Neovascularization of ischemic tissues by gene delivery of the extracellular matrix protein Del-1

The ECM protein Del-1 is one of several novel ECM proteins that accumulate around angiogenic blood vessels in embryonic and tumor tissue and promote angiogenesis in the absence of exogenous growth factors. Del-1 expressed in mouse or rabbit ischemic hind-limb muscle by gene transfer rapidly promotes new blood vessel formation and restores muscle function. This angiogenic ECM protein initiates angiogenesis by binding to integrin αvβ5 on resting endothelium, thereby resulting in expression of the transcription factor Hox D3 and integrin αvβ3. Hox D3 converts resting endothelium to angiogenic endothelium by inducing expression of proangiogenic molecules such as integrin αvβ3. These findings provide evidence for an angiogenic switch that can be initiated in the absence of exogenous growth factors and indicate that the angiogenic matrix protein Del-1 may be a useful tool for the therapy of ischemic disease.

The development of new blood vessels from an existing network of vessels, angiogenesis, occurs during embryonic development, cancer, and arthritis and in response to ischemia (1–7). Novel angiogenesis-inducing therapies have been developed recently to stimulate collateral vessel formation and to improve muscle function in the treatment of occlusive vascular disease, such as myocardial and limb ischemia (5–7).

Angiogenesis requires signals that are transduced from growth factor receptors and integrins (1–4). Soluble growth factors initiate angiogenesis, whereas integrins promote endothelial cell (EC) survival and migration (1–4). Recent studies, however, have identified an alternative mechanism for angiogenesis induction. Several proangiogenic ECM proteins (Del-1, Cyr 61, Nov, connective tissue growth factor/Fisp12, and others) are found in association with vascular endothelium and promote αvβ3 integrin–dependent angiogenesis in the absence of exogenously added growth factors (8–13). One of these “angiomatrix” proteins, Del-1, is a 52-kDa protein that contains three Notch-like EGF domains, an arginine-glycine-aspartic acid (RGD) motif, and two F5/F8 C-terminal domains (13), such as are found in discoidin and factor VIII.

Del-1 promotes EC migration and angiogenesis that is blocked by antagonists of integrin αvβ3 (11). It is unclear how Del-1 or any of the other angiomatrix proteins induce angiogenesis, however, since their known receptor, integrin αvβ3, is poorly expressed on resting ECs and requires prior EC activation (14, 15). Clearly, alternative receptors for Del-1 must exist on ECs. In this report, we evaluate the potential of Del-1 to stimulate therapeutic angiogenesis, and we investigate the mechanisms by which it initiates angiogenesis. We demonstrate that integrin αvβ5 serves as a receptor for Del-1, thereby promoting cell migration and survival. Furthermore, ligation of EC integrin αvβ5 is required for the initiation of angiogenesis by Del-1. Ultimately, Del-1–αvβ5 interactions induce αvβ3 expression and angiogenesis.

Del-1 promotes angiogenesis and functional recovery in models of hind-limb ischemia. The proangiogenic ECM protein Del-1 is expressed by ECs during embryonic and tumor development (11–13). Del-1 promotes angiogenesis in chick, murine, and human in vivo models of angiogenesis (11, 12). In contrast, ECM proteins such as fibronectin, fibrinogen, vitronectin, and collagen are unable to induce angiogenesis (11).

To evaluate the potential of human Del-1 to induce neovessels in ischemic tissues, we induced sustained expression of human Del-1 as well as human VEGF₁₆₅ in mouse tibialis muscle by injection of hDel-1 and hVEGF encoding expression plasmids (Figure 1a). Injection of reporter gene constructs encoding GFP and luciferase demonstrated that optimal expression could be achieved with this plasmid formulation injected at multiple sites in tibialis muscle (18, 19). In vivo expression of hDel-1 and hVEGF₁₆₅ was readily detected in treated murine tibalis anterior muscles, but not in untreated or control-transfected muscle, by immunohistochemical analysis 7 days after injection of formulated plasmid (Figure 1b). Arrows denote fibers that exhibit positive staining. Similar results were obtained in Western blot analysis of muscle lysates (Figure 1c) and RT-PCR studies (data not shown). The Del-1 and VEGF detected was due to induced expression of human genes, since no expression of Del-1 or VEGF was detected in control-transfected tissues. Thus, these studies indicate that gene transfer leads to significant induction of human Del-1 expression in vivo.

Figure 1

Del-1 promotes angiogenesis and restores function to murine ischemic muscle in vivo. (a) Expression constructs used to induce hDel-1 and hVEGF expression in vivo contained a CMV enhancer/promoter (enh/pro), 5′ UTR, an optimized intron (IVS-8), cDNA encoding either VEGF₁₆₅ (pVF1164) or full-length human Del-1 (pDL1599), and the hGH 3′ UTR, as well as a kanamycin resistance gene. (b) Expression of Del-1 or VEGF₁₆₅ in hDel-1 (left) and hVEGF₁₆₅ (right) transgene-injected, noncoding plasmid–injected, and untreated muscle. Photographs were taken at ×100 magnification. Arrows denote fibers exhibiting positive staining. Cntl, control. (c) Lysates of muscle injected with hDel-1 (Del-1) or noncoding control plasmid as well as purified Del-1 (pur.Del-1) were immunoblotted for Del-1 expression using rabbit anti–Del-1 Ab’s. (d) Mouse model of hind-limb ischemia induced by ligation of the femoral artery. (e and f) Capillary/myofiber ratio 7 days after treatment with noncoding (control), hDel-1, or hVEGF₁₆₅ expression plasmids. (e) Cryosections of treated muscle immunostained using a rat anti-mouse CD31 Ab. (f) Mean capillary/myofiber ratio ± SEM for three animals per group (*P < 0.01). (g) Hind-limb ischemia was induced by bilateral ligation of the femoral artery of male CD1 mice. Animals were placed on the treadmill for 5 minutes at 5 m/min and then at 10 m/min thereafter. The speed was increased every 2 minutes until fatigue. Results represent the mean ± SEM (n = 7–8 animals/group) treadmill run time for mice treated with formulated noncoding hDel-1 or hVEGF₁₆₅ expression plasmids. Asterisks indicate statistically significant result relative to noncoding plasmid control (P < 0.05).

