Engineering of multifunctional gels integrating highly efficient growth factor delivery with endothelial cell transplantation

Abstract

Transplantation of Bcl-2-transduced human umbilical vein endothelial cells (ECs) in protein gels into the gastrocnemius muscle improves local reperfusion in immunodeficient mouse hosts with induced hind limb ischemia. We tested the hypothesis that incorporation of local, sustained growth factor delivery could enhance and accelerate this effect. Tissue engineering scaffolds often use synthetic polymers to enable controlled release of proteins, but most synthetic delivery systems have major limitations, most notably hydrophobicity and inefficient protein loading. Here, we report the development of a novel alginate-based delivery system for vascular endothelial growth factor-A₁₆₅ (VEGF) that exhibits superior loading efficiency and physical properties to previous systems in vitro. In vivo, VEGF released from alginate microparticles within protein gels was biologically active and, when combined with EC transplantation, led to increased survival of transplanted cells at 28 days. The composite graft described also improved early (14 days) tissue perfusion and late (28 days) muscle myoglobin expression, a sign of recovery from ischemia, compared with EC transplantation and VEGF delivery separately. We conclude that our improved approach to sustained VEGF delivery in tissue engineering is useful in vivo and that the integration of high efficiency protein delivery enhances the therapeutic effect of protein gel-based EC transplantation.—Jay, S. M., Shepherd, B. R., Bertram, J. P., Pober, J. S., Saltzman, W. M. Engineering of multifunctional gels integrating highly efficient growth factor delivery with endothelial cell transplantation.

Transplantation of healthy cells within a stable vehicle to sites of acute or chronic injury represents a promising strategy for the treatment of disease through regenerative medicine. For example, transplantation of progenitor or fully differentiated endothelial cells (ECs) may be used in therapeutic revascularization as a treatment for ischemia (1). The use of synthetic polymer scaffolds to support cells is a popular approach for tissue engineering; however, many of the most commonly used materials poorly support EC viability or vessel formation, and in general, the therapeutic efficacy of cell transplantation is typically limited by low viability and/or lack of physiological functionality of transplanted cells (2). Furthermore, these synthetic polymers create additional problems, often impeding host cell migration and thereby limiting the overall size of tissues derived from such constructs as well as retarding processes of tissue assembly (3). This is especially detrimental in applications where EC transplantation is needed in that the local microenvironment associated with vascular tissue injury deters recruitment of host ECs (1). In other words, scaffolds designed to promote angiogenesis may actually impede the process.

We have had success in therapeutic revascularization using a hydrated collagen-fibronectin (protein) -based scaffold to support transplantation of human umbilical vein ECs made resistant to apoptosis by transduction of Bcl-2 (Bcl-2-HUVECs; refs. 4, 5). These constructs recruit host smooth muscle cells and pericytes, resulting in the development of a mature (nonleaky) microvascular network, and also elicit a host arteriogenic response that increases perfusion to the implantation site. Although such constructs can enhance reperfusion of an ischemic hind limb in an immunodeficient mouse host, vessel maturation is relatively slow, requiring up to 8 wk. We reasoned that vessel formation could be accelerated and that the therapeutic outcome could be enhanced by delivering exogenous growth factors. However, the collagen-based matrices used in our previous studies are too porous to retain any growth factors of interest (6). Although chemical tethering of proteins to collagen has been demonstrated to result in sustained release (7), the processing steps required for scaffold production are not compatible with cell entrapment. A number of synthetic polymer systems have been reported to provide sustained release of growth factors in the form of matrices or microparticles (8,9,10,11). However, synthetic polymers typically exhibit extremely low efficiency protein delivery (12), making it difficult to achieve therapeutically relevant levels of growth factor without incorporating a significant mass of material, which is undesirable for reasons already mentioned. Microparticles composed of alginate, a hydrophilic natural polymer, have been shown to be effective delivery vehicles for vascular endothelial growth factor-A₁₆₅ (VEGF) (13,14,15). However, the large diameters of previously reported alginate microparticle systems are not ideal for integration into cell-loaded scaffolds, as they would likely induce significant microenvironmental gradient effects, which are especially detrimental in the case of VEGF (16), a growth factor with a narrow therapeutic window.

