INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THE EPIDERMAL GROWTH FACTOR (EGF) family of receptors includes four closely related transmembrane tyrosine kinases: ErbB1 (HER1 or EGF receptor), ErbB2 (HER2 or neu), ErbB3 (HER3), and ErbB4 (HER4). These receptors can form both homo- and heterodimers in a manner that is dependent on the combination of receptors present, the stimulatory ligand, and the intracellular signaling pathways activated (27). This allows them to affect, positively and negatively, a wide variety of cellular functions, including proliferation, migration, differentiation, and cell survival.

The importance of overexpression of these receptors in cellular transformation and tumor metastasis has been elucidated by a number of in vitro and clinical studies (15). Furthermore, targeted deletion of ErbB2, ErbB3, ErbB4, or neuregulin-1 (a ligand for these receptors) in mice leads to developmental abnormalities that are severe in the nervous system and lethal in the cardiovascular system (8, 9, 17, 19, 21, 28). Cardiovascular abnormalities include aborted development of ventricular myocardial trabeculae and the endocardial cushion, the latter of which is dependent on mesenchymal cell growth and endocardial endothelia development.

The ligands for these receptors are primarily synthesized as transmembrane proteins, which can either be cleaved to form a soluble ligand or remain on the cell surface to mediate juxtacrine signaling. These ligands have been divided into three functional groups based on their pattern of ErbB receptor binding and activation. One group (including EGF, transforming growth factor-, and amphiregulin) binds only ErbB1 and can transactivate the other receptors when heterodimerized with ErbB1. A second group (including betacellulin, epiregulin, and heparin-binding EGF) binds both ErbB1 and ErbB4 and transactivates ErbB2 and ErbB3 when heterodimerized with either of these receptors. The last group (including the neuregulins) can bind both ErbB3 and ErbB4 and transactivates ErbB2 and ErbB1 only when expressed in combination with one of these receptors (26).

The neuregulins (also known as heregulins, acetylcholine receptor-inducing activity, Neu Differentiation Factor, or glial growth factor) were originally described as splice variants derived from a single gene (36). More recently, two additional genes also encoding a variety of splice products have been described (3, 4, 37). These ligands play important roles in signaling not only in cardiac development but also in neuronal cell differentiation, signaling at neuromuscular junctions, and epithelial cell morphogenesis [e.g., mammary gland development and wound healing (5), and for a review see Ref. 2]. Recently, neuregulin-1 mRNA has been shown to be expressed by rat coronary microvascular endothelial cells, and neuregulin stimulation of rat primary cardiac myocyte cultures has been demonstrated to result in increased myocyte survival, hypertrophy, and proliferation (38).

Because of the role these receptors play in cardiovascular development, we investigated whether they are present and functional in human vascular endothelium. Here, we show that a subset of these receptors is present in human vascular endothelial cells (EC). EC receptor stimulation by neuregulin leads to proliferation and morphogenic changes, which are consistent with an angiogenic response. Finally, using an in vivo assay, we demonstrate that neuregulin stimulation can lead to growth of new blood vessels.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Materials. The EGF-like domain of human neuregulin-3 (~7 kDa) was purified from a GST fusion construct (courtesy of Drs. M. Kraus and G. Carpenter, Vanderbilt University) by thrombin cleavage (1). Preparations were added to benzamidine-agarose beads (Sigma), 3-4 successive rounds, with a total binding capacity for thrombin 140- to 190-fold over the total amount of thrombin added for the cleavage. The control pGEX peptide was prepared from the empty pGEX-2T vector (Pharmacia Biotech) in an identical fashion. Neuregulin-1 (amino acids 177-241, prepared in Escherichia coli) was a gift of Dr. K. Carraway (Harvard Medical School). Recombinant human thrombin and hirudin were purchased from Sigma; recombinant human EGF and vascular endothelial cell growth factor (VEGF) were from Collaborative Biomedical Products, Becton Dickinson.

Cell isolations and culture. HUVEC were isolated from single donors as previously described (25). Cells were routinely passaged on gelatin-coated plates in medium 199 (M199) with 15% fetal bovine serum (FBS), bovine endothelial cell growth supplement (ECGS, 50 µg/ml), and heparin (100 µg/ml). BT474 cells were obtained from the ATCC and maintained in RPMI 1640 with 10% FBS.

