Kir6.1, a component of an ATP-sensitive potassium channel, regulates natural killer cell development

Abstract

Introduction: Involved in immunity and reproduction, natural killer (NK) cells offer opportunities to develop new immunotherapies to treat infections and cancer or to alleviate pregnancy complications. Most current strategies use cytokines or antibodies to enhance NK-cell function, but none use ion channel modulators, which are widely used in clinical practice to treat hypertension, diabetes, epilepsy, and other conditions. Little is known about ion channels in NK cells.

Results: We show that Kcnj8, which codes for the Kir6.1 subunit of a certain type of ATP-sensitive potassium (K_ATP) channel, is highly expressed in murine splenic and uterine NK cells compared to other K⁺ channels previously identified in NK cells. Kcnj8 expression is highest in the most mature subset of splenic NK cells (CD27^-/CD11b⁺) and in NKG2A⁺ or Ly49C/I⁺ educated uterine NK cells. Using patch clamping, we show that a subset of NK cells expresses a current sensitive to the Kir6.1 blocker PNU-37883A. Kcnj8 does not participate in NK cell degranulation in response to tumor cells in vitro or rejection of tumor cells in vivo, or IFN-γ release. Transcriptomics show that genes previously implicated in NK cell development are amongst those differentially expressed in CD27^-/CD11b⁺ NK cells deficient for Kcnj8. Indeed, we found that mice with NK-cell specific Kcnj8 gene ablation have fewer CD27^-/CD11b⁺ and KLRG-1⁺ NK cells in the bone barrow and spleen.

Discussion: These results show that the K_ATP subunit Kir6.1 has a key role in NK-cell development.

Keywords: ATP-sensitive potassium channels; NK cell development; NK cells; innate immunity; ion channels; potassium channels.

Figure 1

Kcnj8 mRNA expression in NK cells. (A) Kcnj8 is expressed in cytotoxic effector cells in mouse immune cells. Shown are ImmGen ULI RNA seq expression data of bone marrow hematopoietic stem cells (LTHSC_34-BM), splenic germinal center centrocytes (B_GC_CC_Sp), splenic plasma cells (B_PC_Sp), thymic preT DN1 cells (preT_DN1_Th), splenic natural killer T cells (NKT_Sp), splenic CD25hi TRegs (Treg_4_25hi), splenic naïve CD8⁺ T cells (Sp T_8_Nve_Sp), splenic CD8⁺ T cells 7 days after LCMV infection (T8_TE_LCMV_d7_Sp), splenic CD8⁺ T cells 180 days after LCMV infection (T8_Tem_LCMV_d180_Sp), immature thymocytes (Tgd_g2+d1_24a+_Th), Tgd_g2+d17_LN, total splenic dgT cells (Tgd_Sp), splenic CD27^-/CD11b⁺ NK cells (NK_27-11b+_Sp), lamina propria NKp46+ ILC3 cells (ILC3_NKp46+_SI), splenic CD8⁺ cells (DC_8+_Sp), bone marrow neutorophils (GN_BM), thio-induced peritoneal neutrophils (GN_Thio_PC), peritoneal macrophages (MF_PC), thymic medullary epithelial cells (Ep_MEChi_Th), subcutaneous lymphatic endothelial cells (LEC_SLN), and subcutaneous lymphatic pericytes (IAP_SLN). Expression values are normalized by DESeq2. Not shown are immune cell types with expression levels below 0.1. Data are from http://rstats.immgen.org/Skyline/skyline.html. (B) qRT-PCR data showing Kcnj8 expression in isolated splenic NK cells and brain tissue from wild-type C57BL/6 mice and NK-cell specific NKp46-Kir6.1conditional knock-out mice. (C) RNAscope data of a mouse spleen. RNAscope probes used were designed to localize NK cells (Ncr1; red), T cells (Cd3; green), macrophages (Abgre1; magenta), and cells expressing Kcnj8; light blue). Shown in the left are unprocessed images, with DAPI staining in gray scale. Cell segmentation was performed with DAPI as the cell marker and is shown on the right. Detected cells are outlined in white. Cells are pseudo colored based in expression of RNAscope dots. The top and bottom rows are two representative images.

