Introduction

Insulin injection therapy is the therapeutic standard for reducing hyperglycemia in insulin-dependent diabetes mellitus. An alternative approach, pancreas/beta-cell transplantation, can provide glycemic control, but is limited by donor scarcity, the need for long-term immunosuppression, and the short-term survival of transplanted cells.^1,2 Gene therapy to treat diabetes has utilized two approaches: (i) transdifferentiation of nonpancreatic cells into insulin-secreting cells; and (ii) secretion of insulin from surrogate beta-cells. In transdifferentiation, the fate of developmentally related cells (e.g., liver cells) could be redirected toward the pancreatic lineage by transfer of genes encoding for growth or transcription factors.^3,4,5 However, co-expression of other enzymes has limited the potential of this approach.⁵ The second strategy was aimed at using genetically modified surrogate cells, such as intestinal cells,⁶ hepatocytes,⁷ dermal fibroblasts,⁸ myoblasts,⁹ and epidermal keratinocytes¹⁰ to deliver insulin.

Investigators have explored several approaches to regulate insulin delivery from surrogate cells. For one, endocrine cells showed glucose-responsive secretion of insulin as a result of endogenously active regulatory mechanisms.^6,11 However, the co-regulation of other hormones presented a potential complication for clinical applications. The second approach employed glucose-responsive promoter elements to achieve glucose-regulated insulin secretion.^{12,13,14,15,16} However, with insulin secretion controlled at the transcriptional level, this system suffers from slow response to changes in glucose concentration. Recently, a third, novel approach employs a biomaterial, concanavalin A, that can be used for encapsulating insulin-secreting cells in such a manner that when glucose binds to concanavalin A, insulin is released into the bloodstream.¹⁷ Alternatively, secretion from surrogate cells can be controlled by fusing the protein of interest with a self-dimerization mutant of the FK506-binding protein (F_M) and a furin recognition sequence.¹⁸ The resulting fusion protein aggregates and accumulates in the endoplasmic reticulum until the addition of the F_M ligand (rapamycin or its analogs) dissociates the aggregates, thereby allowing the protein (e.g., insulin) to be cleaved by the ubiquitous protease furin and exported from the cells. This system can achieve a relatively rapid release of preformed protein in response to a pharmacological agent. Recently, we reported the modification of human epidermal keratinocytes to express the gene encoding for human proinsulin containing the furin recognition sequences at the A–C and B–C junctions. Primary keratinocytes and bioengineered skin substitutes were able to process proinsulin and secrete active insulin that promoted glucose uptake.¹⁰

In this study, we extend our work to achieve regulatable secretion of insulin by the epidermis in response to an exogenous ligand. As the results show, both epidermal keratinocytes and stratified bioengineered epidermis secreted active insulin within 30 minutes after addition of rapamycin. Moreover, transplanted keratinocytes expressed the transgene for several weeks in vivo and responded to multiple administrations of rapamycin by secreting bioactive insulin that reversed hyperglycemia in streptozotocin (STZ)-treated athymic mice. These results demonstrate that the skin may provide an alternative ectopic site for regulatable insulin delivery in the treatment of diabetes.

Results

Efficient modification of human primary keratinocytes using lentivirus encoding nF_M–insulin fusion protein

Based on the work of Rivera et al.,¹⁸ we generated a fusion protein composed of proinsulin (hppI4) and the self-dimerization mutant of FK506-binding protein (F_M). The resulting fusion protein accumulates in the endoplasmic reticulum until the addition of rapamycin, which dissociates the aggregates and allows the protein to be exported through the constitutive secretory apparatus. The proinsulin gene was modified to contain furin protease cleavage sites in place of the natural convertase sites that are cleaved only in pancreatic -cells and produce active insulin and C-peptide in the ratio 1:1.⁹ In addition, we mutated the tenth amino acid in the proinsulin B-chain from histidine to aspartic acid (hppI4-HD) so as to increase the production of bioactive insulin, as reported earlier in regard to liver cells,^19,20 as well as primary epidermal keratinocytes and skin equivalents.¹⁰

The fusion protein containing three or four F_M domains and hppI4-HD (3F_M-hppI4-HD or 4F_M-hppI4-HD) was cloned into a third-generation lentiviral vector²¹ upstream of the internal ribosome entry site, followed by the enhanced green fluorescence protein (GFP) gene. Internal ribosome entry site allows expression of GFP from the same promoter (cytomegalovirus) as the fusion protein, enabling monitoring of the efficiency of gene delivery and sorting of GFP+ cells using flow cytometry (Figure 1a).

Figure 1.

Efficient lentivirus-mediated gene transfer to human primary keratinocytes. (a) Schematic of lentiviral vector encoding for insulin. The proinsulin gene contains a histidine-to-aspartic acid mutation at the tenth amino acid of the B-chain (hppI4-HD) and is fused with three or four copies of FKBP12 mutant (F_M) through the furin cleavage sequence (FCS). The resulting gene is expressed under the control of the cytomegalovirus (CMV) promoter. In order to enhance gene expression, a herpes simplex virus thymidine kinase 5'-untranslated region (5'-UTR) is inserted downstream of the CMV promoter. For the purpose of determining transduction efficiency, the enhanced green fluorescent protein (GFP) gene was also introduced in the vector beyond the internal ribosome entry site (IRES). (b) Flow cytometry analysis of transduced primary keratinocytes. (c) Bright field and fluorescence imaging of transduced keratinocytes.

Full figure and legend (62K)

Recombinant lentivirus encoding for 4F_M-hppI4-HD was used for transducing human primary keratinocytes on fibronectin at high efficiency,²² yielding up to 80% GFP+ cells (Figure 1b), as determined using flow cytometry. Despite high levels of gene transfer, neither the morphology nor the growth rate of transduced cells appeared to change (Figure 1c and data not shown) over a period of five passages, the time that human primary keratinocytes can be normally cultured before terminal differentiation.

