Gene Therapy (2005) 12, 1752–1760 & 2005 Nature Publishing Group All rights reserved 0969-7128/05 $30.00 www.nature.com/gt RESEARCH ARTICLE Gene delivery to human sweat glands: a model for cystic fibrosis gene therapy

H Lee1,2, DR Koehler1,3, CY Pang1,4,5, RH Levine5,PNg6, DJ Palmer6, PM Quinton7 and J Hu1,2,3 1Research Institute, The Hospital for Sick Children, Toronto, Canada; 2Institute of Medical Science, University of Toronto, Toronto, Canada; 3Department of Laboratory Medicine and Pathology, University of Toronto, Toronto, Canada; 4Department of Physiology, University of Toronto, Toronto, Canada; 5Department of Surgery, University of Toronto, Toronto, Canada; 6Department of Molecular and Human Genetics, Baylor College of Medicine, Houston, TX, USA; and 7Department of Pediatrics, UCSD School of Medicine University of California, San Diego, La Jolla, CA, USA

Gene therapy vectors are mostly studied in cultured cells, application, intradermal injection or submerged culture. We rodents, and sometimes in non-human primates, but it is found that transduction efficiency can be enhanced by useful to test them in human tissue prior to clinical trials. pretreating isolated sweat glands with dispase, which In this study, we investigated the possibility of using human suggests that the basement membrane is a critical barrier sweat glands as a model for testing cystic fibrosis (CF) gene to gene delivery by adenoviral vectors. Using this approach, therapy vectors. Human sweat glands are relatively easy to we showed that Cftr could be efficiently delivered to and obtain from skin biopsy, and can be tested for CFTR function. expressed by the epithelial cells of sweat glands with our Using patients’ sweat glands could provide a safe model to helper-dependent adenoviral vector containing cytokeratin study the efficacy of CF gene therapy. As the first step to 18 regulatory elements. Based on this study we propose that explore using sweat glands as a model for CF gene therapy, sweat glands might be used as an alternative model to study we examined various ex vivo gene delivery methods for a CF gene therapy in humans. helper-dependent adenovirus (HD-Ad) vector. Gene delivery Gene Therapy (2005) 12, 1752–1760. doi:10.1038/ to sweat glands in skin organ culture was studied by topical sj.gt.3302587; published online 21 July 2005

Keywords: gene delivery; Cftr; sweat glands; native tissue; helper-dependent adenovirus vector; basement membrane

