CCL2 is a KIT D816V–dependent modulator of the bone marrow microenvironment in systemic mastocytosis

Myeloproliferative neoplasms (MPNs) are clonal hematopoietic stem cell diseases characterized by an accumulation of mature and immature blood cells and aberrant tyrosine kinase signaling.¹  Increased production of cytokines has been implicated in disease progression and correlates with the symptom burden of patients with classical JAK2 V617F⁺ MPNs.^2-4  Furthermore, marked bone marrow (BM) fibrosis and increased microvessel density in MPN patients have been associated with aberrant cytokine expression and disease progression.^5,6  In addition, reduction of cytokine levels is a major effect of JAK2-targeting tyrosine kinase inhibitor treatment in classical MPNs.⁷  In contrast to the comprehensive data on cytokine expression in classical MPNs, only a few studies addressing the role of cytokines in systemic mastocytosis (SM) have been published.⁸ 

SM is a clonal disease of hematopoietic stem cells characterized by mast cell (MC) infiltration in the BM and/or other organs.⁹  In addition, BM microenvironment alterations, including increased angiogenesis, thickened bone trabeculae, fibrosis, and lymphocytic and eosinophilic infiltrations, are typically found in SM.¹⁰  Most patients present with the somatic KIT mutation D816V.^11,12  The D816V mutant exhibits constitutively active signaling and induces the recruitment of several downstream pathways, including signal transducer and activator of transcription 5 (STAT5), phosphoinositide 3-kinase (PI3K)/protein kinase B (AKT), the mitogen-activated protein kinase (MAPK), and the mammalian target of rapamycin (mTOR) pathways.^13-15  Mutant KIT is considered to be the major driver of pathogenesis in SM. In line with this notion, tyrosine kinase inhibitors targeting KIT D816V showed efficacy in the treatment of advanced SM.^9,16,17  The clinical manifestation of the disease is highly variable, ranging from an indolent course to highly aggressive variants with short survival. The World Health Organization (WHO) classification defines 4 major subtypes of SM that are based on clinical findings: indolent systemic mastocytosis (ISM), mastocytosis with an associated hematologic neoplasm (SM-AHN), aggressive systemic mastocytosis (ASM), and mast cell leukemia (MCL).^18-21  On the basis of clearly reduced survival and poor outcome compared with ISM, the variants SM-AHN, ASM, and MCL are collectively referred to as advanced SM.^22,23 

In contrast to classical MPNs, only a limited number of cytokines have been studied in SM. Interleukin-6 (IL-6) was shown to be increased in SM, and plasma levels of IL-6 correlate with disease severity and progression.^24-28  In particular, IL-6 was described to be increased in the plasma of SM patients and to correlate with serum tryptase levels, severity of symptoms, disease category, severity of BM pathology, organomegaly, the presence of osteoporosis, and the extent of skin involvement.^24-27  A recent study confirmed the prognostic significance of IL-6 in SM.²⁸  High levels of IL-6 were associated with disease severity and also with disease progression in patients with ISM.²⁸  IL-31 has been implicated in the induction of chronic skin inflammation and was also found to be increased in patients with SM.²⁹  We and others have shown that vascular endothelial growth factor and oncostatin M might contribute to changes of the BM microenvironment in SM.^30,31  However, other cytokines potentially triggering angiogenesis and fibrosis, which are involved in the complex interplay between neoplastic MC and stromal cells in the BM of patients with SM have not yet been studied.⁸ 

The chemokine (C-C motif) ligand 2 (CCL2), also referred to as monocyte chemotactic protein 1 (MCP-1), is a pleiotropic cytokine that recruits inflammatory cells to the sites of inflammation.³²  CCL2 primarily acts via the chemokine (C-C motif) receptor 2 (CCR2) that is expressed on monocytes and macrophages, memory T lymphocytes, natural killer cells, and also on various mesenchymal cells.^33-35  In particular, CCL2 has been shown to enhance angiogenesis and fibrosis under physiologic conditions as well as in inflammation and cancer.^35,36  Various solid tumors express CCL2 that in turn promotes tumor growth and metastasis mainly via its paracrine effect on the tumor microenvironment.³⁷  Consequently, targeting of CCL2 showed promising results in cancer models.^35,38-41  By contrast, only limited data are available regarding CCL2 in MPNs. Increased CCL2 serum levels have been described in patients with JAK2 V617F⁺ MPNs,^2,4,42  and MCs were also found to express CCL2.^43,44  However, CCL2 has not been studied in patients with SM, and targeting of CCL2 has not been addressed in MPNs.

