Tyrosine kinase inhibitors (TKIs) are used in the clinical management of hematological neoplasms. Moreover, in solid tumors such as stage 4 neuroblastomas (NB), imatinib showed benefits that might depend on both on-target and immunological off-target effects. We investigated the effects of imatinib and nilotinib on human NK cells, monocytes, and macrophages. High numbers of monocytes died upon exposure to TKI concentrations similar to those achieved in patients. Conversely, NK cells were highly resistant to the TKI cytotoxic effect, were properly activated by immunostimulatory cytokines, and degranulated in the presence of NB cells. In NB, neither drug reduced the expression of ligands for activating NK receptors or upregulated that of HLA class I, B7-H3, PD-L1, and PD-L2, molecules that might limit NK cell function. Interestingly, TKIs modulated the chemokine receptor repertoire of immune cells. Acting at the transcriptional level, they increased the surface expression of CXCR4, an effect observed also in NK cells and monocytes of patients receiving imatinib for chronic myeloid leukemia. Moreover, TKIs reduced the expression of CXCR3 (in NK cells) and CCR1 (in monocytes). Monocytes also decreased the expression of M-CSFR, and low numbers of cells underwent differentiation toward macrophages. M0 and M2 macrophages were highly resistant to TKIs and maintained their phenotypic and functional characteristics. Importantly, also in the presence of TKIs, the M2 immunosuppressive polarization was reverted by TLR engagement, and M1-oriented macrophages fully activated autologous NK cells. Our results contribute to better interpreting the off-target efficacy of TKIs in tumors and to envisaging strategies aimed at facilitating antitumor immune responses.
Virtually all patients with chronic myeloid leukemia (CML) and 20% of adult patients with acute lymphoblastic leukemia carry the translocation between the long arms of chromosomes 9 and 22 that generates the Philadelphia chromosome with a chimeric gene that encodes the fusion protein BCR-ABL, having constitutively activated tyrosine kinase activity. Imatinib mesylate (Glivec) was the first designed protein tyrosine kinase inhibitor (TKI) used in the treatment of CML and BCR-ABL+ ALL (1–3). This drug binds to the protein in a fashion that prevents ATP interacting with the ATP-binding site of the ABL kinase, thereby blocking the tyrosine phosphorylation of proteins involved in BCR-ABL signal transduction (4). Importantly, in imatinib-resistant or intolerant BCR-ABL+ patients, an alternative therapy is based on the administration of nilotinib (Tasigna), a second-generation TKI that, in vitro, is more potent and selective than imatinib as a BCR-ABL tyrosine kinase inhibitor (5–7).
Imatinib and nilotinib also target the discoidin domain receptors 1 and 2, the platelet-derived growth factor receptors α and β (8), the M-CSF receptor (9), and KIT (stem cell growth factor receptor, CD117) (10, 11). KIT is constitutively active in most human gastrointestinal stromal tumors (GISTs), a common mesenchymal neoplasm of the gastrointestinal tract and, in this context, imatinib is registered for the treatment of GISTs (12, 13). Neuroblastomas (NB) represent the most common extracranial solid malignancy of childhood (14); they frequently localize in the abdomen, particularly in the adrenal gland or at lumbar sympathetic ganglia. Metastatic disease (stage 4), very often characterized by bone marrow (BM) infiltration, is diagnosed in at least 50% of NB patients and it is still associated with a dismal prognosis because of a high risk of tumor relapse. Children with either refractory or relapsing metastatic NB have been recently enrolled in a two-stage, phase II clinical trial with imatinib; the drug was well tolerated and showed benefits in a subset of patients, particularly those with BM as the only site of metastatization, low tumor infiltration, and low imatinib exposure (15).
It has been suggested that the efficacy of imatinib in GISTs might depend on both on- and off-target effects, with the latter being related to the antiangiogenic effects and strengthening of antitumor immune responses (16–18). It is conceivable to speculate that also in NB, the efficacy of imatinib might be at least in part related to the enhancement of the anti-NB activity of immune cells, such as NK cells. When properly activated, NK cells exert a potent cytolytic activity and release immunostimulatory cytokines such as IFN-γ that potentiate both innate and adaptive immune responses, thus representing important effectors against both hematological and nonhematological malignancies (19, 20).