Insulin processing and secretion in response to rapamycin treatment

In order to determine whether transduced keratinocytes release insulin in a regulatable manner, cells were treated with vehicle or different concentrations of rapamycin for 4 hours. Insulin release was quantified by measuring the level of C-peptide in the medium. Figure 2a shows that insulin release increased with increasing concentrations of rapamycin, reaching a maximum level at 6 mol/l of ligand. In the absence of rapamycin, C-peptide levels were at least 20 times lower than the maximal levels attained with rapamycin. The addition of the herpes simplex virus thymidine kinase 5'-untranslated region upstream of the nF_M-hppI4-HD gene enhanced insulin production by 60% . Interestingly, the incorporation of a fourth F_M domain did not further increase insulin production (Figure 2a), and the rest of the experiments were performed with 3F_M domains.²³

Figure 2.

Insulin processing and secretion in response to rapamycin (Rapa) treatment. (a) Human primary keratinocytes were transduced with lentivirus encoding for insulin, fused with either three or four copies of F_M domains. The modified cells were treated with vehicle only or with the indicated concentrations of rapamycin for 4 hours. The conditioned medium was harvested and C-peptide levels were determined using enzyme-linked immunosorbent assay. The values are mean SD from a representative experiment (n = 3). (b) Immunostaining for insulin localization (red) in modified keratinocytes treated with either vehicle or 6 mol/l rapamycin for 4 hours. Cell nuclei were counterstained with Hoechst 33258 (blue). The bar represents 10 m. (c) Western blots for insulin in the lysate; (d) the F_M domains in the lysate; or (e) the culture medium of epidermal keratinocytes treated with either vehicle or 6 mol/l rapamycin for the indicated periods. FCS, furin cleavage sequence; GFP, green fluorescent protein; UTR, untranslated region.

Full figure and legend (40K)

In agreement with the enzyme-linked immunosorbent assay results, immunostaining showed insulin accumulation around the cell nucleus. Within 4 hours after addition of rapamycin, however, the staining disappeared (Figure 2b), probably because of insulin release into the medium. Indeed, western blots showed that insulin was present in cell lysates but that its level decreased progressively 4–8 hours after addition of rapamycin (Figure 2c and d). The protein band appeared at 45 kd as expected from the molecular weight of 3F_M (36 kd) plus proinsulin (9 kd). Conversely, at 4 and 8 hours after addition of rapamycin, a 36-kd band was detected in the culture medium, thereby suggesting release of free 3F_M domain from the cells (Figure 2e).

Kinetics of insulin secretion from genetically modified keratinocytes

After the addition of rapamycin (10 mol/l) to keratinocytes that are genetically modified, C-peptide release was detected within 30 minutes, increasing at a constant rate of 0.67 0.05 pmol/million cells/hour (n = 4) over a period of 4 hours. Thereafter, the secretion rate decreased and eventually reached a plateau (Figure 3a).

Figure 3.

Rapid and tightly regulated insulin secretion from transduced keratinocytes in response to repeated rapamycin administration. (a) Insulin release kinetics. Modified cells were treated with 10 mol/l rapamycin, and the culture medium was harvested at the indicated time-points. (b) Modified cells were treated with 2 mol/l rapamycin for 1 hour and replaced with fresh medium without rapamycin for 11 hours. The treatment cycle was repeated three times. (c) Modified cells were treated with 1 mol/l rapamycin for 1 hour and replaced with fresh medium without rapamycin for 6 hours. The treatment cycle was repeated three times. The concentration of C-peptide was measured at the indicated time-points using enzyme-linked immunosorbent assay. The values are mean SD from a representative experiment (n = 3).

Full figure and legend (32K)

Repeated exposure to rapamycin also induced insulin release in engineered keratinocytes treated for 1 hour every 12 hours. The cells secreted most of the C-peptide within the first 2 hours of each cycle, with approximately half of the maximal amounts being secreted during the third hour (Figure 3b). Small amounts of C-peptide were secreted between 3 and 12 hours, thereby suggesting that insulin release returns to baseline value within 2–3 hours after the rapamycin is withdrawn (Figure 3b). Engineered keratinocytes were also able to respond to more frequent exposure to rapamycin every 6 hours (Figure 3c). Insulin was released within 30 minutes after addition of rapamycin, and the level increased 34-fold and 27-fold higher as compared to basal values at 1 and 2 hours, respectively. Thereafter the rate of release decreased, reaching near-basal levels after 3–6 hours. The kinetic profile and the total amount of C-peptide released between 30 minutes and 3 hours were very similar after subsequent additions of rapamycin (7.4 1.2, 7.7 0.4, and 6.7 0.8 pmol/million cells/day), thereby suggesting that epidermal cells are capable of maintaining insulin levels for multiple release cycles.

Insulin-expressing keratinocytes displayed a normal pattern of differentiation and stratification

In order to determine whether genetically modified keratinocytes expressing 3F_M-hppI4-HD retain the ability to stratify into three-dimensional skin substitutes, gene-modified cells were cultured on acellular dermis that was subsequently raised to the air–liquid interface to induce multilayer stratification. As in the case of control cells expressing only GFP, 3F_M-hppI4-HD-expressing cells developed into fully stratified epidermis consisting of basal, suprabasal, granular, and cornified layers (Figure 4a). GFP expression was observed throughout the epidermis with higher levels in the stratum corneum (Figure 4b), possibly because of accumulation of corneocytes and lack of desquamation in vitro. In addition, expression of the differentiation markers keratin 10 (Figure 4c) and involucrin (Figure 4d) were confined to suprabasal and granular layers, respectively, indicating a normal pattern of epidermal differentiation.

Figure 4.

