Gp96 deficiency inhibits NK cell response to IL-15

To investigate the underlying mechanisms of gp96-mediated NK cell development and cytotoxicity, we set out to study whether gp96 regulates the response of NK cells to IL-15, which plays a critical role in NK cell maturation and effector functions²⁸. Bone marrow cells of gp96-deficient mice were cultured in the presence of IL-15 with/without SCF. Unlike WT cells, gp96^–/– bone-marrow cells failed to efficiently develop mature NK cells (Fig. 3A). Moreover, splenocytes from WT and gp96 KO mice were stimulated with different doses of IL-15 in vitro for 1 week. The frequency of the total NK cells among leukocytes was moderately higher in WT mice than that in gp96 KO mice (Fig. 3B), and notably, gp96 deficiency abrogated the increased effect of DX5⁺ mature NK cells by IL-15 treatment as seen in WT NK cells (Fig. 3C). Low proportion of NK cells was detected from splenocytes even after 7 days of culture without IL-15 (Fig. 3C), which may be due to stimulation of trace amounts of autocrine IL-15 and other cytokines by splenocytes²⁹.

Open in a separate window

Fig. 3

NK-specific gp96 deficiency decreases NK cell response to IL-15.

A In vitro NK-cell differentiation from bone-marrow cells. Bone-marrow cells cultured with SCF in the presence (upper panel) or absence (lower panel) of 10ng/ml IL-15 were subjected to the flow cytometric analysis. Frequencies of total NK cells (B) and DX5⁺ NK cells (C) among splenocytes treated with the indicated dose of IL-15 for 7 d. Primary NK cells were isolated and treated with 10ng/ml IL-15 for detection of p-Stat5 (D) and p-S6 (F) levels. E Intracellular staining of p-Stat5 and Stat5 in splenic NK cells (gated on CD3^-NK1.1⁺) after being stimulated with IL-15 (10ng/ml). G, H NK92 cells were stimulated with IL-15 (10ng/ml) after treatment with PU-WS13 or control. At indicated time points, cells were harvested and lysed, followed by immunoblot using a phosphorylated (p)-Stat5, p-S6 antibody. The data are representative of two independent experiments with similar results. Dots represent data from n=5 mice/group. Mean±SD is shown. Statistical significance was determined using two-tailed unpaired t test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. p values: (A) p=0.0003 (NKp, WT BM+SCF vs WT BM+SCF+IL-15), p=0.0168 (NKp, KO BM+SCF vs KO BM+SCF+IL-15), p=0.0005 (mNK, WT BM+SCF vs KO BM+SCF), p=0.0005 (mNK, WT BM+SCF vs WT BM+SCF+IL-15), p=0.0002 (mNK, WT BM+SCF+IL-15 vs KO BM+SCF+IL-15), (B) p=0.0052 (5ng/ml), p=0.0226 (10ng/ml), p=0.0008 (20ng/ml), (C) p=0.0004 (0ng/ml), p<0.0001 (5ng/ml), p<0.0001 (10ng/ml), p<0.0001 (20ng/ml), (F) p=0.0027.

IL-15 promotes NK cell development by activating several downstream signaling pathways, including the JAK-STAT5 and PI3K-ATK-mTOR pathways^9,28. Upon IL-15 stimulation, compared to WT cells, gp96-deficient splenic NK cells showed abruptly decreased phosphorylation of STAT5, as well as p-S6, a reliable marker of mTOR activation (Fig. 3D–F). Similarly, inhibition of gp96 with a specific inhibitor in NK92 cells also decreased the activation of STAT5 and mTOR (Fig. 3G, H).