The NK cell activation depends on the interaction between activating NK receptors and their cognate ligands on tumor targets. In humans, the principal activating receptors include NKp46, NKp30, and NKp44 (collectively termed natural cytotoxicity receptors), DNAM-1, and NKG2D (21, 22). Different ligands have been identified that are upregulated or de novo expressed on the surface of tumor cells. Whereas PVR and nectin-2 are ligands of DNAM-1 (23, 24), and NKG2D interacts with the stress-inducible molecules MICA/B and ULBPs (25), the ligands of natural cytotoxicity receptors are not yet fully defined. NKp46 is still an orphan receptor, whereas B7-H6 (26) and a novel isoform of the mixed-lineage leukemia protein (MLL5) (27) have been identified as ligands of NKp30 and NKp44, respectively. NK cell activity unleashed by the interaction between activating receptors and their ligands is under the control of receptors that recognize optimal levels of self HLA class I molecules on potential targets and transduce inhibitory signals (21, 28). These NK inhibitory receptors include the killer Ig-like receptors (CD158) and the CD94/NKG2A heterodimer.
Either downregulation or loss of HLA class I molecules, which frequently occur in various types of malignancies, including NB, allows an efficient NK-mediated killing of tumors (29). However, it has been shown that several mechanisms might operate in the tumors to escape from immune-mediated surveillance (14, 30). The antitumor activity of NK cells depends on their capability of reaching and infiltrating the tumor tissues. NK cell migration in pathological tissues must be driven by chemokine gradients that are sensed through the expression of chemokine receptors that include CXCR3 (CXCL4, 9, 10, 11 receptor), CXCR1 (CXCL8 receptor), CX3CR1 (CX3CL1 receptor), and CXCR4 (CXCL12 receptor). The latter drives NK cells toward peripheral tissues and is essential for homing and maintenance of NK cells in stromal niches within the BM (31). Upon reaching the tumor microenvironment, NK cells encounter not only neoplastic cells, but also other immune effectors such as macrophages, which are the most represented leukocytes in cancer tissues. Peripheral blood (PB) monocytes express CCR1, are recruited by CCL1 (MIP-1α), and differentiate into macrophages (M0) under the influence of M-CSF (also known CSF1) (32). M0 cells can polarize toward either the M1 immunostimulatory or the M2 “alternative” immunosuppressive functional phenotype, depending on microenvironmental conditions.
The tumor microenvironment has been shown to promote M2 polarization that is characterized by high expression of scavenger receptors, such as CD206, and the release of factors that promote, instead of suppress, tumor growth. These include vascular endothelial growth factor, fibroblast growth factor, matrix metalloproteinases, and immunomodulatory cytokines, such as IL-10 and TGF-β, which in NK cells affect the expression of activating receptors and modulate the chemokine receptor repertoire (33). Importantly, the M2 status is not definitive, and TLR engagement can promote M1 polarization of macrophages, which release high amounts of proinflammatory/immunostimulatory cytokines, such as IL-1β, IL-6, TNF-α, IL-12, and IL-18 and are capable of inducing strong activation of NK cells (34, 35).
In this study, we examined the effect of two TKIs, imatinib and nilotinib, routinely used in the clinical management of several neoplasms, on the survival and function of human NK cells, monocytes, and macrophages, also considering their polarization properties and ability to interact with autologous NK cells.
Materials and Methods
Cells and TKIs
Buffy coats were collected from healthy volunteer blood donors admitted at the blood transfusion center of IRCCS S. Martino-IST after obtaining informed consent, and the study was approved by the Ethics Committee of IRCCS San Martino-IST (39/2012). After standard Ficoll-Paque density-gradient (Euroclone, Pero, Italy), monocytes and NK cells were purified with human monocyte cell isolation kit II or a human NK cell isolation kit, respectively (Miltenyi Biotec, Bergisch Gladbach, Germany). Macrophages (M0) were obtained culturing monocytes for 7 d in 24 lumox multiwell TC-QUALITAET plates (Greiner Bio-One, Frickenhausen, Germany) at a density of 5 × 105/ml and in the presence of 100 ng/ml rM-CSF (PeproTech, London, U.K.). To obtain M2 macrophages (M2), M0 were cultured for an additional 18 h with 20 ng/ml rIL-4 (PeproTech). For coculture experiments, NK cells, simultaneously purified with monocytes, were cultured for 7 d with recombinant human IL-2 (32 U /ml) (Proleukin; Chiron, Emeryville, CA), then harvested, washed, and added to autologous macrophages. To obtain short-term activated NK cells, purified NK cells were cultured overnight with rIL-12 (1 ng/ml) (PeproTech) and rIL-18 (Medical and Biological Laboratories, Nagoya, Japan) at 100 or 20 ng/ml for IFN-γ production or a CD107A degranulation assay, respectively. For TLR stimulation, macrophages were incubated for 24 h with 100 ng/ml LPS from Escherichia coli (Sigma-Aldrich) either in the absence or in the presence of imatinib or nilotinib. M2 macrophages were cocultured overnight with autologous NK cells in the presence of LPS at the 1:1 NK cell/macrophage ratio either in the absence or in the presence of drugs. For macrophage differentiation experiments, monocytes were cultured for 7 d with M-CSF in the absence or in the presence of imatinib or nilotinib, which were added to the cultures daily.