When tested in a mouse model of hind-limb ischemia (Figure 1d), intramuscular injections of formulated hVEGF and hDel-1 plasmids were equally effective in inducing neovessels (Figure 1, e and f) and in restoring hind-limb function (Figure 1g). Injection of hVEGF- or hDel-1–encoding plasmids, but not noncoding plasmid, induced the growth of CD31-positive neovessels in transfected muscle (Figure 1e). Capillary/myofiber ratios in both hVEGF- and hDel-1–treated muscle were approximately 1.7-fold greater (P < 0.01) than in control-treated muscle (Figure 1e). Bilateral ligation of the femoral artery in CD-1 male mice dramatically inhibited mouse performance in a treadmill stress test (P < 0.05). Expression of either Del-1 (P < 0.05) or VEGF (P < 0.05) significantly increased muscle function and performance in the treadmill test, with improved performance lasting over 4 weeks (Figure 1f). These studies indicate that the Del-1 protein can be expressed through gene transfer in vivo and that Del-1 can induce angiogenesis and improve function in murine ischemic muscle.

To determine whether Del-1 promotes angiogenesis and recovery in larger animal models of ischemia, formulated, noncoding hVEGF₁₆₅ or hDel-1 plasmids were injected into ischemic rabbit hind limbs. Unilateral hind-limb ischemia was induced in New Zealand white rabbits by surgical resection of the femoral artery (20). Pilot expression studies using reporter genes (GFP and luciferase) demonstrated that optimal expression could be achieved in the rabbit muscle with the plasmid formulation injected at ten different sites surrounding the medial thigh (Figure 2a). Injection of hVEGF and hDel-1 plasmids resulted in in vivo expression of the transgenes (Figure 2, a and b) and induced the formation of new collateral vessels that cross the medial thigh (Figure 2c). Expression of VEGF or Del-1 genes could be detected in random samplings of injected, but not noninjected, muscle tissues across the medial thigh (Figure 2b). Both hVEGF and hDel-1 induced the formation of threefold more new blood vessels crossing the medial thigh than did noncoding plasmid (P < 0.01) (Figure 2, c and d). A similar index of new blood vessel induction by hVEGF and hDel-1 was indicated by quantification of CD31 expression levels in muscle lysates by immunoblotting (P < 0.01) (Figure 2e). Therefore, in rabbit and murine models of hind-limb ischemia, Del-1 is potent angiogenic factor. Thus, Del-1 is a useful stimulator of in vivo angiogenesis with potential therapeutic value for ischemic disease.

Figure 2

Del-1 induces angiogenesis in rabbit model of ischemia. (a) Angiography was performed after unilateral hind-limb ischemia induced in New Zealand white rabbits (3–4 kg body weight) by surgical resection of the femoral artery (19). X’s indicate the ten sites in and surrounding the medial thigh (midthigh) (bracket) in the upper leg, where plasmid was injected (0.5 mg/site). (b) Sustained transgene expression induced by plasmid injection in muscle was detected 1 month after therapy by RT-PCR for hVEGF₁₆₅ and hDel-1. VEGF or Del-1 gene expression was detected in VEGF- or Del-1–plasmid–injected muscle, respectively, but not in control plasmid–injected muscle. MW, molecular weight standards. (c) Angiography was performed before and 1 month after therapy for each treatment. Brackets indicate medial thigh region. (d) Mean number of new vessels (± SD) crossing the medial thigh (area under the brackets in c) for each treatment 1 month after therapy. Asterisks indicate statistically significant result (P < 0.01). (e) Mean CD31 protein expression (± SD) in muscle tissue 1 month after therapy. CD31 levels within injected muscle were assessed by Western blot analysis of lysates of muscle tissue. Asterisks indicate statistically significant result relative to noncoding plasmid control (P < 0.01). Statistical significance for angiography and CD31 analyses was assessed by one-way ANOVA with plasmid treatment as the main effect.

Integrin αvβ5 is a receptor for Del-1. Since Del-1 and other angiomatrix proteins are potent proangiogenic factors, it is important to understand the mechanisms by which these proteins induce angiogenesis to enable greater control over their roles in pathological or therapeutic angiogenesis. It is unclear how Del-1, or other angiomatrix proteins, initiates angiogenesis, because integrin αvβ3, a known receptor for Del-1, is poorly expressed on quiescent vascular endothelium and is expressed at high levels only after growth-factor stimulation of endothelium (15). To determine if integrins that are expressed on quiescent endothelium, such as αvβ5, can function as receptors for Del-1, we incubated purified integrins plated on microtiter plates with dilutions of purified, biotinylated Del-1 (Figure 3a). Biotinylated, purified Del-1 bound with strong avidity to both purified integrins αvβ3 (Figure 3b) and αvβ5 (Figure 3c) in these in vitro ligand-binding assays, as did biotinylated vitronectin. In contrast, biotinylated collagen did not bind to either of these integrins (Figure 3b). RGD, but not RGE, peptides and EDTA blocked Del-1 binding to both integrins (not shown). Del-1 does not bind purified integrin α5β1, another integrin that plays a role in angiogenesis (17). A 120-kDa fragment of purified fibronectin does bind α5β1 in a saturable manner in this assay (Figure 3d). Thus, Del-1 is a ligand for integrin αvβ5 as well as for αvβ3, but not for α5β1.

