Signature of NK cells that are enriched for IL-10 production in malaria infection history subjects

Plasmodium falciparum infection can trigger high levels of inflammation that lead to fever and sometimes severe disease. People living in malaria-endemic areas gradually develop resistance and control both parasite numbers and the inflammatory response. Adaptive Natural Killer (NK) cells correlate with reduced parasite load and protection from symptoms. However, additional NK cell immunoregulatory roles may lower inflammation and reduce fever induction. We previously found that murine NK cell production of IL-10 can protect mice from experimental cerebral malaria. Human NK cells can also secrete IL-10, but it was unknown what NK cell subsets produce IL-10 and if this is affected by malaria infection history. Here, we show that NK cells from subjects with malaria history make significantly more IL-10 than subjects with no malaria history. We then determined the proportions of NK cells that are cytotoxic and produce interferon g (IFNg) and/or IL-10 and identified a signature of adaptive and checkpoint molecules on IL-10 producing NK cells. Lastly, we find that co-culture with primary monocytes, Plasmodium-infected RBCs, and antibody induces IL-10 production by NK cells. These data suggest that NK cells may help with protection from malaria because of IL-10 induction. Overall design: Single-cell RNA-Seq analysis was carried out using the ddSEQTM Single-Cell Isolator (Bio-Rad, Hercules, CA) and the SureCellTM WTA 3' Library Prep Kit (Illumina, San Diego, CA). Magnetically enriched NK cells from malaria naïve subjects were treated with cytokines IL-15, IL-21, IL-12 or IL-15 alone prior to capture on the ddSeq Single-Cell Isolator. Library preparation was done using SureCellTM WTA 3' Library Prep Kit. Resulting libraries were analyzed for an appropriate length distribution using an Agilent TapeStation, and initially sequenced on the Illumina MiSeq Nano with read lengths: read 1 = 70 bp, index = 8 bp, R2 = 76 bp. Deeper sequencing was then achieved using an Illumina NextSeq 550 High-ouput 2x75 flow cell (total = 6 samples). Per-cell gene expression was quantified for each sample using the BaseSpace® SureCellTM RNA Single-Cell Analysis Workflow v1.2.0 from Illumina, which uses Isas v1.2.6-000184develop for analysis, STAR v2.5.2b for read alignment to the Homo sapiens UCSC hg38 genome, and SAMtools v 1.3. Data was processed downstream using the Seurat single-cell analysis pipeline v4 in R v 4.1.0 (https://doi.org/10.1038/nbt.4096 ). Samples were filtered of low quality (number gene Features < 200) cells before normalization with SCTransform (https://genomebiology.biomedcentral.com/articles/10.1186/s13059-019-1874-1) using 3,000 variable genes, and batch correction using the Seurat integration procedure using the top 30 PCs (Stuart et al., 2019 ; https://www.cell.com/cell/fulltext/S0092-8674(19)30559-8). UMAP visualizations and nearest neighbor clustering (Louvain algorithm) were performed using the top 6 PCs. Purified NK cells were used and single cells that were determined CD3 negative by gene expression were 0 and were labeled as IL10 positive if IL10 raw counts = 1. NK cells that were CD3 negative and IL10 negative were compared to cells that were NK cells that were CD3 neg and IL10 positive (baseline group) with the Wilcoxon rank sum test using their SCT-normalized gene counts (assay=“SCT”, min.pct=0.1, logfc.threshold=0, only.pos=FALSE). Genes were considered differentially expressed if FDR < 0.05. Those DEGs were plotted using the Seurat DotPlot() function.