The NB cell line HTLA-230 was provided by Dr. E. Bogenmann (Children’s Hospital Los Angeles, Los Angeles, CA) (36); the SH-SY5Y cell line was purchased from Banca Biologica and Cell Factory (IRCCS Azienda Ospedaliera Universitaria San Martino-IST, Genoa, Italy). NB cell lines were cultured in the presence of RPMI 1640 medium supplemented with 10% heat-inactivated FCS (Biochrom, Berlin, Germany), 50 mg/ml streptomycin, 50 mg/ml penicillin (Sigma-Aldrich), and 2 mM glutamine (EuroClone). NB cell lines were periodically checked for MYCN amplification by fluorescence in situ hybridization analysis and for morphology, proliferation rate, and mycoplasma contamination, after thawing and within four passages in culture.
The K-562 (chronic myelogenous leukemia) cell line (American Type Culture Collection, Rockville, MD) was maintained in RPMI 1640 medium with 10% FBS (Biowest, Nuaille, France).
Clinical-grade imatinib mesylate (Glivec) and nilotinib (Tasigna) were provided by Novartis Pharma (Basel, Switzerland) to M.V.C. (MTA number 39985). The lyophilized powders were reconstituted in distilled water (imatinib) or DMSO (nilotinib) at the concentration of 10 mg/ml and further diluted in cell culture medium at the indicated work concentrations and times. Medium supplemented with the same percentage of DMSO present at the various nilotinib concentrations was used as a control.
NB and CML patients
The PB samples were collected after informal consent from seven patients diagnosed with high-risk NB either at the disease onset or relapsing (admitted at the Oncology Units of the Istituto Giannina Gaslini, Genoa and Ospedale Bambino Gesù, Rome), two pediatric CML patients (pCML1 and pCML3) (admitted at Ospedale Bambino Gesù, Rome), and three adult CML patients (aCML2, aCML3, and aCML4) (admitted the Hematology Unit, University of Pavia, Pavia). The CML patients were taking imatinib therapy daily.
After standard Ficoll-Paque density gradient separation, PBMCs were treated or not with rIL-12 (1 ng/ml) and rIL-18 (100 ng/ml) in the presence or in the absence of imatinib (6 μg/ml) for 24 h. For the analysis of NK cells and monocytes, PBMCs were stained with a mixture of anti–CD56-PC5 and anti–CD3-FITC (NK cells) and anti-CD14 (monocytes) (Beckman Coulter/Immunotec, Marseille, France). PBMCs derived from healthy volunteer blood donors were used as control.
Flow cytofluorimetric analysis and cytokine release assay
For cytofluorimetric analysis (FACSCalibur; Becton Dickinson, Mountain View, CA), cells were stained with PE-, FITC-, or PC5-conjugated mAbs or with unconjugated mAbs followed by PE-conjugated isotype-specific goat anti-mouse second reagent (SouthernBiotech, Birmingham, AL). Isotype-matched irrelevant mAbs were used as control. Monocytes and macrophages were preincubated for 30 min at 4°C with FcR blocking reagent (Miltenyi Biotec) before specific mAb staining. For apoptosis and necrosis assays, before cytofluorimetric analysis, cells were incubated for 10 min at room temperature with annexin V and TO-PRO-3 iodide (Life Technologies, Carlsbad, CA) to identify apoptotic (annexin V+), necrotic (TO-PRO-3 iodide+), or viable (annexin V−, TO-PRO-3 iodide−) cells. For phagocytosis assays, macrophages were incubated with unopsonized pHrodo E. coli BioParticles or opsonized pHrodo E. coli BioParticles (Life Technologies), at the 1:5 macrophage/BioParticle ratio, at the indicated times and then harvested and analyzed by flow cytometry. On every experimental session, the flow cytometer performances were monitored and the reproducibility of the fluorescence intensity was aligned by calibrated microsphere (Becton Dickinson).
