Immunoglobulin E induces VEGF production in mast cells and potentiates their pro-tumorigenic actions through a Fyn kinase-dependent mechanism

Production of angiogenic factors by MC has been shown to occur few hours after different stimuli, such as hypoxia, antigen or PMA [16, 17]. To investigate if monomeric IgE in the absence of any antigen could induce VEGF secretion in this cell type, two million BMMCs were incubated with a monoclonal anti-DNP IgE (1000 ng/ml) for eight hours at 37°C in BMMC media. Supernatants were then collected and the amount of secreted VEGF was determined by ELISA. The addition of IgE to MC induced a significant secretion of VEGF (53.77 ± 2.15 pg/ml in basal conditions vs 117.16 ± 5.45 pg/ml on IgE-stimulated cells; Figure 1A). To gain insight on the mechanism involved in IgE-induced VEGF production, cells were pre-treated with different pharmacological inhibitors 15 minutes before the addition of IgE and secreted VEGF was measured. VEGF secretion was sensitive to actinomycin D (ActD) and brefeldin A (BFA), indicating that de novo transcription and transport from endoplasmic reticulum to the Golgi apparatus [18] was needed for IgE effects to occur. VEGF production was also affected by tetanus toxin (TTx), which inhibits secretion mediated by toxin-sensitive VAMPs (VAMP-1 and −2) [19], and PP2, that inhibits Src family kinases (Figure 1A).

Two main Src family kinases modulate mediator secretion from MCs. Lyn and Fyn kinases have been shown to be involved in early signaling after FcϵRI crosslinking, leading to the activation of downstream pathways regulating pro-inflammatory cytokine production [7]. In order to test if one of them could be involved in IgE-induced VEGF secretion in BMMC, cells from WT, Lyn −/− and Fyn −/− mice were generated and stimulated with different concentrations of monomeric IgE (Figure 1B). WT BMMCs reached maximal VEGF secretion after the incubation with 1000 ng/million cells while BMMCS generated from Lyn −/− mice did not show an important difference when compared to WT cells. However, BMMCs derived from Fyn −/− mice showed an important defect on IgE-induced VEGF production, since the maximal amount of secreted VEGF in Fyn −/− cells was significantly lower than in WT cells (Figure 1B). We then characterized the time course of VEGF secretion stimulated by IgE in WT and Fyn −/− cells. VEGF in cell supernatants was detectable as early as two hours after stimulation and the maximal amount was reached after eight to twelve hours of stimulation in both cell types. However, Fyn −/− BMMCs did not secrete as much VEGF as WT cells even at longer incubation times (Figure 1C).

In order to investigate the role of Fyn kinase in the IgE-dependent VEGF synthesis, BMMCs derived from WT or Fyn −/− mice were incubated at 37°C in the presence of vehicle or IgE (1000 ng/ml) for 8 hours. Cells were collected using a cytospin centrifuge and processed to detect VEGF by fluorescent immunostaining. Figures 2A and B show that low VEGF positive signal was observed in WT BMMCs and IgE addition increased intracellular VEGF. In contrast, VEGF immunoreactivity was significantly lower in Fyn −/− BMMCs in both basal and IgE-stimulated conditions.

In order to investigate if the absence of Fyn could lead to alterations on IgE-dependent VEGF mRNA synthesis, we analyzed VEGF mRNA content by RT-PCR in WT and Fyn −/− BMMCs after IgE stimulation. Optimal conditions for amplification of VEGF and GAPDH (control) were experimentally determined in our cell system (see Methods section). VEGF₁₆₄ mRNA, the predominant isoform in different cell systems including MC [13], was increased in WT BMMCs after IgE addition and the same effect was observed in Fyn −/− cells. Interestingly, Fyn −/− BMMCs showed slightly higher amounts of VEGF mRNA than WT BMMCs in the absence of any stimulus and also one hour after IgE addition (Figure 2C-D). Together, these results strongly suggested that Fyn kinase could be involved on the translational control of VEGF protein production.