In this study, we identified CCL2 as a KIT D816V–induced cytokine targeting microenvironmental cells in vitro and in vivo. Furthermore, high serum levels of CCL2 were found in patients with SM, and they correlate with advanced disease and poor survival.

Cell culture media and fetal calf serum (FCS) were obtained from Life Technologies (Carlsbad, CA). Recombinant human stem cell factor (SCF) and granulocyte-macrophage colony-stimulating factor were purchased from PeproTech (Rocky Hill, NJ), IL-6 from Miltenyi Biotec (Auburn, CA), and CCL2 from R&D Systems (Minneapolis, MN). Midostaurin and the PI3K/mTOR inhibitor BEZ235 were kindly provided by Dr E. Buchdunger and Dr P.W. Manley (Novartis Pharma AG, Basel, Switzerland). The MEK (defined as MAPK/extracellular signal-regulated kinase [ERK]) inhibitor PD98059 and the PI3K inhibitor LY294002 were purchased from Calbiochem (La Jolla, CA), and the broadly acting STAT inhibitor piceatannol and the inhibitor of κB (IκB) kinase inhibitor TPCA-1 from Tocris Bioscience (Bristol, United Kingdom).

The study was conducted in accordance with the Declaration of Helsinki and approved by the local institutional review board (EK#2010/566, EK#2011/404, EK#2014/1018, and EK#2014/1184). Forty-eight patients with mastocytosis were examined (Table 1). Mastocytosis was diagnosed according to WHO criteria.^18,19  Routine staging included serum tryptase measurement, BM histology, cytologic examination of BM cells, flow cytometry (CD2 and CD25 on BM MCs), and analysis for KIT D816V by melting point analysis after clamp polymerase chain reaction (PCR) and/or quantitative allele-specific PCR.^45,46  Microvessel density was determined by immunohistochemical CD34 staining and counting of all vessels within an examination area of 0.25 mm². Fibrosis was determined by reticulin staining according to the European consensus system (scale of 0 to 3).⁴⁷  Serum samples from age- and sex-matched healthy donors were used as a control cohort. Informed consent was obtained in each case.

TF-1 and Mo7e were cultured in RPMI 1640 medium with 10% FCS and granulocyte-macrophage colony-stimulating factor (5 ng/mL), and HMC-1 cells (KIT D816V⁺ HMC-1.2 clone)⁴⁸  were cultured in Iscove modified Dulbecco medium with 10% FCS. CD34⁺ progenitor cells were isolated from cord blood by magnetic-activated cell sorting (Miltenyi Biotec) and cultured in the presence of SCF and IL-6 as described.⁴⁹  Human umbilical vein endothelial cells (HUVECs) were isolated from umbilical cords according to published techniques and cultured in plates coated with fibronectin (31.25 μg/mL) in M199 medium with 20% FCS and supplements.⁵⁰ 

The coding sequences of wild-type KIT and KIT D816V were cloned into the lentiviral pWPI vector (kindly provided by Didier Trono, Department of Microbiology and Molecular Medicine, University of Geneva, Geneva, Switzerland). Recombinant lentiviruses were produced as described.³¹  TF-1 and Mo7e cells were transduced with wild-type KIT, KIT D816V, or the empty vector and sorted for green fluorescent protein (FACSAria; Becton Dickinson, San Jose, CA).