Introduction Cftr mRNA to 5–10% could rescue the CF phenotype,5,6 although a recent review proposed that a higher level of Cystic fibrosis (CF), the most common genetic disorder in expression may be required at the protein level.7 In the Caucasian populations, is caused by recessive mutations last decade, adenovirus, adeno-associated virus (AAV) in the gene encoding the cystic fibrosis transmembrane and nonviral vectors were developed and used in CF conductance regulator (Cftr).1 CFTR is an ATP-binding gene therapy clinical trials.8 Gene therapy vectors are cassette (ABC) transporter that plays a critical role in salt mostly studied in cultured cells, rodents and other transport and fluid movement by controlling chloride animal models to verify their efficacy before clinical ion transport across epithelia.2 Dysfunction of CFTR trials. However, an appropriate animal model to study leads to disease in various organs including the CFTR functional correction in lung has not been achieved respiratory and digestive tracts. However, the complica- because Cftr knockout mice do not show the pulmonary tions in the lung caused by abnormal airway secretions, pathologic features observed in human CF lung.9 chronic infection and inflammation pose the most serious Although the Cftr knockout mouse models have been threat to CF patients.3 Currently, how the loss of CFTR utilized to evaluate the therapeutic potential of CF gene function causes the clinical manifestations and patholo- therapy,10–12 the lack of pathophysiology in the Cftr gical features of CF lung disease is not fully understood, knockout mouse lung limits the interpretation of experi- so treatment is focused on management of respiratory mental results from CF mouse models. Therefore, an infection and inflammation in the lung.4 alternative model using a native human tissue is CF is considered one of the diseases that could be critically needed to assess the efficacy of CF gene therapy. treated by gene therapy because it is a single gene The eccrine sweat gland (referred to as ‘sweat gland’ disorder and the vector can be delivered directly to the below) is one of the organs affected by CF. The sweat lung through the airway. Moreover, restoration of normal gland is one of the smallest organs in the human body, uniquely developed in humans to control body tempera- ture by evaporation. It has a simple tubular structure Correspondence: Dr J Hu, Lung Biology Research Hospital for Sick Children, 555 University Ave., Toronto, ON, Canada M5G 1X8 with only 2–3 layers of cells. Isotonic primary sweat is Received 20 March 2005; accepted 13 June 2005; published online produced in the lower part of the sweat gland called the 21 July 2005 secretory coil, and, as the primary sweat moves to the Gene delivery to human sweat glands H Lee et al 1753 skin surface through the sweat duct, electrolytes are culture). After 2 days of exposure to the vector, the skin reabsorbed by the epithelial cells from the apical tissue was assayed for b-galactosidase activity deter- membrane. CFTR plays an important role in sweat mined by reaction with X-gal. The transduction effi- secretion and salt reabsorption in sweat glands,13 and ciency was very low and only a few cells in a few sweat its dysfunction in CF patients results in salty sweat. The glands stained with X-gal (Supplementary data, Figure abnormal sweat composition in CF patients can be 8). Those few transduced sweat glands were observed detected by a ‘sweat test’, a common diagnostic method only in the intradermally injected or submersion organ used in CF clinics. Sweat glands do not manifest the cultures. No sweat gland transduction was observed by anatomical alterations or pathology observed in other topical application of virus to skin in air–liquid interface organs affected in the disease, and have been used as a culture. model to study CFTR function and electrolyte transport Most transduced sweat glands were found at sites pathophysiology.14,15 where they had been directly exposed to virus. For There are a number of advantages in using human example, transduced sweat glands were found at the sweat glands as a model to study CF. They are relatively immediate site of the injection or at the excised periphery easy to access from a skin punch biopsy, and the function of the skin tissue in the submerged culture. This of CFTR can be measured electrophysiologically in the observation indicated that the adenovirus vector has apical membrane of the sweat duct ex vivo.16 With these limited ability to penetrate and transduce skin tissue advantages, sweat glands could be a useful model to in the organ culture. In order to increase the number of assess efficacy of CF gene therapy. Delivery and exposed sweat glands from the dermis, we treated the expression of normal Cftr to CF sweat glands is expected skin tissue with collagenase. Although collagenase to restore the normal chloride transport function. Thus, it partially digested the collagen-rich dermal matrix, it left should be possible to evaluate the effectiveness of gene the tissue sticky and did not noticeably increase the transfer and expression by the secretory response, transduction efficiency. Nonetheless, we found that a biochemical analyses of the sweat, or electrophysiologi- few sweat glands at the periphery of the tissue, which cal properties according to well-established methods.16 were completely freed from the dermal matrix by the Furthermore, using sweat glands in studying CF gene collagenase treatment, were transduced by the adenovir- therapy has significant implications for CF clinical trials. al vector (Figure 1). These results using skin organ It is inherently safer and more practical to administer culture suggested that the epidermis and dermal matrix gene therapy vectors to a small punch biopsy or to a are significant barriers to adenoviral vectors. small region of skin, rather than to CF affected organs such as pancreas or lung. However, whether genes can Gene delivery to isolated sweat glands be effectively delivered to human sweat glands remains In an attempt to circumvent the barriers in skin tissue, to be shown. Although sweat glands may provide an we isolated sweat glands from the dermis by mechani- important model to study the effect of CF gene therapy cally mincing the skin.18 The isolated sweat glands were in human native tissues, differences in morphology then incubated with HD-Ad vectors added to the media. and physiology exist between the sweat gland and Inoculating HDD28E4LacZ17 (5 Â 1010 viral particles) into the lung. 100 ml media containing isolated sweat glands appeared As a first step to explore the possibility of using sweat to increase the transduction and expression of the glands as a model for CF gene therapy, we investigated transgene in the sweat glands, highlighted by the large effective gene delivery methods using normal human area of X-gal staining (Figure 2a). However, when the sweat glands. We showed that transgenes can be histologic sections were examined most of the lacZ efficiently delivered to the sweat gland cells by adeno- positive cells were found at the outer layer of the viral gene therapy vectors. Furthermore, Cftr was transduced sweat glands (Figure 2b). In fact, most of the delivered to, and expressed by, the target cells in the X-gal staining was in the remnant interstitial tissue, lumen of sweat glands using a helper-dependent indicating that the HD-Ad did not penetrate to the inner adenovirus (HD-Ad) vector containing an epithelial layer of cells of the sweat gland when applied from the cell-specific promoter. We also found that the basement basal side. This limitation is a problem for using sweat membrane is an important barrier to adenoviral vector gland as a CF gene therapy model because CFTR must be transduction. expressed and function at the luminal membrane.19,20

Results Enhancement of gene delivery to isolated sweat glands Transduction of human sweat glands in skin organ Since sweat gland tubules are only 2–3 cell layers thick, culture we suspected that the poor penetration of adenoviral Considering the small size of sweat glands and their vector was due to some of the remnant dermal ubiquitous distribution in human skin, we first at- component surrounding the sweat glands. We then tempted to transduce sweat glands embedded in the treated the isolated sweat glands with collagenase dermis of full-thickness skin organ culture. A small block (0.1 U/ml) and dispase (50 U/ml) overnight at 41C, of excised full-thickness skin (1 Â1 cm) was maintained before inoculating with the HD-Ad vector (the unit either at the air–liquid interface or submerged in culture definition of the enzymes are defined by the manufac- medium, and exposed to a HD-Ad vector expressing turers and do not correlate). These treatments not only b-galactosidase (HDD28E4LacZ)17 by topical application dramatically increased the overall transduction efficiency or intradermal injection (air–liquid interface culture), or but also resulted in transgene expression from the cells in by incubation in virus-containing media (submersion the lumen (Figure 3a and c). Transduction efficiency was

