The axonal attractant Netrin-1 is an angiogenic factor

Experimental Procedures

Expression Studies. Total RNA was isolated from human umbilical vein endothelial cells (HUVEC), human microvascular endothelial cells (HMVEC), and human aortic VSMCs by using TRIZOL (GIBCO/BRL). Human brain RNA was purchased from BD Biosciences. The RNA was reverse-transcribed with the RETROscript kit (Ambion) according to the manufacturer's instructions. The primer pairs for human DCC were 5′-ACAGGCCTCAAACCAAACAC-3′ (forward) and 5′-ACCTCCATCTCCATGACGAC-3′ (reverse). Those for human neogenin were 5′-ACCCCAGCCTGTGATTAGTG-3′ (forward) and 5′-TGTGATGGTTCAGAGCTTGC-3′ (reverse). The primer pairs for GAPDH were 5′-ACCCAGAAGACTGTGGATGG-3′ (forward) and 5′-TGCTGTAGCCA A ATTCGTTG-3′ (reverse). The following conditions were adopted for amplification of DCC and neogenin: denaturation at 94°C for 30 seconds, annealing at 58°C for 30 seconds, and extension at 72°C for 45 seconds, 35 cycles. The conditions used to amplify GAPDH were denaturation at 94°C for 30 seconds, annealing at 58°C for 30 seconds, and extension at 72°C for 45 seconds, 30 cycles.

Immunohistochemistry and Western Blot Analysis. Immunohistochemistry and Western blot analyses were performed as described (20, 26). Primary antibodies that recognize Netrin-1 (Oncogene), Deleted in Colorectal Cancer (DCC) (Oncogene), Neogenin (Santa Cruz Biotechnology), Unc5H2 (two different antibodies: gifts from Lindsay Hinck, University of California, Santa Cruz; polyclonal antibodies were raised against a GST-ectodomain of Unc5h2 fusion protein), and CD31 (Dako) were used in these studies.

In Vitro Assays. BSA and PDGF-BB were obtained from Sigma. Recombinant human VEGF-165, FGF-2, and Netrin-1 (mouse and chicken) were purchased from R & D Systems. Fibronectin, Laminin-1, Collagen I, and Collagen IV were from BD Biosciences. Human aortic VSMCs, human aortic endothelial cells (HAEC), and HMVEC were cultured according to supplier's instructions (Cambrex, East Rutherford, NJ). Rodent aortic VSMCs were isolated as described (27). All cell biology assays were performed by individuals blinded to treatments and repeated on at least two independent vials of primary cells, with each experiment reproduced three separate times, and each condition within an experiment performed in triplicate.

We used the most direct measurement of mitogenic activity, that is, cell number. Low passage primary endothelial cells or VSMCs were seeded onto 24-well plates at equal density and serum-starved for 16 h. After starvation, factors (BSA, PDGF-BB, VEGF-165, and Netrin-1) were added to the media (EBM containing 0.5% FBS and 0.1% BSA for endothelial cells; DMEM containing 0.1% BSA for VSMCs) at indicated concentrations. After 24, 48, and 72 h of treatment, cell number was measured with a hemacytometer or by counting fixed and stained cells in the well. Fold increase over BSA-treated wells was used to compare different cell types, time points, and test factors. Representative fields of cells stained with hematoxylin and eosin (Sigma) were photographed.

To measure migration, vascular endothelial cells were serum-starved overnight and then seeded at a density of 40,000 cells per well onto 8-μm Transwell inserts (Costar). Test factors were added in serum-free media and placed in the lower chamber. After incubation at 37°C for 3 h, filters were fixed with Zamboni's fixative and stained with a 4′,6-diamidino-2-phenylindole (DAPI) stain kit (Fisher Scientific). The total number of migrated cells was calculated by counting five random fields; this number was then expressed as fold increase. VSMC migration was performed as described above, except 30 ng/ml PDGF-BB was used as the positive control. For blocking experiments, cells were pretreated with 0.1 μg/ml of antibodies that recognize either Neogenin or DCC for 30 min before addition of cells to the upper chamber.

To measure adhesion, 96-well plates were coated with BSA, Netrin-1, Fibronectin, Laminin-1, Collagen I, or Collagen IV for 16-20 h at 4°C. Wells were blocked with PBS containing 1% BSA for 1 h at room temperature before the addition of cells. Cells were harvested by trypsinization and neutralized with growth media, washed twice, and then resuspended in serum-free DMEM containing 0.5% BSA. A total of 5 × 10⁴ cells were added to each well and allowed to attach for 30 min at 37°C in a CO₂ incubator. After washing three times with PBS, cells were fixed, stained with hematoxylin and eosin, and counted. For blocking experiments, cells were pretreated for 30 min with 10 μg/ml of neogenin antibody.

