The Human Ortholog of Granulocyte Macrophage Colony-Stimulating Factor and Interleukin-2 Fusion Protein Induces Potent Ex vivo Natural Killer Cell Activation and Maturation

Natural killer (NK) cells are innate effector cells that react against virus-infected or transformed cells before the onset of adaptive immunity. These cells express an array of inhibitory and activating receptors that sense the quality and quantity of self-MHC class I molecules expressed at the cell surface, which mediate the recognition of target cells and regulate the activation state of NK cells. Upon cytokine stimulation, NK cells become lymphokine-activated killer (LAK) cells that proliferate, produce cytokines [IFNγ and tumor necrosis factor (TNF) α], and upregulate effector molecules such as adhesion molecules, perforin, granzymes, Fas ligand, and TRAIL. These features endow NK cells with the ability to kill tumor cells and act as essential promoters of an adaptive immunity against tumor challenge (1). Indeed, several in vitro as well as in vivo studies indicate that tumor cells are recognized as NK cell targets. NK cells not only directly eliminate tumor cells, but also induce the subsequent development of tumor-specific T-cell responses to parental tumor cells. For example, the potential of NK cell–based therapy has been evoked in patients suffering of acute myeloid leukemia. In this case, adoptive transfer of allogeneic NK cells with mismatched NK inhibitory receptors and HLA class I ligands produce graft-versus-leukemia in the absence of graft-versus-host disease due to the ability of donor NK cells to attack recipient dendritic cells, which are responsible for priming donor T cells and induce graft-versus-host disease (2). Therefore, clinical interest in NK cells as a cancer cell pharmaceutical is rapidly growing, with >200 clinical trials examining either autologous or allogeneic NK cells as well as universal donor NK cell lines. However, there are some theoretical limitations to the use of NK cells for cancer therapy. Tumor cells expressing low levels of MHC class I may fail to trigger NK cell activation (3). NK cells from patients affected by advanced bulky malignancies display impaired cytolytic activity, which can be attributed to tumor-derived suppressive factors such as transforming growth factor β that exert deleterious effects of NK cell effector functions (4). Therefore, the infusion of primed and activated NK cells would be an approach that may enhance the effectiveness of these cells in the treatment of cancer. The use of cytokines and combinations thereof would be an obvious method to activate NK cells ex vivo. Better still, pharmacologic activation of NK cells in a manner unachievable with native cytokines may lead to desirable acquired anticancer properties. In this line of thought, we have previously shown that a murine granulocyte macrophage colony-stimulating factor (GM-CSF)/interleukin-2 (IL-2) fusion protein (a.k.a. GIFT2) displays novel antitumor properties in vivo compared with both cytokines in combination regarding to tumor site recruitment of significant functional NK cell infiltration (5). In the present study, we show that the human ortholog of the GM-CSF/IL-2 fusion protein (hGIFT2) induces potent human NK cell activation and cytotoxicity distinct from that achieved by IL-2 or GM-CSF. The use of hGIFT2 as a means to generate oncolytic NK cells may serve an enhanced cellular platform for cancer cell therapy.

293T cells were cultured in DMEM (Wisent Technologies) supplemented with 10% fetal bovine serum (FBS; Wisent Technologies). The cell lines TF-1, CTLL-2, Daudi, K562, and U266 [American Type Culture Collection (ATCC)] were grown according to ATCC's recommendations. Recombinant proteins (hIL-2 and hGM-CSF, R&D Systems); anti α-tubulin (Santa Cruz Biotechnology); polyclonal anti-phosphorylated Jak1, Jak2, Stat1, Stat3, Stat5, extracellular signal-regulated kinase (ERK), and IkBα, as well as their respective antibodies against full-length proteins (Cell Signaling Technology); anti-human FcR III/II, CD3, CD4, CD8, CD56, NKG2D, NKp44, NKp46, NKp30, CD107a, CD117, CD122, CD69, CD226, CD11b, CD43, KIR, IFNγ, or their isotype control antibodies mIgG1 and mIgG2a for flow cytometry (BD Biosciences); and ELISA kits for human IL-2 and GM-CSF (Invitrogen and eBioscience), and for RANTES, IFNγ, and TNFα (R&D systems). Cytotoxicity detection kit lactate dehydrogenase (LDH; Roche Applied Science).

