KIT receptor is required for mast cell development, survival, and migration toward its ligand stem cell factor (SCF). Many solid tumors express SCF and this leads to mast cell recruitment to tumors and release of mediators linked to tumor angiogenesis, growth, and metastasis. Here, we investigate whether FES protein-tyrosine kinase, a downstream effector of KIT signaling in mast cells, is required for migration of mast cells toward SCF-expressing mammary tumors. Using a novel agarose drop assay for chemotaxis of bone marrow–derived mast cells (BMMC) toward SCF, we found that defects in chemotaxis of fes-null BMMCs correlated with disorganized microtubule networks in polarized cells. FES displayed partial colocalization with microtubules in polarized BMMCs and has at least two direct microtubule binding sites within its N-terminal F-BAR and SH2 domains. An oligomerization-disrupting mutation within the Fer/CIP4 homology-Bin/Amphiphysin/Rvs (F-BAR) domain had no effect on microtubule binding, whereas microtubule binding to the SH2 domain was dependent on the phosphotyrosine-binding pocket. FES involvement in mast cell recruitment to tumors was tested using the AC2M2 mouse mammary carcinoma model. These tumor cells expressed SCF and promoted BMMC recruitment in a KIT- and FES-dependent manner. Engraftment of AC2M2 orthotopic and subcutaneous tumors in control or fes-null mice, revealed a key role for FES in recruitment of mast cells to the tumor periphery. This may contribute to the reduced tumor growth and metastases observed in fes-null mice compared with control mice. Taken together, FES is a potential therapeutic target to limit the progression of tumors with stromal mast cell involvement. Mol Cancer Res; 10(7); 881–91. ©2012 AACR.
This article is featured in Highlights of This Issue, p. 857
There is increasing evidence that during tumor progression, functional interactions occur between tumor cells and immune cells within tissues that can either enhance or suppress tumor growth and metastasis (1). Although immune cells have long been known to infiltrate solid human tumors, their precise roles and prognostic or therapeutic value remains controversial (2). Studies in immunocompromised mouse models have shown that the immune system protects against outgrowth of both spontaneous and carcinogen-induced primary tumors (3–6). In contrast, recruitment of innate immune cells such as macrophages and mast cells to the tumor microenvironment has been linked to creating an inflammatory state driving tumor progression (7–9). Increased density of tumor-associated macrophages (TAM) has been linked to increased tumor vasculature and metastasis in multiple cancer types (10). Mast cells also accumulate at the periphery of tumors and release a variety of mediators that regulate tumor angiogenesis, immunosuppression, and tumor growth (11). Depending on tumor type, stage, and tissue microenvironment, mast cells can either enhance or suppress tumor growth (8). Interestingly, one recent study shows that mast cell targeting can reduce onset of prostate adenocarcinoma, but also leads to development of more highly malignant neuroendocrine tumors in these mice (12). Given the plethora of mast cell mediators that may be produced in the tumor microenvironment, more targeted therapeutic approaches may need to be developed.
KIT receptor (CD117) and its ligand stem cell factor (SCF) initiate critical signals for the differentiation of mast cells from hematopoietic progenitors (13). The SCF/KIT signaling axis remains critical to survival and migration of mature mast cell following homing to connective tissues or the mucosal lining of the airways or gastrointestinal tract (14, 15). Within these tissues, mast cells serve as sentinels against invading pathogens but are also key effector cells for allergic inflammation (16). There is also growing understanding of the contributions of stromal mast cells to tumor progression and immunosuppression. Many epithelial cell–derived cancer cell lines and derived tumors express SCF, and this provides a chemoattractant signal for KIT-expressing mast cells in vitro and in vivo (17, 18). Upon mast cell recruitment, KIT signaling leads to matrix metalloproteinase-9 (MMP-9) production which generates more soluble SCF by cleaving the transmembrane SCF isoform, thus further amplifying mast cell recruitment to the tumor microenvironment (18). Mast cell–derived MMP-9, VEGF, and other mediators can enhance outgrowth of early stage tumors, but may become dispensible in more advanced tumors (19). Thus, interrupting the recruitment of mast cells by targeting the SCF/KIT signaling axis may explain some of the beneficial effects of KIT inhibitor treatments (e.g., imatinib) on non-KIT–expressing tumors (20). Another attractive therapeutic target is the nonreceptor protein-tyrosine kinase (PTK) FES (21), which functions in promoting KIT-driven mast cell migration in vitro (22, 23), and TAM recruitment to mammary tumors in vivo (24).