In this study, we describe the development and characterization of a highly efficient growth factor delivery system and its incorporation into protein gel constructs for EC transplantation. Using an emulsification method, we formulated alginate microparticles entrapping VEGF at high loading efficiency with a smaller size (diameter 10±5 μm) than previously reported. Control over particle size was critical in preventing intragel VEGF concentration gradients, and an ∼10 μm diameter was chosen to facilitate particle dispersion in the gels, while limiting particle migration out of the gels based on computer-generated models (17) and our knowledge of protein diffusion within collagen gels (18, 19). Also, we achieved prolonged release of VEGF by modulating ionic cross-linking agents without cytotoxic effects. In comparison with a popular synthetic polymer delivery system, poly(lactic-coglycolic acid) (PLGA), our particles had a significantly increased VEGF:polymer ratio, reducing the amount of material necessary for release of growth factor at therapeutically relevant levels. This allowed, for the first time, incorporation of VEGF-loaded particles into protein gels entrapping ECs without prevention of cord formation in vitro. Our results indicate an improved therapeutic response for this combinatorial therapy in vivo in a preclinical model of peripheral vascular disease, with increased tissue perfusion at 14 days and enhanced muscle recovery from ischemic injury at 28 days, as evidenced by increased myoglobin expression, compared with either therapy alone.

MATERIALS AND METHODS

Microparticle preparation

Alginate microparticles were prepared using an emulsification technique based on the method of Zheng et al. (20), with several modifications. Alginate from Macrocystis pyrifera [viscosity of ∼250 cP (2% solution, 25°C), ∼50 kDa; Sigma, St. Louis, MO, USA] was purified of endotoxins based on a published procedure (21), with confirmation using an LAL QCL-1000 kit (Lonza, Williamsport, PA, USA). Alginate (18 mg/ml) and hydroxypropylmethylcellulose (Sigma; 2 mg/ml) were codissolved in ultrapure H₂O, followed by the direct dissolution of bovine serum albumin (BSA):VEGF (50:1) at 1 mg/ml, for a theoretical maximum loading of 1000 ng/mg. Three milliliters of the resulting solution was added to 12 ml of iso-octane + 5% (v/v) Span 80 while homogenizing at 17,500 rpm. Tween 80, 0.75 ml of a 30% (v/v) aqueous solution, was then added to the emulsion. After 3 min of mixing, 3 ml of either a filtered 700 mM CaCl₂ or 700 mM ZnCl₂ solution (pH 7) was added at 3 ml/min. After an additional 3 min, 15 ml of 2-propanol was added, and the particles were allowed to cure for 3 min and then spun down at 4000 rpm for 1 min. The supernatant was removed, and the particles were washed 2 times for 8 min in 2-propanol, air-dried, resuspended in sterile endotoxin free H₂O, and lyophilized. ZnCl₂ cross-linked particles were then soaked in M199 medium for 4 h, filtered, and lyophilized overnight. All particles were stored at −20°C until use. ZnCl₂ and CaCl₂ cross-linked particles were used at a ratio of 1:1 for all experiments. For some batches, a small amount of I¹²⁵-conjugated VEGF (Perkin Elmer, Waltham, MA, USA) was added to the bulk VEGF solution before homogenization. PLGA microparticles with an identical theoretical maximum VEGF loading were prepared via double emulsion based on the protocol of Faranesh et al. (10), with several modifications. Briefly, PLGA (50:50 504, M_wof ∼55 kDa; Boehringer Ingelheim, Ridgefield, CT, USA) was dissolved in methylene chloride at 100 mg/ml; 200 μl of a 25 mg/ml BSA:VEGF (50:1) solution in 1× PBS was added dropwise to 1 ml of the PLGA solution while it was homogenizing at 24,000 rpm for 30 s. Five milliliters of a 1% (w/v) poly(vinyl alcohol) (PVA) with 5% (w/v) NaCl solution was subsequently added to the primary emulsion while it was vortexing at high speed for 30 s. The resulting water-in-oil-in-water emulsion was poured into 40 ml of 0.3% (w/v) PVA 5% (w/v) NaCl solution and stirred for 3 h to facilitate evaporation of the methylene chloride. The particles were then spun down at 12,000 rpm in a centrifuge, washed 3 times with double-distilled H₂O, and lyophilized. Blank batches were prepared with 200 μl 1× PBS (without VEGF).

