Primary NK cells were purified (> 95% pure) from peripheral blood mononuclear cells (PBMCs) of healthy human volunteers (n = 26; female n = 12, male n = 14). This research protocol was reviewed and approved by the institutional review board of CHA Bundang Medical Center, CHA University (permit number: CHAMC 2017–01–001). PBMCs were isolated from whole blood by a density gradient with Ficoll-Paque Plus (GE Health Care, Piscataway, NJ) followed by purification using an NK cell isolation kit (Miltenyi Biotec, Germany). Cells were activated in CellGenix® GMP Stem Cell Growth Medium (SCGM, CellGenix, Freiburg, Germany) supplemented with 10% human serum (Sigma-Aldrich, St. Louis, MO) and the cytokines IL-2 (10 or 100 ng/ml), IL-15 (10 ng/ml), IL-27 (10 ng/ml) (Peperotech, Inc. NJ) and IL-18 (10 ng/ml) (R&D Systems, Inc., MN) under standard culture conditions (a humidified 5% CO₂ atmosphere, 37 °C) for 21 days (long-term culture). Fresh culture medium containing cytokines and 10% human serum was added to the flask every 2 to 3 days for 21 days. The initial seeding density of NK cells was a median of 1 × 10⁶ (range: 0.8–1.3 × 10⁶) NK cells in 6-well plates (SPL Life Sciences, KOR). After 7 days, the cells were transferred into T25 flasks, T75 flasks, and T175 flasks (SPL Life Sciences, KOR) with additional media containing cytokines for 21 days.

Morphological appearances of NK cells were observed after 7 days for short-term, 14 days for mid-term, and 21 days for long-term cultures using phase contrast microscopy (Nikon, Tokyo, Japan). Cells were stained with May-Grüwald-Giemsa solutions. Images were analyzed using an ICC50 HD Camera System from Leica (Leica Microsystems).

The human tumor cell line K562 (human chronic myelogenous leukemia, CML) and A2780 (human ovarian cancer cell line) were obtained from the ATCC. Cells were cultured in RPMI 1640 medium (Gibco/Life Technologies, Carlsbad, CA) supplemented with 10% heat-inactivated FBS, 200 mM L-glutamine and 10,000 U/mL Pen/Strep (Gibco/Life Technologies, Carlsbad, CA) at 37 °C in a 5% CO₂ incubator. NK-92, the human NK cell line, was purchased from the ATCC and cultured in Alpha MEM devoid of ribonucleosides and or deoxyribonucleosides, 12.5% FBS, 10,000 U/mL Pen/Strep (Gibco/Life Technologies, Carlsbad, CA), 2 mM L-glutamine (Gibco/Life Technologies, Carlsbad, CA), 0.2 mM inositol (Sigma-Aldrich, St. Louis, MO), 0.02 mM folic acid (Sigma-Aldrich, St. Louis, MO), 0.1 mM 2-mercaptoethanol (Sigma-Aldrich, St. Louis, MO), and 100–200 U/ml recombinant IL-2 (Peperotech, Inc., NJ).

Day 21-expanded NK cells as well as freshly isolated PBMCs and NK-92 cells, were analyzed for phenotypic markers by flow cytometry. Cells were stained in Fluorescence-activated cell sorting (FACS) buffer on ice for 30 min with mixed antibodies. The following fluorescently labeled antibodies were purchased from BD Pharmingen™: anti-human CD3 (HIT3a), CD56 (B159), CD16 (3G8), CD314 (NKG2D; 1D11), CD335 (NKp46; 9E2), CD336 (NKp44; p44–8), CD337 (NKp30; p30–15), CD96 (6F9), CD226 (DNAM-1; DX11), CD19 (HIB19), CD14 (M5E2), KIR2DL1/CD158a (HP-3EA), KIR2DL2/3/CD158b (CH-L), and KIR3DL1/CD158e1 (DX9). The antibodies for human NKG2A/CD159a (#131411), KIR2DL4/CD159d (#181703), KIR3DL2/CD159k (#539304), and KIR3DL3/CD158z (#1136B) were purchased from R&D systems. The antibodies for human CD69 (CH/4) were purchased from Invitrogen, and antibodies for human KIR2DL5A/CD158f (UP-R1) were purchased from Origene. After washing with FACS buffer, cells were analyzed on a FACS Calibur machine (BD Biosciences). Data analysis was carried out using the FlowJo software program (Tree Star Inc., San Carlos, CA).

