Jump to
Abstract
Optimal angiogenic and lymphangiogenic gene therapy requires knowledge of the best growth factors for each purpose. We studied the therapeutic potential of human vascular endothelial growth factor (VEGF) family members VEGF-A, VEGF-B, VEGF-C, and VEGF-D as well as a VEGFR-3–specific mutant (VEGF-C156S) using adenoviral gene transfer in rabbit hindlimb skeletal muscle. The significance of proteolytic processing of VEGF-D was explored using adenoviruses encoding either full-length or mature (ΔNΔC) VEGF-D. Adenoviruses expressing potent VEGFR-2 ligands, VEGF-A and VEGF-DΔNΔC, induced the strongest angiogenesis and vascular permeability effects as assessed by capillary vessel and perfusion measurements, modified Miles assay, and MRI. The most significant feature of angiogenesis induced by both VEGF-A and VEGF-DΔNΔC was a remarkable enlargement of microvessels with efficient recruitment of pericytes suggesting formation of arterioles or venules. VEGF-A also moderately increased capillary density and created glomeruloid bodies, clusters of tortuous vessels, whereas VEGF-DΔNΔC–induced angiogenesis was more diffuse. Vascular smooth muscle cell proliferation occurred in regions with increased plasma protein extravasation, indicating that arteriogenesis may be promoted by VEGF-A and VEGF-DΔNΔC. Full-length VEGF-C and VEGF-D induced predominantly and the selective VEGFR-3 ligand VEGF-C156S exclusively lymphangiogenesis. Unlike angiogenesis, lymphangiogenesis was not dependent on nitric oxide. The VEGFR-1 ligand VEGF-B did not promote either angiogenesis or lymphangiogenesis. Finally, we found a positive correlation between capillary size and vascular permeability. This study compares, for the first time, angiogenesis and lymphangiogenesis induced by gene transfer of different human VEGFs, and shows that VEGF-D is the most potent member when delivered via an adenoviral vector into skeletal muscle.
Angiogenesis (capillary growth), lymphangiogenesis (lymphatic vessel growth), arteriogenesis (enlargement of arteries), and vasculogenesis (in situ formation of blood vessels from vascular stem cells) are crucial for normal embryonic development, growth, and tissue repair. Further understanding of these processes may also contribute to the treatment of many disorders, such as cancer, tissue ischemia, and lymphedema. Vascular endothelial growth factors (VEGFs) are involved in all types of vascular growth.1–6 Currently, the human VEGF family consists of 5 members: VEGF-A, -B, -C, -D, and placenta growth factor (PlGF),7–13 which differ in their ability to bind to VEGF receptors that are primarily expressed in endothelial cells (ECs): VEGFR-1 (Flt-1), VEGFR-2 (KDR/Flk-1), VEGFR-3 (Flt-4), and neuropilin-1. VEGF-A binds to VEGFR-1 and VEGFR-2 as well as to neuropilin-1, whereas PlGF and VEGF-B bind only to VEGFR-1 and neuropilin-1.1,14,15 VEGF-C and VEGF-D are synthesized and secreted as large precursor forms that are proteolytically processed into mature forms comprising the central VEGF homology domain.16,17 Unprocessed forms preferentially signal through VEGFR-3, whereas only the mature forms efficiently trigger VEGFR-2 signaling.16,17 Both VEGF-C and -D have been suggested to be mainly lymphangiogenesis factors; their angiogenic potential has been reported to be considerably weaker than that of VEGF-A.6,13,16 In addition to the naturally occurring forms, Joukov et al18 have generated a mutant factor (VEGF-C156S) that binds to VEGFR-3 but not to VEGFR-2.
VEGF receptors have distinct biological roles. VEGFR-2 is considered to mediate most of the angiogenic, survival, and vascular permeability effects of VEGFs and effects on endothelial progenitor cells (EPCs),1,3,19 whereas the role of VEGFR-1 is controversial. Most studies indicate that VEGFR-1 is mainly a decoy receptor existing also as a soluble form, and that it may downregulate VEGFR-2–mediated mitogenesis.1,19–21 However, others have reported that the VEGFR-1 ligand PlGF mobilizes EPCs, hematopoietic, and inflammatory cells, and that it may amplify VEGFR-2–mediated effects.22 There is extensive evidence that VEGFR-3–mediated signaling alone is sufficient for the growth and maintenance of lymphatic vasculature.2,23,24
Nitric oxide (NO) is crucial in VEGF-A–mediated angiogenesis and vascular permeability,25 but it is not known whether NO is necessary for lymphangiogenesis.