Microparticle characterization

Microparticle sizing was determined in aqueous solution using a Coulter counter (Beckman Coulter Multisizer 3, Fullerton, CA, USA). Batch yields were calculated by weighing particles after lyophilization and comparing the total dry mass to the theoretical maximum. Release profiles were obtained by incubating particles at 3 mg/ml in 24-well plates at 37°C with a 5% CO₂ atmosphere in triplicate. Plates were constantly agitated using an undulating platform and all VEGF concentrations were determined by ELISA (VEGF DuoSet; R&D Systems, Minneapolis, MN, USA). The release medium was 20% FBS-M199 + l-glutamine and penicillin/streptomycin supplements (complete M199). Encapsulation efficiency was determined by collecting samples from release medium for 3 wk, followed by complete dissolution of microparticles to quantify any remaining unreleased VEGF. PLGA microparticles were dissolved in dimethyl sulfoxide, and protein was extracted via phase separation. Alginate microparticles were dissolved in 55 mM sodium citrate. The total amount of VEGF released was then compared with the theoretical maximum loading. For all analyses, particles were pooled from three separate batches.

Cell culture and in vitro bioactivity

HUVECs were routinely grown in complete M199 with endothelial cell growth supplement (ECGS) and were used before passage 6. Bioactivity measurements were made by counting sprouts of HUVECs from microcarriers in fibrin gels exposed to released VEGF from each particle type. Microparticles were incubated in EGM-2 (Lonza, Williamsport, PA, USA) without growth factors for a continuous period of 1, 5, 10, or 15 days. VEGF values were measured by ELISA, and subsequently HUVECs were cultured on Cytodex 3 microcarriers (GE Healthcare, Piscataway, NJ, USA) overnight and suspended within a fibrin gel as previously reported (22). Native VEGF at 10 ng/ml was used as the 100% bioactivity benchmark, and wells with media only (no VEGF) were used as the negative control. An average of 96 microcarriers were evaluated per condition. For cytotoxicity studies, HUVECs were seeded into 24- or 96-well plates at 10,000 cells/well in complete M199 + ECGS and allowed to adhere overnight. Subsequently, in 96-well plates, media were aspirated and replaced with fresh media containing the desired concentration of microparticles. After incubation for 48 h, cell survival was determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (n=4). For experiments examining cytotoxicity of soluble factors only, in 24-well plates media were aspirated and 500 μl of complete M199 + ECGS was added to the wells. An additional 500 μl complete M199 + ECGS was added to cell culture inserts containing the desired amounts of blank particles. After incubation for ∼72 h, and before confluence was observed, cell survival was determined using the MTT assay (n=6).

