NK cells are potent antitumor effector cells (Cerwenka and Lanier, 2001; Ljunggren and Malmberg, 2007; Terme et al., 2008; Vivier et al., 2008). Accordingly, individuals with low NK cell activity display an increased risk to develop cancer (Imai et al., 2000), and high numbers of intratumoral NK cells are often correlated with improved prognosis for cancer patients (Coca et al., 1997; Villegas et al., 2002). Human tumors frequently express low levels of MHC class I molecules that interact with inhibitory NK cell receptors. For instance, alterations in the β2m gene can lead to an almost complete and irreversible lack of MHC class I in melanoma cells (D’Urso et al., 1991). In addition, many tumor cells express high levels of ligands for activating NK cell receptors (Raulet and Guerra, 2009), leading to efficient recognition by NK cells (Vivier et al., 2008; Pegram et al., 2011). So far, NK cell–based therapy was mainly successful in patients suffering from leukemia (Moretta et al., 2011). Acute myeloid leukemia patients that received haploidentical bone marrow grafts from Killer immunoglobulin receptor (KIR)–mismatched donors displayed a significantly increased 5-yr disease-free survival (Ruggeri et al., 2002). In addition, clinical benefits were observed upon infusion of KIR-mismatched NK cells after stem cell transplantation (Passweg et al., 2004; Miller et al., 2005; Geller and Miller, 2011; Geller et al., 2011). However, adoptive transfer of autologous IL-2–activated NK cells in patients suffering from solid tumors such as melanoma or renal cell carcinoma did not result in clinical benefits (Parkhurst et al., 2011). Thus, novel strategies are urgently needed to improve the antitumor activity of transferred NK cells against solid tumors. During certain viral infections (Sun et al., 2009a) and contact hypersensitivity reactions (O’Leary et al., 2006), persistent NK cell subpopulations mounting recall responses were detected, indicating previously unappreciated memory properties of NK cells (Paust and von Andrian, 2011; Sun et al., 2011; Vivier et al., 2011). In addition, NK cells preactivated with IL-12, IL-15, and IL-18 in vitro for 15 h were detectable at high numbers 3 wk after transfer into RAG-1−/− mice and produced high levels of IFN-γ upon restimulation (Cooper et al., 2009). Much lower cell numbers and IFN-γ production were observed when IL-15–preactivated NK cells were transferred. Thus, the activation of NK cells with certain cytokines resulted in an NK cell population with enhanced effector function upon restimulation, indicating that NK cells are able to retain memory of prior activation. Because IL-12/15/18–preactivated NK cells were shown to persist with sustained effector function after restimulation (Cooper et al., 2009), we investigated whether application of IL-12/15/18–preactivated NK cells improves current protocols of immunotherapy of cancer. Our study reveals that a single injection of IL-12/15/18–preactivated NK cells, but neither naive nor of IL-15– or IL-2–pretreated NK cells, combined with radiation therapy (RT), substantially reduced growth of established mouse tumors. Our results raise the possibilities for the development of novel NK cell–based therapeutic strategies for clinical application. RESULTS Adoptive transfer of IL-12/15/18–preactivated NK cells in combination with RT delays growth of established tumors Our study aimed at establishing protocols for the in vitro generation of NK cells that effectively reduce tumor growth upon adoptive transfer. In our tumor model, we applied 106 MHC class I–deficient RMA-S cells s.c. (Kärre et al., 1986), leading to progressive tumor growth. IL-12/15/18–preactivated NK cells were previously shown to persist for 3 wk after adoptive transfer with high effector function upon restimulation (Cooper et al., 2009). To address their therapeutic antitumor activity, 106 syngeneic NK cells preactivated in vitro with IL-12/15/18 or IL-15 alone for 16 h were adoptively transferred at day 7 into RMA-S tumor–bearing mice when the tumor diameter reached ∼5 mm. No therapeutic effect of the adoptively transferred cells was observed (Fig. 1 a). Because RT was shown to improve the antitumor activity of adoptively transferred T cells (Ganss et al., 2002; Gattinoni et al., 2005; Quezada et al., 2010; Xie et al., 2010), we combined NK cell infusion with total body RT of 5 Gy, which represents a sublethal dose of radiation. RMA-S tumor–bearing mice received RT and a single dose of 106 IL-12/15/18–preactivated NK cells on day 7 after tumor inoculation. RT by itself transiently delayed RMA-S tumor growth. Strikingly, adoptive transfer of IL-12/15/18–preactivated NK cells in mice that received RT significantly reduced tumor growth (Fig. 1 b, left) and significantly prolonged survival of recipient mice (Fig. 1 b, right). 22% of treated mice completely rejected tumor and remained tumor free. In contrast, adoptive transfer of control IL-15–pretreated NK cells did not affect the RT-mediated delay of tumor growth (Fig. 1 b). Similarly, transfer of neither naive NK cells nor NK cells expanded with IL-2 improved RT-induced tumor therapy (not depicted). Profound therapeutic antitumor effects of IL-12/15/18–preactivated NK cells were also obtained in a lung metastases model of B16–RAE-1ε melanoma (Fig. 1 c). Overall, our results demonstrate a substantial therapeutic benefit of a single infusion of IL-12/15/18–preactivated NK cells in combination with RT for the treatment of established solid tumors. Figure 1. Adoptive transfer of IL-12/15/18–preactivated NK cells in combination with RT delays RMA-S tumor growth. (a) C57BL/6 mice were s.c. inoculated with 106 RMA-S tumor cells. After 7 d, tumor-bearing mice received 106 NK cells i.v. that were preactivated in vitro with IL-15 or IL-12/15/18 for 16 h. Tumor growth and survival were monitored. The graph of tumor growth displays mean + SEM (n = 5). Survival data are outlined from two experiments (n = 9, 5, and 10 for None, IL-15 NK, and IL-12/15/18 NK groups, respectively) and presented as a Kaplan–Meier survival curve. (b) RMA-S tumor–bearing mice were irradiated with 5 Gy of total body RT on day 7 after tumor cell inoculation. 106 IL-15– or IL-12/15/18–preactivated NK cells were transferred i.v. 3 h after irradiation. The graph of tumor growth displays mean + SEM (n = 5–8). Survival data are outlined from five experiments (n = 24, 38, 11, and 23 for None, RT, RT+IL-15 NK, and RT+IL-12/15/18 NK groups, respectively) and presented as a Kaplan–Meier survival curve. *, P 500 are depicted as 500. Data are pooled from two independent experiments. IL-12/15/18–preactivated NK cells accumulate with high cell numbers in spleen and tumor after adoptive transfer into irradiated mice Our results revealed that the combination with RT was essential for the antitumor activity of transferred IL-12/15/18–preactivated NK cells. To assess the level of lymphodepletion resulting from total body irradiation with 5 Gy, we analyzed numbers of lymphocytes in spleen on day 4 after irradiation. After irradiation, highly decreased amounts of lymphocytes were observed (Fig. 2 a). 8% of CD4+ T cells and 2% of CD8+ T cells of the T cell numbers in nonirradiated mice were detected. The amount of regulatory T cells (Treg cells) was greatly reduced, as well. Figure 2. Adoptive transfer of IL-12/15/18–preactivated NK cells leads to high numbers of transferred NK cells in spleen and tumor. (a) C57BL/6 mice were inoculated with RMA-S tumor cells (day −7). Tumor-bearing mice received RT on day 0 or remained untreated as indicated. 4 d later, the numbers of CD3ε+CD4+, CD3ε+CD8+, CD3ε−NK1.1+, and CD4+FoxP3+ cells in spleen were determined. The graph indicates mean + SD (n = 3–6). Data are representative of two independent experiments. (b) C57BL/6 mice (CD45.2+) were inoculated with RMA-S tumor cells (day −7). Tumor-bearing mice received RT or remained nonirradiated and 106 IL-15– or IL-12/15/18–preactivated NK cells (CD45.1+) on day 0 as indicated. 