Immunoprecipitation and Western blotting. Cell lysates were incubated with 1 µg of each antibody (Ab) as follows: for ErbB1, monoclonal Ab-528 (courtesy of Dr. H. Masui, Sloan Kettering, NY); for ErbB2, Ab-5 (Oncogene Science); for ErbB3, RTJ.2 (Santa Cruz); and for ErbB4, C-18 (Santa Cruz). Immunoprecipitates were Western blotted with the same antibodies used to immunoprecipitate for ErbB3 and ErbB4, and with Ab-2 for ErbB1 (Oncogene Science); Ab-3 for ErbB2 (Oncogene Science); and PY20 for phosphotyrosine (ICN).

Ca²⁺ flux measurements. Cells were loaded with fluo 3-AM in 0.5% pluronic acid, and measurements were made using an ACAS 570 interactive laser cytometer (Meridian) as previously described (25). Data shown are from two to four individual cells in single wells and are representative of three separate experiments using 96-well plates.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Expression of ErbB receptors. Human umbilical vein EC (HUVEC) were evaluated for expression of ErbB1, ErbB2, ErbB3, and ErbB4 by immunoprecipitation and Western blotting. Figure 1A demonstrates that expression of ErbB2, ErbB3, and ErbB4 is easily detected in HUVEC. As noted in other cell types, several different molecular weight species are seen for ErbB4. An ErbB4-specific peptide (Santa Cruz) competes for antibody reactivity with both the higher and lower molecular weight species seen in ErbB4 blots (data not shown), suggesting that these represent alternative, perhaps differentially glycosylated, receptor isoforms. HUVEC lysates do not contain detectable ErbB1 (EGF-r). Because of receptor heterodimerization and apparent required complex formation for signal transduction, this combination of receptors in HUVEC makes their function likely. This pattern of ErbB receptor expression was also seen in lysates prepared from freshly isolated (never cultured) HUVEC, suggesting that the lack of ErbB1 expression is not an artifact of cell culture conditions.

View larger version (40K): [in this window] [in a new window]   Fig. 1.   A: expression of ErbB receptors in human umbilical vein endothelial cells (HUVEC). Confluent monolayers of HUVEC (passages 3-5) in 100-mm dishes were lysed, 2 mg of total protein was immunoprecipitated (only 200 µg of BT474 lysate were used for ErbB2 immunoprecipitation, because this line highly overexpresses ErbB2), and Western blotting performed on the SDS-PAGE-resolved, transferred immunoprecipitates, using antibodies defined in METHODS. B: effect of neuregulin (left) or thrombin (right) on HUVEC ErbB2 tyrosine phosphorylation. Confluent cell monolayers were serum depleted [1% fetal bovine serum (FBS), no endothelial cell growth factor (ECGF)] overnight, followed by treatment with neuregulin-3 (100 ng/ml) for 10 min or recombinant human thrombin (2 U/ml) for 5 min. To determine the amount of phosphorylated ErbB2, lysates containing equal amounts of protein were immunoprecipitated using an anti-ErbB2 antibody and used for Western blotting with an antiphosphotyrosine antibody. Control blots were reprobed with an anti-ErbB2 antibody to demonstrate that equal amounts of total ErbB2 were present in all immunoprecipitates (data not shown). IP, immunoprecipitation; WB, Western blot.

Signaling via ErbB receptors. To determine whether HUVEC ErbB complexes are competent signal transducers, cells were stimulated with 100 ng/ml of neuregulin-3 for 10 min. Figure 1B (left) displays that ErbB2 phosphorylation is induced by neuregulin-3. As neuregulin binds with more avidity to ErbB3 and ErbB4 than to ErbB2 itself, this phosphorylation is likely to occur as a result of neuregulin activation of ErbB2-containing heterodimers.