Gene Therapy Gene delivery to human sweat glands H Lee et al 1754

Figure 2 Transduction of isolated sweat glands. (a) Sweat glands were Figure 1 Transduction of sweat glands in skin organ culture. Human skin isolated from human skin tissue and transduced with HDD28E4LacZ. tissue blocks were pretreated with collagenase and transduced by culturing After 2 days of exposure to the vector, the sweat glands were assayed for in media containing HDD28E4LacZ. After 2 days the tissue blocks were lacZ expression (original magnification  90). (b) The X-gal stained sweat assayed for b-galactosidase activity by reaction with X-gal (blue) and the glands were cryo-sectioned and counter stained with neutral red for transduced sweat glands were isolated and photographed. A part of a sweat histologic analysis (original magnification  400). duct (a) and a whole sweat gland (b) were partially transduced by the vector (original magnification  90). Basement membrane is the major barrier to adenoviral vector delivery To further analyze how the dispase treatment increased correlated with the concentration of the proteases used. transduction efficiency of sweat glands, we examined the The relationship between the degree of enzymatic integrity of the basement membrane after dispase digestion and the transduction efficiency was further treatment by immunostaining for type IV collagen. Type confirmed in other experiments, where the sweat glands IV collagen is the most abundant constituent of basement were treated with dispase for various lengths of time as membranes and assembles a sheet-like network with well as at different temperatures (Supplementary data, laminin to form the scaffold for the basement mem- Figure 9). brane.21 When compared to intact sweat glands (Figure Even though both collagenase and dispase signifi- 4a), treatment with 50 U/ml dispase overnight at 41C cantly enhanced transduction efficiency at the higher significantly reduced type IV collagen in the basement concentrations, they showed different effects on gene membrane (Figure 4b). Digestion for 8 h at 371C delivery at lower concentrations. When sweat glands eliminated most of the type IV collagen at the basement were treated with collagenase at 0.004 U/ml, the trans- membrane (Figure 4c). From these results we conclude duction was limited to the outer cell layer of the glands that the basement membrane is one of the major barriers (Figure 3b). On the other hand, the dispase treatment to gene delivery to sweat glands using adenoviral with the same factor of dilution (2 U/ml) resulted in vectors. fewer transduced glands, but the adenoviral vector delivered the transgene to the cells in the lumen of those glands (Figure 3d). These results suggested that pre- Cftr gene delivery to sweat glands with HD-Ad- treatment with dispase might be a more effective and K18CFTR specific method to improve gene delivery to the lumen of Our laboratory has developed HD-Ad vectors with the sweat glands. control elements from the human cytokeratin 18 gene

Gene Therapy Gene delivery to human sweat glands H Lee et al 1755

Figure 3 Enhanced transduction of sweat glands by enzyme pretreatment. The sweat glands were isolated from human skin tissue and treated with (a) collagenase, 0.1 U/ml; (b) collagenase, 0.004 U/ml; (c) dispase, 50 U/ml; or (d) dispase, 2 U/ml, prior to HDD28E4LacZ exposure. After overnight incubation at 41C in enzymes containing KGM media, sweat glands were transduced with HDD28E4LacZ. The unit definitions of the two enzymes are defined by the manufacturers and do not correlate (gross view magnification  90, histology magnification  200).

(KRT18 or K18) for CF gene therapy.10,22 These vectors similar to that of Cftr.23 K18 has been shown to be may be well suited for CF gene therapy because the K18 expressed in epithelial cells of various organs including promoter can drive cell-specific expression in a pattern sweat glands,24,25 and we confirmed its expression in

Gene Therapy Gene delivery to human sweat glands H Lee et al 1756

Figure 5 Transduction of sweat glands by HD-Ad-K18LacZ. Isolated sweat glands pretreated with dispase were transduced by HD-Ad- K18LacZ. The lacZ expression in the transduced sweat glands was Figure 4 Digestion of the basement membrane by dispase. Sweat glands detected by X-gal staining (a, original magnification  90) and analyzed were isolated and treated with (a) no enzyme, (b) dispase overnight at 41C, by histologic section (b, original magnification  200). In addition, the or (c) dispase for 8 h at 371C. Immunohistochemistry was carried out with X-gal stained tissue sections were subjected to immunostaining with frozen sections of the sweat glands using anti-type IV collagen. Type IV anti-cytokeratin 18 antibody to identify epithelial cells (c, original collagen was detected by DAB (brown) staining against methyl-green magnification  1000). Since the lacZ transgene in HD-Ad-K18LacZ counter staining (green) (original magnification  100). contains a nuclear localization signal, X-gal staining (blue) was observed in the nuclei, whereas K18 was detected (brown) in the cytoplasm.

