Sustained Ca2+ signaling, known as store-operated calcium entry (SOCE), occurs downstream of immunoreceptor engagement and is critical for cytotoxic lymphocyte signaling and effector function. CD8+ T cells require sustained Ca2+ signaling for inflammatory cytokine production and the killing of target cells; however, much less is known about its role in NK cells. In this study, we use mice deficient in stromal interacting molecules 1 and 2, which are required for SOCE, to examine the contribution of sustained Ca2+ signaling to murine NK cell function. Surprisingly, we found that, although SOCE is required for NK cell IFN-γ production in an NFAT-dependent manner, NK cell degranulation/cytotoxicity and tumor rejection in vivo remained intact in the absence of sustained Ca2+ signaling. Our data suggest that mouse NK cells use different signaling mechanisms for cytotoxicity compared with other cytotoxic lymphocytes.
The mobilization of Ca2+ from the extracellular environment into the cytoplasm is important for immune cell activation downstream of activating immunoreceptors. Engagement of these activating receptors leads to the phosphorylation of phospholipase C (PLC), which hydrolyzes phosphatidylinositol-4,5-bisphosphate into the second messengers diacylglycerol (DAG) and inositol-1,4,5-triphosphate (IP3) (1, 2). When IP3 binds to IP3 receptors on the endoplasmic reticulum (ER), ER Ca2+ stores are released into the cytoplasm. ER Ca2+ depletion dissociates Ca2+ ions from the ER Ca2+ sensors stromal interaction molecule (STIM) 1 and 2. STIM1/2 oligomerize and activate ORAI1–3 (Ca2+ release-activated channel [CRAC]) on the plasma membrane to facilitate entry of extracellular Ca2+ into the cytoplasm. This process, known as store-operated calcium entry (SOCE), is important for sustaining high levels of Ca2+ in the cytoplasm after immune activation (1, 3).
STIM1/2-deficient T cells cannot mobilize Ca2+ through the CRAC channel following TCR stimulation (4), leading to defective TCR-mediated proliferation, cytokine production, and degranulation. Because of these defects, STIM1/2-deficient T cells have significantly compromised cytotoxicity against tumors (5) and defective control of acute infections, and they possess reduced memory T cell formation and persistence (6). Like CD8+ T cells, the key functions of NK cells are to produce inflammatory cytokines and perform cell-mediated killing. NK cells also initiate Ca2+ signaling following activating receptor stimulation (7). A few human studies have probed the requirement of SOCE in NK cell functions by examining rare patients with STIM1 or ORAI1 mutations (8, 9). These patients had normal frequencies of NK cells, expression of NK cell–defining markers, LFA-1 activation, and granule polarization. Similar to their T cells, the NK cells from these patients exhibited defective cytokine production and cytotoxicity. These data suggest that NK cells also require sustained Ca2+ signaling for their key effector functions, just like their T cell counterparts.
Although it is generally believed that CD8+ T cells and NK cells use similar mechanisms to generate effector function, recent studies have uncovered key differences in the signaling pathways that dictate NK cell effector function compared with T cells (10–12). Surprisingly, we found that Ca2+ signaling in isolation did not promote key NK cell effector functions, yet the activation of the DAG signaling pathway was sufficient to induce degranulation by mouse NK cells. Moreover, although mouse NK cells displayed defective IFN-γ production in the absence of SOCE, they retained the ability to degranulate and kill target cells. These data suggest that, unlike CD8+ T cells or human NK cells, which need sustained Ca2+ signaling for cytotoxicity, primary murine NK cells do not require SOCE to perform cell-mediated cytotoxicity.
Materials and Methods
STIM1flox/flox STIM2flox/flox mice were generated and generously provided by Dr. A. Rao (La Jolla Institute for Allergy and Immunology), Dr. P. Hogan (La Jolla Institute for Allergy and Immunology), and Dr. M. Oh-hora (Kyushu University) (4). These animals were bred to NKp46iCre/WT or Vav-cre mice (13). Age-matched C57BL/6 (wild-type [WT]) control animals or littermate controls were used for all experiments. Mice were sacrificed and analyzed between 8 and 12 wk of age. Mice were housed in pathogen-free conditions and treated in strict compliance with the Institutional Animal Care and Use Committee regulations at the University of Pennsylvania.
