Author + information
- Received January 30, 2004
- Revision received April 27, 2004
- Accepted May 2, 2004
- Published online August 18, 2004.
- Yi Fu Zhou, MD* (, )
- Eugenio Stabile, MD,
- Jill Walker, MS,
- Matie Shou, MD,
- Richard Baffour, PhD,
- Zuxi Yu, MD, PhD,
- David Rott, MD,
- George D. Yancopoulos, MD,
- John S. Rudge, PhD and
- Stephen E. Epstein, MD
- ↵*Reprint requests and correspondence:
Dr. Yi Fu Zhou, Vascular Biology Laboratory, Cardiovascular Research Institute, GHRB-217, 108 Irving Street, Washington Hospital Center, Washington, DC 20010
Objectives The aim of this research was to test the effects of vascular endothelial growth factor (VEGF)/angiopoietin-1 (Ang-1) on adult hypoperfused tissues.
Background Angiopoietin-1 and VEGF act separately and synergistically in vascular development during embryogenesis. However, little is known regarding their relative roles in collateral development after chronic arterial obstruction and tissue ischemia in the adult.
Methods Central and caudal ear arteries of 32 rabbits were ligated to induce ischemia. At two months, when flow was about 65% of pre-ligation values, we injected intradermally 109plaque-forming unit adenovirus with the following transgenes: Ang-1, VEGF, or a combination of both. Ear perfusion was followed up for four weeks, and vessel leakage was assessed by Evens Blue test.
Results Before injection, flow was 65% of baseline, and endogenous VEGF levels in ischemic tissue were increased. Adenovirus-encoding VEGF gene (Ad.VEGF) at one week caused a visible inflammatory response associated with a 24% flow increase (p = 0.018). Adenovirus-encoding Ang-1 gene (Ad.Ang-1) increased flow 22% (p = 0.004) with no visible inflammation; Ad.VEGF caused three times as much vessel leakage as Ad.Ang-1 (142.5 ± 38 vs. 49.5 ± 9.8 ng Evens Blue/mg tissue; p < 0.001). However, at four weeks, compared with baseline, VEGF decreased flow 18% (p = 0.004), whereas Ang-1 increased tissue perfusion 26% (p < 0.001). This effect was abolished when Ad.Ang-1 was injected with soluble VEGF receptor [Ad.Flt(1-3)-Fc], which blocks VEGF-dependent signaling. Exogenous Ang-1 did not increase perfusion in a normally perfused ear, in which endogenous VEGF is not expressed.
Conclusions Exogenous Ang-1 enhances perfusion in hypoperfused tissues only in the presence of increased levels of endogenous VEGF. Overexpression of VEGF, however, after causing an inflammatory response, does not improve collateral blood flow.
Vessel development is a complex process still incompletely understood. Studies have shown that vascular endothelial growth factor (VEGF) and angiopoietin-1 (Ang-1) are expressed sequentially during embryogenesis, and each of these ligands controls specific, complementary functions that ultimately contribute to the development of mature functional blood vessels (1,2).
Although VEGF is essential for vascular growth during embryogenesis (3), its effects on collateral growth in adult tissues are conflicting (4–6). While it has been reported that exogenous administration of VEGF improves collateral development after acute ischemia in mice, in the same animal model, only transient induction of endogenous VEGF messenger ribonucleic acid transcription has been observed (7). Additionally, because VEGF induces marked increases in vascular permeability and tissue edema, excessive VEGF administration could result in increased interstitial pressure, increased edema, and, consequently, increased ischemia. Of note, two randomized clinical trials have demonstrated no improvement in collateral function after exogenous VEGF administration in patients with coronary artery disease or peripheral artery disease (8,9).
On the other hand, although Ang-1 is considered a major factor inducing vessel maturation during embryogenesis (10), little is known of the role of Ang-1 in collateral development in tissues subjected to chronicflow deprivation. To better assess the relative and simultaneous effects of VEGF and Ang-1 in collateral formation, we tested the effects of their exogenous administration, through gene transfer, in a rodent model of ischemia (11).
