Interleukin-21 differentially affects human natural killer cell subsets


Dr R. Jacobs, Department of Clinical Immunology, Hannover Medical School, Carl-Neuberg-Strasse 1, 30625 Hannover, Germany.
Senior author: Dr Roland Jacobs


Interleukin-21 (IL-21) is a cytokine with pleiotropic effects on various cell types including dendritic cells, B cells, T cells and natural killer (NK) cells. To evaluate if IL-21 affects human NK cell subpopulations in a similar fashion, functional studies were performed on CD56dim and CD56bright NK cells, both bearing IL-21 receptors at identical densities. Stimulation with IL-21 strongly induced proliferation of CD56bright NK cells and cytotoxicity against K562 target cells was preferentially augmented in CD56dim NK cells. In contrast, stimulation with IL-2 and IL-21 alone or in combination failed to induce interferon-γ and tumour necrosis factor-α production in the two NK cell subsets. Intracellular analysis of signal transducer and activator of transcription (STAT) proteins revealed that IL-21 by itself induces phosphorylation of STAT1 and STAT3 in CD56dim NK cells, and to an even higher degree in CD56bright NK cells. In this CD56bright NK cell population alone, IL-2 weakly phosphorylated STAT1 and STAT3, which was further increased when cells were treated with the combination of both cytokines. In contrast, STAT5 was strongly phosphorylated only in CD56bright NK cells by low-dose IL-2, while IL-21 did not affect STAT5 at all. In summary, we present data indicating that the NK-cell-directed cytokines IL-2 and IL-21 not only affect functions in NK cell subpopulations differently but can also act additively.


Natural killer (NK) cells are important effectors of innate immunity because they are capable of lysing tumour or virus-infected cells without previous sensitization.1 In addition, they are able to produce various cytokines and chemokines like interferon-γ (IFN-γ), tumour necrosis factor-α (TNF-α), interleukin-10 (IL-10), macrophage inflammatory protein-1α, and granulocyte–macrophage colony-stimulating factor, and thereby also affect adaptive immune responses and haematopoiesis.2–6 In humans, NK cells are characterized by the expression of CD56 and a lack of CD3. The majority of NK cells coexpress FcγRIII (CD16). According to the expression density of CD56 two subsets can be distinguished, comprising about 90% CD56dim and 10% CD56bright NK cells in peripheral blood. While the first population represents NK cells with high cytotoxic potential expressing killer cell immunoglobulin-like receptors (KIRs) and CD16, the latter appear to be both CD16low/neg KIRneg. Although they secrete various cytokines/chemokines they exhibit only low cytotoxic capacity.7,8 In lymph nodes and tonsils an inverted ratio between CD56dim and CD56bright has been detected. In these tissues, CD56bright NK cells are located in T-cell areas where they can interact with T cells and dendritic cells.9

The functions of NK cells are regulated by balancing the activating and inhibitory signals provided by the respective receptors, which have been shown to be differently expressed by NK cell subsets. In contrast to KIRs, which are almost exclusively present on CD56dim NK cells, lectin-like receptors are expressed by both subsets, indicating that CD56bright NK cells are regulated mainly by this receptor family.7,8,10 The activating natural cytotoxicity receptors NKp30, NKp44 and NKp46 are expressed in variable amounts in both populations. Cellular ligands of natural cytotoxicity receptors are mostly unknown but NKp44 and NKp46 recognize the haemagglutinin of influenza and sendai viruses and NKp30 binds to the pp65 protein of human cytomegalovirus.11–13 Recently, a novel NKp44 ligand (L) has been identified that is induced after infection on CD4+ T cells by a motif of the human immunodeficiency virus type 1 (HIV-1) envelope gp41 protein. NKp44L is suggested to play a key role in the depletion of helper T cells by activated NK cells in HIV infection.14 NKG2D is an activating lectin-like receptor on CD56dim and CD56bright NK cells, which recognizes molecules such as MICA/B and ULBP, which are expressed by stressed cells.