Figure 3

Integrin αvβ5 is a receptor for Del-1. (a) Biotinylated Del-1 (Del-1), vitronectin (VN), or the 120-kDa cell-binding fragment of fibronectin (FN) immunoblotted with goat–antibiotin HRP. (b–d) Biotinylated Del-1, vitronectin, collagen (COL), or biotinylated 120-kDa cell-binding region of fibronectin incubated with purified integrins αvβ5 (b), αvβ3 (c), or α5β1 (d) were quantified by determining absorbance at 405 nm. (e) FACS profile of HT29 colon carcinoma cells incubated with anti-αvβ5, anti-αvβ3, anti-β1, or no primary Ab (neg) followed by FITC-labeled secondary Ab. Data are plotted as log of fluorescence intensity versus cell number times 10^–5. (f) The adhesion of integrin αvβ5-positive/αvβ3-negative HT29 colon carcinoma cells to Del-1 or vitronectin in the presence of culture medium (med), function-blocking anti-integrin Ab’s anti-αvβ3, anti-αvβ5, or anti-integrin β1. (g) FACS profile of HUVECs incubated with anti-αvβ5, anti-αvβ3, or no primary Ab (neg) followed by FITC-labeled secondary Ab, as in e. (h) The adhesion of αvβ5-positive/αvβ3-positive HUVECs on Del-1 or vitronectin in the presence of culture medium, function-blocking anti-integrin Ab’s anti-αvβ3, anti-αvβ5, or anti-integrin α5β1. (i) FACS profile of human microvascular endothelial cells (HMVECs) incubated with anti-αvβ5, anti-αvβ3, or no primary Ab (neg), followed by FITC-labeled secondary Ab, as in e. (j) The adhesion of quiescent HMVECs on Del-1 or vitronectin in the presence of culture medium, function-blocking anti-integrin Ab’s anti-αvβ3, anti-αvβ5, anti-integrin α5β1, or anti-αvβ3 combined with anti-αvβ5 (anti-β3/β5). HPF, high-power field. Magnification, ×200.

To determine whether Del-1 serves as an adhesive ligand for cellular αvβ5, we examined the interaction of Del-1 and αvβ5 in colon carcinoma and ECs. HT29 colon carcinoma cells expressing αvβ5 but not αvβ3 (as determined by FACS analysis; Figure 3e) adhere to Del-1 and to vitronectin (Figure 3f). Adhesion to both proteins can be inhibited completely by function-blocking Ab’s directed against αvβ5 (Figure 3f), but not by isotype-matched Ab’s against αvβ3 or α5β1 integrins. Thus, αvβ5 is a cellular receptor for Del-1. To determine if αvβ5 is a receptor for Del-1 on ECs, we examined the adhesion of HUVECs to Del-1. HUVECs express αvβ5 and αvβ3 in a ratio of 1:4 (Figure 3g). HUVEC attachment to either Del-1 or vitronectin was blocked by anti-αvβ5 and αvβ3 Ab’s, but not by anti-α5β1 Ab’s (Figure 3h). HUVEC adhesion to and migration on Del-1 is 70% αvβ3 dependent and 30% αvβ5 dependent, in proportion to the relative levels of expression of these two integrins on HUVECs. Additionally, αvβ5 serves as a functional receptor for Del-1 on microvascular ECs, the cells that are generally thought to participate in angiogenesis. Integrin αvβ3 and αvβ5 are expressed at equivalent levels on human microvascular ECs (HMVECs) (Figure 3i). Both receptors mediate cell attachment to Del-1 and vitronectin (Figure 3j). Blocking both receptors inhibits essentially all cell attachment to these matrix proteins. Thus, αvβ5 is a functional cell surface receptor for Del-1 on ECs.

Integrin Del-1–αvβ5 interactions induce integrin αvβ3 expression and function. Because integrin αvβ5 is a receptor for Del-1, it may play a role in Del-1–mediated angiogenesis. We examined the relative expression of integrins αvβ5 and αvβ3 during angiogenesis using the CAM model by performing immunohistochemical analysis of integrin expression in CAMs before and after Del-1 stimulation. We used the CAM rather than a murine model because immunological reagents for these integrins cross-react with chicken but not murine tissues. The temporal expression patterns of integrins αvβ5 and αvβ3 are distinct during Del-1–mediated angiogenesis. Integrin αvβ5 is expressed at moderate levels on blood vessels in unstimulated CAMs, and its expression remains unchanged on the vascular endothelium in CAMs after angiogenesis induction by Del-1 (Figure 4, a–c), whether analyzed by immunohistochemistry (Figure 4, a and b) or by Western blot analysis (Figure 4c). In contrast, αvβ5 is first expressed at 12 hours and peaks at 24–36 hours after stimulation by Del-1 (Figure 4, a–c), as has been observed for bFGF-stimulated CAMs (14, 15). Similar results are observed by either immunohistochemical or Western blot analysis. Our studies demonstrate that Del-1, like angiogenic growth factors, induces the de novo expression of αvβ3 in vivo. These results also suggest that αvβ5 could play a role in the initiation of Del-1–mediated angiogenesis because it is constitutively expressed; it may anchor Del-1 to the surface of quiescent vascular endothelium.

