Spatial and temporal control of angiogenesis and arterialization using focal applications of VEGF164 and Ang-1*

Shayn M. Peirce, Richard J. Price, Thomas C. Skalak


Microvascular networks undergo patterning changes that determine and reflect functional adaptations during tissue remodeling. Alterations in network architectures are a result of complex and integrated signaling events. To understand how two growth factor signals interact to stimulate angiogenesis and arterialization, we engineered spatially directed microvascular pattern changes in vivo by using combinations of focally delivered exogenous growth factors. We implanted microdelivery beads containing recombinant vascular endothelial growth factor-164 (VEGF164) and recombinant angiopoietin-1* (Ang-1*) into the dorsal subcutaneous tissue of fully anesthetized male Fischer 344 rats implanted with backpack window chambers, and we quantified vascular patterning changes by using intravital microscopy, a combination of architectural metrics, and immunohistochemistry. Focal delivery of VEGF164 caused spatially directed increases in both the total number and the density of vessels with diameters <25 μm 7 days after microbead implantation. Increases were maintained out to 14 days but were reduced to control values by day 21. The addition of Ang-1* on day 7 maintained these increases out to day 21, induced vessel order ratios comparable to control levels, and was accompanied by increases in the length density of smooth muscle α-actin-positive vessels. We achieved spatial control of patterning changes in vivo by using multisignal stimulation via focal delivery of exogenous growth factor combinations and conclude that Ang-1* administered subsequent to VEGF164 stimulation induces vascular growth while maintaining a network pattern consistent with native patterns that persist in the presence of vehicle control stimulation.

  • microcirculation
  • window chamber
  • microvascular remodeling
  • arteriogenesis
  • vascular tissue engineering
  • vascular endothelial growth factor-164
  • angiopoietin-1

microvascular remodeling in the adult animal underlies adaptive and regulatory events in normal life, such as wound repair (19, 43) and exercise-induced skeletal muscle adaptation (9, 22, 46), as well as in pathological conditions, such as myocardial responses to atherosclerosis (5) and tumor growth (17, 26). This complex process involves angiogenesis (20) or capillary formation from an existing capillary bed, arterialization (6, 33, 40) involving the recruitment and differentiation of contractile perivascular cells, and vessel regression (1, 3, 4, 7). Although epigenetic stimuli may include changes in hemodynamic stresses that impinge on the vessel wall, coordination of these events is ultimately achieved by biochemical signals, such as growth factor proteins. For example, vascular endothelial growth factor (VEGF) mediates angiogenesis by promoting endothelial cell recruitment, proliferation, and tube formation (14, 17, 21), whereas angiopoietin-1 (Ang-1) is thought to promote vessel stabilization and maturation (2, 42).

When a network is structurally remodeled, the architecture, or pattern, of the microvascular bed is altered. The pattern is suggestive of the function served by the particular network, the extent to which it is served, and the stimuli through which the pattern was created (24, 37, 38). Therefore, pattern formation, the assembly of blood vessels into a functional network of arterioles, capillaries, and venules of particular length and diameter distributions, is a central component of microvascular remodeling. Studying induced pattern changes can improve the understanding of how environmental stimuli orchestrate the remodeling process.

Determining the spatial and temporal effects of specific growth factors on microvascular pattern formation also has far-reaching therapeutic possibilities and tissue engineering applications. For example, a thorough understanding of the growth factors implicated in microvascular remodeling combined with their controlled delivery can direct the assembly of unique, tissue-specific networks for collateralization therapy in the rescue and healing of ischemic tissues. Knowledge of the growth factors' roles in pattern formation is critical to engineering tissues that require an adequate blood supply. The objectives of this study are to understand how multisignal combinations of angiogenic growth factors that have previously been shown to have associated roles in microvascular remodeling (2, 42, 44) work in concert to influence network patterning during remodeling and to evaluate whether these factors can spatially and temporally direct angiogenesis and arterialization to focal regions of tissue. We spatially manipulated network patterning in vivo by using focal exogenous delivery of recombinant VEGF164, Ang-1, and combinations of these growth factors, and we studied the ensuing localized patterning changes to obtain a signature of growth factor effects at the network level.