To test the hypothesis that IgE-induced VEGF protein translation could be defective in the absence of Fyn, we evaluated whether VEGF synthesis in Fyn −/− cells could be more sensitive to the protein synthesis inhibitor cycloheximide (CHX) than in WT BMMCs. We found that CXH pre-treatment provoked a decrease on VEGF production in both cell types, but Fyn−/− BMMCs showed an increased sensitivity to this inhibitor (Figure 3A). The main difference was detected at concentrations of 1 μM of CHX, where normalized VEGF production was inhibited by 26.53% of the maximal release in WT cells, compared to 52.49% in Fyn −/− BMMCs.

It has been shown that VEGF synthesis under different stressful conditions (such as hypoxia) depends on the rapid modification of protein translation machinery [20]. In those circumstances, the normal 5′cap-dependent translation is stopped and internal ribosome entry sites (IRES)-dependent translation proceeds on VEGF mRNA [20]. Those adjustments on the translational apparatus can be followed by changes on the phosphorylation of the ribosomal protein S6 (S6) and the dephosphorylation of the eukaryotic initiation factor 4E binding protein 1 (4E-BP1) [21, 22]. To investigate if IgE could induce a stressful condition similar to hypoxia in MC leading to IRES-dependent protein translation, we analyzed the phosphorylation levels of pS6 and 4E-BP1 in WT and Fyn −/− BMMCs stimulated with IgE.

Figure 3B shows that IgE addition to BMMCs induce a rapid phosphorylation of S6 and dephosphorylation of 4E-BP1 in WT cells. In contrast IgE-dependent, dephosphorylation of 4E-BP1 was not detected in the absence of Fyn. Quantitative analysis of western blots performed for Figure 3B is presented on Figure 3C-D. Optical density values of each specific band were normalized with those obtained from actin in the same membrane (see Material and Methods section). Results show that IgE is able to induce changes on the translational machinery of MC, promoting IRES-dependent translation in a Fyn-dependent manner.

In order to evaluate the effect of IgE on the secretion of VEGF in MC and the role of Fyn kinase in that process in vivo, we took advantage of the well-recognized role of MC and VEGF on the development of B16 melanoma tumor [15, 23]. A number of experiments have shown that MC have a positive effect on murine melanoma angiogenesis and tumor growth [23, 24]. For instance, MC-deficient mice (Wv strain) are unable to generate angiogenesis of B16 melanoma tumors and the reconstitution of those animals with WT BMMCs leads to the restoration of tumor development [23]. The number of VEGF positive MC located in the peritumoral area has been found to correlate with increased angiogenesis and poor prognosis of melanoma [15].

We started defining the amount of IgE that could be administered in mice to analyze its effect on melanoma tumor growth. We decided to utilize a dose of IgE that could give a measurable parameter of MC and basophil activation. Passive anaphylactic reactions have been shown to depend on MC activity when initiated with an specific IgE and triggered with the corresponding antigen [25, 26]. Exogenous IgE occupies free FcϵRI receptors on immune cells and those can be crosslinked administrating the antigen to which IgE was synthesized [27].

We determined the concentration of IgE needed for maximal activation of passive anaphylactic reaction in C57BL/6J mice. Monomeric IgE directed against DNP-HSA was intravenously (i.v.) injected to mice and twenty four hours later, antigen was i.v. injected in the presence of Evans blue dye. Twenty minutes after antigen administration, limbs were removed and dye extravasation was measured. As can be observed in Figure 4, the optimal IgE dose to obtain maximal anaphylactic response was 750 ng per mouse and the effect of IgE was long-lasting, since antigen-specific anaphylactic reaction was observed in the limbs even four weeks after a single IgE administration.

To evaluate the effect of IgE on melanoma tumor growth, C57BL6/J (WT) mice were treated with a single i.v. administration of saline or monomeric IgE (750 ng/mouse) and twenty four hours later, mice were subcutaneously (s.c.) inoculated with B16 melanoma cells in one ear pinna and generated tumors were removed after four weeks. Tumors obtained from IgE-treated mice (+IgE) were larger than those obtained from saline-treated mice (−IgE) (Figure 5A). Histopathologic analysis of tumor sections utilizing hematoxilin-eosin (H&E) and toluidine blue (TB) stains showed that those obtained from IgE-treated mice presented significantly higher numbers of blood vessels inside the tumor and MC in the peritumoral area (Figure 5B-D). The role of VEGF on melanoma tumor growth was verified in our system by the use of the neutralizing anti-VEGF antibody Bevacizumab (Beva; Figure 5E). As expected, administration of Beva (10 mg/kg) diminished tumor size. Remarkably, the effect of IgE on tumor growth was prevented by the administration of that agent.

