Abstract: Normal cutaneous wound healing proceeds via an ordered cascade of events that is mediated by specific growth factors, growth factor receptors, and cytokines. In chronic wounds, such as those occurring in the lower extremities of individuals with diabetes, there is a significant deficiency of endogenous growth factors and growth factor receptors. Attempts to overcome these deficiencies through the use of growth factor protein therapeutics have yielded disappointing clinical outcomes. Growth factor gene therapy has therefore attracted attention as a therapeutic alternative to protein therapeutics. It is speculated that a gene therapy approach will prolong the availability of therapeutic proteins, yielding improved healing responses. This review summarizes the various vector systems, transgenes, and animal models that have been employed to assess the feasibility of growth factor gene therapy for cutaneous wound repair. The results have been very promising, warranting ongoing clinical evaluation.

Introduction

The biology of normal wound healing is well characterized and has been the subject of several recent reviews.[1–3] Wound healing proceeds via an ordered cascade of events that is mediated by specific growth factors, growth factor receptors, and cytokines.[4–6] Normal cutaneous wound healing has three phases: inflammation, proliferation, and remodeling (maturation). In chronic wounds, such as lower-extremity neuropathic ulcers occurring in patients with diabetes, the wound healing response is impaired. This impairment is due in part to a deficiency of endogenous growth factors (e.g., platelet-derived growth factor [pdgf], epidermal growth factor [EGF], fibroblast growth factor-2 [FGF-2], transforming growth factor-beta [TGF-b]) and growth factor receptors.[7,8] The deficiency of growth factors in chronic wounds has been linked to elevated levels of matrix metalloproteinases and other neutrophil proteases.[9,10]

The finding of decreased levels of growth factors in nonhealing wounds gave rise to the obvious approach of treating chronic wounds with growth factor proteins. To date, numerous in-vivo studies have been performed with a variety of growth factor proteins (e.g., PDGF, TGF-b, FGF, EGF, keratinocyte growth factor [KGF], vascular endothelial growth factor [VEGF], and insulin-like growth factor [IGF]) in a variety of wound healing models.[4,11] The preclinical studies were very encouraging, demonstrating significant enhancements of wound repair, and biological responses consistent with the known functions of the specific growth factors.[11] In contrast, the clinical success of growth factor proteins has been disappointing.[11–13] To date, only recombinant human platelet-derived growth factor-BB (rhPDGF-BB) has been commercialized for the treatment of diabetic ulcers, and even it yields only modest improvements in healing.[5,14] The disappointing clinical experience with growth factors has been attributed to their short half-lives, degradation by wound proteases, and failure to maintain local protein levels above the therapeutic threshold.[9,10] Consequently, clinical efficacy requires high and frequent dosing, which is prohibitively expensive.

In light of these limitations, gene therapy has attracted significant attention as an alternative, cost-effective approach for cutaneous wound therapy.[12] While achieving long-term expression of a therapeutic gene remains a challenge for gene replacement strategies, only transient gene expression is required for wound repair. Tissue repair cells (i.e., fibroblasts, endothelial cells, inflammatory cells) are the target cells for deoxyribonucleic acid (DNA) uptake. It is speculated that gene therapy will enable production and persistence of growth factors throughout the inflammatory and proliferative phases, yielding improved healing responses compared with protein-based therapy. As the wound remodels, the transfected repair cells will die and transgene expression will cease. In other words, expression of the therapeutic protein will persist only as long as it is needed to promote wound repair.

The two general approaches for gene transfer are ex-vivo and in-vivo DNA delivery. In ex-vivo gene transfer, isolated cells are genetically modified in vitro and then transplanted back into the host. Genetically modified keratinocytes expressing growth factors have been effective in animal models of wound healing.[15,16] While promising, the extensive ex-vivo manipulations and expenses associated this approach have been significant hurdles to clinical development. For in-vivo gene transfer, the genetic material is delivered directly to the target tissue, thus avoiding the manipulations of ex-vivo gene delivery. For cutaneous wound healing applications, the ready accessibility of the target tissue makes in-vivo gene transfer the logical and preferred approach.

