Bioartificial matrices for therapeutic vascularization

Abstract

Therapeutic vascularization remains a significant challenge in regenerative medicine applications. Whether the goal is to induce vascular growth in ischemic tissue or scale up tissue-engineered constructs, the ability to induce the growth of patent, stable vasculature is a critical obstacle. We engineered polyethylene glycol–based bioartificial hydrogel matrices presenting protease-degradable sites, cell-adhesion motifs, and growth factors to induce the growth of vasculature in vivo. Compared to injection of soluble VEGF, these matrices delivered sustained in vivo levels of VEGF over 2 weeks as the matrix degraded. When implanted subcutaneously in rats, degradable constructs containing VEGF and arginine-glycine-aspartic acid tripeptide induced a significant number of vessels to grow into the implant at 2 weeks with increasing vessel density at 4 weeks. The mechanism of enhanced vascularization is likely cell-demanded release of VEGF, as the hydrogels may degrade substantially within 2 weeks. In a mouse model of hind-limb ischemia, delivery of these matrices resulted in significantly increased rate of reperfusion. These results support the application of engineered bioartificial matrices to promote vascularization for directed regenerative therapies.

Reduced vascular perfusion represents a significant cause of death and hospitalization in the United States, including 8 million Americans with peripheral artery disease (1) and 16.8 million Americans with coronary heart disease (2). Clinically viable therapeutic vascularization therapies will lead to better treatment for peripheral vascular disease, ischemic heart disease, and survival of cell and tissue transplants. Therapeutic vascularization has become a major research focus in regenerative medicine with several strategies being pursued (3–10). Most clinical trials have focused on delivery of a single angiogenic gene or growth-factor, although several groups are developing strategies to deliver or activate different progenitor cell types. At present, a large focus of clinical and preclinical work is centered on identifying the ideal angiogenic agent (or combination therapy), delivery strategy, and dosing regimen (11). There is a significant need for a suitable delivery vehicle to study novel vascular therapies in a controlled microenvironment.

Hydrogel matrices enable a high level of control for delivering regenerative therapeutic treatments. Although a large number of synthetic polymers have been shown to be useful implantable hydrogel materials (12, 13), polyethylene glycol diacrylate (PEGDA) provides several advantages for regenerative therapeutics. PEG has a well-established chemistry and long history of safety in vivo. Addition of acrylate groups flanking the PEG chain allows for photo or chemical cross-linking of a PEGDA macromer solution, as well as incorporation of biomolecules such as protease-degradable sites, adhesive ligands, and growth factors (14–16). A major strength of this strategy is the modular “plug-and-play” design of the base hydrogel system, which allows the tailoring of the biochemical and mechanical properties of the delivery vehicle. Due to its high water affinity, PEG is intrinsically resistant to protein adsorption and cell adhesion, providing extremely low background interference with incorporated biofunctionalities. Due to its modular nature and excellent in vivo properties, PEGDA is an appealing platform for delivering regenerative medicine therapies.

PEGDA-based bioartificial matrices have been shown to promote cell survival and endothelial tube formation in vitro (17). PEG vinyl-sulfone matrices containing matrix metalloproteinase (MMP)-degradable, sites, adhesive ligands, and bone morphogenetic protein (BMP-2) promote bone formation in a cranial defect to similar extents as BMP-2-loaded collagen (18). These bioartificial matrices serve simultaneously as engineered delivery vehicles and temporary matrices to support tissue ingrowth and remodeling. The goal of this study was to further the use of modular PEG-based matrices toward developing provascularization therapies. We hypothesized that PEGDA matrices containing a combination of protease-degradable sites, adhesive ligands, and vascular endothelial growth-factor (VEGF) would induce the growth of new vasculature into the implant in vivo and establish the therapeutic potential of this delivery vehicle in a model of hind-limb ischemia.

Results

Bioartificial Hydrogels Exhibit Degradation-Dependent VEGF Release and Support Cell Adhesion Activities.