Modified keratinocytes retain the ability to differentiate and stratify into three-dimensional bioengineered epidermis. Gene-modified human primary keratinocytes were cultured on 3T3/J2 feeder layers and then seeded onto acellular dermis (5 10⁵ cells/cm²) and raised to the air–liquid interface to induce stratification. On day 7, tissues were fixed and embedded in paraffin. Tissue sections were stained with (a) hematoxylin and eosin. Fluorescence images of tissue sections show (b) green fluorescent protein expression, and immunostaining for (c) human K10 (red) or (d) human involucrin (red). (b–d) Cell nuclei were counterstained with Hoechst (blue) (bar = 20 m).

Full figure and legend (68K)

Regulated insulin secretion from bioengineered skin substitutes

As was seen with keratinocytes in culture, skin substitutes too secreted insulin in a rapamycin concentration–dependent manner, reaching maximum secretion in the concentration range between 10 and 30 mol/l (Figure 5a). In the presence of 20 mol/l rapamycin, skin-substitute C-peptide was detected in the medium within 30 minutes to 1 hour after rapamycin administration (Figure 5b). Thereafter, C-peptide was secreted at a constant rate of 0.9 pmol/cm² tissue/hour for 6–7 hours until the stored protein was depleted and the release rate reached a plateau (Figure 5b).

Figure 5.

Regulated insulin secretion from bioengineered epidermis. (a) Gene-modified keratinocytes were seeded on acellular human dermis and raised to air–liquid interface to allow stratification. On day 3, the tissues were treated either with vehicle only or with the indicated concentrations of rapamycin, and the medium was harvested 5 hours later. (b) Insulin secretion kinetics from three-dimensional bioengineered epidermis. On day 3, the tissues were treated at the air–liquid interface with 20 mol/l rapamycin, and conditioned medium was harvested at the indicated time-points. (c) On day 3, the tissues were treated with 5 mol/l rapamycin for 1 hour at the air–liquid interface, washed three times with phosphate-buffered saline, and replaced with fresh medium without rapamycin for 11 hours. The treatment cycle was repeated three times. Conditioned medium was harvested at the indicated times and C-peptide levels were measured using enzyme-linked immunosorbent assay. The values are mean SD from a representative experiment (n = 3).

Full figure and legend (28K)

Skin substitute insulin secretion responses were also investigated after treatment with rapamycin three times at 1-hour intervals. After each exposure to rapamycin, skin substitute C-peptide was secreted mostly within the first 3 hours (Figure 5c), as was seen with keratinocytes in culture. The total C-peptide amounts released during the 3 hours after each successive rapamycin addition were 13.6 0.3, 18.0 1.6, and 17.2 1.8 pmol/cm² tissue/day, respectively. These C-peptide levels were not significantly different from each other, thereby suggesting that skin substitutes are capable of maintaining insulin levels for multiple release responses during a 36-hour period.

Rapamycin-induced insulin secretion reversed hyperglycemia in diabetic nu/nu mice implanted with 3F_M-hppI4-HD keratinocytes

In order to evaluate the ability of modified keratinocytes to regulate glucose levels in vivo, we transplanted 3F_M-hppI4-HD cells into athymic nu/nu mice. Nu/nu mice were injected in the back with 3F_M-hppI4-HD cells (40 10⁶ cells/animal) in a fibrin hydrogel to confine the cells subcutaneously.²⁴ The hydrogels contained keratinocyte growth factor and fibronectin to protect epithelial cells from cytotoxic insults and to enhance cell survival.^25,26

At 10–14 days after the implantation, athymic mice were injected with STZ and rendered diabetic (blood glucose >250 mg/dl).In order to determine whether insulin could be released in response to exogenous ligand administration, the diabetic animals were treated with rapamycin through intraperitoneal injection, and blood was collected at 1- or 2-hour intervals for insulin and glucose measurements. Although the concentrations of plasma insulin varied among the mice, ranging between 1 and 20 pmol/l, insulin was generally detectable within 1 hour and reverted to background levels within 2–4 hours after rapamycin administration (Figure 6a; n = 8). In contrast, no insulin was detected in the blood of mice implanted with cells expressing GFP only, or in 3F_M-hppI4-HD mice treated with vehicle only, without rapamycin.

Figure 6.

Insulin secretion and reversal of hyperglycemia in diabetic athymic nu/nu mice implanted with modified keratinocytes. Diabetic athymic mice with genetically modified cells (40 10⁶ cells) implanted subcutaneously were made to fast for 3 hours and then injected with rapamycin (30 mg/kg body weight). (a) Blood samples were collected at the indicated time-points after the administration of rapamycin, and plasma insulin was measured using enzyme-linked immunosorbent assay. The concentration of insulin at each time-point (I_t) was normalized to the maximum insulin level (I _max) detected in each animal. ^*P < 0.005 for the value at a given time-point versus the value at time-point zero. (b) Blood glucose levels were measured in diabetic athymic mice with genetically modified implanted cells expressing green fluorescent protein (GFP) only (control; n = 4) or 3F_M-hppI4-HD cells (n = 8) at the indicated time-points (G_t), and normalized to the glucose level at time-point zero (G ₀) (the variation across animals ranged from 250 to >600 mg/dl). Rapamycin administration was repeated once a week for three consecutive weeks. ^*P < 0.005 for the value at the indicated time-points versus the value at time-point zero. At 6–8 weeks after the implantation, skin tissue was excised from each animal and processed for histology. Paraffin-embedded tissue sections were stained with (c) hematoxylin and eosin. Fluorescence images of tissue sections show (d) GFP expression, or (e) immunostaining for human K10 (red). (d,e) Cell nuclei were counterstained with Hoechst (blue) (bar = 20 m).