Reagents, flow cytometry, Abs, and data analysis
Akt Inhibitor VIII, Gö6983, and U0126 were purchased from Calbiochem (Darmstadt, Germany), Tocris Bioscience (Bristol, U.K.), and Cell Signaling Technology (Danvers, MA), respectively. Cyclosporine A (CsA) and actinomycin D were purchased from Sigma (St. Louis, MO). For flow cytometry, cells were prepared, stained, and analyzed as previously described (14). Abs for NK cell stimulation and phenotyping were purchased from BD Pharmingen (San Diego, CA), BioLegend (San Diego, CA), eBioscience (San Diego, CA), Bio X Cell (West Lebanon, NH), and Molecular Probes, Invitrogen (Carlsbad, CA). Data were analyzed with FlowJo software (Tree Star, Ashland, OR) and Prism (GraphPad, La Jolla, CA). Rat anti–α-tubulin (clone YL1/2; Bio-Rad, Hercules, CA) and 7-amino-4-chloromethylcoumarin (CMAC; Invitrogen) were used for confocal microscopy.
Primary NK cell–stimulation assays
Total splenocytes were plated in 96-well plates in NK cell media and stimulated as previously described (14). For experiments using Ca2+-free media, splenocytes were plated in Ca2+-sufficient media (MEM with 10% FBS, 1% penicillin/streptomycin, 10 mM HEPES, and 1 × 10−5 2-ME) or Ca2+-free media (S-MEM with 10% FBS, 1% penicillin/streptomycin, 1% l-glutamine, 250 μM EGTA, 10 mM HEPES, and 1 × 10−5 2-ME). Following stimulation, splenocytes were gated on live singlet NK cells (CD3ε−NKp46+DX5+) and assessed by flow cytometry for CD107a and intracellular IFN-γ expression.
Immunofluorescence microscopy and microtubule-organizing center measurements
Immunofluorescence microscopy was carried out essentially as described previously (15). YAC-1 cells were labeled with CMAC and mixed 1:2 with NK cells, and conjugation was induced by centrifugation for 5 min. After an additional 10 min, cells were gently resuspended and incubated on poly-l-lysine–coated coverslips for 10 min. Cells were fixed with 3% paraformaldehyde, permeabilized, and labeled with primary Abs and fluorescent secondary Abs. Conjugates were imaged using a 63× Plan Apo 1.4 NA objective on a spinning disk confocal system (UltraView ERS 6; PerkinElmer, Waltham, MA) equipped with an ORCA-ER camera (Hamamatsu Photonics, Bridgewater, NJ) and Volocity software (v6.1.1; PerkinElmer). Microtubule-organizing center (MTOC) to synapse measurements were performed manually, as follows. The border of the YAC-1 cell was determined based on CMAC fluorescence intensity, and the center of the MTOC was defined as the pixel with the brightest intensity in the 488 channel (anti-tubulin). The distance from the MTOC center to the nearest site on the YAC-1 border was measured using Volocity software. Image preparation was performed with ImageJ.
DX5+ splenocytes from STIM1/2 conditional double-knockout (cDKO) or WT mice were purified using magnetic bead sorting (Miltenyi Biotec). Luciferase-expressing YAC-1 cells were incubated at various E:T ratios with DX5+-enriched splenocytes for 6 h at 37°C in the presence of 1000 U/ml human rIL-2, as previously described (16). Luciferase activity was detected using an IVIS Lumina II imager, and the percentage of specific lysis was calculated as follows: [(minimum − test condition)/(minimum − maximum)] × 100. To measure cytotoxicity against RMA-S cells, CFSE-labeled RMA cells and CellTrace Violet–labeled RMA-S cells were mixed at a 1:1 ratio and cocultured with DX5+ splenocytes at various E:T ratios for 18 h at 37°C in the presence of 1000 U/ml human rIL-2. The live RMA-S/RMA ratio was determined by flow cytometry.