For optimal gene therapy, it is essential to determine which VEGF family member is most suitable for angiogenesis/lymphangiogenesis in a given tissue. To address this issue, we compared adenoviruses encoding human VEGF-A, VEGF-B, VEGF-C, VEGF-D, and VEGF-C156S in rabbit hindlimb skeletal muscle. The objective was to study the direct effects of these VEGFs on local blood and lymphatic vessel growth, blood perfusion, and vascular permeability in the injected muscles and thus establish a basis for the selection of most suitable growth factors for each purpose. In addition, the significance of proteolytic processing of VEGF-D in vivo was studied by using adenoviruses expressing both its full-length and mature (ΔNΔC) forms. We further explored the effects of these growth factors on vascular smooth muscle cells (SMCs) and pericytes, and evaluated the role of NO in vascular permeability, angiogenesis, and lymphangiogenesis induced by VEGF-D in vivo.
Materials and Methods
The superficial femoral artery of New Zealand White rabbits (n=66) was excised and two reentry arteries for collaterals near the knee joint were ligated in order to induce collateral artery growth in the hindlimb as described.4 The nonischemic thigh region lacking confounding endogenous upregulation of VEGF-A and VEGFR-226 was used for the analysis of the effects induced by VEGFs. Seven days after surgery, animals received intramuscular (IM) injections of human clinical grade adenoviruses (total dose 1011 viral particles [vp]) encoding human VEGF-A165,8 VEGF-B167,10 full-length VEGF-C,11 full-length or mature (ΔNΔC) VEGF-D,13 or VEGF-C156S18 (n=3 to 8 in each group). Nω-Nitro-l-arginine methyl ester (L-NAME, NO synthase inhibitor)27 was used to study whether angiogenesis, lymphangiogenesis, and vascular permeability promoted either by the full-length or mature form of VEGF-D are dependent on NO production. Transduced hindlimbs were monitored for vascular permeability and subsequent edema with gadodiamide (GdDTPA-BMA)-contrast agent-enhanced T2*-weighted MRI (MRI)4,26 5 days after gene transfer (GT). Muscle perfusion and vascular permeability were assessed quantitatively with fluorescent microspheres and a modified Miles assay, respectively, 6 days after GT.4 Thereafter, muscle samples were collected for histological analyses.4,26 All animal experiments were approved by the Experimental Animal Committee, University of Kuopio.
For further details, see the expanded Materials and Methods in the online data supplement available at http://www.circresaha.org.
Results
VEGFR-2 Ligands VEGF-A and VEGF-DΔNΔC Induce the Strongest Angiogenesis Effects in Skeletal Muscle
Effects of VEGFs were evaluated in nonischemic muscles in areas outside the needle track in order to exclude the confounding effects of endogenous growth factors induced by injection trauma or ischemia. Control virus (1011 vp/mL) encoding the LacZ marker gene did not induce angiogenesis or significant inflammation in rabbit skeletal muscle (Figures 1a and 1c), whereas an angiogenesis response characterized by a remarkable microvessel enlargement was induced by both AdVEGF-A and AdVEGF-DΔNΔC (Figures 1b, 1d, and 1i). In AdVEGF-DΔNΔC transduced muscles, the size of enlarged vessels sometimes exceeded that of surrounding skeletal myocytes. Adenoviruses encoding full-length VEGF-C and VEGF-D also enlarged the microvessels, although less than AdVEGF-A and AdVEGF-DΔNΔC (Figures 1f and 1h), whereas AdVEGF-B and AdVEGF-C156S had no effects on blood vessels (Figures 1e and 1g). VEGFR-2 and αvβ3 integrin were not detectable with immunostaining in capillaries of intact and AdLacZ transduced muscles (data not shown). In contrast, both VEGF-A (data not shown) and VEGF-DΔNΔC upregulated VEGFR-2 and αvβ3 integrin expression in the ECs (Figures 1j and 1k).