Window chamber apparatus. All experiments were performed using sterile techniques in accordance with the guidelines of the University of Virginia Animal Care and Use Committee. The dorsal skinfold backpack window chamber used in this study (32) allows for the noninvasive intravital observation of a naturally vascularized in vivo tissue throughout a 21-day period (Fig. 1). One hundred and twenty-one 250- to 300-g male Fischer 344 rats were anesthetized with intramuscular injections of ketamine (80 mg/kg body wt), atropine (0.08 mg/kg body wt), and xylazine (8 mg/kg body wt). Dorsal skin was folded along a centerline over the spine and suspended by the chamber. Epidermis and dermis on both sides of the chamber were surgically dissected to expose vascularized subcutaneous tissue, and protective glass shields were placed over the tissue. The total window area enclosed by the backpack chamber and protected from the environment by glass coverslips was 0.785 cm2, and the average thickness of the window tissue was ∼150 μm. After surgery, rats were placed in heated cages and returned to the vivarium. While in their cages, rats were given rat chow and water ad libidum.

Fig. 1.

Left: window chamber implanted into rat dorsal skin (scale bar = 10 mm). Center: window chamber tissue containing intrinsic microvascular networks with a vascular endothelial growth factor-164 (VEGF164) bead-stimulated tissue region shown at higher magnification (right: bead location outlined by white circle; scale bar = 200 μm.)

Noninvasive intravital image acquisition. Images of window chamber tissues were obtained noninvasively. At each time point, rats were anesthetized as described in Window chamber apparatus, glass cover-slips were removed, and the tissues were flushed with 0.9% NaCl sterile saline. Tissue was always imaged from one side of the chamber, selected based on image clarity on day 1. With the use of a ×20 dry objective, images of nine adjacent fields of view were captured and assembled into a single rectangular montaged image measuring 0.09 mm × 1.20 mm that was centered on the alginate bead. Total area imaged per window quadrant was 1.08 mm2, and all of this area was contained within a 1-mm radius of the alginate bead delivery device. Images were also recorded and these recordings were subsequently used during the data collection to indicate flow directions in the network. Visualization of vessels was enhanced by using a 443-nm bandpass hemoglobin optical filter. During measurement acquisition, steps were taken to maintain constant ambient temperature; measurements were obtained in a thermostated room at 22°C.

Tissue harvest, fixation, and histology. Rats were anesthetized and entire window tissues were excised, whole mounted on gelatin-coated glass slides, and fixed in 4% paraformaldehyde overnight in —4°C. Tissues were immunolabeled for smooth muscle α-actin (SMA) using purified Cy3 conjugated clone 1A4 mouse monoclonal anti-SMA (Sigma Biosciences, St. Louis, MO) diluted 1:400 in phosphate-buffered saline containing 0.1% saponin and 2% bovine albumin (Fischer Scientific, Pittsburgh, PA) at pH 7.4 (incubation for 1 h at room temperature). One field of view per window quadrant (area equal to 1.08 mm2) was examined under a ×10 dry objective using a Nikon TE-300 inverted microscope with confocal accessories (Bio-Rad μ-Radiance). Images were digitized and analyzed by using Scion Image software version 4.0.2 (Scion, Frederick, MD), wich was calibrated to measure the length of positive SMA staining along the vessels.