It has also been suggested that in addition to ligand binding-mediated direct activation, ErbB2 signaling may link other receptor-mediated events to cellular processes. This is particularly important for understanding how some inflammatory mediators can elicit a number of cellular responses that are not easily explained by direct signaling through their classical receptor. Thrombin is an excellent example of such a molecule; it is an established modulator of endothelial cell function and has recently been described to elicit cross signaling between its G protein-coupled receptor and ErbB2 (6). Thrombin stimulation also induces HUVEC ErbB2 phosphorylation (Fig. 1B, right). Cross talk with alternate signaling pathways, such as that activated by thrombin, may play a critical role in activation of the ErbB receptors in the absence of direct ErbB ligand binding in EC.

ErbB receptors can also transduce Ca²⁺-mobilizing signals in response to stimulatory antibodies or ligands (14). With the use of an interactive laser cytometer and fluo-3 AM-loaded HUVEC, single cell Ca²⁺ transients were observed in response to both thrombin (Fig. 2A) and neuregulin-3 (Fig. 2, B-D) at the same concentrations that induce ErbB2 tyrosine phosphorylation. As expected, EGF does not induce HUVEC Ca²⁺ fluxes (Fig. 2D), which is consistent with the lack of detectable ErbB1 expression in these cells. Furthermore, neuregulin signaling is ErbB1 independent (27), as noted above. Because the recombinant neuregulin-3 used in these experiments was prepared by thrombin cleavage of a GST fusion protein (expressed in the pGEX vector), and thrombin activity results both in HUVEC Ca²⁺ fluxes and ErbB2 phosphorylation, identical Ca²⁺ signaling experiments were performed in the presence of hirudin, a potent thrombin antagonist. Hirudin completely abolished thrombin-mediated Ca²⁺ transients but had no effect on neuregulin-3 responses (Fig. 2C).

View larger version (3012K): [in this window] [in a new window]   Fig. 2.   Effects of neuregulin and epidermal growth factor (EGF) on HUVEC calcium fluxes. HUVEC monolayers were loaded with flou-3 in 96-well plates. Agonists (or controls) were added at indicated times (arrows). Representative curves are shown from 2 to 3 cells per field, as measured using an interactive laser cytometer. Curves were normalized to the background (=1) fluorescence for each gated cell. Relative intracellular calcium levels (normalized fluorescence) were determined in response to recombinant human thrombin (1 U/ml) (A), neuregulin-3 (500 ng/ml) (B), thrombin (1 U/ml) followed by neuregulin-3 (100 ng/ml) in presence of thrombin inhibitor hirudin (100 U/ml, added just before start of each experiment, i.e., before time 0) (C), recombinant human EGF (500 ng/ml) followed by neuregulin-3 (100 ng/ml) (D), and pGEX control peptide (500 ng/ml) (E).

Effects of neuregulin on HUVEC growth. Because receptor-mediated tyrosine phosphorylation and Ca²⁺ signaling commonly accompany mitogenic responses, the effect of neuregulin-3 on HUVEC growth was evaluated. Cells were plated in growth factor-depleted medium (5% serum without ECGS or heparin, normally required for HUVEC proliferation in vitro) and treated with neuregulin-3, control pGEX peptide, or VEGF for 10 days. Figure 3 demonstrates 10-day HUVEC growth curves, with neuregulin-3 (100 ng/ml) present throughout the experiment, achieving a 25-fold increase in cell number at day 10, similar to that seen with VEGF (10 ng/ml). Because several cytokines (e.g., FGF-4) augment EC growth and induce angiogenesis through VEGF induction and secretion (7), HUVEC growth responses to neuregulin-3 were evaluated in the presence of neutralizing anti-VEGF antibody. Figure 3 shows an antibody-mediated 85% inhibition of VEGF-stimulated EC proliferation, with no effect on neuregulin-stimulated HUVEC growth.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Effects of neuregulin on HUVEC proliferation. Medium control, neuregulin-3 (50 ng/ml), or VEGF (50 ng/ml) were added to serum- and mitogen-depleted (1% FBS, no ECGS) HUVEC in the presence and absence of neutralizing anti-VEGF antibody (0.5 mg/ml) and harvested from single wells (in triplicate) for viable cell counts (trypan blue exclusion). Data points represent means ± SD of well counts.