4 d later, the numbers of host and transferred NK cells were analyzed. (c and d) Host (CD45.1−) and transferred (CD45.1+) NK cells in spleen (c) and tumor (d) are depicted. Cells were gated on CD3ε−NK1.1+ NK cells, and one representative dot plot from each group is shown (top). Numbers indicate percentages of host and transferred NK cells. Numbers of NK cells in spleen and the percentages of NK cells among total cells in the tumor are depicted in the bottom panels. Graphs indicate mean + SD (n = 3). Data are representative of two independent experiments. In a next step, we investigated whether total body RT affected the amounts of IL-15– or IL-12/15/18–preactivated NK cells after adoptive transfer. In these experiments, CD45.1+ NK cells were transferred into tumor-bearing, irradiated or nonirradiated hosts expressing CD45.2, and numbers of host and transferred NK cells were determined in spleen and tumor on day 4 after transfer (Fig. 2 b). Low cell numbers of transferred IL-15–pretreated NK cells were recovered from the spleen of nonirradiated hosts. In irradiated and nonirradiated hosts, similar amounts of transferred IL-15–pretreated NK cells were recovered (Fig. 2 c). In nonirradiated mice, significantly increased numbers of transferred IL-12/15/18–preactivated NK cells compared with IL-15–pretreated NK cells were detected (Fig. 2 c). Importantly, in irradiated mice the numbers of IL-12/15/18–preactivated NK cells further increased by 3.6-fold compared with nonirradiated mice. In irradiated mice, 16 times higher numbers of transferred IL-12/15/18–preactivated NK cells compared with IL-15–pretreated NK cells were detected. In parallel, amounts of host NK cells remained unchanged. Similar results were obtained in blood, lymph node, lung, and liver (not depicted). For direct tumor cell killing, NK cells have to infiltrate the tumor tissue. Fig. 2 d shows that upon irradiation, tumor infiltration of transferred NK cells was improved regardless of whether they were preactivated with IL-12/15/18 or with IL-15 alone. Comparatively few transferred IL-15–pretreated NK cells were detected in the tumors isolated from nonirradiated as well as irradiated hosts. In irradiated hosts, the percentage of transferred IL-12/15/18–preactivated NK cells among total cells in tumor was 20 times higher than in nonirradiated hosts (Fig. 2 d, bottom). Most importantly, the percentage of those cells was 32 times higher compared with IL-15–pretreated NK cells (Fig. 2 d). Overall, our data indicate that irradiation of 5 Gy resulted in higher numbers of transferred NK cells in spleen and tumor. In irradiated mice, IL-12/15/18–preactivated NK cells accumulated at strikingly higher numbers compared with IL-15–pretreated NK cells in spleen and in the tumor tissue already on day 4 after transfer. IL-12/15/18–preactivated NK cells persist for at least 3 mo after adoptive transfer Next, we analyzed transferred NK cells at later time points on day 11 and on day 90 after transfer into tumor-bearing, irradiated mice. Strikingly higher cell numbers of IL-12/15/18–preactivated NK cells were observed 11 d after transfer compared with IL-15–pretreated NK cells in spleen (Fig. 3 a) and tumor (not depicted). Of note, IL-12/15/18–preactivated NK cells were still detectable 3 mo after adoptive transfer in mice that had rejected the tumors and had remained tumor free. The transferred cells were detected in different organs such as spleen, blood, lung, liver (Fig. 3 b), and lymph node (not depicted). Overall, IL-12/15/18–preactivated NK cells can be recovered at high cell numbers after transfer in tumor-bearing, irradiated hosts and persist for at least 3 mo. Figure 3. IL-12/15/18–preactivated NK cells persist in recipient mice. RMA-S tumor–bearing mice received RT and preactivated NK cells as described in Fig. 2 b. 11 or 90 d after NK transfer, the numbers of host and transferred NK cells were analyzed. (a) Host (CD45.1−) and transferred (CD45.1+) NK cells in spleen on day 11 are shown. Cells were gated on CD3ε−NK1.1+ NK cells, and one representative dot plot from each group is shown (top). Numbers indicate percentages of host and transferred NK cells. Numbers of NK cells are depicted in the bottom panels. Graphs indicate mean + SD (n = 5). Data are representative of two independent experiments. (b) Representative dot plots (n = 3) of host (CD45.1−) and transferred (CD45.1+) NK cells on day 