Microvascular network patterning measurements. Functional length density, diameter distribution, and vessel order ratios were used to describe patterning changes and identify the different types of network architectural adaptations. The Strahler ordering scheme (13, 41) was used to classify vessels into different orders. Capillaries, identified under the light microscope based on recognition of their characteristic single-file line of red blood cell flow, were designated as order 1 vessels, and sequential orders were assigned progressively upstream incrementing by branch points at which two vessels of the same order joined. Arteriovenous shunts were designated with an order value equal to 1 less than that of the upstream vessel. Arteriole anastomoses were rare, but when seen, were designated with the order value equal to one greater than the immediate downstream vessel. This ordering scheme for vessel loops preserved the downstream-to-upstream-oriented notation while capturing connectivity changes. Vessel order ratios were computed by dividing the total number of vessels in one order (n) by the total number of vessels in the order succeeding it (n + 1).

Digitized intravital microscopy images were analyzed by using Scion Image software. Vessel segments were traced, and lengths (defined as distances between branch points) and internal diameters (based on blood column widths) were recorded. Order value was assigned to each vessel segment. Lengths of vessels containing SMA-expressing perivascular cells were obtained from immuno-stained histological whole mounts.

Growth factor diffusion characterization. To quantify the amount and spatial distribution of VEGF164 diffusing out of the alginate bead and into the tissue, VEGF164 was iodinated with 0.5 mCi of 125I (Sigma Chemical) by using commercially available IodoBeads according to the protocol described by the manufacturer (Pierce, Rockford, IL) and delivered either via filling the entire chamber volume (1.6 μg/ml) and then sealing the window chamber (superfusion over the entire tissue) or via placement of alginate beads containing 125I-labeled VEGF164, one into each of the four tissue quadrants. Sterile 0.9% NaCl saline solution was dripped over the windows, and glass coverslips were placed on top. The tissues were maintained at room temperature for 2, 23, 47, or 71 h and then exposed to radiographic film for 1 h. Film was digitized, and Scion Image software was used to measure grayscale intensity levels correlated to 125I-labeled VEGF164 concentrations within the tissue. Intensity measurements for 125I-labeled VEGF164 superfusion were obtained along eight equally spaced radial directions from the center of the window chamber outward to the chamber's periphery. Intensity measurements for the bead delivery studies were obtained and averaged along four perpendicular radial directions from the bead outward to a distance of 3 mm from the bead.

Growth factor delivery. Alginate bead delivery devices (39) were placed in 50 μl of 10 μg/ml growth factor; either rat recombinant VEGF164 (R&D Systems) or recombinant Ang-1 (Ang-1*) [containing a modified NH2-terminus (45), Regeneron] in 10 mM phosphate-buffered saline (pH 7.4, Sigma Chemical) overnight and implanted into window tissues of anesthetized rats, one per window quadrant (Fig. 1), 4 days after window chamber implantation surgery (denoted as day 1). Vehicle control was 10 mM phosphate-buffered saline. The 4-day delay allowed the animals to recover from surgery and provided a suitable time period for network equilibration, as determined by initial characterization studies. Therefore, day 1 designates the fifth day after window implantation surgery, and all subsequent time points in this study (up to day 21) are referenced accordingly. Alginate beads were held in stationary positions in the tissue by the natural adhesiveness of the hydrogel throughout the 21-day study. Window quadrants were not physically isolated from neighboring quadrants. One study group was subjected to single time point VEGF164 superfusion, which was delivered in 1.6 μg/ml over the entire tissue surface by filling the chamber volume and then sealing the window chamber. A uniform concentration was maintained over 72 h as determined by characterizing growth factor diffusion (data not shown) using the method described above.

Statistical analysis. All results are presented in the form of means ± SD. All comparisons were made by using the statistical analysis tools provided by SigmaPlot 5.0 (SPPS). Data were tested for normality and analyzed by one-way ANOVA, followed by nonpaired Tukey's t-test. Significance was asserted at P < 0.05.