Regenerative biomaterials designed to integrate with the host tissue must provide mechanisms for cell invasion. We engineered bioartificial hydrogel matrices to contain MMP-sensitive cross-links, adhesive ligands, and the growth factor VEGF (Fig. 1A). MMP-degradable PEGDA macromer (A-PEG-GPQ-PEG-A) was synthesized by reacting the proteolytically cleavable peptide GPQGIWGQK (18) with acrylate-PEG₃₄₀₀-NHS, a primary amine reactive cross-linker. PEG-acrylate functionalized arginine-glycine-aspartic acid tripeptide (RGD) adhesive ligand (A-PEG-RGD) was synthesized by reacting acrylate-PEG₃₄₀₀-NHS with the cell-adhesive peptide GRGDSPC. PEG-acrylate functionalized VEGF (A-PEG-VEGF) was synthesized by reacting acrylate-PEG₃₄₀₀-maleimide, a sulfohydryl-reacting cross-linker, with VEGF₁₂₁-cys, a growth factor containing an additional C-terminal cysteine (19). A-PEG-VEGF activity was assessed by endothelial cell proliferation assay and found to be equivalent to commercially available VEGF₁₆₅. Hydrogel matrices consisting of A-PEG-GPQ-PEG-A, A-PEG-RGD, and A-PEG-VEGF, were generated by polymerizing the acrylate end-functional groups with low-intensity UV light in the presence of the photoinitiator Ciba Irgacure 2959 (Fig. 1A and C). Nondegradable matrices consisted of photopolymerized PEGDA.

Fig. 1.

Design of PEGDA-based bioartificial matrices. (A) Bioactive ligands are functionalized with PEG-acrylate group: MMP-degradable cross-linker is functionalized with two PEG-acrylates, adhesive ligands and growth factors are mono-PEG-acrylated. Additon of a photoinitiator to an aqueous solution of PEGylated precursors results in formation of a cross-linked hydrogel after exposure to UV light by polymerization of acrylate end groups. (B) Biomolecules functionalized with PEG-acrylate have increased molecular weight when run on an SDS-PAGE gel. (i) Fluorescence of FITC label, Lane 1: RGD-FITC, Lane 2: A-PEG-RGD-FITC. (ii) BaCl/I stain for PEG, Lane 1: A-PEG-NHS, Lane 2 A-PEG-GPQ-PEG-A (iii) Coomassie, Lane 1: VEGF₁₂₁-cys, Lane 2: A-PEG-VEGF₁₂₁-cys. (C) Macroscopic view of a bioartificial PEGDA hydrogel cast in an 8 × 2 mm silicone mold.

Successful functionalization of bioactive ligands with acrylate-PEG–reactive group was confirmed by demonstrating increased molecular weight distributions (Fig. 1B). Cells express MMPs as they invade tissue and biological matrices such as collagen. The degradation profile of the bioartificial matrix is an important aspect because it is designed to maintain structural integrity in vivo and only degrade subsequent to tissue invasion. Susceptibility of A-PEG-GPQ-PEG-A hydrogels to MMP-mediated degradation was confirmed by progressive dissolution during incubation with active MMP-2 enzyme or collagenase-I over 20 h (Fig. 2A). Gel degradation was slowed by addition of the MMP-2 inhibitor oleoyl-N-hydroxylamide. Untreated gels with MMP-degradable sequences swelled slightly over 20 h with no weight loss and remained intact for 4 weeks in PBS. These results demonstrate that controlled degradation of synthetic hydrogels can be achieved by incorporating MMP-sensitive cross-links.

Fig. 2.

In vitro results for degradation and cell spreading. (A) Degradation of bioartificial hydrogels incubated in collagenase-1 or MMP-2. Gels are seen to degrade after several hours of incubation, and degradation is decreased by addition of inhibitor and gels in PBS do not degrade. (B) Viability stain of NIH3T3 fibroblasts seeded in hydrogel formulations spread in bioartificial matrices with degradable sequences and adhesive sites: (i) type I collagen, (ii) PEGDA (iii) A-PEG-GPQ-PEG-A, and (iv) A-PEG-GPQ-PEG-A + A-PEG-RGD.