In addition to the receptors mentioned above, various soluble factors modulate the functional capabilities and development of NK cells. In particular, IL-2, IL-12 and IL-15 affect NK biology. Interleukin-21 is a cytokine with pleiotropic effects on various cell types.15 The corresponding IL-21 receptor (R) is expressed on B cells, T cells, dendritic cells and equally on CD56dim and CD56bright NK cells.8 Signalling of IL-21R is mediated via the common γ-chain (CD132), which is also a component of IL-2R, IL-7R, and IL-15R.16 Functionally, IL-21 is involved in the inhibition of class switching from immunoglobulin G (IgG) to IgE in B cells, thus playing a role in autoimmune diseases like allergic asthma.17 In NK cells, IL-21 enhances cytotoxicity in vivo and in vitro.18 In the presence of IL-15 or IL-18, IL-21 is able to induce the production of IL-10 and IFN-γ.19,20 A recent study demonstrated the development of CD56dim NK cells from CD34+ haematopoietic stem cells in the presence of flt3L, stem cell factor, IL-15 and IL-21. In contrast, without IL-21, the generation of CD56bright NK cells lacking CD16 and KIRs is favoured.21

To evaluate the effects on both NK cell subpopulations, we analysed the response of sorted CD56dim and CD56bright NK cells to IL-21 in isolation or in combination with low doses of IL-2, serving only as costimulating factor. The experiments revealed that IL-21 differentially affects the regulation of activation, cytotoxic capability, and proliferation of NK cells. In parallel, signalling via IL-21R and IL-2R was traced by intracellular staining of phosphorylated STAT proteins using flow cytometry. This study demonstrates that CD56dim and CD56bright NK cells favour different signalling pathways initiated by stimulation with IL-21 and/or IL-2.

Materials and methods

Isolation of lymphocytes

Peripheral blood mononuclear cells (PBMC) were isolated from leucocyte filters obtained from the Institute of Blood Transfusion at Hannover Medical School. The filters were disconnected from blood-bag systems from healthy donors who gave their informed consent. The approval of the local ethical committee was received for this study. The filters were flushed with phosphate-buffered saline (PBS) against original blood flow and the PBMCs were isolated by Ficoll–Hypaque centrifugation and fluorescence-activated cell sorting (FACS) analysis was performed using CD3, CD20 and CD56 monoclonal antibodies (mAbs; BD Pharmingen, Heidelberg, Germany). Donor PBMCs were depleted from CD3+, CD19+, CD14+ and CD15+ cells using mAb coupled to magnetic beads (Dynal, Hamburg, Germany). The prepurified cells were then stained for CD3 and CD56 and sorted with a FACStarplus cell sorter (Becton Dickinson, Heidelberg, Germany). Purity of sorted CD3 CD56+ NK cells was re-analysed using a FACSCalibur (Becton Dickinson).

Antibodies used for flow cytometry and cell depletion

Cell surface markers were analysed using the following mAbs: CD56 conjugated with fluorescein isothiocyanate (CD56-FITC) (NCAM1.2), CD56 conjugated with phycoerythrin (CD56-PE) (My31), CD56 conjugated with allophycocyanin (CD56-APC) (B159), CD3-FITC (UCHT1), CD3-APC (UCHT1), CD20-FITC (2H7), CD69-PE (L78), pSTAT1 conjugated with ALEXAFluor 647 (4a), pSTAT3-ALEXAFluor 647 (4/P-Stat3), pSTAT5-ALEXAFluor 647 (47), granzyme A-PE (CB9), perforin-FITC (δG9), CD94-FITC (HP-3D9), CD158a-FITC (EB6), CD158b-FITC (GL183), and TNF-α-FITC (MAb11) (BD Pharmingen, Heidelberg, Germany); NKp30-PE (Z25), NKp44-PE (Z231), NKp46-PE (BAB281), CD11a-FITC (25·3), and CD18-FITC (7E4) (Coulter Immunotech, Hamburg, Germany); IL-21R-PE (#152512) (R & D Systems, Wiesbaden, Germany); granzyme K-FITC (24C3) (Immunotools, Friesoythe, Germany); IFN-γ-ALEXAFluor 488 (B27) (Caltag, Hamburg, Germany); and CD25-FITC (ACT-1) (DAKO, Hamburg, Germany). Phenotyping was performed using 3 × 105 cells/100 μl in 96-well, U-bottomed plates. Each well was supplemented with 10 μl human IgG (Intraglobin F, Biotest, Dreieich, Germany) to prevent non-specific binding via Fc receptors. Antibodies were added at the recommended concentrations and incubated for 15 min at 4°. The cells were washed three times with PBS/0·1% bovine serum albumin (BSA; Sigma-Aldrich, Seelze, Germany) and analysed using a FACSCalibur cytometer. Labelled isotype-matched antibodies were run in parallel as controls.

