Sequential actions of EOMES and T-BET promote stepwise maturation of natural killer cells

EOMES and T-BET are related T-box transcription factors that control natural killer (NK) cell development. Here we demonstrate that EOMES and T-BET regulate largely distinct gene sets during this process. EOMES is dominantly expressed in immature NK cells and drives early lineage specification by inducing hallmark receptors and functions. By contrast, T-BET is dominant in mature NK cells, where it induces responsiveness to IL-12 and represses the cell cycle, likely through transcriptional repressors. Regardless, many genes with distinct functions are co-regulated by the two transcription factors. By generating two gene-modified mice facilitating chromatin immunoprecipitation of endogenous EOMES and T-BET, we show a strong overlap in their DNA binding targets, as well as extensive epigenetic changes during NK cell differentiation. Our data thus suggest that EOMES and T-BET may distinctly govern, via differential expression and co-factors recruitment, NK cell maturation by inserting partially overlapping epigenetic regulations.

Reproducibility of ChIPs. We compared the number of reads in detected peaks in the replicates. BEDtools makewindows v2.26.0 74 was used to compute all non-overlapping 10Kb long bins along the mouse genome. BEDtools intersect was used to count the number of reads falling into each bin for all IP samples. Read counts per bin are presented in a scatterplot and the Spearman correlation coefficient is computed. Heatmaps were generated using Deeptools v3.5 75d using the tool bamCoverage to generate bigwigs files with a step of 10 nt. Bigwig files were normalized using the RPGC method. Then, the tool Deeptools computeMatrix v3.5 was used to generate a count matrix at the positions of interest and finally the tools Deeptools plotHeatmap v3.5 and plotProfile v3.5 were used to generate heatmaps and mean profile plots. Data presented on the heatmap and mean profile are pooled by condition.
ChIP-seq and RNA-seq data that support the findings of this study have been deposited in the Geo repository with the accession code GSE168242 (https:// www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE168242).
ChIP-seq datasets. The following published datasets were downloaded from GEO or SRA Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.

Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Life sciences study design
All studies must disclose on these points even when the disclosure is negative. We did not perform sample size calculations. Sample size was determined to be adequate based on the magnitude and consistency of measurable differences between groups, as well as feasibility of performing highly technical experiments with a rare cell population.
In all experiments NK cells were analyzed by flow cytometry, we performed at least three biological replicates for each group in each experiment. In general, the number of biological replicates largely exceeded this number. For RNAseq experiments, three replicates were used in each group. For the ChIP-seq experiment, we only used two replicates because this type of experiment is technically very challenging and requires lots of mice for a single experiment.
We did not exclude data All experiments were reproduced at least three times as stated above.
This is not relevant in our experiment, as all comparisons were performed between defined groups ie control vs knockout mice.
Blinding wass not performed during data collection. However, RNA-seq data and ATAC-seq samples were processed by separate scientists and each data set was analyzed by separate bioinformaticians.

Data access links
May remain private before publication.

Files in database submission
Genome browser session (e.g. UCSC) Methodology