MF, LLL, and LJS designed research. MF performed research. LLL and LJS contributed reagents and analytic tools. MF, LLL, and LJS analyzed data. MF and LJS wrote the paper. MF, LLL, and LJS approved the final manuscript.
LLL and the University of California (San Francisco, CA) have licensed intellectual property rights relating to NKG2D for commercial applications. The authors declare that no other competing interests exist.
Ectromelia virus (ECTV) is an orthopoxvirus (OPV) that causes mousepox, the murine equivalent of human smallpox. C57BL/6 (B6) mice are naturally resistant to mousepox due to the concerted action of innate and adaptive immune responses. Previous studies have shown that natural killer (NK) cells are a component of innate immunity that is essential for the B6 mice resistance to mousepox. However, the mechanism of NK cell–mediated resistance to OPV disease remains undefined. Here we show that B6 mice resistance to mousepox requires the direct cytolytic function of NK cells, as well as their ability to boost the T cell response. Furthermore, we show that the activating receptor NKG2D is required for optimal NK cell–mediated resistance to disease and lethality. Together, our results have important implication towards the understanding of natural resistance to pathogenic viral infections.
Ectromelia virus (ECTV) causes mousepox, a murine disease that is the equivalent of human smallpox. ECTV normally penetrates through the periphery but rapidly spreads through the lymphatic system to vital organs. In mousepox-sensitive strains of mice, ECTV infection culminates with either rapid death or overt symptoms of mousepox due to very high loads that the virus reaches in vital organs, particularly the liver. However, some strains of mice such as C57BL/6 (B6) and 129 also become infected with ECTV but naturally resist mousepox by controlling the virus loads in vital organs and clearing the virus without clinical symptoms of disease. Natural killer (NK) cells are cells of the innate immune system previously shown to play an important role in natural resistance to mousepox. However, how NK cells protect from this disease is still unknown. In this paper we show that NK cells directly contribute to antiviral defenses by curbing virus dissemination to vital organs and also indirectly by augmenting the antiviral T cell response. We also demonstrate that optimal protection requires the activating NK cell receptor NKG2D which facilitates killing of ECTV-infected cells. Our work has important implications for the understanding of natural resistance to viral disease.
Ectromelia virus (ECTV), the causative agent of mousepox, is an orthopoxvirus (OPV) with host specificity for the mouse. It is genetically very similar to vaccinia virus (VACV), the human pathogen variola virus (the agent of smallpox), and monkeypox virus [
Although all mouse strains can be infected with ECTV, the outcome of the infection following footpad inoculation varies. Some sensitive strains, such as DBA/2, A/J, and BALB/c, develop mousepox and suffer high mortality during the first 2 wk post-infection, whereas other strains, such as C57BL/6 (B6), clear the infection without visible symptoms of systemic disease [
Natural killer (NK) cells are innate effector cells serving as a first line of defense against certain viral infections and tumors [
The activation of NK cells is regulated by a balance of signals transduced by activating and inhibitory receptors [
In this study, we show that NK cells directly contribute to antiviral defenses by curbing virus dissemination to central organs and also indirectly by augmenting the antiviral T cell response. We also demonstrate that the activating receptor NKG2D is involved in the NK cell–mediated resistance to mousepox, whereas NK1.1 is not.
B6 mice are normally resistant to mousepox ([
NK Cells Are Required for Resistance to Mousepox during Early ECTV Infection
Intact or NK cell–depleted (treated with anti-NK1.1 mAb PK136) B6 mice were infected with 3,000 pfu ECTV. (A) 7 d PI, the absolute numbers of live lymphocytes in the spleen were determined by trypan blue exclusion. Data correspond to the average ± SD of pooled spleens of three mice from five independent experiments.
(B) 7 d PI, virus titers in spleen and liver were determined by plaque assay. Data correspond to the average ± SD of six individual mice from two independent experiments.
(C) B6 mice were infected with ECTV in the footpad. The NK cell response, as determined by the percentage of NK cells expressing intracellular IFN-γ, at the indicated times PI was determined in the indicated organs. Data correspond to pools of three mice and are representative of three similar experiments.