Figure 1. AdVEGF-A and AdVEGF-DΔNΔC induce strong capillary enlargement in skeletal muscle. a through i, CD31 immunostaining of semimembranosus muscles transduced with adenoviruses encoding the indicated proteins 6 days after GT. Asterisks indicate the needle tracks. VEGFR-2 (j) and αvβ3 integrin (k) are upregulated on ECs after AdVEGF-DΔNΔC GT as shown by specific immunostainings. Scale bar=100 μm in a and b, 50 μm in c through i, and 25 μm in j and k.
Interestingly, there were differences in the angiogenesis patterns promoted by AdVEGF-A and AdVEGF-DΔNΔC. The strongest effects after AdVEGF-A treatment occurred in connective tissue between muscle bundles and in muscle fascias, whereas angiogenesis stimulated by AdVEGF-DΔNΔC was more diffuse (Figures 2b, 2c, 2f, and 2g). Furthermore, AdVEGF-A generated more glomeruloid bodies (clusters of tortuous vessels28) than AdVEGF-DΔNΔC (Figure 2b).
Figure 2. VEGFs elicit powerful effects on blood and lymphatic ECs, pericytes, and SMCs. a through l, CD31 (ECs, blue)+αSMA (pericytes and SMCs, brown) double immunostaining of transduced semimembranosus muscles 6 days after GT. V indicates vein; A, artery; and L, lymphatic vessel (distinguished from blood vessels by the lack of αSMA-positive perivascular cells). a through d, Low-power overviews (scale bar=100 μm; 25 μm in the insert of b). a, AdLacZ. b, After AdVEGF-A, angiogenesis response localizes, especially to the surfaces of muscle bundles (arrowheads indicate glomeruloid bodies, shown in the insert) and muscle fascia (asterisk) and involves growth of α-SMA–positive vessels. Lymphatic vessels do not respond to VEGF-A (arrow). c, AdVEGF-DΔNΔC–induced angiogenesis is more diffuse than that induced by AdVEGF-A and consists mainly of an enlargement of α-SMA–positive microvessels. d, In AdVEGF-D-treated muscle, lymphatic vessels (arrows) have sprouted between muscle bundles. e through j, Higher magnification images show distinct features of the responses to different VEGFs (scale bar=100 μm). e, AdLacZ. f, AdVEGF-A induces strong angiogenesis especially in connective tissue but not lymphangiogenesis. g, After AdVEGF-DΔNΔC, capillaries are enlarged (arrowhead) but also lymphatic vessels show additional sprouts (arrow) from bigger lymphatics (L). h, Intense lymphatic sprouting in AdVEGF-D (arrow). i, Mainly lymphangiogenesis (arrow) but also weak capillary enlargement (arrowhead) in AdVEGF-C. j, AdVEGF-Cl56S induces selectively lymphangiogenesis (arrow). k through p, Scale bar=25 μm. k, Normal-sized capillaries in AdLacZ control. Arrowheads indicate α-SMA–positive pericytes. l, Enlarged microvessels resemble arterioles or venules more than capillaries after AdVEGF-DΔNΔC. Most vessels have numerous α-SMA–positive pericytes (arrowheads) but occasionally vessels lack a complete coverage (arrow). m through p, CD31 (blue)+BrdU (brown) double immunostainings 6 days after GT. m, AdLacZ. n, Both ECs (arrowheads) and pericytes (arrows) proliferate after AdVEGF-DΔNΔC. o, One dividing cell in the wall of a vein (arrow) but not in the artery. p, Abundant number of proliferating SMCs in a collateral artery after AdVEGF-DΔNΔC.
VEGFR-3 Ligands VEGF-C, VEGF-C156S, and VEGF-D Promote Lymphangiogenesis
Histochemical staining of lymphatic endothelial cells for 5′ nucleosidase activity29 and an intraarterial injection of Ricinus Communis lectin (data not shown) revealed that the CD31-positive but α-SMA–negative vessels in the interstitial connective tissue between the muscle bundles were lymphatics, not blood vessels (Figures 2d through 2j).