In order to assess the functionality of the adhesive and degradation components of 3D hydrogels prior to implantation, NIH3T3 fibroblasts were encapsulated in gel formulations containing MMP-degradable sites, RGD, neither, or both. Cells remained rounded in gels containing only RGD or MMP-degradable sites but exhibited a spread morphology in gels containing both RGD and MMP-degradable sites, similar to cells in type I collagen (Fig. 2B). These results indicate that both adhesive ligands and MMP-degradable sites are necessary for cells to spread within the bioartificial 3D environment and are in agreement with published in vitro studies (20). The intrinsically low background signal of PEG permits detection of sensitive engineered biofunctional effects, and subsequent focused manipulation of these effects.

Bioartificial Matrices with VEGF Promote Subcutaneous Vessel Growth in Rats.

VEGF release from the hydrogel was measured in vivo by labeling A-PEG-VEGF with the amine-reactive infrared dye indocyanine green (ICG)-sulfo-OSu. ICG was observed to have significantly higher fluorescence after exposure to UV-photocross-linking compared to other dyes tested. Preformed constructs consisting of 10% PEGDA or A-PEG-GPQ-PEG-A and 2.8 μmol/mL A-PEG-RGD + 80 μg/mL A-PEG-VEGF-ICG were implanted subcutaneously in male Lewis rats. Control rats received subcutaneous injections of A-PEG-VEGF-ICG in PBS or matrices with no VEGF. The fluorescent signal in the rats was measured on a Caliper Xenogen IVIS Lumina bioluminescent imaging system at 0, 1, 3, 7, and 14 days post implantation. The fluorescent signal in the implant was quantified by gating a region of interest (ROI) around the periphery of the implant and subtracting the average background counts in the surrounding tissue from the average total counts in the implant, as background intensity varied between animals. A steady decline in signal intensity for the soluble VEGF injection was observed, with 90% of the signal lost by 2 weeks of implantation (Fig. 3A and B). In contrast, the degradable hydrogel matrix exhibited constant VEGF levels during the first 2 days followed by a gradual decrease over the next 12 days. The nondegradable matrix with VEGF showed nearly constant levels of VEGF over the 2-week implantation period. The soluble VEGF group had much higher initial fluorescence because this group was not exposed to UV light. We found that ICG exposure to UV light without the addition of PEG chains to absorb photoinitiator-generated radicals resulted in elimination of any detectable ICG signal. Samples containing unlabeled VEGF were undistinguishable from background fluorescence. Upon retrieval of the implants at day 14, the degradable samples were seen to have partially degenerated, and in some animals the gel was entirely dissociated. The nondegradable samples remained completely intact. New vessels were seen to grow in the tissue surrounding the implants with a general trend of a higher density of small blood vessels in tissue immediately adjacent to the implant. Larger and more regular vessels were seen growing into and around degradable implants, whereas nondegradable implants induced a large number of small vessels in the surrounding tissue.

Fig. 3.

Degradation of subcutaneous implants containing ICG-labeled VEGF. (A) Quantification of VEGF fluorescent signal in implants shows early release for degradable matrix and late release for nondegradable matrix. Soluble injection shows continuous decline in signal strength. Initial soluble signal is significantly higher as it was not attenuated by subjection to cross-linking conditions. (B) Representative images from IVIS scanning of fluorescently labeled VEGF in degradable implants, nondegradable implants, and PBS injection. Number indicates average counts per unit area within ROI.

We next evaluated the ability of these bioartificial matrices to promote vascularization in a subcutaneous implantation site. Hydrogels were photopolymerized in cylindrical silicone molds (9-mm diameter, 2 mm thick) around polycaprolactone (PCL) mesh disks to allow for visualization and retrieval of implants. Gel formulations included unfunctionalized PEGDA, MMP-degradable, MMP-degradable + RGD, or MMP-degradable + RGD + VEGF. Gel concentrations of 10% (wt/vol) PEGDA or A-PEG-GPQ-PEG-A (21), 2.8 μmol/mL A-PEG-RGD (22), and 80 μg/mL A-PEG-VEGF (19) were used. Hydrogel constructs were implanted dorsally in male Lewis rats. At 2 or 4 weeks, subjects were perfused with a radio-opaque silicon-based vascular contrast agent to gather quantitative and 3D structural data on patent blood vessel ingrowth (23). Examination of explanted constructs revealed that hydrogels remained intact except for degradation by invading blood vessels, even after 4 weeks in vivo. Micro–computer tomography (CT) analysis of the scanned constructs was gated within the periphery of the PCL ring to ensure only vessels inside the hydrogel were measured (Fig. 4A). Evaluation of the scanned explants revealed approximately 6-fold increased vascular density at 2 weeks and 12-fold increased vascular density at 4 weeks for degradable gels containing adhesive ligands, degradable sites, and VEGF compared to all other groups (Fig. 4B and C). The patency of the neovasculature was shown to be connected to the host circulatory system because contrast agent perfused through the aorta reached the vessels in the implant. These results validate the in vivo vascularization potential of the engineered hydrogel constructs.