When B16 melanoma cells were inoculated in MC-deficient mice (Wsh), the size of the tumors after 28 days was significantly lower than the observed in WT mice (Figure 6A-B). As expected, normal tumor size was recovered after the reconstitution of Wsh animals with BMMCs derived from WT mice (Wsh Rec WT). Interestingly, monomeric IgE importantly increased the size of melanoma tumors on WT and Wsh Rec WT mice but this effect was not observed in Wsh animals treated with IgE. The effect of IgE on melanoma tumor growth positively correlated with the capacity of IgE to induce anaphylactic reactions, since Evans blue extravasation in Wsh Rec WT mice was similar to the observed in C57BL6/J mice under the same treatment (data not shown).

Histological analysis of biopsies of Wsh and Wsh Rec WT-derived tumors, in the presence or absence of IgE, was performed by staining slides with H&E and TB dyes. The number of blood vessels and peritumoral MC was significantly higher in Wsh Rec WT than in Wsh mice (Figure 6 C-D).

To investigate if Fyn kinase in MC could be involved in IgE-induced melanoma growth and angiogenesis, Wsh mice were reconstituted with BMMCs from Fyn −/− mice (Wsh Rec Fyn−/−). IgE-dependent tumor growth was significantly impaired in mice reconstituted with Fyn−/− BMMCs compared to mice reconstituted with WT cells (Figure 7A-B). A decreased number of blood vessels and MC were observed in tumor biopsies obtained from Wsh Rec Fyn−/− mice compared to those obtained from Wsh Rec WT animals (Figure 7C).

Salts, inhibitors and buffer components, as well as anti-dinitrophenyl (DNP) IgE (clone SPE-7) and DNP coupled to human serum albumin (DNP-HSA) were purchased from Sigma-Aldrich. Recombinant IL-3 was purchased from Peprotech.

C57BL/6J (wild type: WT; stock No. 000664), mast cell-deficient B.6Cg-Kit ^W-sh (Wsh; stock No. 005051), 129/Sv-Lyn^tm1Sor/J (stock No. 003204) and 129-Fyn^tm1Sor/J mice (stock No. 002271) were purchased from Jackson Laboratories. 129-Fyn^tm1Sor/J mice were back-crossed with C57BL/6J at least five times in our animal facilities to obtain Fyn-deficient animals (Fyn −/−) on the C57BL/6J genetic background. Genotyping of each animal was performed by PCR from genomic DNA utilizing the primers and conditions suggested by the provider. Animals were maintained under pathogen-free conditions and treated in accordance with NIH guidelines. Experiments were performed with age-matched male mice of at least 8 weeks old and were approved by the animal ethics committee of Cinvestav (CICUAL, protocol 032–02 and 0478–10). When necessary, mice were euthanized by CO₂ inhalation.

Bone marrow-derived mast cells (BMMCs) were generated extracting the bone marrow from both tibias of mice four to eight weeks old. Bone marrow was cultured in BMMC media (RPMI 1640 supplemented with 20 ng/mL IL-3, 0.1 mM non-essential aminoacids, 50 μM β-mercaptoethanol, 25 mM HEPES pH 7.4, 1mM pyruvate, antibiotic/antimycotic and 10% FBS) during four weeks and after that, FcϵRI receptor expression was analyzed by flow cytometry utilizing a specific positive for the IgE receptor as described [41]. To confirm functionality of BMMC cultures, routinely two million of WT or Fyn −/− BMMCs were incubated with 100 ng/ml of IgE for 20 min in 1 ml of Tyrode’s-BSA buffer (20 mM HEPES pH 7.4, 135 mM NaCl, 5 mM KCl 5 mM, 1.8 mM CaCl₂, 1 mM MgCl₂, 5.6 mM glucose, 0.05% bovine serum albumin; BSA) at 37°C. After incubation, different concentrations of the specific antigen (human serum albumin coupled to DNP) were added. After 30 minutes at 37°C, cell supernatants were collected and tested for β-hexosaminidase activity as described [15].