Figure 4

Del-1 interactions with αvβ5 regulate integrin αvβ3 expression. (a and b) Cryosections of CAMs stimulated with Del-1 were incubated with Ab’s against integrin αvβ3 or αvβ5 in combination with Ab’s against vWF. Integrin expression was detected with a rhodamine-conjugated secondary Ab and vWF with an FITC-conjugated secondary Ab. Integrin expression on blood vessels is seen as orange-yellow structures. Representative sections were photographed at ×200. White bar indicates 10 μm. (b) Average integrin expression levels for each integrin in vessels were quantified by determining the average number of red pixels in vessels that was greater than background and dividing by the average number of green pixels greater than background. (c) Integrins αvβ3 and αvβ5 were immunoprecipitated from lysate CAMs stimulated with bFGF or Del-1. Immunoprecipitates were immunoblotted with Ab’s against the cytoplasmic tails of the β3 and β5 subunits. Ratios of αvβ3 to αvβ5 were determined by densitometry. (d) CAMs were stimulated with Del-1 in the presence or absence of saline, anti-αvβ5, or W6/32 control isotype-matched Ab’s. Cryosections of CAMs were then incubated in anti-αvβ3 and anti-vWF Ab’s. Integrin αvβ3 expression was detected with a rhodamine-conjugated secondary Ab and vWF with an FITC-conjugated secondary Ab. Representative sections were photographed at ×200. White bar indicates 10 μm. Integrin αvβ3 expression on blood vessels is seen in yellow. Blood vessels are indicated with arrows. cIgG, control IgG. (e) Expression levels of integrin αvβ3 in vessels were quantified as in b.

To evaluate the role of αvβ5 in Del-1 induction of αvβ3 expression, we treated Del-1–stimulated CAMs with saline, αvβ5 function-blocking mAb (P1F6), or isotype-matched control Ab’s prior to performing immunohistochemical analysis for αvβ3 expression. Anti-αvβ5 Ab’s inhibited αvβ3 expression on CAMs 80% (Figure 4, d and e). These results show that αvβ5 is required for Del-1–mediated induction of EC αvβ3 expression.

We further investigated the roles of αv integrins in Del-1–mediated angiogenesis by testing the effects of function-blocking Ab’s to both αv integrins in CAM angiogenesis assays. Both anti-αvβ3 and αvβ5 Ab’s inhibited Del-1–induced angiogenesis at least 85% (P < 0.01), whether we applied purified recombinant Del-1 protein (Figure 5, a and b) or Del-1–expressing adenovirus (Figure 5, c and d) to CAMs. We hypothesized that if αvβ5 is required to induce αvβ3 expression during Del-1–mediated angiogenesis, then αvβ5 regulates the early, but not the later phases, of Del-1–mediated angiogenesis. In fact, αvβ5 Ab’s more potently inhibit Del-1–mediated angiogenesis when applied early (at 0 hours) rather than late (24–48 hours) during angiogenesis (Figure 5e). In contrast, αvβ3 plays a later role in Del-1–mediated angiogenesis, because it is not significantly expressed until 12–24 hours after stimulation (Figure 4, a–c). Furthermore, unlike αvβ5 Ab’s, αvβ3 function-blocking Ab’s inhibited angiogenesis even if applied as late as 24 hours after angiogenesis initiation (Figure 5e). Because the in ovo chick embryo model of angiogenesis is a closed system, Ab’s cannot be removed from the system once they enter it. Thus, anti-αvβ3 Ab’s applied at 0 hours remain in the circulation and inhibit angiogenesis even though αvβ3 is expressed only after 24 hours of stimulation. These results indicate that both αvβ3 and αvβ5 play important roles in Del-1–induced angiogenesis, although at different time points.

Figure 5

Del-1–mediated angiogenesis requires integrins αvβ3 and αvβ5. (a and b) Del-1–stimulated chick CAMs were treated with saline, anti-αvβ3, anti-αvβ5, or control IgG (W6/32), and blood vessel branch points were counted (a) or photographed at ×10 magnification (b). Ten replicate embryos were tested per treatment group. (c and d) CAMs were treated with 10⁷ PFU of Del-1 or β-galactosidase (LacZ) expressing adenovirus, and blood vessel branch points were counted (c) or photographed at ×10 magnification (d). (e) Del-1–stimulated chick CAMs were treated with saline, anti-αvβ3, anti-αvβ5, or control IgG (W6/32) at 0, 24, and 48 hours. Blood vessel branch points were counted. Ten replicate embryos were tested per treatment group. Asterisks indicate statistically significant result relative to Del-1 stimulation (P < 0.01).

To determine if αvβ5 and αvβ3 expression are required to promote Del-1–induced angiogenesis, we evaluated the ability of Del-1 and bFGF to induce angiogenesis in integrin β5-null mice and in integrin β3-null mice (16). These mice exhibit apparently normal angiogenesis during development (16). The bFGF induced an increase in CD31-positive vascular structures to the same extent in β5^–/– mice (n = 10) as it did in age-matched sibling β5^+/– mice (n = 3) and in wild-type β5^+/+ mice (n = 5) (Figure 6, a and b). While Del-1 induced significant angiogenesis in β5^+/– (n = 3) and β5^+/+ (n = 5) mice, however, it was substantially less effective in inducing angiogenesis in β5^–/– mice (n = 10; P < 0.0001; Figure 6, a and b). Interestingly, bFGF induced angiogenesis in β3^–/– mice (n = 11) to approximately the same extent as it did in age-matched F₂ β3^+/+ hybrid mice of the same genetic background (C57BL6 × 129SF2/J) (n = 6) and in wild-type C57BL/6J β3^+/+ mice (n = 8) (Figure 6, c and d); however, while Del-1 induced significant angiogenesis in C57BL6 and F₂ hybrid β3^+/+ mice, it was ineffective in inducing angiogenesis in β3^–/– mice (n = 11; P < 0.0001). Taken together, these studies indicate that expression of integrin αvβ5 as well as integrin αvβ3 is required for Del-1–mediated angiogenesis.