Blood was drawn from healthy donors, after informed consent had been obtained, into heparin- or citrate-coated CPT tubes (BD Biosciences), and peripheral blood mononuclear cells (PBMC) were isolated by density centrifugation according to the manufacturer's recommendation. NK cells were obtained by negative purification using the untouched NK cells isolation kit (Miltenyi Biotec). Cells were treated with NK cell biotin-antibody cocktail and magnetic microbeads reagents and were depleted by magnetic cell sorting system with an autoMACS column (Miltenyi Biotec) according to the manufacturer's instructions. NK cell population (CD56⁺CD3⁻) purity assessed by flow cytometry was 96% (Supplementary Data S1). Three different donors were used for each experiment and an average of two different experiments was performed with the same donor.

The human GM-CSF cDNA (Invivogen) was modified to remove the 3′ nucleotides encoding the stop codon and subsequently cloned in frame with the cDNA encoding the human IL-2 devoid of start codon to generate the cDNA for hGIFT2 fusokine. The hGIFT2 was incorporated into a bicistronic vector (pCMV), allowing the expression of both hGIFT2 and green fluorescent protein (6). The nucleotide sequence of the hGIFT2 cDNA was confirmed by DNA sequencing at the Guelph Molecular Supercenter. Concentrated media conditioned by 293 T cells stably transfected with pCMV encoding for hGIFT2 was used to test the bioactivity of hGIFT2. The IL-2–responsive CTLL-2 or GM-CSF–responsive TF-1 cell lines were plated at a density of 10⁵ cells per well in a 96-well plate and treated with increasing concentration of cytokines for 48 h. Cell proliferation was assessed with a MTT assay.

PBMC were cultured with hGIFT2 and equimolar concentrations of cytokine controls (IL-2, GM-CSF, and IL-2 combined with GM-CSF, 2 pmol). After 7 d in culture, the cells were resuspended in PBS with 2% FBS and incubated with anti-human FcR III/II for 30 min. Subsequently, cells were labeled with a mixture of conjugated antibodies specific for cell surface markers (CD3, CD4, CD8, CD56, and their respective isotype controls) for 1 h at 4°C and analyzed by flow cytometry using a Becton Dickinson FACScan.

For cell surface marker staining, NK cells stimulated with hGIFT2 and equimolar concentrations of cytokine for 7 d were resuspended in PBS with 2% FBS, incubated with anti-human FcR III/II for 30 min, and labeled with conjugated antibodies specific for NK cell activating receptors (NKp30, NKp44, NKp46, NKG2D, CD226), inhibitory receptors (CD43, KIR), maturation markers (CD11b, CD117), NK cell activation markers (CD69, CD107a), and IL-2 receptor components (α subunit CD25, β subunit CD122, and γ subunit CD132). The expression of these cell surface markers was determined by FACS Calibur cytometer (BD) and analyzed using Cellquest software (BD). For IFNγ intracellular staining, NK cells previously stimulated with cytokines for 2 h were treated with Brefeldin A for 4 h, were fixed, were permeabilized, and were stained with IFNγ-conjugated antibody. For Annexin V and propidium iodide staining, NK cells cultured with cytokines (2 pmol) for 7 d were incubated with Annexin V/propidium iodide for 15 min at room temperature and analyzed by flow cytometry. For carboxyfluorescein succinimidyl ester staining, NK cells prestained with CFSE were cultured with cytokines for 4 d, and cell proliferation was assessed by flow cytometry. The data were analyzed using Cellquest software (Becton & Dickinson).