Animals and viruses
Male New Zealand white rabbits of average weight, 3.3 kg, were housed one per cage and allowed free access to water and commercial rabbit food. The current protocol was approved by the Institutional Animal Care and Use Committee of Medstar Research Institute, and the investigation conforms to the Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85-23, revised 1996). The recombinant adenovirus expressing human Ang-1*, mouse VEGF164, and mouse VEGF-soluble receptor [Flt(1-3)-Fc] were kindly supplied by Regeneron Pharmaceuticals Inc. (Tarrytown, New York). All viruses were propagated in 293 cells, purified and titered by standard method, then stored in −80°C until use. The increased vascular permeability 48 h after Ad.VEGF164 transfection confirmed the activity of mouse VEGF on rabbit tissues.
Rabbits were anesthetized with intramuscular ketamine (35 mg/kg) and xylazine (5 mg/kg). Both ears were shaved and sterilized by applying Betadine and then 75% alcohol. Local anesthesia was administered by injecting 2% lidocaine subcutaneously around the surgical areas at the base of the ears; 1-cm incisions were made in the skin above the central and caudal arteries. These two arteries were then carefully isolated from the surrounding tissue using forceps. Two sutures were tied around each artery approximately 2 mm apart. The arteries were then cut in between the sutures to create ear ischemia (11).
Assessment of tissue flow
Repeated ear blood flow measurements over the regions of interest were obtained at pre-ligation, immediately after, and serially over eight weeks using Laser Doppler Perfusion Imaging (Moor Instruments, Wilmington, Delaware) (12). Rabbits were anesthetized as above, and measurements were performed at constant room temperature (73°F to 74°F) to minimize temperature-dependent flow changes in the ear subcutaneous tissue. Palette 2 color-code setting with a range of 118 to 1,000 was used to scan and review images. Tissue perfusion index (expressed as flux index) was averaged from regions of interest. For each animal, baseline, pre-ligation blood flow was taken to be 100%. Post-ligation flux index was calculated as a percentage of the pre-ligation value.
Effects of Ang-1 vs. VEGF on tissue perfusion
Arteries were ligated as described above in both ears of 32 rabbits. Eight weeks later, animals were randomly divided into groups injected intradermally with an adenovirus (109plaque forming unit [PFU] total; total volume, 200 μl) containing: 1) the Ang-1 transgene (n = 8); 2) the VEGF transgene (n = 8); 3) both Ang-1 and VEGF (n = 8); or 4) the same adenovirus without a transgene insert (Ad.Null, n = 8). Flow was measured pre-injection, one, two, three, and four weeks after injection. All post-injection values were divided by pre-injection ones and were expressed as percentage of the pre-injection tissue flux index.
To test the effects of Ang-1 on tissue perfusion in the absence of endogenous VEGF, a separate set of experiments was performed: adenovirus-encoding Ang-1 gene (Ad.Ang-1) was injected simultaneously with an adenoviral vector containing the VEGF soluble receptor gene [Flt(1-3)-Fc], or with a control adenoviral vector (Ad.Null) into both ears of three animals with chronically impaired perfusion. Finally, the effects of Ad.Ang-1 and adenovirus-encoding VEGF gene (Ad.VEGF) on nonischemic ear tissues were also tested in separate groups of normal rabbits (three animals in each group).
Vascular permeability after injection of Ad.Ang-1 or Ad.VEGF was measured using Evens Blue in a separate group of rabbits (13). Evens Blue, 20 mg/kg (dissolved in phosphate-buffered saline-containing heparin [100 U/ml]), was injected into the femoral artery of anesthetized rabbits; 30 min after injection, four pieces of tissues were removed from each ear along the injected area, weighed, placed in 2 ml formamide, and incubated at 50°C for 24 h for Evens Blue extraction. Evens Blue concentrations in tissue extracts were measured at 630 nm against standard Evens Blue concentrations using a micro-plate reader. Concentrations were then normalized by nanogram per milligram ear tissue. Assays were performed on one rabbit from each group at 3, 7, and 14 days after adenovirus injection.
Tissue samples were collected seven days after injection and were cut into two parts: one part was fixed with 10% formalin and paraffin embedded, then sectioned for hematoxyline and eosin staining; the second part was fixed in 2.5% electron microscope grade glutaraldehyde followed by Osmium after fixation and embedded in Epon for electron microscopic analysis. Ultrathin sections were cut, stained with lead citrate and uranyl acetate, and examined by transmission electron microscopy (JEOL 1200 EX) (14). Some sections were subjected to Kajikawa staining to identify the details of the elastic fiber and collagen. Cross-sectional thickness of the ear, as observed under bright field microscopy, was used as an index of tissue edema.
