Inhibition of Angiogenesis by Interleukin 4

Human recombinant basic fibroblast growth factor (bFGF), murine recombinant IL-4 (muIL-4), and goat anti– human IL-4 were from R & D Systems (Minneapolis, MN). Human recombinant IL-4 (huIL-4, specific activity 5 × 10⁶ U/mg) was from Peprotech Inc. (Rocky Hill, NJ), as was human IL-13. Additional muIL-4 was purchased from Sigma Chemical Corp. (St. Louis, MO). Neutralizing rat mAb 11B11 (30) was a gift from Millennium Pharmaceuticals (Cambridge, MA) and was used as an ascites fluid. The muIL-4 used for systemic treatment of mice was generously supplied by Schering Plough Research Institute (Kenilworth, NJ). It had a specific activity of 2.24 × 10⁹ U/ mg and was >99% pure as judged by silver stained SDS-PAGE reducing gels.

Mouse mammary adenocarcinoma line K485 (31) and derivatives transfected with pSV7Neo (F1-1) or with pLT.IL-4 and pSV7Neo (D2-B1, E2A5, and E2A6; all described in reference 32) were grown in DME supplemented with 10% fetal bovine serum (FBS) and 2 mM glutamine. Serum-free conditioned media were collected as previously described (33), concentrated using a membrane with a 3-kD cut off, and then the protein was assayed with a Bio-Rad kit (Bio-Rad Laboratories, Hercules, CA).

T_H0 supernatants were generated from short-term spleen cell cultures derived from BALB/c congenic αβ T cell receptor transgenic mice (D011.10) in which >85% of the CD4 T cells are specific for ovalbumin. Erythrocyte-free splenic cells (4 × 10⁶/ml) from 8–10-wk-old mice were cultured with 18 μM ovalbumin in 24-well culture plates in Click's media (Irvine Scientific, Santa Ana, CA) supplemented with 5 × 10⁵ 2-mercaptoethanol, 3 mM glutamine, and 1% Nutridoma (a serum supplement from Boehringer Mannheim Corp., Indianapolis, IN). After 72 h of culture at 37°C in 10% CO₂, supernatants were pooled, sterile filtered, and assayed for IL-4 and for angiogenic activity. All supernatants were dialyzed against PBS before use in angiogenesis assays as undialyzed Click's medium was stimulatory. Culture supernatants were assayed for IL-4 and IFN-γ using commercial ELISA kits (Endogen, Woburn, MA) and data analyzed with SoftMax Pro software (Molecular Devices, Palo Alto, CA). Test samples were diluted to fall within a standard curve containing five to seven dilution points on each assay plate.

Human dermal microvascular endothelial cells, (HMVEC; Clonetics, San Diego, CA) were grown in endothelial basal medium supplemented with amphotericin and gentamycin sulfate, each at 50 μg/ml, 0.1 μg/ml human epidermal growth factor, 0.01 μg/ml hydrocortisone, 120 μg/ml bovine brain extract, and 5% FBS and, unless stated otherwise were used at passage <10 after starvation overnight in the same media without serum containing 0.1% BSA. Bovine adrenal capillary endothelial cells, BACE (a gift from Judah Folkman, Harvard University) were grown in DME supplemented with 10% donor calf serum, 2 mM glutamine, and endothelial cell mitogen at 50 μg/ml and used at passage 15 after starving overnight in serum-free media containing 0.1% BSA. Cells were harvested, suspended in DME with 0.1% BSA and plated at 1.75 × 10⁴ cells/well on the lower surface of a gelatinized membrane with 5-μm pores in a modified Boyden chamber (Neuroprobe, Cabin John, MD). After attachment, chambers were inverted, test material added to the top of the well and chambers incubated 3–4 h at 37°C to allow migration. Membranes were recovered, fixed, stained, and the number of cells migrated to the top of the filter in 10 high powered fields counted. All samples were tested in quadruplicate. Conditioned media from tumor cells were used at 20 μg/ml, bFGF at 10 ng/ml, and anti–IL-4 at 10–20 μg/ml.

HMVEC were plated into 96-well plates at 2.5 × 10³ cells/well in Clonetics EBM containing 2% FBS and gentamycin and allowed to adhere for 4–6 h. Test substances were then added, cells incubated for 72 h, and growth measured using Cell Titer 96 (Promega Corp., Madison, WI).

Neovascularization in the rat cornea was assayed as previously described (34). Hydron pellets containing, where indicated, bFGF at 100 ng/ml, murine IL-4 at 100 ng/ml, human IL-4 at 0.1–100 ng/ml, and/or anti-murine IL-4 at 40 μg/ml were implanted into the avascular cornea of the rat. 7 d later vessels were filled with colloidal carbon, the corneas were excised, and the growth of new vessels was assessed. Vigorous ingrowth from the limbus towards the pellet was considered a positive response.

