This study was approved (approval no. 329/10) by the Ethics Committee of the Goethe University Frankfurt (Frankfurt, Germany) and was performed in accordance with the Declaration of Helsinki with written informed consent given by every participant. NK cells were isolated from freshly generated donor buffy coats provided by the German Red Cross Blood Donation Service (DRK-Blutspendedienst Baden-Württemberg-Hessen, Frankfurt, Germany), using immunomagnetic negative selection (EasySep™ Human NK Cell Enrichment kit, StemCell Technologies, Vancouver, BC, Canada) according to the manufacturer instructions.

Briefly, PBMCs obtained from buffy coats by density gradient centrifugation were diluted to a final concentration of 100 × 10⁶ cells/mL. Then, 50 µL of enrichment cocktail were applied per milliliter cell suspension and incubated for 10 min. Subsequently, 100 µL of microbead suspension were added and incubated for another 5 min. After incubation for 2.5 min in “The Big Easy” EasySep™ Magnet, the NK cell-enriched suspension was decanted into a new tube.

Purified primary NK cells were cultured at a concentration of 2 × 10⁶ cells/mL in X-VIVO 10 medium (Lonza Group Ltd., Basel, CH) supplemented with 5% heat inactivated human fresh frozen plasma (FFP; provided by DRK-Blutspendedienst Baden-Württemberg-Hessen, Frankfurt, Germany) and 100 U/mL penicillin and 100 µg/mL streptomycin (Gibco, New York, NY, USA). Correlating to the batch, the cells were provided with IL-2 (ProleukinS, Novartis Pharmaceuticals, Horsham, UK), 100 U/mL (IL-2¹⁰⁰) or 1,000 U/mL (IL-2¹⁰⁰⁰), 10 ng/mL IL-15 (IL-15), 25 ng/mL IL-21 (both PeproTech, Rocky Hill, CT, USA) (IL-21) or combinations of those (IL-2¹⁰⁰ + 15, IL-2¹⁰⁰ + 15 + 21, IL-15 + 21). Every 3–4 days, half of the medium was replaced by fresh medium containing the corresponding cytokines or combinations. One batch was treated with IL-15 and boosted with IL-21 only 3–4 days before harvest and analysis (IL-15 + 21_boost).

Chronic myologenous erythroleukemia cell line K562 and RMS cell lines RH30 and RD (45) were purchased from the American Type Culture Collection (Manassas, VA, USA). Cells were maintained in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 10% heat inactivated fetal calf serum (Invitrogen, Paisley, UK), 100 U/mL penicillin and 100 µg/mL streptomycin. The cells were splitted twice a week.

GFP/luciferase-expressing RD cells (RD^GFP/Luc) were generated via lentiviral transduction using vector particles pseudotyped with vesicular stomatitis virus G protein that were produced using the transfer plasmid pSEW-luc2, which encodes firefly luciferase and enhanced green fluorescent protein linked via a 2A peptide (46). GFP positive cells were enriched by fluorescence activated cell sorting (FACS) using a FACSAria II™ device (BD Biosciences, San Jose, CA, USA). Culture conditions for transduced cells were the same as for non-transduced cells.

In order to check the quality of enriched NK cells and to monitor the phenotype of ex vivo expanded NK cells, samples were analyzed with a FACSCanto 10c™ system (BD Biosciences). Post-harvesting cells were resuspended in FACS buffer containing CellWASH (BD Biosciences), 0.5% bovine serum albumin (Sigma Aldrich, Taufkirchen, Germany) and 0.01% NaN₃ (0.1 M, Sigma Aldrich).

Intracellular staining was accomplished using formaldehyde (AppliChem GmbH, Darmstadt, Germany) for fixation and 90% methanol for membrane perforation.

