Abstract
For gene therapy purposes, the skin is an attractive organ to target for systemic delivery of therapeutic proteins to treat systemic diseases, skin diseases, or skin cancer. To achieve long-term stable expression of a therapeutic gene in keratinocytes (KC), we have developed an approach using a bicistronic retroviral vector expressing the desired therapeutic gene linked to a selectable marker (multidrug resistant gene, MDR) that is then introduced into KC and fibroblasts (FB) to create genetically modified human skin equivalent (HSE). After grafting the HSE onto immunocompromised mice, topical colchicine treatment is used to select and enrich for genetically modified keratinocyte stem cells (KSC) that express MDR and are resistant to colchicine’s antimitotic effects. Both the apparatus for topical colchicine delivery and the colchicine doses have been optimized for application to human skin. This approach can be validated by systemic delivery of therapeutic factors such as erythropoietin and the antihypertensive atrial natriuretic peptide.
Keywords
Introduction
Human skin is the largest organ of the human body and is a good candidate for gene therapy. Human skin consists of 3 distinct layers including, from top to bottom, the epidermis, the dermis, and the subcutaneous tissue (Kanitakis, 2002). The epidermis is mainly composed of keratinocyte (KC, 90% to 95%); other cells in the epidermis include melanocytes, Langerhans cells, and Merkel cells (Kanitakis, 2002). The epidermis allows the establishment of a protective barrier against physical, chemical, or biological damage by continuously renewing itself every 21 to 28 days (Bergstresser and Costner, 2003; Halprin, 1972). The dermis contains a mixture of collagen, several different cell types (fibroblasts [FB], dermal dendrocytes, mast cells), as well as blood vessels (endothelial cells) and nerve fibers (Kanitakis, 2002). Its role is mainly to support and protect the epidermis and its appendages. The subcutaneous tissue contains fat and blood vessels (Khavari et al., 2002) and plays an important role in thermoregulation, insulation, provision of energy (nutritional store), and protection from mechanical injuries.
Gene therapy is powerful technology to introduce genes into target organs for a therapeutic purpose. Virtually all the human body organs can be targeted by gene therapy, but among these organs, the skin is one of the most attractive because of its size and easy accessibility. Additional advantages include the ability to monitor gene expression and to remove genetically modified cells if untoward side effects occur. Since the genetic basis of many inheritable skin diseases is known, a number of preclinical gene therapy studies have been conducted to treat acquired and inherited skin diseases. In cell culture and animal models (immunocompromised mice and dogs), efficient gene transfer of the corrective gene into skin cells has been used to treat junctional epidermolysis bullosa (JEB; Dellambra et al., 1998; Seitz et al., 1999; Spirito et al., 2006; Vailly et al., 1998), dystrophic epidermolysis bullosa (Baldeschi et al., 2003; Chen et al., 2002; Chen et al., 2000; Ortiz-Urda et al., 2003; Ortiz-Urda et al., 2002), xeroderma pigmentosum (Arnaudeau-Begard et al., 2003), X-linked ichtyosis (Freiberg et al., 1997; Jensen et al., 1993), as well as lamellar ichthyosis (Choate et al., 1996). In a recent clinical trial, a normal laminin 5-β3 gene was introduced into KC from a JEB patient using a retroviral vector, and the genetically modified KC was subsequently grafted back onto the patient. This ex vivo gene therapy approach (Figure 1) was able to revert the disease phenotype (Mavilio et al., 2006).