Figure 6

Integrin αvβ5 is required for Del-1– but not bFGF-mediated angiogenesis. (a) Angiogenesis was induced by injection of growth factor–reduced Matrigel containing bFGF or Del-1 in homozygous integrin β5-null (β5^–/–) mice and age-matched heterozygous (β5^+/–) sibling mice as well as β5^+/+ mice. Cryosections of Matrigel plugs were immunostained with Ab’s directed against murine CD31. (b) CD31-positive vessels were quantified in five ×200 microscopic fields per plug, and the mean ± SEM for each treatment group was determined. Asterisk indicates statistically significant result relative to wild-type controls (P < 0.001). (c) Angiogenesis was induced by injection of growth factor–reduced Matrigel containing bFGF or Del-1 in homozygous integrin β3-null (β3^–/–) mice, age- and strain-matched C57BL/6 × 129S F₂ hybrid β3-positive mice (F₂), as well as age-matched β3-positive C57BL/6 mice (C57). Cryosections of Matrigel plugs were immunostained with Ab’s directed against murine CD31. (d) CD31-positive vessels were quantified in five ×200 microscopic fields per plug and the mean × SEM for each treatment group was determined. Asterisk indicates statistically significant result relative to wild-type controls (P < 0.001).

Del-1–αvβ5 interaction regulates HOX gene expression. Some angiogenic growth factors, such as bFGF, induce EC expression of the Hox D3 homeobox gene, which in turn converts ECs from the resting to the angiogenic state, in part by upregulating expression of the β3 integrin subunit and urokinase plasminogen activator (24). Therefore, we examined the possibility that the αvβ5–Del-1 interaction regulates Hox D3 transcription and, thereby, αvβ3 expression. To examine the regulation of Hox D3 and αvβ3 expression by Del-1 in vitro, we used ECs that express lower levels of αvβ3. Since ECs in culture express significant levels of αvβ3, ECs were plated on layers of basement membrane (growth factor–reduced Matrigel) composed of collagen, laminin, and proteoglycans to induce withdrawal from the cell cycle and suppress αvβ3 expression (24–29). As they withdraw from the cell cycle, ECs on basement membrane express less cyclin D1, but more cell cycle inhibitor, p21^cip, than proliferating ECs, indicating that the cells have withdrawn from the cell cycle (Figure 7a). ECs on basement membrane also show reduced integrin αvβ3 protein (Figure 7a) and reduced integrin β3 and Hox D3 message expression (Figure 7b). Hox D3 message levels were reduced 50%, whereas β3 message levels were reduced 75%. Addition of Del-1 to these cultures, however, suppresses differentiation and increases gene expression of integrin β3 and Hox D3 (Figure 7b; P < 0.015), indicating that Del-1 suppresses withdrawal of ECs from the cell cycle. These studies therefore indicate that Del-1 stimulates expression of HoxD3 and αvβ3 in ECs.

Figure 7

Del-1–αvβ5 interactions induce Hox D3 expression. (a) ECs were plated in monolayers (–BM) or on tissue-culture plates coated with thin layers of basement membrane (+BM) prior to cell lysis and Western blot analysis for cyclin D1, p21^cip, and integrins β3 and β5. (b) RT-PCR was performed for Hox D3, integrins β3 and β5, and GADPH (not shown). Relative levels of each PCR product were determined by densitometry in comparison with GADPH expression levels for each treatment. Asterisks indicate statistical significance relative to monolayer culture (–BM) (P < 0.02). (c) ECs were plated on endothelial monolayers in serum culture, BM + Del-1 (Del), or on BM + Del-1 with anti-αvβ5 function-blocking Ab’s (Del + P1F6), or control Ab’s (Del + cIgG). Integrin β5 and β5 protein expression levels were determined by Western blot analysis. Average expression levels for each integrin were determined in three experiments by densitometry. (d) RT-PCR was performed on RNA isolated from cells treated as in c for Hox D3, integrins β3 and β5, and GADPH. Relative levels of each PCR product were determined by densitometry in comparison with GADPH expression levels for each treatment. Asterisks indicate statistical significance relative to Del-1 stimulation (P < 0.02).

To determine whether Hox D3 gene expression depends on αvβ5−Del-1 interactions, we added function-blocking anti-αvβ5 Ab’s to Del-1–stimulated EC basement membrane cultures. These Ab’s significantly reduced β3 protein expression (Figure 7c; P < 0.01) and inhibited Hox D3 and integrin β3 gene expression 25–50% (Figure 7d; P < 0.02). These studies indicate that αvβ5 recognition of Del-1 is required to trigger conversion of the EC to the proliferative state. In contrast, vitronectin, another αvβ5 ligand, does not have a similar effect (not shown). Thus, integrin αvβ5 specifically regulates the EC response to the proangiogenic protein Del-1.

To determine whether Hox D3 is required for Del-1–mediated angiogenesis, we evaluated the effects of Hox D3 antisense on angiogenesis in the CAM. Because human and chicken Hox D3 genes are highly conserved, with overall DNA sequence identity of 83%, human Hox D3 sense and anti-sense constructs (24) were applied to CAMs stimulated by bFGF or Del-1. Direct application of DNA to CAMs results in significant gene expression in proliferating cells such as ECs (30). Expression of the Hox D3 antisense, but not sense, construct in chick CAMs significantly blocked angiogenesis induced by bFGF and Del-1 (Figure 8a; P < 0.01). Chicken Hox D3 gene expression was induced by bFGF and Del-1, but was significantly blocked by Hox D3 antisense constructs, indicating that the antisense had inhibited expression of the chick Hox D3 gene (Figure 8, b and c). Expression of chicken and human Hox D3 in the CAM was quantified by RT-PCR (Figure 8, b and d), while GFP expression was verified by Western blot analysis of CAM lysates (Figure 8e). Hox D3 antisense suppressed integrin β3 but not β5 expression in both bFGF- and Del-1–stimulated CAMs (Figure 8e). These studies indicate that Del-1–mediated angiogenesis requires activation of an angiogenic program that leads to Hox D3 expression and subsequently to αvβ3 expression.