The following antibodies were used: CD3-APC (#UCHT1), TRAIL-R-APC [#DJR2-4(7-8)], FAS-BV421 (#DX2), CD56-FITC (clone #HCD56), FAS-L-PE (#NOK-1), TRAIL-PE (#RIK-2), CD19-PerCP (#HIB19), CD16-PE/Cy7 (#3G8) all from Biolegend (San Diego, CA, USA); CD3-V450 (#UCHT1), CD19-V450 (#HIB19), CD14-V450 (#MφP9), CD45-BV510 (#HI30), NKp30-AF488 (#P30-15), DNAM-1-FITC (#DX11), NKp44-PE (#P44-8.1) CD45-APC (#2D1), CD137/4-1BB-APC (#4B4-1), CD107a-APC/H7 (#H4A3), IFN-γ-FITC (#B27), pAKT-AF647 (#F29-763), pERK1/2-AF647 (#20A), from BD Biosciences; CD56-APC/AF700, NKG2D-APC (#ON72), CD11a/LFA-1-FITC (#25.3) from Beckman Coulter Immunotech (Brea, CA, USA); CD45-PE (#HI30) from Invitrogen (Carslbad, CA, USA); and NKp46-APC (#9E2), KIR2D-FITC (#NKVFS1), CD158e/k-PE (#5.133), NKG2A-APC (#Z199) from Miltenyi Biotec (Bergisch-Gladbach, Germany). Depending on the panel Zombie Violet Fixable Viability Kit (BioLegend), 7AAD (BD Biosciences) or strongly diluted DAPI were used for live/death discrimination.

Data were acquired on a FACSCanto 10c™ instrument (BD Biosciences, San Jose, CA, USA) and analyzed using Flowjo (Tree Star Inc., Ashland, OR, USA).

To investigate the killing capacity of ex vivo expanded NK cells a FACS-based cytotoxicity assay was employed. NK effector cells were harvested after 6 days of cytokine stimulation. Target cells were harvested and stained for 5 min with Celltrace CFSE (Molecular Probes, Eugene, OR, USA) in a final concentration of 5 µM. After all cells had been washed with Dulbecco’s Phosphate Buffered Saline (DPBS, Gibco), they were resuspended in X-VIVO 10 medium supplemented with 5% heat inactivated human FFP, 100 U/mL penicillin and 100 µg/mL streptomycin. NK cells and target cells were combined in a U-bottom 96-well plate at effector to target (E:T) ratios of 1:1, 5:1, and 10:1, adjusted to 25,000 target cells per well in a total volume of 200 µL. After 5 h of co-incubation the supernatant was removed, cells were harvested and resuspended in a 1:6,000 dilution of DAPI for live/death discrimination. From each well, the same amount of target cells was acquired using a FACSCanto 10c™ device. Samples exclusively containing target cells served as spontaneous lysis controls. Spontaneous lysis was subtracted from each sample to obtain specific lysis values. All experiments were conducted in triplicates for each NK cell donor.

To address additional antibody-dependent cellular cytotoxicity (ADCC) by NK cells, in a separate experiment, 10 µg/mL of anti-ErbB2 antibody Trastuzumab (Herceptin, ROCHE, Mannheim, Germany) were added and cytotoxicity compared to NK cell cytotoxicity in the absence of Trastuzumab.

Conjugation capacity of stimulated NK cells was addressed by staining NK cells with Celltrace CFSE, while target cells were stained with Celltrace Calcein Violet AM for 20 min at 4°C. After intensive washing, effector and target cells were co-incubated for 0–90 min, then shortly vortexed and fixed with 1–2% formaldehyde. Flow cytometry data were acquired on a FACSCanto 10c™ instrument.

Due to limited availability of NK cell numbers, cytotoxicity and conjugation assays were performed with cells from other donors than were used for proliferation assays.

Degranulation potential of cytokine stimulated NK cells was assessed as described (47), with cells harvested on day 6 of cultivation. Cells were washed and resuspended in fresh X-VIVO 10 medium supplemented with 5% heat-inactivated human FFP, 100 U/mL penicillin, and 100 µg/mL streptomycin. After 1 h, cells were incubated with anti-human CD107a, followed by an additional hour of incubation with GolgiStop™ (BD Biosciences). Cells were washed, blocked with human IgG and stained with Zombie Violet™ Fixable Viability Kit for live/death discrimination. Post washing, cells were stained for CD45, CD56, and CD16, fixed with formaldehyde solution (2% final concentration) and permeabilized with saponin buffer [0.2% saponin, 1% bovine serum albumin (both Sigma Aldrich) in DPBS]. In the end, cells were stained intracellularly with anti-human IFN-γ, washed, and measured by flow cytometry.

For Phosflow analysis freshly harvested cells were stained on the surface for CD45, CD3, and CD56, fixed and permeablilized with 90% methanol and intracellularly stained for pAKT, pERK1/2, and pMAPK. Flow cytometry data acquisition was performed on a FACSCanto 10c™ instrument.

Due to limited availability of NK cells numbers, degranulation assays were performed with cell preparations from different donors than proliferation or cytotoxicity assays.