The skin is also suitable as a vehicle for systemic delivery of therapeutic molecules because of its high blood supply, which exceeds metabolic needs approximately 10-fold (Odland, 1983). In 1989, the Taichman group demonstrated that unmodified human epidermal sheets grafted onto immunocompromised mice or rats were able to produce and secrete a protein, human apolipoprotein, that was detectable systemically (Fenjves et al., 1989). Others studies have used transgenic mouse skin expressing TNF-α (Cheng et al., 1992), insulin-like growth factor II (Da Costa et al., 1994), or coagulation factor VIII (Fakharzadeh et al., 2000) in KC to show that human epidermis can produce and secrete therapeutic molecules systemically. Also, human or pig KC have been genetically modified in culture (ex vivo approach) and then transplanted as an epidermal sheet or human skin equivalent (HSE) onto immunocompromised mice to secrete coagulation factor IX (Gerrard et al., 1993), human growth hormone (Jensen et al., 1994), β-human chorionic gonadotropin (Krueger et al., 1994), or vascular endothelial growth factor (Rio et al., 1999). Alternatively, fibroblasts (FB) can be used for systemic gene delivery using an ex vivo approach. Genetically modified FB have been used for the systemic delivery of collagen type VII (Goto et al., 2006; Ortiz-Urda et al., 2003; Woodley et al., 2007), transferrin (Petersen et al., 1995), β-glucuronidase (Moullier et al., 1993), and alpha-galactosidase A (Medin et al., 1996). Although these studies have shown promising results and have demonstrated the feasibility of using skin for systemic delivery of protein products, they have not achieved long-term expression of the therapeutic protein. In this overview, we will discuss the gene therapy strategy that we have developed to achieve long-term expression of a therapeutic gene in human skin.
Skin Gene Therapy Approaches
Two different approaches exist to introduce new genetic material into skin cells, the in vivo and ex vivo approaches (Figure 1). In these 2 approaches, both viral and nonviral gene transfer can be used. For in vivo skin gene therapy, the therapeutic gene can be directly introduced into the skin using direct injection, liposomal or gene gun delivery of viral vectors, and naked DNA (Alexander and Akhurst, 1995; Domashenko et al. 2000; Hengge et al., 1995; Williams et al., 1991). We have chosen to use an ex vivo approach to achieve long-term expression of therapeutic proteins in the skin. In the ex vivo approach, KC are first isolated from a patient’s skin biopsy and then are genetically modified during in vitro culture by viral vectors to introduce the therapeutic gene. The bioengineered HSE is created in culture by seeding the gene-transduced KC onto a collagen matrix containing autologous FB (Garlick and Taichman, 1994; Ponec, 1991). This genetically modified HSE can then be grafted back onto the patient. This ex vivo approach is routinely used in our gene therapy studies.
Selection of Keratinocyte Stem Cells Using Multidrug Resistance Gene
Since the epidermis is a renewable tissue, keratinocyte stem cells (KSC) are required for the ongoing replacement of KC lost during normal homeostasis. Nucleotide-labeling experiments have determined that KSC are located in the bulge area of hair follicles, where they are concentrated, and in the basal layer of interfollicular epidermis (Bickenbach and Holbrook, 1987; Lavker et al., 1993). As shown in Figure 2, 1 KSC will give rise to 1 cell that will retain the KSC phenotype, while the other will become a transient amplifying (TA) cell. The TA cells are the short-lived progenitor cells, which undergo a finite number of cell divisions in the basal layer of the epidermis before starting to differentiate (Bickenbach et al., 2006). Once TA cells stop dividing, they begin the process of differentiation and stratification, forming the suprabasal layers and the stratum corneum (Figure 2). Since all differentiating KC will eventually be lost from the epidermis, the only cells that persist for a long time in the epidermis are believed to be the KSC.
To achieve a long-term expression of our therapeutic gene in a renewable tissue such as epidermis, we need to target the KSC population. Since there are no specific cell surface markers available to identify and isolate viable KCS while in culture, we have developed a method to select for the genetically modified KSC using the multidrug resistance (MDR) selectable marker gene, combined with topical colchicine selection. MDR is a transmembrane efflux transporter (p-glycoprotein) for a wide variety of cytotoxic drugs such as vincristine, colchicine, taxol, and others (Gottesman and Pastan, 1993; Higgins, 1993). Colchicine is an antimitotic agent that blocks cells in mitosis, and after a prolonged exposure, cells will eventually undergo apoptosis (Takano et al., 1993; Tsukidate et al., 1993).