Gp96-deficient NK cells have specific and shared transcriptome signatures with Eomes-deficient cells

To further investigate the molecular characteristics of spleen NK cells from Ncr1^Cregp96^fl/fl mice, we performed 10× genomics scRNA-seq. A total of 15699 single NK cells (7349 WT and 8350 Ncr1^Cregp96^fl/fl NK cells) were analyzed, and a mean of 1376 gene transcripts in WT and 1368 in Ncr1^Cregp96^fl/fl NK cells was detected. Unsupervised clustering of all sequenced NK cells revealed six distinct clusters based on transcript signatures (Fig. 4A), with each cluster expressing unique gene markers (Fig. 4B). NK cells from gp96^–/– and WT mice exhibited distinct cluster patterns (Fig. 4C and Fig. ^S2A). Of note, around 29% of the gp96^-/- NK cells were distributed in cluster 3, in comparison with only 3% of WT cells. In contrast, cluster 4 contained 0.3% of gp96^–/– and 18% WT NK cells. These clusters displayed key features in gene expression changes along NK cell maturation, including expressions of genes for effector molecules (Gzmb and Prf1) and genes for maturation markers (CD11b/Itgam, Cma1 and Klrg1) and immature markers (Cd7, and Xcl1)³⁰ (Fig. 4D). Compared to cluster 3, cluster 4 exhibited higher expressions of Itgam, Klrg1, Cma1 and Cd27.

Open in a separate window

Fig. 4

High-throughput single-cell RNA-seq of NK cells reveals gp96^–/– clusters with distinct transcriptional signatures.

A A t-Distributed Stochastic Neighbor Embedding (tSNE) and graph visualization of the 15699 single NK cells defining 6 clusters. B Heatmap of marker genes in scRNA-seq clusters. Columns: single cells. Rows: cluster marker genes. Representative genes that are differentially expressed are on the left. C Cell number of Ncr1^Cregp96^fl/fl and WT NK cells within each cluster. D Feature dot plot showing the relative expression levels of the indicated genes in each cluster from (A). E Velocity analysis of the origin and direction of NK cell maturation. Velocity fields were projected onto the t-SNE plot. F KEGG analysis of DEGs for indicated clusters. Statistical significance was determined using one-sided Fisher’s Exact Test with adjustments for multiple comparisons typically using methods like the Benjamini-Hochberg procedure. Selected KEGG terms with adjusted P values<0.05 are shown.

Next, we traced cell fate and reconstructed cell lineage direction using the RNA velocity approach. NK maturation in WT cells followed a single main branch (cluster 3/5 through 2 to cluster 0/1 and 4) without significant division. By contrast, velocity analysis revealed a different branch via cluster 3 through 4 to cluster 0/1 under gp96 knocking out condition (Fig. 4E), which is speculated that cluster 3 may be blocked in a relatively immature developmental stage. Since cluster 3 and cluster 4 are the different major clusters between WT and gp96 KO cells in both cell number and maturation process (Fig. 4C, E, Fig. ^S2D), we performed differential gene expression analysis among them. As seen in Fig. 4F, natural killer-mediated cytotoxicity was significantly enriched in cluster 4 (P.adj= 8.4×10⁻⁹) but not significantly enriched in cluster 3 (P.adj=0.076) by KEGG enrichment. Taken together, these results showed that NK-specific gp96 deficiency reduces its maturation and cytotoxic function.

To understand the molecular mechanisms whereby gp96 controls NK cell differentiation, we analyzed the expression of key transcriptional factors^11,31 in NK lineage development in gp96-deficient vs. WT NK cells. T-bet and Eomes play major roles in controlling NK cell maturation and function¹¹. Using the mouse model expressing endogenously tagged Eomes¹², splenic NKs were isolated for ChIP analysis. WT and Eomes^-/- NK cells¹² were performed for RNA-seq to identify differentially expressed genes. Combination the ChIP-seq analysis with RNA-seq results was used to identify direct Eomes target genes. The overlap between Eomes-bound (ChIP-seq) and Eomes-regulated genes (RNA-seq) was identified as direct Eomes target genes. Analysis of the Eomes target genes revealed that the majority of upregulated transcripts in Eomes knockout NK cells were also upregulated in gp96 depleted NK cells, and most of the downregulated transcripts were downregulated simultaneously (Fig. 5A). By contrast, no such a trend was seen in T-bet knockout NK cells (Fig. ^S3A). In addition, genes differentially expressed between gp96 knockout and WT NK cells were correlated to those of Eomes^–/– and Eomes⁺ NK cells³² (Fig. 5B). There was a significant decrease of Eomes levels in gp96-depleted NK cells compared to wild-type cells, as determined by flow cytometry (Fig. 5C and Fig. ^S3B) and western blot analysis (Fig. 5D). No obvious changes of Id2 and T-bet were observed under gp96 deficiency (Fig. ^S3B). Moreover, overexpression of Eomes largely rescued the effect of gp96 KO on DX5 expression in NK cells (Fig. 5E). These data indicate that gp96 may manipulate NK cell maturation mainly via Eomes.