In Vivo Assays. Chorioallantoic membrane assay (CAM) was performed as described (28, 29). Briefly, fertilized Leghorn chicken eggs were incubated under conditions of constant humidity (60%) at 37°C for 72 h. On days 3-4, eggs were opened into sterile clingwrap hammocks and incubated at 37°C with 2.0% CO₂ and 90% relative humidity for 4 more days. At days 7-8, test samples were diluted 1:1 in sterile 1.0% methylcellulose and applied to 1-mm³ Gelfoam sponges (Upjohn). The sponges were overlayed with a 3 × 3 mm 250-μm-pore nylon mesh (Sefar America, Depew, NY) and cured on 13-mm Thermanox coverslips (Nalge Nunc), for 2 h at room temperature under sterile conditions. Assembled test samples were placed on the CAMs and incubated at 37°C for a treatment period of 72 h. The CAMs were then fixed in 4% paraformaldehyde/2% glutaraldehyde/PBS and photographed on a Leica MZ 12.5 equipped with a Zeiss Axiocam digital camera. The average number of microvessels surrounding the perimeter of the implanted mesh in each image was calculated by image j (NIH image) or a blinded reviewer and expressed as fold increase over BSA-treated CAMs.

For the murine corneal micropocket assay, hydron (Hydro Med, Cranbury, NJ) pellets containing sucralfate (Sigma), and the indicated test factors were prepared and implanted as described (30, 31). Briefly, 7- to 8-week-old male C57BL/6 mice were anesthetized with i.p. avertin (Sigma). The eyes were topically anesthetized with proparacaine, and an incision was made lateral to the pupil with a no. 10 surgical blade. A corneal micropocket was dissected from this incision toward the limbus with a von Graefe knife no. 3. A pellet was implanted in this micropocket, and topical erythromycin was applied. Five to six days after pellet implantation, neovascularization was quantitated by visualization with a slit lamp microscope. The neovascularized area was computed with the formula 2π/10 × clock hours × vessel length (mm).

Results

To determine whether Netrin-1 is a vascular mitogen, we measured cell number after addition of Netrin-1 to primary endothelial cells or VSMCs. We compared the responses of endothelial cells to Netrin-1 versus VEGF, a known endothelial mitogen. As shown in Fig. 1, primary HAEC and HMVEC proliferate in response to Netrin-1. Netrin-1 elicited the same level of proliferation as VEGF in both cell types, although the response of HMVEC consistently exceeded that of HAEC (Fig. 1A). Dose-response experiments revealed that the activity of Netrin-1 peaked at 0.7 nM (50 ng/ml) and produced similar levels of endothelial proliferation as 0.6 nM (10 ng/ml) VEGF (Fig. 1B). Netrin-1 also serves as a mitogen for VSMCs, producing similar levels of proliferation as PDGF, a known VSMC mitogen (Fig. 1 C and D). In contrast to endothelial cells, the mitogenic activity of Netrin-1 on VSMC did not peak at 50 ng/ml. Direct cell count was confirmed by using the nonradioactive colorimetric assay WST-1, and [³H]thymidine assays confirmed stimulation of DNA synthesis (data not shown).

To assess the capacity of Netrin-1 to induce migration, we measured the movement of cells through a filter toward a chamber filled with test factors. First, we compared the response of endothelial cells to Netrin-1, VEGF, and BSA. Fig. 2 A and B shows that Netrin-1 induces HAEC and HMVEC migration in a dose-dependent fashion. The migration of endothelial cells toward Netrin-1 was directional, or chemotactic, because eliminating the gradient between the two chambers blocked migration (Fig. 2C). Second, we showed that Netrin-1 was chemotactic for human and rat VSMC (Fig. 2 D-F) with a potency similar to PDGF. As observed with the mitogenic properties of Netrin-1, there were subtle differences between the responses of endothelial cells and VSMC, as VSMC migration to Netrin-1 peaked at 50 ng/ml. When nonvascular cell types such as NIH 3T3 or COS cells were used in our cell culture assays, they did not respond to Netrin-1 (data not shown). Thus, Netrin-1 is a potent mitogen and chemoattractant for both endothelial cells and VSMC with a specific activity comparable to VEGF or PDGF.