NK cells (1 × 10⁵; effector cells) previously stimulated with hGIFT2 or cytokine controls for 3 d were washed twice and cocultured with 1 × 10⁴ target cells (Daudi, K562, U266) at an effector/target cell ratio of 10:1 for 4 h. Supernatants were collected and used to measure the LDH released from the cytosol of damaged or lysed cells using a cytotoxicity detection LDH kit. As controls, we measured the spontaneous LDH released from target and effector cells. The percentage of cytotoxicity was calculated according to the formula:

Target Maximum: maximum LDH released from Triton X-100–treated target cells.

For signaling analysis, media conditioned by stably transfected 293 T cells expressing hGIFT2, as well as media conditioned by nontransfected cells containing equimolar concentration of cytokine controls (IL-2, GM-CSF, or IL-2 combined with GM-CSF, 5 pmol each), were added to 10⁶ NK cells for 20 min before being lysed and probed by Western blot with rabbit anti-phosphorylated Jak1, Jak2, Stat1, Stat3 and Stat5, ERK, and Akt antibodies. Antibodies against total proteins were used as loading controls as well as anti–α-tubulin antibody. Three independent experiments were performed with NK cells from different donors.

P values were calculated by paired Student's t test. P < 0.05 was considered statistically significant.

The human GMCSF/IL-2 (hGIFT2) construct was created by cloning a modified human GM-CSF cDNA missing the nucleotides encoding for the stop codon in frame with the 5′ end of the human IL-2 cDNA devoid of its start codon. The final fusokine hGIFT2 cDNA encodes for a single polypeptide chain of 302 amino acids (Fig. 1A). Denaturing immunoblotting performed on the supernatant of 293T cells stably transfected to express hGIFT2 (5 ng per million cells in 24 h) showed that the chimeric protein was secreted and has an apparent molecular weight of 45 kDa by SDS-PAGE under reducing conditions (Fig. 1B). We detected and quantified hGIFT2 by IL-2 and GM-CSF ELISA and we found that 1 pmol (45 ng) of hGIFT2 is equivalent to 97 units, whereas 1 pmol (15 ng) of recombinant IL-2 is equivalent to 221 units (Fig. 1C). Both techniques confirm that only the full-length hGIFT2 fusokine of predicted molecular weight is secreted by transfected cells.

The bioactivity of GM-CSF and IL-2 domains as part of hGIFT2 was confirmed by proliferation of GM-CSF–dependent cell line, TF-1 (Fig. 2A), and IL-2–dependent cell line, CTLL-2 (Fig. 2B). The proliferation level of each cell line was determined by MTT incorporation for the last 4 hours of incubation with cytokines. As result, hGIFT2 induces similar proliferation levels of these cell lines compared with equimolar concentration of each cytokine controls, which indicate that both GM-CSF and IL-2 as part of the fusion preserve the ability to induce an appropriate proliferative response.

The effect of hGIFT2 on lymphocytes was determined by culturing PBMCs with equimolar concentrations of cytokines (2 pmol). After 7 days in culture, we observed that hGIFT2 promotes a significant increase in the number of NK cells compared with equimolar concentration of cytokine controls, whereas the number of CD8, CD4, and NKT cells did not show significant differences with each cytokine treatment (Fig. 3A). Similarly, we observed a significant increase of NK cell percentage in the hGIFT2-treated PBMC fraction from three different donors (Fig. 3B). Based on these results, we investigated whether hGIFT2 induces NK cell proliferation or enhances NK cell survival. To test these hypotheses, purified NK cells were prestained with CFSE, cultured with equimolar concentrations of cytokines, and analyzed by flow cytometry. After 4 days in culture, IL-2 as well as IL-2 combined with GM-CSF induces NK cell proliferation (Fig. 3C). However, after 7 days in culture with cytokine controls the NK cell number decreased, although the media was changed at day 4 to avoid cell starvation. In contrast, hGIFT2 does not induce NK cell proliferation but rather protect NK cells from apoptosis (Fig. 3C and D). Based on these results and in agreement with previous reports (7, 8), we suggest that IL-2–activated NK cells undergo apoptosis in long term culture.