Adenovirus encoding the full-length VEGF-D induced the strongest lymphatic vessel growth as almost all the interstitial connective tissue in transduced muscles was filled with lymphatic vasculature 6 days after GT (Figures 2d and 2h). AdVEGF-DΔNΔC, AdVEGF-C, and AdVEGF-C156S were also potent lymphangiogenesis inducers (Figures 2g, 2i, and 2j). Remarkably, the adenovirus encoding the VEGFR-3 selective mutant VEGF-C156S stimulated exclusively lymphangiogenesis, whereas AdVEGF-A (Figures 2f and 2j) and AdVEGF-B (data not shown) had no effects on lymphatic vessels.
VEGF-A and VEGF-DΔNΔC Induce Arteriogenesis
In AdLacZ muscles, only a portion of the capillaries had α-SMA–positive pericytes and there were few proliferating SMCs (Figures 2k and 2o). In contrast, both in AdVEGF-A– and AdVEGF-DΔNΔC–transduced muscles, nearly all enlarged capillaries had a complete or almost complete α-SMA–positive pericyte coverage, indicating that α-SMA expression was upregulated (Figures 2b, 2c, and 2l). Furthermore, BrdU labeling showed a high proliferation rate of ECs and pericytes after AdVEGF-A (data not shown) and AdVEGF-DΔNΔC GT (Figure 2n). In these muscles, collateral arteries showed active remodeling with abundant numbers of proliferating SMCs (Figure 2p), especially in regions of plasma protein extravasation (see following sections).
AdVEGF-A and AdVEGF-DΔNΔC Increase Vascular Permeability and Cause Edema
Extensive vascular permeability and subsequent edema were observed in AdVEGF-A and AdVEGF-DΔNΔC transduced muscles 5 days after GT by contrast agent–enhanced MRI (Figures 3b and 3g). Extravasated contrast agent was detected under the skin, in the semimembranosus muscle and in its fascias, and in the fat tissue between the medial and lateral muscle compartments. In contrast, AdLacZ or other AdVEGFs did not induce significant vascular permeability effects (Figures 3a, 3c through 3f, 4a, 4c, and 4⇓e). The modified Miles assay demonstrated strong and quite diffuse vascular permeability in AdVEGF-DΔNΔC transduced muscles (Figure 4b). Microscopically, extravasated plasma proteins were retained, especially in connective tissue between muscle bundles, around large blood vessels, as well as within muscle fibers surrounded by enlarged, leaky microvessels (Figures 4d and 4f).
Figure 3. GdDTPA-BMA enhanced T2*-weighted transversal MR images of rabbit mid-thighs demonstrate increased vascular permeability in response to different VEGFs 5 days after GT (transduced limb on the left, intact on the right). No vascular leakage or edema but only the scar of hindlimb operation (arrow) is visible as bright GdDTPA-BMA contrast in (a) AdLacZ control. b, Vascular leakage and edema (arrowhead) in AdVEGF-A transduced muscle. c through f, No significant edema after GT with the indicated viruses. g, Extensive edema after AdVEGF-DΔNΔC GT (arrowhead). h, AdVEGF-DΔNΔC+NO synthase inhibitor (L-NAME).
Figure 4. Extravasated plasma proteins in AdVEGF-DΔNΔC transduced semimembranosus muscles 6 days after GT as illustrated by the modified Miles assay. a, AdLacZ control muscle. b, Blue color demonstrates extravasated proteins in AdVEGF-DΔNΔC–treated muscle. c through f, Protein-bound Evans blue detected in histological sections as red fluorescence. Blue nuclear staining with DAPI. c, No extravasated plasma proteins around a small artery (arrowhead) in AdLacZ control. d, In AdVEGF-DΔNΔC muscle, plasma proteins are detected around and in the wall of an artery (arrowhead), but not in areas where capillaries are not enlarged (right side of the image). e, No capillary vessel leakiness in the AdLacZ control. f, In regions of active angiogenesis and capillary enlargement (arrowheads), muscle contains abundant amounts of extravasated plasma proteins. Bar=1cm in a and b; 100 μm in c through f.