Protein gel implants

Collagen-fibronectin gels were prepared as described previously (4). For in vitro evaluation, microparticles were suspended in collagen-fibronectin solution with Bcl-2-HUVECs at 1.68 mg/ml in 48-well plates. Solutions were incubated at 37°C for 20 min to allow for collagen polymerization, an equal volume of complete M199+ECGS was added, and gels were incubated for 24 h. Bcl-2-HUVEC viability was assessed by direct cell counting (trypan blue exclusion) after digestion of gels with 1 mg/ml collagenase at 37°C for 2 h in triplicate. For subcutaneous implants in C57Bl/6 mice, gels were prepared identically, with the exception of excluding Bcl-2-HUVECs, and implanted into dorsal subcutaneous pockets after anesthesia. Each mouse received three gel implants: without particles, with blank particles, and with VEGF-releasing particles. Four mice were evaluated for both PLGA and alginate. Gels were harvested after 2 wk and processed by standard methods. Vessel counts were assessed using hematoxylin and eosin (H&E) and Bandeiraea simplicifolia lectin (BS-l) stained sections by a blinded observer. Experiments involving an established model of hind limb ischemia in C.B-17 SCID/bg mice were performed as described previously (4). Blood flow analysis, clinical score determination, arteriography, and statistical analyses were all performed as described by Yu et al. (23), unless otherwise noted. For experiments involving I¹²⁵-VEGF, C57Bl/6 underwent subcutaneous implantation of gels as described above. Mice were euthanized, and serum was withdrawn at specific time points. Serum was allowed to coagulate and spun down at 15,000 g for 30 min. I¹²⁵-VEGF content was determined using a gamma counter (Beckman Gamma 5500B, Fullerton, CA, USA). Histological and immunohistochemical analyses were performed by routine methods, as described in Schechner et al. (5). Primary antibodies used included F4/80 (1:100; Abcam, Cambridge, MA, USA), Bcl-2 (1:100; BD Pharmingen, San Jose, CA, USA), BS-l (1:100; Vector, Burlingame, CA, USA; 1:100), and myoglobin (Biomeda, Foster City, CA, USA; 1:200). All experiments were performed under protocols approved by the Yale Human Investigation and Yale Animal Care and Use Committees.

RESULTS

Particle morphology and VEGF release

Alginate microparticles displayed a more uniform size distribution and a significantly smaller average diameter (10±5 μm) than PLGA particles (30±18 μm; Fig. 1A). Although PLGA particles can be fabricated under specific conditions to diameters of 100 nm or lower, the smallest particles we were able to produce with a total protein loading of 5% (w/w), while retaining bioactivity of the encapsulated protein, had an average diameter of ∼30 μm in agreement with published data (10, 24). The use of an emulsification process led to a substantial increase in effective VEGF encapsulation efficiency in alginate compared with PLGA (Fig. 1B), while the batch yields for each process were similar (alginate=75±3% and PLGA=71±2%; n=3). Survival of nontransduced HUVECs in direct contact with blank microparticles of both PLGA and alginate in vitro was nearly identical at all concentrations examined (data not shown), and neither alginate nor PLGA produced soluble factors that were cytotoxic to HUVECs in vitro at concentrations up to 10 mg/ml (not shown).

Figure 1.

Alginate microparticles release bioactive VEGF at high efficiency. A) Size distribution of PLGA microparticles (mean=30±18 μm) and alginate microparticles (mean=10±5 μm) encapsulating VEGF. B) Loading and encapsulation efficiency of VEGF. C) Bioactivity of encapsulated VEGF. VEGF bioactivity, as measured via sprout formation, for alginate (solid line) and PLGA (dashed line) formulations compared with native (never encapsulated) VEGF (top thin dotted line) and media without VEGF (bottom thin dotted line), average of 96 microcarriers evaluated per condition. D) VEGF release from alginate is tunable via modulation of ionic cross-linking agents, as seen in different release profiles for particles cross-linked with 700 mM CaCl₂ (dashed line), 700 mM ZnCl₂ (gray line), or a 1:1 mixed population (black line). For all panels, particles were pooled from 3 separate batches (n = 3). Data are mean + se; *P < 0.05; ***P < 0.001.