Cytokine content in the supernatants of NK cells or macrophages was quantified by ELISA kits: IFN-γ, TNF-α, IL-12p40/p70 (Life Technologies), and IL-18 (Medical and Biological Laboratories).
CD107a degranulation assay
HTLA-230 or cytokine-activated NK cells were cultured for 24 h in medium alone or supplemented with different concentrations of imatinib or nilotinib. Then cells were cocultured for 3 h at the E:T ratio of 1:1 in the presence of anti–CD107a-PE mAb. M2 macrophages were cocultured for 24 h with autologous NK cells in the presence of LPS at the 1:1 NK/macrophage ratio either in the absence or in the presence of drugs. NK cells were then collected and incubated with the HTLA-230 cell line at the E:T ratio of 1:1 for 3 h in the presence of anti–CD107a-PE mAb. Before flow cytometry, NK cells were stained with a mixture of anti–CD56-PC5 and anti–CD3-FITC.
PBMCs derived from CML patients, treated or not with I6, or PBMCs derived from healthy donors were cocultured with the K562 cell line for 3 h in the presence of anti–CD107a-PE mAb, taking in consideration the percentage of NK cells and using a NK/target ratio of 1:1. Before flow cytometry, NK cells were stained with a mixture of anti–CD56-PC5 and anti–CD3-FITC.
Total RNA was extracted from NK cells, monocytes, and SH-SY5Y cells using the miRCURY RNA isolation kit—cell and plant (Exiqon), according to the manufacturer’s guidelines. RNA (300 ng) was reverse transcribed using the SuperScript VILO cDNA synthesis kit (Invitrogen). Real-time PCR was performed using specific TaqMan gene expression assays (Applied Biosystems). CXCR4 gene expression was normalized to HPRT1 gene expression. Experiments were performed in triplicate.
The following mAbs were produced in our laboratory: BAB281 (IgG1, anti-NKp46), AZ20 (IgG1, anti-NKp30), Z231 (IgG1, anti-NKp44), BAT221 (IgG1, anti-NKG2D), KRA236 (IgG1, anti-DNAM-1), MA127 (IgG1, anti-NTBA), PP35 (IgG1, anti-2B4), C227 (IgG1, anti-CD69), MAR93 (IgG1, anti-CD25), M5A10 (IgG1, anti-PVR), L14 (IgG2a, anti–Nectin-2), 5B14 (IgM, anti–4IgB7-H3), BAM195 (IgG1, anti-MICA), A6136 (IgM) and 6A4 (IgG1) (anti–HLA class I), D1.12 (IgG2a anti–HLA-DR), and C127 (IgG1 anti-CD16). Anti–PD-L1.3.1 (IgG1, anti-PD-L1) and anti–PD-L2 (IgG1, anti–PD-L2) mAbs were produced in D. Olive’s laboratory. Anti-CD14 (IgG2a) and a mixture of anti–CD56-PC5 and anti–CD3-FITC (IgG1) and anti–CD20-FITC were purchased from Beckman Coulter/Immunotec; anti-CD64 (IgG1), anti–CD80-PE, anti–CD107A-PE, and anti–CD206-FITC were purchased from BD Biosciences (San Diego, CA); anti–human IL-18 (IgG1) and anti–CX3CR1-PE (rat, IgG2b) were purchased from Medical and Biological Laboratories; anti-CCR1 (IgG2b), anti-CCR7 (IgG2a), anti-CXCR4 (IgG2b), anti-CXCR3 (IgG1), anti–CD204-PE, and anti–M-CSFR (IgG1) were purchased from R&D Systems (Minneapolis, MN); anti–CD31-PE, anti–CD33-PE, anti–CD36-PE, and anti–CCR6-PE were purchased from Miltenyi Biotec; anti-CXCR1 (IgG1) and anti-ULBP3 (166510, IgG2a) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA); and anti–annexin V-FITC was purchased from eBioscience (San Diego, CA). All Abs were of mouse origin, unless otherwise specified. PE and FITC isotype-matched mouse (BD Biosciences) (Miltenyi Biotec) or rat (Medical and Biological Laboratories) mAbs were used as negative controls.
A Wilcoxon–Mann–Whitney p value test (nonparametric significance test) was employed. Graphic representation and statistical analysis were performed using PASW Statistics version 20.0 software (formerly SPSS Statistics) (IBM, Milan, Italy) and GraphPad Prism 6 (GraphPad Software, La Jolla, CA).