human sweat glands by immunostaining (Supplemen- tary data, Figure 10). expression level was moderate compared to that We tested whether the K18 control elements could achieved by HDD28E4LacZ, and showed a punctate drive an expression pattern similar to that of native Cftr transduction pattern (Figure 5a). The histologic examina- in sweat glands. Isolated sweat glands treated with tion of HD-Ad-K18LacZ transduced sweat glands dispase were exposed to HD-Ad-K18LacZ vector, which showed cell type-specific expression of the transgene contained the b-galactosidase gene with a nuclear (Figure 5b). Unlike sweat glands transduced with localization signal under the control of the K18 promo- HDD28E4LacZ, where all of the cells expressed the ter.22 As shown in Figure 5 transgene expression by HD- transgene, HD-Ad-K18LacZ induced lacZ expression Ad-K18LacZ in sweat glands was distinctive from that of specifically in the luminal cells. The condition of the human cytomegalovirus (CMV) promoter-driven histologic sections of isolated sweat glands after dispase vector HDD28E4LacZ (Figure 3a and c). In general, the treatment and ex vivo culture prevented clear morpholo-

Gene Therapy Gene delivery to human sweat glands H Lee et al 1757 HD-Ad-K18-CFTR 1 2M3 4 K18 promoter K18 intron 1

337 305 K18 exon 1 Cftr PvuII transgene Cftr mRNA 178 127 endogenous Cftr mRNA 5’ UTR Figure 6 Expression of Cftr by sweat glands transduced with HD-Ad- K18-CFTR. A total of 50 sweat glands were treated with dispase overnight and transduced with HD-Ad-K18-CFTR for 2 days. The total RNA was isolated and cDNA was generated using a common reverse primer (black arrow head). Forward primers designed specifically at the K18 exon 1 (gray arrow head) or 50 UTR of Cftr (unfilled arrow head) were used in PCR (35 cycles) to detect the transgene Cftr (Lane 2, 305 bp) or endogenous Cftr (Lane 3, 337 bp) expression, respectively. The transgene expression was confirmed by restriction enzyme (PvuII) digestion (Lane 1). Lane 4 is RT- PCR product (25 cycles) of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). gical identification of the cell types on sections. However, the epithelial cell-specific expression of the transgene was confirmed by double staining to detect K18 (by immunostaining) in the cytoplasm and LacZ (by X-gal staining) in the nucleus (Figure 5c). The expression of lacZ transgene in luminal cells in this study was similar to the Cftr expression pattern detected in normal human sweat glands by immunohistochemistry in other stu- dies.19,26 These results show that the HD-Ad vector with K18 control elements can drive an epithelial cell specific transgene expression in sweat glands. Finally, we transduced sweat glands with HD-Ad- K18CFTR, our HD-Ad containing the human Cftr gene tagged with a VSV-G epitope under control of the K18 promoter.10 Cftr expression was detected by RT-PCR using transgene-specific primers designed across K18 intron 1, which allowed us to distinguish the PCR product of the transgene Cftr mRNA from that of the endogenous Cftr mRNA or the vector DNA (Figure 6). Furthermore, we confirmed Cftr gene expression in the epithelial cells of the lumen by immunostaining the sweat gland sections with anti-VSV-G antibody 3 days after transduction (Figure 7b). CFTR-VSV-G was detected Figure 7 Gene delivery and expression of Cftr. Sweat glands were in the perinuclear region and at the plasma membrane transduced with (a) no virus, or (b) HD-Ad-K18CFTR, and Cftr of transduced cells in the sweat glands (Figure 7c), expression was detected by immunostaining with anti-VSV-G antibody consistent with our immunohistochemistry results for (brown) against methyl-green counter staining (original magnification the endogenous CFTR of the human sweat gland  200). (c) is at a higher magnification of (b) showing intracellular (Supplementary data, Figure 11) as well as previous localization of transduced Cftr gene product (original magnification reports of the intracellular localization of CFTR.27  1000).

Discussion therapy.29 A number of gene delivery studies with This study investigated gene delivery to human sweat adenoviral vectors showed that it is possible to achieve glands to explore the possibility of using sweat glands as a significant transduction of mouse epidermis in vivo, a model for CF gene therapy. Prior to this study there and skin appendages such as hair follicles and sebaceous have been no reports of gene delivery directed to human glands can be transduced in mouse skin.30–32 The sweat glands. One study reported partial transduction of discrepancy between the transduction we observed with sweat glands when an AAV vector was intradermally our human skin organ culture and what others have injected into pig skin.28 Using various methods we observed in vivo may reflect inherent differences between attempted to deliver genes using an HD-Ad vector to eccrine sweat glands and other skin appendages, mouse human skin in organ culture. We found that gene versus human species differences, and/or differences delivery to sweat glands in intact human skin was very between skin in organ culture versus skin in vivo. inefficient ex vivo. In general, topical gene delivery has Our observations indicated that the basement mem- been considered to be difficult for cutaneous gene brane is a critical barrier to the delivery of adenoviral