To measure systemic effects of IL-4 in the mouse, five BALB/c mice (Jackson Laboratories, Bar Harbor, ME) were given intraperitoneal injections twice a day to give a dose of 15 μg/d/animal of murine recombinant IL-4. This dose was chosen in consultation with Dr. Bob Coffman (DNAX, Palo Alto, CA) to be saturating but not yet toxic. Five mice were treated similarly with vehicle saline only. On the second day, hydron/sucralfate pellets containing bFGF at 50 ng/pellet were implanted into their corneas as described (35), and vessels were observed using a slit lamp on days 3 and 5 after implantation.

When tested in the classic rat cornea assay, locally released IL-4 prevented vessels from sprouting from the surrounding vascular limbus into the normally avascular cornea in response to the inducer basic fibroblast growth factor (bFGF; Fig. 1; Table 1). Both murine and human IL-4 were active in this rodent assay and inhibition was dose dependent and sensitive to a mAb against muIL-4, 11B1, which is known to neutralize other activities of IL-4 (36). As concentrations of IL-4 were reduced, inhibitory activity fell off and at the lowest concentration of 0.1 ng/ml IL-4 by itself displayed a weak ability to induce neovascularization. Simple diffusion calculations indicate that the concentration of a compound that actually reaches the vascular limbus in such cornea assays is reduced by 10–50-fold suggesting that the stimulatory concentration of IL-4 at the endothelial cell is <0.01 ng/ml.

Systemic IL-4 also had an antiangiogenic effect. When mice received intraperitoneal injections of muIL-4, they became unable to mount a corneal angiogenic response to an implant containing an inducing concentration of bFGF (Table 2).

To determine if the antiangiogenic effect of IL-4 was due to a direct interaction of the cytokine with endothelial cells, its effects on cultured cells were tested. muIL-4 inhibited the migration of bovine adrenal capillary endothelial cells towards bFGF at concentrations of 1 ng/ml and greater (Fig. 2 A). huIL-4 gave an identical dose–response curve using these cells (data not shown). When their receptor message levels were measured by semiquantitative RT-PCR using primers described in references 23 and 37, human capillary endothelial cells expressed IL-4Rα and IL-13Rα subunits. These cells did not express the IL-4Rγc, the receptor subunit thought to be involved in the species-specific action of IL-4 on lymphocytes (6). The message levels of the receptors were not altered by treatment of the cells with IL-4 or with bFGF.

huIL-13, the cytokine thought to activate the same dimeric receptor on endothelial cells as does IL-4 (38), was also inhibitory (Fig. 2,B) although a higher concentration was required. Differential effects of IL-4 and IL-13 on cells like endothelial cells that lack the γc subunit of the IL-4 receptor may be explained by overproduction of IL-13Rβ, a recently identified molecule that binds IL-13 with high affinity but fails to transduce signals and may sequester IL-13 (39). Both muIL-4 and huIL-4 also blocked the migration of human microvascular cells (Fig. 2,C), although these cells required a cytokine concentrations of >10 ng/ml to achieve complete inhibition. As has been seen before with other inhibitory cytokines (40), both IL-4 and IL-13 exhibited biphasic dose-response curves. Exceedingly low doses stimulated the migration of endothelial cells (Fig. 2, A–C). huIL-4 was also effective at blocking the migration of human microvascular endothelial cells towards 100 pg/ml of the inducer VEGF and did so with an ED₅₀ similar to that with which it inhibited migration towards bFGF (data not shown).

IL-4 did not have any inhibitory effect on the growth of endothelial cells. The proliferation of human dermal capillary endothelial cells (Fig. 3) and of large vessel human umbilical vein endothelial cells (data not shown) was unaffected over a wide range of doses of huIL-4.

To determine if IL-4 is present in biological fluids in sufficient concentrations to be antiangiogenic, supernatants from two types of cells were tested. Serum-free conditioned media were collected from mouse mammary carcinoma tumor cell line, K485, and from its subclones that expressed IL-4 and as a result are known to produce slower growing tumors in vivo (32). Media from a vector-transfected control (F1-1, making no detectable IL-4, <0.001 ng IL-4/μg protein) and from two IL-4–transfected subclones that expressed low levels of IL-4 (E2A5 producing 0.18 ng IL-4/μg protein and E2A6 producing 0.06 ng IL-4/μg protein) were angiogenic and not sensitive to neutralizing antibody against the cytokine (Fig. 4,A). If this concentration of IL-4 were used by itself in a migration assay it would be weakly stimulatory. In contrast, medium conditioned by the IL-4 transfectant that produced high levels of IL-4 (D2B1 secreting 15 ng IL-4/μg protein), the line that was most severely retarded in in vivo tumorigenicity assays (32), was antiangiogenic despite the background of tumor angiogenic factors (Fig. 4, A–C; Table 3). The D2B1 conditioned medium blocked migration in vitro (Fig. 4,A) even towards media conditioned by the tumorigenic parent (Fig. 4,B) as well as neovascularization in vivo (Table 3; Fig. 4 C) induced by bFGF. IL-4 was the major inhibitor in this medium for its neutralization revealed underlying angiogenic activity and rendered the samples unable to inhibit angiogenesis induced by bFGF.

In a second experiment, supernatants of stimulated murine T_H0 cells were tested for angiogenic activity. These supernatants that contained 21 ng/ml of IFN-γ and 7.7 ng/ml of IL-4 were antiangiogenic due to the presence of IL-4 (Fig. 5). When IL-4 was neutralized they became able to induce the migration of capillary endothelial cells and were no longer able to inhibit migration induced by bFGF.