Purified human NK cells pretreated with hGIFT2 displayed an enhanced activation profile compared with NK cells stimulated with equimolar concentration of cytokine controls, based on the expression levels of the early activation marker (CD69) and a marker of degranulation (CD107a; Fig. 4A). To evaluate the responsiveness of NK cells to hGIFT2, we analyzed the expression levels of the IL-2 trimeric receptor subunits (α, β, γ). IL-2Rα and IL-2Rγ expression levels were not modified by any cytokine treatment (data not shown), whereas IL-2Rβ expression was upregulated by hGIFT2 and downregulated by GM-CSF (Fig. 4A). Intracellular staining for IFNγ indicates that hGIFT2 induces a robust expression of IFNγ in the first 6 hours of cytokine treatment (Fig. 4B,, left). By ELISA, we quantified that hGIFT2 promotes >37-fold greater IFNγ production than equimolar concentration of IL-2 after 3 days in culture (Fig. 4B,, right), as well as significant secretion of RANTES (Fig. 4C,, bottom). However, we did not detect significant differences regarding TNFα production (Fig. 4C , top). We also observed that GM-CSF inhibits NK cell–derived IFNγ, TNFα, and RANTES production. These results indicate that hGIFT2 displays superior properties as a tool for NK cell activation and overcomes GM-CSF–mediated immunosuppression. In addition, hGIFT2 similarly affect both CD56^bright and CD56^dim subsets of NK cells, which upon hGIFT2 stimulation become highly activated and granulated (Supplementary Data S2).

Spontaneous cytotoxic activity is mainly triggered by the activating receptors natural cytotoxicity receptors (NCR; NKp30, NKp44, NKp46), NKG2D, and leukocyte adhesion molecule DNAX accessory molecule 1 [DNAM-1 (CD226); refs. 1, 3]. hGIFT2-stimulated NK cells upregulate NKp44, NKp46, and DNAM-1 (CD226) expression (Fig. 5A) but not NKG2D (Supplementary Data S3A) and NKp30 (data not shown) expression. In addition, hGIFT2 downregulates the expression of the receptor for stem cell factor c-kit (CD117) and upregulates the expression of integrin CD11b (macrophage antigen-1; Fig. 5B). hGIFT2 downregulates the expression of the inhibitory receptor CD43 and does not alter the expression level of the inhibitory receptors KIR (Supplementary Data S3B and C). This phenotype indicates that hGIFT2-stimulated NK cells undergo a maturation process, which correlated with significant greater cytotoxicity than cytokine controls against both NK-resistant Daudi cells and NK-sensitive K562 and U266 tumor cells. As previously reported (9), we also observed that GM-CSF significantly suppresses NK cell cytotoxicity (Fig. 5C). After 7 days in culture with cytokines, a 30% decline of cytotoxicity was observed in hGIFT2-stimulated NK cells, and a 60% reduction of killing ability in IL-2–stimulated NK cells (Supplementary Data S4).

NK cells express components of both IL-2 and GM-CSF receptors (10). The intracellular signaling of IL-2R occurs through the β chain [JAK1/signal transducers and activators of transcription (STAT) 3] and the γ chain (JAK3/STAT5), whereas the GM-CSF receptor signals through its β chain (JAK2/STAT5). To decipher the molecular mechanism underpinning the effect of hGIFT2 on NK cells, we analyzed the activation status of JAK kinases and STAT transcription factors downstream of IL-2 and GM-CSF receptors, respectively. We observed that hGIFT2 induces robust hyperphosphorylation of JAK1 and JAK2, and this is associated with greater activation of STAT1, STAT3, and STAT5 when compared with controls. In contrast, GM-CSF completely abrogates STAT5 phosphorylation and inhibits IL-2–mediated STAT5 activation (Fig. 6A and B). We also measured the activation status of other signaling pathways downstream of IL-2 and GM-CSF receptors, such as MKK/ERK and AKT pathways (11, 12). AKT pathway was more activated by hGIFT2 than cytokine controls (Fig. 6D), which supports the observation that hGIFT2 enhances NK survival. In contrast, the activation status of NF-κB and MKK/ERK did not show significant differences in hGIFT2-stimulated NK cells compared with equimolar concentrations of cytokine controls (data not shown).