Baseline remodeling characterization. To assess baseline remodeling in response to the window chamber implantation surgery, patterning metrics (±SD) in 12 animals were characterized after window chamber implantation (Fig. 2). Functional length densities were calculated in different diameter groupings (0–25, 26–50, 51–75, and 76–100 μm) (data not shown). The only significant difference (P < 0.05) in functional length density from day 1 was seen on day 3, during which the functional length density of the smallest diameter vessels (<25 μm) increased (Fig. 2A). By day 5 this value had returned to that seen on day 1 and remained statistically similar to values on day 1 throughout the remainder of the study. There was no significant difference in functional length density at any time compared with day 1 in any other diameter category (data not shown).

Fig. 2.

Functional length density of vessels with diameters <25 μm (A) and percentage of vessels with diameters <25 μm (B) increases significantly on day 3 compared with day 1, but levels return to initial levels by day 5. *Significantly different from day 1, P < 0.05; values are means ± SD; n = 12 window quadrants.

Diameter distributions were calculated as a percentage of the total number of vessels (arterioles, capillaries, and venules) in a particular diameter grouping. At day 3, the percentage of vessels in the smallest diameter group had increased compared with day 1, but this value returned to the initial values by day 5 and remained there (Fig. 2B). Furthermore, there was no statistically significant difference in any of the vessel order ratios at any time point compared with day 1 or in SMA coverage length at any time points compared with day 1 (data not shown).

Growth factor superfusion. We hypothesized that the patterning effects elicited by focal exogenous growth factor exposure, wherein a concentration gradient was delivered over a duration of 72 h, would be distinct from those elicited by a single time point, spatially homogeneous growth factor exposure, wherein the spatial gradient and time-release delivery was eliminated. To test this hypothesis, one study group received a single superfusion of the entire tissue surface of either vehicle control or VEGF164. The concentration of superfused VEGF164 (1.6 μg/ml) was approximately equal to that experienced by a region of tissue 0.5 mm from the alginate beads after 3 h of bead implantation. This concentration was selected for superfusion because it was relevant to subsequent VEGF164 alginate bead stimulation and fell within a range previously shown to invoke a significant architectural remodeling response (27). There were no significant differences in the average functional length density of vessels with diameters <25 μm, percentage of vessels in this diameter category, order ratios, or SMA coverage lengths between microvascular beds stimulated with vehicle control (delivered via superfusion or alginate beads) versus those treated with VEGF164 superfusion (data not shown).

Growth factor diffusion characterization and delivery vehicle placement. 125I-labeled VEGF164 is released from the alginate bead into the tissue gradually over a 72-h time period (Fig. 3A). All of the growth factor remains in the tissue within 1 mm of the alginate bead, and a concentration gradient exists at all four time points.

Fig. 3.

A: diffusion of protein from the bead into the tissue at different time points as characterized by 125I-labeled VEGF164. Initial concentration of VEGF164 in beads was 10 μg/ml. Values are means ± SD; n = 8 window quadrants. B: bead placement for 4-region insertion scheme. One bead is contained by each quadrant, as represented by the small circles, with gray representing a growth factor-containing bead and white indicating a bead containing vehicle control. Dashed lines represent maximum diffusion distance (1 mm) according to diffusion characterization experiments. C: bead placement for 6-region insertion scheme. Small circles represent beads containing either growth factor (gray) or vehicle control (white). Dashed circles represent maximum diffusion distance (1 mm) according to diffusion characterization experiments. Hexagonal orientation allows adjacent bead placement with respect to the previously measured diffusion distance.

To test the spatial sensitivity of the delivery device in eliciting a localized response, we utilized a six-region bead implantation scheme (Fig. 3C), wherein alginate beads incubated in either 10 μg/ml VEGF164 or vehicle control (phosphate-buffered saline, pH 7.4) were placed into one of each of the six tissue regions in a hexagonal configuration. In this configuration VEGF164-bead spacing (2 mm apart) was identical to twice the maximum distance of growth factor diffusion (1 mm from the bead). We hypothesized that if the patterning response was limited to the region bounded by the VEGF164 diffusion, there would be comparable responses among the three regions stimulated by VEGF164-beads that would be distinct from the responses seen in regions stimulated by control beads, and we would have achieved spatially directed microvascular patterning.