90 after transfer of IL-12/15/18–preactivated NK cells are shown from spleen, blood, lung, and liver of animals that had rejected the tumors. Cells were gated on CD3ε−NK1.1+ NK cells. Data are representative of two independent experiments. After adoptive transfer, IL-12/15/18–preactivated NK cells display rapid proliferation that is dependent on endogenous IL-2 After adoptive transfer, IL-12/15/18–preactivated NK cells were found at high cell numbers in different organs in tumor-bearing, irradiated mice. The highest ratio between transferred and host NK cells was in the lung, possibly because of the i.v. injection of NK cells for adoptive transfer (Fig. 4 a). In all other organs, transferred and host NK cells were detected at similar ratios, indicating that transferred cells distributed without tropism for certain organs. To investigate whether high numbers of transferred NK cells were associated with proliferation, we transferred CSFE-labeled IL-15– or IL-12/15/18–pretreated NK cells into tumor-bearing, irradiated mice. 4 d after transfer, >95% of IL-12/15/18–preactivated NK cells had proliferated with more than eight daughter generations in spleen (Fig. 4 b), blood, lymph node, lung, liver, and tumor (not depicted). In contrast, very low percentages of IL-15–pretreated NK cells had proliferated at this early time point in spleen (Fig. 4 b) and other organs (not depicted). Figure 4. Rapid proliferation of IL-12/15/18–preactivated NK cells in recipient mice is dependent on endogenous IL-2. NK cells (CD45.1+) were pretreated with IL-15 or IL-12/15/18 for 16 h and labeled with CFSE. RMA-S tumor–bearing mice received RT and 106 CFSE-labeled NK cells 3 h after irradiation as described in Fig. 2 b. (a) 4 d later, host and transferred cells were analyzed in the spleen, blood, lymph node, lung, liver, and tumor. The graph depicts the ratio between the numbers of transferred and host NK cells (mean + SD; n = 5). Data are representative of two independent experiments. (b) 4 d after transfer, in vivo proliferation of transferred NK cells was analyzed. Histograms were gated on CD3ε−NK1.1+CD45.1+ transferred NK cells, and one representative histogram from each group is shown (n = 5). Percentages of cells that proliferated are depicted. Data are representative of three independent experiments. (c) NK cells were treated with IL-15 or IL-12/15/18 for 16 h and stained with anti-CD25, -CD122, and -CD132 mAbs (bold black lines) or the isotype control (filled histograms). Data are representative of three independent experiments. (d) CFSE-labeled IL-12/15/18–pretreated NK cells (CD45.1+) were transferred into RMA-S tumor–bearing, irradiated mice (CD45.2+) as described in Fig. 2 b. Anti–IL-2, –IL-15, or –IL-12 mAb was applied as described in Materials and methods. 4 d after adoptive transfer, in vivo proliferation of transferred NK cells was analyzed. Cells were gated on CD3ε−NK1.1+CD45.1+ transferred NK cells, and one representative histogram from each group is shown (n = 3–5). The replication index indicating the fold expansion of the proliferating cells was calculated by FlowJo. Numbers of transferred NK cells in spleen are depicted. Graphs indicate mean + SD (n = 3–5). Data are representative of two independent experiments. (e) IL-2−/− mice or littermates (CD45.2+; top) and IL-15−/− or wild-type mice (CD45.2+; bottom) received RT and 106 CFSE-labeled preactivated NK cells (CD45.1+) 3 h after irradiation. 4 d later, CFSE dilution of transferred NK cells from spleen was determined. Data shown were gated on CD3ε−NK1.1+CD45.1+ transferred NK cells. One representative histogram from each group is shown (n = 3–5). To determine factors involved in the rapid proliferation, expression of cell surface molecules implicated in NK cell proliferation was analyzed on preactivated NK cells before adoptive transfer. Our results reveal elevated expression of CD25 (IL-2R α-chain) and CD132 (IL-2R γ-chain) and slightly lower expression of CD122 (IL-2R β-chain) on IL-12/15/18–preactivated NK cells compared with IL-15–pretreated NK cells (Fig. 4 c). Upon culture of MACS-sorted NK cells in IL-12 or IL-18, low levels