Angiogenesis and Lymphangiogenesis Profiles of Adenovirally Delivered VEGFs
A quantitative comparison of capillary density and mean area, total capillary area, plasma protein extravasation, regional muscle perfusion, and total lymphatic vessel area in the transduced muscles 6 days after GT gave an angiogenesis or lymphangiogenesis profile for each VEGF. Surprisingly, only AdVEGF-A GT significantly increased the capillary density (Figure 5a). In contrast, AdVEGF-DΔNΔC enlarged the mean capillary area as much as 14-fold as compared with AdLacZ control without inducing any increase in the capillary number (Figures 5a and 5b). The corresponding increase in the mean capillary size was 6.4-fold with AdVEGF-A. Effects on microvessel enlargement reached a statistical significance also with AdVEGF-C and AdVEGF-D with 2.5- and 2.2-fold increases, respectively (Figure 5b).
Figure 5. Quantitative effects of different AdVEGFs on angiogenesis, vascular permeability, muscle perfusion, and lymphangiogenesis. All measurements were done from transduced semimembranosus muscles 6 days after GT. a, Capillary density (capillaries/mm2). Only AdVEGF-A GT resulted in a significant increase in the capillary number as compared with AdLacZ control. b, Mean capillary area (μm2, black bars) and vascular permeability ratio between transduced and intact contralateral muscle (gray bars). Note that AdVEGF-A and AdVEGF-DΔNΔC strongly increased the mean capillary area and vascular permeability ratio over AdLacZ. Also AdVEGF-C and AdVEGF-D induced some capillary enlargement but not significant plasma protein extravasation. c, Total area of muscle covered by capillary lumens (percent, black bars) and regional perfusion ratio between transduced and intact contralateral muscle (gray bars). d, Total area of muscle covered by lymphatic vasculature. *P<0.05, **P<0.01, and ***P<0.001 vs AdLacZ.
AdVEGF-DΔNΔC and AdVEGF-A GT resulted in dramatic 24- and 13-fold increases, respectively, in plasma protein extravasation as measured by the modified Miles assay (Figure 5b). Although AdVEGF-C and AdVEGF-D had weak effects on capillary enlargement, they did not significantly enhance plasma protein extravasation.
About 6% and 8% of the AdVEGF-A– and AdVEGF-DΔNΔC–transduced muscles were covered by microvessel lumens, respectively (Figure 5c). These figures were much greater than in AdVEGF-C– (1.8%), AdVEGF-D– (1.7%), and AdLacZ-treated muscles (0.7%). Regional perfusion was increased accordingly in the transduced muscles. AdVEGF-A, AdVEGF-DΔNΔC, and AdVEGF-C induced statistically significant 4.0-, 3.2-, and 2.0-fold increases in perfusion, respectively (Figure 5c). As shown in Figure 5d, lymphatic vessel area of muscles (%) was strongly increased with adenoviruses expressing the VEGFR-3 ligands VEGF-C (1.5%), VEGF-C156S (1.3%), VEGF-D (2.6%), and VEGF-DΔNΔC (1.4%) as compared with AdLacZ (0.12%).
Finally, angiogenesis and lymphangiogenesis indices were calculated for each VEGF. These indices illustrate the balance between blood and lymphatic vessel growth (Figure 6). VEGF-A was the only to induce angiogenesis but not lymphangiogenesis, whereas full-length VEGF-C and VEGF-D were found to be mainly, and VEGF-C156S exclusively, lymphangiogenic. The mature form VEGF-DΔNΔC induced significant growth of both vessel types. VEGF-B was inefficient both for angiogenesis and lymphangiogenesis.
Figure 6. Summary of angiogenesis and lymphangiogenesis profiles of human VEGFs. Fold inductions in total capillary area (angiogenesis) and total lymphatic vessel area (lymphangiogenesis) of AdVEGF-treated muscles as compared with AdLacZ 6 days after adenoviral GT.