For knockdown of KIT, a lentiviral vector pLKO.2 encoding a short hairpin RNA targeting KIT was used.³¹  For knockdown of CCL2 or nuclear factor κB1 (NF-κB1), hairpins in a microRNA-E (miR-E) backbone were cloned into the pRRL-based SGEP vector as described.⁵¹  Guide sequences of the hairpins are provided in supplemental Table 1, available on the Blood Web site. Production of recombinant vesicular stomatitis virus G glycoprotein pseudotyped lentiviruses and transduction of target cells was performed as described.⁵² 

Real-time PCR for the initial cytokine screen was performed as described.⁵²  Results were expressed as ΔCt values (ΔCt = Ct_ABL − Ct_cytokine). In addition, CCL2 messenger RNA (mRNA) copy numbers were absolutely quantified on a Biorad Real-Time PCR System (Biorad, Foster City, CA) using iTaq SYBR Green Supermix with ROX (Biorad, Hercules, CA) and plasmid standards. CCL2 copy numbers were normalized to Abelson murine leukemia viral oncogene homolog 1 (ABL1) copy numbers and expressed as percent of ABL.⁵³  Primer sequences are provided in supplemental Table 2.

To determine the levels of CCL2 protein in cell culture supernatants, an enzyme-linked immunosorbent assay (ELISA; R&D Systems) was used. The lower limit of detection was 50 pg/mL. Patients’ sera were analyzed by a commercial precoated ELISA (R&D Systems) according to the recommendations of the manufacturer. The lower limit of detection was 31.3 pg/mL.

Whole-cell extracts were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis, and immunoblotting was carried out by using antibodies against KIT, phospho-KIT, and β-actin as described.⁵⁴  Cytoplasmatic and nuclear extracts were isolated by using NE-PER Nuclear and Cytoplasmic Extraction Kit (Thermo Scientific, Waltham, MA) and probed with antibodies against NF-κB, IκB, phospho-IκB, glyceraldehyde-3-phosphate dehydrogenase, and histone H3. A description of antibodies used in this study is provided in supplemental Table 3.

Surface expression of CCR2 was determined by using a brilliant violet 421-conjugated CCR2 antibody (clone K036C2, BioLegend) or an isotype matched control antibody. Cells were analyzed by flow cytometry on a FACSCanto II (Becton Dickinson) with FlowJo software (Tree Star, Ashland, OR).⁴⁹  CD117⁺⁺ primary MCs were enriched from BM aspirates by fluorescence-activated cell sorting.

Immunohistochemistry was performed on paraffin-embedded, formalin-fixed BM sections with a Ventana Benchmark Ultra stainer (Ventana, Tucson, AZ) with extended heat-induced epitope retrieval with CC1 buffer and the ultraView Universal DAB Detection Kit (Ventana). Xenotransplant tumor sections were stained with hematoxylin and eosin, chromotrope-aniline blue (CAB), or by immunohistochemistry for Ki-67 or von Willebrand factor. Fibrosis and inflammation in the tumors was graded from 0 to 4. A description of antibodies used in this study is provided in supplemental Table 3.

Polycarbonate membranes (8 μM pore size, 0.7 cm² growth area, Corning Incorporated, Corning, NY) were coated with fibronectin (31.25 μg/mL). The lower chambers were filled with medium with or without supernatant from HMC-1 cells. In selected experiments, conditioned medium was pre-incubated with anti-CCL2 antibody (mouse immunoglobulin G1, clone 24822, R&D Systems) or an isotype control antibody for 1 hour at 37°C. HUVECs (2 × 10⁴) were loaded into the top chamber of each well and incubated for 4 hours. Cells attached to the lower side of the membrane were fixed on ice with methanol, washed with phosphate-buffered saline, and stained with 4′,6-diamidino-2-phenylindole. The number of migrated cells was evaluated by fluorescence microscopy (Axio Imager A1 microscope; Zeiss, Oberkochen, Germany).

HUVECs were grown to confluence in 24-well plates, and a scratch was introduced to the monolayer. Medium with or without supernatant from HMC-1 cells was added to the cell layer. In selected experiments, conditioned medium was pre-incubated with anti-CCL2 antibody as described above. Cell layers were examined after 24 hours of incubation with an Axiovert 25 inverted microscope (Zeiss). Migrated cells were counted by using ImageJ software.⁵⁵  Results were expressed as percent of control.

Animal studies were approved by the local institutional review committee for animal research (approval No. GZ 68.205/0218-II/3b/2012). In brief, NOD-SCID IL-2Rgamma-null (NSG) mice (The Jackson Laboratory, Bar Harbor, ME) were subcutaneously injected with HMC-1 cells (1 × 10⁷) after lentiviral knockdown of CCL2 (n = 6) or transduction with a nontargeting control vector (n = 5). Each mouse received 2 independent injections into each lower flank. After 3 weeks, tumor nodules were palpable, and tumor size was measured every other day with a slide caliper and calculated by using the formula a² × b/2 where a is the tumor length and b is the tumor width. After 34 days, mice were euthanized and subcutaneous tumors were weighed, formalin fixed, and paraffin embedded.

Data were analyzed by using GraphPad Prism software (GraphPad Software, La Jolla, CA). To compare differences, Student t test (for 2 groups) or one-way analysis of variance (followed by Bonferroni-adjusted post hoc analysis for more than 2 groups) was applied for parametric data, and the Mann-Whitney U test (for 2 groups) or the Kruskal-Wallis test (followed by Dunn’s post hoc analysis for more than 2 groups) was applied for nonparametric data. The survival was estimated by the Kaplan-Meier method and the log-rank test. Results were considered significantly different when the P value was <.05. The reference range for CCL2 was calculated from the healthy control cohort as mean ± 2 standard deviations (SD).

To screen for KIT D816V–dependent expression of cytokines potentially involved in the pathogenesis of SM, expression levels of 43 cytokines were determined in 2 growth factor–dependent human cell lines, Mo7e and TF-1, after lentiviral transduction of KIT D816V (Figure 1A). KIT D816V induced expression of OSM, LIF, IL-8, TNF-α, VEGF, CCL2, and CCL5 in at least 1 cell line model (Figure 1B). In this study, we addressed the role of CCL2 in the pathogenesis of SM. KIT D816V significantly induced CCL2 mRNA expression (Figure 1C) and CCL2 protein secretion (Figure 1D). In line with this finding, CCL2 mRNA levels increased substantially during SCF-mediated differentiation of cord blood–derived MCs (Figure 1E). Moreover, the highest CCL2 levels were observed in the human neoplastic MC line HMC-1 harboring the KIT D816V mutation (Figure 1E). To confirm that CCL2 expression in neoplastic MCs is dependent on KIT D816V, HMC-1 cells were treated with midostaurin, a multikinase inhibitor reportedly suppressing the kinase activity of mutant KIT. Midostaurin was found to reduce expression of CCL2 in HMC-1 cells in a dose-dependent manner (Figure 1F). Similar effects were observed after RNA interference (RNAi) –mediated knockdown of KIT in HMC-1 cells (Figure 1G). Together, these data show that KIT D816V promotes expression of CCL2.

To assess the role of signaling pathways downstream of KIT D816V on CCL2 expression, pharmacologic inhibitors targeting key signaling molecules known to be activated by KIT D816V, namely MAPK, PI3K, and STAT5, were applied. Piceatannol, a compound directed against multiple targets, including STAT5, the MAPK inhibitor PD98059, and the rather unspecific PI3K inhibitor LY294002 reduced expression of CCL2 in HMC-1 cells (Figure 2A). In contrast, the more specific PI3K inhibitor BEZ235 showed no effect (Figure 2A). This suggests that LY294002 does not inhibit CCL2 expression via the PI3K pathway. Piceatannol and LY294002 have also been shown to affect NF-κB signaling.^56,57  Therefore, a specific NF-κB inhibitor, TPCA-1, was used to analyze the dependency of CCL2 expression on NF-κB in HMC-1 cells. Indeed, treatment of HMC-1 cells with TPCA-1 reduced the nuclear fraction of NF-κB and also the phosphorylation of the NF-κB inhibitor IκB (Figure 2B) and resulted in a dose-dependent reduction of CCL2 expression (Figure 2C). In addition, we silenced expression of NF-κB1 in neoplastic MCs by RNAi. Knockdown of NF-κB1 significantly reduced expression of CCL2, confirming the specificity of the effect (Figure 2D). Complementary to this observation, lentiviral expression of KIT D816V resulted in activation of the NF-κB pathway in TF-1 cells, as demonstrated by nuclear translocation of NF-κB and phosphorylation of IκB (Figure 2E). Together, our data show that KIT D816V promotes expression of CCL2 through activation of NF-κB in neoplastic MCs.

CCL2 has been reported to promote angiogenesis and thus may be important for the BM alterations associated with SM. Therefore, we analyzed the effect of conditioned medium from HMC-1 cells on HUVECs. As expected, HUVECs were found to express the CCL2 receptor CCR2 (Figure 3A) and to migrate against recombinant CCL2 in a modified Boyden chamber assay. Conditioned medium from neoplastic MCs also induced migration of HUVECs, and pre-incubation with a neutralizing antibody against CCL2 significantly reduced their migratory response (Figure 3B). Furthermore, the supernatant of KIT D816V⁺ HMC-1 cells enhanced migration of HUVECs in a wound-healing assay. Again, a neutralizing antibody against CCL2 abolished this effect (Figure 3C-D). In addition, we silenced expression of CCL2 in neoplastic MCs by RNAi to further confirm specificity. Knockdown of CCL2 reduced the effect of conditioned medium of neoplastic MCs to the baseline level in the wound-healing assay (Figure 3E; supplemental Figure 1). Together, these data suggest that CCL2 might contribute to BM microenvironment alterations in SM patients.

The impact of CCL2 on angiogenesis and tumor growth in vivo was assessed in a xenotransplantation model using NSG mice. Subcutaneous injection of HMC-1 cells with or without knockdown of CCL2 resulted in the formation of tumors within 23 days. Tumor volumes were measured between day 23 and 34 and showed a significant difference between CCL2-expressing and CCL2-knockdown cells (Figure 4A). After 34 days, tumors were isolated and analyzed for weight and microarchitecture. Tumors of HMC-1 cells after CCL2 knockdown were significantly lighter in comparison with the controls (Figure 4B). In addition, tumors showed a different microarchitecture after knockdown of CCL2 with a reduction in necrotic areas and leukocyte infiltration, a lower proliferation rate as determined by staining with Ki-67, and less tumor stroma (Figure 4C-G). In particular, CAB staining showed a striking difference in collagen deposition in the tumors (Figure 4E,H), and immunohistochemistry for von Willebrand factor indicated a reduced microvessel density after knockdown (Figure 4F,I). Overall, silencing of CCL2 expression in neoplastic MCs resulted in a more benign tumor stroma with concomitant reduction of tumor cell proliferation and tumor mass in vivo.

To study the significance of our findings in human disease, CCL2 serum levels were measured in a cohort of patients with mastocytosis (Table 1). Patients with mastocytosis showed significantly higher levels of CCL2 (450.9 ± 364.9 pg/mL [mean ± SD]; n = 48) compared with controls (282 ± 92.9 pg/mL [mean ± SD]; P = .0027) (Figure 5A). To investigate whether CCL2 was expressed by neoplastic MCs, we examined BM sections obtained from patients with SM. As assessed by immunohistochemistry, tryptase-positive spindle-shaped neoplastic MCs were found to react with anti-CCL2 antibody (Figure 5C-D). Finally, highly purified (fluorescence-activated cell sorting) primary neoplastic MCs of patients with advanced SM (n = 2) expressed substantial amounts of CCL2 by quantitative PCR (ASM, 1783 ± 26; MCL, 337 ± 84; CCL2 mRNA levels percent of ABL [mean ± SD]).

CCL2 serum levels in advanced SM were significantly higher compared with indolent SM and cutaneous mastocytosis (Figure 5B). CCL2 levels showed only modest correlation to serum tryptase levels, MC infiltration in the BM, KIT D816V allele burden, and BM fibrosis (supplemental Figure 2). In addition, no difference in CCL2 levels was observed when comparing patients with or without mediator-related symptoms (supplemental Figure 2). We next examined the impact of serum CCL2 levels on survival and used the upper reference range as a cutoff to separate two prognostically distinct groups. Mastocytosis patients with CCL2 levels above 468 pg/mL (n = 15) showed a median survival of 2.8 years, whereas the median survival was not reached in the group of patients with CCL2 levels below 468 pg/mL (n = 33; log-rank test, P < .001; Figure 5E). In the group of ISM patients, a significant survival difference was observed (P = .0011; Figure 5F). Together, our data show that high CCL2 is associated with advanced disease and poor survival in mastocytosis.

The activating KIT D816V mutation has been implicated in the pathogenesis of SM and represents a minor diagnostic SM criterion. However, mechanisms underlying the biological effects of the mutated KIT receptor are still not fully understood. The important interplay of neoplastic (stem) cells and altered tumor microenvironment is well known in solid cancers and has also been addressed in classical MPNs. Interactions of neoplastic cells and their microenvironment contribute to therapy resistance, tumor stem cell homeostasis, metastasis, aggressive tumor growth, and paraneoplastic symptoms, among other things. Data focusing on the role of the microenvironment in SM are limited. Cytokines have been shown to affect the microenvironment in SM and are considered a major driver of disease progression and clinical symptoms in JAK2 V617F⁺ MPNs.³¹ 

KIT D816V activates various downstream signaling pathways that modulate gene expression and drive increased cytokine expression. We identified NF-κB signaling to be critical for KIT D816V–dependent expression of CCL2. CCL2 is a well-known NF-κB target gene.^58,59  However, NF-κB has not been prominently discussed as a major signaling pathway downstream of KIT D816V in SM. Tanaka et al⁶⁰  showed constitutive activation of NF-κB signaling in KIT D816V⁺ HMC-1 cells. These data are in line with our observation of the pivotal role of NF-κB signaling for CCL2 expression and suggest a pathophysiological relevance for NF-κB signaling in SM. In addition, targeting strategies for NF-κB in SM have already been suggested.⁶⁰ 

Our in vitro data clearly show that KIT D816V triggers expression of CCL2. Although the mutation is found in >90% of SM patients, marked differences in CCL2 serum levels between patients were observed. The number of MCs, the expression of CCL2 in other (neoplastic) cells, or additional genetic mutations in MCs might influence the CCL2 serum level. First, we observed only a loose correlation between CCL2 and surrogate parameters of the MC burden, such as serum tryptase and the degree of MC infiltration in the BM. Second, several cell types are capable of expressing CCL2, and increased CCL2 has been described in various myeloid neoplasms.^4,42  Multilineage involvement of KIT D816V being present not only in neoplastic MCs but also in clonal myeloid cells is a common finding in advanced SM.⁶¹  Thus, it is likely that clonal non-MCs contribute to increased CCL2 levels, especially in SM-AHN. Third, recent sequencing studies in patients with advanced SM have shown several additional somatic mutations in addition to KIT D816V.⁶²  Additional studies that investigate the mechanisms of cooperation between KIT D816V and additional oncoproteins to elucidate their joint effect on cytokine expression, microenvironment alterations, and pathogenesis in SM are quite interesting.

With regard to the clinical relevance of CCL2 serum measurement, we observed increased CCL2 levels in patients with mastocytosis and the highest levels in patients with advanced SM. Thus, measurement of CCL2 might contribute to risk assessment in SM. Comparable data have been published for other cytokines.²⁸  The observed difference in overall survival might be partly explained by the enrichment of patients with advanced SM in the group with high CCL2 levels. However, a significant survival difference was also observed in the group of ISM patients. Thus, our data indicate an association of high CCL2 levels with advanced disease and poor survival in mastocytosis. Nevertheless, the clinical significance of CCL2 measurement in addition to established risk factors as well as other cytokine profiles needs to be assessed in forthcoming studies that include larger numbers of patients.

We found that MC-derived CCL2 promotes migration of endothelial cells and alterations of the tumor microenvironment in SM in vitro and in vivo. In particular, tumors showed vascularization and fibrosis, and profound effects on the microenvironment were observed after knockdown of CCL2. Thus, valuable data regarding the interaction of neoplastic MC and tumor microenvironment could be obtained. Currently, other human MC lines, such as the recently described ROSA and MCPV, are being investigated in our laboratory.^49,63  Preliminary data indicate that these MC lines produce CCL2-positive tumors as well. Additional in vivo studies on the interaction of neoplastic MCs and BM cells are warranted to further increase our understanding of the pathophysiologic role of CCL2 in SM.

The concept of targeting CCL2 in SM is in line with data from several studies showing a significant increase in survival and inhibition of metastasis by targeting CCL2 in solid cancers.^35,38-40  The effect can be partly explained by direct effects on endothelial cells and fibroblasts. In addition, targeting of CCL2 also reduces infiltration with tumor-associated macrophages and thus indirectly affects other cells of the tumor microenvironment.⁶⁴  Although targeting of CCL2 showed promising results in various tumor models, one recent study raised concerns about the acceleration of breast cancer metastasis by promoting angiogenesis after cessation of CCL2 inhibition.⁴¹  This further strengthens the important role of CCL2 in tumor progression but needs to be carefully addressed when applying this targeting strategy to the patients. Our data indicate that targeting of cytokines, including CCL2, might also be a promising strategy in advanced SM, complementary to the direct targeting of KIT D816V.

In summary, the chemokine CCL2 promotes alterations of the tumor microenvironment in SM in vitro and in vivo. High serum levels of CCL2 in patients correlate with advanced disease and poor survival. Our data imply that CCL2 is another important player in the complex interplay of neoplastic MCs and BM microenvironment and thus a potential therapeutic target in SM.

The authors thank Eva Schrefl and Sophie Frank (Department of Laboratory Medicine), Günther Hofbauer and Andreas Spittler (Cell Sorting Core Facility), and Sabine Rauscher and Marion Gröger (Imaging Core Facility) for technical support. They also thank Helmuth Haslacher, Thomas Perkmann, and Peter Quehenberger (Department of Laboratory Medicine) for contributing serum samples for the control cohort and Gertrude Krainz for proofreading the manuscript (Department of Pathology, Medical University of Vienna, Vienna, Austria for all).

This study was supported by Austrian National Science Fund (FWF) projects P26079-B13 and F4704-B20. L.K. has received funding from FWF projects P26011 and P29251 and MSCA-ITN-2015-ETN ALKATRAS, No. 675712.

Contribution: G.G., N.W., A.B., K.S., G.E., S.R., J.Z., L.K., V.S., M.M., and G.H. performed in vitro and in vivo experiments and analyzed the data; G.G., A.-I.S, T.P.-K., L.M., W.R.S., P.V., and G.H. obtained and analyzed clinical data; G.G., P.V., M.M., and G.H. designed the study and wrote the paper; and all authors revised and approved the manuscript.

Conflict-of-interest disclosure: P.V. served as a consultant in a global Novartis trial investigating the effects of midostaurin in patients with advanced systemic mastocytosis and received honoraria and research grants from Novartis, Blueprint, and Deciphera. G.H. and W.R.S. received honoraria from Novartis. The remaining authors declare no competing financial interests.

Correspondence: Gregor Hoermann, Department of Laboratory Medicine, Medical University of Vienna, Waehringer Guertel 18-20, Leitstelle 5H, A-1090 Vienna, Austria; e-mail: gregor.hoermann@meduniwien.ac.at.