Open in a separate window

Fig. 5

Effect of gp96 on Eomes expression in NK cells.

A The fold changes indicated the difference in relative transcript expression of Eomes-bound genes in DX5⁺ splenic NK cells between WT compared with Ncr1^Cregp96^fl/fl mice, as determined by RNA-seq. Y-axis showed Log2 fold changes of WT vs. KO mice. B Correlation analysis of Log₂ fold changes of Eomes-targeted genes between gp96^–/– NK versus WT NK and Eomes^–/– versus WT NK. C Flow cytometry analysis of levels of Eomes between Ncr1^Cregp96^fl/fl and WT NK cells. D Western blot analysis of Eomes levels between Ncr1^Cregp96^fl/fl and WT NK cells from spleen and bone marrow (BM). E Flow cytometry analysis of DX5⁺ NK cell percentage in WT or gp96 KO BM hematopoietic stem cells infected with Eomes or control Lentiviral vector. Mean±SD of three replicates is shown. Statistical significance was determined using two-tailed unpaired t test. *p<0.05, ***p<0.001, ****p<0.0001. p values: (B) p=0.0004, (C) p<0.0001 (spleen), p=0.0002 (BM), (D) p<0.0001 (spleen), p<0.0001 (BM), (E) p=0.0003 (WT-control vs KO-control), p=0.0117 (KO-control vs KO-Eomes).

Cellular gp96 binds to and protects Eomes from autophagy-lysosome-mediated degradation

Flow cytometry analyses show that expressions of gp96 and Eomes had a similar trend during NK maturation, which were high in CD11b⁺CD27⁺, and then decreased in CD27^- CD11b⁺ both in mouse spleen NK cells (Fig. 6A) and human PBMC NK cells (Fig. ^S4A). A confocal microscopy analysis further showed that cytoplasm gp96 levels were associated with NK maturation stages (Fig. 6B). The mRNA level of Eomes was not significantly reduced under gp96 depletion in sorted splenic NK cells (Fig. ^{S4B, C}), indicating that gp96 affects Eomes expression at the posttranscriptional level. Confocal microscopy results showed that gp96 colocalized with Eomes dominantly in the nucleus and minorly in cytosol but not in the ER of CD11b⁺CD27⁺ NK cells (Fig. 6C–E, and Supplementary Movie ¹, ³).

Open in a separate window

Fig. 6

Gp96 inhibits autophagy-lysosome-mediated degradation of Eomes.

A Flow cytometry analysis of Eomes and gp96 levels among CD27 single positive, double positive (DP), CD11b single positive NK cells in mouse spleen. Dots represent data from n=5 mice/group. B Immunostaining of gp96 in spleen NK cells sorted from WT mice. The mean fluorescence intensity of gp96 was analyzed by Imaris 9.7. Bar, 10μm. C–E CD11b⁺CD27⁺ double positive NK cells were sorted from WT mice and subsequently stained for nucleus (Hoechst), Eomes, gp96, Calregulin and α-tubulin for confocal microscopy analysis. Scale bars, 3μm for original shots (C–E) and 1μm for magnified views (C). Z-stack images of gp96 and Eomes in whole cell, cytosol and nucleus, respectively. Percentage of ROI colocalized for cytosol and nucleus were obtained by Imaris. Percentages of colocalization of gp96 and Trim28 in the nucleus and cytosol were calculated, respectively. Scale bars, 3μm. Dots represent data from n=3 fields. (D). F Western blot analysis of Eomes levels in spleen NK cells sorted from WT and gp96 KO mice. Cells were treated with 50μg/ml CHX for the indicated times. The band intensity at 0h in WT NK cells was arbitrarily taken as 1.0. G Western blot analysis of Eomes levels in 293 cells transfected with Flag-gp96 or the empty vector. Cells were treated with 50μg/ml CHX for the indicated times. The band intensity at 0h in vector cells was arbitrarily taken as 1.0. H 293 cells stably expressing Eomes (left) and NK cells from gp96 KO mice (right) were treated with either 40µM CQ or 10µM MG132 for 6h and subjected to western blotting. I WT and Atg5 knockout 293 cells were transfected with His-Eomes and Flag-gp96 or a control vector. Cells were lysed and subjected to western blotting. J WT, Atg5 and gp96/Atg5 double knockout 293 cells were transfected with His-Eomes and subjected to western blotting. K WT and gp96 knockout HEK293 cells were transfected with His-Eomes. Cells were grown for 24h and treated with 20nM of BafA1 for 4h, followed by IP-Western analyses. L WT and gp96 knockout 293 cells were co-transfected with His-Eomes and GFP-P62. Cells were then treated as (H) and subjected to IP-Western analyses. The data are representative of two independent experiments with similar results. Mean±SD is shown. Statistical significance was determined using two-tailed unpaired t test. *p<0.05, **p<0.01, ****p<0.0001. p values: (A) p<0.0001 (Eomes, CD27 SP vs DP), p<0.0001 (Eomes, DP vs CD11b SP), p=0.0193 (gp96, CD27 SP vs DP), p=0.0391 (gp96, DP vs CD11b SP), (B) p<0.0001 (CD27 SP vs DP), p=0.0097 (DP vs CD11b SP).