Unlabelled antibodies used for enrichment of NK cells were: CD14 and CD19 (Immunotools), CD15 (BD Pharmingen), and CD3, which was purified from OKT3 hybridoma supernatants in the Department of Clinical Immunology, Hannover Medical School. Magnetic beads M450 (Dynal) were coupled to 10 μg antibody/108 beads. For depletion, a concentration of approximately two beads was calculated per target cell. After 15 min of incubation, the tube was placed into a magnetic field that concentrated labelled cells towards the wall of the tube. Enriched NK cells in the supernatant were transferred into a second tube for subsequent analyses.

Intracellular staining

Analysis of intracellular granzyme A, granzyme K and perforin was preceded by surface staining 3 × 105 cell/50 μl PBS/BSA using CD56 and CD3 antibodies at 4°. After 30 min, cells were washed twice and resuspended in PBS supplemented with 4% paraformaldehyde at room temperature. Ten minutes later, cells were washed and 50 μl saponin buffer (PBS supplemented with 0·1% saponin and 0·01 m HEPES) was pipetted before adding antibodies against the intracellular proteins granzyme A, granzyme K and perforin. After 30 min incubation at 4°, the cells were washed three times and analysed by flow cytometry.

Before the intracellular staining for production of IFN-γ and TNF-α, cells were stimulated with cytokines as described below and brefeldin A (Sigma-Aldrich, Munich, Germany) was added after 1 hr of culture in a humidified incubator at 37° to inhibit secretion and achieve accumulation of the cytokines produced. After an additional 23 hr of incubation, cells were surface stained as described above with CD56 and CD3 antibodies. After fixation with 4% paraformaldehyde (Merck, Darmstadt, Germany), cells were perforated with 0·1% saponin (Riedel-de Haen, Seelze, Germany) in PBS and antibodies directed against the appropriate cytokines were added. After 20 min incubation and three washes, cells were analysed as described above.

Stimulation with cytokines

Freshly isolated PBMCs (3 × 106/ml) or highly purified (> 98%) CD56dim and CD56bright NK cells (2·5 × 105/ml) were resuspended in RPMI-1640 medium supplemented with 10% fetal calf serum (FCS), 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mm l-glutamine, and 1 mm sodium pyruvate (R10 medium). Cells were then activated with 50 U/ml IL-2 (Eurocetus, Amsterdam, the Netherlands) and/or 20 ng/ml IL-21 (R & D Systems, Wiesbaden, Germany) for 20 hr. Activation markers CD25 and CD69 were determined by FACS analysis.

Proliferation assay

Triplicates of sorted CD56dim and CD56bright NK cells (2·5 × 105/ml) were incubated in R10 medium supplemented with 50 U/ml IL-2 and/or 20 ng/ml IL-21. After 48 hr, cultures were pulsed with 0·4 μCi [3H]thymidine (Amersham Biosciences, Braunschweig, Germany) and incubated for another 24 hr. After harvesting, incorporated DNA was measured in a beta-counter (Perkin Elmer, Rodgau, Germany).

Cytotoxicity assay

Sorted CD56dim and CD56bright NK cells (1·5 × 105/ml) were activated for 20 hr with 50 U/ml IL-2 and/or 20 ng/ml IL-21. A standard 4-hr 51Cr-release (Amersham Biosciences) assay was performed against K562 target cells as described elsewhere.22 Effector : target (E : T) ratios of 5 : 1, 2·5 : 1 and 1 : 1 were used and lytic units (LU20/107 cells) of the assays were calculated according to the methods established by Bryant et al.23

Conjugate-forming assay

Freshly isolated PBMCs (3 × 106/ml) were incubated with 50 U/ml IL-2 and/or 20 ng/ml IL-21 for 20 hr. Activated cells were stained with CD3 PE and CD56 APC monoclonal antibodies. After two washes, the cells were again adjusted to 3 × 106/ml and 100 μl was added to 200 μl washed K562 cells (1 × 106/ml). The cell suspension was lightly centrifuged at 100 g for 3 min and incubated for 15 min at 37° in a humidified atmosphere. Cells were then gently mixed and analysed with a FACS by gating on PBMCs and K562 cells, excluding CD3+ T cells. K562 target cells display a discrete autofluorescence signal in FL1, thus conjugates can be identified as cell clusters fluorescing green (autofluorescence) and blue (APC) simultaneously. By applying appropriate gates the percentage of CD56bright or CD56dim cells forming conjugates was assessed.