RMA and RMA-S tumor models
Short-term and long-term tumor-rejection assays were performed, as previously described (17), in WT and cDKO (NKp46iCre/WT) mice. For NK depletion and IFN-γ neutralization, C57BL/6 mice were injected i.p. with 200 μg of anti-PK136 (NK1.1) and with 500 μg of anti–IFN-γ Ab at 24 and 1 h before tumor challenge, respectively. For short-term assays, the ratio of RMA-S (CFSE+)/RMA (CellTrace Violet+) cells was calculated.
Results and Discussion
Activation of the DAG signaling pathway in the absence of extracellular Ca2+ is sufficient to induce NK cell degranulation
To individually test the contribution of the signaling pathways downstream of PLCγ to NK cell activation, we stimulated primary splenic C57BL/6 (WT) NK cells with the Ca2+ ionophore, ionomycin (Iono), or with the DAG analog, PMA. Although a small fraction of NK cells stimulated with Iono alone expressed IFN-γ, it did not induce any cell surface CD107a expression, a marker for degranulation (Fig. 1A–C). Similarly, PMA alone negligibly upregulated IFN-γ expression on NK cells. However, we unexpectedly found that PMA robustly induced cell surface CD107a expression by NK cells (Fig. 1A–C). Although PMA alone is not expected to elevate cytoplasmic Ca2+ through the activation of SOCE, it was possible that extracellular Ca2+ entering the cell from other channels contributed to the effects of PMA-induced degranulation. However, the proportion of NK cells expressing CD107a in Ca2+-sufficient media (2 mM Ca2+) and Ca2+-free media was similar (Fig. 1D), suggesting that PMA-derived signals can prompt NK cell degranulation in a Ca2+-independent manner.
Extracellular calcium entry is required for IFN-γ expression but not for degranulation and cytotoxicity by NK cells
Next, we stimulated WT splenocytes with plate-bound Abs against NK cell–activating receptors in the presence or absence of extracellular Ca2+. Compared with NK cells stimulated in Ca2+-sufficient media, the proportion of NK cells expressing IFN-γ was severely reduced (80–90% inhibited) when activated in Ca2+-free media (Fig. 2A). In contrast, the degranulation of NK cells activated in Ca2+-free media was relatively preserved, ranging from a 15 to 40% decrease in the proportion of CD107a-expressing NK cells, depending on the activating receptor (Fig. 2A). Still, in general, there was relative preservation of NK cell degranulation compared with IFN-γ production from NK cells stimulated in Ca2+-free conditions.
We next tested the role of extracellular Ca2+ entry in NK cell function using NKp46iCre to delete STIM1 and STIM2 proteins in mature NK cells (STIM1/2 cDKO). Although the initial burst of intracellular Ca2+ is controlled by IP3-mediated release of Ca2+ from ER stores, sustained Ca2+ signaling requires extracellular Ca2+ mobilization through the CRAC channel, a process that is controlled by STIM1 and STIM2 proteins. cDKO mice displayed similar splenic NK cell percentages and numbers, but an increased proportion of the most mature NK cell population (CD11bhiCD27lo), compared with WT control mice (data not shown). Importantly, STIM1/2 cDKO NK cells expressed major NK cell–activating receptors (NK1.1, 2B4, NKG2D, and Ly49D) at a level comparable to WT controls (data not shown). Similar to NK cells stimulated in Ca2+-free conditions, STIM1/2 cDKO NK cells displayed significantly defective IFN-γ production, but intact degranulation, compared with WT NK cells upon stimulation with various activating receptors (Fig. 2B, 2C). This phenotype was also verified using Vav-cre to delete STIM1/2 in all hematopoietic cells (STIM1/2–Vav-cre mice). STIM1/2–Vav-cre mice displayed similar splenic NK cell percentages and numbers and normal NK cell development compared with WT control mice (Supplemental Fig. 1A–C). Importantly, NK cells from STIM1/2–Vav-cre mice showed relatively preserved degranulation and intact cytotoxicity, but severely impaired IFN-γ production (Supplemental Fig. 1D, 1E). Together, these data suggest that NK cell cytotoxic function is intact in the absence of SOCE.