(D) Representative flow cytometric analysis of D-LN from mice infected for 2 d with ECTV and from control uninfected mice. Upper panel: Dot plots indicating the proportion of NK cells (NK1.1+, CD3ɛ−). Lower panel: GzB and IFN-γ production by gated NK cells (NK1.1+/CD3ɛ− gate of the upper panels). Data correspond to pools of three mice and are representative of at least five experiments. >98% of cells stained with control Ig were in the lower left quadrant of the dot plots (not shown).
(E) B6 mice were infected with ECTV, and NK cell (NK1.1+, CD3ɛ−) proliferation was determined at different times PI in the indicated organs by using a 3-h BrdU incorporation assay. Data are representative of three independent experiments.
(F) Flow cytometric analysis of D-LN from mice infected for 2 d with ECTV and from control uninfected mice. Upper panels: Plots are gated on NK cells (NK1.1+, CD3ɛ−). Lower panels: Gated on R1, R2, and R3 populations from the upper-right (infected) plot. Data are representative of three independent experiments.
(G) Intact B6 mice and B6 mice depleted of NK cells (treated with anti-NK1.1) or T cells (treated with anti-CD4 and anti-CD8) were infected with ECTV and virus titers in spleen and liver were determined on day 3 PI. Data are the average ± SD of six individual mice from two experiments.
(H) Wild-type and IFN-γ-deficient B6 mice were infected with ECTV in the footpad, and virus titers were determined 3 d PI. Data are the average ± SD of three individual mice and is representative of two individual experiments.
(I) Wild-type and Pf-deficient B6 mice were infected with ECTV in the footpad and virus titers were determined 3 d PI. Data are the average ± SD of three individual mice and are representative of two individual experiments.
We next evaluated the NK cell response at different times PI by flow cytometry. NK cell production of IFN-γ and granzyme B (GzB) in the draining lymph nodes (D-LN) was already induced 24 h PI, reaching the peak at 48 h PI when ∼20% of the NK cells produced IFN-γ and 43% produced GzB (
To distinguish whether the increase in NK cell number in D-LN resulted from recruitment and/or proliferation, we inoculated mice at different stages of infection with BrdU IP and sacrificed the mice 3 h later to determine BrdU incorporation by flow cytometric analysis. Very few NK cells incorporated BrdU in the D-LN during the first 3 d PI, suggesting increased migration to the D-LN at early times PI. In fact, on day 3 and 5 PI, the BrdU incorporation in NK cells in spleen and liver was higher than in the D-LN (
A recent report by Hayakawa and Smyth revealed that peripheral NK1.1+ NK cells can be divided into three subsets according to their expression of CD11b and CD27, which they designated R1 (CD27+, CD11b−), R2 (CD27+, CD11b+), and R3 (CD27−, CD11b+), that represent NK cells at distinct developmental stages [
We hypothesized that the observed early recruitment to and activation of NK cells in the D-LN may contribute to the prevention of virus dissemination to central organs. Thus, we depleted NK cells in B6 mice with anti-NK1.1 mAb (PK136) and determined viral titers in spleen and liver on day 3 PI. To rule out a role for T cells at this stage we also depleted both CD4+ and CD8+ T cells by using a combination of anti-CD4 (GK1.5) and anti-CD8 (2.43) mAbs. Depletion of NK cells resulted in >103-fold increase in viral loads in the spleen, while a much lower increase in virus titers was observed in mice depleted of T cells. The slight increase in virus titers in the T cell–depleted mice was not statistically significant (
NK cells can control viral infections by producing antiviral cytokines, such as IFN-γ, or by perforin (Pf)-mediated killing of infected cells [
Our previous studies and those of others showed that an optimal CD8+ T cell response is required for resistance to mousepox [
(A) Intact or NK cell–depleted B6 mice were infected with 3,000 pfu ECTV. T cell proliferation on day 5 PI was determined in the D-LN by using a 3-h BrdU incorporation assay. Upper panel: gated on CD8+ T cells; lower panel: gated on CD4+ T cells.
(B) Intact or NK cell–depleted B6 mice were infected with ECTV. Production of IFN-γ and GzB by CD8+ T cells in spleen on day 7 PI was determined. Plots are gated on CD8+ T cells. Data are representative of three independent experiments.