Angiogenesis but not Lymphangiogenesis Is Dependent on NO Production
NO synthase inhibitor L-NAME significantly blocked capillary enlargement and increases in vascular permeability after AdVEGF-DΔNΔC treatment as shown in Figures 3h and 7⇓a. In contrast, NO synthase inhibition did not affect lymphangiogenesis induced by full-length VEGF-D (Figure 7b).
Figure 7. Angiogenesis, but not lymphangiogenesis, is dependent on NO production. Plasma protein extravasation correlates with capillary size in transduced muscles. a, NO synthase inhibitor (L-NAME) significantly blocked the angiogenesis (black bars) and vascular permeability (gray bars) effects induced by AdVEGF-DΔNΔC. *P<0.05 vs AdVEGF-DΔNΔC treatment alone. b, Lymphangiogenesis induced by AdVEGF-D is not affected by L-NAME. c, Vascular permeability ratio in muscles transduced with different AdVEGFs or AdLacZ plotted against the mean capillary area of the same muscle shows a positive correlation (r=0.81, P<0.01). d, Positive correlation also exists between regional perfusion and the total capillary area (r=0.61, P<0.01).
Vascular Permeability Correlates With Microvessel Size
A positive correlation was found when the mean microvessel area calculated from each transduced muscle was plotted against the respective vascular permeability ratio (Figure 7c). A positive correlation also existed between the total capillary area and the regional muscle perfusion, indicating that the anatomical observation about capillary vessel growth is in line with a physiological measure of blood flow (Figure 7d). However, the correlation was best with physiological (small) capillary sizes because microspheres (diameter 15 μm) do not likely get stuck in strongly enlarged capillaries (diameter >15 μm), which may underestimate perfusion values obtained with this method.
Discussion
We analyzed the human VEGF family members VEGF-A, VEGF-B, VEGF-C, VEGF-D, VEGF-DΔNΔC, and a VEGFR-3–specific mutant VEGF-C156S for their effects on angiogenesis, lymphangiogenesis, and vascular permeability in a rabbit hindlimb model using adenoviral GT. We found striking inductions in blood vessel growth and vascular permeability with adenoviruses encoding VEGF-A and VEGF-DΔNΔC. Intriguingly, AdVEGF-A and AdVEGF-DΔNΔC induced not only EC, but also pericyte and SMC proliferation, and 3- to 4-fold increases in muscle perfusion at rest. A positive correlation was observed between capillary size and vascular permeability. Adenoviruses expressing the full-length VEGF-C and VEGF-D induced mainly, and the VEGFR-3–specific mutant VEGF-C156S exclusively, lymphangiogenesis. AdVEGF-D generated the most powerful lymphatic vessel growth, which, unlike angiogenesis, was not affected by NO synthase inhibition. Finally, we found that the VEGFR-1 ligand VEGF-B was ineffective both for angiogenesis and lymphangiogenesis.
We performed GT of different VEGFs in the nonischemic thigh muscles in this rabbit model of collateral artery growth4 because necrosis, inflammation, and expression of endogenous growth factors caused by ischemia26 may mask differences between transduced growth factors and make this kind of comparison impossible. Furthermore, therapies designed for augmentation of arteriogenesis should also be effective in nonischemic muscles upstream to ischemia, which is usually the site of collateral growth.