Two million BMMCs were incubated with distinct concentrations of IgE for different times at 37°C in 1 ml of BMMC media supplemented with a protease inhibitor cocktail (mini-complete, Roche). VEGF was determined in the supernatants utilizing ELISA kits from Peprotech and Invitrogen. Actinomicyn D (Act D; 5 μg/ml), Brefeldin A (BFA; 5 μg/ml), Tetanus toxin (TTx; 10 ng/ml) and PP2 (10 μM) were added to the cells 15 min previous to the stimulation [15, 42] with 1000 ng/ml IgE for 8 h at 37°C.

Two million BMMCs were incubated with 1000 ng/ml IgE at 37°C for different times. Total RNA was isolated utilizing Tri-Reagent and isopropanol RNA precipitation as described [43]. cDNA synthesis was performed starting with 5 μg of total RNA utilizing the Superscript First Strand System for RT-PCR from Invitrogen, following the instructions included. Reported primers were used to amplify VEGF [44] and GAPDH [45], utilizing one tenth of the cDNA reaction volume. Standardization of the number of cycles needed to obtain linear amplification of VEGF and GAPDH mRNAs in our samples was performed experimentally. Linear amplification for VEGF was obtained between 30 and 40 cycles, whereas for GAPDH it was between 25 and 30 cycles. Then, conditions for VEGF amplification were 94°C for 1.5 min; 35 cycles of 94°C for 30 sec, 58°C for 45 sec and 72°C for 45 sec and additional extension at 72°C for 10 min. Parameters for GAPDH amplification were the same but only 28 cycles of amplification were used for this gene. RT-PCR products were separated in 2% agarose gels. Photographs of each gel were quantified utilizing the LabWorks Image Acquisition and Analysis software (v 4.5) installed in a UVP image analyzer.

Two million BMMCs were incubated with 1000 ng/ml IgE for 8 h in 2 ml of BMMC media supplemented with a protease inhibitor cocktail (Roche) at 37°C. Cells were collected by centrifugation using a cytospin, then washed, fixed in cold acetone for 5 min at 4°C, washed again and blocked with 3% BSA, 2% goat serum and 0.01% Tween-20 in PBS for 30 min at RT. Cells were incubated with an anti-VEGF rabbit polyclonal serum (1:50; sc-507, Santa Cruz) diluted in blocking solution overnight at 4°C. After washes with PBS, slides were treated with an Alexa 568-anti-goat secondary antibody (1:500, Invitrogen) diluted in blocking solution for one hour. Finally, an incubation with 4′, 6-diamidino-2-phenylindole, dihydrochloride (DAPI, Invitrogen) for 5 min at RT was performed and slides were properly mounted. A confocal microscope (Olympus FluoView FV1000) with a 405-nm laser for DAPI and 543–nm laser for Alexa 568 was used. Fluorophores were imaged using a sequential line scan, with detection bands set at 405 to 461 nm for DAPI stain and 543 to 603 nm for Alexa 568. Each image was saved at a resolution of 512 × 512 pixel image size and analyzed. Average fluorescence intensities of VEGF per cell were measured from 13 to 33 cells per experimental group utilizing the software LabWorks 4.5.

Two million BMMCs were incubated with 1000 ng/ml IgE for 0, 15, 30 and 60 min in Tyrode’s buffer without BSA [16] at 37° [46]. Cell pellets were isolated and lysed on Laemmli 1X buffer. In general, fifty micrograms of total MC protein was separated in SDS-PAGE and transferred to PVDF to conduct western blot using reported solutions and methods [41]. Antibodies against phospho-pS6 ribosomal protein (Ser 235/236, 1:1000) and phospho-4E-BP1 (Thr37/46, 1:1,000) were purchased to Cell Signaling Technology. For loading control, membranes were incubated with an anti-β-actin antibody (1:10,000) from Santa Cruz. Densitometric analysis was performed utilizing an Epichemi Darkroom from UVP Systems, with the LabWorks 4.5 software.