Gene Therapy Gene delivery to human sweat glands H Lee et al 1758 vectors from the basal side of sweat glands. The changes in sweat secretion function using sweat glands basement membrane has been suggested as a physical from CF patients. barrier to herpes simplex virus (HSV) vectors,33 and The study of functional efficacy is an important step Weeks et al34 showed that the intradermal injection of toward a successful CF gene therapy. In addition to HSV below the epidermal basement membrane did not molecular end points, such as mRNA and protein cause pathogenesis. In Duchenne muscular dystrophy expression, clinical trials often relied on nasal electrical gene therapy, the ineffectiveness of viral vectors to potential difference to detect correction of the chloride transduce mature muscle has been a major impedi- transport function.8 However, measuring nasal potential ment.35 Huard et al33 reported that the more mature difference is known to be a variable technique39 and basement membrane of the muscle fiber in adult mice, there have been conflicting results questioning the compared to newborn mice, acted as barrier for HSV sensitivity of the method for measuring CFTR func- vector transduction. Van Deutekom et al35 also showed tion.40,41 Although the nasal epithelium is more closely that treating myofibers with streptokinase, a fibrinolytic related to the lung epithelium in its electrophysiology, drug, enhanced transduction efficiency of the HSV vector the sweat gland could provide an alternative model to in mature skeletal muscle, suggesting basement mem- assess the efficacy of the gene therapy vector in an organ brane and surrounding extracellular matrix as a physical without disease pathology. One difficulty would be the barrier. Interestingly, Pruchnic et al36 reported that AAV currently less widespread use of sweat gland electophy- could transduce mature as well as immature muscle siology techniques14–16 compared to nasal potential fibers, and proposed that the small size of AAV, difference measurement. compared to other viral vectors, might have contributed In addition to quantitative assays for the functional its ability to penetrate the basement membrane. evaluation of gene therapy, the sweat gland model could We were able to significantly enhance adenoviral gene provide advantages in clinical trials. Gene delivery to delivery to sweat glands by dispase pretreatment. Saito et sweat glands from the skin surface or to sweat glands al37 used collagenase to enhance the gene transfer to hair isolated from a punch biopsy would be more convenient follicle from mouse and suggested that partial digestion than delivery to the nasal cavity or the lung. More of the dermis might have increased the transduction importantly, ex vivo transduction of sweat glands, or efficiency. We also achieved comparable transduction transduction of sweat glands in a small region of skin, efficiencies with sufficient collagenase treatment instead would not be a major safety and ethics concern of dispase. However, dispase treatments were more compared to applying viral vectors to pathologically effective than the collagenase at higher dilutions (Figure compromised organs such as the CF lung. The sweat 3b and d). Commercially available collagenase and gland model can also provide a way to predict dispase are extracts from bacterial cultures and exhibit compliance to the gene therapy for each individual various proteinase activities, but dispase is considered an patient, by testing his own sweat glands prior to effective reagent for digestion of basement components treatment of the CF affected organ. With these potential such as fibronectin and type IV collagen.38 Our immu- clinical implications, the sweat gland could be consid- nostaining for type IV collagen demonstrated that ered as an alternative model to study efficacy of CF gene degradation of the basement membrane was correlated therapy in humans. with enhanced transduction efficiency. Although we used dispase treatment to enhance gene delivery from the basal side of the sweat gland in our ex vivo study, Materials and methods an alternative method could be used to deliver genes to sweat glands without the enzyme pretreatment, to Helper-dependent adenoviral vectors preserve tissue integrity. One of the plausible methodol- Construction of the HDD28E4LacZ, HD-Ad-K18LacZ ogies is microperfusion. Microperfusion of gland tubules and HD-Ad-K18CFTR HD-Ad vectors has been de- has been used to study electrophysiology and CFTR scribed previously. HDD28E4LacZ contains bacterial lacZ functions in sweat ducts.14,15 This technique would allow under control of the CMV immediate-early promoter.17 delivery of the vector directly to the lumen of sweat HD-Ad-K18LacZ and HD-Ad-K18CFTR have 2.5 kb glands instead of from the basal side, without enzyme of genomic sequence upstream from the cytokeratin 18 pretreatment. Once the vector is delivered to the sweat (K18) gene containing the promoter plus first intron of tubule, the chloride conductance could be quantitatively K18 followed by a translation enhancer.10,22 HD-Ad- measured to evaluate CFTR function in the sweat K18LacZ contains a nuclear localization signal from SV40 gland.16 antigen.22 HD-Ad-K18CFTR has the vesticular stomatitis In this study, we have demonstrated that it is possible virus G glycoprotein (VSV-G) epitope tag at the C- to deliver Cftr to sweat glands using an HD-Ad, and to terminus of CFTR.10 The HD-Ads were prepared as express CFTR in the epithelial cells in the lumen. Our previously described.17 The final concentration of virus results provide the basis to study the functional efficacy was 5–8 Â 1012 viral particles/ml. of CF gene therapy using human sweat glands, and we propose that sweat glands could be used as an Preparation of full thickness skin and gene delivery alternative model. Delivery of Cftr to sweat glands from using skin organ culture a CF patient should restore normal CFTR function of Human skin tissue was obtained from abdominal b-adrenergically stimulated secretion, lower sweat salt pannus excised from otherwise healthy subjects under- concentrations, and enhance ClÀ conductance in sweat going dermolipectomy. A clinical protocol (HSC protocol ducts. These changes can be quantitatively evaluated by # 001960042) was approved for ex vivo experiments on biochemical and electrophysiological methods.14–16 the skin tissue. The hypodermal fat was removed from Future studies should be directed toward detecting these the skin and the full-thickness skin was briefly washed