Figure 2

Single cell RNA seq data were obtained using NK cells of mouse spleen that were isolated using a negative selection method. Data analysis was performed with the Seurat package. (A) Sixteen separate cell clusters were identified by graph-based clustering algorithm. Non-linear dimensional reduction was performed and data are displayed as t-distributed Stochastic Neighbor Embedding (tSNE) plots. (B) Marker feature tSNE plots show lack of expression of CD3e (CD3E) in populations 0 to 3, and expression of the NK cell markers Klrb1c (NK1.1), Ncr1 (NKp46), Cd27 (CD27) and Itgam (CD11b). Also shown is high expression of Klra3 (Ly49C) in mature NK cells (population 0).

Figure 3

Kcnj8 mRNA expression is highest in the mature CD11b⁺ NK cells in the mouse spleen. These data focus on populations 0 to 3 that are enriched in NK cell makers, identified in Figure 2 . (A) Single cell violin plots are shown for Kcnj8 (Kir6.1), and the NK cell maturity markers Itgam (CD11b), and Cd27 (CD27). (B) Single cell violin plots of the scRNA seq data were generated to show the expression in populations 0 to 3 of ion channel genes Kcnj8 (Kir6.1), Kcnj11 (Kir6.2), Abcc9 (SUR2), Kcna3 (Kv1.3), Kcnn4 (Kv1.3), Kcnk6 (TWIK-2), Mcoln1 (TRPML1), Mcoln2 (TRPML2), and Trpm2 (TRPM2). Note that Abcc8 (SUR1) had no gene counts in this dataset.

Figure 4

Kcnj8 is expressed at higher levels in mature, educated uterine NK cells. Volcano plot showing selected DEGs in NKG2A+ [(A), n=9] and LY49C/I+ [(B), n=6] educated uNK compared to uneducated uNK cells (LY49C/I- NKG2A- n=9). Red color indicates statistical significance (FDR<0.05). Fold change expressed in log2 relative to uneducated uNK cells. A list of all DEG with their FC and FDR is accessible in Supplementary Tables S5A and S5B.

Figure 5

Patch clamp current recordings in isolated mouse spleen NK cells. NK cells, isolated from wild-type mouse splenocytes by negative selection, were subjected to whole-cell patch clamp recordings. (A) An example of recordings of a cell that exhibited time-dependent currents upon depolarization with little inward rectification. Shown on the left are recordings made in the absence of drug, 3 min after application of 100 µM pinacidil, and 3 min after application of 10 µM PNU-37883A (in the continued presence of pinacidil). (B) Depicted is the current, measured at the end of the voltage step, as a function of applied voltage of this cell, demonstrating pronounced outward rectification. (C) A typical example of a cell with a distinctly different current profile, that showed little time dependence and the presence of strong inward currents upon hyperpolarization. (D) The current-voltage relationship of this cell was essentially linear, and the current was strongly blocked by PNU-37883A.

Figure 6

NK cell degranulate effectively in both WT and NK-cell specific Kcnj8 KO mice. (A) Flow cytometry gating strategy to detect CD107a+ (degranulating) NK cells in a co-culture of NK cells and YAC-1 cells. NK cells, isolated from mouse spleen, were co-cultured for 2 h with YAC−1 target cells in the presence of the K_ATP channel opener (30 µM pinacidil), a K_ATP channel blocker (1 µM glibenclamide). Phorbol 12-myristate 13-acetate (PMA; 80 nM) and ionomycin (1.3 µM) was used as a positive control. The no drug group contained solvent only (<0.01% DMSO). (B) Summary data of all experiments using the K_ATP channel drugs. (C) Spleen NK cells isolated from WT or NK-specific Kcnj8 KO mice were co-cultured with or without the target cell lines, RMA, RMA-S and RMA-RAE1y. Flow cytometry was performed to detect the percentage of CD107a+ NK cells (as a surrogate marker of degranulation). The data are from 3 separate experiments.

Figure 7

NK cells reject tumor cells equally in vivo in WT and NK- specified Kcnj8 KO mice. Mice were injected I.P. with a mixture of RMA-S, RMA and RMA-RAE1y cells, pre-stained with fluorescent markers for later visualization and identification. The RMA-Rae1γ cells are preferential NK cell targets due to high expression of MHC class I molecules. After 48 h, cells were isolated from the peritoneum and subjected to flow cytometry to quantify the relative amounts of target cells. (A) Cells from the peritoneal fluid of four WT and four NK- specified Kcnj8 KO mice were downsampled to equal cell counts of viable cells per group (40.000 cells per group) in FlowJo and concatenated to a single file prior to unbiased dimensionality reduction approach, visualized as tSNE plots. The clusters were determined by protein expression levels of forward and side scatter (FSC & SSC), H2Kb, CD3, CFSE, FarRed, NK1.1 and CD69. RMA-S cells were not detected. The RMA (red) and RMA-RAE1y (green) cell populations each comprised about 5% of the number of cells in both the WT and KO mice. (B) Depicted is the percentage of RMA and RMA-RAE1y cells found in peritoneal fluid of WT and KO mice. Experiment was repeated twice with similar results.

Figure 8

Openers or blockers of K_ATP channels do not affect store-operated Ca²⁺ entry (SOCE) in mouse NK cells. Cytosolic Ca²⁺ of isolated splenic NK cells was recorded with Fluo4/AM. (A) Depicted are representative F/F0 traces in different wells of the same experiment with NK cells pretreated for 15 min with pinacidil (100 µM), TRAM-34 (10 µM), PNU-37883A (10 µM) or with no drug (<0.1% DMSO). Store depletion was accomplished by 1 μM thapsigargin and SOCE was initiated by adding 200 µM CaCl₂ to the external solution. (B) Summary data of experiments performed on different days (n=3 mice). p=0.84 with 1W-ANOVA.

Figure 9

Data of a bulk RNA seq experiment performed with mouse splenic cells, NK cells isolated by cell sorting by first using NK markers NK1.1 and NKp46, then NK cell maturity markers CD27 and CD11b (n=2 each of WT and tamoxifen-induced NK cell-specific Kcnj8 KO mice). Normalized reads and differentially expressed genes were calculated using the DEseq2 package. Normalized gene expression was averaged within populations. (A) Principal component analysis demonstrates segregation of the three populations 2, 1 and 0 (respectively CD27⁺/CD11b^-, CD27⁺/CD11b⁺ and CD27^-/CD11b⁺). (B) Mapping of RNA seq reads against the mouse mm10 genome was performed with IGV (Integrative Genomics Viewer). Note that the Kcnj8 gene at the bottom of the panel is oriented in the reverse direction (3’ to 5’). (C) Normalized Cd27 (CD27), Itgam (CD11b), and Kcnj8 (Kir6.1) reads in the three different populations for wild-type (WT) and mice with NK cell-specific Kcnj8 deficiency (p<0.05 with the Mann-Whitney Rank Sum Test). Note that, although Kcnj8 reads remain in the KO mice, they are from exons 1 and 3 as shown in panel (B).

Figure 10

NK cell maturation in bone marrow and spleen of NK-specific Kcnj8 KO and WT mice. (A) NK cells, isolated from bone marrow, were subjected to flow cytometry. Data of WT and constitutive NK-cell specific Kcnj8 deficient mice are contrasted. A tSNE plot of differentially expressed proteins identified six populations of NK cells with differential protein expression (population 0 – population 5). NK maturation markers in each population is shown. WT is shown on the top and KO is the bottom. The frequency of each of the NK cell populations are compared between WT and KO mice. *p<0.05; **p<0.001 using a Student’s t-test. (B) Isolated spleen NK cells from WT mice, or mice with tamoxifen-induced NK-cell specific Kcnj8-deficiency, were subjected to flow cytometry. NK cells were selected based on expression of NK1.1 and NKp46. Cells were further selected based on expression of CD27 and CD11b. (C) Summary data from 3-4 separate mice in each group is depicted as bar graphs. p<0.05 with 2W-ANOVA, followed by a Dunnett’s t-test.