FACS analysis of phosphorylated STAT proteins

PBMCs (3 × 106/ml) of seven different donors were stimulated with IL-2 and/or IL-21 as indicated above. After 1 hr, equal volumes of 4% paraformaldehyde (Merck, Darmstadt, Germany) were added and incubated for an additional 10 min at 37°. Cells were pelleted and 1 ml of ice-cold 90% methanol (Mallinckrodt Baker B.V., Deventer, the Netherlands) was added. After 30 min incubation on ice, cells were washed twice with PBS (Biochrom, Berlin, Germany)/1% FCS (PAA, Linz, Austria) and resuspended in 100 μl PBS/FCS. Stainings of pSTAT proteins and surface markers were performed for 1 hr at room temperature until the cells were washed and analysed by flow cytometry. Phenotypic data were processed using Cell Quest Pro (Becton Dickinson) or Summit 4·0 (Dako Cytomation, Hamburg, Germany) software.


Statistical analyses were performed as t-tests using GraphPad Prism V4·03 software. Levels of significance are given as P-values (*P < 0·05, **P < 0·01, ***P < 0·001).


IL-21R surface expression on NK cell subsets

CD56dim and CD56bright NK cells express several surface molecules, which can mediate stimulating signals. To compare the response patterns of NK cell subsets towards activation with ‘physiological’ agents, we screened for suitable receptors, which are expressed to a similar degree on CD56dim and CD56bright NK cells. Gene array analysis revealed equal levels of IL-21R mRNA in both NK cell subsets, thus matching this criterion (Fig. 1a). Expression data are publicly accessible at EMBO-EBI ArrayExpress database (URL: under accession number: E-MEXP-380. After activation with phorbol 12-myristate 13-acetate (PMA)/ionomycin the expression of this transcript was marginally increased but levels were still comparable in both subsets.8 In addition, FACS analysis of resting and activated PBMCs confirmed low density of IL-21R on both NK cell subsets (Fig. 1b,c).

Figure 1.

 Expression of IL-21R on NK cell subsets. (a) Affymetrix array analysis revealed positive signals for IL-21R-specific mRNA in resting and activated CD56dim (open blocks) and CD56bright (solid blocks) NK cells using Probe Set 219971_at. Expression data are publicly accessible at EMBO-EBI ArrayExpress database (URL: under accession number: E-MEXP-380. Flow cytometry revealed a low surface expression of IL-21R on NK cells. Histograms of CD56dim (b) and CD56bright (c) subsets document the expression of IL-21R on resting (grey-filled) and activated (solid line) NK cells as compared to the isotype controls (dotted line).

IL-21 induces differential regulation of CD69 and CD25 in NK cell subsets

To assess if signalling via IL-21R is comparable between CD56dim and CD56bright NK cells, PBMCs were stimulated either with IL-21 alone or with IL-21 in combination with IL-2 in suboptimal doses. The optimal concentration of IL-21 for the assays was predetermined by a dilution series (data not shown). Activation markers CD69 and CD25 (IL-2R α-chain) were analysed after 24 hr stimulation with the respective stimuli (Fig. 2). A strong up-regulation of CD69 was found on CD56dim NK cells in the presence of IL-21 alone as compared to medium control (P < 0·01). There was no further enhancement when costimulated with IL-2. CD56bright NK cells exhibited a slight but significant up-regulation of CD69 only when activated with both cytokines simultaneously (P < 0·01) (Fig. 2a).

Figure 2.

 Regulation of activation markers by IL-21 and/or IL-2. PBMCs of healthy donors (n = 7) were stimulated with IL-21 and/or IL-2 and activation markers CD69 (a) and CD25 (b) were analysed using flow cytometry. The percentage of receptor-bearing CD56dim (open blocks) and CD56bright (solid blocks) NK cells are shown. Solid connecting lines represent significant changes as compared to a medium control, whereas dotted lines depict significant differences between CD56dim and CD56bright NK cells following activation with the respective stimulus. For comparisons between the two populations, values were adjusted for medium levels ***P < 0·001; **P < 0·01; *P < 0·05; ± SEM.