The release of VEGF from PLGA microparticles was characterized by a large burst and a lack of substantial sustained release. In the best formulations we produced, >92% of the total released VEGF from PLGA (∼175 ng VEGF/mg PLGA; n=3) was released within the first 24 h, and <25 ng VEGF/mg PLGA were released during the subsequent 3 wk in vitro. VEGF release from alginate particles exhibited <45% release in the first 24 h (Fig. 1D). Over 600 ng VEGF/mg alginate was released from alginate microparticles, a total more than 3 times greater than that released from PLGA particles. VEGF bioactivity was vastly improved in alginate microparticles when compared with PLGA (Fig. 1C; n=3). In alginate particles, substantial bioactivity, as assessed by sprout formation, was still evident after >2 wk, while bioactivity of VEGF from PLGA microparticles declined sharply over time and was not significantly different from negative control values at 2 wk. Furthermore, we found that modulating ionic cross-linkers in the alginate particle fabrication process allowed for control over release kinetics, similar to the way release from PLGA can be affected by changing the lactic acid:glycolic acid ratio. By mixing populations of particles cross-linked by Ca²⁺ and Zn²⁺ ions, the release of VEGF from alginate was tuned to produce the most amenable profile for potential therapeutic benefit (Fig. 1D; n=3).

In vivo bioactivity of encapsulated VEGF

Release of VEGF from alginate microparticles embedded within acellular protein gels resulted in a significant angiogenic response in vivo (Fig. 2). The increase in VEGF release associated with alginate particles relative to PLGA corresponded to a significant improvement in vascular density (Fig. 2C) and vascular caliber (Fig. 2D; n=4) in vivo, as quantified using H&E (Fig. 2A, B) and BS-l stained sections. Analysis of F4/80 stained sections indicated no apparent differences in implant cellularity or capsular thickness (not shown), suggesting negligible variation in inflammatory response. To control for the high BSA content in the particle formulations, protein gels containing alginate particles loaded with BSA alone were analyzed. However, the responses to these particles were negligibly different than those to blank particles. Serum VEGF levels in animals receiving implants containing I¹²⁵-conjugated VEGF were not substantially above background and were significantly reduced compared with those in animals receiving subcutaneous injections of free I¹²⁵-VEGF (Fig. 2E), indicating local, and not systemic, delivery, a critical feature for VEGF due to its role in tumorigenesis (25).

Figure 2.

Alginate microparticle/protein gel composite grafts release biologically active VEGF in vivo. A, B) Representative H&E-stained sections from subcutaneous tissue surrounding 14-day explants of gels implanted into C57BL/6 mice containing either alginate (A) or PLGA microparticles (B; ×200). C) Tissue surrounding gels containing alginate microparticles encapsulating VEGF was more highly vascularized than that surrounding control gels. *P < 0.05; n = 8. D) Vascular caliber was increased in vessels associated with gels containing alginate particles encapsulating VEGF when compared with PLGA. ***P < 0.001; n = 4. E) Release of I¹²⁵-VEGF from alginate microparticles in acellular gels did not raise overall serum levels of VEGF, compared with subcutaneous injection of free VEGF, indicating local, and not systemic, therapy. Data are mean ± se; *P < 0.05; n = 4. Scale bars = 50 μm.

Incorporation of particles into Bcl-2-HUVEC protein gels

Integration of alginate microparticles into protein gels resulted in sustained VEGF release for up to 10 days in vitro (Fig. 3A; n=3). Inclusion of Bcl-2-HUVECs into protein gels containing alginate microparticles resulted in a significant alteration of in vitro release kinetics associated with a time-dependent decrease in cell viability (Fig. 3A; n=3). Cell viability was also dependent on particle concentration within the gel; encapsulation of VEGF was essential in maintaining in vitro cell viability at higher microparticle concentrations (Fig. 3B; n=3). At a particle concentration of 1.68 mg/ml, incorporation of blank or VEGF-containing alginate microparticles into Bcl-2-HUVEC-containing gels did not result in a significant change in cell viability over long time periods in vitro (Fig. 3C; n=3). Qualitatively, Bcl-2-HUVEC cord formation was unaffected by the presence of alginate microparticles, even at high concentrations (1.68 mg/ml), shown in Fig. 3D–F. These in vitro findings, combined with in vivo bioactivity, suggest that alginate microparticles can be used in conjunction with cell transplantation.