To gain insight into the mechanism whereby NK cells become activated during ECTV infection, we explored whether various known NK cell–activating receptors are involved in resistance to mousepox. We first focused on NK1.1 and on the activating receptors of the Ly49 family. We focused on these receptors because NK1.1 is not encoded in mousepox-susceptible DBA/2 and BALB/c mice. Furthermore, it has been speculated that NK1.1 might be
NKG2D Is Involved in NK Cell–Mediated Resistance to Mousepox
The results above suggested, but did not formally prove that Ly49H, Ly49D, and/or NK1.1 are not required for the resistance of B6 mice to mousepox, because it remained possible that Ly49H, Ly49D, and/or NK1.1 play a role in B6 resistance but that other activating receptors substitute their function in 129/Sve mice. Although it is not possible to definitively rule out the participation Ly49H or Ly49D in B6 resistance with currently available reagents, we took advantage of FcɛRIγ-deficient B6 mice [
We addressed whether the activating receptor NKG2D might be involved in mousepox resistance. NKG2D is expressed by NK cells and some T cells, including γδ-TcR+ T cells and activated CD8+ T cells. Although not polymorphic, this activating receptor is expressed at somewhat lower levels on activated NK cells of mousepox-susceptible NOD and BALB/c mice than in B6 mice ([
In mice, NKG2D signals through either DAP10 or DAP12 adapter proteins, whereas Ly49H and D signal through DAP12 [
To get further insights into the role of NKG2D and its adapters in resistance to mousepox, we determined cell numbers and virus titers 7 d PI in NKG2D-blocked as well as in DAP12- and DAP10-deficient mice. Results showed that NKG2D blockade and DAP12- or DAP10-deficiency resulted in significantly decreased splenic lymphocytosis as compared with wild type, untreated B6 mice (
(A) The indicated mice were infected with 3,000 pfu ECTV and the absolute number of lymphocytes in their spleens was determined on day 7 PI by trypan blue exclusion. An uninfected control is also shown. Data are the average ± SD of three pooled mice per group from at least three individual experiments.
(B) Virus titers in spleen of the indicated mice on day 7 PI. Data are the average ± SD of three pooled mice per group from at least three individual experiments.
NKG2D is expressed by NK cells [
(A) B6 mice were infected with 3,000 pfu ECTV. On day 5 PI, mice were pulsed with BrdU for 3 h and their spleens analyzed by flow cytometry. Plots are gated on CD3ɛ− NK1.1+ cells. Data correspond to pools of three mice from three individual experiments.
(B) Increased viral titers of infected mice with NKG2D blockade in vivo. Intact B6 mice, B6 mice with NKG2D blockade, B6 mice depleted of NK cells, B6 mice depleted of T cells (treated with anti-CD4 + anti-CD8 mAbs), or depleted of T cells and with NKG2D blockade were infected with 3,000 pfu ECTV and the viral titers in spleen were determined on day 3 PI. Data are the average ± SD of three individual mice and are representative of two similar experiments.
(C) The indicated mice were infected with 3,000 pfu ECTV, and the NK responses in the D-LN were determined at 2 d PI. Upper panel: Dot plot showing the proportion of NK cells (DX5+, CD3e−) in the D-LN of infected and control uninfected mice. Lower panel: GzB and IFN-γ production by NK cells. Cells correspond to the DX5+, CD3ɛ− gate of the upper panels. Data correspond to pools of three mice and are representative of three similar experiments.
(D) NK cells were purified from spleens (5 d PI) of ECTV-infected intact or NKG2D-blocked mice and stained as indicated. Filled histogram: isotype-matched control Ig; gray line: NKG2D-blocked mice; black line: intact mice.
(E) NK cells were purified from spleens of intact or NKG2D-blocked mice as indicated and were used as effectors in a 4-h 51Cr release assay against the indicated targets. Data correspond to pools of three mice and are representative of three experiments.
(F) As in (D), but the NK cells were purified from untreated mice and a neutralizing anti-NKG2D mAb was added to the in vitro cytotoxicity assays, as indicated.