In contrast, CD25 was down-regulated on CD56dim and CD56bright NK cells in the presence of IL-2 alone, but up-regulated after activation with IL-21 (Fig. 2b).

Other NK-cell-specific markers were investigated in parallel under the same conditions but no changes in the expression of IL-21R, KIRs (CD158a, CD158b), natural cytotoxicity receptors (NKp30, NKp44, NKp46), CD94, CD16 or CD56 were detected, neither after 24 hr nor after 7 days of culture. Adhesion molecules (e.g. CD11a and CD18) and cytolytic molecules (granzyme A, granzyme K, and perforin) also remained unaffected by the cytokine treatment (data not shown).

Production of IFN-γ and TNF-α

In all experiments CD56dim and CD56bright NK cells failed to produce IFN-γ and TNF-α. In contrast to PMA/ionomycin, which was used for control, IL-2, IL-21, or the combination of both cytokines, could not induce cytokine production. This was further confirmed by quantitative polymerase chain reaction (iCycler, Bio-Rad, Hercules, CA) using IFN-γ-specific and TNFα-specific primers and sorted CD56dim and CD56bright NK cells, which were cultivated for 24 hr with IL-2 and/or IL-21 in the absence of the transport inhibitor brefeldin A (data not shown).

IL-21-dependent proliferation of NK cells

CD56dim and CD56bright NK cells were sorted using FACStarPlus with high purity (> 98%) and proved to be free of any contamination of the respective other NK cell subset. After 72 hr incubation with IL-2 and/or IL-21, proliferation was determined by [3H]thymidine assay (Fig. 3). CD56bright NK cells displayed a strong proliferative response towards IL-2 that was further increased by coincubation with IL-21 (P < 0·05). IL-21 alone failed to induce proliferation of either NK cell subset. CD56dim NK cells that were treated the same way exerted only poor proliferation but the effects of combined IL-2 and IL-21 were comparable to results of the CD56bright subset, although on a lower level.

Figure 3.

 Patterns of IL-21- and/or IL-2-induced proliferation of NK cell subsets. Proliferation of sorted NK cell subsets in response to activation with IL-21 and/or IL-2 was investigated by [3H]thymidine assay. Mean values of seven assays ± SEM are shown. Asterisks above solid lines represent significant increases as compared to the medium control. Asterisks above dotted lines illustrate significant differences between CD56dim (open blocks) and CD56bright (solid blocks) NK cells activated with the same mediator. For comparisons between the two populations values were adjusted for medium levels. ***P < 0·001; **P < 0·01; *P < 0·05; ± SEM.

Cytotoxicity of CD56dim NK cells is strongly up-regulated by IL-21

The cytotoxic potential of highly purified NK cell subsets was assessed after 24 hr stimulation with IL-2 and/or IL-21 (Fig. 4a). As expected, CD56dim NK cells exerted stronger spontaneous cytotoxicity against K562 target cells in comparison to CD56bright NK cells. IL-21 was able to slightly increase cytotoxicity of CD56dim and CD56bright NK cells, whereas a low dose of IL-2 was only effective in CD56bright NK cells in this respect. In combination, IL-21 and IL-2 caused a more than doubling of the lytic units of both subsets as compared to the respective medium controls. Again, the effect of the cytokines was similar in both subsets but CD56bright NK cells retained a lower cytotoxic potential under all conditions.

Figure 4.

 IL-21 and/or IL-2 enhance cytotoxicity and conjugate formation. (a) Sorted NK cell subsets were activated with IL-21 and/or IL-2 for 20 hr and cytotoxicity against K562 was determined in standard 4-hr 51Cr-release assays; lytic units (LU20/107 cells) are depicted. (b) The impact of IL-21 and/or IL-2 on conjugate forming ability of CD56dim (open blocks) and CD56bright (solid blocks) NK cells complexed to target cells was measured using flow cytometry and gating on respective NK subsets. Mean values of NK cells bound to targets are given in percentage ± SEM (n = 6).