Next, we explored the mechanism of gp96 KO-induced Eomes reduction. Depletion of gp96 in splenic NK cells and HEK293 cells stably expressing Eomes caused a moderate decrease in Eomes stability (Fig. 6F, Fig. ^S4D), whereas ectopic expression of gp96 caused an obvious increase in Eomes stability (Fig. 6G). Expression of Eomes was mostly increased in the presence of a lysosome inhibitor CQ, but not proteasome inhibitor MG132 in both primary NK and HEK293 cells (Fig. 6H), suggesting lysosome mediated degradation of Eomes. Moreover, the effect of gp96 on Eomes levels was abolished in the absence of Atg5 (Fig. 6I, J), which plays a vital role in autophagosome formation by promoting the conversion of LC3-I to LC3-II³³. Gp96 KO facilitated Lys 63-linked, but not Lys 48-linked, polyubiquitin chains of Eomes protein (Fig. 6K, Fig. ^S4E) after autophagy inhibition with Baf A1. Since autophagy receptors play an essential role in selective autophagy as scaffolding proteins for recognizing ubiquitinated substrates, we determined whether the classical autophagy receptor SQSTM1 (P62) is involved in Eomes degradation. We found that the SQSTM1 interacted with Eomes, and the interaction was increased in the absence of gp96, and vice versa (Fig. 6L, Fig. ^S4F). Taken together, our results show that gp96 inhibits Eomes polyubiquitin and the subsequent autophagy-lysosome-mediated degradation.

Cell lines and plasmid construction

B16F10 cells from our laboratory were cultured in-house. The HEK293 cell line, NK92 human NK cell lines, MC38 colon cancer cells, and LLC lung cancer cells were purchased from the Cell Resource Center, Peking Union Medical College. The cell line was checked free of mycoplasma contamination by PCR and culture. Its species’ origin was confirmed with PCR. The cell line’s identity was authenticated with STR profiling (FBI, CODIS). All the results can be viewed on the website (http://cellresource.cn). B16F10 cells were cultured in RPMI 1640 medium with 10% fetal bovine serum (FBS), penicillin, and streptomycin (100 IU/ml). HEK293, MC38, and LLC cells were cultured in Dulbecco’s modified Eagle medium with 10% FBS, penicillin, and streptomycin (100 IU/ml). NK92 cells were cultured in α-MEM with 12.5% FBS, 12.5% horse serum, penicillin, and streptomycin (100 IU/ml), and IL-2 (10ng/ml). All plasmids used in this paper were purchased by iGene Biotechnology Co., Ltd.