Figure 3.

Incorporation of alginate microparticles into cellularized protein gels. A) Bcl-2-HUVECs within collagen-fibronectin gels (dashed line), when compared with acellular gels (solid line), sequester VEGF (lines, right axis) as cell viability decreases (bars, left axis; n=3). B) Bcl-2-HUVEC viability (lines, right axis) is dependent on alginate microparticle concentration, with released VEGF (bars, left axis) functioning as a survival factor at higher particle concentrations. Bcl-2-HUVECs in gels with VEGF-containing particles (solid line) are more viable than those in gels with blank particles (dashed line) at identical concentrations (n=3). C–F) Bcl-2-HUVEC viability in protein gels over time is unaffected by incorporated alginate microparticles both with VEGF (large dashed line) and without VEGF (small dashed line), compared with gels without particles (solid line), both quantitatively (C) and qualitatively (D–F). Cord formation (arrows, D–F) is evident in Bcl-2-HUVECs alone (D), Bcl-2-HUVECs with blank alginate microparticles (1.68 mg/ml; E), and Bcl-2-HUVECs with VEGF-entrapping alginate microparticles (1.68 mg/ml; F). Data are mean ± sd; ×200; *P < 0.05. Scale bars = 50 μm.

Composite graft performance in hind limb ischemia

We examined the effect of incorporating alginate microparticles, either with or without VEGF, into acellular or Bcl-2-HUVEC-containing protein gels in a model of hind limb ischemia. At 14 days, we observed a significant increase in muscle reperfusion associated with grafts containing both Bcl-2-HUVECs and alginate microparticles encapsulating VEGF (ECμV) relative to blank microparticle control grafts (Fig. 4A; n=7). Overall recovery of perfusion at 28 days did not appear to be significantly affected by any of the treatments applied, indicating that the implant affected the kinetics of reperfusion more than the overall angiogenic response (Fig. 4B, C; n=7). Histological analysis revealed a significant increase in midgraft vascular density at 28 days associated with alginate microparticles, although not necessarily with VEGF (Fig. 4D; n=4). Inflammation is known to cause dramatic increases in local vascular density (26); however, F4/80 staining of midgraft tissue sections revealed little macrophage infiltration into any of the grafts at 28 days (data not shown). Staining for Bcl-2 indicated that the increase in vascular density was almost entirely due to an increase in Bcl-2-positive vessels (Fig. 4E–H; n=4) and not to a significant infiltration of host vessels to the graft, as verified by a lack of BS-l-positive vascular structures (not shown). Analysis of muscle tissue immediately surrounding the graft site at 28 days demonstrated a significant increase in myoglobin expression, a sign of muscle recovery, associated with ECμV grafts (Fig. 5; n=4). Significance vs. negative control was also attained for grafts containing Bcl-2-HUVEC and alginate microparticles without VEGF (ECμB), indicating the possibility of VEGF-independent, Bcl-2-mediated events in facilitating recovery.

Figure 4.

Integration of VEGF delivery into protein gels increases perfusion and cell survival in vivo. A) Hind limb perfusion data collected via a Doppler apparatus indicate that implants combining Bcl-2-HUVECs and VEGF-containing alginate microparticles (ECμV) facilitate improvement in reperfusion kinetics after ischemic injury (n=8). B) Data obtained from arteriograms indicate an early host arteriogenic response in all therapeutic grafts at 14 days, with a trend toward increased perfusion over ligation control (Lx) for Bcl-2-HUVECs alone (EC) and Bcl-2-HUVECs with VEGF-containing microparticles (ECμV) grafts at 28 days. C) Clinical scoring illustrates improved movement and leg condition in ECμV grafts at 28 days. D–G) ECμV implants display improved vascular density compared with ECs, VEGF-containing alginate particles alone (μV), or blank alginate particles alone (μB; *P<0.05; n=8). Representative sections from collagen gel implants containing ECs (E), Bcl-2-HUVECs and blank alginate microparticles (ECμB; F), and ECμV (G). H) Inclusion of alginate microparticles enhances survival of Bcl-2+ cells (red, arrows) within collagen gels at 28 days (×200; n=8). Data are mean ± se. Scale bars = 50 μm.