of CD25 were induced. The presence of both IL-12 and IL-18 was required to induce high expression of CD25 on purified NK cells during the activation (not depicted). CD127 (IL-7R α-chain) and CD28 were not detectable (not depicted). Because IL-2R chains were highly expressed on IL-12/15/18–preactivated NK cells, we determined the requirement of IL-2 (binding to the IL-2R α-, β-, and γ-chains) and IL-15 (binding to the IL-2R β- and γ-chains) for their rapid in vivo proliferation. IL-12/15/18–preactivated NK cells showed significantly less proliferation in spleen after IL-2 neutralization (Fig. 4 d) or upon transfer into IL-2−/− mice (Fig. 4 e, top). Neutralization of IL-15 did not significantly affect their rapid proliferation (Fig. 4 d). Accordingly, proliferation was not reduced in IL-15−/− hosts (Fig. 4 e, bottom). IL-12 has been implicated in the proliferation of Ly49H+ NK cells during MCMV infection (Sun et al., 2009b). In our experimental model, neutralization of IL-12 in the recipient mice did not delay proliferation of transferred NK cells. Importantly, upon in vivo neutralization of IL-2, but not of IL-15 or IL-12, significantly lower numbers of transferred NK cells were detected (Fig. 4 d). Together, our data reveal high expression of CD25 on IL-12/15/18–preactivated NK cells before transfer and an indispensible role of IL-2 for their rapid proliferation in vivo. The rapid proliferation of IL-12/15/18–preactivated NK cells after adoptive transfer depends on the presence of CD4+ T cells that produce IL-2 In a next step, we investigated the source of IL-2 in our experimental model. Fig. 5 a shows that on day 4 after irradiation and NK cell transfer, IL-2 production in spleen was mainly detected in CD4+ T cells. Other cell populations (CD4−) produced negligible levels of IL-2. Because T cells are the main source of IL-2, depletion of CD4+ or CD8+ T cells was performed using the respective antibodies. Depletion of CD4+ T cells resulted in a pronounced reduction in proliferation of transferred IL-12/15/18–preactivated NK cells (Fig. 5 b) that was comparable with IL-2 neutralization (Fig. 4 d). Significantly lower numbers of transferred NK cells were detected in spleen (Fig. 5 b). Depletion of CD8+ T cells did not significantly affect the rapid proliferation (Fig. 5 b) but significantly reduced the numbers of transferred NK cells, although to a much lesser extent compared with CD4+ T cell depletion. Of note, CFSE dilution of transferred IL-12/15/18–preactivated NK cells on day 4 after transfer remained unchanged in diphtheria toxin–treated DC-deficient CD11c.DOG mice (not depicted; Hochweller et al., 2008). Overall, our results indicate that IL-2 and host CD4+ T cells promote the rapid proliferation and expansion of IL-12/15/18–preactivated NK cells in tumor-bearing, irradiated hosts. Figure 5. The rapid proliferation of IL-12/15/18–preactivated NK cells is dependent on CD4 + T–producing IL-2. CFSE-labeled IL-12/15/18–pretreated NK (CD45.1+) cells were transferred into RMA-S tumor–bearing, irradiated mice (CD45.2+) as described in Fig. 2 b. Anti-CD4 or anti-CD8 mAb was applied as described in Materials and methods. 4 d later, spleen cells were analyzed. (a) Spleen cells were restimulated by PMA/ionomycin, and IL-2 expression was determined by intracellular staining. One representative dot plot gated on total splenocytes is shown. Numbers indicate percentages among total splenocytes. Percentages of IL-2–producing CD4+ or CD4− cells (after subtraction of the isotype control) in spleen are shown in the graph (mean + SD; n = 3). n.d., not detectable. Data are representative of three independent experiments. (b) CFSE-dilution in spleen was determined. Data shown were gated on CD3ε−NK1.1+CD45.1+ transferred NK cells, and one representative histogram from each group is shown (n = 3). The replication index of transferred NK cells was calculated by FlowJo. Numbers of transferred NK cells in spleen are shown. Graphs indicate mean + SD (n = 3). Data are representative of three independent experiments. IL-12/15/18–preactivated NK cells display a mature phenotype and potent effector function