The methods by which DNA can be delivered in-vivo fall into one of three categories: biologic, chemical, and physical. The biologic method employs derivatives of naturally occurring viruses and exploits the highly efficient natural mechanisms by which these viruses enter cells, transport their DNA payloads to the nucleus, and use the cell’s machinery to activate viral gene expression. Four classes of virus have received the most attention as potential gene therapy vectors: retroviruses (including lentivirus), adeno-associated viruses (AAV), adenoviruses (Ad), and herpes simplex virus type 1 (HSV-1) (Table 1). For gene therapy vectors, regions of the wild-type viral genome required for replication are deleted, and the transgene of interest (and associated regulatory regions) are inserted. Of course, viral replication is required for vector manufacture, and therefore the missing viral functions must be provided in trans during production. This generally is accomplished by manufacturing the virus in a packaging cell line that has been genetically engineered to express the missing viral proteins or by co-infection of cells with a helper virus. Both approaches have associated risks. Recombination can occur between the viral genome and packaging cell DNA, creating a replication competent viral contaminant, and great care must be taken to remove all helper virus from vector
preparations.

Table 1

Retroviruses

Retroviruses are enveloped viruses with a 7 to 11kb single-stranded ribonucleic acid (RNA) genome. Different subclasses of retrovirus can be identified based on their genomic complexity.[17] Oncoretroviruses are the simplest retroviruses, consisting of three genes: the gag gene, which encodes core proteins; the pol gene, which encodes reverse transcriptase; and the env gene, which encodes the viral envelope protein.[18] Oncoretroviruses are exemplified by Murine Leukemia Virus (MLV) and infect only dividing cells. For use as gene therapy vectors, the gag, pol, and env genes are removed and replaced with the therapeutic transgene. Retroviruses are limited in the size of transgene that can be inserted and still allow for productive packaging (<6kb). Lentiviruses are a subclass of retroviruses that include the human immunodeficiency virus (HIV-1). While their basic genome structure (gag-pol-env) is the same as oncoretroviruses, lentiviruses have additional genes with protein products that enable infection of nondividing as well as dividing cells.[18] Therefore, lentiviruses have a broader potential target cell population compared with oncoretroviruses.

Retroviruses infect cells via cell surface receptors. The RNA genome is reverse transcribed into double-stranded (ds) DNA, which then stably integrates into the host cell genome.[17] Historically, retrovirus integration was assumed to be random. However, recent studies suggest that HIV-1 and MLV favor genes as sites of integration.[19,20] The integrative capacity of retroviral vectors makes them appealing for applications in which long-term transgene expression is required. However, the ability to integrate into the genome of target cells raises the theoretical risk of insertional mutagenesis. Another risk associated with retroviruses is that of generating a replication competent retrovirus (RCR) during manufacture.[17,21] In spite of these potential risks, retroviruses are the most commonly used vector in gene therapy clinical trials (~34%).[22]

Adenovirus

Adenoviruses (Ad) are nonenveloped viruses with a DNA genome of approximately 36kb in size. Adenoviruses exist as more than 50 serotypes; however, serotypes 2 and 5 are most commonly employed as gene therapy vectors.[23] Replication-defective Ad vectors are constructed by replacement of the E1 region with transgene sequences. There are upper limits for Ad genome sizes that can be productively packaged (105% of wild-type genome), and therefore additional viral genes can be deleted to accommodate large transgenes.[24]

Adenoviral vectors have several advantages over other viral vector systems, including efficient transduction of dividing and nondividing cells and transduction efficiencies of greater than 95 percent.[25] Adenoviruses are internalized via receptor-mediated endocytosis, a process mediated by the coxsackie adenovirus receptor (CAR) and integrins. Once in the nucleus, the Ad genome does not integrate into the host cell genome. Consequently, transgene expression is transient, and there is no risk for insertional mutagenesis. Drawbacks of Ad vectors include cytotoxicity of viral proteins and host cellular immune responses, resulting in local inflammation and destruction of transduced cells. Previous exposure to adenovirus is common in the human population, and there is the theoretical chance that pre-existing antibodies will limit the vector’s effectiveness when re-administration and/or long-term expression of the vector is required.[26] However, when delivered to a localized tissue site (i.e., intramuscular or intratumoral injection), effective repeat administration of Ad vectors has been demonstrated.[27,28]

Adenoviruses have been used in approximately 27 percent of all gene therapy clinical trials[22] and are good candidate vectors for indications requiring short-term, high levels of transgene expression (i.e., cutaneous wound repair).[26]

Adeno-Associated Virus (AAV)