The data above showed that NK cells are the main population-controlling early virus dissemination to visceral organs, and that absence of NKG2D signaling results in increased susceptibility to mousepox. However, NKG2D is not only expressed by NK cells but also by some T cells. We hypothesized that if NKG2D has a role in NK cell–mediated resistance to mousepox, NKG2D blockade should result in increased early virus dissemination to organs independent of T cells. To test this hypothesis, we measured early viral titers in the spleens (day 3 PI) of mice treated with the anti-NKG2D mAb CX5. To rule out a role of NKG2D on T cells at this time, we also included groups of mice that were T cell–depleted or NKG2D-blocked and T cell–depleted. As shown in
We next tested whether NKG2D blockade and DAP10- or DAP12-deficiency affected the recruitment of NK cells to D-LN (
Because NK cell recruitment and activation in the D-LN was not diminished by NKG2D blockade or DAP12 and DAP10 deficiency (
NKG2D-mediated killing requires the recognition of ligands on the surface of target cells. The ligands of NKG2D are host cell–encoded MHC class I–like proteins that are expressed by tumors and stressed cells and also following infection of cells with some viruses. Identified cellular ligands for NKG2D in mice include H60, MULT1, and Rae-1 [
(A) MEFs were infected with 0.5 pfu ECTV 189898-p7.5-EGFP for 18 h. Cells were analyzed for staining with the indicated reagents after gating for EGFP− cells (uninfected) and EGFP+ cells (infected). Data correspond to one typical experiment from three similar experiments. Shaded area, infected cells stained with isotype-matched control Ig or secondary Ab alone; black line, infected cells stained with the indicated reagent; gray line, non-infected cells stained with the indicated reagent.
(B) qRT-PCR was performed as described. Data were normalized to the amount of β-actin mRNA. Data are representative of two similar experiments.
In this study, we confirmed previous reports demonstrating that NK cells are required for natural resistance to mousepox [
To determine whether and when NK cells are required for resistance to mousepox we depleted mice of NK cells with either anti-asialo GM1 or anti-NK1.1 Abs at different times PI. Although separately these two approaches have caveats, together they provide conclusive evidence that the presence of NK cells during the first 4 d PI, but not beyond day 5 PI, is essential for resistance to mousepox. First, even though anti-NK1.1 Ab depletes NKT cells, the loss of mousepox resistance upon Ab treatment cannot be due to the elimination of NKT cells because anti-asialo GM1 Ab does not eliminate this population [
We also followed the kinetics of NK proliferation and activation in spleen, liver, and D-LN of ECTV-infected mice. Interestingly, although we detected some proliferation, we did not find activated NK cells in liver and spleen on day 3 PI. This was despite the fact that NK cell–mediated control of virus loads in spleen and liver already occurred at this time PI. In fact, the peak of NK cell activation in spleen and liver took place on day 5 PI. On the other hand, the proportion of total NK cells, as well as the proportion of activated NK cells in the D-LN, peaked as early as day 2 PI, and these parameters were still substantially elevated on day 3 PI, notwithstanding that the proliferation of NK cells at these times was not yet detectable in D-LN. In addition, we found that the increase in NK cell numbers in the D-LN on day 2 PI was mostly due to an increase in mature NK cells and that these mature NK cells were preferentially activated. Together, these data suggest that the prompt NK cell response in the D-LN is responsible for the early control of virus loads in central organs and that this response is mostly due to the recruitment and activation of mature NK cells rather than their expansion. To spread to central organs from the footpad, ECTV must pass through the D-LN [
In addition to NK cells, strong adaptive T cell responses are essential for resistance to mousepox. Our experiments here show that NK cells have a direct role in controlling ECTV because they reduced virus loads on day 3 PI when the T cell response is still undetectable (data not shown and [
Previous work by Delano and Brownstein showed that
Because mouse NKG2D signals through either the adapter protein DAP12 or DAP10, we further tested the susceptibility of mice lacking one or the other adapter. Of interest, when considering virus titers and spleen cellularity, the susceptibility of these two strains of mice was intermediate between intact and NKG2D mAb–blocked B6 mice. This indicates that for resistance to mousepox, the two adapters likely have overlapping, but not completely redundant, effects. Still, DAP12 seems to be the preferred adapter because DAP12-deficient mice were more susceptible to mousepox than DAP10-deficient mice, although it is possible that other DAP12-associated receptors might contribute to resistance to mousepox. Our results predict that DAP10 + DAP12 double-deficient mice would be highly susceptible to mousepox. Unfortunately, these mice are not yet available.