To develop this model and demonstrate the feasibility of this approach, a monocistronic (or single gene) retroviral vector containing only the MDR gene was initially used to genetically target human KC and FB during ex vivo culture. These genetically modified KC and FB were then used to bioengineer human skin in culture. The hypothesis is that after grafting the MDR-modified HSE onto immunocompromised mice, topical colchicine will select for MDR-positive cells when applied to the skin and achieve long-term expression of the transgene. Previously, we have shown in an in vitro skin organ culture that we can highly enrich for KC expressing MDR when the MDR-modified HSE is treated with 50 ng/ml of colchicine (Pfutzner et al., 1999). More recently, we have shown in an in vivo animal model that a MDR-modified HSE grafted onto immunocompromised mice can be topically selected with colchicine, resulting in sustained high percentages of MDR-positive KC in vivo as well as a high MDR expression per cell, compared with MDR expression in control HSE that was not treated with colchicine (Pfutzner et al., 2002). In Figure 3A, HSE expressing MDR in 50% of KC (MDR+KC) at time 0 has been grafted onto immunocompromised mice and then topically selected 3 times per week with either vehicle control only or 200 μg/g colchicine cream. HSE was harvested at 9 and 15 weeks postgrafting, and the percentage of KC expressing MDR (MDR+KC) was determined by flow cytometry using an anti-MDR antibody. At 9 weeks, flow activated cell sorting (FACS) analysis of the grafts that were topically selected with colchicine contained 50 ± 26% (n = 8) MDR+KC compared to 13 ± 5% (n = 6; p = .002) in the vehicle control. At 15 weeks, colchicine-treated grafts contained 44 ± 15% (n = 8) MDR+KC, whereas the percentage of MDR+KC in vehicle control grafts continued to decline to 7 ± 4% (n = 7; p < .0001). In Figure 3B, the mean fluorescence intensity (MFI) or MDR-expression measured in MDR+KC is significantly higher in grafts selected with colchicine (16 ± 3 MFI) compared to the vehicle control (11 ± 2 MFI; p = .006). Although topical colchicine selection prevents a decrease in the percentage of MDR+KC (Figure 3A), topical colchicine did not increase the percentage of MDR+KC past 50%, even during long-term colchicine selection. One potential explanation why greater MDR+KC enrichment did not occur is that an insufficient topical colchicine dose (200 μg/g) was applied to the MDR-modified HSE.
Colchicine Dose Titration
The optimal dose of colchicine was previously determined by a topically applied dose titration onto mouse skin and human skin grafts. A dose of approximately 0.1 g of colchicine cream with increasing concentrations of 0, 100, 200, and 500 μg of colchicine per 1 g of cream were topically applied 3 times a week for a period of 4 weeks. Histological analysis has shown that the mouse skin was significantly more sensitive to topical colchicine effects than grafted human skin, perhaps because of structural differences between mouse and human skin, and this differential effect was a limiting feature of our model for colchicine delivery and selection. The maximal dose that could be topically applied to human skin grafts without causing ulceration to surrounding mouse skin was 0.1 g of 200 μg/g of colchicine cream 3 times a week. At that dose, we could see marked epidermal acanthosis (hyperproliferation of KC) in mouse skin and about 450/1,000 mouse basal cells blocked in mitosis without disrupting the epidermal integrity of the mouse skin (Puftzner et al., 2002). In human skin, basal cells are blocked in mitosis 7 times less than in mouse skin (66/1,000 vs. 450/1,000 basal cells, respectively; Puftzner et al., 2002). If a concentration higher than 200 μg/g is used to treat human skin graft, mouse skin ulceration is observed, and human skin graft is eventually lost.
To increase the topical colchicine dose delivered to the HSE grafted onto the back of immunocompromised mice, we have designed and constructed a delivery system in which only grafted human skin is exposed to the topically applied colchicine cream. As shown in Figure 4, a chamber for colchicine application is constructed on an immunocompromised mouse previously grafted with human skin. A punch hole the size of the human skin graft is made in the first transparent dressing that is wrapped around the mouse to expose the human skin graft (Figure 4A). Loctite Super Glue Control Gel is applied around the edge of the opening for human skin, and a piece of silicone gel containing a hole in the center is added to the top of the transparent dressing (Figure 4B). To keep this all together, another transparent dressing is wrapped around the mouse, again with an opening so the human skin graft is accessible (Figure 4C). The last step is to add an adhesive bandage with a hole for access and a second adhesive bandage cover that can be easily removed for when colchicine cream is applied (Figure 4D). The advantages of this system are that the mouse skin is not directly exposed to the colchicine cream, and anesthesia of the mouse is not required for topical application of each colchicine dose. This delivery system was durable and remained functional for 5 to 6 weeks with 2 to 3 colchicine applications per week.