Western blot and co-immunoprecipitation assay

The stimulated NK92 cells or sorted splenic NK cells were harvested, and an equal number of cells were directly lysed. Primary and secondary antibodies used in western blot and were listed in Supplementary Table ³. For endogenous Co-IP, cells were lysed with 0.25% NP-40 lysis buffer (20mM Tris-HCl, 125mM NaCl, 5mM MgCl2, 0.2mM EDTA, 12% Glycerol, and 0.25% Nonidet P-40). Equal amounts of total protein were incubated with indicated antibodies or normal mouse IgG overnight at 4°C, and then 40μl of protein A/G beads were added for an additional 2h of incubation. For exogenous Co-IP, anti-HA beads (or anti-flag beads, or anti-myc beads) were added to equal amounts of total protein and incubated for 3h. Beads were centrifuged (800rpm for 2min) and washed five times using wash buffer (20mM Tris-HCl, 125mM NaCl, 5mM MgCl₂, 0.2mM EDTA, and 0.1% Nonidet P-40). The beads were heated at 100°C for 10min before SDS-PAGE and immunoblotting.

Cell fractionation

Cell fractions were harvested using Plasma Membrane Protein Isolation and Cell Fractionation Kit (Invitrogen) following the manufacturer’s instructions. Fourty micrograms of protein were analyzed by SDS-PAGE and western blot.

Immunostaining for confocal analysis

Murine NK cells were isolated from the spleen by magnetic cell sorting using EasySep Mouse NK Cell Isolation Kit (Stemcell) following the manufacturer’s instructions. A total of 50-90% pure NK cells were obtained using this procedure. Cells were then subsequently sorted into different subsets using a FACSAria Cell Sorter (Becton–Dickinson, San Jose, USA). After 4% PFA fixation for 15min, blocking (3% BSA in PBS) for 30min at RT was performed. Immunostainings were performed after a permeabilization step with 0.05% Triton X-100 for 30min. Primary antibodies were added to the cells for one-hour incubation at RT. After washes with PBS, cells were incubated with the appropriate secondary antibodies along with Hoechst for 30min at RT. After washes with PBS, cells were observed with a Leica SP8 laser scanning confocal microscope. The images and relative quantification were processed using Imaris (ver 9.7.2).

Z-stack images of gp96 and Trim28 in whole cell, cytosol and nucleus were generated by Imaris. Before analysis, a new surface for nucleus was created using the Hoechst signal. Nucleic signal and cytosol signal were obtained by the mask channel function in Imaris. Signals retained in the cell nucleus are considered as nuclear gp96 or Trim28, while signals outside the nucleus are regarded as cytosolic gp96 or Trim28. The resulting images were processed by analyzing colocalization, and the percentage of ROI colocalized for cytosol and nucleus were obtained respectively. A total of 3 fields randomly were selected and quantified in each assay. The 3D videos were provided in the supplementary materials. Primary and secondary antibodies used in immunofluorescence were listed in Supplementary Table ³.

In vivo ubiquitination assays

For the in vivo ubiquitination assay, HEK293 cells were co-transfected with expressing plasmids encoding Myc-Eomes and Trim28. Cells were grown overnight and treated with 20nM Bafilomycin A1 (BafA1) for 4h before harvesting. Cell lysates were immunoprecipitated using anti-Myc resin, followed by Western blot analyses with ubiquitin, ubiquitin-Lys 48-only or ubiquitin-Lys 63-only.

Flow cytometry

The spleen and LNs were collected and single-cell suspensions of lymphocytes were prepared by mechanical disruption in 1640 medium supplemented with 2% FBS. Peripheral blood mononuclear cells (PBMC) were freshly isolated from healthy donors. Written informed consent was obtained for each participant according to institutional guidelines. All experiments were approved by the Fifth Medical Center of PLA General Hospital (permit number KY-2020-11-5-1). Single-cell suspensions were stained with the appropriate monoclonal antibody in phosphate-buffered saline containing 5% serum. To detect phosphorylated signaling proteins, NK cells were fixed with Phosflow Lyse/Fix buffer, followed by permeabilization with Phosflow Perm buffer III (BD) and staining with antibodies. All other intracellular proteins were stained according to the manufacturer’s instructions using Foxp3/Transcription Factor Staining Buffer Set Kit (eBioscience). Fortessa and FACSAria III (BD Biosciences, San Diego, USA) were used for analysis and cell sorting, with dead cells excluded by the LIVE/DEAD Fixable Violet Dead Cell Stain Kit (Invitrogen, Carlsbad, USA). Primary antibodies used in flow cytometry were listed in Supplementary Table ⁴.