AAVs are nonenveloped parvoviruses with a single-stranded DNA genome approximately 5kb in size. The AAV genome consists of two genes: rep, which encodes proteins that control replication and integration into the host genome, and cap, which encodes the capsid structural proteins.[29] AAV normally requires a helper virus, such as Ad, to mediate productive infections. AAV interacts with cellular receptors to enter cells. After internalization, the virus is trafficked through the endocytic pathway and transported to the nucleus where vector genomes are released.[30,31] Recombinant AAV vectors (rAAV) can infect dividing and nondividing cells and can express transgenes transiently from an episomal state or more permanently follow random integration into the host cell genome (wild-type AAV integrates at a specific locus on chromosome 19).[23,31] As is the case for Ad, there is a high prevalence of neutralizing antibodies against AAV in the human population, which potentially limits their utility in applications requiring repeated administration. On the other hand, since rAAVs are devoid of viral genes, they elicit limited cellular immune and inflammatory responses compared to adenoviral vectors.[30,31] A major limitation of AAV is its packaging capacity (<=4.5kb transgene). AAV is in use in 2.4 percent of current gene therapy trials.[22]

Herpes Simplex Virus

HSV-1 is an enveloped virus with a 152kb double-stranded DNA genome consisting of more than 80 genes.[29] It infects both dividing and non-dividing cells and does not integrate into the host cell genome. Because of their large genome size, 30–50 kb transgenes can be packaged into recombinant HSV-1 vectors, which are currently used in less than one percent of gene therapy trials.[22]

Non-Viral Vectors

Non-viral vectors (plasmid DNA) circumvent some of the drawbacks associated with the use of viral vectors. For example, large pieces of foreign DNA can be accommodated, and there is no concern regarding vector integration, recombination, or replication.[32] It is also generally accepted that non-viral vectors are less immunogenic than their viral counterparts. Nonetheless, there are potential problems associated with the use of non-viral vectors. Specifically, it has been demonstrated that immune cells are stimulated by unmethylated CpG residues in the plasmid DNA.[33,34] On the other hand, while viral vectors possess efficient mechanisms for delivery of DNA into cells, efficient delivery of nonviral vectors into target cells has proven a significant challenge. Therefore, much work has been devoted to developing non-viral vector systems that mimic the efficiency of DNA delivery exhibited by viral vectors.

Non-viral vectors can be delivered in-vivo by physical or chemical means. Physical delivery techniques include direct injection, particle-mediated gene delivery, and electroporation. The simplest and least risky technique for non-viral DNA delivery is direct injection into the target tissue. Direct injection of naked plasmid DNA is the method of choice in approximately 11 percent of gene therapy trials.[22]

The precise mechanisms by which naked DNA is taken up by cells remains unclear, and the efficiency of DNA transfection is generally low compared to that of viral vectors. Therefore, various physical and chemical methods have been developed in an attempt to increase the efficiency of plasmid DNA delivery and target cell transfection.

Physical methods include electroporation and particle-mediated gene delivery.[35,36] Particle-mediated gene transfer involves bombarding tissues with microparticles (typically gold or tungsten) that are coated with DNA. Using a device known as a “gene gun,” the particles are driven into cells by an electrical or pressure-induced force. The DNA payload is delivered directly into the cytoplasm of cells, thus bypassing the endosomal compartment and avoiding degradation by endosomal enzymes.[37] However, DNA is not stable in the cytoplasmic compartment either, succumbing to degradation by cytosolic nucleases with an apparent half life of 50 to 90 minutes.[38] As is the case for other nonviral delivery systems, transgene expression is transient, and transfection efficiency is low compared to viral systems.

Furthermore, the particles are nonbiodegradable and may eventually induce tissue damage. Particle-mediated gene delivery has been used in less than one percent of all gene therapy trials.22

Electroporation involves injection of the DNA into the target tissue, followed by the application of electric pulses and permeabilization of cell membranes.[37,39] The DNA then enters the cytoplasm of cells. Although this method has demonstrated success in pre-clinical studies and has attractive potential for wound healing applications, it has yet to be tested in the clinical setting.

Numerous chemical methods have been developed for non-viral gene transfer. These approaches include modifying the DNA payload with lipids, polymers, and proteins, including receptor-specific ligands.[32] Dramatic improvements in gene delivery have been demonstrated in vitro, but development of liposome-DNA complexes has progressed the furthest, and are utilized in approximately 12 percent of all current gene therapy protocols.[22,32] Cationic lipids are positively-charged, associate electrostatically with negatively charged plasmid DNA, and facilitate association with negatively charged cell surfaces. The complex enters the cell via endocytosis, and certain lipids can facilitate endosomal escape and delivery of the DNA to the cytoplasm.[40]

In-Vivo Studies

Numerous in-vivo studies, employing various DNA vectors, transgenes, modes of DNA delivery, and animal models have been conducted to determine the feasibility of using gene therapy to promote cutaneous wound repair (summarized in Table 2). Viral vectors encoding growth factors have demonstrated great promise. Gu, et al., summarizes the preclinical studies conducted in support of an ongoing Phase I clinical trial for the treatment of chronic diabetic ulcers of the lower extremity using an Ad5 vector encoding PDGF-B.41