PB NK cells are more resistant than monocytes to the cytotoxic effects of imatinib and nilotinib
The HTLA-230 human NB cell line was treated with concentrations of either imatinib or nilotinib ranging from 120 to 0.3 μg/ml (Fig. 1A, Supplemental Fig. 1A) and after 24 h assessed for apoptosis and/or necrosis by staining with annexin V and TO-PRO-3 iodide and flow cytometry analysis. Whereas the cells were unaffected by nilotinib, with imatinib concentrations ≥15 μg/ml, increasing numbers of necrotic cells were detected and all cells died upon exposure to concentrations of 120 μg/ml.
We analyzed the influence of decreasing concentration of drugs on the survival of circulating monocytes and NK cells. Monocytes and NK cells were purified from PB of healthy donors and their viability was assessed after 24 h of treatment with one or another drug (Fig. 1B, 1C, Supplemental Fig. 1A). At high concentration (≥15 μg/ml), monocytes were susceptible to treatment with both drugs, whereas NK cells were killed by imatinib but virtually resistant to treatment with nilotinib, even at the highest concentration used. At drug concentration ≤6 μg/ml, a low percentage of dead monocytes was detected after treatment with imatinib, whereas with nilotinib high numbers of cells underwent apoptosis and/or necrosis, particularly at 6 μg/ml. NK cells were poorly susceptible to treatment with a low concentration of imatinib and virtually resistant to nilotinib.
To confirm that NK cells were less susceptible than monocytes to the cytotoxic effect of TKIs, cells were treated daily with a drug concentration of ≤6 μg/ml and analyzed for apoptosis/necrosis after 24, 48, and 72 h (Supplemental Fig. 1B). The NB cell line was resistant to treatment with TKIs. The percentage of monocytes undergoing apoptosis/necrosis increased at 48 h, and with drug concentrations ≥3 μg/ml most cells died on day 3. On the contrary, NK cells remained resistant to all nilotinib concentrations even after 3 d of treatment. Two days of treatment with imatinib minimally affected NK cell survival, and relevant numbers of dead cells were detected on day 3 only. This, however, occurred in some donors, whereas in others NK cells were poorly susceptible to imatinib cytotoxicity even after 3 d of treatment.
Imatinib and nilotinib modify the chemokine receptor repertoire of monocytes and NK cells
We analyzed the phenotype of monocytes and NK cells that had been treated with a concentration of ≤6 μg/ml, gating on annexin V and TO-PRO-3 iodide double-negative viable cells. Neither drug altered the expression of a large panel of surface markers, typical of one or another cell type (34, 35) (Supplemental Fig. 1C). On the contrary, imatinib and nilotinib significantly modified the chemokine receptor repertoire of immune cells (Fig. 2A, Supplemental Fig. 1C). In particular, both monocytes and NK cells showed strong upregulation of the expression of CXCR4. Moreover, at 6 μg/ml, downregulation of CCR1 or CXCR3 was detected in monocytes and NK cells, respectively. Neither drug modified the expression of other chemokine receptors, including CX3CR1, which is expressed by both monocytes and NK cells (Supplemental Fig. 1C).
The drug-mediated upregulation of CXCR4 was observed not only in monocytes and NK cells but also in PB T and B cells (Fig. 2B). Notably, neither drug induced significant modulation of CXCR4 expression in NB cell lines that constitutively expressed different levels of the chemokine receptor (Fig. 2B, Supplemental Fig. 1D).
With regard to the mechanism responsible for TKI-mediated upregulation of CXCR4 in immune cells, preliminary results showed regulation at the transcriptional level. Indeed, according to the surface phenotype, imatinib and nilotinib increased CXCR4 mRNA expression in NK cells and monocytes, but not in NB cell lines (Fig. 2C).
Imatinib and nilotinib did not affect or marginally impair cytokine-mediated activation of NK cells
Activated NK cells display cytolytic activity against NB cells, and DNAM-1/PVR interactions play a predominant role in tumor recognition (29). We analyzed whether drug concentrations of ≤6 μg/ml could impair the capability of NK cells to degranulate in the presence of NB. In NB, neither drug modified the expression of ligands for the activating NK receptors DNAM-1 and NKG2D, or induced upregulation of molecules that could limit the NK cell function (Supplemental Fig. 1D). These include HLA class I, B7-H3 (37), and PD-L1 and PD-L2 (38), ligands of the PD-1 inhibitory receptor (39).