Fig. 4.

Micro-CT images of bioartificial matrices implanted subcutaneously in rats perfused with Microfil radio-opaque contrast agent. (A) GPQ + RGD + VEGF implants showing vasculature in surrounding tissue growing into implant, gray volume defines hydrogel. (B) Quantification of vascular volume/total implant volume. ± SEM (C) Representative scans from nondegradable implants with no adhesive ligands (PEGDA), nondegradable with adhesive ligands (PEGDA + RGD), degradable A-PEG-GPQ-PEG-A with no adhesive ligands (GPQ), degradable A-PEG-GPQ-PEG-A with adhesive A-PEG-RGD (GPQ + RGD), and degradable A-PEG-GPQ-PEG-A with adhesive A-PEG-RGD and A-PEG-VEGF (GPQ + RGD + VEGF) N = 5.

Bioartificial Matrices with VEGF Increase Rate of Perfusion in a Mouse Model of Hind-Limb Ischemia.

Depending on the therapeutic application, bioartificial matrices can be delivered as preformed constructs containing cells and other regenerative agents such as growth factors, or they can be delivered as macromer solutions and polymerized in situ. To test the ability of PEG-based bioartificial matrices to improve reperfusion rates in a model of peripheral limb ischemia, we chose to deliver the matrix as a macromer solution and polymerize it in situ. By delivering the matrix as a macromer solution followed by in situ polymerization, we integrated the matrix more deeply within the target tissue than application of a preformed construct would have allowed. Macromer solutions of the PEGDA-based bioartificial matrix components were formulated to include RGD and MMP-degradable sites, with or without the addition of VEGF. Our preliminary studies indicated that better cell invasion and tissue integration occurs at lower concentrations of PEG hydrogel. We therefore reduced the concentration of A-PEG-GPQ-PEG-A to 5% (vol/vol) for this study, but kept the concentrations of A-PEG-RGD and A-PEG-VEGF at 2.8 μmol/mL and 80 μg/mL, respectively. The left leg femoral artery of 8–9 week old strain-129 mice was ligated and excised in compliance with a well-established model of peripheral limb ischemia (24–27). The right leg was left undisturbed to serve as a control reference. During femoral artery excision surgery, mice received (i) no treatment, or injections of (ii) PBS (vehicle control), (iii) soluble A-PEG-VEGF, (iv) soluble hydrogel precursors, or (v) soluble hydrogel precursors with A-PEG-VEGF at three sites (50 µL each) in the muscle groups surrounding the femoral artery region. Injected precursor solutions were immediately polymerized with low-intensity UV light exposure. The precursor solution injected into the muscle was seen to fully polymerize in situ during necropsy of test mice, indicating that the UV light penetrated the tissue deeply enough to cross-link the precursor solution. Preliminary studies established that exposure to the same UV wavelength, intensity, and duration on skin, muscle, or small bowel mesentery resulted in no detectable inflammation, tissue damage, or increased angiogenesis. At 4 and 7 days postsurgery, mice were imaged on a laser Doppler perfusion imaging (LDPI) system to quantitatively analyze perfusion to the peripheral limb. The perfusion ratio of the ischemic limb compared to the nonischemic limb in each animal was taken as the measurement for comparison in both the foot and the leg (ankle to proximal ligation) (Fig. 5). At day 4, the mice receiving matrix with VEGF showed a trend of increased perfusion in the leg although the differences were not statistically significant (p = 0.056), and no trend was seen in the feet, although the PBS control group is slightly elevated. By day 7, animals receiving hydrogels with VEGF exhibited a 50% increase in perfusion to the legs and a 100% increase in perfusion to the feet compared to untreated subjects.