FES (also known as Fps) is a PTK that is highly expressed in innate immune cells, and signals downstream of receptors for cytokines, growth factors, IgE, and bacterial products to regulate innate immunity (25–27). FES, and the related kinase FER, have a unique domain organization among PTKs, including the N-terminal Fer/CIP4 homology-Bin/Amphiphysin/Rvs (F-BAR), and extended F-BAR (FX) domains, that mediate membrane targeting and oligomerization (28, 29). These domains are followed by SH2 and PTK domains that interact with one another to regulate FES kinase activity (30). FES kinase activity may also be regulated by in cis interactions involving the N-terminal domain, as suggested by an oligomerization-disrupting mutation (L145P) within the F-BAR domain that causes hyperactivation of FES and increased localization to microtubules (31). Although the role of FES targeting to microtubules is not well defined, mutation of the phosphotyrosine (pY)-binding pocket within the SH2 domain was found to largely disrupt FES localization to microtubules (31). The FX domain of FER and FES also promotes kinase activation via binding phosphatidic acid produced by phospholipase D2 to promote cell motility and myeloid differentiation, respectively (32, 33). In leukemias driven by oncogenic KIT receptors (e.g., D816V), FES was identified as a key effector of growth and survival signaling and a potential therapeutic target (34). Complicating this possibility are findings suggesting both tumor promoting and suppressing roles of FES, which may depend upon the tumor type and stage (24, 35–37).
SCF-induced dimerization of wild-type KIT receptor triggers FES activation, which involves binding of the FES SH2 domain to the phosphorylated juxtamembrane region of the KIT receptor (22, 23). In mast cells, FES activation leads to filamentous actin (F-actin) reorganization and phosphorylation of actin regulatory protein HS1 (22). Compared with control bone marrow–derived mast cells (BMMC), fes-null (fes−/−) BMMCs show defects in SCF-induced cell adhesion, polarization, and motility (22). Because the SCF/KIT signaling axis is central to the recruitment of mast cells to tumors, and linked to tumor progression, we tested the role of FES in tumor-associated mast cells. Here, we provide further insights into regulation and function of FES in mast cell polarization and motility toward SCF, including SCF produced by a metastatic mammary carcinoma model. Tumor engraftment studies support a key role for FES in mast cell recruitment to mammary tumors, and suggest that FES expression in mast cells contributes to tumor progression in this mouse model via stromal effects.
Materials and Methods
The previously reported fes-null (fes−/−) strain (27), was maintained on a 129/SvJ background for BMMC studies. Control and fes−/− mice were also maintained on a nu/nu (NCr-Foxn1nu/nu) background for tumor studies, as previously reported (24). Mice were housed in the Animal Care Facility at Queen's University (Kingston, ON, Canada). All procedures carried out were according to the guidelines of the Canadian Council on Animal Care, with the approval of the university animal care committee.
Commercial reagents included: recombinant murine SCF (Peprotech), anti-SCF neutralizing antibody (Millipore), KIT inhibitor (SU11652, Calbiochem), CellTracker green and CellTracker orange (Invitrogen), anti-α/β-tubulin (Cell Signaling Technology), Alexa Fluor 488–conjugated goat anti-rabbit IgG (Invitrogen), Alexa Fluor 633–conjugated goat anti-mouse IgG (Invitrogen), bovine fibronectin (Roche Diagnostics), normal goat serum (Sigma-Aldrich), and porcine tubulin (Cytoskeleton). Noncommercial reagents included rabbit anti-FES (38).
The highly metastatic AC2M2 mouse mammary carcinoma cell line (39) was cultured as described (24). BMMC cultures were established by culturing femoral bone marrow cells from wild-type (fes+/+) and fes−/− mice (129/SvJ background; 4–8 weeks of age) in the presence interleukin (IL)-3 and SCF (5% conditioned medias), as previously described (22). After 4 weeks of culturing nonadherent cells, BMMC maturity was assessed using flow cytometry to detect surface expression of FcϵRI and Kit as described (22).