Western blotting was deployed for the assessment of production and release of apoptosis-mediating perforin and granzyme B. NK cells and culture supernatants were harvested on day 6 of cytokine stimulation. Cells were lysed using RIPA buffer supplemented with cOmplete™ Protease Inhibitor Cocktail (ROCHE) followed by sonification. Protein concentrations were determined via Bradford assay (Protein Assay Dye Reagent Concentrate, Bio-Rad, Munich, Germany). Separation of proteins was accomplished by SDS-PAGE followed by semi-dry blotting onto polyvinylidenfluoride membranes. Before antibody application, membranes were blocked with 5% skim milk powder in DPBS. Mouse monoclonal antibodies against human perforin (1:1,000, LifeSpan BioSciences, Seattle, WA, USA), granzyme B (1:200, #2C5, Santa Cruz, Heidelberg, Germany) and a rabbit antibody against human γ-tubulin (1:2,000, Sigma Aldrich) were used as primary antibodies. HRP conjugated rabbit anti-mouse IgG (1:15,000, Sigma Aldrich) and goat-anti rabbit (1:16,000, Sigma Aldrich) served as secondary antibodies. The antigen–antibody complexes were detected using an ECL-chemiluminescence system (Pierce™ ECL, Thermo Scientific, Waltham, MA, USA) according to the manufacturer’s instructions, and visualized using X-ray films (Fujifilm, Tokyo, Japan).

Due to limited availability of NK cell numbers, immunoblot assays were performed with cells from other donors than used for proliferation or functional assays.

Cytokine secretion was examined by cytometric bead array analyses (CBA) on supernatants of stimulated NK cells on the sixth day of cultivation using BD CBA Flex Sets for IFN-γ, tumor necrosis factor (TNF)-α, macrophage inflammatory protein (MIP)-1α, monocyte chemoattractant protein (MCP)-1, IL-8, IL-10, and granulocyte-macrophage colony-stimulating factor (GM-CSF) (BD Biosciences). The tests were performed according to the manufacturer’s instructions using a mixture of PE-conjugated antibodies against the cytokines listed above. Data were acquired with the BD FACSVerse™ Bioanalyzer and were quantitated using the FCAP Array™ software (v3.0.1; BD Biosciences).

The in vivo experiments were approved by the government committee (Regierungspräsidium Darmstadt, Darmstadt, Germany) and were conducted in accordance to the requirements of the German Animal Welfare Act.

Female, 12- to 16-week-old NOD/SCID/IL-2Rγc^−/− mice (NSG) mice were injected subcutaneously with 10⁵ luciferase expressing RD^GFP/Luc cells. After visual detection of tumor nodes about 3 weeks post-cell injection, mice were imaged by performing a Cone-Beam CT (CBCT) operating at 65 kV, 0.5 mA and irradiated while immobilized with 2.5% isoflurane anesthesia (AbbVie, Wiesbaden, Germany) using a Small Animal Radiation Research Platform (SARRP, Xstrahl Ltd., Camberley, UK). CBCT images were transferred to MuriPlan™ Software and individual isocenters were selected for targeted radiation therapy applying a two-field geometry. Fractionated single doses of 2.5 Gy using a 1–10-mm collimated beam operating at 175 kV, 15 mA were applied four times a week to reach a total dose of 27.5 Gy. Post-termination of RT, adoptive transfer of IL-15 expanded and IL-21 boosted NK cells was accomplished in three injections, once a week. NK cells were purified from 11 buffy coats as described above and maintained separately during cultivation in 25 cm² suspension cell culture flasks (Cellstar, Frickenhausen, Germany) using the IL-15 + 21_boost protocol as described before. After 10–11, 17–18, and 24–25 days of ex vivo expansion and stimulation, NK cells were harvested, washed, and pooled. The NK cells were injected intravenously via the tail vein with 10⁷ NK cells in a total volume of 100 µL per mouse. Due to limited availability of NK cell numbers, in vivo application was performed with cells from other donors than used for proliferation or functional assays.

Tumor growth was monitored by caliper measurements and bioluminescence imaging (BLI) using an IVIS Lumina II system (Perkin Elmer, Waltham, MA, USA). For the latter method, mice were anesthetized by isoflurane inhalation and subcutaneously injected with 150 µg of in vivo grade VivoGlo™ luciferin (Promega, Madison, WI, USA) dissolved in 100 µL DPBS per mouse. Images were acquired after an incubation time of 15 min. BLI data analysis was performed using Living Image^® software (Perkin Elmer).

