Adaptive immune features of natural killer cells

Abstract

In an adaptive immune response, naive T cells proliferate during infection and generate long-lived memory cells that undergo secondary expansion after a repeat encounter with the same pathogen. Although natural killer (NK) cells have traditionally been classified as cells of the innate immune system, they share many similarities with cytotoxic T lymphocytes. We use a mouse model of cytomegalovirus infection to show that, like T cells, NK cells bearing the virus-specific Ly49H receptor proliferate 100-fold in the spleen and 1,000-fold in the liver after infection. After a contraction phase, Ly49H-positive NK cells reside in lymphoid and non-lymphoid organs for several months. These self-renewing 'memory' NK cells rapidly degranulate and produce cytokines on reactivation. Adoptive transfer of these NK cells into naive animals followed by viral challenge results in a robust secondary expansion and protective immunity. These findings reveal properties of NK cells that were previously attributed only to cells of the adaptive immune system.

Figure 1. Preferential expansion of wild-type but not DAP12-deficient NK cells during MCMV infection

Mixed bone marrow chimeric mice (1:1 mixture of wild-type (CD45.1⁺) and DAP12-deficient (CD45.2⁺) cells) were infected with MCMV. a, Upper, percentages of wild-type and DAP12-deficient NK cells (gated on CD3⁻, NK1.1⁺). Lower, Ly49H and KLRG1 on wild-type NK cells. b, Upper, BrdU incorporation (day 7 PI) of wild-type (solid lines) and DAP12-deficient (dotted lines) NK cells. Lower, BrdU incorporation by Ly49H⁺ (solid lines) and Ly49H⁻ (dotted lines) wild-type NK cells. c, d Percentages of Ly49H⁺ cells within the wild-type NK cell population following infection with MCMV (c) or MCMV-Δm157 (d). Data are representative of 3 experiments with 3–5 mice per time point.

Figure 2. Robust proliferation of adoptively transferred wild-type NK cells in DAP12-deficient mice following MCMV infection

a, 10⁵ wild-type NK cells (CD45.1⁺) were transferred into DAP12-deficient mice (CD45.2⁺) and infected with MCMV. b, Transferred NK cells (CD45.1⁺) within the total CD3⁻ NK1.1⁺ gated population analyzed for Ly49H, CD69, NKG2D, and intracellular IFN-γ. c, Percentages of transferred CD45.1⁺ Ly49H⁺ NK cells within the total CD3⁻ NK1.1⁺ population after infection. d, CFSE-labeled wild-type NK cells (5×10⁵) were transferred into DAP12-deficient hosts. Ly49H⁺ and Ly49H⁻ NK cells were analyzed after infection. e, Percentages (left graph) and absolute numbers (right graph) of transferred CD45.1⁺ NK cells in DAP12-deficient or wild-type B6 recipients after infection. Error bars display s.e.m. (n = 3–5). Data are representative of 5 experiments.

Figure 3. Expansion and contraction of NK cells in lymphoid and non-lymphoid tissues results in “memory” NK cells

a–b, Following adoptive transfer of 10⁵ (squares) or 10⁴ (circles) wild-type NK cells (CD45.1⁺) into DAP12-deficient mice and MCMV infection, percentages (left) and absolute numbers (right) of Ly49H⁺ NK cells in spleen (a) and liver (b). Error bars display s.e.m. (n = 3–5). Data are representative of 3 experiments. c, Fold expansions of Ly49H⁺ NK cells over 7 days of infection were calculated in B6 mice, in 1:1 and 1:5 wild-type:DAP12-deficient bone marrow chimeric mice, and following adoptive transfer of wild-type NK cells into DAP12-deficient mice. Error bars display s.e.m. from 3 experiments in each group of mice indicated.

Figure 4. Function and phenotype of “memory” NK cells

a, 10⁵ wild-type NK cells (CD45.1⁺) transferred into DAP12-deficient mice (CD45.2⁺) were analyzed 70 days after MCMV infection. Left, percentage of CD45.1⁺ Ly49H⁺ cells within total NK cell population. Right, percentages of CD45.1⁺ NK cells producing IFN-γ after stimulation. b, “Memory” NK cells compared to naïve NK cells from uninfected mice after anti-NK1.1 stimulation. Percentages of Ly49H⁺ NK cells (gated on total NK cells) producing IFN-γ. Right, percentages of Ly49H⁺ NK cells producing IFN-γ (4 mice/group, horizontal bar is mean, p = 0.0009). c, LAMP-1 on “memory” NK cells (day 55 PI) versus naïve NK cells (gated on Ly49H⁺ NK cells) after anti-NK1.1 stimulation. d, Surface markers on “memory” NK cells (day 45 PI) versus naïve NK cells. e, NK cells from day 7 MCMV-infected Yeti mice were transferred into DAP12-deficient Rag2^−/− mice. YFP in “memory” Ly49H⁺ NK cells after 28 days compared to Ly49H⁺ NK cells and T cells in uninfected Yeti mouse.

Figure 5. Secondary expansion and protective immunity in “memory” NK cells

a, 10⁵ wild-type NK cells (CD45.1⁺) transferred to DAP12-deficient mice (CD45.2⁺) were isolated 40–50 days after primary MCMV infection and transferred into DAP12-deficient mice (CD45.2⁺). Following infection of the second host, Ly49H⁺ NK cells were analyzed. b, Percentages of transferred “memory” Ly49H⁺ NK cells in the second host. c, Percentages (left) and absolute number (right) of Ly49H⁺ NK cells within the total NK cell population in second host. Error bars display s.e.m. (n = 3–5). Data are representative of 5 experiments. d, Analysis of CSFE-labeled “memory” and naïve NK cells transferred into DAP12-deficient recipient mice (day 3 and 6 PI). e, Expansion and contraction of “memory” and naïve NK cells (1×10⁴ input) shown as a percentages (left) and absolute number (right) of Ly49H⁺ NK cells within the total NK cell population. Error bars display s.e.m. (n = 3–5). Data are representative of 3 experiments. f, Survival of DAP12-deficient neonatal mice receiving 1×10⁴ or 1×10⁵ naive, or 1×10⁴ “memory” NK cells (or PBS as control) followed by MCMV infection, with or without anti-Ly49H blocking. Data were pooled from 3 experiments.

Figure 5. Secondary expansion and protective immunity in “memory” NK cells

a, 10⁵ wild-type NK cells (CD45.1⁺) transferred to DAP12-deficient mice (CD45.2⁺) were isolated 40–50 days after primary MCMV infection and transferred into DAP12-deficient mice (CD45.2⁺). Following infection of the second host, Ly49H⁺ NK cells were analyzed. b, Percentages of transferred “memory” Ly49H⁺ NK cells in the second host. c, Percentages (left) and absolute number (right) of Ly49H⁺ NK cells within the total NK cell population in second host. Error bars display s.e.m. (n = 3–5). Data are representative of 5 experiments. d, Analysis of CSFE-labeled “memory” and naïve NK cells transferred into DAP12-deficient recipient mice (day 3 and 6 PI). e, Expansion and contraction of “memory” and naïve NK cells (1×10⁴ input) shown as a percentages (left) and absolute number (right) of Ly49H⁺ NK cells within the total NK cell population. Error bars display s.e.m. (n = 3–5). Data are representative of 3 experiments. f, Survival of DAP12-deficient neonatal mice receiving 1×10⁴ or 1×10⁵ naive, or 1×10⁴ “memory” NK cells (or PBS as control) followed by MCMV infection, with or without anti-Ly49H blocking. Data were pooled from 3 experiments.