Gene Therapy Gene delivery to human sweat glands H Lee et al 1759 with PBS. For skin organ culture study the full-thickness Immunohistochemisty skin was cut into 1 Â1 cm pieces and either submerged The transduced sweat glands were treated in fixative in media or placed on the Transwell culture system solution (as above) for 30 min at 41C and embedded in (Costar, Corning, NY, USA), such that the dermis OCT for cryosection, or in paraffin after a series of short contacted the medium while the epidermis remained dehydration steps in 70% (20 min), 90% (20 min), 100% exposed to the air. RPMI 1640 containing 10% FBS was ethanol (2 Â 20 min), and xylene (2 Â 30 min). Both frozen used as the culture media. The skin was transduced and paraffin sections were cut 5 mm in thickness and by topical application or injection of adenovirus, or paraffin sections were deparaffinized prior to immuno- by addition of adenovirus to the submersion culture staining. Tissue sections were rehydrated in PBS, treated medium. For topical application, 10 ml of virus solution, with 0.2% Triton X-100 in PBS for 4 min, quenched with 10 containing 2.5 Â 10 virus particles, was carefully laid on 0.3% H2O2 for 10 min, and blocked with blocking buffer the epidermis in the center of the skin sample. For (2% BSA, 5% goat serum) for 30 min. Then, monoclonal injection, the same number of virus particles (in 5 ml) was antibody against VSV-G (1:2000 dilution, Boehringer intradermally injected using a 30G needle. For submer- Mannheim, Indianapolis, IN, USA) or human type IV sion culture, the same number of virus particles was collagen (1:2500 dilution, Chemicon, Boronia, Australia) added to 1 ml media, which contained the skin sample. diluted in blocking buffer was applied to the section. The The tissue was incubated for 2 days in a 5% CO2 primary antibodies were detected using the Vectastain incubator at 371C and then stained with X-gal. In order to Kit and DAB Substrate Kit for Peroxidase (Vector digest the dermal matrix, full-thickness skin was placed Laboratories, Burlingame, CA, USA), according to the in 2 mg/ml of collagenase (274 U/mg) (Worthington manufacturer’s recommendations. Finally, the sections Biochemicals, Lakewood, NJ, USA) in aMEM with 1% were counter stained with either neutral red or methyl FBS for 2 h at 371C. green. For K18/LacZ double staining, the X-gal stained sweat glands were embedded in paraffin and subjected to immunostaining with anti-cytokeratin 18 antibody Isolation of sweat glands and transduction (1:20 dilution, RDI, Flanders, NJ, USA). Sweat glands were isolated under a dissection micro- scope after mincing 1 Â 0.5 cm full-thickness skin with fine surgical scissors.18 The isolated sweat glands were RT-PCR maintained in KGM media (Clonetics, Walkersville, MD, A total of 50 isolated sweat glands were transduced with USA) during isolation and the transduction experiment. 1.24 Â 1011 particles of HD-Ad-K18-CFTR in 100 mlof The sweat glands were digested in 100 ml of KGM KGM for 2 days as described above. The total RNA was containing 10% dispase (5000 U/ml) (Collaborative isolated in 500 ml of TRIzol (Invitrogen, Carlsbad, CA, Research, Bedford, MA, USA) overnight at 41C, unless USA), according to the manufacturer’s recommendations specified otherwise. Alternatively, collagenase H following tissue disruption by a cycle of freezing and (0.2 U/ml) (Roche, Mannheim, Germany), diluted in thawing and vigorous pippetting. cDNA was generated 0 KGM, was used to compare transduction efficiencies. using a common reverse primer (5 -ccgaagggcattaat- 0 The unit definitions of dispase and collagenase were gagtttagg-3 ) in the coding region of Cftr, and the forward 0 defined by the manufacturers and do not correlate. Five primers designed specifically at the K18 exon 1 (5 - 0 0 to ten isolated sweat glands in each well of a 96-well dish GTCCGCAAAGCCTGAGTCCTGTCC-3 )or5 UTR of 0 0 were washed twice in PBS and cultured in 100 mlof Cftr (5 -ATCACAGCAGGTCAGAGAAAAAG-3 ) were KGM. The HD-Ad (2.5–5 Â 1010 particles) was added used in PCR to detect the transgene or endogenous Cftr directly to the media and the culture maintained for expression, respectively. The expression of glyceralde- 2 days, except when specified otherwise. hyde-3-phosphate dehydrogenase (GAPDH) was also detected by RT-PCR using a reverse primer (50-GATG ATGACCCGTTTGGCTCC-30) and a forward primer X-gal staining (50-CTGGTAAAGTGGATATTGTTGCC-30). The PCR re- Sweat glands transduced with HD-Ad vectors containing actions were performed for 35 cycles, except for GAPDH, lacZ were assayed for b-galactosidase activity by reaction whose PCR reaction was performed for 25 cycles. with X-gal. The sweat glands were briefly washed in PBS and fixed in fixative solution (1% formaldehyde, 0.1% glutaraldehyde, 2 mM MgCl2,5mM EGTA in 0.1 M Acknowledgements sodium phosphate buffer, pH 7.8) by incubating for 30 min at 41C with gentle mixing. After washing twice We thank Bernard G Martin for preparation of the for 15 min in washing solution (2 mM MgCl2, 0.02% NP- helper-dependent adenoviral vectors, and Dr Jeffery 40, 0.01% deoxycholate in 0.1 M sodium phosphate Wang and Mr Kirk Tayler for technical assistance. We buffer, pH 7.8) at 41C, the tissue was stained with X-gal also thank Dr Hugh O’Bradovich and Dr Johanna solution (wash solution plus 5 mM K4Fe(CN)6:3H2O, Rommens for critical comments during this study. This