Results were analyzed using repeated measures one-way ANOVA with the Geisser–Greenhouse correction. For in vivo experiments, two-way ANOVA was used to compare the tumor-growth curves of different treatment groups. Statistical calculations were performed using GraphPad Prism v6 (GraphPad, La Jolla, CA, USA), and p values <0.05 were considered statistically significant.

The primary goal of ex vivo expansion of NK cells is to yield high cell numbers for adoptive transfer with an appropriate quality and optimal cytotoxic functionality of the NK cell product. To this end, NK cells were isolated by an immunomagnetic negative selection (as described in Section “Materials and Methods”) and expanded ex vivo under feeder-cell free cultivation conditions addressing the impact of different cytokines. NK cells were cultured with IL-2¹⁰⁰ (100 U/mL), IL-2¹⁰⁰⁰ (1000 U/mL), IL-15 (10 ng/mL), IL-21 (25 ng/mL), combinations of these cytokines or IL-15 and a 3-day IL-21 boost (IL-15 + 21_boost). Purity of the enriched NK cells was determined and cell numbers as well as viability were monitored over a period of 4–6 weeks (Figures 1A–D).

The average content of starting material after NK cell purification consisted of 89% NK cells. Purity of the CD3^negCD56^pos NK cells decreased slightly on day 3, but increased afterward during the culturing process. With protocols using IL-15, IL-15 + 21, and IL-15 + 21_boost, the purity reached over 95% from day 10 on. In cultures containing exclusively IL-2 also CD3^pos, cells expanded and diminished NK cell purity especially at late culture time points after 2–3 weeks (data not shown).

Remarkably, all protocols involving the addition of IL-15, such as IL-15 alone, IL-2¹⁰⁰ + 15, and IL-15 + 21_boost, led to a high increase in the number of NK cells over the first 10 days reaching a plateau until day 21. Permanent stimulation with IL-21 (IL-2¹⁰⁰ + 15 + 21 and IL-15 + 21) also provoked rapid but less distinct expansion. In general, all expansion rates decreased slowly after 3 weeks of ex vivo cultivation (Figure 1A).

High levels of IL-2 (IL-2¹⁰⁰⁰) evoked proliferation of NK cells during the first 6 days, but average expansion rates declined subsequently. Cultivation in the presence of low IL-2 levels (IL-2¹⁰⁰) or IL-21 alone did not induce proliferation, instead NK cells died (Figures 1A–D).

Irrespective of donor-dependent differences, stimulation with IL-15, IL-2¹⁰⁰ + 15, and IL-15 + 21_boost performed better than all other stimulation protocols. Of note, permanent exposure to IL-21 dampened IL-15-driven expansion, while a short boost with IL-21 did not disturb proliferation of NK cells, but, in some cases, even increased the expansion rate. With the IL-15 + 21_boost protocol, expansion rates ranged between 2- and 10-fold depending on the donor, exhibiting an average 4.5-fold increase on day 10 of culture (Figure 1C).

Percentages of viable cells decreased during the first days, but then recovered upon further stimulation (Figure 1B). Comparisons on day 10 of cultivation indicated significantly higher frequencies of viable cells in the product of IL-15-stimulated cells than in products obtained from expansion protocols with permanent addition of IL-21. Nevertheless, a short boost with IL-21 did not significantly affect cell viability compared to IL-15 mono treatment (Figure 1D).

Next, we investigated changes in the phenotype and subset composition of NK cells triggered by cytokine-induced ex vivo expansion. To monitor their maturation state, NK cells were analyzed for the distribution of the CD56^highCD16^neg and CD56^dimCD16^pos subset over 4–6 weeks (Figures 2A,B). As all stimulation protocols induced an upregulation of the NK cell marker CD56, only a discrimination of CD16^pos and CD16^neg NK cells was implemented for further analysis of the maturation state.

Depending on the expansion protocol, the frequency of CD16^pos NK cells was reduced. Only protocols with permanent IL-21 exposure (IL-21, IL-15 + 21, and IL-2¹⁰⁰ + 15 + 21) maintained a mature phenotype (Figure 2A). This effect became more pronounced the longer the ex vivo expansion endured (Figures 2A,B). In parallel with the decrease in mature phenotype, the proportion of less mature CD16^neg NK cells was increased (Figure 2B).