Full figure and legend (69K)

Insulin secretion coincided with changes in blood glucose levels. Blood glucose concentrations of diabetic athymic mice transplanted with 3F_M-hppI4-HD keratinocytes decreased significantly to euglycemic levels (80–150 mg/dl) within 2–4 hours after rapamycin, even in the animals with blood glucose levels as high as 600 mg/dl (Figure 6b; Table 1). In contrast, blood glucose levels in control STZ-treated mice with GFP-expressing keratinocytes (n = 4) were not affected by rapamycin administration. Similar results were obtained after a second (n = 5) and third (n = 2) administration of rapamycin (Figure 6b), thereby indicating that implanted cells could respond to the ligand in a repeatable manner. Because clearance from the body could take 4–5 days,²⁷ rapamycin was administered only once a week so as to ensure complete clearance of the ligand between consecutive treatments. During that time, several animals lost >20% of their body weight because of severe hyperglycemia and had to be killed (in accordance to the Institutional Animal Care and Use Committee protocol). As a result, the animals available for the second and third applications of rapamycin were fewer in number.

Table 1 - Glucose levels in the plasma of streptozotocin-treated mice right before each cycle of rapamycin administration.

Full table (19K)

Six to eight weeks after the implantation, the animals were killed and the skin tissues containing the implanted cells, when excised and evaluated histologically, showed normal morphology (Figure 6c). Fluorescence microscopy showed that the implanted cells appeared green because of GFP expression (Figure 6d), thereby indicating sustained expression of the transgene in vivo. In addition, immunostaining for human keratin 10 verified that the cells were human keratinocytes (Figure 6e).

Discussion

Diabetes mellitus affects millions of people, and the search for alternative means of producing and releasing insulin in human cells is fueled by the desire to achieve euglycemic control to the maximum extent possible. The continuous delivery of baseline levels of insulin by genetically modified cells has been shown to have beneficial effects on glucose levels.^7,28,29,30 Moreover, intramuscular³¹ and liver³² injections of lentivirus-insulin in STZ-treated rats have been shown to decrease hyperglycemia and increase survival in diabetic animals. Although these studies suggested that lentiviral gene transfer can be used in the treatment of type 1 diabetes, they were not designed to provide a method for regulated insulin delivery.

In this study, regulated insulin delivery is achieved using epidermal keratinocytes and bioengineered skin secreting insulin in response to an exogenous ligand, and is shown to reverse hyperglycemia in severely diabetic mice. The lentiviral insulin constructs efficiently transduced keratinocytes, which expressed insulin constitutively under a cytomegalovirus promoter, as evidenced by GFP co-expression. Rapamycin at low concentrations rapidly reduced stored insulin levels in the transduced keratinocytes, with the marker of proinsulin cleavage, C-peptide, as well as the fusion protein epitope 3F_M, appearing in the extracellular media. The risk of hypoglycemia in this approach is low, given that the concentration of insulin in the serum of animals was measured as being between 1 and 20 pmol/l (or 0.16–3.3 mU/l), which is low compared to the level of venous plasma insulin observed after a carbohydrate meal (100 mU/l). Interestingly, the tight regulation imposed by the use of 3F_M domains prevents constitutive secretion of insulin (Supplementary Figure S1), thereby minimizing the risk of hypoglycemia without compromising the rate of insulin secretion (Supplementary Figure S2). Although active rapamycin was utilized in these studies for proof-of-concept, it is anticipated that future studies will employ activity-deficient rapamycin analogs as they become available.

One concern was the potential cellular toxicity because of intracellular protein aggregation. Cell stress was not apparent as evidenced by a series of observations. First, the proliferation of genetically modified cells was not significantly affected. Second, the percentage of GFP-positive cells did not change with time (data not shown), thereby indicating that transduced and nontransduced cells proliferated at the same rate. Third, the program of keratinocyte differentiation was unchanged, with insulin-expressing keratinocytes developing into a normal, fully stratified epidermis. Four, after transplantation into nu/nu mice, keratinocytes secreted insulin and regulated glucose levels for 6–8 weeks, thereby indicating that cytotoxicity did not become apparent during that period.

Taking the perspective that bioengineered skin cells may have direct application in diabetes management in the future, it is useful to compare the current pharmacological insulin replacement therapies with the engineered keratinocyte model. Regular insulin injection kinetics show that the time of onset is 30 minutes, and peak activity lasts from 1.5 to 4 hours, with an approximate duration of action of 5–8 hours in humans. Even the new rapidly acting aspart or lispro insulins have a time of onset of 10–20 minutes, and a shortened duration of action of 2–5 hours.³³ Insulin release from engineered keratinocytes compares favorably with injected insulin kinetics, with rapamycin-induced insulin release beginning within 30 minutes and extending to 3 hours at reduced rates. Similarly, skin substitutes showed rapamycin-induced C-peptide release within 30 minutes to 1 hour, with peak insulin levels at 1–2 hours, and continued secretion for several hours. Most important, the in vivo mouse studies demonstrated that rapamycin elicited insulin release from modified injected keratinocytes within 1 hour, with peak insulin levels occurring at 1–2 hours, while circulating insulin levels decreased to background levels within 4 hours after stimulation. These data illustrate that insulin-engineered skin cells can be regulated to produce and release bioactive insulin only after administration of the ligand and with kinetics closely approximating those achieved with direct injection of rapid-acting and short-duration hormone preparations. Moreover, insulin release from transplanted gene modified 3F_M-hppI4-HD keratinocytes resulted in significant rapid reduction of plasma glucose levels in hyperglycemic mice.

Our results suggest that skin may be an efficient ectopic site for insulin delivery. Based on the average level of insulin production by skin substitutes after rapamycin administration (56 pmol/cm² of tissue/day), the stability of C-peptide in serum (half-life = 174 hours³⁴) and the clearance of C-peptide in blood (half-life =60 minutes³⁵), the area of skin tissue required for producing insulin at equal molar amounts as from transplanted islets (1.0 ng/ml of serum^36,37) is 190 cm². This estimated area of skin graft is <1% of the total skin surface area of the average adult (2 m²), thereby suggesting that bioengineered skin may provide an efficient vehicle for insulin release. Moreover, skin substitutes secreted insulin in response to repeated stimulation and showed they were capable of producing and releasing insulin over prolonged periods according to need for controlling blood glucose.