Netrin-1 has significant structural similarities to the laminins, proteins known to promote cellular adhesion (24). To test cellular adhesion, we coated cell culture wells with BSA, Netrin-1, or extracellular matrix proteins with known adhesive properties. Vascular cells were placed on the coated wells and washed after 30 min. The number of cells that remained attached to the well was counted. We found that VSMC, but not endothelial cells, adhere to Netrin-1 (Fig. 3). Moreover, Netrin-1 promoted VSMC adhesion as effectively as other vascular matrix proteins, such as Laminin-1, Collagen I, and Collagen IV. Thus, in contrast to migration and proliferation assays in which endothelial cells and VSMC both respond to Netrin-1, there is a sharp distinction between the ability of endothelial cells and VSMC to adhere to Netrin-1.

Axonal attraction by Netrin-1 is mediated by DCC receptor family members that include DCC and Neogenin (2). To determine whether these receptors are involved in the vascular response to Netrin-1, we first examined whether mRNA or protein was detected in either HMVEC or human VSMC. We failed to detect DCC transcript or protein in endothelial cells by using RT-PCR, Northern blot, or Western blot analysis (Fig. 4 A and B and data not shown). However, we were able to detect a neogenin transcript and gene product in VSMC, (Fig. 4 A and B and data not shown). In a specific subset of nerve cells, Netrin-1 can function as a repulsive cue, and this activity is mediated by a heterodimer comprised of DCC and a coreceptor, Unc5H2 (7). Although we have confirmed reports by others that Unc5H2 is expressed during murine embryonic angiogenesis (32), we did not detect expression of Unc5H2 in endothelial cell or VSMC lines by Northern or Western blot analysis. By RT-PCR, a faint amplified product for Unc5H2 was detected only after 35 cycles (data not shown). These expression data suggest that vascular expression of Unc5H2 may be different in mouse embryonic development versus the adult human cell lines. We examined whether Neogenin mediates Netrin-1 signaling in VSMC by using both migration and adhesion assays. Functional blocking Neogenin antibodies inhibited Netrin-1-mediated migration and adhesion of VSMC (Fig. 4 C and D). The effect was specific, because these antibodies failed to block either PDGF-induced migration of VSMC or Netrin-1-induced migration of primary endothelial cells that do not express neogenin. Thus, Neogenin mediates Netrin-1 signaling in VSMC.

To determine whether Netrin-1 functions as an angiogenic factor in vivo, we measured the response of blood vessels to Netrin-1. In the first model, gelfoam sponges soaked in 1 μg of BSA, Netrin-1, FGF, or VEGF were placed on the chorioallantoic membrane of chick yolk sacs, and the number of vessels invading these sponges was measured after 72 h. Netrin-1 and VEGF each stimulated a 2-fold increase in angiogenesis over BSA control sponges (Fig. 5). In a second model, hydron pellets with equimolar amounts of either Netrin-1 (200 ng) or VEGF (50 ng) were placed on murine corneas and induced equivalent angiogenic responses after 5 days (Fig. 6 A and B). Control pellets elicited no blood vessel formation. Thus, Netrin-1 can function as an angiogenic factor in both in vivo and in vitro assays. Interestingly, when equimolar amounts of the two factors were applied together, the response was robust and greater than the sum of the responses to each of the factors individually. To examine the relationship of Netrin-1 and VEGF in our in vitro assay, low doses of Netrin and VEGF were applied in a migration assay. Migration to 10 ng/ml Netrin-1 or 2.5 ng/ml VEGF was <2-fold relative to the BSA control. However, when Netrin-1 and VEGF were combined, the response was greater than the sum of each factor individually (Fig. 6C). Together, these data suggest that Netrin-1 and VEGF may act synergistically to induce angiogenesis both in vitro and in vivo.

Discussion

The superimposable patterns of blood vessels and nerves suggests that distal targets might secrete factors that attract both vessels and nerves (14, 15). Here, we show that the neural attractant, Netrin-1, has a previously undescribed role in promoting angiogenesis. First, cell culture experiments demonstrate that Netrin-1 specifically stimulates proliferation, migration, and adhesion of vascular cells with potencies comparable to VEGF, PDGF, or Fibronectin. Second, Neogenin mediates netrin signaling in VSMC. The lack of expression of the known netrin receptors such as DCC, Neogenin, or Unc5H family members suggest that these factors do not mediate the proangiogenic effects of Netrin-1 on endothelial cells. Recent data indicate a role for integrins in Netrin-1-mediated adhesion (24); however, Netrin-1 does not promote endothelial adhesion, and blocking integrin signaling has no effect on the response of endothelial cells to Netrin-1 (data not shown). Together, these data suggest that an unidentified netrin receptor activates endothelial cell proliferation and migration. Different netrin receptors for endothelial cells and VSMCs may explain the differences in the response of these two cell types to Netrin-1 in migration and adhesion assays. Finally, the chorioallantoic membrane and the murine corneal micropocket experiments demonstrate that Netrin-1 activates blood vessel growth in vivo. Thus, here we report a canonical neural guidance factor that attracts axons and promotes angiogenesis in vitro and in vivo.