The clinical effectiveness of GM-CSF and IL-2 combination for the treatment of cancer is controversial. A phase II clinical study of moderate-dose IL-2 and GM-CSF in patients with metastatic or unresectable renal carcinoma fails to manifest the synergistic therapeutic effect of these two cytokines (13). Indeed, patients receiving GM-CSF and IL-2 treatment display an impaired LAK activity compared with IL-2 alone, and specifically, GM-CSF has been shown to downregulate certain aspects of innate immune response such as NK cell cytotoxicity (14). In the present study, we have shown that hGIFT2 displays unheralded novel properties that overcome GM-CSF–mediated NK cell suppression and promotes potent NK cell activation. hGIFT2 upregulates the expression of the early activation marker CD69 and promotes greater NK cell degranulation based on acquired surface expression of LAMP-1 (CD107a). The potent immune stimulatory effects of hGIFT2 on NK cells are evidenced by a substantial upregulation of NK activating receptors NKp44 and NKp46. The human triggering receptors responsible for NK cell cytotoxicity include the NCRs and NKG2D. NCRs belong to the Ig superfamily and include NKp46 and NKp30, which are expressed by both resting and activated NK cells, and NKp44, which is expressed only by activated NK cells (15–17). Although the cellular ligands on tumor cells for NCRs are currently unknown, NCRs have been identified as crucial receptors for target cell recognition and induction of NK cell cytotoxicity toward a wide range of cancer cells. Substantial evidence indicates that there is a direct correlation between NCR surface densities and the ability of NK cells to kill different tumor target cells (18). hGIFT2 slightly upregulates the expression of DNAM-1 (CD226), which is a transmembrane glycoprotein that mediates lymphocyte adhesion and signaling. This activating receptor recognizes PVR (CD155) and Nectin-2 (CD112), two members of the lectin family that are highly expressed on carcinomas, melanomas, and some hematopoietic cell lines (19, 20). Consequently, hGIFT2-stimulated NK promote significant cytotoxic activity against all target tumor cells analyzed. Interestingly, hGIFT2-treated NK cells were able to lyse “NK-resistant” Daudi cells (Burkitt's lymphoma). These tumor cells are negative for MHC class I expression, but expresses MHC class II, which correlates with the increase resistant to NK-mediated lysis (21). As expected, hGIFT2-stimulated NK cells also promote greater cell lysis of NK-sensitive K562 and U266 cell lines. K562 leukemia cells (chronic myelogenous leukemia) are negative for both MHC class I and II expression, whereas U266 cells (myeloma), although are sensitive to NK-mediated cell cytotoxicity, express high levels of MHC class I (22), which supports “the missing self” hypothesis that NK cells kill certain targets because they fail to express adequate levels of MHC gene products.