A possible route of insulin delivery may involve transplantation of bioengineered skin prepared from genetically modified autologous keratinocytes. Alternatively, direct administration of lentivirus may provide an easier route of delivery, and this has been used effectively for correcting genetic skin diseases, e.g., epidermolysis bullosa.^38,39,40 Interestingly, novel techniques are being developed, e.g., painless microneedles,^41,42,43 that may achieve in vivo gene delivery to large areas of the skin in a controlled (control of depth and area of delivery) and painless manner. Insulin could be released by topical administration of the inducer agent according to requirement, thereby decreasing the effective dose and minimizing potential systemic effects. The potential for reducing the frequency of injections of the vector, and the possibility of topical application of the inducer appear very attractive, because this approach can minimize the number of injections, ultimately increasing patient comfort and compliance. However, the potential for oncogene activation at lentiviral vector integration sites remains a serious concern. This concern may be somewhat alleviated by the possibility of surgical skin tissue removal should adverse effects occur. The use of vectors that do not integrate (e.g., adenovirus) but can provide transgene expression for several weeks, or of lentiviral vectors that are designed to integrate at specific sites in the human genome,^44,45 may provide safer alternatives.

Materials and Methods

Vector construction and virus preparation. The third-generation lentiviral system has been described elsewhere.²¹ The segment FCS-hppI4HD-IRES-GFP was amplified from vector pQC(hppI4HD)IG¹⁰ by PCR using primers 1+2 (Table 2). Subsequently, the amplified segment was used for substituting GFP in pCSCG lentiviral vector by digestion with AgeI–XhoI cloning sites, yielding vector pCSC-FCS-hppI4HD-IG. Segments containing the mutant form of FK506 binding protein 12 (F_M) with and without 5'-untranslated region were amplified from vector pC4S1-FM4-FCS-hGH (Ariad, Cambridge, MA) by PCR, using primers 3+5 and 4+5, respectively. Because the vector pC4S1-FM4-FCS-hGH contained four copies of F_M and the reverse primer could bind to 3'-end of any of the F_M domains, multiple PCR products containing one to four F_M domains were generated (5'-UTR-nF_M and nF_M, where n = 1–4). Subsequently, PCR products containing two to four F_M domains were gel-purified and cloned into pCSC-FCS-hppI4HD-IG at the NheI and AgeI sites to form pCSC-5'-UTR-nF_M-FCS-HD-IG and pCSC-nF_M-FCS-HD-IG, respectively (n = 2, 3, or 4). All the PCRs were carried out with High Fidelity PLUS polymerase (Roche, Indianapolis, IN) in accordance with the manufacturer's protocol. The sequences of all the cloned genes were confirmed by sequencing with ABI PRISM 3130XL genetic analyzers (Applied Biosystems, Foster City, CA).

Table 2 - PCR conditions and primers. Restriction enzyme sites are underlined and the furin cleavage site (FCS) is shown in bold.

Full table (27K)

For the preparation of lentivirus, 15 g each of the constructed plasmids described earlier, together with 5 g pMDL-g/p, 3 g pSRV-rev, and 1.5 g pMD2.G were transiently co-transfected into 293T/17 cells (American Type Culture Collection, Manassas, VA) in a 75-cm² tissue culture flask. Transfection was performed in accordance with standard calcium–phosphate precipitation protocol⁴⁶ when cells reached 70–80% confluence. After 24 hours, the transfected cells were washed with phosphate-buffered saline (PBS), and the culture medium was replaced with medium containing 5 mmol/l sodium butyrate (Aldrich, St. Louis, MO). The viruses were harvested every 24 hours, and two batches of viruses were collected from each transfection. The viruses were filtered through 0.45-m filter (Millipore, Bedford, MA) and immediately concentrated by ultracentrifugation at 50,000g and 4 °C for 2 hours. The pellet was resuspended in keratinocytes-SFM (Gibco BRL, Grand Island, NY) to yield purified and concentrated virus stock (25).

Cell culture and lentivirus transduction. Cell culture, preparation of skin equivalents, virus transduction, and evaluation of transduction efficiency by flow cytometry were carried out as described earlier.^22,24

Rapamycin treatment. For the in vitro experiments, rapamycin (Tecoland, Edison, NJ) was dissolved in ethanol (Pharmco, Shelbyville, KY) to yield a stock concentration of 10 mmol/l. The stock solution was diluted in cell culture medium to the indicated concentrations and used for inducing insulin secretion by genetically modified keratinocytes. For the in vivo experiments, rapamycin was first dissolved in N,N- dimethylacetamide (TCI America, Portland, OR) to yield a stock concentration of 0.208 mg/ml. The stock solution was then diluted to the indicated concentrations in a mixture of 4% (vol/vol) N,N- dimethylacetamide, 10% (vol/vol) polyethylene glycol (average molecular weight 400), and 17% (vol/vol) polyoxyethylene sorbitan monooleate (Mallinckrodt Baker, Phillipsburg, NJ).⁴⁷

C-peptide enzyme-linked immunosorbent assay. Conditioned media from samples were harvested and stored at –75 °C until use. The amount of C-peptide in each sample was measured using an ELISA kit for human C-peptide (human C-peptide ELISA kit; Millipore, Billerica, MA).

Western blot and immunofluorescence. The detailed protocol for western blot has been described earlier.⁴⁸ The antibodies used were: guinea pig anti-human insulin [1:500 in blocking solution (5% milk in Tris-buffered saline); Abcam, Cambridge, MA], rabbit anti-FKBP12 (1:3,000 for cell lysates, 1:1,000 for culture medium, both in blocking solution; Affinity BioReagents, Golden, CO), and mouse anti-human -actin (1:5,000 in blocking solution; Sigma, St. Louis, MO). After overnight incubation at 4 °C, horseradish peroxidase–conjugated secondary antibodies including donkey anti-guinea pig immunoglobulin G (IgG) (H+L) (0.16 ng/ml in blocking solution; Jackson ImmunoResearch, West Grove, PA), goat anti-rabbit IgG (1:2,000 in blocking solution; Cell Signaling Technology, Danvers, MA), or goat anti-mouse IgG (1:2,000 in blocking solution; Cell Signaling Technology, Danvers, MA) were added and incubated at room temperature (RT) for 1 hour. Protein bands were detected by chemiluminescence (LumiGLO; KLP, Gaithersburg, MD) and exposed on film.