Figure 5.

Myoglobin immunohistochemistry on excised hind limb tissue. A–E) Representative sections from collagen gel implants secured within the gastrocnemius of C.B.-17/Scid-Bg mice after femoral artery ligation containing Bcl-2-HUVECs (EC; A), blank alginate microparticles (μB; B), VEGF containing alginate microparticles (μV; C), Bcl-2-HUVECs and blank alginate microparticles (ECμB; D), and Bcl-2-HUVECs and VEGF-containing alginate microparticles (ECμV; E). F) Implants containing both Bcl-2-HUVECs and VEGF-containing alginate microparticles display increased myoglobin expression (red, arrows) in muscle tissue immediately adjacent to the implant at 28 days (n=8). Data are mean ± se. *P < 0.05; **P < 0.01; ***P < 0.001. Scale bars = 50 μm.

DISCUSSION

Rapid therapeutic revascularization of acutely ischemic tissue may help to reduce tissue injury. Attempts to accomplish revascularization solely through delivery of growth factors have largely failed in clinical trials (27, 28); EC transplantation is an alternative approach. We previously reported (4) the capacity of transplanted Bcl-2-HUVECs to form an organized vascular network resulting in enhanced tissue perfusion in the setting of experimental femoral artery ligation. We hypothesized that integration of local, sustained growth factor delivery would improve the therapeutic response.

The concept of combining growth factor delivery with cell transplantation is not a new one, having been explored by our group (29) and others (30, 31). However, the present study demonstrates several significant advances. First, by using a protein gel system, we avoid the use of proinflammatory scaffold materials that may disrupt or delay the formation of neotissue (3). Second, we apply growth factor delivery to Bcl-2-transduced endothelial cells, which outperform nontransduced endothelial cells in terms of induction of host angiogenic (5, 32) and arteriogenic (4) responses. Finally, we have developed a superior approach for encapsulation and delivery of VEGF that is compatible with protein-gel based systems, overcoming a major limiting factor in the use of these constructs for tissue engineering applications (6). Cumulatively, we believe these advances confer a major advantage over previously reported approaches.

The preponderance of data collected in this study suggests that, in the context of VEGF delivery integrated with cell transplantation, alginate microparticles prepared via an emulsification method exhibit superior performance to PLGA microparticles of similar size, as shown in Fig. 1. PLGA is an obvious choice for protein delivery applications because of its biocompatibility, but in this case, the requirement for a relatively high total protein loading (5% w/w) nullified some potentially advantageous qualities of PLGA, namely, the versatility of fabrication into nanoparticles. Although it is possible to create high-loading PLGA-based delivery systems for some proteins, such as nerve growth factor (8), reports of PLGA nanoparticles encapsulating therapeutic proteins are scarce due to problems with loading efficiency, burst release, and formulation instability (24). In our hands, bioactivity of VEGF was not retained over time on release from PLGA nanoparticles, consistent with previous reports (33). Furthermore, The PLGA microparticles produced in this study were consistent with those previously reported in terms of size, encapsulation efficiency, and total protein released (10, 34); in each of these categories, alginate microparticles displayed a more desirable profile (Fig. 1A, B). In addition, the alginate microparticles fabricated for this study were significantly smaller than previously reported alginate beads entrapping VEGF, making them ideally suited for use as a depot delivery system (35) and are, to our knowledge, the first reported <20 μm alginate particles encapsulating VEGF. Moreover, the tuning of VEGF release rate via modulation of ionic cross-linkers is a novel concept that allowed for improved sustained release of growth factor (Fig. 1D).