At each time point (7, 14 and 21 days), 500 μl of cell supernatants were collected and frozen at − 80 °C. Cytokine analysis was performed according to the manufacturer’s instructions using a Human IFN-ɣ Quantikine ELISA Kit (R&D Systems, Minneapolis, MN). Absorbance was read at 450/540 nm using a SpectraMax L Microplate Reader (Molecular Devices, Sunnyvale, CA). Concentration was calculated using the standard provided with the kits. Means and standard deviations of concentrations in triplicate samples were compared by t-test.

Calcein (calcein AM, AM; acetoxymethyl) is a fluorescent dye that can be used to determine cell viability in eukaryotic cells (C3099, Thermo Fisher Scientific). K562 target cells were used to demonstrate the NK cell-mediated cytotoxicity detection method using calcein AM. Target cells were stained by incubating them in 5 μM calcein AM staining media (1 mg/ ml stock) in complete RPMI media for 30 min at 37 °C in 5% CO₂. K562 tumor targets (T, loaded with calcein) and primary NK cells (E, effector cells) stimulated with IL-15/IL-18/IL-27 for 21 days were seeded together in a cell culture slide (8 chambered, SPL Life Sciences, Korea) at different E:T ratios (10:1, 5:1, 2.5:1, 1.25:1 and 0:1) in duplicate. Cells were cocultured at 37 °C for 4 h. After 4 h, live imaging of NK cell cytotoxicity on tumor targets was performed using a Zeiss LSM 510 microscope (Carl Zeiss Microscopy, LLC). The cytotoxicity % was calculated by the equation: cytotoxicity % = (dead target cell count / dead + live target cell count) × 100.

Cells were washed with cold PBS, and 200 μl of the cell suspension (1 × 10⁴ cells/ml) was added to a slide funnel and fixed in a cold methanol-acetone solution for 5 min. The slides were washed three times with PBS and then blocked with blocking solution (10% normal goat serum, Jackson ImmunoResearch Laboratories Inc., West Grove, PA) for 1 h. The cells were then washed with PBS and permeabilized in 0.1% Triton® X-100 (in PBS) for 20 min. Protein was detected using anti-perforin (1:100, Abcam) and anti-granzyme B (1:100, Abcam) antibodies for 24 h at 4 °C, followed by several washes in PBS. Incubation was repeated with an appropriate fluorescein isothiocyanate and Texas Red-conjugated secondary antibody (1:100, Jackson ImmunoResearch Laboratories Inc., West Grove, PA), and cell nuclei were stained with 4,6-diamidino-2-phenylindole. Coverslips were mounted onto slides with Vectashield Mounting Medium (Vector Laboratories, Inc., Burlingame, CA) and examined using a Zeiss LSM 510 confocal microscope (Carl Zeiss, Inc.).

Cells were lysed in ice-cold RIPA buffer (#9806, Cell Signaling Technology, Inc. Danvers, MA). Before isolation of total protein, NK cells were gently removed  by washing with 1 X cold PBS.  For Western blotting, anti-caspase-3, anti-caspase-8, and anti-caspase-9 antibodies from Cell Signaling; anti-perforin (CB5.4) and anti-granzyme B antibodies from Abcam; and anti-β actin antibodies from Santa Cruz Biotechnology were used. Secondary antibodies and ECL reagents were obtained from GE Healthcare. Signals were visualized by a G: BOX-CHEMI-XT4 Gel Documentation System.