Effects of neuregulin on HUVEC tube formation in collagen gels. As suggested above, EC growth plays a central role in numerous physiological and pathological processes. For example, endothelial proliferation is required during the neovessel formation (angiogenesis) of wound healing. An in vitro tube formation assay was used to begin assessing whether neuregulin provides an angiogenic stimulus. Figure 4 demonstrates that within 24-36 h HUVEC grown in a two-dimensional collagen gel matrix form elongated tube structures in response to VEGF (Fig. 4B), neuregulin-3 (Fig. 4D), and neuregulin-1 (Fig. 4F) but not to control peptide (Fig. 4A). As with the EC proliferation assays, the neutralizing anti-VEGF antibody had no effect on neuregulin-induced tube formation (Fig. 4E), whereas it largely abrogated the VEGF response (Fig. 4C). These findings demonstrate that endothelial proliferative and in vitro angiogenic responses to neuregulin do not require VEGF activity.

View larger version (91K): [in this window] [in a new window]   Fig. 4.   HUVEC tube formation in collagen gels. Control pGEX peptide (100 ng/ml) (A), vascular endothelial cell growth factor (VEGF, 50 ng/ml) (B), VEGF (50 ng/ml) + anti-VEGF (1 µg/ml) (C), neuregulin-3 (50 ng/ml) (D), neuregulin-3 + anti-VEGF (1 µg/ml) (E), or neuregulin-1 (100 ng/ml) (F) were added in ECGS-free medium to confluent HUVEC monolayers overlaid with type I collagen (1.5 mg/ml). Photomicrographs were obtained 24 h after stimulus addition. Patches of detaching, round refractile cells, as well as areas of degenerating tubelike structures, are seen in A and C. Results are representative of 4 separate experiments. Magnification ×100.

In vivo angiogenesis induced by neuregulin. Although in vitro tube formation provides a useful method for testing the effects of defined factors on angiogenesis involving isolated EC, the relevance of this model to in vivo angiogenesis is less clear. Therefore, an established rat corneal angiogenesis model (23) was used, in which Hydron pellets containing various doses of neuregulin, VEGF, or control pGEX peptide were surgically placed in the eyes of rats, and corneal neovascularization was assessed 7 days later. As little as 10 ng of neuregulin induced corneal neovessels (Fig. 5B), with a marked angiogenic response to 50 ng (Fig. 5C), similar to that seen with VEGF (25 ng) (Table 1). A dose response to neuregulin-3 was observed (Table 1) with significant corneal inflammation observed only at the highest dose tested (200 ng). Thus the angiogenic response observed with lower neuregulin doses is not due to the elaboration of cytokines from invading inflammatory cells.

View larger version (27K): [in this window] [in a new window]   Fig. 5.   Effect of neuregulin on rat corneal angiogenesis. Pellets containing pGEX control peptide (50 ng/pellet) (A), neuregulin-3 (10 ng/pellet) (B), or neuregulin-3 (50 ng/pellet) (C) were implanted in intracorneal pocket. Corneal sections were performed 7 days after implantation. Outgrowths of neovessels can be seen extending from cornea (bottom) toward implanted pellets (top) in B and C.

View this table: [in this window] [in a new window]   Table 1.   Neovascular responses induced in rat corneas by neuregulin-3

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Receptor tyrosine kinases of the EGF-r family transduce mitogenic signals in numerous cell types. This has been most extensively studied in neoplastic cells and nervous system development. Our findings provide the first evidence that ErbB receptors (apart from EGF-r) transduce biochemical and functional signals in EC and that neuregulin is angiogenic. The presence of ErbB2, ErbB3, and ErbB4, but not ErbB1, in HUVEC is consistent with endothelial responses to neuregulin. Neuregulin-1 has been shown to bind to ErbB3 and ErbB4 and stimulate tyrosine kinase activity of ErbB2 and ErbB4 in an ErbB1-independent manner (27). In all other cell types previously studied, ErbB2 phosphorylation induced by neuregulin appears to occur through association with ErbB3 or ErbB4 (which preferentially heterodimerize with ErbB2), because cells expressing only ErbB2 homodimers appear incapable of sufficient ligand binding to promote kinase activation (11, 27).