A prerequisite for effective cytolysis, and thus a possible regulatory element, is the binding of effector cells to respective targets by forming conjugates. To analyse whether or not the observed regulation of cytotoxic capacity is mediated by changes in the ability to form conjugates, the interaction of effector and target cells was explored by FACS analysis. Flow cytometry revealed that the formation of conjugates was modified by the treatment, although expression of adhesion molecules was not regulated (data not shown). Both cytokines enhanced conjugate formation and displayed a cumulative effect when administered together (Fig. 4b). The percentage of CD56dim NK cells forming conjugates with target cells was marginally higher compared to the CD56bright subset.

IL-21 preferentially activates STAT3 in human NK cells

The limited number of CD56bright NK cells in peripheral blood made comparative analyses of signal transduction pathways using Western blots not feasible. Therefore, signalling molecules were measured intracellularly using flow cytometry. The PBMCs activated with IL-2 and/or IL-21 were analysed for phosphorylation of STAT1, STAT3 and STAT5. As positive controls, cells were stimulated with well-established STAT inducers: IFN-γ (10 ng/ml) for STAT1, IL-6 (5 ng/ml) for STAT3, and IL-2 (1000 U/ml) for STAT5. Kinetic experiments revealed that 1 hr of stimulation was optimal for the analysis of STAT phosphorylation in all cell populations (data not shown). IFN-γ induced phosphorylation of STAT1 as assessed by gating on monocytes according to forward scatter versus side scatter properties of the cells (Fig. 5a). For analyses of pSTAT3 and pSTAT5, cells were gated on lymphocytes revealing a slight induction of pSTAT3 by IL-6 (Fig. 5b) and a strong induction of pSTAT5 by IL-2 (Fig. 5c) within 1 hr.

Figure 5.

 Intracellular staining of pSTAT proteins. PBMCs were stimulated with cytokines as indicated and analysed for phosphorylation of STAT1 (a), STAT3 (b) and STAT5 (c) by flow cytometry. Grey-filled histograms represent medium controls and open histograms represent the respective stimulus.

Activation of PBMCs with IL-21 induced a significant increase of STAT1 phosphorylation in CD56dim and CD56bright NK cells when compared with medium control (P < 0·01) (Fig. 6a,d). In CD56bright NK cells, phosphorylation was marginally further enhanced when stimulated by both cytokines. IL-2 selectively triggered the phosphorylation of STAT1 in CD56bright but not in CD56dim NK cells. Under all three conditions the enhancement of STAT1 phosphorylation was significantly higher in CD56bright NK cells.

Figure 6.

FACS analyses of phosphorylated STATs in NK cell populations. PBMCs were stimulated with cytokines as indicated (1: IL-2, 2: IL-21, 3: IL-2 + IL-21, filled grey area: medium control) and analysed for phosphorylation of STAT1 (a), STAT3 (b), and STAT5 (c) by gating on CD56dim (left panel) and CD56bright (right panel) NK cells. Bar diagrams depict MFIs of pSTAT1 (d), pSTAT3 (e), and pSTAT5 (f) as assessed for CD56dim (open blocks) and CD56bright (solid blocks) NK cells (n = 7). Asterisks above solid connecting lines represent significant enhancement compared to medium control. Asterisks above dotted lines illustrate significant differences between CD56dim and CD56bright NK cells activated with the same stimulus. For comparisons between the two populations values were adjusted for medium levels; ***P < 0·001; **P < 0·01; *P < 0·05; ± SEM.

Phosphorylation of STAT3 was particularly induced by IL-21 and costimulation with both cytokines, as assessed from the mean fluorescence intensity (MFI) in CD56dim and CD56bright NK cells (Fig. 6b,e). Compared to the respective medium controls, the increases reached levels of significance in both NK cell subsets (P < 0·001). Costimulation with IL-2 and IL-21 revealed stronger phosphorylation in CD56bright NK cells compared to the CD56dim subset (P < 0·01). STAT3 was also phosphorylated in CD56bright when PBMCs were stimulated with IL-2 alone (P < 0·01). In contrast, STAT3 remained unaffected in IL-2-stimulated CD56dim NK cells.

Activation with IL-2 induced a moderate increase of STAT5 phosphorylation in CD56dim NK cells and a considerably more pronounced increase in CD56bright NK cells (Fig. 6c,f).

These analyses demonstrate a preferential phosphorylation of STAT3 by IL-21. The same stimulus failed to induce any phosphorylation of STAT5 and affected STAT1 only marginally. In contrast, IL-2 activated mainly the STAT5 pathway with stronger induction of STAT5 phosphorylation in CD56bright NK cells as compared to the CD56dim NK subset.