The cytotoxicity of NK cells was determined using a CytoTox Glo™ Cytotoxicity Assay (Promega Corporation, Madison, WI) according to the manufacturer’s instructions. Briefly, K562 cells were seeded at a density of 5.0 × 10⁴ cells per well in 96-well plates (SPL Life Science, KOR). Then, 1.0 × 10⁶ expanded NK cells and NK-92 cells were resuspended in SCGM complete medium, and four serial dilutions (2-fold) were performed. Aliquots from each NK cell serial dilution containing 5.0 × 10⁵, 2.5 × 10⁵, 12.5 × 10⁵ and 6.25 × 10⁴ cells were added per well in a 96-well plate in duplicate. After 4 h of incubation, 50 μl of CytoTox Glo™ Cytotoxicity Assay reagent was added to all wells. This assay used a luminogenic peptide substrate to detect dead cells by selectively measuring ‘dead-cell protease activity.’ The luminescent signal that reflects cytotoxicity was measured using a SpectraMax L Microplate Reader (Molecular Devices, Sunnyvale, CA). Cytotoxicity was calculated by dividing the luminescent dead-cell signal by the total cell luminescence value.

A two-tailed paired t-test was performed to analyze data using GraphPad Prism 5.0 software (GraphPad Software Inc., San Diego, CA) as indicated in the figure legends. Statistical analysis was performed using a t-test adjusted with Benjamini and Hochberg procedure, and the ANOVA test adjusted with Bonferroni posttests. P values of less than 0.05 were considered statistically significant. The results are expressed as the mean ± SD and were obtained from two or three independent experiments.

In this study, we evaluated the characterization of NK cells derived from PBMCs from 7-, 14-, and 21-day cultures. First, we isolated peripheral blood NK cells from healthy donors (HDs). Twenty-six healthy adult donors enrolled in the study. Twelve donors were females with a mean age of 33.25 ± 5.429 years (n = 12, mean ± SD); the range was 20–40 years. Fourteen donors were males with a mean age of 35.07 ± 5.313 years (n = 14, mean ± SD); the range was 20–49 years (Additional file 1: Table S1).

In peripheral blood, human NK cells comprise approximately 10–15% of the total cells. The results from female donors show that the mean total viable PBMC count at presentation was 2.080 ± 0.871 × 10⁸ / 100 ml (mean ± SD, range: 1.070–4.0 × 10⁸ / 100 ml), including 9.34% (range: 5.374–13.931%) CD3-CD56^dim NK cells (Additional file 2: Figure S1). The mean total viable PBMC count from male donors was 1.869 ± 0.617 × 10⁸ / 100 ml (mean ± SD, range: 1.02–2.64 × 10⁸ /100 ml), including 12.62% (range: 6.786–23.137%) CD3-CD56^dim NK cells. Our data show that the NK cell populations did not significantly differ between the age groups of males and females. After purification of NK cells, we obtained 93.68 ± 4.30% CD3-CD56^dim NK cells. On day 21, NK cells stimulated with single cytokines or a combination were composed of highly enriched CD3 − CD56^dim (96.34 ± 2.41%) or CD56^dim CD16+ (93.10 ± 3.38%) NK cells (Additional file 2: Figure S2), with minimal contamination by CD3+ T cells (0.11 ± 0.21%), CD14+ monocytes (0.63 ± 0.32%) or CD19+ B cells (0.79 ± 0.54%).

To improve the ex vivo proliferation of human NK cells, we optimized the culture medium using different types of cytokines, such as IL-2, IL-15, IL-18, IL-27, and cytokine combinations. To test this approach, we purified CD3-CD56⁺ NK cells from PBMCs (median purity: 90.8%, range: 93.1–97.5%).

Purified NK cells were cultured with IL-2, IL-15 or cytokine combinations, such as IL-15/IL-18, IL-15/IL-27, IL-18/IL-27, and IL-15/IL-18/IL-27, for 7 days for short-term, 14 days for mid-term, and 21 days for long-term cultures (Fig. 1a).Fig1Fig. 1