RNA-seq and analysis

CD3^-NK1.1⁺DX5⁺ NK cells were sorted to a typical purity of >98%. RNA-seq and bioinformatics analysis were conducted by Shanghai Biotechnology Corporation.

Single-cell RNA sequencing and data processing

Freshly isolated splenocytes were stained with anti-CD3 (17A2), and anti-NK1.1 (PK136) antibodies and NK cells were sorted to >99% purity using a BD FACSAria™ III. Splenic NK cells pooled from three mice of each indicated genotype were enriched by FACS for library preparation, and scRNA-seq was performed by Shanghai Biotechnology Corporation. The final library pool was sequenced on the Illumina NovaSeq 6000 instrument using 150-base pair paired-end reads. Raw sequencing data were converted to FASTQ files and aligned to the mouse genome reference sequence (GRCH38). The 10X Genomics Cell Ranger (version 3.0.1) was used to demultiplex samples, process barcodes, and generate a digital gene-cell matrix from this data. The output was then imported into the Seurat (version 3.1.5) R toolkit for quality control and downstream analysis. The percentage of ribosome genes is 0.326 and 0.438% in gp96^fl/fl and Ncr1^Cregp96^fl/fl cells, respectively. Cells with less than 200 genes and cells with above 20% mitochondrial genes were removed as low-quality cells. Cells with abnormally high unique molecular identifiers or gene numbers were also identified as putative doublets and removed. After quality control and filtering steps, 7349 cells from gp96^fl/fl mice and 8350 cells from Ncr1^Cregp96^fl/fl mice were used for further analyses. The Louvain algorithm was used for the unsupervised computational analysis of scRNA-seq data.

Generation of BM chimeras

To generate mixed BM chimeras, BM cells were isolated from the femurs and tibias of CD45.1⁺WT and CD45.2⁺ Ncr1^Cregp96^fl/fl mice. WT recipient mice (CD45.1⁺CD45.2⁺) were sub-lethally gamma-irradiated (7Gy) and intravenously transplanted with a mixture (1:1) of WT (CD45.1) and KO (Ncr1^Cregp96^fl/fl) donor bone marrow cells (5×10⁶). NK cells were analyzed 8 weeks after reconstitution.

Cytolytic assays

For cytolytic assays against YAC-1 cells, CFSE–labeled YAC-1 cells were cocultured with effector cells (mouse splenic NK cells) at the indicated effector:target (E:T) ratios for 4hours. Next, cell mixtures were stained with 7-AAD to calculate killing efficiency as the percentage of 7-AAD⁺ YAC-1 cells with flow cytometry. To determine NK cell cytotoxic activity, 2 × 10⁶ CFSE-labeled YAC-1 cells were intraperitoneally injected into WT and KO mice, respectively. Cells in the peritoneal cavity were harvested, and CFSE-labeled YAC-1 cells were measured 24h post-injection.

Transfection of primary NK cells

Murine NK cells were sorted from the spleen using EasySep Mouse NK Cell Isolation Kit (Stemcell) following the manufacturer’s instructions. 3 × 10⁶ NK cells were placed into each well of a 12-well culture plate and then transfected using Advanced DNA RNA Transfection Reagent (ZETA LIFE, AD600025). 10μg plasmid was directly mixed with transfection reagent according to 1:1 relationship, then use a pipette to blow 10-15 times to mix. After incubation at room temperature for 15minutes, the transfection complex was prepared. Add transfection complex to the cell culture plate and mix gently, place in the CO₂ incubator and continue to culture. Cells were grown for 48h, followed by IP-Western analyses.