Next we analyzed whether TKI conditioning could affect cytokine-mediated activation of NK cells. As shown in Fig. 3, the presence of imatinib and 0.3 μg/ml nilotinib did not significantly impact the acquisition of the CD69 activation marker, and NK cells in the presence of NB showed a degranulation capability comparable to that observed in the absence of TKIs. The only exception was represented by nilotinib used at 6 μg/ml, which hampered degranulation capability, with this effect being paralleled by a reduced upregulation of CD69.
Interestingly, as observed in PB NK cells, all drug concentrations also induced significant upregulation of CXCR4 in cytokine-stimulated NK cells (Fig. 3B).
M0 and M2 macrophages are more resistant to TKIs than monocytes and maintain the capability of polarizing toward M1
We investigated whether imatinib or nilotinib could affect the survival and function of macrophages, in particular of those polarized toward M2 that might display a tumor-promoting role (Fig. 4, Supplemental Fig. 2A–D). Purified PB monocytes were cultured with M-CSF and on day 7 differentiated macrophages (M0) were polarized toward M2 in the presence of IL-4. M0 and M2 macrophages were treated with different concentrations of imatinib or nilotinib and analyzed for apoptosis and/or necrosis by flow cytometry. M0 and M2 were more resistant to TKI exposure (Fig. 4, Supplemental Fig. 2A–D) as compared with monocytes (see Fig. 1B, Supplemental Fig. 1A, 1B). Indeed, imatinib concentrations ≥30 μg/ml were necessary to induce apoptosis/necrosis in a relevant number of macrophages, and cell viability was only marginally affected by TKI concentration up to 15 μg/ml. Moreover, macrophages were highly resistant to all concentrations of nilotinib (Fig. 4, Supplemental Fig. 2A).
The treatment with drug concentrations ≤6 μg/ml did not damage the phagocytosis capability of macrophages, which showed levels comparable to controls (Supplemental Fig. 2B), or their ability to polarize toward M1 via TLR stimulation. Indeed, also in the presence of drugs, LPS stimulation increased the expression of CD80 and HLA class I and abolished that of the membrane-bound form of IL-18 (mIL-18) (40) (Supplemental Fig. 2C). Moreover, after TLR stimulation, M0 and M2 released proinflammatory/immunostimulatory cytokines such as IL-12, TNF-α, and, according to the loss of mIL-18, soluble IL-18 (Supplemental Fig. 2D).
In the presence of imatinib and nilotinib, TLR-stimulated M2 macrophages promote NK cell activation
We analyzed whether imatinib or nilotinib could influence the capability of M1-polarizing macrophages to induce the activation of PB NK cells. Purified PB NK cells were cocultured with autologous M2 macrophages stimulated with LPS, either in the absence or in the presence of TKI concentration ≤6 μg/ml. After 24 h, NK cells were recovered and evaluated for the phenotypic and functional characteristics (Fig. 5, Supplemental Fig. 2E).
In agreement with previously published data (34, 35), TLR-stimulated M2 induced in NK cells upregulation of the expression of CD69, CD25, and CCR7, enhancement of degranulation capability, and the release of high amounts of IFN-γ. In the presence of imatinib, NK cells maintained the capability to upregulate the CD69 and CD25 expression, to degranulate in the presence of the NB cell line, and released IFN-γ at levels comparable to those observed with the classical rIL-12 plus rIL-18 cytokine stimulation. In the presence of nilotinib at 0.3 μg/ml, NK cells showed CD69 and CD25 upregulation, as well as increased degranulation and IFN-γ production. On the contrary, at a concentration of 6 μg/ml, nilotinib significantly impaired the acquisition of CD69 and the increase of degranulation capability.
In the presence of either TKI, CCR7 expression displayed considerable variability among donors, and only in some cases an increased percentage of CCR7+ cells was detected, particularly when using imatinib.
Imatinib and nilotinib affect the survival of monocytes differentiating toward macrophages in the presence of M-CSF
We also evaluated whether imatinib or nilotinib could affect the capacity of monocytes to differentiate toward macrophages in the presence of M-CSF.
Purified PB monocytes were differentiated toward M0 with M-CSF in the presence of TKIs. Cells were analyzed by flow cytometry for cell survival rate and, gating on viable cells, for the expression of surface markers typical of macrophages (Fig. 6A, Supplemental Fig. 3A, 3B). At drug concentration of 6 μg/ml most cells underwent apoptosis/necrosis. At 0.3 μg/ml, high percentages of viable cells were detected, which progressively acquired a phenotype typical of macrophages. In particular, cells reduced the expression of CD14 and HLA II, increased that of CD16, and de novo expressed NTB-A, B7-H3, CD204, and mIL-18 (Supplemental Fig. 3B) (34, 40).