Figure 8

Hox D3 is required for Del-1–induced angiogenesis. (a) CAMs of 10-day-old chicken embryos stimulated by PBS, bFGF, or Del-1 were transfected with either N1-GFP reporter vector or Hox D3 sense (S), or Hox D3 antisense (AS) DNA expression constructs and N1-GFP. CAMs were fixed prior to excision, and blood vessel branch points (vessel branch pts) were quantified. (b) Expression levels of the chicken (chick) Hox D3, human Hox D3, and chicken GADPH genes in treated CAM tissues were determined by RT-PCR. (c and d) Relative levels of chick Hox D3 (c) and human Hox D3 (d). PCR products were determined by densitometry in comparison with GADPH expression levels for each treatment. (e) Expression levels of integrin β3 and β5, as well as GFP in CAMs stimulated with Del-1 or bFGF and treated with saline (PBS), Hox D3 antisense or sense were determined by Western blot analysis. Average expression levels for each integrin were determined in three separate experiments by densitometry. Asterisks indicate statistical significance relative to bFGF or Del-1 stimulation (P < 0.01).

Overexpression of Hox D3 promotes vascular proliferation and leads to hemangioma formation in CAMs (23). To determine whether αvβ5 function is required once Hox D3 is expressed, we transduced CAMs with Hox D3 and negative control retroviruses in the presence or absence of αvβ5 or αvβ3 antagonists. Hox D3 transduction induced a hemangioma-like angiogenic response in the CAM that was completely blocked by Ab’s directed against integrin αvβ3 (P < 0.01; Figure 9, a and b). In contrast, anti-αvβ5 Ab’s had no effect on the Hox D3–induced angiogenesis (Figure 9, a and b). Control viruses did not induce hemangiomas in the CAM (Figure 9, a and b). Taken together, these results demonstrate that Del-1–αvβ5 interactions regulate Hox D3 expression and that Hox D3 then regulates αvβ3 expression and function during angiogenesis.

Figure 9

Hox D3 is downstream of αvβ5 but upstream of αvβ3. (a) Q4dh cells transfected with Hox D3 expressing control virus (neg cntl) retroviral constructs were cultured on replicate CAMs in the presence of saline (HOX D3 virus), anti-αvβ3 (Hox D3 + LM609), anti-αvβ5 (Hox D3 + P1F6), or control IgG (W6/32, not shown). CAMs were excised after fixation, and representative CAMs were photographed. Each treatment group contained ten to 12 replicate CAMs. (b) Blood vessel branch points on filter disks were quantified and graphed as mean ± SEM. Asterisks indicate statistical significance relative to Hox D3 virus stimulation (P < 0.01).

Taken together, these results demonstrate that Del-1 stimulates angiogenesis by binding to quiescent EC integrin αvβ5, thereby inducing Hox D3–regulated expression of integrin αvβ3. These studies delineate a novel angiogenesis pathway induced by the ECM protein Del-1.

The studies presented here indicate that the angiogenic protein, Del-1, regulates a complex pathway of angiogenesis that can result in beneficial angiogenesis in response to ischemia. Integrin αvβ5 on resting endothelium binds Del-1 and triggers a program leading to EC activation and expression of the Hox D3 homeobox gene and, subsequently, integrin αvβ3.

Our studies show that Del-1–αvβ5 interactions induce αvβ3 expression and promote the angiogenic phenotype in vitro and in vivo. Basement membrane cultures were used in these studies to evaluate the role of Del-1 in the regulation of Hox D3 and αvβ3 expression in vitro. In vivo, quiescent ECs can be stimulated to undergo angiogenesis by various growth factors and by the angiomatrix protein Del-1. ECs begin to proliferate, upregulate cyclin D1, and express αvβ3. Upon completion of angiogenesis, endothelial tubes are formed, and expression of cyclin D1 and αvβ3 are downregulated. Thin layers of basement membrane (growth factor–reduced Matrigel) composed of collagen, laminin, and proteoglycans were used in this study to force ECs into a differentiated state in vitro, allowing us to examine the effect of Del-1 on nonproliferative, differentiated ECs that express low levels of integrin αvβ3. In vitro basement membrane cultures cause withdrawal from the cell cycle and suppression of αvβ3 (24, 25). Our studies indicate that Del-1 interaction with αvβ5 stimulates EC Hox D3 and αvβ3 expression in vitro and in vivo. Our in vivo studies indicate that these proteins are required for Del-1–mediated angiogenesis.

The studies presented here indicate that Del-1–induced angiogenesis depends upon integrins αvβ5 and αvβ3 because genetic deletion of either integrin, as well as antagonists of these integrins, inhibit Del-1–mediated angiogenesis. Mice lacking integrins αvβ5 and αvβ3 exhibit embryonic and adult angiogenesis, suggesting that these integrins are not required for some forms of embryonic and tumor-induced angiogenesis (31, 32). These studies are in apparent conflict with studies indicating that αvβ3 integrin ligation is required to provide survival signals to proliferating ECs during angiogenesis in vivo in normal animals (14, 15, 33, 34). Such studies have shown that when expressed, αvβ3 must be ligated or cell death is induced by a caspase 8–dependent process termed integrin-mediated death (33, 34). Mice lacking αvβ3 exhibit enhanced tumor angiogenesis (31), possibly because a key “death receptor” is absent (33, 34). Our studies indicate, however, that αvβ5 and αvβ3 are required for Del-1–mediated angiogenesis because Del-1–induced angiogenesis does not occur in the genetic absence of either integrin.

While Del-1 angiogenesis depends upon integrin αvβ3 function in genetically normal animals, it is not clear if Del-1 itself serves as a ligand for αvβ3 during angiogenesis. In vitro studies indicate that Del-1 is a ligand for αvβ3 and that it and αvβ3 are coordinately expressed in association with angiogenic blood vessels. It is possible that Del-1 may regulate angiogenesis by binding first to αvβ5 and then to αvβ3 or that Del-1 may also regulate the expression or generation by proteolysis of additional αvβ3 ligands.