after transfer into irradiated recipients Before and after adoptive transfer, IL-12/15/18–preactivated NK cells displayed a mature phenotype characterized by CD11bhighCD27low, KLRG1high, and CD43high expression (Fig. 6, a and b) that was similar to memory NK cells during MCMV infection (Sun et al., 2009a). In contrast, IL-15–pretreated NK cells expressed lower levels of KLRG1 and CD43 before and after transfer. Moreover, before transfer, IL-12/15/18–preactivated NK cells produced substantially higher levels of IFN-γ, granzyme B, and perforin compared with IL-15–pretreated NK cells (Fig. 7 a). Figure 6. IL-12/15/18–preactivated NK cells display a mature phenotype before and after transfer. NK cells (CD45.1+) were treated with IL-15 or IL-12/15/18 for 16 h and transferred into tumor-bearing, irradiated mice (CD45.2+; n = 5) as described in Fig. 2 b. (a) Dot plots show the expression of CD11b and CD27 on splenic naive NK cells (gated on CD3ε−NK1.1+) and IL-15– or IL-12/15/18–preactivated NK cells before (gated on CD3ε−NK1.1+) and after transfer (gated on CD3ε−NK1.1+CD45.1+ transferred NK cells) on day 11. Numbers indicate percentages. (b) Histograms show expression of KLRG1 and CD43 on NK cells. Numbers indicate geometric mean fluorescence. All data are representative of two independent experiments. Figure 7. IL-12/15/18–preactivated NK cells show high effector function before and after transfer into irradiated recipients. (a) NK cells were pretreated with IL-15 or IL-12/15/18 for 16 h. Expression of IFN-γ, granzyme B, and perforin (bold black lines) or the isotype control (filled histograms) was determined by flow cytometry. Data are representative of three independent experiments. (b and c) IL-15– or IL-12/15/18–pretreated NK cells (CD45.1+) were transferred into untreated (No RT) or irradiated (RT) tumor-bearing mice (CD45.2+) as described in Fig. 2 b. 4 d later, splenocytes were restimulated by RMA-S cells and stained with anti–IFN-γ, anti–granzyme B, and anti-perforin mAbs (bold black lines) or the isotype control (filled histograms; b). Dot plots (left) depict whole splenocytes. Histograms (right) are gated on CD45.1+NK1.1+ cells as indicated in the left panel. Numbers indicate percentages of IFN-γ–producing cells among the transferred NK cells or mean fluorescence intensity (MFI) of granzyme B and perforin expression. One representative dot plot or histogram from each group is shown. (c) Numbers of IFN-γ–producing transferred NK cells and mean fluorescence intensity of granzyme B and perforin detected in transferred NK cells are depicted. Graphs indicate mean + SD (n = 3). Data shown are representative of two independent experiments. (d) NK cells from wild-type, IFN-γ−/−, or perforin−/− mice were stimulated with IL-12/15/18 for 16 h and transferred into tumor-bearing, irradiated mice 7 d after tumor inoculation. Survival data are outlined from two experiments (n = 15, 9, 10, and 10 for RT, RT+NK wild-type, RT+NK IFN-γ−/−, and RT+NK perforin−/− groups, respectively) and presented as a Kaplan–Meier survival curve. ** , P 90%. Tumor cells and mouse tumor models. The MHC class I–deficient lymphoma cell line RMA-S was cultured in RPMI-1640 (Sigma-Aldrich) supplemented with 10% FCS, 1% l-glutamine, 1% penicillin, and 1% streptomycin (Invitrogen). Mice were s.c. injected with 106 RMA-S lymphoma cells that were washed three times in PBS. On day 7 after tumor cell inoculation, tumor-bearing mice were treated with 5 Gy of total body RT (0.49 Gy/min). 106 IL-15– or IL-12/15/18–treated syngeneic NK cells were i.v. injected ∼3 h after irradiation. Tumor diameters were measured by a caliper. Mice were euthanized when the tumors reached the mean diameter of 1.5 cm. The tumor volume was calculated as large diameter × small diameter × depth. The melanoma cell line B16 ectopically expressing RAE-1ε was cultured in DMEM (Sigma-Aldrich), 10% FCS, 1% l-glutamine, 1% penicillin, and 1% streptomycin (Invitrogen). Mice were i.v. injected with 106 B16–RAE-1ε cells and received total body irradiation and adoptive transfer of NK cells at day 7 after tumor cell inoculation as described for the RMA-S tumor