A crucial step in NK cell–mediated cytotoxicity is the polarization of the MTOC to the NK/target cell synapse (18, 19). We assessed MTOC polarization of WT and STIM1/2 cDKO NK cells when mixed with the NK cell–sensitive YAC-1 target cell. Polarization of the MTOC to the target cell synapse was quantitatively similar between WT and STIM1/2 cDKO NK cells (Fig. 2D). The preserved degranulation and MTOC polarization response correlated with intact cytotoxic function of STIM1/2 cDKO NK cells, because STIM1/2 cDKO NK cells killed YAC-1 target cells as effectively as WT NK cells (Fig. 2E).
NK cells from STIM1/2 cDKO mice exhibit cytotoxic function in vivo
We next examined the capacity of STIM1/2 cDKO NK cells to mediate cytotoxicity and clear tumor cells in vivo. We first tested the ability of STIM1/2 cDKO mice to acutely clear the RMA-S (TAP-deficient/MHC-Ilo, NK cell–susceptible) and parental RMA (MHC-I+, NK cell–resistant) tumor cells lines. We injected RMA-S/RMA cells i.v. at a 3:1 ratio into WT, STIM1/2 cDKO, or NK cell–depleted WT animals and measured how this ratio changed 18 h posttumor challenge. As expected, the RMA-S/RMA ratio skewed significantly toward RMA cells in WT mice compared with NK cell–depleted WT mice (Fig. 3A). Although the RMA-S/RMA ratio was significantly higher in STIM1/2 cDKO mice compared with WT mice, the ratio was still significantly lower than in NK cell–depleted WT mice, suggesting that STIM1/2 cDKO NK cells were capable of acutely killing RMA-S cells in vivo, albeit less effectively than WT NK cells (Fig. 3A). The reduced cytotoxicity exhibited by STIM1/2 cDKO NK cells was unlikely to be related to defective IFN-γ production, because IFN-γ neutralization had no effect on skewing the RMA-S/RMA ratio (Supplemental Fig. 2A). Instead, as opposed to YAC-1 target cells, STIM1/2 cDKO NK cells killed RMA-S cells less effectively than WT NK cells, because STIM1/2 cDKO NK cells displayed reduced cytotoxicity when cocultured with a mix of RMA and RMA-S cells in vitro (Supplemental Fig. 2B, 2C).
To measure long-term tumor growth, we injected equivalent numbers of RMA or RMA-S cells s.c. into WT, cDKO, and NK cell–depleted WT mice. Eleven days after tumor challenge, the tumor size of RMA-S cells was smaller than RMA cells, which was normalized by NK cell depletion. Importantly, no difference in tumor size was observed between WT and cDKO mice injected with RMA or RMA-S tumor cells, demonstrating that NK cells lacking STIM1/2 limit tumor growth comparably to WT controls (Fig. 3B). Together with our in vitro results, these data strongly suggest that sustained Ca2+ entry through SOCE is relatively dispensable for NK cell degranulation and cytotoxic function.
DAG-mediated signaling pathways drive NK cell degranulation in a transcription-independent manner
We next interrogated the mechanistic explanation for why sustained mobilization of extracellular Ca2+ is required for NK cell IFN-γ production but not for NK cell degranulation. We reasoned that IFN-γ production might be defective in the absence of SOCE because the induction of IFN-γ, but not degranulation, requires the Ca2+-activated transcription factor NFAT. To test this possibility, we blocked transcription with actinomycin D and, more specifically, abrogated the activation of NFAT using a calcineurin inhibitor (CsA). Indeed, although the proportion of NK cells producing IFN-γ was decreased significantly in NK cells treated with actinomycin D or with CsA, these inhibitors had no effect on degranulation (Fig. 4A).