Western immunoblot analysis
The VEGF and Ang-1 concentrations in ear tissues were assessed by Western immunoblots. Tissue samples were homogenized in ice-cold RIPA lysis buffer at room temperature for 2 min, and then centrifuged (14,000 rpm, 4°C) for 10 min (15). Tissue pellets were removed, and the protein-containing supernatant was collected and kept at −20°C until use. Samples containing 10 μg protein were loaded and separated on denaturing 10% SDS-PAGE and then blotted onto nitrocellulose (Invitrogen, Carlsbad, California). Blots were blocked overnight at 4°C with 5% nonfat dry milk in 0.1% TBS-T and then incubated with anti-VEGF antibody (dilution 1:500, Chemicon, Temecula, California) or an antibody that recognizes Ang-1* (dilution 1:10,000, kindly supplied by Regeneron, Tarrytown, New York). Specific proteins were detected by enhanced chemiluminescence (Pierce, Rockford, Illinois).
Repeated measures analysis of variance analysis was performed for the multiple time point tissue perfusion comparisons between groups, and two-tailed paired Student ttests were used for the single time point comparison before and after administration of transgenes. Two-tailed unpaired Student ttests were used when appropriate. Values are expressed as mean ± SD.
Immediately after ligation, flux index declined in all animals by 75% to 80%. Spontaneous flux index recovery reached two-thirds of baseline levels by day 21, and did not change over the following five weeks (Fig. 1a).All experimental treatment with recombinant vectors was performed two months after ligation, a time, as noted, when spontaneous flow recovery was complete and stabilized, and hypoperfusion persisted. At this time Western blot analysis showed that endogenous VEGF levels were still increased compared with baseline levels (Fig. 1b). In the four groups of animals (Ang-1-only, VEGF-only, Ang-1+VEGF, Ad.Null), no significant differences in the levels of hypoperfusion were observed before injecting recombinant vectors.
Four days after Ad.VEGF injection, tissue VEGF levels dramatically increased compared with pre-injection levels. Likewise, injection of the Ad.Ang-1 viral vector into ischemic rabbit tissue markedly increased expression of Ang-1 (Fig. 1c).
At one week after injection, flow in the VEGF-only group increased 24% (p = 0.0014). Overall perfusion images showed diffuse areas of redness (depicting flux) (Fig. 2a).The Ang-1-only group demonstrated a 22% flow increase (p = 0.004), but red flux areas appeared confined within separate vessels (Fig 2a). At the end of follow-up (four weeks after injection), perfusion was 18% below baseline (pre-injection) (p = 0.004) in the VEGF-treated animals (Fig 2b). In contrast, perfusion in the Ang-1-only group remained elevated (26% above baseline, p < 0.001) (Fig 2b). The combination of VEGF and Ang-1 did not increase the effects caused by Ang-1 alone, and the Ad.Null group showed a flat line with no significant differences from baseline (Fig. 2b).
When recombinant vectors were injected into normally perfused ears, Ad.Ang-1 did not alter tissue flow; conversely, Ad.VEGF elicited the same pattern of inflammation as it did in the ischemic rabbit ear, which was accompanied by a transient but significant flow increase (Fig. 3a).
To test whether the increased endogenous VEGF levels present in chronically hypoperfused ears contribute to the collateral-enhancing effects of Ang-1, we simultaneously injected Ad.Ang-1 and Ad.Flt(1-3)-Fc, which encodes a soluble VEGF receptor that binds VEGF and thereby abolishes its signaling (16). The Ad.Ang-1-induced increase in flow was blocked, an effect not seen when Ad.Ang-1 was simultaneously injected with control adenoviral vector (Ad.Null) (Fig. 3b).
When vessel leakage was assessed three days after virus injection, Evens Blue concentration was significantly higher in the VEGF-only group than in the Ang-1-only group (33.7 ± 14.0 vs. 7.5 ± 3.6 ng Evens Blue/mg tissue, p < 0.001). In the Ang-1+VEGF group, concentration was 24.6 ± 18.6 ng/mg, less than the VEGF-only group (p < 0.05), but still higher than the Ang-1 group (p < 0.001) (Fig 4a).At seven days, the VEGF-only group showed greater vessel leakage versus Ang-1 (142.5 ± 38 vs. 49.5 ± 9.8 ng Evens Blue/mg tissue, p < 0.001). However, at 14 days there was no significant difference between the VEGF and Ang-1-only groups (13.4 ± 21.8 vs. 13.4 ± 7.8 ng Evens Blue/mg tissue) (Fig. 4a).