For immunofluorescence, cells were cultured in Dulbecco's modified Eagle's medium supplemented with 5% insulin-free dialyzed fetal bovine serum (Gibco, Grand Island, NY) for 24 hours, washed five times with PBS, and fixed with acetone for 5 minutes at –20 °C. After being fixed, the cells were washed with ice-cold PBS twice and blocked with 10% normal goat serum (Vector Laboratories, Burlingame, CA) + 1% bovine serum albumin (EMD Chemicals, San Diego, CA) in PBS for 1 hour at RT. Thereafter, the cells were incubated with guinea pig anti-human insulin (1:100 in 1% bovine serum albumin) overnight at 4 °C. Alexa 594–conjugated goat anti-guinea pig IgG (4 g/ml in 1% bovine serum albumin, Molecular Probes, Grand Island, NY) was applied after primary antibody incubation (1 hour, RT). The cell nuclei were stained with Hoechst 33258 (25 g/ml in TNE buffer: 10 mmol/l Tris, 2 mol/l NaCl, 1 mmol/l EDTA, pH 7.4; Molecular Probes, Grand Island, NY) for 10 minutes at RT. Fluorescent images were obtained using an inverted fluorescence microscope (Nikon Diaphot, Nikon).

Histology and immunohistochemistry. Skin tissues were embedded in paraffin and stained with hematoxylin and eosin as described earlier.²² For immunohistochemical analysis, paraffin-embedded tissue sections were de-paraffinized and subjected to antigen retrieval by trypsin digestion (type II-porcine pancreas; Sigma, St. Louis, MO). Briefly, a solution containing 100 g/ml trypsin and 100 g/ml CaCl₂ in ddH₂O (pH 7.8) was heated to 37 °C and used for treating tissue sections for 30 minutes at RT. After being washed twice in water, the tissue sections were blocked in PBS/10% normal goat serum for 1 hour at RT and primary antibodies were added for 1 hour at RT. The primary antibodies used were mouse anti-human cytokeratin 10 (1:100 in blocking solution; Novocastra, MA), and mouse anti-human involucrin (1:8,000 in blocking solution; Sigma, St. Louis, MO). After four washes with PBS, Alexa 594–conjugated goat anti-mouse IgG (4 g/ml in blocking solution; Molecular Probes) was added for 30 minutes at RT. After four washes with PBS, the nuclei were stained with Hoechst 33258 (25 g/ml) for 2–5 minutes and then washed twice with PBS. Fluorescent images were obtained using an inverted fluorescence microscope and analyzed using ImageJ (version 1.37; National Institutes of Health, Bethesda, MD).

Transplantation of the gene-modified keratinocytes into diabetic mice. Human epidermal keratinocytes (100 million cells/ml) were resuspended in 400 l PBS containing fibronectin (22 g/ml; EMD Chemicals, San Diego, CA), keratinocyte growth factor (56 g/ml; Amgen, Thousand Oaks, CA), and thrombin (2.5 U/ml; Sigma, St. Louis, MO). The cell-containing solution was mixed with 100 l of fibrinogen (12.5 mg/ml; Enzyme Research Laboratories, South Bend, IN) and injected subcutaneously into the backs of athymic nu/nu mice (Harlan Sprague Dawley, Indianapolis, IN) where it gelled within 5–10 seconds. Ten to fourteen days after implantation, the mice were rendered diabetic by treatment with STZ (250 mg/kg body weight; Calbiochem, San Diego, CA). Thereafter, the body weight of each animal was monitored twice a week. The mice that lost >15–20% of body weight were killed, in line with the Institutional Animal Care and Use Committee protocol. When mice became diabetic (glucose level >250 mg/dl), they were made to fast for 3 hours and then injected intraperitoneally with 100 l rapamycin (30 mg/kg body weight). Blood samples were collected using the cheek pouch or submandibular bleeding methods at the indicated time-points, and glucose levels were measured using a glucose meter (One Touch Ultra Smart; Lifescan, Milpitas, CA). The concentration of insulin was measured in blood plasma using an Ultra Sensitive Insulin ELISA (Mercodia, Winston Salem, NC).

Statistical analysis. Statistical analysis of the data was performed using a two-tailed Student's t -test ( = 0.05) using Microsoft Excel (Microsoft, Redwood, CA).

