector to target ratios. Selection for mutant cells that escaped lysis by NK cells resulted in enrichment of guide RNAs targeting genes that confer vulnerability (e.g., ligand of an NK activation receptor). Mutant cells that gained sensitivity to NK cells and were killed more rapidly could be identified by the depletion of guide RNAs targeting genes that confer resistance (e.g., cancer-promoting genes). One of the genes with the strongest evasion scores in a genome-wide CRISPR screen for resistance to NK cells was SPPL3. SPPL3 is an intramembrane protease located in the Golgi apparatus that cleaves glycosyltransferases in their transmembrane region and releases their catalytic domain. The activity of SPPL3 reduces the accumulation of intracellular glycosyl-transferases and the extent to which proteins are glycosylated during transport through the Golgi. Accordingly, SPPL3 deletion in the B-cell lymphoblastoma cell line 221 led to an increase in complex N-glycans on plasma membrane proteins. Biochemical and functional experiments revealed that increased glycosylation of ligands for NK activation receptors NKG2D and CD2 interfered with receptor binding and lysis of the target cells by NK cells. Analysis of ligands for NK receptors in SPPL3-KO cells revealed slower migration during gel electrophoresis, consistent with higher glycosylation. Pharmacological inhibition of the N-glycan pathwayat a step preceding branching mediated by MGAT transferasesreversed the slower gel migration of ligands and restored a normal sensitivity of SPPL3-KO cells to lysis by NK cells. Binding of the clinical CD20 antibody rituximab was also impaired on SPPL3-KO cells. This suggests that escape from treatment of mature B-cell non-Hodgkin's lymphoma (NHL) and mature B-cell acute leukemia (B-AL) with rituximab can occur not only by downregulation of CD20 expression, but also by enhanced CD20 glycosylation. This should be an important consideration in the selection of antibodies for both monoclonal antibody therapies and single-chain variable fragments (scFvs) inserted into CARs. Previous research had linked the loss of a chromatin regions encompassing SPPL3 to reduced infiltration of T cells in lung adenocarcinoma. This may be due to elevated glycosylation and disruptions in ligand-receptor interactions. Our study suggests that NK cells may also be impaired in this respect. Unlike mutations in specific ligands for NK receptors, the impact of SPPL3 mutations in cancer is much broader. Having identified SPPL3 as a major regulator of sensitivity to NK cells, we went on to determine how a higher abundance of membrane-tethered intracellular glycosyltransferases impacts glycosylation of cell surface proteins and how this leads to target cell resistance. Was it mainly a quantitative effect due to higher numbers of glycosylated proteins and of glycosylation sites on proteins? Or was it more specific, due to changes in the type of N-glycans? To address this, SPPL3-KO cells were subjected to another round of selection by coculture with NK cells. This time a sgRNA library focused on genes that control glycosylation pathways was used to identify the type of glycans that may be responsible for resistance to NK cells. Two independent screens with this library identified MGAT2, which adds GlcNAc on the second branch after the trimming of high mannose precursors, and B3GNT2, a transferase specialized in the extension of branched N-glycans by adding GlcNAc in a b1,3 linkage to an unsialylated terminal galactose. B3GNT2 has also been identified as a driver of resistance of human melanoma cells to T-cell mediated cytotoxicity. B3GNT2 prefers to elongate the fourth branch, which is added by MGAT5. B3GNT2 competes with sialyltransferases for the same substrate, namely one of the 4 terminal Gal on tetra-antennary N-glycans. Sialyltransferases get a head start, as they sialylate any terminal Gal, including those on partially branched N-glycans, while B3GNT2 prefers tetra-antennary N-glycans. Remarkably, the resistance to NK-mediated lysis conferred by SPPL3-KO was further increased by deletion of ST6GAL1. As ST6GAL1 specializes in a2,6-linked sialylation of terminal Gal on branched N-glycans, its absence provides more opportunities for branch extension by B3GNT2. We performed a systematic analysis of peptide-specific recognition of 4 HLA-C ligands by 5 KIR members, including two activating KIR, and demonstrate that KIR can display exquisite peptide-specificity, recognizing just 1% of peptides screened in the case of KIR2DS4. Our data can explain the limitation of the old model and suggests a new model to interpret KIR-HLA-I associations with disease. We categorically show that KIR previously defined as weak (including activating KIR) are not intrinsically weak but are specific for peptide sequences that are rare on normal cells. Using a refined peptide motif for KIR binding we searched human, viral, and bacterial proteomes to identify potential ligands. We validated a functional peptide, conserved among several pathogenic bacteria, that stimulated primary KIR2DS4+ NK cells. We propose that differences in peptide-specificity is the key element that distinguishes one KIR-HLA-I interaction from another. Interpretation of disease associations with specific KIR-HLA-I combinations demands analysis of immunopeptidome data from tissues and cells implicated in the disease. Our work should stimulate those working on antigen presentation and HLA immuno-peptidomics to focus on the contribution of HLA-I bound peptides to the function of innate peptide-specific receptors. KIRs positively and negatively regulate NK cell responses upon recognition of HLA-I ligands. The numerous disease associations with specific combinations of KIR and HLA-I ligand imply that KIR-HLA-I interactions are not simply generic but that individual KIR members have unique properties, which must be identified and understood. We measured over 3,500 different KIR-peptide:HLA-C interactions and uncovered substantial differences in peptide-specificity among KIRs. Notably, 2DL1-C2-HLA-C was unique in being largely peptide agnostic, while four other KIR2D-HLA-C interactions had a sequence dependence ranging from highly selective to exquisitely specific. Further, intrinsic KIR affinity for HLA-C is similar between inhibitory and activating KIR, rejecting the proposition that activating KIR are weak receptors. Finally, we demonstrate that predicting peptide KIR ligands is feasible, paving the way for tools that can identify KIR ligands in immunopeptidomes from physiologically relevant tissue sites. Our data cement the KIR in the emerging family of germline-encoded NK cell receptors with specificity for HLA-peptide complexes. The task ahead is to identify specific KIR peptide-ligands in cells and tissues that lay at the core of the KIR-HLA associations with disease.