Our data strongly suggest a role for Netrin-1 in angiogenic growth stimulation. We found that netrin-1 is also expressed in nonneural tissues during the early stages of murine embryonic development when angiogenesis is underway. High levels of netrin-1 mRNA are evident within the developing somites that flank intersomitic vessel tracts (data not shown). Given these results, one might predict that early vascular development would be defective in embryos lacking netrin-1. To determine whether Netrin-1 plays a role in developmental angiogenesis, we examined two model systems in which netrin activity was attenuated. First, we examined murine embryos that were homozygous for a strong hypomorphic allele of netrin-1 at embryonic days 9.5 and 15 and postnatal day 0.5, but found no evidence of primary cardiovascular abnormalities. Similarly, morpholino knockdown experiments of Netrin-1a and 1b in zebrafish (the only known netrin orthologues in zebrafish) did not alter cardiovascular development (data not shown). There are two possible explanations for these observations. First, there are at least three netrin family members, netrin-1, netrin-2, and netrin-4, as well as two genes encoding Netrin-G1 and Netrin-G2. These proteins have all been implicated in directing axonal projections, although only mice lacking netrin-1 have been described (5, 22, 33-36). Thus, functional redundancy in the vascular system is plausible, and generation of multiple mutant mice, and intercrosses thereof, will be required to assess the roles of netrin family members in embryonic vascular development. A second explanation for the apparent lack of an embryonic vascular phenotype in netrin-1 hypomorphs is the possibility that there are differences in the potency of netrin-1 between adult and embryonic vascular cells. Examination of adult tissues revealed that Netrin-1 protein is expressed within and surrounding blood vessels (data not shown). The data we present herein were derived entirely from adult endothelial cells and VSMCs. To address the possibility of differences between embryonic and adult cells, it will be necessary to purify embryonic endothelial cells and assess their behavior to Netrin-1 treatment directly.

Not only does Netrin-1 induce angiogenesis, it augments the angiogenic activity of VEGF in both in vitro and in vivo systems (Fig. 6). Recent reports indicate that VEGF is a neurotrophic factor, suggesting that the combination of VEGF and netrin family members may work in synergy to induce the vascularization and innervation of distal targets. Although none of the other neural guidance cues are known to have mitogenic activity individually or in combination with VEGF, there is evidence that axonal repulsive cues such as semaphorins, slits, and ephrins either inhibit endothelial migration or block angiogenesis (17, 20, 37-39). Thus, it is likely that neural attractive and repulsive cues, and pro- and antiangiogenic peptides, may act as shared guidance signals that coordinate the patterning of parallel networks such as blood vessels and nerves, and potentially epithelial tubes (23, 25). If guidance signaling programs share regulatory factors in both vascular and neuronal systems, we would expect that, in certain biological contexts, netrins may also act as repulsive or antiangiogenic agents. This expectation is based on evidence that, in axonal guidance, netrin can be bifunctional, mediating attraction and repulsion, dependent on expression and heterodimerization of the DCC and Unc5H2 receptors (7). Future studies will be needed to determine whether netrins have dual roles in vascular guidance.

Due to their therapeutic potential, there is considerable interest in the identification of new vascular mitogens. The major families of vascular mitogens include multiple genes and isoforms of the VEGF, PDGF, and FGF families. The biologic and therapeutic roles of these factors have been studied extensively (40, 41). The recent identification of endocrine gland-selective VEGF, EG-VEGF, raises the possibility that new classes of tissue-specific endothelial mitogens may await discovery (42). Similarly, we would not be surprised if Netrin-1 is simply the first of many neural guidance factors to induce axonal projection and to function as a potent vascular mitogen. The potential for treating ischemic vascular diseases, cancer, or retinopathy by targeting netrins is evident and needs to be explored. To fully realize this potential, the identity of the endothelial receptor responsible for the angiogenic activity of Netrin-1 and the interplay between each netrin family member with other angiogenic peptides and neural guidance factors must be defined.