References

REFERENCES

1. Chernajovsky, Y, Dreja, H, Daly, G, Annenkov, A, Gould, D, Adams, G et al. (2000). Immuno- and genetic therapy in autoimmune diseases. Genes Immun 1: 295–307. | Article | PubMed | ChemPort |
2. Bonner-Weir, S and Sharma, A (2002). Pancreatic stem cells. J Pathol 197: 519–526. | Article | PubMed | ISI |
3. Zalzman, M, Gupta, S, Giri, RK, Berkovich, I, Sappal, BS, Karnieli, O et al. (2003). Reversal of hyperglycemia in mice by using human expandable insulin-producing cells differentiated from fetal liver progenitor cells. Proc Natl Acad Sci USA 100: 7253–7258. | Article | PubMed | ChemPort |
4. Ferber, S, Halkin, A, Cohen, H, Ber, I, Einav, Y, Goldberg, I et al. (2000). Pancreatic and duodenal homeobox gene 1 induces expression of insulin genes in liver and ameliorates streptozotocin-induced hyperglycemia. Nat Med 6: 568–572. | Article | PubMed | ISI | ChemPort |
5. Kojima, H, Fujimiya, M, Matsumura, K, Younan, P, Imaeda, H, Maeda, M et al. (2003). NeuroD-betacellulin gene therapy induces islet neogenesis in the liver and reverses diabetes in mice. Nat Med 9: 596–603. | Article | PubMed | ISI | ChemPort |
6. Cheung, AT, Dayanandan, B, Lewis, JT, Korbutt, GS, Rajotte, RV, Bryer-Ash, M et al. (2000). Glucose-dependent insulin release from genetically engineered K cells. Science 290: 1959–1962. | Article | PubMed | ISI | ChemPort |
7. Kolodka, TM, Finegold, M, Moss, L and Woo, SL (1995). Gene therapy for diabetes mellitus in rats by hepatic expression of insulin. Proc Natl Acad Sci USA 92: 3293–3297. | Article | PubMed | ChemPort |
8. Falqui, L, Martinenghi, S, Severini, GM, Corbella, P, Taglietti, MV, Arcelloni, C et al. (1999). Reversal of diabetes in mice by implantation of human fibroblasts genetically engineered to release mature human insulin. Hum Gene Ther 10: 1753–1762. | Article | PubMed | ChemPort |
9. Shaw, JA, Delday, MI, Hart, AW, Docherty, HM, Maltin, CA and Docherty, K (2002). Secretion of bioactive human insulin following plasmid-mediated gene transfer to non-neuroendocrine cell lines, primary cultures and rat skeletal muscle in vivo. J Endocrinol 172: 653–672. | Article | PubMed | ChemPort |
10. Lei, P, Ogunade, A, Kirkwood, KL, Laychock, SG and Andreadis, ST (2007). Efficient production of bioactive insulin from human epidermal keratinocytes and tissue-engineered skin substitutes: implications for treatment of diabetes. Tissue Eng 13: 2119–2131. | Article | PubMed | ChemPort |
11. Wu, L, Nicholson, W, Wu, CY, Xu, M, McGaha, A, Shiota, M et al. (2003). Engineering physiologically regulated insulin secretion in non-beta cells by expressing glucagon-like peptide 1 receptor. Gene Ther 10: 1712–1720. | Article | PubMed | ChemPort |
12. Thulé, PM, Liu, J and Phillips, LS (2000). Glucose regulated production of human insulin in rat hepatocytes. Gene Ther 7: 205–214. | Article | PubMed | ISI | ChemPort |
13. Yang, YW and Kotin, RM (2000). Glucose-responsive gene delivery in pancreatic islet cells via recombinant adeno-associated viral vectors. Pharm Res 17: 1056–1061. | Article | PubMed | ChemPort |
14. Chen, R, Meseck, ML and Woo, SL (2001). Auto-regulated hepatic insulin gene expression in type 1 diabetic rats. Mol Ther 3: 584–590. | Article | PubMed | ISI | ChemPort |
15. Alam, T and Sollinger, HW (2002). Glucose-regulated insulin production in hepatocytes. Transplantation 74: 1781–1787. | Article | PubMed | ISI | ChemPort |
16. Olson, DE, Paveglio, SA, Huey, PU, Porter, MH and Thulé, PM (2003). Glucose-responsive hepatic insulin gene therapy of spontaneously diabetic BB/Wor rats. Hum Gene Ther 14: 1401–1413. | Article | PubMed | ChemPort |
17. Cheng, SY, Constantinidis, I and Sambanis, A (2006). Use of glucose-responsive material to regulate insulin release from constitutively secreting cells. Biotechnol Bioeng 93: 1079–1088. | Article | PubMed | ChemPort |
18. Rivera, VM, Wang, X, Wardwell, S, Courage, NL, Volchuk, A, Keenan, T et al. (2000). Regulation of protein secretion through controlled aggregation in the endoplasmic reticulum. Science 287: 826–830. | Article | PubMed | ISI | ChemPort |
19. Burke, GT, Hu, SQ, Ohta, N, Schwartz, GP, Zong, L and Katsoyannis, PG (1990). Superactive insulins. Biochem Biophys Res Commun 173: 982–987. | Article | PubMed | ChemPort |
20. Schwartz, GP, Burke, GT and Katsoyannis, PG (1987). A superactive insulin: [B10-aspartic acid]insulin(human). Proc Natl Acad Sci USA 84: 6408–6411. | Article | PubMed | ChemPort |
21. Miyoshi, H, Blömer, U, Takahashi, M, Gage, FH and Verma, IM (1998). Development of a self-inactivating lentivirus vector. J Virol 72: 8150–8157. | PubMed | ISI | ChemPort |
22. Bajaj, BG, Lei, P and Andreadis, ST (2005). Efficient gene transfer to human epidermal keratinocytes on fibronectin: in vitro evidence for transduction of epidermal stem cells. Mol Ther 11: 969–979. | Article | PubMed | ChemPort |
23. Kumar, M, Keller, B, Makalou, N and Sutton, RE (2001). Systematic determination of the packaging limit of lentiviral vectors. Hum Gene Ther 12: 1893–1905. | Article | PubMed | ChemPort |
24. Geer, DJ, Swartz, DD and Andreadis, ST (2005). Biomimetic delivery of keratinocyte growth factor upon cellular demand for accelerated wound healing in vitro and in vivo. Am J Pathol 167: 1575–1586. | PubMed | ChemPort |