model. On day 14, lungs were dissected and fixed in Bouin’s buffer (Sigma-Aldrich), and numbers of nodules were counted under a dissecting microscope (S8AP0; Leica). Preparation of single-cell suspension from tumors. Tumors were removed, cut into small pieces, and digested with 0.5 mg/ml hyaluronidase (Sigma-Aldrich) and 0.5 mg/ml DNase I (Sigma-Aldrich) at 37°C for 30 min. Percentages of NK cells were calculated among all viable cells in the tumors (gated on 7-AAD− population, including both tumor cells and tumor-infiltrating leukocytes). Ex vivo stimulation of mouse NK cells. Cells were isolated from spleen of treated mice and co-cultured with RMA-S cells (106 spleen cells/5 × 105 RMA-S cells) for 22 h. GolgiStop (BD) was added 4 h before the end of co-culture. Cells were stained for surface markers, fixed, and permeabilized (eBioscience), followed by intracellular staining of IFN-γ and granzyme B. For intracellular staining of IL-2, splenocytes were restimulated with 50 ng/ml PMA (Sigma-Aldrich) and 500 ng/ml ionomycin (Sigma-Aldrich) for 4 h in the presence of GolgiStop. In vivo proliferation assay. In vitro activated NK cells were labeled with 1.5 µM CFSE (Sigma-Aldrich) at room temperature for 15 min. After three washes with PBS, cells were transferred into tumor-bearing, irradiated mice. 4 d later, single-cell suspensions from spleen and other organs were prepared, stained, and analyzed by flow cytometry. The replication index indicating the fold expansion of the proliferating cells was calculated by FlowJo. 500 µg anti–IL-2 (S4B6 and JES6-1A12, 1:1; Bio X Cell), 200 µg anti-CD4 (GK1.5; Bio X Cell), and 200 µg anti-CD8 (2.43; Bio X Cell) mAbs were i.p. injected 2 d before adoptive transfer of CFSE-labeled NK cells and every second day. 750 µg anti–IL-12 (C17.8; Bio X Cell) mAb was i.p. injected 1 d before NK cell infusion; 25 µg anti–IL-15/IL-15Rα (GRW15PLZ; eBioscience) mAb was i.p. injected 1 d before NK cell infusion and subsequently every second day. Immunohistochemistry. Freshly isolated tumors were embedded in O.C.T. (optimal cutting temperature) compound, frozen in liquid nitrogen, and stored at −80°C until use. Cryosections of 6 µm in thickness were air dried for 10 min at room temperature and fixed for 10 min at 4°C with acetone. The following primary antibodies (at 1:200 dilution) were used: rat anti-CD4 (H129.19), rat anti-CD8 (53-6.7), and mouse anti-CD45.1-Biotin (A20). The anti–rat Ig HRP Detection kit (BD) was used for detection according to the manufacturer’s protocol. Sections were counterstained with Hematoxylin (Mayer’s hemalam solution; Applichem). Images were digitally captured on a BX51 microscope (Olympus) and imaged using cell^D software (Olympus). Human NK cells. PBMCs from healthy donors were isolated by Ficoll separation (LSM 1077 lymphocyte separation medium; PAA). NK cells were purified by negative selection (Human NK cell isolation kit; Miltenyi Biotec) with a purity of CD3−CD56+ NK cells >95%. NK cells were preactivated in SCGM medium (CellGenix) containing 20% human serum (PAA), 1% penicillin, and 1% streptomycin (Invitrogen) with 10 ng/ml IL-12 (PeproTech), 20 ng/ml IL-15, and 100 ng/ml IL-18 (MBL) for 16 h. To assess in vitro proliferation, preactivated NK cells were labeled with 2 µM CFSE (Sigma-Aldrich) and cultured in 100 U/ml recombinant human IL-2 (National Institutes of Health; day 0). On days 2, 4, 6, and 8, cells were counted using a hemocytometer, and CFSE dilution was analyzed by flow cytometry on a FACSCalibur (BD). Dead cells were excluded by gating on 7-AAD− cells. For IFN-γ production, NK cells were harvested on days 4 and 8 and restimulated with either 10 ng/ml IL-12 and 50 ng/ml IL-15 or K562 cells in the presence of 100 U/ml IL-2 at an E/T ratio of 1:1. Supernatants were harvested after 24 h, and IFN-γ was measured by ELISA (BioLegend). Statistics. The statistical significance of results from experimental groups in comparison with control groups was determined by the Student’s t test. Survival data were analyzed with the log-rank test. All tests were two tailed, and P < 0.05 was considered to be statistically significant.