In order to evaluate the role of IgE on melanoma tumor growth, we first determined the amount of IgE that could be needed to occupy the free FcϵRI receptor on the surface of mast cells in vivo. C57BL/6J mice were i.v. injected with saline solution or, 25, 50, 250, 500, 750 and 1000 ng of anti-DNP IgE dissolved in 100 μL of sterile saline solution. Then, a passive anaphylaxis assay [25, 26] was performed by i.v administering 100 μg DNP-HSA in 100 μl Evans blue dye (0.25%). Twenty minutes later, animals were euthanized and the limbs were removed. The amount of extravased dye in the tissue was measured as previously described [25, 26]. To test the amount of free IgE after administration, C57BL/6J mice were i.v. injected with 0, 25, 50, 250, 500, 750 and 1000 ng of anti-DNP IgE and trunk blood was collected 0, 1, 7, 14, 21 and 28 days later. Each blood sample was centrifuged at 21,000 x g for 20 min at 4°C and the resultant serum was stored at −80°C. IgE levels were determined utilizing an ELISA kit (BD Biosciences).

MC-deficient (Wsh) mice (eight to twelve weeks old) were reconstituted with MC as described [47]. Briefly, mice were i.v. injected with BMMCs (2×10⁶ cells) in Tyrode’s buffer without BSA and four weeks after adoptive transfer of BMMCs MC reconstitution was confirmed by histological analysis of ear pinna and passive cutaneous anaphylactic reactions. Wsh mice were reconstituted with BMMCs obtained from C57BL/6J (Wsh Rec WT) or C57BL/6J-Fyn−/− (Wsh Rec Fyn−/−) mice. Animals were inoculated with melanoma cells four to six weeks after reconstitution and tumor growth (see following section) was compared with age-matched WT and non-reconstituted animals.

B16-F1 melanoma cells obtained from ATCC (CRL-6323; low metastatic variant) were generously donated by Dr. Guadalupe Reyes, Cinvestav, and cultured as suggested by the provider. Briefly, cells were propagated in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% FBS, 100 U of penicillin and 100 μg/ml of streptomycin [23]. For tumor generation, mice were s.c inoculated into the left ear pinna with 0.5 × 10⁶ B16 melanoma cells in Tyrode’s buffer without BSA and the right ear pinna with vehicle [23]. Tumor weight development was monitored on daily basis and tumors were removed 28 days after inoculation to register weight differences between left and right ear pinna. Normal development of the tumors was routinely confirmed by histological analysis, where melanoma cells, blood vessel formation and infiltrating inflammatory cells were detected by specific dyes at different times after inoculation [48]. To test the influence of IgE on melanoma tumor growth, mice were i.v. administered with IgE (750 ng) or vehicle 24 hours previous to melanoma cell inoculation. When necessary, mice were treated with bevacizumab 10 mg/kg s.c., biweekly, starting 24 h after inoculation of melanoma cells (modified from [49]).

Biopsy specimens of the ears inoculated with vehicle or B16 melanoma cells were fixed in neutral buffered formalin (10% formaldehyde in 10 mM phosphate buffer pH 7.0) prior to paraffin embedding. Sections of 2.5 μm were cut and examined. For each treated animal (n = 2-7 per group), some ear sections were stained with hematoxylin and eosin (H&E) and other sections were stained with 0.1% toluidine blue (TB), pH 1.0 [50] in order to analyze blood vessels and mast cells, respectively. Ear pinna mast cells were counted by dividing the whole pinna area in consecutive fields length extending from the base to the tip (7.3–12.62 mm) under the 40X objective. Mast cells numbers are expressed as mean ± SEM per length of ear pinna. Images of histological sections were obtained with a microscope (Zeiss Axiostar) coupled to a digital camera (Power Shot A640). For each mouse, two sections were analyzed and counts were averaged. Blood vessel quantification was performed on 15 high-powered images of randomly selected areas of each paraffin section stained by H&E (x200). All histological measurements were done by two observers on blinded samples (control or tumor-bearing with or without IgE). In sections stained by H&E, the absolute number of large and small vessels per ear pinna was monitored by morphological analysis according to Carmeliet, 2003, i.e. vessels containing only the endothelial cell layer were considered small vessels, whereas those with the smooth muscle cell layer were considered to be large blood vessels [51].

Unless specified otherwise, all data are expressed as the mean ± SEM of at least three independent experiments. Student’s t-test or Two-Way ANOVA followed by Student-Newman Keuls test was used to evaluate the significance of the differences. Statistically significant was set at P < 0.05.