Table 2

Adenoviral vectors encoding vascular endothelial growth factor (VEGF) have also been tested for wound healing activity. Topical treatment of excisional wounds in chemically-induced diabetic mice with an aqueous solution of AdVEGF165 promoted angiogenesis at the site of injury, increased granulation tissue formation, and accelerated wound closure when compared to controls.[42] Similarly, AAV encoding VEGF165 (rAAV-VEGF165) promotes angiogenesis and accelerated wound healing following intradermal injection into full-thickness excisional wounds in rats,[43] full-thickness incisional wounds in diabetic mice,[44] and partial-thickness scald burns in normal mice.[45]

A single, topical application of an Ad vector encoding human iNOS has been shown to reverse the delayed healing of full-thickness excisional wounds in iNOS knock-out mice,[46] and topical treatment of excisional wounds on the ears of New Zealand white rabbits with an Ad vector encoding hepatocyte growth factor (AdHGF) resulted in accelerated wound closure and prevention of scar formation.[47]

Treatment of ischemic rabbit ear wounds with a single intradermal injection of AdPDGF-B resulted in a large amount of granulation tissue and partial reepithelization.[48] In this same study, granulation tissue formation and re-epithelization in response to injection with PDGF-BB protein were not significantly different from responses to injection of vehicle alone. In a subsequent study, the effects of an immune response to AdPDGF-B on the wound healing response was assessed. Ischemic wounds on one rabbit ear were treated by injection of AdPDGF-B followed by treatment five weeks later of wounds on the second ear. While an immune response was elicited following the first set of injections, the wound healing response to the second set of injections was comparable to that seen in naïve animals.[49]

Plasmid DNAs encoding growth factors have also been shown to enhance in-vivo wound healing. Direct injection of an aqueous solution of a plasmid DNA encoding TGF-b1 into subcutaneously implanted polyvinyl alcohol (PVA) sponges resulted in increased collagen content compared with sponges injected with a reporter gene plasmid.[50] Similarly, a combination of topical application and subcutaneous injection of soluble aFGF plasmid DNA resulted in increased tensile strength, increased cellular density, and decreased scar formation in incisional wounds in diabetic mice.[51] However, the variable responses prompted the authors to investigate particle-mediated gene delivery as a more efficient transfection system. They found that particle-mediated transfer of TGF-b1 or EGF cDNAs enhances wound repair in rat and porcine wounds, respectively.[50,52] Similarly, particle-mediated delivery of PDGF-A or PDGF-B cDNAs to rat skin increased the tensile strength of full-thickness incisional wounds compared with wounds transfected with a reporter gene construct.[53]

Liposome-mediated DNA delivery has also been assessed in wound healing models. Topical application of lipofectamine-plasmid complexes containing the cDNA for FGF-1 (aFGF) results in an increased rate of wound closure, increased cellular density, and decreased scar formation in excisional wounds in diabetic mice.[51] Subcutaneous injections of liposome-plasmid complexes containing the cDNA for KGF or IGF-1 have demonstrated dose-dependent improvements in reepithelization, collagen deposition, and angiogenesis in burn wounds in rats.[54,55]

Gene-Activated Matrices

Despite the encouraging preclinical data, the approaches described thus far have not progressed to clinical evaluation due to low gene transfer efficiencies and labor-intensive methodologies. To overcome the limitations of current therapeutic modalities, the authors have developed a strategy for tissue repair and regeneration that uses biocompatible/biodegradable matrices to deliver DNA vectors, an approach called gene activated matrix (GAM).[56] Key to this approach is the ability of the matrix to retain the vector at the treatment site and be infiltrated by tissue repair cells for subsequent uptake and expression of the transgene. In addition to being permissive for cellular infiltration, association of the DNA vector with the matrix is desirable in order to ensure prolonged availability of the vector to incoming repair cells. In-vitro studies have demonstrated that the relative amounts of released and matrix-associated vector can be manipulated by altering the composition of the matrix.[57,58] The matrix can also localize the transgene product to the site of delivery, thus further enhancing wound healing potential.[58]

Stimulation of wound healing with GAMs has been demonstrated in several in-vivo models using both viral and non-viral DNA vectors. Gu, et al., describe the pre-clinical development of a formulation consisting of an E1-deleted Ad5 vector encoding human PDGF-B in 2.6 percent Type I collagen (AdPDGF-B/GAM).41 AdPDGF-B/GAM is currently undergoing evaluation in a Phase I study for the treatment of chronic diabetic ulcers of the lower extremity.