In the adhesion and migration assays, endogenous FES and α/β-tubulin immunofluorescence staining were carried out on fes+/+ and fes−/− BMMCs attached to fibronectin-coated coverslips (20 ng/mL). Cells were fixed with 4% paraformaldehyde, permeabilized with 0.2% Tween-20, blocked using 5% goat serum, and incubated with rabbit anti-FES/FER (1:50) and mouse anti-α/β-tubulin (1:50) at 4°C overnight. Following extensive washing in PBS/0.2% Tween-20, coverslips were incubated with Alexa Fluor 488–conjugated goat anti-rabbit IgG (1:400) and Alexa Fluor 633–conjugated goat anti-mouse IgG (1:400) at room temperature for an hour. After washing 3 times with PBS/0.2% Tween-20, coverslips were mounted on slides and analyzed by confocal microscopy (HCX PL APO DIC 63×/1.32 Oil CS objective; Leica TCS SP2 MultiPhoton).
Glutathione S-transferase fusion protein expression and purification
Plasmids encoding glutathione S-transferase (GST) fusions to FES N-terminal (residues 1-459) and C-terminal (460-822) fragments of FES were previously described (29, 37). The L145P and R483E mutations were generated in pGEX-FES 1–459 and pGEX-FES 460–822 plasmids using the QuikChange Site-Directed Mutagenesis Kit (Invitrogen) and verified by sequencing. All GST fusion proteins were expressed and purified from BL21 cells using glutathione-Sepharose beads according to the manufacturer's instructions (GE Healthcare Ltd.). All purified GST fusion proteins were dialyzed in PBS, and aggregates removed by ultracentrifugation (60,000 × g) before experiments.
Soluble porcine tubulin purchased from Cytoskeleton (Cat#T240) was adjusted to 1 mmol/L MgGTP, depolymerized on ice for 20 minutes, clarified by centrifugation at 13,300 × g for 30 minutes and the supernatant fraction was allowed to repolymerize in a 34°C water bath for 20 minutes. At this point 10 μL of PME buffer (10 mmol/L Pipes pH6.9, 5 mmol/L MgCl, 1 mmol/L EGTA, 0.1 mmol/L MgGTP, 1 mmol/L DTT) supplemented with 40 μmol/L taxol was added and the mixture was incubated for an additional 20 minutes. Polymerized microtubules were pelleted by centrifugation 13,300 × g for 30 minutes at room temperature and microtubules were resuspended in ATPase buffer (20 mmol/L HEPES pH with KOH to 7.2, 5 mmol/L MgAc, 0.1 mmol/L EGTA, 0.1 mmol/L EDTA, 25 mmol/L KAc, 1 mmol/L DTT). For analysis of FES–microtubule interactions, microtubules (0–5 μmol/L) were incubated with 4 μmol/L FES protein in ATPase buffer (100 μL final volume). Bound proteins were pelleted by ultracentrifugation at 100,000 × g for 1 hour and both supernatant and pellet fractions were analyzed separately by SDS-PAGE and stained with Coomassie brilliant blue. Gel images were scanned, and densitometry was carried out using Corel Photopaint imaging software (histogram function).
Analysis of SCF expression in AC2M2 cells
RNA was extracted from AC2M2 and 4T1 mouse mammary carcinoma cells (3 × 106) using TRIzol according to the manufacturer's instructions (Gibco-BRL). The extracted RNA was immediately converted to cDNA using SuperScript II Reverse Transcriptase (RT) according to manufacturer's instructions (Invitrogen). Mouse SCF gene expression was assessed by PCR using primers that amplify nucleotide 436–1,002 of murine SCF (forward: 5′-GCTTGACTACTCTTCTGGAC-3′ and reverse: 5′-CTCTCTCTTTCTGTTGCAAC-3′), as described previously (18). Primers for β-actin were previously described (40). The PCR reactions were set up using 100 ng cDNA, 5 pmol primers, Taq polymerase, and the PCR mixture (2.5 μL 10× PCR buffer, 1.25 μL, 50 mmol/L MgCl2, 0.5 μL of 10 mmol/L dNTP mix) made to a final volume of 25 μL with double distilled water. PCR was carried out at annealing temperature of 65°C for 40 cycles using an iCycler (Bio-Rad).