Fig. 5.

Hind-limb perfusion in mice with ligated femoral artery and vein. (A) LDPI imaging of limb perfusion at day 7 responding to treatment conditions: no treatment, PBS injection, soluble A-PEG-VEGF₁₂₁-cys injection, degradable A-PEG-GPQ-PEG-A with adhesive A-PEG-RGDand degradable A-PEG-GPQ-PEG-A with adhesive A-PEG-RGD and A-PEG-VEGF₁₂₁-cys. (B) Quantification of perfusion ratio (normal leg : ischemic leg) at days 4 and 7 postsurgery. Error bars represent standard error of the mean (N = 10).

Discussion

Provascularization therapy remains a significant challenge in regenerative medicine. We adapted the PEGDA platform for in vivo vascularization applications and demonstrated efficacy with both construct implantation and in situ polymerization in subcutaneous and ischemia models. We first demonstrated ability to generate the PEGylated matrix components and form a hydrogel construct by UV photopolymerization. We showed that the hydrogels degrade in the presence of enzymes typically expressed by invading cells but not in buffer solution. We also showed that cells in the matrix require both adhesive sequences and degradable sites in order to spread within the bioartificial gel. This result indicates that the bioartificial material is a suitable matrix to promote tissue ingrowth and remodeling.

We showed that incorporation of VEGF in engineered hydrogel matrices modulated in vivo release kinetics. In gels containing proteolytically degradable cross-links, the matrix is designed to only release VEGF as the matrix is digested by invading cells. Degradable hydrogel matrices exhibited constant VEGF levels during the first 2 days followed by a gradual decrease over the next 12 days, whereas nondegradable matrices with VEGF showed nearly constant levels of VEGF over the 2-week implantation period. These results indicate that incorporation of degradable cross-links in the hydrogel controlled the release of VEGF from the matrix. In contrast, injected soluble VEGF levels steady declined with 90% of the signal lost by 2 weeks of implantation as expected for diffusion and turnover.

We next tested the ability of precast constructs to vascularize in a rat subcutaneous implant model. Engineered hydrogels containing MMP cleavage sites, RGD, and bound VEGF significantly enhanced vessel ingrowth by 2 weeks with increasing vasculature at 4 weeks. Importantly, the patency of this observed vascular ingrowth at 4 weeks was preserved, as shown by the ability of contrast agent perfused through the aorta to reach vessels within the implant. Matrices lacking either VEGF or RGD showed minimal tissue invasion. Nondegradable matrices failed to integrate with the host tissue on any level. We attribute the presence of long-term patent vessels at 2 and 4 weeks to controlled VEGF release and bioavailability from the degradable hydrogel.

The subcutaneous implantation experiment used 10% (wt/vol) hydrogels, consistent with the polymer density used in published reports (18, 20, 21). While conducting preliminary research for another in vivo study, we discovered that lower density hydrogels are potentially more useful for vascularization purposes because vessel invasion is faster and of higher density in more-easily degradable gels. Many researchers are interested in vascular-inductive matrices for cell-transplant applications, in which case vessel ingrowth will need to occur within a matter of days and not weeks to avoid ischemic-related die-off of transplanted cells. Future studies with bioartificial matrices for construct implantation should examine ways to improve vessel density and ingrowth-rate; one method being through reducing the amount of material that needs to be degraded for cell invasion.