Because Del-1 and a few other novel ECM proteins, but not all ECM proteins, induce angiogenesis, it is possible that these proteins also interact directly with growth factor receptors as well as integrins. Such interactions could account for the differential effect of the αvβ5 ligands Del-1 and vitronectin in angiogenesis models. While both matrix proteins mediate αvβ5-dependent adhesion and migration, only Del-1 induces EC proliferation and angiogenesis. Del-1 may exert EC-specific activities by binding to receptors found on ECs and not on other cells. For example, Del-1 does not promote HT29 colon carcinoma cell proliferation, nor does it upregulate integrin β3 or Hox D3 expression in these cells (data not shown), in contrast to its effects on ECs. Additionally, it is possible that Del-1 binding to EC integrins induces expression of proangiogenic growth factors. To understand how Del-1 regulates angiogenesis, it will be important to delineate the difference in signal transduction induced by this and other ECM proteins.

A proangiogenic ECM protein may play an important role during localized embryonic and pathological angiogenesis. The local deposition of Del-1 could allow locally controlled vascular proliferation. Once in the extracellular space, Del-1 may associate with other ECM proteins, providing a scaffold upon which ECs migrate during angiogenesis.

We have identified a novel mechanism that accounts for the induction of neovascularization by Del-1 and perhaps other proangiogenic ECM proteins. Del-1 ligation of integrin αvβ5 induces Hox D3 expression, thereby promoting integrin αvβ3 expression and function and the invasion and growth of new blood vessels. Del-1 thus regulates a program of angiogenesis characterized by expression and function of the transcription factor Hox D3. As a proangiogenic factor that promotes recovery from ischemic injury, Del-1 may be a useful tool for the management of ischemic diseases.

This work was supported by grants from the National Cancer Institute (CA71619 and CA83133) to J.A. Varner and by a grant from the National Heart Lung and Blood Institute to T. Quertermous (NHLB52168). We would like to thank Richard Hynes for the β3–null mice; Dean Sheppard for the generous gift of integrin β5^+/– and β5^–/– mice; and David Cheresh, Mark Ginsberg, and Glen Nemerow for gifts of Ab’s and purified integrins. We would like to thank Eric Rabinovsky for assistance with the rabbit model of the hind limb. We would also like to thank Yuhong Zhu, Hong Yin, Gaylen Howell, Elizabeth Wilson, Mary Thiesse, Jijun Wang, Isaac Van Sligtenhorst, and Wendy Nelson for excellent technical assistance.

Conflict of interest: Mike Coleman is an employee of Valentis Inc. Thomas Quertermous is the inventor of Del-1, which is licensed to Valentis Inc. and for which he receives licensing fees.

Nonstandard abbreviations used: endothelial cell (EC); arginine-glycine-aspartic acid (RGD); human VEGF₁₆₅ (hVEGF₁₆₅); human Del-1 (hDel-1); chorioallantoic membrane (CAM); avian retrovirus expressing human Hox D3 (CK-D3); CK virus without the insert (CK); green fluorescent protein (GFP); human umbilical vein EC (HUVEC); human microvascular EC (HMVEC).