NKG2D is expressed by most NK cells but also by activated CD8+ T cells. In fact, we have observed that in ECTV-infected B6 mice a large proportion of virus-specific CD8+ T cells express NKG2D beginning 7 d PI (unpublished data). Thus, the experiments reported here do not rule out the additional contribution of NKG2D in the T cell–mediated resistance to mousepox, a very interesting possibility that we are currently investigating. Nevertheless, our experiments clearly implicate NKG2D in the NK cell–mediated response to ECTV because in vivo NKG2D blockade resulted in enhanced virus loads in central organs before the onset of the T cell response and because in vivo NKG2D blockade reduced the cytotoxicity of ECTV-activated NK cells. Still, several lines of evidence indicate that NKG2D is not the only activating receptor involved in the anti-ECTV NK cell–mediated response and that its most likely role is as a co-stimulator that facilitates cytotoxicity rather than being required for the initial NK cell activation: (1) NK cell depletion was much more effective than NKG2D blockade at rendering B6 mice susceptible to mousepox; (2) NKG2D-blockade did not affect NK cell proliferation (data not shown), IFN-γ and GzB production by NK cells, or recruitment of NK cells into the D-LN of ECTV-infected mice; (3) In vivo and in vitro NKG2D blockade significantly decreased, but did not abrogate, the cytotoxicity of ECTV-activated NK cells. Ongoing studies in our laboratory are aimed at identifying other activating receptor(s) and signaling pathway(s) required for NK cell–mediated resistance to mousepox.
In summary, our work demonstrates that NK cells contribute to the natural resistance of B6 mice to mousepox by using direct effector functions (most likely the killing of infected cells) to curb virus dissemination and by supporting a strong adaptive T cell response. Moreover, our data suggest that the activating receptor NKG2D, but not NK1.1 or Ly49 family members, has a role in this NK cell–mediated resistance to mousepox by promoting optimal NK cell–mediated killing. Thus, our data provide substantial insights into the mechanisms of natural resistance to ECTV and possibly other OPV infections. Together, our work furthers our understanding of host-pathogen interactions and the mechanisms whereby NK cells protect from viral disease.
YAC-1 and CHO-K1 cell lines were obtained from Dr. Kerry Campbell (Fox Chase Cancer Center, Philadelphia, Pennsylvania), and A9, MC57G, and BSC-1 cells were obtained from the ATCC. MEF cells were made from day 11 to 13 embryos from B6 mice. As standard tissue culture media, we used RPMI-10 that consisted of RPMI-1640 medium (Invitrogen) supplemented with 10% fetal calf serum (Sigma), 100 IU/ml penicillin and 100 μg/ml streptomycin (Invitrogen), 10 mM Hepes buffer (Invitrogen), and 0.05 mM 2-mercaptoethanol (Sigma). MEF were grown in DMEM medium containing 15% fetal calf serum. When indicated, RPMI 2.5 (as above but with 2.5% FCS) was used. When required, 10 U/ml interleukin 2 (IL2) was added to RPMI 10 (RPMI 10-IL2). All cells were grown at 37 °C and 5% CO2.
The production of ECTV stocks for infection of mice and the determination of titers in stocks and organs were done as described previously [
For ECTV infection of cells, 3–5 × 105 cells were plated in 6-well plates and cultured overnight to allow cells to adhere. The cells were then infected with 0.5 pfu ECTV/cell for 18 h, collected, washed, stained, and analyzed for surface expression of various markers. For ECTV infection of peritoneal cells, the mice were euthanized by halothane inhalation and injected i.p. with PBS, the abdomen massaged gently, and the peritoneal cells were collected by aspiration and washed.