We evaluated whether the reduced numbers of surviving and differentiating monocytes could correlate with a drug-mediated reduction of the expression of the M-CSF receptor (CSF1R) (Fig. 6B, Supplemental Fig. 3C). Purified PB monocytes were treated with imatinib and nilotinib for 24 h and analyzed by flow cytometry. Exposure to TKI concentrations of 6 μg/ml strongly decreased both the percentage of monocytes expressing M-CSFR and its surface density. On the contrary, treatment with 0.3 μg/ml did not significantly affect M-CSFR expression.
Effect of imatinib on immune cells from high-risk NB and CML patients
PBMCs were obtained from seven children with high-risk NB either at the disease onset or relapsing and analyzed for CXCR4 expression. In these patients, NK cells and monocytes constitutively expressed levels of the chemokine receptor comparable to those of healthy donors. In vitro treatment with imatinib (6 μg/ml) induced upregulation of CXCR4 expression in both cell types (Fig. 7A, Supplemental Fig. 4A).
We extended the analysis to PBMCs from three adult and two pediatric patients receiving imatinib for CML. CML NK cells and monocytes displayed a higher CXCR4 surface expression as compared with healthy donors (Fig. 7B, Supplemental Fig. 4B). The chemokine receptor surface density further increased when CML PBMCs were treated in vitro with imatinib at the concentration of 6 μg/ml (Fig. 7B, Supplemental Fig. 4B). Moreover, the treatment downregulated the expression of CCR1 and M-CSFR in monocytes (Fig. 7C, Supplemental Fig. 4B). Unlike what was observed in healthy donors (see Fig. 2A, Supplemental Fig. 1C), in vitro imatinib exposure did not decrease the expression of CXCR3 in CML NK cells (Supplemental Fig. 4B).
NK cells of CML patients were also analyzed for degranulation in the presence of the prototypical K562 target cell line (Fig. 7D, Supplemental Fig. 4C). The degranulation capability of resting NK cells did not significantly differ from that of healthy donors. Moreover, the in vitro treatment with imatinib did not affect the degranulation capability of resting and cytokine-stimulated CML NK cells. Notably, in some patients NK cells appeared to degranulate better in the presence than in the absence of the drug.
More than 20 TKIs are approved as cancer therapies for patients with Ph+ hematological malignancies, melanoma, breast cancer, non–small cell lung cancer, and colorectal cancer. Recently it has been demonstrated that imatinib provides clinical benefits in GISTs (18) and NB (15). In particular, in GISTs, benefits were associated with drug-dependent off-target effects, such as modulation of immune responses, and in NB a similar mechanism was postulated to occur. This prompted us to analyze the effect of two TKIs, namely imatinib and nilotinib, on NK cells, which represent pivotal cytolytic effectors in antitumor responses. Moreover, owing to the existence of a crosstalk between NK cells and macrophages (34), the analysis was extended to monocytes and macrophages, with particular attention to those characterized by tumor-promoting M2 polarization. TKI concentrations ≥15 μg/ml are higher than those achievable in adult patients in which plasma concentrations between 1 and 3 μg/ml are presently recommended (1). In pediatric patients, who showed higher tolerability than adults, plasma concentrations of imatinib ranged from 2 to 7 μg/ml, with lower values associated with better responses (15). We thus selected drug concentrations ≤6 μg/ml that, under our experimental conditions, were devoid of direct effects on NB cells and we analyzed their influence on phenotype and function of immune cells.
PB (resting) NK cells were highly resistant to the cytotoxic effect of the drugs, as <10% of cells died when exposed to imatinib and virtually all survived when exposed to nilotinib. Moreover, with the exception of nilotinib used at 6 μg/ml, NK cells maintained the ability to be activated by immunostimulatory cytokines, increasing the expression of activation markers and acquiring a degranulation capability in the presence of NB cells, comparable to that detected in the absence of drugs. In this context, in NB cells neither drug substantially modified the expression of ligands for the activating NK receptors (14) or induced upregulation of HLA class I and of other molecules involved in the immune checkpoints (38, 41, 42). Importantly, the drugs did not substantially alter the expression in NK cells of activating receptors, including DNAM-1 and NKp30, which are crucial for recognition and killing of both leukemia (23) and NB cells (29). Of note, a certain degree of variability in NKp30 surface expression upon TKI exposure was detected among donors. This could be explained by changes in the expression levels of the NKp30 receptor isoforms (a, b, and c) (43). This interpretive hypothesis is under further investigation, because it could correlate with differences in responses to drug treatments. Indeed, recent data showed a reduced response to imatinib in GIST patients with the NKp30 isoform c, which induces production of IL-10 instead of IFN-γ and TNF-α (18).