Cytokine regulation of the proliferation and cytotoxicity of primary NK cells. a CD3-CD56+ NK cells (0.6–1.8 × 10⁶ / well) isolated from PBMCs were cultured in the indicated cytokines for 21 days. Primary NK cells were imaged using an inverted microscope and counted. Representative images of aggregates of growing primary NK cells in different culture conditions. Bars represent 500 μm; original magnification × 40. b The graph represents the total NK cell number of each group. Symbols indicate cytokine treatment groups (n = 3 / group): IL-2 (▲), IL-15 (▼), IL-15/18 (◆), IL-15/27 (○), IL-18/27 (□), and IL-15/18/27 (red). c. Fold expansion of NK cell numbers compared with those on day 0 following culture of CD3-CD56+ NK cells with the indicated cytokines. The graphs show the mean ± SD. * P < 0.05, ** P < 0.01, and *** P < 0.001, compared with day 0. d NK cell viability. The viable cell numbers were determined by trypan blue staining on days 7, 14 and 21. * P < 0.05, compared with day 14. e NK cytotoxicity assays of various cytokine-stimulated NK cells with K562 target cells on days 7, 14 and 21. The E:T ratios ranged from 0:1 to 10:1. After 4 h of incubation at 37 °C, the lysis of target cells was measured by ELISA. E:T indicates the effector-to-target ratio. The cytolytic activity of human NK cells stimulated with IL-15/18/27 toward K562 cells was significantly increased (*P < 0.05, **P < 0.01, *** P < 0.001) compared with that of resting NK cells (day 0) at the same E:T ratio. f Supernatants were analyzed for IFN-ɣ secretion by ELISA. The data presented are the mean ± SD of three separate experiments., *** P < 0.001, compared with IL-2 treated group

We examined the effect of combined treatment with IL-15, IL-18, and IL-27 on the expression of NK cell receptors. We investigated the expression pattern of the NK cell activating receptors CD16, NKG2D, and CD69, the natural cytotoxicity receptor (NCR) family, such as NKp44, NKp30, NKp46, DNAM-1, and CD69 (Fig. 2a), and inhibitory receptors [killer-cell immunoglobulin-like receptors (KIRs), such as KIR2DL1, KIR2DL2/3, KIR2DL4, KIR2DL5A, KIR3DL1/2/3, CD94/NKG2A and CD96] (Fig. 2b) on NK cells before and after cytokine-mediated NK cell expansion using flow cytometry and compared these values to those of NK-92 and NK-92MI cell lines (Additional file 2: Figure S4). As shown in Fig. 2a, after 21 days of culture with IL-15/IL-18/IL-27, the expression levels of most NCR family members, such as NKG2D, NKp44, NKp30, and CD69, were increased compared to those in resting NK cells. The expression of one of the coreceptors, CD226, was also increased after culture for 21 days. However, CD335 (NKp46) expression was decreased (Fig. 2c). The expression level of the most inhibitory receptors (Fig. 2d) did not change during cytokine treatment, except for the expression of KIR2DL2/3 and CD96, which was increased in expanded NK cells [KIR2DL2/3 (day 0: 20.75 ± 16.01; day 21: 44.60 ± 20.54) and CD96 (day 0: 4.16 ± 5.57; day 21: 47.63 ± 28.07)] (Additional file 1: Table S2). We also observed IL-2^Hi, IL-2 alone, and IL-15 alone groups (Additional file 1: Table S3). The expression of one of the NCRs, NKp30, was increased in the IL-2- and IL-15-only stimulated groups. Interestingly, KIR2DL2/3 expression was observed only in the stimulated IL-2^Hi or IL-2-only groups.Fig2Fig. 2

Flow cytometry analysis of NK cell receptors on human primary NK cells. (A) Cell surface expression of the indicated molecules on primary human NK cells on day 0 and day 21. Cells were stimulated with cytokine combinations (IL-15, IL-18, and IL-27) for 21 days. Primary NK cells from healthy donors were stained for expression of NK cell activating receptors (a) and inhibitory receptors (b), as indicated. Histograms show representative examples of NK cell receptor expression (shadow area) and show the percentage of NK cells positive for a given receptor relative to the isotype control (gray lines). NK cells were gated as viable, single, CD3-CD56+ cells. (c-d) Statistical analysis for the difference in NK cell receptor expression between day 0 and day 21. Significant differences are indicated in the graph as follows: *P < 0.05 and ** P < 0.01, compared with day 0. n.s: not significant

To investigate whether the morphological changes were the result of cytokine stimulation after 7, 14 and 21 days, we performed Wright-Giemsa staining. As shown in Fig. 3a, Giemsa staining revealed that IL-2, IL-15, IL-15/IL-18, IL-15/IL-27, and IL-15/IL-18/IL-27 affected morphological changes in CD3-CD56+ NK cells during culture, including increased size and cytoplasmic granularity. With the assumption that these cytoplasmic granules are cytolytic agents of NK cells, such as perforin and granzyme B, we analyzed the subcellular locations of these proteins using immunofluorescence staining (Fig. 3b).Fig3Fig. 3