Depletion of NK cells was performed by i.p. inoculation of 200 μg anti-NK1.1 mAb PK136 or 20 μl anti-asialo GM1 antisera (Wako), as indicated. Antibody treatment was done 1 d before or on the indicated days after virus infection. For depletion of T cells, mice were injected i.p. with 200 μg anti-CD4 mAb GK1.5 and 200 μg anti-CD8 mAb 2.43 1 d before infection. For NKG2D blockade mice were inoculated with purified 200 μg CX5 Ab 1 d before and 2 d PI.
Mice were exsanguinated from the orbital cavity to decrease the amount of blood in the liver. The liver was removed and passed through a cell strainer (BD Falcon) to obtain a single cell suspension. The cells were resuspended in 35% Percoll solution (in PBS) containing 100 U/ml heparin and centrifuged at 830 ×
At the indicated time post-infection (PI), mice were injected with 2 mg BrdU i.p. 3 h later, spleens and lymph nodes (LNs) were removed and made into single cell suspensions. The liver lymphocytes were obtained as described above. The cells were then stained for cell surface molecules, fixed, and permeabilized using the Cytofix/Cytoperm kit (BD Pharmingen) according to the manufacturer's instructions, incubated with DNase at 37 °C for 1 h, and subsequently stained with FITC-conjugated anti-BrdU mAb (eBiosciences).
Determination of cytokine production by intracellular staining was done as described previously [
NK cells were purified from spleens using anti-CD49b-conjugated microbeads and a LS column (Miltenyi Biotec) according to the manufacturer's instructions and were stained for flow cytometric analysis with PE-conjugated anti-NKG2D mAb CX5. NK cells were resuspended in RPMI 10, and serially diluted in round-bottom 96-well plates in triplicate in 100 μl/well. The indicated target cells were prepared by incubation with 200 μl 51Cr (0.1 mCi) in 100 μl of FCS for 2 h. Cells were thoroughly washed, resuspended in RPMI 10, and 50 μl (5×103 targets) were added to the wells containing effector cells. The plates were incubated at 37 °C for 4 h. Where indicated, 20 μg/ml of a neutralizing anti-NKG2D mAb (clone 191004, R&D Systems) was added at the initiation of cytotoxicity assays. 50 μl supernatants were transferred to white 96-well plates containing 75-μl Microscint-40 scintillation fluid (PerkinElmer). Controls included wells with target cells alone for spontaneous release and wells with target cells and 1% Triton-X for maximal release. Radioactivity was measured by using a Packard Topcount instrument (PerkinElmer). Specific lysis was determined by using the formula [(experimental release − spontaneous release)/(full release − spontaneous release)]×100. Three animals were used per experimental group.
Primers and probes for Rae-1 were purchased from Applied Biosystems. The amplicon was 95 bp and the sequence of the probe was GGAAAAGCCAAGATCAACCTCAAGG. The primers and probe for β-actin were synthesized at the DNA Synthesis Facility in the Fox Chase Cancer Center. The primers and probe used for β-actin were: sense, 5′-CACCGAGGCCCCCCT-3′; anti-sense, 5′-CAGCCTGGATGGCTACGTACA-3′, and the probe was 5′-6-FAM-AACCCTAAGGCCAACCGTGAAAAGATGA-BHQ1–3′. Total RNA extracted from infected cell lines and LNs of infected mice was treated with Dnase I, and the first-strand cDNA was synthesized by using random primers. qRT-PCR was carried out by using the ABI 7500 (Applied Biosystems). The cycling conditions for real-time PCR were: 50 °C for 10 min, followed by 45 cycles of 95 °C for 30 s, and 60 °C for 2 min. Data were analyzed by using the Sequence Detection v1.2 Analysis Software (Applied Biosystems).
We used a two-tailed
We thank FCCC Tissue Culture, Quantitative Real Time PCR, Flow Cytometry, Hybridoma, Biostatistics and Bioinformatics, Histopathology, and Laboratory Animal Facilities. We thank Dr. Andres Klein-Szanto for histopathology analysis, Drs. David Hesslein, Joe Sun, and Mark Orr for assistance with the mutant mice and comments on the manuscript, Drs. Kerry Campbell and Glenn Rall for comments on the manuscript, and Holly Gillin for secretarial assistance.