5mM K3Fe(CN)6, 1 mg/ml 5-bromo-4-choloro-3-indolyl work was supported by an Operating Grants from the b-D-galactopyranoside) for 2 h at 371C. Samples were Canadian Cystic Fibrosis Foundation, an Operating then rinsed in 70% ethanol, and further fixed in 4% Grant from the Canadian Institutes of Health Research paraformaldehyde for 1 h at 41C. We used same methods to JH and to CYP (MOP 8048), the Sellers Fund to JH and for skin tissue but allowed longer incubation times. The CYP, and an Operating Grant from the National stained tissue samples were sectioned after embedded Institutes of Health (P50 HL59314) to PN. JH is a CCFF in paraffin or OCT (Tissue-Tek, Torrance, CA, USA) and Scholar and holds a Premier’s Research Excellence counter stained with neutral red. Award of Ontario, Canada.

Gene Therapy Gene delivery to human sweat glands H Lee et al 1760 References 22 Toietta G et al. Reduced inflammation and improved airway expression using helper-dependent adenoviral vectors with a 1 Quinton PM. Physiological basis of cystic fibrosis: a historical K18 promoter. Mol Ther 2003; 7: 649–658. perspective. Physiol Rev 1999; 79: S3–S22. 23 Chow YH et al. Targeting transgene expression to airway 2 Sheppard DN, Welsh MJ. Structure and function of the CFTR epithelia and submucosal glands, prominent sites of human chloride channel. Physiol Rev 1999; 79: S23–S45. CFTR expression. Mol Ther 2000; 2: 359–367. 3 Boucher RC. An overview of the pathogenesis of cystic fibrosis 24 Kozaki M et al. Immunohistochemical analysis of cytokeratin lung disease. Adv Drug Deliv Rev 2002; 54: 1359–1371. expression in dog skin. J Vet Med Sci 2001; 63: 1–4. 4 Gibson RL, Burns JL, Ramsey BW. Pathophysiology and 25 Watanabe S, Hirose M, Sato S, Takahashi H. Immunohistochem- management of pulmonary infections in cystic fibrosis. Am J ical analysis of cytokeratin expression in eccrine spiradenoma: Respir Crit Care Med 2003; 168: 918–951. similarities to the transitional portions between secretory 5 Dorin JR et al. A demonstration using mouse models that segments and coiled ducts of eccrine glands. Br J Dermatol successful gene therapy for cystic fibrosis requires only partial 1994; 131: 799–807. gene correction. Gene Therapy 1996; 3: 797–801. 26 Claass A et al. Applicability of different antibodies for 6 Kerem B, Kerem E. The molecular basis for disease variability in immunohistochemical localization of CFTR in sweat glands cystic fibrosis. Eur J Hum Genet 1996; 4: 65–73. from healthy controls and from patients with cystic fibrosis. 7 Amaral MD. Processing of CFTR: Traversing the cellular maze – J Histochem Cytochem 2000; 48: 831–837. how much CFTR needs to go through to avoid cystic fibrosis? 27 Bradbury NA. Intracellular CFTR: localization and function. Pediatr Pulmonol 2005; 39: 479–491. Physiol Rev 1999; 79: S175–S191. 8 Albelda SM, Wiewrodt R, Zuckerman JB. Gene therapy for lung 28 Hengge UR, Mirmohammadsadegh A. Adeno-associated virus disease: hype or hope? Ann Intern Med 2000; 132: 649–660. expresses transgenes in hair follicles and epidermis. Mol Ther 9 Grubb BR, Boucher RC. Pathophysiology of gene-targeted 2000; 2: 188–194. mouse models for cystic fibrosis. Physiol Rev 1999; 79: S193–S214. 29 Ghazizadeh S, Taichman LB. Virus-mediated gene transfer for 10 Koehler DR et al. Protection of Cftr knockout mice from acute cutaneous gene therapy. Hum Gene Ther 2000; 11: 2247–2251. lung infection by a helper-dependent adenoviral vector expres- 30 Li L, Hoffman RM. The feasibility of targeted selective gene sing Cftr in airway epithelia. Proc Natl Acad Sci USA 2003; 100: therapy of the hair follicle. Nat Med 1995; 1: 705–706. 