To analyse whether CD56dim and CD56bright NK cell populations are differentially regulated upon stimulation with the same stimulus, we screened CD56dim and CD56bright NK cells for receptors susceptible to physiological activation in both subsets. Receptors like CD16, CD2 or the IL-2 receptor complex, which are known to activate NK cells, were excluded because different results were expected as the result of differential surface expression between the two subsets.24 Thus, we screened recent gene array experiments of resting and activated CD56dim and CD56bright NK cells, respectively.8 The query revealed equal expression of IL-21R mRNA both in resting and activated CD56dim and CD56bright NK cells. FACS analyses confirmed similar expression patterns of IL-21R on the surface of both NK cell subsets. This prompted us to explore the impact of IL-21 on NK cell subpopulations. Since IL-2 is a well-known potent activator of NK cells, we also investigated whether low doses of this cytokine can modulate IL-21-induced response patterns of CD56dim and CD56bright NK cells.

At first, PBMCs were activated with IL-21 alone or in combination with IL-2, which was used in suboptimal doses and served only as a costimulating agent. After 24 hr of stimulation, expression analysis of activation markers CD69 and CD25 revealed different regulation in the two NK cell subsets. CD56dim NK cells strongly up-regulated CD69 in the presence of IL-21 alone or in combination with IL-2, while CD56bright NK cells displayed only marginal enhancement of this marker when both cytokines were present in culture. In contrast, CD25 was preferentially enhanced on CD56bright NK cells after stimulation with IL-21 alone. IL-2 induced a decrease of CD25, especially on the CD56bright NK cell subset. This could be the result of internalization of this receptor chain, which might serve as a regulatory feedback mechanism.25,26

Functionally, we investigated cytokine production (IFN-γ and TNF-α), proliferation and cytotoxicity of highly purified NK cell subsets after stimulation with IL-21 and IL-2. The physiological stimuli affected NK-cell-mediated functions differently. The cytokines IL-2 and IL-21 alone or in combination were insufficient to induce IFN-γ and TNF-α in the two NK cell subsets. Since intracellular staining requires the addition of a transport inhibitor, which could have negatively affected the mild stimulation regimen, we also analysed sorted CD56dim and CD56bright NK cells on the mRNA level. In this case, cells were stimulated with IL-2 and/or IL-21 in the absence of any transport inhibitor. The lack of any specific mRNA as assessed by quantitative PCR confirmed that IL-2 and/or IL-21 were not able to induce IFN-γ and TNF-α, which might be because of the stimulation protocol using low concentrations.

In contrast to cytokine production, other functional capacities of NK cells were affected by stimulation with IL-2 and/or IL-21. CD56bright NK cells exhibited proliferative responses in the presence of low-dose IL-2, probably because of expression of the high-affinity IL-2 receptor complex. The response was further enhanced after incubation with IL-21. This could be caused by the observed positive effect of IL-21 on CD25 expression, thus increasing the number of high-affinity IL-2 receptor complexes. CD56dim NK cells exhibited a low proliferative response towards IL-21 and IL-2 but exerted increased cytotoxicity when incubated with IL-21 alone. This is in line with a study demonstrating that the cytokine augments the cytotoxicity of human NK cells in vitro.18 In addition, we observed an additive effect when CD56dim NK cells were stimulated with IL-2 and IL-21 in combination. Under the same conditions, cytotoxicity of CD56bright NK cells was only marginally up-regulated.

To unravel the underlying mechanisms of the enhanced cytotoxicity that was mainly observed in CD56dim NK cells, we investigated some contributing factors, like the expression of cytolytic molecules, but we did not find any regulation of granzyme A, granzyme K and perforin in NK cells after activation with IL-21 and/or IL-2. Thus regulation of cytolytic substances does not contribute to enhanced cytotoxicity induced by IL-21.