Furthermore, cytokine stimulation induced upregulation of activating and inhibitory receptors on the NK cell surface (Figure 2C). Interestingly, different cytokine combinations led to a significant increase of apoptosis inducing TRAIL or FAS ligand (FAS-L) in line with enhanced expression of TRAIL-R and FAS. We further observed that the presence of IL-21 had an additive effect with IL-15 regarding an enhanced surface expression of TRAIL, TRAIL-R, DNAM-1, FAS, FAS-L, NKp46, but also inhibitory NKG2A. For NKG2D, NKp30, NKp44 and inhibitory KIR2D and CD158 e/k (KIR3DL1/DL2), IL-21 counteracted the up-regulating effect of IL-15.

TRAIL, DNAM-1, NKp46, and NKG2A were rapidly upregulated upon short exposure to IL-21 if cells were pre-stimulated with IL-15 (IL-15 + 21_boost) and reached expression levels similar to those of NK cells stimulated permanently with IL-15 and IL-21 (IL-15 + 21). By contrast, changes in the levels of other surface markers were less rapid, showing significant differences for cells stimulated following the different protocols.

Almost all CD16^neg NK cells expressed two of the three activating receptors, NKG2D, DNAM-1, and NKp44, only a small fraction of these cells expressed only one (NKG2D) and an even smaller fraction none. After 6 days of NK cell expansion, almost all CD16^neg NK cells co-expressed all three receptors (Figure S1 in Supplementary Material, left panels). Among the CD16^pos fraction, half of it expressed only NKG2D and half co-expressed NKG2D and DNAM-1. After expansion with IL-15 alone or an additional short-term IL-21 boost, on day 6 of culturing, half of the CD16^pos fraction expressed all three receptors, and half co-expressed two. Permanent exposure to IL-21 further increased the fraction of CD16^pos NK cells that co-expressed all three receptors. Altogether, no prominent difference in the co-expression of NKG2D, NKp44, and DNAM-1 was observed upon expansion with the three different protocols (Figure S1 in Supplementary Material, right panel).

To analyze the kinetics of the receptor expression induced by the different expansion protocols based on the use of IL-15 and/or additional IL-21, flow cytometry data were acquired at three time points, before stimulation, early at day 6, and late at day 18 (Figure S2 in Supplementary Material). Permanent presence of IL-21 diminished the expression of NKG2D, NKp30, and NKp44 on the NK cell surface. In contrast, it had hardly any influence, when applied after an IL-15-induced expansion phase. Expression of NKp46 was only slightly reduced by long-term IL-21 treatment. The higher expression level of DNAM-1, aroused by permanent presence of IL-21, stayed stable over time (Figure S2A in Supplementary Material).

Inhibitory receptors of the KIR family (KIR2D and CD158e/k/KIR3D) did not show any relevant changes during NK cell expansion, while NKG2A was strongly upregulated during NK cell expansion, independent from the cytokines used (Figure S2B in Supplementary Material).

Besides DNAM-1, expression of the adhesion molecules LFA-1 (CD11a) and 4-1BB (CD137) was addressed. Expression of LFA-1 was early increased under all tested cytokine conditions. In contrast, 4-1BB (CD137) was upregulated only after a longer culture period (Figure S2C in Supplementary Material). Accordingly, no prominent differences in conjugate formation were observed comparing the NK cells of all three tested cytokine expansion protocols (Figure S2D in Supplementary Material).

Besides an increase in the yield of of NK cells, another aim of the ex vivo expansion is to obtain a therapeutic cell product that mediates optimal anti-tumor activity. In case of NK cell immunotherapeutics, cytotoxicity is a valuable indicator of functionality. Accordingly, cytotoxicity of cytokine-expanded NK cells was tested against targets such as the erythroleukemia cell line K562 and the RMS cell lines RD and RH30. The RD cell line represents the embryonal subtype of RMS while RH30 represents the alveolar subtype, which is more difficult to treat.

A short boost with IL-21 increased the cytotoxic effect of NK cells toward all three cell lines compared to stimulation with IL-15 alone or permanent exposure to IL-21 (Figure 3). An increase in specific lysis of approximately 10% was achieved by an IL-21 boost for all three cell lines as compared to sole IL-15 stimulation.

Cytotoxicity of expanded NK cells decreased with longer culture periods and reduced from 60% on day 6 to 40% on day 10 and 14 (E:T = 10:1). Still cytotoxicity of continuously expanded cells was superior to cytotoxicity of NK cells that were cryoconserved after 6 days of cytokine expansion (Figure S3A in Supplementary Material).