In other studies, Doukas, et al., formulated FGF gene vectors in collagen-based matrices and examined them in a skeletal muscle excisional wound model.[59] Increased angiogenesis, arteriogenesis, and myotube regeneration were observed following the delivery of either FGF-2 or FGF-6 genes to muscle wounds, when compared with reporter gene formulations. Similar responses were seen with both plasmid and Ad5-based vectors. Delivery of the FGF-2 protein as a collagen formulation failed to produce responses similar to the FGF genes.

Enhancement of granulation tissue formation and vascularization has been observed following subcutaneous implantation of poly(lactide-co-glycolide) (PLG) sponges containing a plasmid encoding PDGF-BB compared with sponges containing a reporter gene plasmid.[60] This response was dependent on the matrix and the transgene in that an aqueous formulation of the plasmid yielded no significant increases in granulation tissue or vessel number. GAM formulations of PDGF-B plasmid DNA have also demonstrated activity in collagen-based matrices. Collagen matrices containing either PDGF-A or PDGF-B DNA promoted increased granulation tissue formation and increased reepithelization of ischemic rabbit ear wounds when compared with controls.[61]

Over the years, pedicled transverse rectus abdominus musculocutaneous (TRAM) flaps have become increasingly popular for tissue reconstruction, especially as required to cover difficult wounds, such as diabetic ulcers, leg wounds, and for breast reconstruction following mastectomy.[62,63] The authors have demonstrated that prophylactic injection of collagen-embedded PDGF-B, but not FGF2, plasmid DNA markedly increases flap survival in a rat TRAM flap model (Figure 1).[64] The DNA/matrix formulations were delivered subcutaneously into the skin paddles seven days prior to flap elevation, and tissues were harvested seven days after flap elevation. Both PDGF-B and FGF-2 stimulated an increase in the flap capillary network but only PDGF-B DNA improved flap survival.

Figure 1

Collagen-immobilized plasmid DNA encoding PDGF-B induces TRAM flap survival. Collagen ± DNA (800µl, 6mg/mL DNA) was delivered seven days prior to flap elevation. Tissues were harvested seven days post- flap elevation. *P<0.0001 vs. collagen and collagen + pLuc (N=6–7)

Matrix-mediated gene therapy is also compatible with DNA vector modifications designed to increase levels and/or cell-type specificity of transgene expression. For example, conjugation of Ad to FGF-2 ablates the native tropism of the virus and redirects its uptake through high affinity cell surface FGF receptors.65 Given that FGF receptors are overexpressed on tissue repair cells, FGF-2 targeting of Ad5 vectors might allow for improved wound healing responses at reduced vector doses. Using the PVA sponge model, the authors have demonstrated that FGF-2-targeting of an Ad5 vector encoding PDGF-B generates increases in granulation tissue and angiogenesis comparable to non-targeted vectors at a log lower dose.57

Conclusions

The clinical development of growth factor protein therapeutics for wound repair has been hampered by the inherent instability of the proteins, especially in the hostile environment of a wound. Consequently, high and frequent dosing of protein is required for a therapeutic effect, thus making protein therapy labor-intensive and prohibitively expensive. Indeed, despite years of effort, only recombinant human platelet-derived growth factor-BB (rhPDGF-BB) has been commercialized for the treatment of diabetic ulcers, and even it yields only modest improvements in healing.[5,14] Therefore, the need for improved therapeutic modalities is evident.

Significant progress has been made in recent years in the development of gene therapeutics for cutaneous wound repair. Gene therapy offers the potential for sustained local availability of therapeutic proteins, decreased frequency of dosing, and therefore improved therapeutic responses at reduced costs. Matrix-mediated gene delivery may provide further efficacy and cost benefits. A phase I study currently in progress is evaluating the efficacy of a topically applied formulation of AdPDGF-B in collagen. This study represents the first topical gene therapy trial to be conducted.

While this review has focused on the application of growth factor gene therapy to cutaneous wounds, gene therapy offers promise in numerous other areas of tissue repair and regeneration, including therapeutic angiogenesis, bone fracture repair, nerve repair, cartilage repair, ligament repair, and surgical applications.[66,67] Furthermore, while growth factor genes are logical payloads for tissue repair, other candidate genes include growth factor receptors, protease inhibitors, extracellular matrix proteins, transcription factors, etc. In other words, gene therapy for tissue repair is most likely in its infancy.