The hydrophilicity of the alginate microparticles allowed for their simple integration into protein gels, which, in the absence of cells, resulted in the sustained release of bioactive VEGF that was capable of inducing a host angiogenic response (Figs. 2, 3A). Addition of particles to cell-loaded gels was accomplished without cytotoxic effects or disruption of cord formation in vitro (Fig. 3C–F) and resulted in an alteration in the VEGF release profile that correlated with cell viability in vitro (Fig. 3A). This indicates that ECs suspended in these composite gels were able to access VEGF and use it as a survival factor, with VEGF loading playing a critical role in maintaining EC viability in gels in vitro (Fig. 3B). Correlatively, our in vivo data indicate that the incorporation of alginate microparticles into proteins gels for cell transplantation leads to an increase in transplanted cell survival (Fig. 4E–H), and furthermore that release of VEGF from the incorporated particles results in improved muscle reperfusion (Fig. 4A) and recovery (Fig. 5) from induced ischemia in a mouse model. Importantly, this increase in cell survival is seen at 28 days. Although VEGF delivery has been demonstrated to improve transplanted cell viability in the short term (36), long term cell survival associated with VEGF delivery has been less encouraging (37). Also, our system delivers VEGF within the concentration range known to be beneficial for normal vascular development (16) at a lower total dose than other VEGF delivery systems used in the same model (38).

The observed increase in myoglobin expression (Fig. 5), accompanied by the improved perfusion seen at 14 days (Fig. 4A), as a result of treatment with ECμV grafts suggests enhanced muscle recovery from ischemia in C.B-17 SCID/bg mice. Immunohistochemical data indicate that this recovery is partially VEGF dependent, as evidenced by the significant difference between the ECμV and ECμB groups (Fig. 5), which concurs with previous reports that implicate VEGF as a stimulus for muscle recovery in general (39,40,41) and for enhanced myoglobin expression specifically (42). Interestingly, we also observed that the presence of alginate microparticles within gels, which correlated with increased survival of Bcl-2-positive cells regardless of VEGF-entrapment, resulted in increased myoglobin expression, as evidenced by the result for the ECμB group in Fig. 5, implicating either an alternative biochemical pathway or a possible mechanical stimulus. Qualitatively, no significant change in mechanical strength or handling characteristics was observed on incorporation of particles into gels, and attempts to elucidate a difference using rheometry and a tensiometer proved unsuccessful. Furthermore, we examined the possibility that alginate particles could directly interact with ECs in vitro and in vivo and did not observe any specific association beyond that which would be expected by chance. The apparent correlation between increased survival of Bcl-2-HUVECs and increased myoglobin expression in the absence of VEGF was not anticipated. Although interplay between VEGF and Bcl-2 has been shown to be synergistic with respect to generation of an angiogenic response (32) that may affect muscle recovery, the functionality of Bcl-2-HUVECs acting alone in this regard is unknown. These data indicate the possibility that the contribution of Bcl-2-HUVECs to host recovery may be made through secretion of paracrine factors that either initiate signaling events directly or recruit additional cells to the injury site.

The incorporation of a highly efficient delivery system for growth factors within a protein gel optimized for cell transplantation greatly enhances the versatility of such constructs for other applications. A major limitation of collagen-based systems for cell transplantation and tissue engineering is the inability to incorporate sustained delivery of therapeutic agents (6). Despite increasing efforts to develop biomimetic engineered scaffolds, collagen possesses unique physiological properties that make it an ideal vehicle for cell transplantation (43). Using a composite system, we observed that VEGF delivery combined with Bcl-2-HUVEC transplantation resulted in a more efficacious treatment for ischemic injury than either factor alone. The VEGF delivery system described displays physico-chemical properties and protein loading levels that enable tissue engineering applications, such as the combination of microparticulate growth factor delivery with cell transplantation in a protein gel setting, that are not practical using previously reported particle systems. Such composite grafts may be used for the development of primary treatment strategies for critical tissue ischemia or in the production of large-scale bioengineered solid organs, which require rapid, sustained tissue perfusion.