Despite the decrease in CD16^pos NK cell numbers, no decrease of ADCC was observed with NK cells expanded with the IL-15 or IL-15 + 21_boost protocols compared to NK cells expanded with the IL-15 + 21 protocol in an assay against ErbB2^pos RD cells supplemented with anti-ErbB2 antibody Trastuzumab. All expansion protocols resulted in an additional lysis of about 5 to 10% by ADCC at an E:T ratio of 10:1 (Figure S3B in Supplementary Material).

To execute cytotoxicity upon activation by target cell recognition, one basic mechanism of NK cells is to degranulate and release cytolytic enzymes. In order to assess the impact of IL-21 to the cytolytic capacity and the underpinning mechanism, we stained for CD107a (lysosomal-associated membrane protein 1), which is located in the membranes of lytic granules and is expressed on the NK cell surface upon degranulation when lytic granules fuse with the outer membrane (47, 48).

Exposure to IL-21 during cultivation significantly increased the number of degranulating cells compared to IL-15 mono-treatment. This effect was rapidly induced after a short boost with IL-21. Permanent stimulation with IL-21 evoked similar results, suggesting that degranulation activity is maintained in the continuous presence of IL-21 (Figure 4A).

To further investigate if granule exocytosis is associated with cytolytic functionality, the expression of the pore-forming protein perforin and the serine protease granzyme B was addressed by immunoblotting. Indeed, expression and release of both proteins were induced strongly upon continuous stimulation with IL-21, but not subsequent to a short boost with IL-21 (Figures 4B–D). In particular, permanent presence of IL-21 in the culture provoked roughly a doubling of the relative expression of granzyme B and perforin in comparison to short-term or no IL-21 (Figures 4C,D).

In accordance with the slower induction of de novo production of granzyme B and perforin after short-term compared with continuous exposure to IL-21, apoptosis-inducing enzymes were released to a lesser extent despite similar degranulation levels were observed under both conditions (Figures 4A,B).

To further evaluate if this increase of degranulation was correlated with an increase of PI3K pathway signaling, we measured the phosphorylation of AKT and ERK1/2. Both were upregulated using IL-15 for expansion and even more with additional IL-21 (Figure S4 in Supplementary Material).

Natural killer cells can mediate direct cytotoxicity but also have an immunoregulatory function. The latter is achieved by secretion of cytokines. Hence, we analyzed the release of different cytokines after ex vivo expansion of NK cells using cytometric bead arrays (Figure 5).

Secretion of TNF-α was induced slightly by IL-15 and increased significantly upon stimulation with IL-21. Remarkably, after short-term exposure to IL-21, the TNF-α level was even higher in comparison with a continuous exposure to IL-21. A similar effect was observed for GM-CSF, although to a lesser extent. IFN-γ concentrations were significantly elevated with both protocols that utilized IL-21 causing a 10-fold increase in IFN-γ levels compared to IL-15 alone. Similarly, MIP-1α was secreted to a higher extent in the presence of IL-21 compared with IL-15 alone, independent of the duration of administration (Figure 5).

Levels of MCP-1 were similar for all three stimulation protocols, while IL-8 release was gradually more induced by IL-21 as compared to IL-15 alone and was further increased depending on the duration of IL-21 exposure. Also the levels of anti-inflammatory IL-10 were increased depending on the duration of IL-21 presence. In contrast, IL-15 application alone almost prevented secretion of IL-10 by NK cells (Figure 5).

Given that NK cells cultured with the IL-15 + 21_boost protocol exhibited the highest expansion rate and cytotoxic activity toward different target cells, we finally aimed to validate the in vivo effectiveness of NK cells expanded with the two-phase protocol in a RMS xenograft model. To this end, NOD/SCID/IL-2Rγc⁻ (NSG) mice were subcutaneously injected with the RD^GFP/Luc cell line and after establishing tumors, mice were locally irradiated by an image guided high precision local RT (Figure 6A). Subsequently, adoptive transfer of IL-15 + 21_boost continuously expanded NK cells was performed by three consecutive injections of 10⁷ NK cells. Tumor burden was monitored by caliper measurement (not shown) or BLI analysis of eight mice per group over 79 days (Figure 6B). Animals treated with RT monotherapy displayed a significant decrease in luciferase intensity indicating a growth retardation (Figures 6C,D). Following combined RT and NK cell therapy, the cytostatic effect was even more pronounced as compared to untreated or irradiated tumor-bearing controls (Figures 6B–E).