1. Varner, J. 1997. The role of vascular cell integrins αvβ3 and αvβ5 in angiogenesis. In Regulation of Angiogenesis. I. Goldberg and E.M. Rosen, editors. Birkhauser Verlag. Basel, Switzerland. 361–390.
2. Carmeliet, P, Jain, RK. Angiogenesis in cancer and disease. Nature. 2000. 407:249-257.
   View this article via: PubMed CrossRef Google Scholar
3. Eliceiri, BP, Cheresh, DA. The role of αv integrins during angiogenesis: insights into potential mechanisms of action and clinical development. J. Clin. Invest. 1999. 103:1227-1230.
   View this article via: JCI PubMed CrossRef Google Scholar
4. Yancopoulos, GD, et al. Vascular-specific growth factors and blood vessel formation. Nature. 2000. 407:242-248.
   View this article via: PubMed CrossRef Google Scholar
5. Baumgartner, I, et al. Constitutive expression of VEGF 165 following intramuscular gene transfer promotes collateral vessel development in patients with critical limb ischemia. Circulation. 1998. 97:1114-1123.
   View this article via: PubMed Google Scholar
6. Banai, S, et al. Angiogenic-induced enhancement of collateral blood flow to ischemic myocardium by vascular endothelial growth factor in dogs. Circulation. 1994. 89:2183-2189.
   View this article via: PubMed Google Scholar
7. Asahara, T, et al. VEGF contributes to postnatal neovascularization by mobilizing bone marrow derived endothelial progenitor cells. EMBO J. 1999. 18:3964-3972.
   View this article via: PubMed CrossRef Google Scholar
8. Babic, AM, Kireeva, ML, Kolesnikova, TV, Lau, LF. Cyr 61, a product of growth-factor inducible immediate early gene, promotes angiogenesis and tumor growth. Proc. Natl. Acad. Sci. U. S. A. 1998. 95:6355-6360.
   View this article via: PubMed CrossRef Google Scholar
9. Babic, AM, Chen, C-C, Lau, LF. Fisp12/mouse connective tissue growth factor mediates endothelial cell adhesion and migration through integrin αvβ3, promotes endothelial cell survival and induces angiogenesis in vivo. Mol. Cell. Biol. 1999. 19:2958-2966.
   View this article via: PubMed Google Scholar
10. Kireeva, ML, Lam, SC-T, Lau, LF. Adhesion of human umbilical vein endothelial cells to the immediate early gene product Cyr61 is mediated through integrin αvβ3. J. Biol. Chem. 1998. 273:3090-3096.
    View this article via: PubMed CrossRef Google Scholar
11. Penta, K, et al. Del-1 induces integrin signaling and angiogenesis by ligation of αvβ3. J. Biol. Chem. 1999. 274:11101-11109.
    View this article via: PubMed CrossRef Google Scholar
12. Aoka, Y, et al. The embryonic angiogenic factor Del 1 accelerates tumor growth by enhancing vascular formation. Microvasc. Res. 2002. 64:148-161.
    View this article via: PubMed CrossRef Google Scholar
13. Hidai, C, et al. Cloning and characterization of developmental endothelial locus-1: an embryonic endothelial cell protein that binds the αvβ3 integrin receptor. Genes Dev. 1998. 12:21-33.
    View this article via: PubMed CrossRef Google Scholar
14. Brooks, PC, et al. Integrin αvβ3 antagonists promote tumor regression by inducing apoptosis of angiogenic blood vessels. Cell. 1994. 7:1157-1164.
15. Eliceiri, BP, Klemke, R, Stromblad, S, Cheresh, DA. Integrin αvβ3 requirement for sustained mitogen-activated protein kinase activity during angiogenesis. J. Cell Biol. 1998. 140:1255-1263.
    View this article via: PubMed CrossRef Google Scholar
16. Huang, X, Griffiths, M, Wu, J, Farese (Jr), RV, Sheppard, D. Normal development, wound healing, and adenovirus susceptibility in beta5-deficient mice. Mol. Cell. Biol. 2000. 20:755-759.
    View this article via: PubMed CrossRef Google Scholar
17. Eliceiri, BP, et al. Selective requirement for Src kinases during VEGF-induced angiogenesis and vascular permeability. Mol. Cell. 1999. 4:915-924.
    View this article via: PubMed CrossRef Google Scholar
18. Wang, J, Wilson, E, Van Sligtenhorst, I, Fix, M, Coleman, M. In vivo gene expression in muscle using a polymer delivery system. Mol. Ther. 2001. 3:S65. (Abstr.)
19. Hartikka, J, et al. Electroporation-facilitated delivery of plasmid DNA in skeletal muscle: plasmid dependence of muscle damage and effect of poloxamer 188. Mol. Ther. 2001. 4:407-415.
    View this article via: PubMed CrossRef Google Scholar
20. Takeshita, S, et al. Therapeutic angiogenesis. A single intraarterial bolus of vascular endothelial growth factor augments revascularization in a rabbit ischemic hind limb model. J. Clin. Invest. 1994. 93:662-670.
    View this article via: JCI PubMed CrossRef Google Scholar
21. Yatohgo, T, Izumi, M, Kashiwagi, H, Hayashi, M. Novel purification of vitronectin from human plasma by heparin affinity chromatography. Cell Struct. Funct. 1988. 13:281-292.
    View this article via: PubMed CrossRef Google Scholar
22. Kim, S, Harris, M, Varner, J. Regulation of angiogenesis in vivo by ligation of integrin α5β1 with the central cell binding domain of fibronectin. Am. J. Pathol. 2000. 156:1345-1362.
    View this article via: PubMed Google Scholar
23. Orlando, RA, Cheresh, DA. Arginine-glycine-aspartic acid binding leading to molecular stabilization between integrin αvβ3 and its ligand. J. Biol. Chem. 1991. 266:19543-19550.
    View this article via: PubMed Google Scholar
24. Boudreau, N, Andrews, C, Srebow, A, Ravanpay, A, Cheresh, DA. Induction of the angiogenic phenotype by Hox D3. J. Cell Biol. 1997. 139:257-264.
    View this article via: PubMed CrossRef Google Scholar
25. Myers, C, Charbonneau, A, Boudreau, N. Homeobox B3 promotes capillary morphogenesis and angiogenesis. J. Cell Biol. 2000. 148:343-351.
    View this article via: PubMed CrossRef Google Scholar
26. Kubota, Y, Kleinman, HK, Martin, GR, Lawley, TJ. Role of laminin and basement membrane in the morphological differentiation of human endothelial cells into capillary-like structures. J. Cell Biol. 1988. 107:1589-1598.
    View this article via: PubMed CrossRef Google Scholar
27. Boudreau, N, Werb, Z, Bissell, MJ. Suppression of apoptosis by basement membrane requires three dimensional tissue organization and withdrawal from the cell cycle. Proc. Natl. Acad. Sci. U. S. A. 1996. 93:3509-3513.
    View this article via: PubMed CrossRef Google Scholar
28. Donovan, D, Brown, NJ, Bishop, ET, Lewis, CE. Comparison of three in vitro human angiogenesis assays with capillaries formed in vivo. Angiogenesis. 2001. 4:113-121.
    View this article via: PubMed CrossRef Google Scholar
29. Feng, X, Clark, RAF, Galanakis, D, Tonnesen, MC. Fibrin and collagen differentially regulate human dermal microvascular cell integrins: stabilization of αvβ3 mRNA. J. Invest. Derm. 1999. 113:913-919.
    View this article via: PubMed CrossRef Google Scholar
30. Kim, S, Bakre, M, Yin, H, Varner, J. Inhibition of endothelial cell survival and angiogenesis by protein kinase A. J. Clin. Invest. 2002. 110:933-941. doi:10.1172/JCI200214268.
    View this article via: JCI PubMed Google Scholar
31. Hodivala-Dilke, KM, et al. β3-integrin-deficient mice are a model for Glanzmann thrombasthenia showing placental defects and reduced survival. J. Clin. Invest. 1999. 103:229-238.
    View this article via: JCI PubMed CrossRef Google Scholar
32. Reynolds, LE, et al. Enhanced pathological angiogenesis in mice lacking β3 and β5 integrins. Nature Med. 2002. 8:229-238.
33. Stupack, DG, Puente, XS, Butsaboualoy, S, Storgard, CM, Cheresh, DA. Apoptosis of adherent cells by recruitment of caspase-8 to unligated integrins. J. Cell Biol. 2001. 155:459-470.
    View this article via: PubMed CrossRef Google Scholar
34. Cheresh, DA, Stupack, DG. Integrin-mediated death: an explanation of the integrin-knockout phenotype? Nature Med. 2002. 8:1-2.
    View this article via: PubMed CrossRef Google Scholar