NK cells stimulated with IL-2 for 7 days display a remarkable reduction of cytotoxicity activity against myeloma cells. According to previous reports that IL-2 induces NK cells apoptosis in long-term culture (7, 8), we suggest that the reduction of NK cell cytotoxicity over time is due to a decrease of IL-2–stimulated NK viability. On the other hand, hGIFT2 does not induce NK cell proliferation compared with equimolar concentration of IL-2, but rather protects NK cells from apoptosis, which result in a net increase of NK cell number over time. In addition, hGIFT2 prompts NK cells to secrete substantial amounts of IFNγ (37-fold greater than equimolar concentrations of IL-2) as well as RANTES. The extremely high production of IFNγ may restrain the proliferation of NK cells and at the same time prevent NK cell apoptosis, since IFNγ has been reported to induce cell cycle arrest in several cell types (23). In LAK cells, IFNγ in synergy with IL-2 acts as inhibitor for LAK cell proliferation but not differentiation (24). We also found that distinct from IL-2, hGIFT2 induces a NK cell maturation process characterized by the downregulation of the expression of the receptor for stem cell factor c-kit (CD117), which is expressed by NK cell precursors but is largely diminished on matured NK cells (25, 26). CD11b integrin upregulation constitute another marker of NK cell maturation (27). As molecular mechanisms underpinning hGIFT2 effect on effector cells, hGIFT2 not only upregulates IL-2Rβ (CD122) expression but also induces a hyperactivation of JAK1 and JAK2 kinases associated to β chains of both IL-2 and GM-CSF receptors, which lead to a robust phosphorylation of STAT1, STAT3, and STAT5 transcription factors. By comparing the levels of STAT5 phosphorylation in hGIFT2-stimulated NK cells with equimolar concentration of GM-CSF, we observed that hGIFT2 overcomes GM-CSF–mediated suppressive signals on NK cells. These results suggest that hGIFT2 has higher avidity for NK cells than single cytokine control maybe due to a bivalent binding of hGIFT2 to both receptors expressed on NK cells. Interestingly, hGIFT2 fusokine promotes greater activation of phosphoinositide 3-kinase/Akt in NK cells. Activated Akt in turn phosphorylates a number of downstream target genes that prevent apoptosis and promote cell survival, which could be the mechanism used by hGIFT2 to protect NK cell from apoptosis and to enhance survival.

Several clinical trials using ex vivo IL-2–expanded and activated autologous NK cells have shown apparent limited effectiveness in the therapy of cancer, including melanoma, renal carcinoma, lung carcinoma, ovarian, and brain cancer (28). The infusion of high doses of IL-2 not only is associated with severe toxic side effects, but also sensitizes NK cells to apoptosis when they are in contact with the vascular endothelium, which may cause a decrease of NK cell migration to the tumor (29, 30). Interestingly, allogeneic NK cells characterized by a KIR/KIR-ligand incompatibility display more cytotoxicity against solid tumors and leukemia (31, 32). Remarkably, patients with acute myeloid leukemia treated with alloreactive NK cells benefit from higher rates of engraftment, survival, and reduced incidence of GvHD (33). The risk of GvHD is a consequence of the inefficient priming of alloreactive donor T cells consequent to NK cell–mediated killing of recipient antigen-presenting cells (34). However, alloreactive NK cells arising from hematopoietic stem cell transplantation require several weeks to mature posttransplantation, which may contribute to leukemia relapse (35). Therefore, we may speculate that the infusion of mature NK cells would more effectively prevent leukemic relapse, promote engraftment by eliminating patient lympho-hematopoietic cells, and reduce GvHD by lysing patient antigen-presenting cells. Therefore, biochemical NK priming that enhances their survival, effector function, and maturation would be desirable in the setting of cancer cell immunotherapy—features afforded by NK cell treatment with hGIFT2.

In summary, NK cells are the only lymphoid cells that express components of both IL-2 and GM-CSF receptors (10). Therefore, the selective effect of hGIFT2 on NK cells suggest that hGIFT2 binds to these cytokine receptors in an atypical fashion, which triggers aberrant signals downstream of both IL-2 and GM-CSF receptors, leading to the hyperphosphorylation of receptor-associated JAK/STAT proteins. Consequently, hyperactivated NK cells produce large amount of IFNγ and RANTES and overexpress NCRs involved in NK cell cytotoxicity while contemporaneously acquiring features of mature and apoptosis-resistant cells. The use of GIFT2-primed autologous NK cells may therefore be of interest in cell therapy of cancer.

No potential conflicts of interest were disclosed.

Grant support: Canadian Institute for Health Research operating grant MOP-15017. C. Penafuerte is a recipient of Montreal Centre for Experimental Therapeutics in Cancer Scholarship and U.S. Army Graduate study Scholarship. J. Galipeau is a Fonds de recherché en santé du Québec chercheur-boursier senior.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Manaf Bouchentouf, Dr. N. Eliopoulos, and Dr. Patrick Williams for technical advice and materials.