Using this setup for colchicine delivery, a new colchicine dose titration for human skin grafted onto immunocompromised mouse was performed with the following concentrations: 0, 150, 450, 900, and 1,800 μg/g. At 450 μg/g, a histological image shows a marked epidermal acanthosis in the human skin grafts, and the amount of human basal cells blocked in mitosis have reached their maximum without disrupting the epidermal integrity of human skin (Figure 5C). At 900 μg/g, the number of human basal cells blocked in mitosis is comparable to the 450 μg/g dose, but inflammatory cells are more prevalent, suggesting that this dose has toxic effects on the skin (Figure 5C and 5D). At 1,800 μg/g, the human epidermis is completely lost (Figure 5E). We have thus determined that the optimal concentration is 450 μg/g for human skin. At this concentration, without the colchicine delivery system that we have constructed, the mouse skin integrity is completely disrupted (Figure 5F). Importantly, with this delivery system, the mouse skin surrounding the human skin grafts is not damaged or ulcerated by these higher colchicine concentrations (data not shown), enabling delivery of higher colchicine dose to the MDR-modified HSE.
Model for Systemic Delivery of Therapeutic Proteins
To demonstrate that skin can be used for long-term systemic delivery of therapeutic molecules, bicistronic (2 genes) retroviral vectors can be created that link expression of MDR to a desired therapeutic gene. Bicistronic retroviral vectors are characterized by the expression of 2 different genes controlled by a single promoter. A single mRNA coding for the two genes (MDR and a therapeutic gene) is expressed, with the 2 genes separated by an internal ribosome entry sequence, which allows protein translation of both genes. The gene in first position is usually expressed to a greater degree that the gene in second position; hence, it is important that the selectable marker gene be in the second position. The advantage of the bicistronic vector is that transduced and selected KC and FB that express MDR will also express the therapeutic gene. Additionally, selection will maintain expression of the therapeutic gene at a higher level for long time periods. Previously, MDR has been used successfully as a selectable marker to target hematopoietic progenitor cells (Aran et al., 1996; Galipeau et al., 1997; Hildinger et al., 1998; Licht et al., 1999; Licht et al., 2000; Sugimoto et al., 1995; Sugimoto et al., 2003; Zhou et al., 1998).
We have developed a bicistronic retroviral vector expressing atrial natriuretic peptide (ANP) and MDR in a genetically modified HSE. ANP is a peptide hormone synthetized mainly in cardiomyocytes and secreted in response to volume expansion and pressure overload. In target tissues, ANP induces a signal transduction pathway in a receptor-dependent manner to decrease blood pressure by inhibiting the renin-angiotensin-aldosterone pathway, relaxing blood vessels (vasodilation) and/or decreasing the cardiac output (Levin et al., 1998; Wijeyaratne and Moult, 1993). Infusion of ANP can decrease blood pressure in hypertensive patients and animal models, but stable long-term expression of ANP is required for clinical application (Lin et al., 1999; Wilkins et al., 1997). ANP-expressing HSE grafted onto hypertensive patients could be a good alternative to the infusions. Another group is also developing the bicistronic retroviral vector system to express different therapeutic genes, such as erythropoietin, using the ex vivo skin gene therapy approach (Scheidemann et al., 2006). The goal of these studies is to show that genetically modified HSE can be grafted onto immunocompromised mice, and therapeutic genes can be expressed and secreted systemically for a long period of time using topical colchicine selection to ultimately treat systemic diseases.
Conclusions
Human skin can be genetically modified to express proteins that are not usually made by this tissue. By targeting the KC progenitor cells, especially KSC, with a bicistronic retroviral vector containing a selectable marker such as MDR combined with topical colchicine application, an enrichment and long-term expression of the transgene can be achieved. The addition of a therapeutic gene to this system has great potential to treat acquired and inherited skin diseases as well as systemic illnesses such as hypertension.
Footnotes
Acknowledgements
We thank Carole Yee and Girish Patel for their advice and suggestions to improve this manuscript. This research was supported by the Intramural Research Program of the National Institutes of Health, National Cancer Institute, Center for Cancer Research. JPT was a recipient of a postdoctoral fellowship from the Canadian Institutes of Health Research (CIHR).