The downstream signaling effects of neuregulin in EC share several similarities with other signaling pathways, such as those activated by fibroblast growth factor (FGF) or VEGF, known to regulate angiogenesis. For example, the induction of calcium fluxes in EC after neuregulin treatment may be a critical part of angiogenic signaling, because inhibitors of ligand-stimulated Ca²⁺ influx have been shown to prevent FGF-induced angiogenesis (16). Furthermore, a recent study demonstrated that neutralizing antibodies directed against ErbB2 reduce tumor cell VEGF secretion, and expression of the oncogenic form of ErbB2 increases NIH/3T3 VEGF secretion (24). This raises the interesting possibility that neuregulin-ErbB signaling activates angiogenesis by inducing EC VEGF secretion and VEGF receptor signaling. However, our data show that anti-VEGF antibodies, in concentrations sufficient to block high levels of VEGF, had no effect on neuregulin-stimulated HUVEC growth or tube formation, demonstrating that these responses are VEGF independent.

Our findings provide the first evidence that ErbB receptors (apart from EGF-r) transduce biochemical and functional signals in EC and that neuregulin is angiogenic. This is particularly interesting in the context of recent studies demonstrating lethal cardiovascular abnormalities in neuregulin-1, ErbB2, ErbB3, or ErbB4 homozygous-deleted (null) mice (2). This information also supports the newly emerging concept that blood vessel and neuronal development share common signaling pathways, as has been shown for neuropilin-1, ELK, and CXCR4 receptor signaling (23, 29, 32, 34, 39). Furthermore, the finding that administration of nerve growth factor to human corneas results in neovascularization as well as healing of neurotrophic ulcers suggests that there is overlap, or perhaps even cross talk, between pathways promoting nerve growth and angiogenesis in humans (18).

Data suggesting that receptors responsible for angiogenesis have opposing effects depending on the ligand present are emerging from several diverse fields. In the case of the neuropilin-1 receptor, stimulation by the ligand Semaphorin III appears to repel neuronal cell growth, whereas VEGF stimulation promotes blood vessel growth (13, 22, 32). This interesting dichotomy of responses also exists for the Tie2 receptor, which promotes angiogenesis when liganded by angiopoietin-1 and disrupts blood vessel formation when liganded by angiopoietin-2 (20). The current repertoire of ligands for the ErbB receptors includes EGF, amphiregulin, betacellulin, neuregulin, and others. Although these ligands have been shown to have differential effects on ErbB signaling, no such ligand-dependent opposing effects have been demonstrated for the mammalian ErbB receptor family. However, the Drosophila ErbB receptor homologue DER is positively regulated by its ligand Vein (a neuregulin homologue) and negatively regulated by its ligand Argos in a manner that is dependent on the EGF-like domain of the stimulatory ligand (30, 31). Although there is no currently known mammalian homologue of Argos, the high level of conservation of these signaling pathways raises the possibility that an anti-angiogenic ErbB receptor ligand exists in mammals. Pro- and anti-angiogenic regulation of this pathway may have important developmental and clinical implications.

Neuregulin mRNA has been shown to be present in cells from both the endocardial endothelium of mice and the coronary microvascular endothelium of rats (3, 37, 38). These sources of neuregulin may play an important role in paracrine and autocrine signaling in the heart, regulating both myocyte and endothelial cell structure and function. The finding that ErbB2 and ErbB4 expression is maintained in adult rat hearts (38) and our data showing that neuregulin can induce neovascularization in adult rats suggest that this signaling pathway may function both in embryonic and adult cardiovascular tissues.

Finally, neuregulin stimulation of the ErbB2, ErbB3, and ErbB4 receptors results in activation of several signaling pathways that are shared by other tyrosine kinase receptors involved in angiogenesis, such as the VEGF receptors (Flt-1 and Flk-1/KDR) and the Eph receptors (ELK and Eck), including the mitogen-activated protein kinase and c-Jun kinase pathways (10, 12, 33). These shared signaling pathways may provide valuable clues about the fundamental mechanisms regulating endothelial cell proliferation, migration, and angiogenesis.