Lastly, we implemented in situ polymerization of the bioartificial matrix in a mouse hind-limb ischemia functional model. Macromer solutions of matrix and matrix + VEGF injected into the muscle in areas made ischemic by femoral artery ligation were polymerized in situ with UV light. Blood perfusion to the ischemic limb was measured by LDPI and found to be greatest in animals that received matrices with bound VEGF at day 7 postsurgery. The result that the engineered matrix containing VEGF performs better than injection of soluble VEGF is noteworthy because it indicates that the delivery vehicle is acting synergistically to amplify the effect of the growth factor. It is presumed that the increased perfusion is due to growth-factor sequestration in the matrix, resulting in prolonged exposure that persists as the matrix is degraded and remodeled, as shown with sustained release in the in vivo degradation experiment. Furthermore, the adhesive ligands and degradable sequences in the matrix are designed to interact with endothelial cells undergoing angiogenesis. In reperfusion to the leg, the matrix alone (no VEGF) performed as well as soluble VEGF injection, indicating that the engineered adhesive and degradable hydrogel matrix itself has a beneficial healing or supportive effect. These results demonstrate the effective use a bioartificial hydrogel to act synergistically as a directive scaffold and a growth-factor delivery vehicle.

Several noteworthy studies indicate that delivery of VEGF has limited therapeutic success at achieving long-term, stable, vascular growth in humans (3). Although a strong stimulator of vascular growth, VEGF administration alone has limited ability to induce the growth of larger vessels. Long-duration exposure is necessary to produce stable microvasculature that does not regress after withdrawal of the VEGF stimulus (28, 29). In this study, we achieved steady vascular ingrowth continuing to a midrange time point of 4 weeks when the study was terminated. Based on the VEGF release results, we attribute the persistent vascularization to the conjugation of VEGF to the matrix, where it is only released in a proteolysis-dependent manner as opposed to diffuse or bolus injection delivery methods. This result is consistent with other reports that have used sustained-release strategies (19, 30, 31). By binding the growth factor to the matrix, a persistent provascularization signal is generated that does not quickly ramp up and fade away as in a soluble delivery method. The modular nature of bioartificial matrices makes studying the effects of other provascularization factors in the same controlled environment straightforward. Future studies incorporating more or different factors may be able to achieve even more robust healing effects.

The use of RGD as an adhesive ligand is often disputed because it is known to have lower adhesion performance or selectivity than several other proteins and protein fragments (32). However, RGD is a ligand for αvβ3 integrin, an adhesion receptor highly expressed in endothelial cells undergoing angiogenesis (33), and our results have demonstrated that RGD supports vessel invasion into degradable bioartificial matrices. The PEGDA platform is amenable to experimentation with other types of adhesive molecules. For example, ephrin was demonstrated to support endothelial cell adhesion and tubulogenesis in vitro (21). Other adhesive ligands such as the collagen-mimetic peptide GFOGER (34), the fibronectin fragment FNIII7-10 (35), and sequences from the ECM protein laminin can be used with the bioartificial system for studying other types of regenerative applications such as neurite outgrowth, epithelial morphogenesis (36), and osteogenesis. We emphasize that our strategy is not for synthetic biomimetic matrices to fully recapitulate the complete biological activities of native ECM. However, from an engineering perspective, bioartificial matrices provide many advantages that make their use for regenerative therapeutics attractive. Most importantly, bioartificial systems provide a high level of control to the designer using reproducible and synthetic components in modular plug-and-play architecture. A major conceptual aspect of these bioartificial matrices is their application as engineered platforms for directed cell invasion by incorporating bioactive adhesion motifs and enzyme-specific cleavage sites rather than serving as simple polymeric growth-factor reservoirs. In the present implantation studies, explant analyses suggest that the engineered matrix does not remain long term and is fully degraded over a period of weeks as new tissue forms. Based on this observation, the dominant mechanism of vascularization is likely cell-demanded, proteolysis-dependent release of VEGF. Further studies into the growth-factor release kinetics in vivo and cellular activity at the tissue-hydrogel interface are warranted to better characterize the specific nature of the observed therapeutic effects.

Materials and Methods

Matrix Synthesis and Degradation Profile.