Increased cytolytic granule accumulation in expanded NK cells. a The morphology of expanded NK cells. Cytospin preparations of NK cells (day 0, resting NK cells) and cytokine-activated NK cells (day 21) were stained by Wright-Giemsa staining. Representative images from cultures at 7, 14, and 21 days are shown. Original magnification × 400. b Immunofluorescence staining for perforin and granzyme B for NK cells before (day 0, upper panel) and after culture (day 21, lower panel) in the presence of a cytokine combination (IL-15/IL-18/IL-27). DAPI was used to stain the nuclei (blue). The negative control using secondary antibody (anti-IgG) only demonstrated low nonspecific binding of the secondary antibody. The data shown are representative of three independent experiments. Bars represent 10 μm. Original magnification × 100. c Western blotting analysis for perforin and granzyme B. NK cells were stimulated with cytokines (IL-15, IL-15/IL-18, and IL-15/IL-18/IL-27) for 21 days. The graph represents the relative expression of each protein. Protein bands were quantitated by densitometric analysis. The ratio of the intensity of protein bands relative to that of β-actin was calculated. Experiments were repeated three times with similar results. *P < 0.05, compared with cells cultured in the presence of IL-15 alone

Next, we investigated whether expanded NK cells could directly kill cancer cells, using a chamber slide as a coculture system. After 4 h of coculture with K562 and expanded NK cells, we performed a calcein release assay. Figure 4a shows representative bright-field (middle panel) and fluorescence images (upper panel) at each E: T ratio for the target cells. The number of brightly fluorescent live target cells gradually decreased with increasing number of expanded NK cells (E: T ratio), while all of the control target cells (K562 alone) were brightly fluorescent. At an E: T ratio of 5:1, nearly total lysis of K562 cells was observed. To derive live cell counts, the fluorescence intensity of the target cells was assessed at different E: T ratios so that target cells with lower fluorescence signals, such as those lacking calcein release and apoptotic bodies, were excluded from the live (bright) cell counts (Fig. 4a, b). In Fig. 4b, we demonstrated that K562 cells showed dramatic morphological changes typical of apoptotic death, such as blebbing and cell shrinkage, after 4 h of coculture with expanded NK cells.Fig4Fig. 4

Measurement of NK cell cytotoxicity by imaging cytometry. a K562 target cells were stained with calcein AM. After 4 h of incubation with 21-day-expanded NK cells (stimulated with IL-15/18/27), fluorescence images show progressive loss of fluorescence intensity of the K562 cells at various E:T ratios. Representative bright field, calcein, and overlay images showing E:T ratio-dependent target cell killing. Original magnification × 100. The graph represents the percentage of viable or dead target cells. *** P < 0.001 was considered signigicant. b A high power view of the calcein AM assay showing the progress of NK cell killing. Nearly all of the target cells were killed in a 10:1 effector-to-target cell sample, while calcein AM-labeled K562 cells were not killed in the control image. Bright-field and fluorescence overlay images of calcein show K562 cells undergoing apoptotic death following interaction with NK cells. The images were derived from a Zeiss LSM 510 microscope (left). The graph (right) represents cytotoxicity against K562 cells with expanded NK cells on days 7, 14 and 21 over the 4 h of the assay. *P < 0.05, ** P < 0.01, ***P <0.001 compared with day 0. Symbols indicate cytokine treatment groups (n = 3 / group): day 0 (▲), day 7 (▼), day 14 (◆), day 21 (red ○), and K562 cells only (●). c Immunoblot analysis for caspase-8, − 9 and − 3 activation. K562 cells were cocultured with primary NK cells for 4 h. Immunoblotting was performed with antibodies specific for caspase-8, − 9 and − 3 and their cleaved forms. β-actin was used as an internal standard. d Protein bands were quantitated by densitometric analysis. The ratio of the intensity of protein bands relative to that of β-actin was calculated. Bar graph represents the relative expression of cleaved caspase-8, − 9 and − 3 proteins. Experiments were repeated three times with similar results