Because DAG-derived signals were sufficient to drive NK cell degranulation in the absence of extracellular Ca2+, we sought to determine the signaling pathways downstream of DAG that are needed for NK cell degranulation when Ca2+-dependent pathways were eliminated. DAG signals through three major downstream signaling pathways that involve the activation of ERK, AKT, and PKC. To test the relative contribution of these signaling pathways, we pharmacologically inhibited ERK (MEK inhibitor U0126), AKT (AKT1/2 inhibitor), and PKC (PKC inhibitor Gö6983) in splenocytes from WT and STIM1/2 cDKO mice that were stimulated with the anti–NK1.1 Ab. Following ERK inhibition, the proportion of STIM1/2 cDKO NK cells expressing CD107a was higher than that of WT NK cells at each concentration, suggesting that WT NK cells rely on ERK for degranulation more than STIM1/2 cDKO NK cells (Fig. 4B). AKT was equally required by WT and STIM1/2 cDKO NK cells for degranulation (Fig. 4C). In contrast, cDKO NK cells were more sensitive to PKC inhibition by Gö6983 compared with WT NK cells (Fig. 4D). This shows that, in the absence of sustained Ca2+ signaling, NK cells heavily rely on DAG-mediated PKC activation for the degranulation response.
Our data indicate that, in the absence of SOCE, NK cells have the ability to degranulate and perform cell-mediated killing. Although sustained Ca2+ signaling is required for IFN-γ production in an NFAT-dependent manner, it is surprising that NK cell–mediated killing is achievable in the absence of SOCE. This does not disqualify Ca2+ from being important in this process, because STIM1/2 cDKO NK cells undergo ER-mediated Ca2+ release, have mitochondrial retention of Ca2+, and may express other Ca2+ channels, such as transient receptor potential channels (20). It is clear from our data that SOCE is not required for NK cell killing as it is for CD8+ cytotoxic T cells (4, 5).
Previous studies have shown that Ca2+-independent pathways are important for degranulation (17, 21–24). PKC, which is directly activated by DAG, may be critically important for degranulation in the absence of extracellular Ca2+. It has been shown that DAG-dependent signals drive MTOC and granule polarization in T cells (25–27). This is in line with our data showing that STIM1/2 cDKO NK cells polarize their MTOCs normally. Key synaptic and exocytosis proteins depend on PKC for function. PKC phosphorylates the synaptic proteins synaptotagmin, syntaxin 4, VAMP, Munc18, and SNAP25, all of which are critical for the fusing of lysosomes to the plasma membrane and release of its contents (28). Although many of these proteins also have C2 Ca2+-binding domains, PKC isozymes may be the chief factor in the final steps of NK cell degranulation. It is important to note that, in contrast to our mouse NK cell data, NK cells from human patients with ORAI1 and STIM1 mutations are unable to degranulate. Interestingly, however, neither human nor mouse NK cells require SOCE for the MTOC-polarization process (8). Thus, although DAG-mediated signals also drive lytic granule docking and fusion in mouse NK cells, human NK cells appear to require SOCE for these final steps of degranulation.
In summary, our data show that primary mouse NK cell degranulation and cell-mediated killing do not depend on SOCE and, furthermore, point to DAG-mediated downstream signaling molecules, such as PKC, as the key driver of this process. Our studies show that degranulation, a key cell biological process that has long been thought to be driven primarily by Ca2+ signals, can occur in the absence of sustained Ca2+ entry in mouse NK cells.
We thank Dr. Anjana Rao, Dr. Patrick Hogan, and Dr. Masatsugu Oh-hora for providing us with STIM1/2 floxed mice for this study.
This work was supported by grants from the National Institutes of Health (NIH; R01-HL107589, R01-HL111501, R01-AI121250, R21-AI117282) and the American Asthma Foundation. J.F.-B. was supported by National Institute of Allergy and Infectious Diseases/NIH/Health and Human Services Grant T32-AI055428.
The online version of this article contains supplemental material.
Abbreviations used in this article:
Ca2+ release-activated channel
store-operated calcium entry
stromal interaction molecule
The authors have no financial conflicts of interest.