Peptides were custom prepared by a commercial manufacturer (AAPPTEC) and supplied at 95% purity. Two molar equivalents of A-PEG-SCM (Creative PEGworks) per mole of GPQGIWGQK were dissolved in toluene and evaporated to a thick oil; the molar ratio of A-PEG-SCM to GRGDSPC and GGRGDSPGGK-carboxyfluorescein (RGD-FITC) was 1∶3. The evaporated oil was dissolved in dimethylformamide to bring the concentration of A-PEG-SCM to 50 mg/mL. The peptides to be PEGylated were added along with 1 M equivalent of triethanolamine per mole of A-PEG-SCM and reacted for 4 h. The product was precipitated in ether and dried, then dissolved in diH₂O, sterile filtered, and purified by dialysis. Products were lyophilized and stored at -20 °C. PEGylated products were run on a 20% SDS-PAGE gel to check for molecular weight increase. A-PEG-RGD-FITC was visualized directly by fluorescence to confirm small peptide PEGylation. A-PEG-GPQ-PEG-A was visualized by barium chloride/iodine stain described in ref. 37. Gels were degraded in vitro with 20 mU/mL collagenase-I (Sigma), 200 pM MMP-2 (Calbiochem), or 200 pM MMP-2 + 40 μM oleoyl-N-hydroxylamide (Calbiochem). Weight loss was determined by wet weight percent change.

VEGF Synthesis.

BL21 Star(DE3) Escherichia coli (Invitrogen) were transformed with VEGF₁₂₁-cys plasmid grown in DH5α E. coli (Invitrogen) following manufacturer’s instructions. Transformed BL21 Star(DE3) cells were induced with 1 mM IPTG at OD₆₀₀ = 0.8. After 4 h, cells were pelleted and frozen at -20 °C. Pelleted cells were thawed and lysed in bacterial protein extraction reagent (B-PER) protein extraction reagent (Pierce). Inclusion bodies containing VEGF₁₂₁-cys were separated from soluble protein by centrifugation at 15,000  × g and were solubilized for 1 h on ice in 10 mL B-PER reagent + 6 M urea. VEGF₁₂₁-cys was purified by 6× His-Bind column (Pierce) following manufacturer’s instructions with addition of 6 M urea to wash and elution buffers. 2 mM DTT was added to the elution fraction, which was sequentially dialyzed for 24 h against 4 M urea, 1 mM EDTA, 150 mM NaCl, 25 mM Tris-HCl, pH 7.5, followed by dialysis against 2 M urea, 1 mM EDTA, 150 mM NaCl, 25 mM Tris-HCl, pH 7.5, and dialysis against 1 × PBS. Endotoxin was removed from protein solution by a Detoxi-Gel Endotoxin Removing Column (Pierce), and endotoxin levels were verified to be below 0.1 endotoxin units/mL by Limulus Amebocyte Lysate colorimetric assay (Lonza). Protein was concentrated to 1 mg/mL and stored at -80 °C in 50% glycerol. VEGF₁₂₁-cys purity was verified by SDS-PAGE with Coomassie staining/Western blot and specificity by ELISA detection. VEGF₁₂₁-cys was functionalized by incubation overnight at 4 °C with 50 M excess acrylate-PEG-maleimide (Laysan Bio, custom order) in PBS. Functionalization was verified by molecular weight increase seen by SDS-PAGE and Coomassie staining. A-PEG-VEGF₁₂₁-cys activity was verified by addition to endothelial basal media [molecular cellular and developmental biology (MCDB)-121, 5% FBS, +ascorbate, L-glutamine] at concentration intervals and observing the effect on endothelial cell proliferation as compared to VEGF₁₆₅ (Invitrogen).

3D Construct Cell Seeding.

A-PEG-GPQ-PEG-A was dissolved in PBS at 10% (wt/vol) with 0.05% Irgacure 2959 (Ciba) photoinitiator. NIH3T3 cells were added at 300,000 cells/mL. 9 × 1 mm silicone isolator wells (Grace Bio Labs) were adhered to glass slides and filled with 70 µL PEGDA solution per well. Macromer solutions were cross-linked by exposure to 365-nm UV light at 10 mW/cm² for 12 min and placed in 1 mL of DMEM + 10%FBS and incubated for 24 h. Cells in 3D culture were stained with calcein acetoxymethyl ester for imaging and to ensure viabilility.

In Vivo Degradation Study.

