Regulatory T cells control NK cells in an insulitic lesion by depriving them of IL-2

Abstract

Regulatory T (T reg) cells control progression to autoimmune diabetes in the BDC2.5/NOD mouse model by reining in natural killer (NK) cells that infiltrate the pancreatic islets, inhibiting both their proliferation and production of diabetogenic interferon-γ. In this study, we have explored the molecular mechanisms underlying this NK-T reg cell axis, following leads from a kinetic exploration of gene expression changes early after punctual perturbation of T reg cells in BDC2.5/NOD mice. Results from gene signature analyses, quantification of STAT5 phosphorylation levels, cytokine neutralization experiments, cytokine supplementation studies, and evaluations of intracellular cytokine levels collectively argue for a scenario in which T reg cells regulate NK cell functions by controlling the bioavailability of limiting amounts of IL-2 in the islets, generated mainly by infiltrating CD4(+) T cells. This scenario represents a previously unappreciated intertwining of the innate and adaptive immune systems: CD4(+) T cells priming NK cells to provoke a destructive T effector cell response. Our findings highlight the need to consider potential effects on NK cells when designing therapeutic strategies based on manipulation of IL-2 levels or targets.

Figures

Figure 1.

Gene expression changes in pancreatic NK cells soon after T reg cell perturbation. NK cells were sorted from the pancreatic infiltrate 24 h after DT injection into BDC2.5/NOD mice with or without DTR expressed in T reg cells, and were profiled by microarray. (A) Transcript changes. Differential gene-expression (FC) values for DTR+ (n = 3) and DTR− mice (n = 6) (y axis) versus their two-class mean (x axis). Induced (>2-fold) genes are highlighted in red and repressed (>2-fold) genes are highlighted in blue. (B and C) Activation features. FC/FC plots of the same NK cell data as in A versus NK cells responding to different activation stimuli. Y axis, FC values for DTR+ versus DTR− mice; x axis, FC values for splenic NK cells cultured with (n = 3) or without (n = 2) IL-12 + IL-18 (see Materials and methods for details; B) or splenic NK cells from C57/BL6 mice 24 h after being infected with MCMV or not (n = 3; C). (D) Time course of transcript changes. Carried out as in A, except additional time points at 8 and 15 h were examined. FC/FC plots comparing gene expression differentials for pancreatic NK cells from DTR+ and DTR− BDC2.5/NOD mice (y axis) versus splenic NK cells from the same animals (x axis). Multiple replicates of cellular populations were collected (usually n = 3–6) and averaged. Highlighting in A–D represents the same set of genes.

Figure 2.

The role of TGF-β signaling. (A, left) Representative flow cytometry profiles of IFN-γ expression in NOD-derived splenic NK cells cultured with (red) or without (gray) IL-12 + IL-18 in the absence of TGF-β, or with both cytokines and TGF-β (blue). (right) Summary data from three independent experiments. Mean ± SD. The p-value was calculated using the two-tailed unpaired Students’ t test. (B) Reciprocal transcript changes promoted by the loss of TGF-β and T reg cells. FC/FC plot comparing cytokine-activated, cultured, splenic NK cells ± TGF-β (x axis, n = 3) versus pancreatic NK cells from BDC2.5/NOD mice ± T reg ablation (y axis, same data as in Fig. 1 A). Dashed red line, linear regression. (C) FC/FC plot comparing transcriptional profiles of cytokine-activated, cultured, splenic NK cells ± TGF-β (y axis, as in Fig. 2 B) versus NK cells ± cytokine-activation (x axis, as in Fig. 1 B). Activation-independent TGF-β–induced genes were highlighted in pink and were superimposed (D in pink) on a volcano plot (p-value vs. fold change) of NK cell transcripts from BDC2.5/NOD mice depleted or not of T reg cells (same data as in Fig. 1). The number of signature genes up-regulated (right) or down-regulated (left) 24 h after T reg cell perturbation are indicated. P-value from the χ2 test. (E and F) Summary flow cytometry data for BDC2.5/NOD mice treated for 24 h with anti–TGF-β versus either an isotype-control mAb or PBS (combined). Percentage of NK cells of lymphocytes (E) and percentage of IFN-γ+ NK cells (F) for two or three independent experiments. Mean ± SD. Dotted line (mean) and gray shading (SD) represents values achieved after T reg cell ablation (a composite of multiple independent experiments).

Figure 3.

An IL-2 footprint is induced by T reg ablation. (A) IL-2–induced (orange) and –repressed (blue) gene transcripts (Marzec et al., 2008) are superimposed on volcano plots (p-value versus FC) of pancreatic NK cell transcripts up-regulated (to the right) or down-regulated (to the left) by perturbation of T reg cells (data from Fig. 1) at 8 h (top) and 24 h (bottom). P-values calculated using the χ2 test. (B) Representative flow cytometry plots for NK cells isolated from the pancreas of BDC2.5/NOD mice 24 h after T reg ablation or not. (C) Summary data for three to four independent experiments with mean ± SD. (D, left) Representative flow cytometry plots for pancreas-infiltrating NK cells after T reg cell ablation; (right) summary for the percentage of CD25-expressing versus CD25-negative NK cells simultaneously expressing IFN-γ from at least three independent experiments.

Figure 4.

Neutralization of IL-2 prevents the activation of pancreatic NK cells in response to T reg cell ablation. (A–D) The pancreatic infiltrate from BDC2.5/NOD mice (DTR+ or DTR− control littermates) was analyzed 24 h after DT injection ± anti–IL-2 mAb JES6-1 (or isotype control) co-injection. (left) Representative flow cytometry plots. (right) Summary data for fraction and numbers with mean ± SD from three to four independent experiments. (A) NK cells (NKp46+CD3−CD19−) and (B) Foxp3+ T reg cells (Foxp3+CD4+CD3+CD19−) in the pancreatic infiltrate (FSC/SSC lymphocyte gated). (C) IFN-γ–producing NK cells among total NK cells in the pancreatic infiltrate. (D) CD25 expression on pancreatic NK cells.

Figure 5.

Supplementation with IL-2 induces early IFN-γ production and eventual accumulation of NK cells in the pancreatic lesion, as well as clinical diabetes. Pancreatic infiltrate from BDC2.5/NOD mice was analyzed 24 h after treatment with control PBS, IL-2–S4B6 complexes, or mutant IL-2 analogue Super-2, or 4 d after injection of IL-2–S4B6 complexes. (A, left) Representative flow cytometry data. (middle) Summary data for fraction of NK cells in the FSC/SSC lymphocyte gate. (right) Cell number summary data. (B) Analogous data for IFN-γ–producing NK cells. Mean ± SD, at least three independent experiments for both panels. (C) Summary diabetes data for 2 cohorts after 3 consecutive treatments with IL-2–S4B6 complexes (n = 14 IL-2–S4B6 complex; n = 9 PBS).

Figure 6.

IL-2 production in the pancreas before and after T reg ablation. (A) Representative flow cytometry plots for pancreatic infiltrate from BDC2.5/NOD mice stained and analyzed for IL-2 by intracellular flow cytometry. An extended FSC/SSC gate was taken to include both lymphocytes and leukocytes along with a live/dead stain to exclude dead cells. CD45+IL-2+ cells were analyzed for CD3 and CD4 expression. (B) Summary data for pancreatic infiltrate from mice depleted of T reg cells (8 h with DT, DTR+) or not (8 h with DT, DTR−). Mean ± SD, at least three independent experiments. (C) Summary data, as in B, gated on CD4+ T cells.