Electron microscopy showed that Ad.VEGF increases the amount of cytoplasmic vesicles (Fig. 4b, right) and loosens the intercellular conjunctions of endothelial cells as seen by small “gaps” between adjacent endothelial cells (Fig. 4b, left). These were not found in Ad.Ang-1- or Ad.Null-treated animals.
At one week after injection, Ad.VEGF induced localized inflammation, with purple discoloration of the skin, and edema. In contrast, Ad.Ang-1 produced only slight redness (Fig. 4c, top panel) and no edema (5.9 ± 0.3 mm vs. 2.5 ± 0.4 mm in cross-sectional thickness of VEGF vs. Ang-1; p < 0.0001; normal = 2.1 ± 0.1 mm). Moreover, regional temperature was significantly higher in Ad.VEGF- vs. Ad.Ang-1-treated ears (35.7 ± 3.5°C vs. 30.6 ± 1.7°C; p = 0.0005). This inflammatory response emerged two to three days after injection, peaked at seven days, and gradually declined over two weeks. In the Ad.VEGF/Ad.Ang-1 combined group, the ear showed a similar, but less intense, inflammatory response compared with the Ad.VEGF group, and in the Ad.Null group the ear showed inflammation similar to Ad.Ang-1 (data not shown); Ad.VEGF caused mononuclear cell infiltration of the injected area (42.3 ± 10.6 mononuclear cells per high-power view) (Fig. 4d). In contrast, after Ad.Ang-1 or Ad.Null injection, few mononuclear cells were found (6.3 ± 2.5 and 5.9 ± 3.3) (Fig. 4d).
The results of this investigation confirm that the relation of single proangiogenic factors to collateral development is complex and that their effects differ. In particular, we found that, in a model of chronic tissue hypoperfusion, both VEGF and Ang-1 increased flow 20% to 25% at one week after injection. However, the patterns of tissue perfusion, as assessed by laser Doppler, differed: Ang-1-induced flow increase was localized to larger vessels, with no visible inflammatory response; quite the reverse, VEGF produced a diffuse mottled increase in flow velocity that was associated with more pronounced swelling, vessel leakage, and inflammatory cell infiltration. At four weeks, both the inflammatory perfusion pattern and increased perfusion disappeared in the VEGF-treated ears; in fact, flow actually was lower than pre-treatment values. In contrast, the Ang-1-induced improvement was maintained at four weeks, with tissue perfusion 26% higher than that present during pre-treatment. This effect was dependent on the presence of elevated levels of endogenous VEGF in the hypoperfused tissue. The combination of Ad.VEGF and Ad.Ang-1 did not increase the effects caused by Ang-1 alone.
The model we employed is one of chronic hypoperfusion (11,17) in which endogenous VEGF levels are already increased. In this setting, additional VEGF, as noted, does not increase collateral perfusion. Moreover, when endogenous VEGF signaling is blocked by coadministering the gene encoding a soluble VEGF receptor, Ang-1-mediated flow increases are abolished (Fig. 3b). Consistent with this finding is the observation that, in normally perfused ears (in which endogenous VEGF levels are undetectable) (Fig. 1b), Ad.Ang-1 is unable to improve collateral blood flow (Fig. 3a).
Our data, therefore, indicate that a delicate equipoise is required between VEGF tissue concentrations and the ability of VEGF and of Ang-1 to enhance collateral perfusion-increased endogenous VEGF levels, present in chronic tissue hypoperfusion, are pivotal in creating a proarteriogenic milieu for Ang-1, but excessive VEGF expression can have deleterious effects. Other results are also compatible with this cautionary note regarding VEGF administration; thus, VEGF overexpression induced by gene transfer in a mouse hindlimb model of acute ischemia was also associated with deleterious effects (6).