25. Gomperts, BN, Belperio, JA, Fishbein, MC, Keane, MP, Burdick, MD and Strieter, RM (2007). Keratinocyte growth factor improves repair in the injured tracheal epithelium. Am J Respir Cell Mol Biol 37: 48–56. | Article | PubMed | ChemPort |
26. Ray, P (2005). Protection of epithelial cells by keratinocyte growth factor signaling. Proc Am Thorac Soc 2: 221–225. | Article | PubMed | ChemPort |
27. Palaparthy, R, Pradhan, R, Chan, J, Wang, Q, Ji, Q, Achari, R et al. (2005). Pharmacokinetics and safety of ABT-578, a sirolimus (rapamycin) analogue, after single intravenous bolus injection in healthy male volunteers. Clin Drug Investig 25: 491–498. | Article | PubMed | ChemPort |
28. Valera, A, Fillat, C, Costa, C, Sabater, J, Visa, J, Pujol, A et al. (1994). Regulated expression of human insulin in the liver of transgenic mice corrects diabetic alterations. FASEB J 8: 440–447. | PubMed | ChemPort |
29. Docherty, K (1997). Gene therapy for diabetes mellitus. Clin Sci (Lond) 92: 321–330. | PubMed | ChemPort |
30. Bailey, CJ, Davies, EL and Docherty, K (1999). Prospects for insulin delivery by ex-vivo somatic cell gene therapy. J Mol Med 77: 244–249. | Article | PubMed | ChemPort |
31. Oh, TK, Li, MZ and Kim, ST (2006). Gene therapy for diabetes mellitus in rats by intramuscular injection of lentivirus containing insulin gene. Diabetes Res Clin Pract 71: 233–240. | Article | PubMed | ChemPort |
32. Ren, B, O'Brien, BA, Swan, MA, Koina, ME, Nassif, N, Wei, MQ et al. (2007). Long-term correction of diabetes in rats after lentiviral hepatic insulin gene therapy. Diabetologia 50: 1910–1920. | Article | PubMed | ChemPort |
33. Davis, SN and Granner, DK (eds) (2001). Insulin, Oral Hypoglycemic Agents, and the Pharmacology of the Endocrine Pancreas. McGraw-Hill: New York.
34. Melles, E, Bergman, T, Alvelius, G, Jonsson, A, Ekberg, K, Wahren, J et al. (2003). Proinsulin C-peptide and its C-terminal pentapeptide: degradation in human serum and Schiff base formation with subsequent CO₂ incorporation. Cell Mol Life Sci 60: 1019–1025. | Article | PubMed | ChemPort |
35. Fredriksson, A, Ekberg, K, Ingvar, M, Johansson, BL, Wahren, J and Stone-Elander, S (2002). In vivo biodistribution and pharmacokinetics of ¹⁸F-labeled human C-peptide: evaluation in monkeys using positron emission tomography. Life Sci 71: 1361–1370. | Article | PubMed | ChemPort |
36. Fiorina, P, Folli, F, Bertuzzi, F, Maffi, P, Finzi, G, Venturini, M et al. (2003). Long-term beneficial effect of islet transplantation on diabetic macro-/microangiopathy in type 1 diabetic kidney-transplanted patients. Diabetes Care 26: 1129–1136. | Article | PubMed |
37. Fiorina, P, Folli, F, Maffi, P, Placidi, C, Venturini, M, Finzi, G et al. (2003). Islet transplantation improves vascular diabetic complications in patients with diabetes who underwent kidney transplantation: a comparison between kidney-pancreas and kidney-alone transplantation. Transplantation 75: 1296–1301. | Article | PubMed |
38. Chen, M, Kasahara, N, Keene, DR, Chan, L, Hoeffler, WK, Finlay, D et al. (2002). Restoration of type VII collagen expression and function in dystrophic epidermolysis bullosa. Nat Genet 32: 670–675. | Article | PubMed | ISI | ChemPort |
39. Woodley, DT, Keene, DR, Atha, T, Huang, Y, Ram, R, Kasahara, N et al. (2004). Intradermal injection of lentiviral vectors corrects regenerated human dystrophic epidermolysis bullosa skin tissue in vivo. Mol Ther 10: 318–326. | Article | PubMed | ISI | ChemPort |
40. Siprashvili, Z and Khavari, PA (2004). Lentivectors for regulated and reversible cutaneous gene delivery. Mol Ther 9: 93–100. | Article | PubMed | ISI | ChemPort |
41. McAllister, DV, Allen, MG and Prausnitz, MR (2000). Microfabricated microneedles for gene and drug delivery. Annu Rev Biomed Eng 2: 289–313. | Article | PubMed | ISI | ChemPort |
42. Prausnitz, MR (2001). Overcoming skin's barrier: the search for effective and user-friendly drug delivery. Diabetes Technol Ther 3: 233–236. | Article | PubMed | ChemPort |
43. Kaushik, S, Hord, AH, Denson, DD, McAllister, DV, Smitra, S, Allen, MG et al. (2001). Lack of pain associated with microfabricated microneedles. Anesth Analg 92: 502–504. | Article | PubMed | ISI | ChemPort |
44. Yant, SR, Huang, Y, Akache, B and Kay, MA (2007). Site-directed transposon integration in human cells. Nucleic Acids Res 35: e50. | Article | PubMed | ChemPort |
45. Moehle, EA, Rock, JM, Lee, YL, Jouvenot, Y, DeKelver, RC, Gregory, PD et al. (2007). Targeted gene addition into a specified location in the human genome using designed zinc finger nucleases. Proc Natl Acad Sci USA 104: 3055–3060. | Article | PubMed | ChemPort |
46. Tiscornia, G, Singer, O and Verma, IM (2006). Production and purification of lentiviral vectors. Nat Protoc 1: 241–245. | Article | PubMed | ChemPort |
47. Rivera, VM, Ye, X, Courage, NL, Sachar, J, Cerasoli, F Jr., Wilson, JM et al. (1999). Long-term regulated expression of growth hormone in mice after intramuscular gene transfer. Proc Natl Acad Sci USA 96: 8657–8662. | Article | PubMed | ChemPort |
48. Koria, P and Andreadis, ST (2007). KGF promotes integrin alpha5 expression through CCAAT/enhancer-binding protein-beta. Am J Physiol Cell Physiol 293: C1020–C1031. | Article | PubMed | ChemPort |

Acknowledgments

This work was supported by a grant from the National Institutes of Health R01 DK068699-01 (to S.T.A) and DK068700-02 (to S.G.L.) and a grant from the Integrative Research and Creative Activities Fund of the University of Buffalo.