Conditioned medium from AC2M2 cells (30 × 106 cells, 3 days, passaged 1:5). Immunoblot analysis of SCF expression was conducted on AC2M2 conditioned medium using anti-SCF (Millipore, 1:1,000). Recombinant SCF was used as a positive control.
AC2M2 conditioned medium–induced migration of BMMCs
Transwell filters (3-μm pore; BD Biosciences) were coated with 20 μg/mL fibronectin for 2 hours at 37°C, rinsed once with PBS, and used for the migration assays with starved BMMCs for 4 hours. Cytokine-starved BMMCs (4 × 105) were added to the upper chambers and the AC2M2 conditioned medium was added to wells below. To assess the contribution of the SCF/Kit axis, anti-SCF neutralizing antibody was added to the AC2M2 conditioned medium in the lower well (1 μg/mL), or Kit inhibitor (1 μmol/L) was added to the starved BMMCs in the upper Transwell chambers. After incubation at 37°C for 4 hours, cells on the upper surface of the membrane were removed with a cotton swab and migrated cells on the bottom of the membrane were counted. The cells were then fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100/PBS and 4′,6-diamino-2-phenylindole (DAPI; 300 nmol/L). BMMCs were visualized by fluorescence microscopy, and the number of cells was recorded for 4 fields of view per membrane at ×10 magnification to calculate relative migration. The experiments were all done in triplicate.
BMMC agarose drop migration assays
Agarose drop chemotaxis assays were previously described for adherent cancer cells (41) and were adapted for use with BMMCs as described here. BMMCs were starved of IL-3 and SCF O/N in BMMC starvation medium [Iscove's Modified Dulbecco's Medium, 10% (vol/vol) FBS, 1% (vol/vol) antimicrobial-antimycotic solution, 1% (vol/vol) glutamine, 1 mmol/L sodium pyruvate, 1% (vol/vol) nonessential amino acids, 50 μmol/L α-monothioylglycolate]. In the agarose drop assays, coverslips were coated with 20 ng/mL fibronectin O/N at room temperature, and afterwards a prewarmed 0.5% agarose drop containing SCF (30 ng/mL) was added to the center of the coverslip (10 μL) and placed at 4°C for 15 minutes. BMMCs (5 × 105 cells) were added to each well and allowed to migrate 16 to 18 hours. Afterwards, the cells were fixed and microtubules were stained as described earlier.
For live-cell imaging, wild-type and fes−/− mast cells were stained with CellTracker green or CellTracker orange (Invitrogen; 10 μmol/L), respectively. After 45 minutes, BMMCs were resuspended in cytokine-free starvation medium and were used in migration assays within 24 hours. Stained cells were mixed at a ratio of 1:1 (106 cells in total) and seeded on a coverslip with an SCF-containing agarose drop. Images in green and red channels were acquired at the edge of the agarose drop every 2 minutes for 8 to 16 hours using a WaveFX spinning disc confocal microscope (×10 objective, 21 z-axis slices; Quorum Technologies Inc.). Images and videos were compiled using Metamorph Software and migration analyses conducted using ImageJ manual tracking plug-in (NIH, Bethesda, MD).
Tumor engraftment models and mast cell staining
AC2M2 cells (7,500 cells) were injected into either the fourth mammary gland of age-matched, female wild-type and fes−/− mice (nu/nu background) as described previously (24) or injected subcutaneously (50,000 cells in PBS) in the lower flank of the back. Tumor growth at the injection site was measured by ultrasound imaging (VisualSonics Vevo 770) at 16, 19, and 21 or 23 days after engraftment. On days 21 to 23, the mice were euthanized and the tumors along with surrounding stromal tissue were removed, formalin-fixed, and paraffin-embedded. Sections were cut (4 μm), deparaffinized with toluene and rehydrated through an ethanol gradient, followed by staining with 1% Alcian blue (pH 2.5 in 3% acetic acid) and 0.1% nuclear fast red (Kernechtrot; Poly Scientific R&D Corp.). Images were acquired on a microscope equipped with a CCD camera, and Alcian-blue positive mast cells were counted at the tumor periphery (≤10 cell diameters) for 4 fields of view for each mouse tissue (fields around tumor margins were selected randomly for analysis).