The mechanisms responsible for the effects we observed are undoubtedly complex; Ang-1 activity is essential to endothelial cell survival, vascular branching, and pericyte recruitment. In a newly forming vessel, Ang-1 can be produced by mural cells (the abluminal cells of the vasculature, which include pericytes in the microvasculature and smooth muscle cells in large vessels), and once secreted will stabilize nascent vessels and make them leak-resistant (18). The interaction between Ang-1 and its receptor (Tie2) activates intracellular kinases, a step critical for vascular remodeling and maturation (19); Ang-1-induced angiogenesis is also dependent on endothelial-derived nitric oxide, which derives from the PI3-kinase/Akt-mediated activation of endothelial nitric oxide synthase (20).
Likewise, several activities are also necessary for the angiogenesis effects of VEGF. Among these are the chemoattractant effects of VEGF on inflammatory cells, and its capacity to induce vascular permeability factor (21,22). A coordinated infiltration of lymphocytes and macrophages appears to be crucial for regulating collateral vessel growth, probably through their release of soluble proangiogenic factors (23,24). In our study, however, exogenous VEGF administration appeared to induce excessive inflammatory cell infiltration that ultimately did not lead to sustained increases in collateral perfusion (Figs. 2a and 2b).
A limitation of our study is the absence of an anatomical assessment of the extent of collateralization in the tissues after ischemia and after each of the gene therapy-based treatments. We also provide no mechanistic insights into how exogenous administration of Ang-1 to an adult ischemic tissue modifies the structure, size, and function of ischemia-induced collaterals.
Additionally, VEGF is expressed in various isoforms in response to different stimuli; high-molecular-weight isoforms bind to the matrix and act at the local level, while low-molecular-weight isoforms are soluble and have paracrine functions. The relative role of each of these in regulating blood vessel formation is only partially known. The lack of effect of recombinant VEGF expression, observed in our model, could be related to the fact that expression of only one of these isoforms is not sufficient to induce collateral development. It should be noted, however, that soluble isoforms of VEGF can hypothetically be produced from exogenously introduced VEGF (i.e., through gene therapy) by plasmin-mediated post-transcriptional cleavage (21). This issue is important and warrants more detailed investigations.
The considerations discussed above, taken together with the fact that VEGF administration did not improve primary end points in two recently reported clinical angiogenesis trials (8,9), may have important clinical implications. They suggest that strategies involving administration of a single angiogenic agent may not result in optimal angiogenesis. Rather, it appears that an optimal angiogenic response will require multiple agents. Such a strategy may entail using an initial agent aimed at inducing vessel growth (i.e., VEGF) and a subsequent agent to promote stability and function (i.e., Ang-1), as suggested by the present study. However, a strategy relying on more complex interactions of multiple agents, such as might be achieved with administering a gene encoding a master-switch molecule or by using cell therapy (25), may be necessary. The issues raised by these considerations will undoubtedly be a major focus of future experimental efforts.
The authors would like to thank Daniel A. Canos and Hongsheng Wu for performing statistical analyses, and Dr. Cheol Lee and Ms. Sophia Rushton-Reid for their assistance in preparing the manuscript and graphics.
Supported by the research funding of our own institution, the MedStar Research Institute.
- Abbreviations and acronyms
- adenovirus-encoding Ang-1 gene
- adenovirus-encoding soluble VEGF receptor gene
- control adenovirus without transgene insert
- adenovirus-encoding VEGF gene
- soluble VEGF receptor gene
- plaque-forming unit
- vascular endothelial growth factor
- Received January 30, 2004.
- Revision received April 27, 2004.
- Accepted May 2, 2004.
- American College of Cardiology Foundation
- Sato K.,
- Wu T.,
- Laham R.J,
- et al.
- Chae J.K.,
- Kim I.,
- Lim S.T,
- et al.
- Masaki I.,
- Yonemitsu Y.,
- Yamashita A,
- et al.
- Lee C.W.,
- Stabile E.,
- Kinnaird T,
- et al.
- Henry T.D.,
- Annex B.H.,
- McKendall G.R,
- et al.
- Rajagopalan S.,
- Mohler E.R. 3rd.,
- Lederman R.J,
- et al.
- Stabile E.,
- Zhou Y.F.,
- Saji M,
- et al.
- Arras M.,
- Ito W.D.,
- Scholz D,
- et al.
- Stabile E.,
- Burnett M.S.,
- Watkins C,
- et al.
- Kinnaird T.,
- Stabile E.,
- Burnett M.S,
- et al.