Isolation of progenitor stem cells and NK differentiation in vitro

Bone marrow cells were obtained from WT or KO mice. Progenitor stem cells (PSCs) were enriched using an EasySep Mouse Progenitor Stem Cell Negative Selection kit (StemCell Technologies, Canada). Transduction was conducted immediately after the enrichment of PSCs using the spin protocol. PSCs (0.3 × 10⁶) were placed into each well of a 48-well culture plate. After transduction, the virus-containing supernatant was removed. The transduced PSCs were cultured in NK differentiation conditioned medium⁴⁶ RPMI 1640 supplemented with 10% FBS, 1% PSG, 1.6mmol/l 2-mercaptoethanol, 0.5ng/ml of murine IL-7, 30ng/ml of stromal cell factor, and 50ng/ml FMS-related tyrosine kinase 3 ligand. On day 3, after transduction, 0.5ml of fresh media was added. On day 5, cells were pelleted, old media was removed, and cultured in 0.5ml of complete RPMI media containing 30ng/ml of IL-15. On day 8, 0.5ml of fresh media was added. On day 10, old media was replaced with fresh media containing 30ng/ml of IL-15 and 1000 U/ml of IL-2. On day 14, differentiated NK cells were analyzed by flow cytometry.

Virus production and transduction protocols

Bone marrow progenitor cells and primary NK cells were enriched by magnetic cell sorting. The transduction of cells was performed in the following protocol. Cells were pelleted at 400×g for 5min at room temperature in 1.5ml screw-cap tubes and resuspended in 0.25ml of viral supernatant (MOI=10) in the presence of 8μg/ml of Polybrene (Sigma, St. Louis, MO). Cells were incubated for 2h at 37°C and 5% CO₂. At the end of transduction, the virus-containing supernatant was removed.

BM NK differentiation in vitro

Bone marrow cells were cultured in RPMI1640 medium containing 10% fetal calf serum and 100ng/ml stem-cell factor (SCF), with or without 50ng/ml rIL15 (Genzyme). Flow cytometric and cytolytic analyses were done after culture for 10 days.

Tumor models

Single-cell suspension of MC38 colon cancer or LLC lung cancer cells was injected subcutaneously into female WT or KO mice (2 × 10⁵ cells per mouse). Mice were euthanized on days 21 to 28 following tumor injection for analysis of tumor-infiltrating lymphocytes. Tumor volumes were monitored with a caliper and calculated using the formula: V (in mm³)=0.5 (ab²), where a is the longest diameter and b is the shortest diameter.

All animal experiments were performed in strict accordance with institutional guidelines on the handling of laboratory animals. Mice were euthanized when the maximum tumor size (diameter: 2.0cm) had been reached. The diameter of the tumor is measured by a caliper. The health of the animals was monitored every other day and there were no unexpected deaths. To minimized mice suffering and distress, the only tumor was implanted in the subcutaneous site in the flank on each animal. Criteria for early termination are listed as below: If the tumor meets the tumor size limitations; If the tumor becomes ulcerated or necrotic; If the animal’s ability to eat or drink is compromised; If the animal’s ability to ambulate normally or breathe is impaired. Cervical dislocation was applied in the study to minimized animal suffering and distress.

Isolation of tumor-infiltrating lymphocytes

Tumor-infiltrating lymphocytes (TILs) were isolated by dissociating tumor tissue in the presence of collagenase I (0.1% w/v, Sigma) and DNase (0.005% w/v, Sigma) for 1h before centrifugation on a discontinuous Percoll gradient (GE Healthcare). After centrifugation, the white opaque layer at the interface between the two Percoll solutions, mostly containing leukocytes, is collected. Isolated cells were then used in various assays of NK cell function.

Intravenous lung metastasis assay

For lung experimental-metastasis studies, 2.5 × 10⁵ B16F10 cells were injected intravenously into WT or KO mice. The number of B16F10 melanoma surface nodules in the lungs of each mouse was counted. Two weeks after the injection, lung tumor nodules were counted.

Quantification and statistical analysis

FACS data were collected and processed using FACS software (FlowJo, version 10). GraphPad Prism 6 software was used to analyze data using a two-tailed paired Student’s t test or two-way ANOVA. Data are represented as mean±SD. P<0.05 were considered significant.

We would like to thank Tong Zhao for FACS sorting and technical help. We thank Xiaolan Zhang for performing the confocal laser scanning microscopy and SIM imaging. This work was supported by a grant from National Key R&D Program of China (2022YFC2304203), the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB29040000), the Industrial innovation team grant from Foshan Industrial Technology Research Institute (2018HS), and grants from the National Natural Science Foundation of China (81871297, 81903142 and 32070163), Foshan High-level Hospital construction DengFeng plan, and Guangdong Province biomedical innovation platform construction project tumor immunotherapy.