In another approach to explain the increased cytotoxicity we performed conjugate forming assays, revealing that particularly for CD56dim NK cells there was an enhanced capacity to form conjugates with K562 target cells. The more effective binding of targets might contribute to increased cytotoxicity. Conjugates are formed and stabilized via the interaction of adhesion molecules and respective ligands. Therefore, the expression of some potential candidates was analysed. However, we could not detect any changes in density of the adhesion molecules CD11a/CD18 (LFA-1), CD2, or CD58, which are known to be involved in conjugate formation between target and effector cells. This does not, however, definitively exclude a regulation of target binding by these adhesion molecules. For example, it is known that conformational changes in LFA-1 and not its increased surface expression, cause the enhancement of conjugate-forming ability.27,28 Moreover, we demonstrated up-regulation of CD69 on CD56dim NK cells. This molecule also mediates binding to target cells and may be responsible for the effects observed in this study.29

In human peripheral blood the CD56bright NK cells represent a small population, comprising only about 1% of all lymphocytes. Thus comparative Western blot analyses of signal transduction pathways used by CD56dim and CD56bright NK cells were not feasible. Therefore, we investigated phosphorylation of three signalling molecules (STAT1, STAT3, STAT5) after activation with IL-21 and/or IL-2 by intracellular staining and flow cytometry. Positive control cells were stimulated with IFN-γ (STAT1), IL-6 (STAT3), and IL-2 (STAT5), which have been described to phosphorylate the respective molecules given in brackets.30

Interluekin-21 alone and in combination with IL-2 induced the phosphorylation of STAT1, which among other factors regulates the natural cytotoxicity but not the ADCC of NK cells and may therefore play a role in the enhanced cytotoxic potential observed in this study.31

Furthermore, STAT3 was strongly phosphorylated by IL-21, as well as the combination of IL-21 and IL-2. An additive effect of both cytokines was found in CD56bright NK cells. STAT3 is involved in the regulation of proliferation. Its activation may explain the enhanced proliferative response, which was detected in both NK cell subsets.32,33 STAT5 was strongly phosphorylated in CD56bright in the presence of low dose IL-2. IL-21 did not further affect STAT5 phosphorylation. This response is possibly mediated by the high-affinity IL-2R, consisting of CD25, CD122 and CD132, which is only expressed on CD56bright but not CD56dim NK cells. The latter subset displayed marginal phosphorylation of STAT5 after activation with 50 U/ml IL-2. Similar to STAT3, STAT5 is also known to regulate proliferation and both STATs together may be responsible for the strong proliferative response observed for CD56bright NK cells.32,33

To fulfil their functions, STAT proteins must form hetero- or homodimers, which are essential for their translocation into the nucleus and their binding to respective DNA motifs. Which STAT monomers interact with each other and the exact properties of the complexes are not yet completely understood. We argue that IL-21 and/or IL-2, respectively, mediate the formation of particular dimers, which are responsible for differential regulation of the functions determined in CD56dim and CD56bright NK cells.

In summary, this study revealed that IL-21 alone or in combination with IL-2 differentially affects function and signal transduction in CD56dim and CD56bright NK cells. Sources of these cytokines are mainly T cells, supporting the view that T and NK cells communicate via these factors. The additive effect mediated by IL-21 and low-dose IL-2 may be of great interest for clinical applications. In several mouse models IL-21 has been shown to elicit a strong antitumour effect in T and NK cells against various malignancies.34–37 On the other hand, high-dose IL-2 administration is effective against metastatic melanoma and renal cell carcinoma, and has also been tried for supporting therapeutic regimens in HIV patients. Treatment is often accompanied by complications and severity of side-effects of IL-2 varies with the dose, route and schedule of administration.38,39 Based on our results, coadministration of IL-21 and reduced doses of IL-2 seem to be feasible for inducing strong immunological activation with diminished side effects. This appears to be even more attractive as a mouse model with combination therapy with IL-21 and low-dose IL-2 resulted in long-term tumour-free survival by improving the function of tumour-antigen-specific CD8+ T cells.40 Another study demonstrated that IL-21-stimulated NK cells mediate increased cytotoxicity against antibody-coated breast cancer cells and drive the migration of naive and activated T cells in vitro.41 Thus, coadministration of IL-21 and low-dose IL-2 could be a potent cancer therapy by inducing both an immediate innate antitumour response by NK cells and a subsequent long lasting immunity mediated by tumour-specific CD8+ T cells.


We thank Rachel Thomas for thoroughly reading and improving the manuscript. This study was supported by the priority programme SPP1110 of the Deutsche Forschungsgesellschaft (grant JA1058) and Graduate Program of Lower Saxony.