A-PEG-VEGF₁₂₁-cys was labeled by overnight incubation at 4 °C with 200 M excess of the dye ICG-sulfo-OSu (Dojindo). Excess dye was removed by two rounds of gel filtration through Zeba Spin desalting columns (Pierce). 150-µL constructs consisting of 10% (wt/vol) A-PEG-GPQ-PEG-A, 2.8 μmol/mL A-PEG-RGD, 80 μg/mL ICG-VEGF with 0.05% Irgacure 2959 in PBS were cast in 9 × 2 mm molds and polymerized with a 10 min exposure to UV light. Constructs consisting of 10%PEGDA + RGD + ICG-VEGFwere used as nondegradable controls, and constructs with unlabeled A-PEG-VEGF₁₂₁ were used as imaging controls. Implants were equilibrated in PBS for 24 h to leach out any unbound ligand. Five rats per condition each received two dorsal subcutaneous implants or 150-µL subcutaneous injections of soluble ICG-VEGF. Rats were imaged at 745-nm excitation, 840-nm emission, and 60-s exposure time in a 700 series Xenogen IVIS machine. Images were analyzed for background-subtracted average photon counts within an ROI gated over the implant site.

Subcutaneous Vascularization, Microfil Perfusion, and Micro-CT Imaging.

Rings of 2-mm thickness PCL mesh (66% porosity, 300–500-um pore size) were made with concentric 8- and 5-mm sterile biopsy punches. The macroporous nature of these meshes readily allows tissue and vascular ingrowth. 100-µL constructs consisting of 10% (wt/vol) A-PEG-GPQ-PEG-A, 2.8 μmol/mL A-PEG-RGD, 80 μg/mL A-PEG-VEGF with 0.05% Irgacure 2959 in PBS were cast in 9 × 2 mm molds containing the PCL rings. Implants were equilibrated in PBS for 24 h. 16 male Lewis rats each received four randomized dorsal subcutaneous implants. After sacrifice by CO₂ inhalation, rats were perfused with 0.9% saline + 4 mg/mL papaverine hydrocholoride (Sigma), followed by 0.9% saline and 10% neutral buffered formalin. After fixation, 30 mL of 80% (vol/vol) diluted MV-122 Microfil (Flowtec) was injected into the aorta with a syringe and allowed to polymerize overnight before implant retrieval. Explants were scanned at 16-µm resolution with a Scanco μCT-40 micro-CT machine. ROIs were gated inside the edge of the PCL rings.

Hind-Limb Ischemia and LDPI Imaging.

At 8–9 weeks of age, male 129 mice (Charles River) were anesthetized with intraperitoneal injections of xylazine (10 mg/kg) and ketamine (80 mg/kg). A unilateral incision was made over the left medial thigh of the mouse. The superficial femoral artery and vein were ligated proximal to the caudally branching deep femoral artery and proximal to the branching of the tibial arteries. The portion of the artery and vein between the ligation points was excised. 150 μL of PBS, A-PEG-VEGF suspended in PBS, or nonpolymerized A-PEG-GPQ-PEG-A (5% wt/vol) + A-PEG-RGD (2.8 μmol/mL) (with or without 80 μg/mL A-PEG-VEGF) +0.05% Irgacure 2959 were injected into three sites in the ischemic muscle. The muscle was then exposed to 15 min of UV light (365 nm, 10 mW/cm²), which induced polymerization of the matrix. As an additional control, no solution was injected to the muscle, and the mice were not exposed to UV light. The skin was closed with interrupted silk sutures.

LDPI (Moor Instruments) was used to evaluate the perfusion in the ischemic and nonischemic legs at 4 ms/pixel scan speed, 256 × 256 resolution in arbitrary perfusion units. Perfusion was estimated (1) in the feet and (2) in the ischemic portion of the legs not including the feet. The nonischemic legs and feet were used as controls. The results were reported as ratios of surgery to nonsurgery leg/foot for each animal to account for natural variation in vasodilation between animals.

Statistics.

Statistical analyses were performed using one-way ANOVA with Tukey’s test for post hoc comparisons. For LDPI analysis, the perfusion values for the contra-lateral nonoperated leg were used as covariants. A p-value of 0.05 was considered significant.