Our hypothesis was confirmed by the data reported in Fig. 4. Tissue regions stimulated with microdelivery beads containing VEGF164 (regions 4, 5, and 6) were statistically similar to one another in all metrics, whereas control-treated tissue regions (regions 1, 2, and 3) were also similar to one another in all metrics at all time points. Stimulation with VEGF164-containing alginate beads produced nearly twofold increases in functional length density in the smallest diameter vessels (<25 μm) at two time points compared with control stimulation: days 7 and 14 (Fig. 4A). However, by day 21 functional length density in the VEGF164 bead-stimulated tissues was decreased to control bead-stimulated levels. Similarly, the percentage of vessels with diameters <25 μm was significantly increased at days 7 and 14 in the VEGF164 bead-stimulated tissue over that in the control bead-stimulated case, but by day 21 this percentage was reduced to the control level (Fig. 4B). Interestingly, the ratio values for 1:2 order vessels, reflecting the number of capillaries to the number of vessels located one branchpoint upstream, were also significantly increased at days 7 and 14 for the VEGF164 bead-stimulated tissue over those in the control bead-stimulated case, although this ratio returned to the control by day 21 (Fig. 4C). SMA coverage lengths in tissues stimulated with VEGF164, as a percentage of total vessel length, were comparable to control percentages at all time points throughout the study (Fig. 4D). Furthermore, VEGF164 bead-stimulated tissue did not differ significantly from the control treatment in functional length density, diameter distribution, order ratio, or SMA coverage at any time point in vessels larger than 25 μm (data not shown). Finally, the six-region bead insertion scheme generated similar results to those generated by using a four-region implantation scheme (Fig. 3B; data not shown), affirming the hypothesis that the network patterning changes were limited to the regions of tissue stimulated by the growth factor gradients.

Fig. 4.

Functional length density of vessels with diameters <25 μm (A), percentage of vessels with diameters <25 μm (B), order ratios for order 1:order 2 (C), and smooth muscle α-actin (SMA) coverage lengths of vessels with diameters <5 μm (D) for tissue regions treated with VEGF164 beads or control beads placed into the 6 regions. *Significantly different from corresponding control region: 3 vs. 4, 2 vs. 5, 1 vs. 6, P < 0.05; values are means ± SD; n = 6 tissue regions.

Directed patterning changes stimulated with VEGF164 vs. Ang-1*. To study the effects of Ang-1* on microvascular network patterning compared with VEGF164, 12 rats implanted with dorsal skin backpack windows received alginate beads containing VEGF164 or Ang-1* and vehicle control, implanted one per tissue quadrant (Fig. 3B) on day 1. These interventions generated localized network patterning changes that were distinct and spatially limited to the quadrants containing the microdelivery device (Fig. 5).

Fig. 5.

Microvascular networks stimulated with vehicle, VEGF164 or recombinant angiopoietin-1 (Ang-1*) delivered via alginate beads. Alginate beads are outlined by the circle (scale bar = 100 μm).

On days 7 and 14, VEGF164 caused significantly increased functional length densities (Fig. 6A) and increased percentages of <25 μm diameter vessels (Fig. 6B) compared with stimulation with control beads at these time points. At no time point was there a significant difference in either of these metrics in tissue quadrants stimulated by Ang-1* compared with those stimulated with vehicle control. Ratio values of 1:2 order were increased in the VEGF164 bead-stimulated group over the control-treated group after 7 and 14 days (Fig. 6C), but there was no significant difference between the control and Ang-1*-stimulated groups at any time point for this order ratio. SMA coverage length in <25 μm diameter vessels at days 4 and 7 was significantly increased in the Ang-1*-stimulated group compared with the control bead-stimulated group (Fig. 6D), but no differences were seen in the VEGF164-stimulated group.

Fig. 6.