Whereas this in vivo study supports the role for VEGFR-2 signaling as the major regulator of angiogenesis and vascular permeability, our data suggest that VEGFR-1 signaling alone may not be capable of initiating angiogenesis because its ligand VEGF-B was ineffective. Nevertheless, distinct histological characteristics were associated with AdVEGF-A–induced angiogenesis in comparison with that obtained with AdVEGF-DΔNΔC. AdVEGF-A increased capillary density and induced the formation of glomeruloid bodies28 more frequently than AdVEGF-DΔNΔC, which exerted more diffuse effects and strongly enlarged the preexisting microvessels. These findings may relate to the fact that VEGF-A binds to VEGFR-1 and VEGFR-2,1 whereas VEGF-DΔNΔC binds to VEGFR-2 and VEGFR-3.13,17 However, it is more likely that the different affinities of VEGF-A and VEGF-DΔNΔC for the extracellular matrix are responsible for these biological differences. In contrast to VEGF-A165, there is no heparan-binding domain in the sequence of VEGF-DΔNΔC.1,13 Furthermore, the selective VEGFR-1 and VEGFR-3 ligands, VEGF-B and VEGF-C156S, respectively, appeared inert toward blood vessels. Nevertheless, the possibility of subtle interspecies differences on receptor activation, which could modify the biological outcomes, cannot be excluded because human VEGFs were tested in the rabbit. However, according to our previous studies and this work, human and mouse VEGF-A and human VEGF-C have all shown expected potency in rabbits.4,30 Further studies are needed to clarify the role of VEGFR-1 in angiogenesis because PlGF, another VEGFR-1 ligand, is angiogenic at least in some circumstances.22
As expected from in vitro receptor binding profiles11,13,17,23 and recent in vivo work,2 we found that VEGFs activating VEGFR-3 induced lymphangiogenesis in rabbit skeletal muscle. The counting of lymphatic vessels was done from CD31 and α-SMA double-immunostained sections because of the possibility that molecular markers used for the detection of lymphatic endothelium, such as VEGFR-3 and LYVE-1, may also be expressed on activated ECs of blood vessels or may not be expressed in all lymphatic ECs.6,23,31,32 VEGF-C stimulated predominantly, and the VEGFR-3–specific mutant VEGF-C156S exclusively, lymphatic vessel growth. However, lymphangiogenesis induced by the full-length VEGF-D was strongest as nearly all the space between muscle bundles was filled with lymphatics in AdVEGF-D–treated muscles. The ability of the processed form VEGF-DΔNΔC to induce substantially more angiogenesis and less lymphangiogenesis than the full-length form can be explained by the fact that proteolytic processing increases its affinity toward VEGFR-2 more (290-fold) than toward VEGFR-3 (40-fold).17 Our data also suggest that, at least in nonischemic skeletal muscle, VEGF-D is not efficiently cleaved by proteases. Taken together, our results implicate that VEGF-C, VEGF-D, and VEGF-C156S could be applicable in the treatment of lymphatic disorders such as primary and secondary lymphedema.24
Blood vessels formed in response to VEGF-A have been suggested to lack pericytes,33 and it has been proposed that because of this defect they are prone to regression when VEGF-A levels are decreased.34 However, our data demonstrate that the expanded vessels induced both by AdVEGF-A and AdVEGF-DΔNΔC are accompanied by an α-SMA–positive pericyte coverage, which, together with the large size (diameter up to 50 μm), suggests a shift from the midcapillary phenotype toward arterioles, venules, or possibly arteriovenous shunts. Nevertheless, in spite of the efficient pericyte recruitment, our recent study shows that at least in nonischemic skeletal muscle, where excess blood perfusion is not necessary, the enlarged capillaries induced by AdVEGF-A or AdFGF-4 regress after the termination of gene expression.4 To the best of our knowledge, it has not been shown that VEGFs could induce SMC or pericyte proliferation in vivo and thus these actions are likely indirect. Arterial SMCs and pericyte proliferation occurred in areas of increased plasma protein extravasation, which suggests that protein efflux from the vasculature to the extravascular space may contribute to these processes.4 For example, extravasation of plasma proteins triggers the clotting cascade, leading to deposition of fibrin gel in the interstitial space. In addition to causing edema, the hydrated fibrin gel is proangiogenic because it provides matrix for cell migration and growth.35 Additional growth factors that are mitogenic for SMCs, such as PDGFs, may also be upregulated. Furthermore, increased shear stress in the enlarged microvessels and enhanced integrin signaling36 may further contribute to the transformation of capillaries toward bigger vessels.