15364–15369. 31 Lu B et al. Topical application of viral vectors for epidermal gene 11 Limberis M, Anson DS, Fuller M, Parsons DW. Recovery of transfer. J Invest Dermatol 1997; 108: 803–808. airway cystic fibrosis transmembrane conductance regulator 32 Domashenko A, Gupta S, Cotsarelis G. Efficient delivery of function in mice with cystic fibrosis after single-dose lentivirus- transgenes to human hair follicle progenitor cells using topical mediated gene transfer. Hum Gene Ther 2002; 13: 1961–1970. lipoplex. Nat Biotechnol 2000; 18: 420–423. 12 Grubb BR et al. Inefficient gene transfer by adenovirus vector to 33 Huard J et al. The basal lamina is a physical barrier to herpes cystic fibrosis airway epithelia of mice and humans. Nature 1994; simplex virus-mediated gene delivery to mature muscle fibers. 371: 802–806. J Virol 1996; 70: 8117–8123. 13 Quinton PM. The sweat gland. In: Yankaskas JR, Knowles MR 34 Weeks BS, Ramchandran RS, Hopkins JJ, Friedman HM. Herpes (eds). Cystic Fibrosis in Adults. Lippincott-Raven: Philadelphia, simplex virus type-1 and -2 pathogenesis is restricted by the 1999, pp 419–437. epidermal basement membrane. Arch Virol 2000; 145: 385–396. 14 Quinton PM, Reddy MM. Control of CFTR chloride conductance 35 van Deutekom JC, Hoffman EP, Huard J. Muscle maturation: by ATP levels through non-hydrolytic binding. Nature 1992; 360: implications for gene therapy. Mol Med Today 1998; 4: 214–220. 79–81. 36 Pruchnic R et al. The use of adeno-associated virus to circumvent 15 Reddy MM, Quinton PM. Control of dynamic CFTR selectivity the maturation-dependent viral transduction of muscle fibers. by glutamate and ATP in epithelial cells. Nature 2003; 423: Hum Gene Ther 2000; 11: 521–536. 756–760. 37 Saito N et al. High efficiency genetic modification of hair 16 Sato K, Sato F. Methods for studying eccrine sweat gland follicles and growing hair shafts. Proc Natl Acad Sci USA 2002; 99: function in vivo and in vitro. Methods Enzymol 1990; 192: 583–599. 13120–13124. 17 Palmer D, Ng P. Improved system for helper-dependent 38 Stenn KS et al. Dispase, a neutral protease from Bacillus polymyxa, adenoviral vector production. Mol Ther 2003; 8: 846–852. is a powerful fibronectinase and type IV collagenase. J Invest 18 Lee CM, Jones CJ, Kealey T. Biochemical and ultrastructural Dermatol 1989; 93: 287–290. studies of human eccrine sweat glands isolated by shearing and 39 Ahrens RC et al. Use of nasal potential difference and sweat maintained for seven days. J Cell Sci 1984; 72: 259–274. chloride as outcome measures in multicenter clinical trials in 19 Sato F, Sato K. cAMP-dependent Cl(À) channel protein (CFTR) subjects with cystic fibrosis. Pediatr Pulmonol 2002; 33: 142–150. and its mRNA are expressed in the secretory portion of human 40 Walker LC et al. Relationship between airway ion transport and eccrine sweat gland. J Histochem Cytochem 2000; 48: 345–354. a mild pulmonary disease mutation in CFTR. Am J Respir Crit 20 Kartner N et al. Mislocalization of delta F508 CFTR in cystic Care Med 1997; 155: 1684–1689. fibrosis sweat gland. Nat Genet 1992; 1: 321–327. 41 Ho LP et al. Correlation between nasal potential difference 21 Kalluri R. Basement membranes: structure, assembly and role in measurements, genotype and clinical condition in patients with tumour angiogenesis. Nat Rev Cancer 2003; 3: 422–433. cystic fibrosis. Eur Respir J 1997; 10: 2018–2022.

Supplementary Information accompanies the paper on Gene Therapy website (http://www.nature.com/gt).

Gene Therapy