Functional length density (A), percentage of vessels with diameters <25 μm (B), order ratios for order 1:order 2 (C), and SMA coverage lengths (D) of <25 μm diameter vessels for windows treated with microdelivery beads containing VEGF164, vehicle control, or Ang-1* on day 1. *Significantly different from control at that time point, P < 0.05; values are means ± SD; n = 6 treated quadrants.

Directed patterning changes stimulated by sequential applications of VEGF164 and Ang-1*. To study how temporal association of VEGF164 and Ang-1* could direct localized angiogenesis and the perivascular cell recruitment process associated with arterialization, 36 rats were divided into three equal groups, and growth factor or control treatment was administered via alginate bead on day 1 and day 7 by removing the initial microdelivery bead and implanting the second bead in the same location. One group received VEGF164 followed by vehicle control, another group received VEGF164 followed by Ang-1*, and a third group received VEGF164 followed by VEGF164 treatment (Fig. 7). On days 7 and 14, length densities (Fig. 8A) and percentages of <25 μm diameter vessels (Fig. 8B) were significantly increased for VEGF164/VEGF164, VEGF164/Ang-1*, and VEGF164/control groups compared with control/control. For the VEGF164/VEGF164 and VEGF164/Ang-1* groups, functional length density remained significantly elevated compared with control-control at day 21. However, the length density of VEGF164/control-stimulated vessels in this diameter category returned to control/control levels.

Fig. 7.

Bead implantation time course for two-treatment study and hypothesized remodeling responses.

Fig. 8.

Functional length density (A), percentage of vessels with diameters <25 μm (B), order ratios for order 1:order 2 (C), and SMA coverage lengths (D) of <25 μm diameter vessels for windows treated with two doses (day 1/day 7) via microdelivery beads, as follows: vehicle control/vehicle control, VEGF164/Ang-1*, VEGF164/VEGF164, or VEGF164/vehicle control. *Significantly different from control at that time point; †significantly different from VEGF164/VEGF164-treated group at that time point, P < 0.05; values are means ± SD; n = 6 treated quadrants.

There was a statistically significant increase in the 1:2 order ratio values of all treatment groups compared with control/control values on day 7 (Fig. 8C). On day 14, the VEGF164/VEGF164-stimulated tissue ratio and VEGF164/control-stimulated tissue ratio remained elevated above control levels; however, the 1:2 order ratio from VEGF164/Ang-1*-stimulated tissues had decreased to control levels. By day 21 there were no significant differences between growth factor-treated study groups and control/control-treated study group. The length of <25 μm diameter vessels coated with SMA-positive cells was significantly increased on day 14 in the VEGF164/Ang-1*-stimulated tissues compared with control-control and VEGF164-VEGF164-stimulated tissues. VEGF164/control and VEGF164/VEGF164 groups did not differ significantly from the control/control treatment group at any time point.


Previous investigations of microvascular remodeling have been focused on whole network adaptation in response to growth factor delivery by using large-scale exogenous applications of single angiogenic growth factors. In most of these studies the remodeling response was quantified by measuring the changes in total functional vascular density over time (10, 31). The current study uses a detailed pattern analysis approach to study microvascular remodeling patterning changes in response to spatially and temporally controlled focal applications of growth factors. In addition to capillary growth, vessel maturation via the acquisition of a perivascular cell coating, representing the early stages of the arterialization process, was also examined by using a combination of metrics that collectively describe pattern alterations. Unlike most previous studies, this research was conducted in healthy adult tissue, of which its natural response to wounding is microvascular remodeling. Consequently, the cells and signaling processes involved in microvascular remodeling are assumed to be both functional and naturally occurring. Microvascular network patterning is the result of many processes, including angiogenesis, arterialization, and regression, which are coordinated by multiple signaling events (33) and up to 1,000 genes (21). As molecular signals work in concert to elicit a cascade of cellular behaviors that contribute to this continuum, we developed experiments to examine the effects of stimulating remodeling using a temporal combination of VEGF164 and Ang-1*.