The NO synthase inhibitor L-NAME27 significantly blocked AdVEGF-DΔNΔC–induced vascular effects including capillary enlargement and extravasation of plasma proteins, indicating that, like in the case of VEGF-A,19,25 NO is crucial for blood vascular effects stimulated by VEGF-DΔNΔC. Furthermore, VEGFR-2 and αvβ3 integrin were upregulated on ECs by VEGF-DΔNΔC, suggesting that its angiogenic signaling mechanisms are similar to those of VEGF-A.1 Furthermore, efficient angiogenesis and VEGFR-2 upregulation by its ligands in nonischemic skeletal muscle shows that ischemia is not requisite for blood vessel growth. In contrast to angiogenesis, inhibition of NOS did not affect lymphangiogenesis. To the best of our knowledge, this is the first demonstration that NO synthases are not crucial for lymphangiogenesis. Perivascular cells may be important for such a specific requirement of NO in angiogenesis as blood vascular, but not small lymphatic vessels, have a pericyte coverage.37
We found an interesting correlation between mean capillary size and plasma protein extravasation. VEGFR-2 ligands induce plasma protein extravasation probably by multiple mechanisms. For example, they increase EC surface area, capillary blood flow and pressure, and also have direct effects on EC ultrastructure and vesicle transportation.38 As shown in this study, cell proliferation is involved in capillary enlargement. Thus, in addition to hemodynamic changes in enlarged microvessels, an increased number of intercellular clefts between dividing ECs, and possibly decreased integrity of the basement membrane may be of importance in vessel leakiness occurring 5 to 6 days after AdVEGF GT. In contrast to this kind of vascular permeability related to angiogenesis, ultrastructural changes and vesicle transport in ECs may be more important in acute plasma protein leakage induced by VEGFs.4 Ang-1 has been reported to improve the EC barrier function to plasma proteins.39 However, our data suggest that edema after VEGF GT could be reduced by hindering the excess enlargement of developing vessels. On the other hand, this may not be possible without the restriction of EC and SMC proliferation, which would likely compromise the success of therapeutic angiogenesis. Furthermore, as discussed above, increased plasma protein extravasation may be essential for efficient angiogenesis35 and arteriogenesis.
Our findings have several important implications. Firstly, they show that at least in nonischemic skeletal muscle the angiogenic effects of VEGF family members comprise microvessel enlargement, pericyte recruitment, and increases in vascular permeability, and not necessarily increases in capillary density. This relationship between angiogenesis and vascular permeability could be used for the judgment of successful gene transfer and for the evaluation of edema after clinical VEGF gene therapy with noninvasive imaging methods such as MRI or ultrasound. Secondly, unlike angiogenesis, lymphangiogenesis is not dependent on VEGFR-2 and NO. Most importantly, this unique comparison of the biological effects generated by different human VEGF family members in vivo provides a classification of human VEGFs as blood or lymphatic vessel growth factors depending on their receptor specificity. The VEGFR-1 ligand VEGF-B is not able to trigger either angiogenesis or lymphangiogenesis in rabbit skeletal muscle, whereas VEGFR-2 and VEGFR-3 ligands are strongly angiogenic and lymphangiogenic, respectively. In conclusion, VEGF-A, VEGF-C, VEGF-C156S, VEGF-D, and VEGF-DΔNΔC have potential as vascular therapeutics and the most suitable VEGF for each treatment can be chosen based on the need for angiogenesis and/or lymphangiogenesis.
Acknowledgments
This study was supported by grants from the Ludwig Institute for Cancer Research, the Sigrid Juselius Foundation, Academy of Finland, Novo Nordisk Foundation, and Finnish Medical Foundation. Marc Achen and Steven Stacker are supported by the National Health and Medical Research Council of Australia and the Anti-Cancer Council of Victoria. The α-SMA antibody was a generous gift from Dr Giulio Gabbiani (Department of Pathology, Centre Medical Universitaire, Geneva, Switzerland). Anne Martikainen, Mervi Nieminen, Janne Kokkonen, Marja Poikolainen, and Dr Martin Kavec are acknowledged for their expert technical help.
Footnotes
-
↵*Both authors contributed equally to this study.
- Received October 1, 2002.
- Revision received April 14, 2003.
- Accepted April 14, 2003.
References
This Issue
Jump to
Article Tools
Share this Article
- VEGF-D Is the Strongest Angiogenic and Lymphangiogenic Effector Among VEGFs Delivered Into Skeletal Muscle via Adenoviruses