Data presented above demonstrate that IL-4 is a potent inhibitor of angiogenesis in vivo when introduced locally as well as when injected systemically. It is among the most potent inhibitors identified to date. Its ED₅₀ measured in the migration assay was between 0.015 and 0.15 nM, better than that measured in the same assay for other potent antiangiogenic agents such as thrombospondin-1 (0.5 nM, see reference 40) and angiostatin (7 nM, see reference 41). The dose of the 13-kD IL--4 that was needed to inhibit neovascularization systemically in the mouse was a relatively modest 0.5 mg/kg/d which compares favorably with the doses needed to produce the same effect with other inhibitors of angiogenesis including the 5–10 mg/kg/d needed for 450-kD thrombospondin-1 (20), the 6–100 mg/kg/d required for 38–45-kD angiostatin (42, 43) and 20 mg/kg/d for 20-kD endostatin (44).

The inhibition of angiogenesis is a newly recognized function for IL-4 and one that helps to explain a number of its previously noted activities. Our ability to drive mice into an antiangiogenic state with IL-4 injections suggests that a systemic, IL-4–dependent antiangiogenic effect may contribute to the ability of IL-4 secreting cells to inhibit the growth of distant tumors (45, 17). The inhibition of angiogenesis may also play a role in the ability of IL-4 to ameliorate arthritis in animals (46) and reduce the cartilage degradation in patients (47) that results in part from invading endothelial cells in the synovial pannus. The local antiangiogenic activity of IL-4 could account for the decrease in tumor vessel density observed in IL-4 secreting gliomas (21) and should enhance its effectiveness as a gene therapy agent against tumor metastases (48).

IL-4 inhibits angiogenesis by acting directly on endothelial cells for it was able to block the migration of cultured cells, as are most other direct inhibitors of angiogenesis (49). At very low doses, IL-4 was stimulatory to endothelial cells in vitro and able to induce a weak neovascularization response in vivo. Such a biphasic dose–response curve is unusual although not unprecedented. TGF-β1 also has biphasic effects on endothelial cell migration, stimulating at picomolar doses but inhibiting at nanomolar concentrations (40). Although this phenomenon of stimulation giving way to inhibition has not been explained, there is an explanation for one example of the inverse situation where inhibition of migration of endothelial cells at low concentrations gives way to stimulation at high doses. This occurs with the inhibitor of angiogenesis thrombospondin-1 and can be explained by the sequential activation of two distinct receptors. An inhibitory receptor is active at low doses, but it is easily cleared from the cells so that at high doses a second more stable receptor with lower affinity is engaged and transduces a stimulatory signal (50, 51). A dual receptor hypothesis for IL-4 is supported by the observation that IL-13 replicated the activity of IL-4 in the inducing phase of the dose–response curve with reasonable fidelity yet required 10-fold more protein to reproduce the inhibitory portion. Such an observation could be explained if endothelial cells expressed a stimulatory receptor that had equal affinity for IL-4 and IL-13 in addition to a second receptor, which transduces inhibitory signals, that interacted more effectively with IL-4 than with IL-13.

In contrast to the situation in hematopoietic cells where IL-4 displays tight species specificity between mouse and human, no species limitations were observed in our experiments where human and mouse cytokines were equally effective on bovine and human endothelial cells and human IL-4 was effective in the rat cornea assay. It is possible that the species specificity that has been defined in lymphoid cells depends on the γc subunit of the receptor that is absent from endothelial cells where it is replaced by an IL-13 receptor subunit.

Despite its effects on migration, IL-4 did not appear to have any effect on the growth of endothelial cells. Although both the migration and the mitogenesis of endothelial cells are frequently blocked by agents that inhibit angiogenesis, there is a small class of inhibitors that do not seem to have major effects on mitogenesis (49). However, it may be premature to assign IL-4 to this class given the fact that another report, using microvascular endothelial cells that were brought more severely to quiescence before the start of the experiments, has described a modest mitogenic effect of IL-4 (29).

Although data presented here demonstrate most clearly the inhibition of angiogenesis induced by bFGF, IL-4 was also able to block angiogenesis induced by other factors for it was effective against tumor cell conditioned media, against the mix of inducers present in the T_H0 cell supernatant as well as against media conditioned by activated macrophages (data not shown).

In identifying IL-4 as a new inhibitor of angiogenesis these results provide a new mechanism to explain its demonstrated antitumor activity at both local and distant sites, and suggest, as has been demonstrated for other inhibitors of angiogenesis (52), that it may be particularly useful in enhancing the effect of cytotoxic antitumor therapies. In addition, the ability of IL-4 to be either a positive or a negative influence on neovascularization must now be taken into account when explaining the induction and resolution of the inflammatory angiogenesis that accompanies so many pathological processes.

bFGF

basic fibroblast growth factor

FBS

fetal bovine serum

HMVEC

human dermal microvascular endothelial cells

hu

human recombinant

mu

murine recombinant

VCAM-1

vascular cell adhesion molecule 1