One main finding of this study is that temporally controlled focal applications of exogenous VEGF164 and Ang-1* administered via alginate microbeads spatially direct microvascular patterning changes in vivo. Specifically, VEGF164 increases the number and total length of <25 μm diameter vessels and increases the numbers of capillaries relative to the microvessels, presumably arterioles, in the 15- to 30-μm diameter range directly upstream from them (Fig. 6). These patterning changes are spatially localized to within a radial distance of 1 mm from the growth factor delivery source, whereas regions of tissue neighboring the microbead are unaffected. These data are consistent with previous results that illustrate the ability of VEGF to induce capillary sprouting (8, 14, 29, 35) and increase tissue vascularity in healthy tissue (12).

Although total vessel length increases are significant throughout day 14, these patterning changes are transient and regress to control values by day 21 of the study (Fig. 6A). We suggest the regression seen at this time point is in part due to the lack of growth factor stimulus present at this later time point, because the detectable diffusion gradient was shown to disappear after the first 3 days. We suggest that because the growth of newly formed microvessels is not accompanied by an increased metabolic demand, their presence is not essential to the function of the tissue. Therefore, the newly formed vessels may be eliminated by the natural regression process (4, 27), leaving only those vessels required for meeting metabolic demands. Consistent with our observations is experimental proof that short-term stimulation by VEGF causes transient remodeling responses (11). We hypothesize that beyond the 21-day time point network, architectures remodel to accommodate the metabolic needs of the tissue, whereas metabolically unnecessary vessels regress without additional subsequent treatments of exogenous VEGF164.

After 4 days, single application of Ang-1* transiently increased the percentage of vessels covered in SMA-positive cells (Fig. 6D). Interestingly, VEGF164 also stimulated a subtle prearterialization response at days 7 and 14 through the recruitment of perivascular cells expressing SMA. This conclusion is supported by the fact that VEGF164 induced significant total length density increases at these time points, but the length of vessels coated by SMA as a percentage of total vessel length did not change (Fig. 6D). Therefore, the total length of SMA-positive cell coverage increased to maintain this percentage near a constant value during these time points. This observation is consistent with previous work that suggests VEGF164 has the ability to protect endothelial cells from undergoing apoptosis (3) and promote vessel maintenance (16, 20). Exogenous treatment with VEGF164 alone has been shown to induce the formation of numerous enlarged vessels in the skin, which later remodel by acquiring a smooth muscle cell coat and function as medium-sized arteries (36). Despite changes in SMA-positive cell coverage, 1:2 order ratio values (which are increased in response to VEGF164 treatment) did not reflect an increase in arteriole relative to capillary growth (Fig. 6C). These data are not contradictory, however, because order ratios reflect branching patterns and are independent of vessel length density changes.

In achieving spatially directed network patterning, we demonstrated the importance of delivering a focal dose of growth factor by subjecting window tissues to a negative control of a single-time point superfusion of VEGF164. Although the concentration (protein/tissue volume) of superfused growth factor was comparable to the average concentration experienced by tissue within the region of focal VEGF164 diffusion, as confirmed by the radioisotope-labeled VEGF164 study, the superfusion delivery method did not generate significant patterning changes. In fact, a superfusion of VEGF164 caused responses similar to those seen when vehicle control was superfused, and neither treatment significantly altered network patterning. Therefore, we suggest that the localized diffusion gradient provided by the alginate microdelivery device is necessary for eliciting significant patterning changes. This conclusion has important implications in tissue engineering and therapies where it would be advantageous to direct vascular remodeling to or within localized tissue regions.