Importantly, both TKIs revealed a striking capacity to modulate the chemokine receptor repertoire of NK cells. In particular, they strongly increased CXCR4 expression in both resting and activated NK cells, and imatinib at the highest concentration used decreased that of CXCR3. Interestingly, increased CXCR4 expression upon TKI exposure was detected also in PB T and B lymphocytes, but it did not occur in NB cells. These data suggest that imatinib and nilotinib through CXCR3 downregulation and CXCR4 upregulation might disfavor the recruitment of NK cells toward peripheral tissues, while supporting their homing into BM niches. This might potentiate the NK immune surveillance against tumor cells that, as occur in stage 4 NB patients, metastasize in BM. It is of note that, different from what has been described for TGF-β (33), imatinib and nilotinib did not modify the expression of CX3CR1, which participates in the recruitment and extravasation of NK cells into BM.
Increased CXCR4 expression upon TKI exposure also occurred in PB monocytes and was accompanied by downregulation of CCR1, which participates in their extravasation in peripheral tissues. Monocytes were more susceptible than NK cells to drug exposure and, in particular, at 6 μg/ml, significant numbers of cells underwent apoptosis/necrosis with imatinib and up to 60% died with nilotinib. This effect was not prevented by stimulation with M-CSF, the major factor involved in monocyte differentiation toward macrophages (44). In keeping with the low survival rate and the reduced number of cells undergoing differentiation, in the presence of TKIs, monocytes strongly decreased the expression of the M-CSFR. These results are in line with previous data showing that M-CSF is fundamental for survival of monocytes (45, 46) and that imatinib impairs the signal transduction pathway of the M-CSFR (47). Our data suggest that imatinib and nilotinib could decrease in vivo the half-life of circulating monocytes, as well as their recruitment and macrophage differentiation in tissues, possibly reducing the generation of macrophages that in the tumor microenvironment might acquire M2 tumor-promoting function. Interestingly, it has been shown that imatinib inhibited the early phase of the differentiation of myeloid-derived suppressor cells and reduced their number in CML patients (48), whereas it did not affect differentiation toward dendritic cells (49).
Macrophages, both unpolarized (M0) and M2, were highly resistant to imatinib and nilotinib. A high number of cells survived to TKI exposure and maintained their phenotypic and functional characteristics, thus showing that drugs did not induce or modify macrophage polarization. Importantly, in the presence of TKIs, M2 maintained the capability to revert their immunosuppressive functional phenotype toward M1 after TLR engagement. Indeed, they released high amounts of proinflammatory and immunostimulatory cytokines. Moreover, with the exception of nilotinib used at 6 μg/ml, M1-polarized M2 macrophages acquired the capability of fully activating NK cells that increased the expression of the α-chain of the IL-2 receptor, the degranulation capability in the presence of NB, and released a high amount of IFN-γ, a cytokine crucial for amplification of both innate and adaptive Th1 antitumor immune responses.
Collectively, our results shed light on the influence of imatinib mesylate and nilotinib on NK cells and monocyte/macrophage survival, function, and expression of chemokine receptors, contributing both to better interpreting the efficacy of these agents in tumors and to envisaging strategies aimed at facilitating an antitumor immune response rather than to promoting a direct effect on tumor cells.
We thank Dr. Paul W. Manley (Novartis Institutes for BioMedical Research, Basel, Switzerland) for helpful discussions.
This work was supported by Associazione Italiana per la Ricerca sul Cancro Investigator Grant 15704 (to A.M.) and Special Program Molecular Clinical Oncology 5 per 1000 Grant 9962 (to A.M. and F.L.). F.B. is the recipient of Associazione Italiana per la Ricerca sul Cancro Fellowship IG.15704.
The online version of this article contains supplemental material.
Abbreviations used in this article:
chronic myeloid leukemia
gastrointestinal stromal tumor
the membrane-bound form of IL-18
tyrosine kinase inhibitor.
A.M. is a founder and shareholder of Innate Pharma (Marseille, France). The remaining authors have no financial conflicts of interest.