Differences between genotypes in assays were tested for statistical significance using a paired Student t test (Microsoft Excel; P < 0.05 was considered to be statistically significant).
FES promotes velocity and persistence of mast cell chemotaxis toward SCF in an agarose drop assay
To better define the contributions of FES to SCF/KIT-driven chemotaxis in BMMCs, we adapted an agarose drop assay recently developed for carcinoma cells migrating toward EGF (41). The modified assay is depicted in Fig. 1A and involves addition of BMMCs that were starved of cytokines to wells containing fibronectin-coated glass coverslips on which an agarose drop (with or without SCF) resides. KIT signaling to β1 integrins promotes adhesion of mast cells (42), and this was reflected in formation of a ring of BMMCs surrounding the edge of the SCF-embedded agarose drops (Fig. 1B). As expected, far fewer BMMCs adhered at random positions on coverslips with control agarose drops lacking SCF (data not shown). Next, we compared phenotypes of wild-type (fes+/+) and fes−/− BMMCs in this chemotaxis assay. Similar numbers of adherent cells surrounding the SCF-containing agarose drops were observed for both genotypes (Fig. 1B). While a number of wild-type BMMCs had migrated under the SCF-containing agarose drops, this was significantly reduced in fes−/− BMMCs (Fig. 1B and D). The cell shape, as revealed by staining of microtubules, indicated that BMMCs adopt a polarized cell phenotype around the SCF-containing agarose drop (Fig. 1B, see arrows). We observed a significant reduction in the numbers of polarized cells for fes−/− BMMCs in this assay (Fig. 1C). Thus, consistent with our previous study (22), FES signaling downstream of KIT promotes cytoskeletal reorganization leading to a polarized mast cell capable of directional migration.
To extend on these results, we further adapted the SCF agarose drop assay to compare chemotaxis of BMMCs of either genotype within the same field by loading BMMCs with either CellTracker green or orange dyes (Fig. 2A). A field with similar proportion of fes+/+ (green) and fes−/− (red) BMMCs was imaged by time-lapse confocal microscopy, which revealed a striking reduction in numbers of fes−/− BMMCs migrating under the agarose drop compared with control (Fig. 2B and see Supplementary Video S1). Subsequent tracking of individual migrating cells revealed a significant defect in both the average distance of cell migration and the velocity of FES-deficient mast cells compared with control cells (Fig. 2C and D).
FES localizes to microtubules and promotes their organization in mast cells
Previous studies have shown that FES colocalizes with microtubules, and this is enhanced by generating activating mutations (e.g., L145P), or by coexpressing FES with Src family tyrosine kinases (31). Because microtubules are reoriented in polarized, motile leukocytes (43), we examined whether FES may function in this process to promote mast cell polarization and migration. First, we examined FES localization in polarized BMMCs relative to microtubules and the cell nuclei (stained with rabbit anti-FES, mouse anti-α/β-tubulin, and DAPI, respectively). We detected FES localization at the leading edge, and punctate structures proximal to the microtubule-organizing center (MTOC) and the trailing end (uropod) of the cell (Fig. 3A). While FES does not show extensive colocalization with microtubules, some colocalization was observed proximal to the MTOC and in the uropod of the cell (Fig. 3A, see intensity profile below for selected region indicated by dashed line). To determine whether FES regulates microtubule organization in mast cells, we stained microtubules within fes+/+ and fes−/− BMMCs subjected to the SCF agarose drop assay. Wild-type BMMCs displayed polarized cell phenotypes with a large leading edge containing arrays of microtubules emanating from the MTOC in the perinuclear region (Fig. 3B). Parallel arrays of microtubules were also observed in the uropod of fes+/+ BMMCs (Fig. 3B, see intensity traces i–iii below corresponding to regions indicated by the dashed lines). Under the same staining conditions, fewer fes−/− BMMCs displayed a polarized phenotype, and the microtubule staining intensity was reduced compared with control cells (Fig. 3B, see intensity traces iv–vi below corresponding to regions indicated by the dashed lines). This may reflect that FES functions to regulate microtubule organization in mast cells directly, or that FES functions to regulate processes required before microtubule reorganization.
Previous studies of FES localization or binding to microtubules provided evidence that this interaction can be regulated by mutations in the N-terminal and SH2 domains of FES (31, 44). However, no studies have tested for direct interactions between FES and microtubules. Here, we divided FES into N-terminal (1–459) and C-terminal (460–822) fragments (Fig. 4A). Together with the wild-type sequences, we generated an L145P mutations that blocks oligomerization and promotes FES kinase activity (45). Also, we mutated the pY-binding pocket within the SH2 domain (R483E), as this mutation reduced FES colocalization with microtubules in cells (31). All constructs were purified as GST fusions and then tested in microtubule-binding assays. We detected approximately 20% of FES 1–459 within the pellet fraction in the presence of microtubules, and virtually none in the pellet in the absence of tubulin (Fig. 4B and C). Under the same conditions, none of the GST control protein was found in the pellet fraction (data not shown). The FES L145P mutant also showed a similar capacity for binding microtubules (Fig. 4B and C), suggesting this binding is not dependent on the predicted α-helical F-BAR domain of FES (29). It is also worth noting, that incubation of FES 1–459, and not GST or 460–822, caused depolymerisation of microtubules, as evidenced by increased amount of tubulin in the soluble fraction (Fig. 4B). The C-terminal domains of FES (460–822) also bound to microtubules, with approximately 70% maximal binding (Fig. 4B and C). This binding was lost with the R483E mutation (Fig. 4B and C), suggesting that FES interacts most strongly with pY sites of microtubules. Consistently, our preparations of porcine microtubules showed pY reactivity (data not shown). Also, binding of FES 460–822 to microtubules was similar for a construct with a kinase-inactivating mutation (K588R; data not shown). Future experiments will be required to determine the putative binding sites in microtubules and the kinases involved in generating these FES-binding sites.
SCF production by AC2M2 mouse mammary carcinoma promotes mast cell migration
Next, we wished to establish whether FES is a potential therapeutic target in pathologies associated with aberrant SCF/KIT signaling, such as during tumor progression. Because FES has already been implicated in TAM recruitment to AC2M2 mouse metastatic mammary tumors in mice (24), we tested whether these cells express SCF. Indeed, SCF mRNA was detected by RT-PCR in AC2M2 cells (Fig. 5A). As shown previously (18), 4T1 mouse mammary carcinoma also express SCF (Fig. 5A; β-actin served as a control). Conditioned medium from AC2M2 cells was collected to test for secretion of SCF. Immunoblot with SCF antisera showed that SCF is present in AC2M2 conditioned medium, which comigrated with recombinant murine SCF (Fig. 5B). To test whether conditioned medium from AC2M2 cells served as a chemoattractant for mast cells, we used modified Transwell chamber migration assays with BMMCs treated with or without a KIT inhibitor, migrating toward AC2M2 conditioned medium treated with or without a neutralizing SCF antibody (Fig. 5C). Indeed, AC2M2 conditioned medium served as a chemoattractant for wild-type BMMCs (∼5-fold increase in motility), whereas fes−/− BMMCs showed a significant defect in AC2M2 conditioned medium–induced motility (Fig. 5D). Importantly, preincubation of AC2M2 conditioned medium with SCF-neutralizing antibody completely suppressed the motility of wild-type BMMCs (Fig. 5D). As expected, KIT inhibitor pretreatment of BMMCs had a similar effect (Fig. 5D). Together, these findings identify another epithelial-derived tumor model that attracts mast cells, with potential relevance to tumor progression and metastasis studies in mice.
FES is required for mast cell recruitment and enhanced growth of mammary tumors in mice
As reported recently, FES contributes to mammary tumor progression and metastases via effects on mammary stromal cells, such as macrophages (24). However, the potential role of FES in mast cell recruitment to tumors has not been reported. We used the same approach as Zhang and colleagues, to engraft AC2M2 cells in mice that were either fes+/+ or fes−/− (on a nu/nu background). In subcutaneous AC2M2 tumors, we identified skin mast cells near the tumor periphery in mice of both genotypes (Supplementary Fig. S1A). Scoring of the tumor-associated mast cells revealed a significant reduction in fes−/− mice compared with wild-type (Supplementary Fig. S1B), whereas no defects in resident skin mast cells were detected in nontumor-bearing mice (Supplementary Fig. S1C). Tracking tumor volumes in these mice revealed that although FES is not required for tumor growth, the tumors trended toward smaller volumes at the required endpoint (Supplementary Fig. S1D).
To test for effects of FES on the mammary stroma during tumor progression, we conducted orthotopic engraftment of AC2M2 cells. Consistent with our results for subcutaneous tumors, we observed a significant reduction in tumor-associated mast cells within the mammary tissues of tumor-bearing fes−/− mice compared with fes+/+ mice (Fig. 6A). Quantification of these results revealed a significant reduction in fes−/− mast cells to the mammary tumors (Fig. 6B). This was not due to a defect in resident mast cell population in these mice as the numbers of mammary gland mast cells was unchanged between wild-type and fes−/− mice (Fig. 6C and D). In this orthotopic engraftment AC2M2 model, we observed a correlation between the numbers of tumor-associated mast cells and the tumor volumes (Fig. 6B and E). Together, with our previous results showing reduced metastasis in mice lacking FES (24), these results implicate FES in promoting tumor progression and metastasis via enhancing the recruitment of both TAMs (24) and mast cells (this study).
In this study, we report that FES is a potential nontumor target in breast cancer, to limit the recruitment of mast cells and tumor progression. This builds on a recent study showing that FES promotes TAMs recruitment to mammary tumors, leading to enhanced tumor vasculature, tumor growth, and metastasis (24). Here, we report a highly tractable model for SCF-induced mast cell adhesion, polarization, and chemotaxis using SCF-embedded agarose drops. Using both fixed and live cell imaging, we show that FES promotes both the persistence of motility and velocity of mast cells. The subcellular localization of FES in polarized mast cells suggest that FES signals in various compartments, including the leading edge, MTOC-proximal structures, and within the uropod. There is partial colocalization with microtubules, which may be mediated by 2 separate binding sites in FES, and we showed that FES is also required for motility of mast cells toward SCF-producing tumor cells in vitro and tumors in vivo. Overall, these results implicate FES as a potential therapeutic target within the tumor stroma, to limit tumor progression and metastases.
Our study shows that FES possesses 2 independent modes of direct binding to microtubules mediated through the F-BAR/FX domains, and the SH2 domain. This functional redundancy strongly suggests that FES may act as a microtubule-associated protein (MAP), whereby its role in vivo may involve recruitment to microtubules and signaling to microtubules or MAPs. It was previously shown that a functional SH2 domain was required for FES L145P colocalization with microtubules in COS7 cells (31). The microtubule binding assays support that the SH2 domain appears to mediate the majority of microtubule binding. Although the phosphorylated tyrosine residues in microtubules that bind FES have not been identified, it is worth noting that our microtubule preparations contained detectable tubulin pY (data not shown). Previous studies by Laurent and colleagues (2004) have also shown that FES phosphorylates tubulin dimers and promotes their polymerization into microtubule-like structures (31). How FES regulates tubulin polymerization and interacts with microtubules in more physiologic conditions awaits further study. It is worth noting that in our study, the FES 1–459 fragment appeared to depolymerize the microtubules during binding assays (more in supernatant), and this was not observed with the FES C-terminus or GST control. This is consistent with differences in the organization of microtubules in fes−/− BMMCs subjected to the SCF agarose drop assay (this study), and may coincide with a soluble tubulin-binding site within the FES N-terminal domain (31). In future, it will be interesting to determine whether FES interactions with microtubules actually alter microtubule stability and nucleation, and whether these in vitro observations have roles in FES-dependent physiologic cellular responses like cell motility.
FES has been reported to localize to different cellular compartments, including the nucleus (46, 47), plasma membrane (29), the trans-Golgi (48), and localized to microtubules (31, 44). These different pools of FES likely signal to distinct substrates involved in cell differentiation, vesicle trafficking, focal adhesions, and cell migration. Systematic studies of FES regulation and substrate identification are required to fully elucidate these FES signaling pathways. Previous results have suggested that FES is constitutively oligomeric, but usually maintained in an inactive state (44). The L145P mutant, designed to be at the center of the first predicted CC domain of FES (45), is located near the tips of the F-BAR dimer generated by homology modeling of the FES F-BAR domain on the structure of FBP17 (29). A detailed characterization of the effects of L145P mutation on F-BAR domain function will be important to fully understand that how FES is regulated by this domain (25). The L145P mutant was shown to reduce or abolish oligomerization (44, 45), while increasing FES kinase activity and localization to microtubules (31, 49). Binding of FES F-BAR/FX domains to membrane phospholipids can also promote kinase activation (29, 32). Solving a structure of the complete FES kinase would undoubtedly help in understanding how the N-terminal domains regulate the SH2-kinase domain module; however, this will require overcoming the challenge of generating high quality crystals of full-length FES.
A number of key signaling proteins are required for SCF to trigger mast cell migration downstream of c-KIT, including class IA phosphoinositide 3-kinase, Rac GTPases, and FYN kinase (25, 42, 50, 51). FES kinase also plays a significant role in BMMCs (22) and in tumor-associated mast cells (this study) and macrophages (24). The relative contribution of FES signaling in these 2 cell types or the tumor vasculature will require further studies. For example, irradiated fes−/− mice could be reconstituted with wild-type or fes−/− bone marrow and tested in similar tumor engraftment studies. This would address the contributions of FES signaling within immune cells in this model. Also, new models of mast cell deficiency should allow for more accurate testing of the contributions of connective tissue mast cells to tumor growth, without potentially misleading results caused by effects of KIT mutations (e.g., W-sh/W-sh or W/Wv) on other cell types (52). Reconstitution of these mice with wild-type and fes−/− mast cells would allow for testing of the relative contribution of FES signaling in mast cells on tumor progression. Studies of lipopolysaccharide-treated fes-null mice implicate FES in regulating cytokine production and leukocyte extravasation, which correlated with defects in TLR4 internalization in macrophages (53, 54). Wei and colleagues recently showed that along with SCF, tumor cells shed ligands for TLR4, and that costimulation of KIT and TLR4 signaling in tumor-associated mast cells leads to enhanced secretion of cytokines (e.g., VEGF and IL-10) that enhance tumor progression (55). Future studies of the FES signaling axis in the tumor microenvironment will be required to fully understand the potential for therapeutic targeting of FES. Given the importance of some of these mediators on the adaptive immune response, it will be important to test the effects of targeting FES in immune competent mice. These studies will require the development of specific FES inhibitors, which are currently unavailable.
Mast cell infiltration of invasive carcinomas can lead to a multitude of phenotypic changes that can either enhance or suppress tumor growth (11, 19, 55). Links between immune cells and tumor progression have identified a number of secreted mediators, including growth factors [(platelet-derived growth factor) PDGF, EGF], immunosuppressors (TGF-β, adenosine), angiogenic factors (VEGF, angiotensin, IL-1β), and tissue remodeling proteases (tryptases, MMPs; ref. 56, 57). Thus, targeting the recruitment of mast cells to the tumor microenvironment may help suppress growth of the tumor vasculature and enhance recognition of tumors by the immune system. Targeting the SCF/KIT axis, or downstream mediators like FES, within the tumor microenvironment may have benefits for patients with KIT-negative mammary tumors (8). However, a recent study in a prostate cancer model suggests that targeting mast cells may alter not just tumor growth but also the tumor phenotype (12, 19). Thus, more preclinical studies with improved models of mast cell deficiency (52) will shed new light on contributions of mast cells and their myriad effectors on tumor progression and metastasis.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: E. Kwok, S. Zhang, P.A. Greer, J.S. Allingham, A.W. Craig
Development of methodology: E. Kwok, S. Zhang, P.A. Greer
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Kwok, S. Everingham, S. Zhang, P.A. Greer, J.S. Allingham, A.W. Craig
Analysis and interpretation of data: E. Kwok, S. Zhang, P.A. Greer, J.S Allingham, A.W. Craig
Writing, review, and/or revision of manuscript: E. Kwok, S. Zhang, P.A. Greer, J.S. Allingham, A.W. Craig
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Zhang, J.S. Allingham
Study supervision: P.A. Greer, J.S. Allingham, A.W. Craig
The authors thank Jalna Meens for technical assistance with mouse injections and tissue staining and Bruce Elliott and Craig laboratory members for helpful discussions.
This work was supported by operating grants from Canadian Institutes for Health Research (MOP82882) and Canadian Breast Cancer Foundation to A.W.B. Craig, and a Canadian Cancer Society grant to P.A. Greer.
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.