Ang-1 and VEGF are thought to play coordinated roles in vascular development (25, 40) and remodeling (15, 23, 30). It has been suggested that the production of VEGF and the angiopoietins must be spatially and temporally correlated to ensure a normal network architecture remodeling response (28). Transgenic mice overexpressing VEGF and Ang-1 exhibit increased dermal microvessel density compared with mice overexpressing either VEGF or Ang-1 alone. Microvessels in double transgenics were tortuous, capillary-like, and had increased diameters and lengths. However, unlike VEGF-overexpressing vessels, they were not excessively leaky (44). When Ang-1 was administered exogenously in a mouse corneal pocket, it did not induce significant capillary sprouting, but when coadministered with VEGF, Ang-1 significantly increased vascular density and perivascular cell recruitment. Furthermore, perivascular cell recruitment in VEGF-induced neovascularization was dependent on Ang-1; without it, vessels were unable to recruit perivascular cells during the first 6 days (2). These studies suggest Ang-1 functions concomitantly with VEGF to form mature functioning networks.

By sequentially applying VEGF164 and Ang-1*, we induced prolonged patterning changes that were distinct. The transient patterning response, as assessed by functional length density and diameter distribution, that was stimulated by a single dose of VEGF164 on day 1 was maintained to day 21 by delivering a subsequent dose of either VEGF164 or Ang-1* on day 7 (Fig. 8, A and B). Sequential treatments of VEGF164 almost doubled in the number of capillaries relative to the number of arterioles directly upstream (Fig. 8C). However, a secondary treatment with Ang-1* returned to and maintained the 1:2 order ratio to control levels by day 14. This change was accompanied by an increase in SMA-positive vessel length (Fig. 8D). These data suggest that the initial VEGF164 dose induced angiogenesis, and specifically increased total length of capillaries, whereas the secondary dose of Ang-1* maintained the normal vessel hierarchy and was associated with the recruitment of SMA-positive cells.

In a recent study, exogenous delivery of Ang-1* restored normal hierarchical vasculatures in retinas of murine neonates where perivascular mural cells were previously blocked from being recruited by injecting an antagonistic antibody against PDGF-receptor-β (45). Without exogenous Ang-1*, vessels were leaky, and the network pattern hierarchy was abnormal with increased total vascular density and enlarged trunk and capillary vessels. Delivery of Ang-1* did not reinstate perivascular cell recruitment in this model system, suggesting that this growth factor maintains endothelial cell integrity by imparting its effects directly on endothelial cells and not via perivascular cell-endothelial cell interactions. However, this study does not discount the possibility that Ang-1 stabilization of endothelial cell integrity may permit or even promote other downstream growth factor signals that facilitate perivascular cell recruitment. Our data showing that Ang-1* promoted perivascular recruitment and reestablished the 1:2 order ratio when used in conjunction with exogenous VEGF164 support this hypothesis.

In conclusion, previous work has suggested that growth factors function in a coordinated manner during angiogenesis, arterialization, and regression (6, 18, 33, 34), yet how they interact to produce pattern changes at the network level is not well understood. The in vivo approach of this study employs a series of metrics to describe whole network patterning changes in response to exogenous stimulation by spatially and temporally controlled applications of VEGF164, Ang-1*, and combinations of these growth factors. We have provided evidence for a coordinated role of VEGF164 and Ang-1* in generating remodeled networks with native patterning characteristics. In addition to studying the remodeling effects of multisignal combinations of growth factors, we have also demonstrated the ability to perform spatially directed angiogenesis and arterialization, a capability of importance to therapeutic vascularization strategies and tissue engineering.


The authors thank George D. Yancopoulos and Regeneron Pharmaceuticals for donating the Ang-1* growth factor protein, the laboratory of Dr. Joel Linden for supervising the radioisotope-labeling portion of this study, and Dr. Ioamnis Constantinidis of the University of Florida for supplying the alginate beads.


Sources of support for this study include National Heart, Lung, and Blood Institute Grants 5 T32 HL-07284-25 to S. M. Peirce, HL-52309 and HL-65958 to T. C. Skalak, and HL-66307 to R. J. Price.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract