¹Department of Biochemistry, The University of Hong Kong, China

²Department of Dermatology, University of Muenster, Germany

Correspondence to: K SE Cheah, Department of Biochemistry, 5, Sassoon Road, Li Shu Fan Building, The University of Hong Kong, China

We report the isolation of a cosmid clone containing the entire human COL7A1 gene in one piece. The ability of the genomic sequences within this clone to direct tissue-specific expression of human collagen VII in transgenic mice was tested. The data show that the gene construct is capable of directing expression of collagen VII in the skin of fetal and neonatal transgenic mice. Expression of COL7A1 in these mice was widespread, in a pattern consistent with that found in human tissues and was in parallel with that of the endogenous mouse gene. Immunostaining, using human-specific antibodies, showed that human collagen VII protein was present at the skin basement membrane zone of the transgenic mice. Dermal extracts from 19-month-old transgenic mice contained mature human collagen VII protein, and fibroblasts derived from skin biopsies of these mice actively synthesized human collagen VII. The demonstration of successful and stable expression of human collagen VII in in vivo gene transfer is the first step towards the future development of therapeutic protocols for the rescue of keratinocyte function in severe blistering diseases such as dystrophic epidermolysis bullosa. Gene Therapy (2000) 7, 1631-1639.

dystrophic epidermolysis bullosa; human type VII collagen; gene therapy; skin disease; transgenic mice

Introduction

Anchoring fibrils attach the epidermal basement membrane of the skin to the underlying dermal connective tissue.^1,2 These fibrils are functionally deficient in hereditary dystrophic epidermolysis bullosa (DEB), a clinically and biologically heterogeneous group of dominant and recessive blistering skin disorders (reviewed in Refs ³ and ⁴). Clinical hallmarks of the disorders are blistering of the skin and external mucous membranes in response to minimal trauma and scarring of the skin. Typical morphological findings include dermal-epidermal cleavage below the basement membrane, and absence or structural abnormalities of the anchoring fibrils.

The major component of anchoring fibrils is collagen VII which is a large macromolecule (approximately 1000 kDa) synthesized and secreted by keratinocytes, and to a lesser extent by fibroblasts, in the skin. Collagen VII is a homotrimer of three 1(VII) chains which associate to form a triple helix.¹ Its precursor, the procollagen molecule has three domains: a central triple helical domain consisting of (Gly-X-Y) repeats and two non-helical globular (NC-1, NC-2) domains. Procollagen VII molecules associate via their carboxyl termini following which the NC-2 domains are cleaved off.⁵ The resulting collagen VII molecules further condense to form the anchoring fibril in which the NC-1 domains are inserted and encapsulated in the basement membrane of the lamina densa and also in amorphous electron-dense structures that are termed anchoring plaques. An extensive scaffold of anchoring fibrils provides structural stability of the sub-basal lamina of the skin by entrapping other matrix components. The genes for human (COL7A1) collagen VII in chromosome 3p21 and mouse (Col7a1) collagen VII in chromosome 9 have been cloned and are highly conserved, both encoding a polypeptide of 2944 amino acids.^6,7,8,9

Collagen VII is of particular interest for the pathology of DEB because mutations in the human gene result in the disease. In milder DEB subtypes, collagen VII protein is expressed but the anchoring fibrils are morphologically altered.^3,4,10 In the most severe DEB form, both collagen VII and anchoring fibrils are absent from the skin,^11,12,13,14 and the clinical course of the disease is dramatic with extensive blistering and scarring which leads to fusion of the digits of the hands and feet, so called mitten hands. Therefore, this recessively inherited severe subtype is also called mutilating DEB.

To date more than 100 distinct COL7A1 mutations have been identified in DEB families (reviewed in Refs ⁴ and ¹⁵). Targeted inactivation of the Col7a1 gene in mice recently has been shown to result in the recapitulation of many of the phenotypic features of human recessive DEB (RDEB), confirming collagen VII deficiency as the primary cause of the disease.¹⁶ These findings also indicate that the mouse can be a useful model for RDEB.

Apart from being able to identify the molecular defect in DEB, and provide prenatal diagnosis, it would be very important to develop methods of treating the disease by gene therapy. Successful gene therapy for DEB will depend on the restoration of collagen VII in the anchoring fibrils of the skin by sustained synthesis. Despite detailed knowledge of the gene structure and biosynthesis of collagen VII, no effective gene therapy currently exists for this severe skin disorder. Difficulties associated with successful gene transfer to epidermal keratinocytes include limitations in the extent of surgical grafting that can be achieved and low and transient expression. These limitations are associated with the inability to achieve sustained gene expression by long-lived keratinocyte stem cells and their descendents (reviewed in Ref. ¹⁷).

Central to any approach to treat EB by gene therapy is the availability of clones containing the complete COL7A1 gene as one transcriptional unit. COL7A1 mRNA is approximately 9.2 kb and the gene consists of 118 exons, spanning 30 kb. The size of the gene and its transcript and the achieving tissue-specificity of gene expression therefore pose additional technical challenges for the development of in vivo approaches for stable gene transfer of COL7A1 in the epidermis. In this study we report the isolation of a cosmid clone containing the entire human COL7A1 gene in one piece. We have tested the ability of the genomic sequences within this clone to direct tissue-specific expression of human collagen VII in transgenic mice. The data show that the gene construct is capable of directing expression of collagen VII in the skin of fetal and neonatal transgenic mice. Expression of COL7A1 in the transgenic mice was in a pattern consistent with that found in human tissues and was in parallel with that of the endogenous mouse gene. Furthermore, we show that human collagen VII can be detected in the skin of transgenic mice even 19 months after birth.

Results

Isolation of the complete COL7A1 within a single genomic cosmid clone

An intact human COL7A1 gene within a single clone would be very useful for the development of gene therapy vectors for DEB. In order to isolate such a clone two human cosmid libraries were screened for the COL7A1 gene. Three cosmids were isolated and characterized by restriction enzyme mapping (see Materials and methods). The presence of the 5' and 3' ends of COL7A1 within one cosmid, HC7C61, was detected using oligonucleotide probes designed from the published COL7A1 sequences.⁶ Subclones from HC7C61 were sequenced and comparison of sequences, with the published sequence of COL7A1, showed that it contained the complete human COL7A1 gene with 3.0 kb 5' and 4.0 kb 3' flanking DNA by (Figure 1a). Southern analysis on human genomic DNA using HC7C61 as a hybridization probe showed an identical restriction enzyme pattern with that of HC7C61 except for the end fragments of the cosmid, confirming that the cosmid was not rearranged (Figure 1b).

Generation of transgenic mice expressing the COL7A1 gene

To determine whether HC7C61 contained a functional human COL7A1 gene which could be expressed in a tissue-specific manner, we generated transgenic mice carrying the cosmid, HC7C61. Three transgenic founders were produced. Two of these were collected at stage E16.5 for collagen VII immunostaining analyses. A third transgenic founder line (T61) was established for further study. Genomic DNA of a transgenic and a non-transgenic littermate and of human tissue were digested with HindIII and EcoRI and then hybridized with HC7C61 probe (Figure 1b). The restriction enzyme fragment pattern for T61 and human genomic DNA was identical confirming that the transgene was intact and was not rearranged.

Expression of the human COL7A1 transgene in T61 mice was tested for by RT-PCR using total RNA extracted from different tissues of a 19-month-old T61 and oligonucleotide primers, which were specific for the human COL7A1 gene (Figure 2). A specific product of the expected size for the human cDNA (423 bp), distinct from that expected from amplification of genomic sequence (667 bp), was obtained for RNA isolated from several tissues such as the spleen, heart, kidney, liver, lung and skin (Figure 2a). No RT-PCR product was generated with RNA from a non-transgenic littermate, confirming that the PCR was specific to the human gene (Figure 2a). These RT-PCR products hybridized to the human COL7A1 probe, pBE in Southern blot analyses, indicating that they are human COL7A1 specific (Figure 2a). Similar results were obtained for RNA isolated from different tissues of a 3 days post partum (dpp) transgenic mouse (data not shown).

The major site of expression of collagen VII is in the skin.^1,18,19,20 However despite reports of expression in other tissues,^{20,21,22,23,24,25} the tissue range of expression of collagen VII has not been extensively studied in fetal and neonatal stages. To determine if the expression of the transgene in other sites was the result of inappropriate expression or was in parallel with a wide pattern of expression of the human gene, the presence of collagen VII mRNA in human 21-week fetal (Figure 2b) and non-transgenic littermate mouse tissues (Figure 2c) was examined and compared. As shown in Figure 2b, expression of COL7A1 mRNA was found in human fetal eye, spleen, brain, heart, kidney, liver, lung and skin. In a 3-day neonatal wild-type mouse and E16.5 wild-type mouse fetuses, Col7a1 expression was for tested by RT-PCR using oligonucleotide primers, which were specific for the mouse Col7a1 gene. A specific product of the expected size for the mouse cDNA (152 bp) was produced in total RNA isolated from eye, brain, heart, kidney, liver, lung and skin and E16.5 mouse fetus, distinct from that expected from the amplification of genomic sequences (625 bp) (Figure 2c). Therefore there was agreement in the pattern of expression of the transgene with that found normally for human COL7A1 and for the endogenous mouse gene. These data suggest that HC7C61 contains sufficient regulatory DNA sequences to direct appropriate expression of COL7A1 in transgenic mice.

Expression of human collagen VII protein in the skin of the COL7A1 transgenic mouse

To determine if the human COL7A1 transgene could direct synthesis of collagen VII protein, immunostaining of cryosections of transgenic mouse skin was performed. Sections from human skin, the two independent transgenic founder E16.5 mouse fetuses and their non-transgenic littermate were tested for the presence of human collagen VII. Rabbit-anti-human antibody (1.3a-VII) specific for human collagen VII was used. A strong positive signal was found in human skin at the junction of the epidermis and the dermis (Figure 3b). A positive signal was also found at the dermal-epidermal junction in both transgenic fetuses (Figure 3f). A staining was also seen around the hair follicle (Figure 3f) and a weak signal in the tongue region along the surface lining (Figure 3j). Non-transgenic mouse sections did not stain with the antibody confirming its specificity to human collagen VII (Figure 3d and h). These data suggest that the transgene can produce human collagen VII protein.

To determine if human collagen VII protein synthesized in the skin of the transgenic mice was stable, skin biopsies obtained from dorsal skin of a 19-month-old T61 transgenic mouse and of a non-transgenic littermate were analyzed for expression of the transgene by immunostaining and immunoblotting of dermal extracts. Human collagen VII was expressed and deposited in the transgenic mouse skin, along the dermal-epidermal junction (Figure 4aiii) and the adnexal structures, such as the hair follicles (Figure 4aiv). The staining was distinct, however, less intense than with antibodies to mouse collagen VII (Figure 4ai). In contrast, staining of the skin of a non-transgenic littermate with antibodies to human collagen VII remained negative (Figure 4aii).

For immunoblot analysis, the epidermis and dermis were separated chemically from each other, extracted with chaotropic buffers and immunoblotted with antibodies to human collagen VII. Dermis extracts of the transgenic mouse contained a 290 kDa immunoreactive band corresponding to human collagen VII, whereas extracts from the control mice remained negative, demonstrating the expression of the human transgene in vivo (Figure 4b). The size of the immunoreactive polypeptides corresponded to that of mature human collagen VII chains, indicating that in the transgenic mouse skin the processing of human procollagen VII to collagen VII had occurred in the normal fashion. A further indication that the transgene product was normally processed and deposited in the matrix was the fact that the epidermal extracts did not contain collagen VII.

When skin fibroblasts from a 19-month-old transgenic mouse were stained with antibodies to human collagen VII, a distinct positive signal was observed in some cells indicating sustained expression of the human collagen (Figure 4ciii and iv). The staining was strong in the rough endoplasmic reticulum of the cells, reflecting active collagen biosynthesis. The secreted protein diffuses into the medium.^5,14 In agreement with the observations on the skin, the fibroblasts of normal littermates remained negative with antibodies to human collagen VII (Figure 4cii).

The presence of human collagen VII in the transgenic mice did not cause any obvious functional abnormality. The tail skin of the transgenic mice was subjected to repeatedly mechanical shearing stress, but it remained intact without any signs of blistering. This observation suggested that the coexistence of human and mouse collagen VII had not compromised the mechanical properties of the skin.

Discussion

The ability to achieve tissue-specific and stable expression of the COL7A1 gene and its product in vivo are key prerequisites for any protocol aimed at successful restoration of collagen VII to DEB skin by gene transfer. Since the epidermis is a tissue which undergoes constant renewal, sustained presence of collagen VII via an in vivo gene transfer protocol would be optimal for successful therapy. The challenges facing gene therapy research therefore are not only the design of vectors for efficient delivery but also solving the problem of correctly expressing the transferred gene (reviewed in Ref. ²⁶). There have been several attempts at gene transfer to epidermal keratinocytes (reviewed in Refs ¹⁷ and ²⁶). Most of these approaches have used retroviral vectors, but also adeno-associated virus vectors and a hemagglutinating virus of the Japan-liposome system have been applied.^27,28 Retrovirus-mediated gene transfer of the LAMB3 gene to cultured keratinocytes from patients with junctional EB (JEB) resulted in correct synthesis, processing and deposition of laminin 5 and restoration of assembly of the hemidesmosomes missing in JEB skin.²⁹ In another ex vivo study, the COL17A1 cDNA was used to restore BP180/collagen XVII protein synthesis in primary JEB keratinocytes.³⁰ Grafted keratinocytes regenerated skin in an immunodeficient mouse model. However subepidermal bulla formed and restoration was lost after 1 month.³⁰ Recombinant retroviral vectors have also been used successfully for in vivo gene transfer of a lacZ reporter gene to the epidermis. Expression of the reporter transgene was noted in immune tolerant mice up to 16 weeks but persisted only to 3 weeks in immune competent mice.¹⁷

As a first step towards developing a gene therapy approach to DEB, we have used transgenesis in mice to establish a model to test the feasibility of achieving tissue-specific and stable expression of human collagen VII by gene transfer. A transgenic mouse approach is especially important because it is possible to determine if the regulatory sequences of the gene vector are sufficient to drive appropriate tissue-specific expression. The importance of sequence elements for tissue-specific expression of genes in vivo is often inferred from co-transfection assays in cultured cells. While such studies are valuable, differing results may be obtained depending on whether transfection is stable or transient. It is also not possible to identify elements important for tissue- and stage-specificity of expression by this approach. Experiments using transgenic animals to recapitulate the normal pattern of gene expression have shown that in certain cases, such as elastase, (I) and 1 (II) collagen genes, that DNA sequences identified as essential for gene regulation in transfected cells may not have any functional role in vivo.^31,32,33

The activation and repression of gene expression are usually controlled at the level of transcription by multiple positive and negative regulatory elements which lie not only in 5' flanking sequences but may also be located within introns, exons and the 3' flanking DNA.^33,34,35 In addition, transgenic mouse studies have shown that gene expression of transgenes are facilitated and more appropriate by the inclusion of intron sequences (reviewed in Ref. ³⁶). Although Chen and others have proposed using a collagen VII 'minigene' with an internal 2 kb deletion for corrective gene transfer of collagen VII in DEB keratinocytes,³⁷ several reports on the genetic defects in DEB have clearly demonstrated that deletions in the COL7A1 gene are associated with a pathologic phenotype.^3,4,15,38 Therefore, despite technical problems with the delivery of large constructs, therapeutic approaches using a deleted gene are not likely to result in the desired normal phenotype. For these reasons we chose to isolate and use a genomic clone containing the complete COL7A1 gene rather than cDNA sequences in our study.

The tissue distribution of expression of human COL7A1 in the transgenic mice was more widespread than expected. However this expression of the transgene closely mimicked the endogenous mouse Col7a1 gene and similarly, COL7A1 mRNAs were found in equivalent human fetal tissues. These results are in broad agreement with immunohistochemical studies on adult human tissues, where low levels of collagen VII protein have been found in brain,²⁴ intestine,²² urinary bladder and prostate,^21,23 endothelial cells,²⁵ and basement membranes surrounding or underlying the epithelia of the larynx, oesophagus, trachea, vagina and lung carcinoma.^20,21 The widespread expression of COL7A1 in internal organs may be characteristic of the fetal and neonatal state and faithful reproduction of the expression pattern by the transgene is of importance since internal epithelia are also affected in DEB. Indeed, patients with severe DEB often have involvement of the oesophagus, larynx, upper respiratory tract and/or certain urogenital epithelia.³⁸

One major problem for keratinocyte gene therapy with retroviral vectors is inability to achieve sustained transgene expression in vivo.^39,40 The presence of human collagen VII in the skin of 19-month-old transgenic mice suggests that the expressing cells are not removed by immunocompetent mice or that synthesis continues over time, perhaps by stem cells. In a recent study in which human keratinocytes marked with a recombinant retroviral lacZ transgene were grafted to full thickness wounds in nude mice, -gal-positive cells which expressed keratinocyte markers persisted over a 40-week period, pointing to the existence of epidermal keratinocyte stem cells in the graft.⁴¹ Indeed, positive identification of keratinocyte stem cells and their stable transduction with genes for structural proteins would be optimal for gene therapy for a number of genodermatoses.^42,43,44

A prerequisite for demonstration of utility of the human COL7A1 gene vector for therapy is that appropriate tissue pattern of expression can be recapitulated in the transgenic animal. Our study shows for the first time that the human COL7A1 gene can be successfully expressed in the skin of transgenic mice and that the collagen VII protein, that is synthesized and processed appropriately, is stable and present continually along the basement membrane zone in detectable amounts 19 months after birth. Although the human and mouse collagen VII are very similar, with 85% amino acid identity,⁹ these small differences could have potentially destabilized the anchoring fibrils. It was encouraging that no adverse effects of expressing human collagen VII on skin integrity in mice were seen and appropriate processing of the protein was found.

Keratinocytes are believed to be the main source of collagen VII in vivo (see Refs ¹ and ⁴). However, under certain situations fibroblasts can be induced to produce this collagen, e.g. in cell culture or under pathological conditions such as severe scleroderma of the skin,⁴⁵ fibroblasts have been shown to express collagen VII. Unfortunately, keratinocyte cultures could not be initiated from the skin of aged mice any more, but we showed that synthesis of human collagen was sustained in fibroblasts derived from the skin of 19-month-old transgenic mice. This indicated long-term stable expression of the transgene, and it is possible that in vivo both cell types contribute to the synthesis of collagen VII and the assembly of the anchoring fibrils.

A successful gene therapy approach for DEB could involve the ex vivo transfection of the patient's cultured keratinocytes with COL7A1 gene vector, followed by grafting of the cells on to eroded areas of skin. For at least two reasons, this strategy of the phenotypic reversion of epidermal cells defective for collagen VII opens perspectives in the long-term treatment of recessive DEB. First, cultured keratinocytes are already used routinely to make autologous grafts for patients suffering from large skin or mucosal defects, e.g. after burn injuries.⁴⁶ Second, collagen VII is a component of highly ordered supramolecular structures in situ. After its secretion by the keratinocytes, collagen VII polymerizes into anchoring fibrils which are crosslinked by tissue transglutaminase in the extracellular matrix.⁴⁷ Anchoring fibrils and other post-translationally modified fibrillar suprastructures in the skin are very stable and have a slow turnover, a fact that is advantageous for long duration of the therapeutic effects.^1,48 Therefore, a relatively low level or medium-term expression of the transgene product may be sufficient for restoration of the epidermal-dermal adhesion for longer periods of time. This hypothesis is supported by molecular genetic studies on families with recessive DEB. Heterozygous carriers of COL7A1 null mutations synthesize only one allelic gene product, i.e. have only 50% of collagen VII in the skin but are phenotypically and functionally completely normal.^4,15 This observation suggests that the goal of high efficiency gene transfer that restores normal gene expression to all cells is not necessary for a corrective gene therapy of severe recessive DEB. Instead, a 50% level of normal collagen VII expression in the keratinocyte grafts should produce good therapeutic effects.

The successful restoration of hemidesmosomes in JEB keratinocytes²⁹ suggests that resurfacing blister wounds of DEB patients with artificial epidermis obtained from DEB keratinocytes stably transfected with a genomic COL7A1 vector could result in the formation of stable anchoring complexes. Alternatively, direct in vivo approaches by way of either intradermal injection^49,50 of genomic COL7A1 vector or topical administration of naked DNA to skin via hair follicles⁵¹ may result in adsorption and expression in the epidermis. The demonstration of successful and stable expression of human collagen VII protein in in vivo gene transfer in transgenic mice is the first step towards such a goal.

Materials and methods

Human fetal materials

Human tissues from a 21-week fetus were obtained from elective therapeutic terminations of pregnancy. Approval was obtained from the Ethical Committee of the Medical Faculty, The University of Hong Kong, for work with fresh embryonic materials.

Subcloned probes

5.2 kb BamHI and EcoRI human COL7A1 genomic fragment spanning from intron 2 to 19 (Figure 1a).

MC42EHc: 566 bp EcoRI and HincII mouse Col7a1 genomic fragment spanning from intron 2 to intron 3 (Figure 2c).

Oligonucleotide primers

The following pairs of oligonucleotide primers were used for polymerase chain reactions (PCR) and for reverse transcription PCR (RT-PCR). To facilitate cloning of the PCR products, an EcoRI adaptor (underlined) was added to the 5' end of some of the primers. (1) Amplification between human COL7A1 exon 9 and 12 in transgenic mouse tissues (Figure 2a) and human fetal tissues (Figure 2b): P1, specific to exon 9: 5'-GGGTTCAGTGTTGCTGCGTG-3' (nt 2945-2964); P2, specific to exon 12, for both PCR and RT-PCR): 5'-CTCGGTGGCTTGCAGGTCTG-3' (complementary to nt 3593-3612). (2) Amplification between mouse Col7a1 exon 1 to exon 2 in different mouse tissues (Figure 2c): PCR (R15): 5'-AAGAATTCTGCGCTGCAGAGATCCTGAT-3' (nt 31-50); RT-PCR (R14): 5'-AAGAATTCAATCCTCG CACTTCTCGAAA-3' (complementary to nt 660-679).

Isolation of genomic clones for human COL7A1

Hybridization conditions for library screening were at 65°C in 3 ´ SSC, 0.1% SDS, 12% polyethylene glycol 8000 (Sigma, St Louis, MO, USA), 250 g/ml heparin (Sigma) (1 ´ SCC: 0.15 M sodium citrate, pH 7.5). Post-hybridization washes of filters were to 0.1% SSC, 0.1% SDS at 65°C. The cosmid clone (cosHcol7) was isolated by screening a human cosmid library⁵² for the COL7A1 gene using a cDNA clone, pCG54, encoding the chicken 1 (I) collagen gene.⁵³ Sequence analysis of a subclone of cosHcol7 which hybridized to pCG54 showed that it contained sequences encoding the 'hinge' region of collagen VII (a short 39 amino acid non-Gly-X-Y sequence interruption of the triple helical domain,⁷ (data not shown). A combination of Southern blot hybridization analyses using oligonucleotide primers and subclones (a gift of Dr J Uitto, Jefferson University, USA) as probes showed that cosHcol7 contained sequences extending 400 bp 5' to exon 4 and 8 kb beyond the 3' end of the gene (data not shown). Further human genomic clones including HC7C61 were isolated using the subclone of cosHcol7, encoding the hinge region of collagen VII, to screen a human acute lymphocytic leukemia cell line (HPBALL) genomic library.⁵⁴

Generation of transgenic mice

Transgenic mice were produced by pronuclear injection of HC7C61 DNA linearized at the ClaI site, into one-cell zygotes of F1 (CBA ´ C57BL6) hybrid mice as described.⁵⁵ Transgenic mice were identified by PCR using oligonucleotide primers P1 and P2 on genomic DNA isolated from tail tips (data not shown). PCR reactions, containing 2.5 U Taq polymerase (Life Technologies, Hong Kong, China), were denatured at 94°C for 3 min, followed by 30 cycles of 94°C for 1-min denaturing, 55°C for 1-min annealing and 72°C for 1-min elongation. To ensure that there was no rearrangement of the transgene, genomic DNA isolated from the transgenic mice was analyzed by Southern blot analysis with HC7C61 as probe (Figure 1). The hybridization and post-hybridization conditions were the same as described for library screening (see above).

RNA isolation and expression analyses

Total human and mouse RNAs were prepared by the lithium chloride-urea differential precipitation method.⁵⁶ Transgenic mouse RNAs were isolated from 19-month-old mouse T61 and its non-transgenic littermate. Wild-type mouse RNAs were isolated from E16.5 fetuses and 3-day-old neonate different tissues. Human RNAs were isolated from 21-week human fetal tissues. Expression of the human and mouse COL7A1/Col7a1 gene was assessed by RT-PCR using primers P1 and P2 (human) and R14 and R15 (mouse). For RT-PCR, first-strand cDNA was transcribed from 5 g of total human RNA or total mouse RNA using Superscript II RNase H-Reverse Transcriptase (200 units, Life Technologies) with a corresponding antisense primer (see above) in manufacturer's buffer at 45°C for 1 h. Second strand DNA synthesis was carried out with 1/20 of the RT product at 72°C for 30 min with the corresponding 5' sense primer (see above) with Taq DNA polymerase (5 units, Taq DNA polymerase, recombinant, Life Technologies). PCRs were carried out with 1-min denaturation reaction at 94°C, 1-min annealing reaction ranging from 54°C to 58°C depending on the Tm of primers, 1-min extension reaction at 72°C for 30 cycles. As controls, PCRs were performed with the same set of RNAs but without reverse transcriptase to test for genomic contamination. The identities of the RT-PCR products were analyzed by Southern blotting using specific probes for human gene: pBE, and for mouse: MC42EHc for RT-PCR products of P1-P2, and R14-R15, respectively.

Immunostaining

Human fetal skin, wild-type and transgenic E16.5 mouse fetuses and skin biopsies from a 19-month transgenic mouse and its non-transgenic littermate were snap-frozen in liquid nitrogen and processed for immunofluorescence or -peroxidase staining. Immunostainings were carried out on 5-7-m cryosections with standard techniques, using an antibody to human collagen VII that did not cross-react with mouse. For the human fetal skin and E16.5 mouse fetuses, the cryosections were air dried, rinsed with PBS and incubated at 4°C for 16 h with 1.3a-VII (diluted 1:2 in PBS/BSA), a rabbit-anti-human, polyclonal antibody against human collagen VII.⁵⁷ After washing in PBS, goat anti-rabbit fluorescein-conjugated secondary antibody (Vector Laboratories, Burlingame, CA, USA), diluted 1:50 in PBS/BSA, was applied for 30 min at room temperature. After further washing in PBS, the sections were mounted with Vector Shield (Vector Laboratories) and photographed under a fluorescence microscope (Zeiss Axioskop; Zeiss, Jena, Germany) using Kodak Ektachrome ASA400 (Kodak, Rochester, New York, USA). For 19-month mouse skin, sections were incubated with antibodies to recombinant NC-1 domain of human collagen VII overnight at room temperature and with the EnVision+ peroxidase-labelled goat-anti-rabbit second antibodies (Dako, Carpinteria, CA, USA) for 2 h.

Skin fibroblasts of a 19-month-old transgenic mouse and a normal littermate were initiated from skin biopsies with standard techniques using the outgrowth method and DMEM (Life Technologies) with 10% fetal calf serum. Primary skin fibroblasts were trypsinized, seeded on glass coverslips and grown to subconfluency. They were then cultured for 48 h in the presence of 50 g/ml of ascorbic acid to enhance collagen synthesis. After fixation with methanol at -20°C for 15 min the cells were reacted with antibodies to human collagen VII and peroxidase-labeled second antibodies as described above.

Protein extraction and immunoblot analyses

For extraction of collagen VII from mouse skin, the epidermis and dermis were separated with incubation in a buffer containing 1 M NaCl overnight.⁵⁸ The dermis was extracted with a neutral buffer containing 8 M urea, 2% SDS, 0.020 M Tris-HCl, pH 6.8 and a mixture of proteinase inhibitors.¹⁴ The proteins were separated on SDS-PAGE using gels with 4.5-15.0% polyacrylamide gradients and transferred on to nitrocellulose. The samples were subjected to immunoblotting using human-specific collagen VII antibodies and alkaline phosphatase-linked anti-rabbit second antibodies (Sigma, Deisenhofen, Germany).

Acknowledgements

This study was supported by research grants from the Dystrophic Epidermolysis Bullosa Research Association (DEBRA), The University of Hong Kong Committee for Research and Conference Grants, the Hong Kong Research Grants Council-DAAD Joint Research Scheme and the German Research Council (DFG, grants Br 1475/2-3 and SFB 492/A3). We are also very grateful for skilled technical assistance from Sandra YY Wong and Margit Schubert.

1 Burgeson RE. Type VII collagen, anchoring fibrils and epidermolysis bullosa. J Invest Dermatol 1993; 101: 252-255,

2 Shimizu H et al. Immunohistochemical, ultrastructural, andmolecular features of Kindler syndrome distinguish it from dystrophic epidermolysis bullosa. Arch Dermatol 1997; 133: 1111-1117,

3 Bruckner-Tuderman L. Hereditary skin diseases of anchoring fibrils. J Dermatol Sci 1999; 20: 122-133,

4 Bruckner-Tuderman L, Hopfner B, Hammami-Hauasli N. Biology of anchoring fibrils: lessons from dystrophic epidermolysis bullosa. Matrix Biol 1999; 18: 43-54,

5 Bruckner-Tuderman L et al. Immunohistochemical and mutation analyses demonstrate that procollagen VII is processed to collagen VII through removal of the NC-2 domain. J Cell Biol 1995; 131: 551-559,

6 Christiano AM et al. Structural organization of the human type VII collagen gene (COL7A1), composed of more exons than any previously characterized gene. Genomics 1994; 21: 169-179, Article MEDLINE

7 Christiano AM, Greenspan DS, Lee S, Uitto J. Cloning of human type VII collagen. Complete primary sequence of the alpha 1(VII) chain and identification of intragenic polymorphisms. J Biol Chem 1994; 269: 20256-20262,

8 Li K et al. cDNA cloning and chromosomal mapping of the mouse type VII collagen gene (Col7a1): evidence for rapid evolutionary divergence of the gene. Genomics 1993; 16: 733-739,

9 Kivirikko S, Li K, Christiano AM, Uitto J. Cloning of mouse type VII collagen reveals evolutionary conservation of functional protein domains and genomic organization. J Invest Dermatol 1996; 106: 1300-1306,

10 McGrath JA et al. Structural variations in anchoring fibrils in dystrophic epidermolysis bullosa: correlation with type VII collagen expression. J Invest Dermatol 1993; 100: 366-372,

11 Heagerty AH et al. Identification of an epidermal basement membrane defect in recessive forms of dystrophic epidermolysis bullosa by LH 7:2 monoclonal antibody: use in diagnosis. Br J Dermatol 1986; 115: 125-131,

12 Leigh IM et al. Type VII collagen is a normal component of epidermal basement membrane, which shows altered expression in recessive dystrophic epidermolysis bullosa (published erratum appears in J Invest Dermatol 1989; 92: 135). J Invest Dermatol 1988; 90: 639-642,

13 Bruckner-Tuderman L et al. Lack of type VII collagen in unaffected skin of patients with severe recessive dystrophic epidermolysis bullosa. Dermatologica 1988; 176: 57-64,

14 Bruckner-Tuderman L, Mitsuhashi Y, Schnyder UW, Bruckner P. Anchoring fibrils and type VII collagen are absent from skin in severe recessive dystrophic epidermolysis bullosa. J Invest Dermatol 1989; 93: 3-9,

15 Pulkkinen L, Uitto J. Mutation analysis and molecular genetics of epidermolysis bullosa. Matrix Biol 1999; 18: 29-42,

16 Heinonen S et al. Targeted inactivation of the type VII collagen gene (Col7a1) in mice results in severe blistering phenotype: a model for recessive dystrophic epidermolysis bullosa. J Cell Sci 1999; 112: 3641-3648,

17 Ghazizadeh S, Harrington R, Taichman L. In vivo transduction of mouse epidermis with recombinant retroviral vectors: implications for cutaneous gene therapy. Gene Therapy 1999; 6: 1267-1275,

18 Uitto J, Chung Honet LC, Christiano AM. Molecular biology and pathology of type VII collagen. Exp Dermatol 1992; 1: 2-11,

19 Uitto J, Christiano AM. Dystrophic forms of epidermolysis bullosa. Semin Dermatol 1993; 12: 191-201,

20 Wetzels RHW et al. Distribution patterns of type VII collagen in normal and malignant human tissues. Am J Pathol 1991; 139: 451-459,

21 Leigh I, Purkis PE, Bruckner-Tuderman L. LH 7.2 monoclonal antibody detects type VII collagen in the basement membranes of ectodermally derived epithelia including skin. Epithelia 1987; 1: 17-29,

22 Leivo I et al. Anchoring complex components laminin-5 and type VII collagen in intestine: association with migrating and differentiating enterocytes. J Histochem Cytochem 1996; 44: 1267-1277,

23 Nagle RB et al. Expression of hemidesmosomal and extracellular matrix proteins by normal and malignant human prostate tissue. Am J Pathol 1995; 146: 1498-1507,

24 Paulus W et al. Expression of type VII collagen, the major anchoring fibril component, in normal and neoplastic human nervous system. Virch Arch 1995; 426: 199-202,

25 Ryynanen J et al. Type VII collagen gene expression in human umbilical tissue and cells. Lab Invest 1993; 69: 300-304,

26 Khavari PA. Gene therapy for genetic skin disease. J Invest Dermatol 1998; 110: 462-467, Article MEDLINE

27 Sawamura D et al. In vivo transfer of a foreign gene to keratinocytes using the hemagglutinating virus of Japan-liposome method. J Invest Dermatol 1997; 108: 195-199, MEDLINE

28 Braun-Falco M, Doenecke A, Smola H, Hallek M. Efficient gene transfer into human keratinocytes with recombinant adeno-associated virus vectors. Gene Therapy 1999; 6: 432-441, MEDLINE

29 Vailly J et al. Corrective gene transfer of keratinocytes from patients with junctional epidermolysis bullosa restores assembly of hemidesmosomes in reconstructed epithelia. Gene Therapy 1998; 5: 1322-1332,

30 Seitz CS et al. BP180 gene delivery in junctional epidermolysis bullosa. Gene Therapy 1999; 6: 42-47,

31 Sokolov BP et al. Tissue-specific expression of the gene for type I procollagen (COL1A1) in transgenic mice. Only 476 base pairs of the promoter are required if collagen genes are used as reporters. J Biol Chem 1995; 270: 9622-9629,

32 Swift GH, Kruse F, MacDonald RJ, Hammer RE. Differential requirements for cell-specific elastase I enhancer domains in transfected cells and transgenic mice. Genes Dev 1989; 3: 687-696,

33 Leung KKH et al. Different cis-regulatory DNA elements mediate developmental stage- and tissue-specific expression of the human COL2A1 transgene. J Cell Biol 1998; 141: 1291-1300,

34 Darnell JEJ. Variety in the level of gene control in eukaryotic cells. Nature 1982; 297: 365-371,

35 Gutman A, Gilthorpe J, Rigby PW. Multiple positive andnegative regulatory elements in the promoter of the mouse homeobox gene Hoxb-4. Mol Cell Biol 1994; 14: 8143-8154,

36 Bonifer C et al. Prerequisites for tissue specific and position independent expression of a gene locus in transgenic mice. J Mol Med 1996; 74: 663-671,

37 Chen M et al. Corrective gene transfer in dystrophic epidermolysis bullosa. J Invest Dermatol 1999; 112: 552 (Abstr. 175),

38 Fine JD, Bauer EA, McGuire J, Moshell A. Epidermolysis Bullosa: Clinical, Epidemiologic and Laboratory Advances, and the Findings of the National Epidermolysis Bullosa Registry. Johns Hopkins University Press: Baltimore, 1999,

39 Deng H, Lin Q, Khavari PA. Sustainable cutaneous gene delivery. Nat Biotechnol 1997; 15: 1388-1391, MEDLINE

40 Krueger GG, Morgan JR, Petersen MJ. Biologic aspects of expression of stably integrated transgenes in cells of the skin in vitro and in vivo. Proc Assoc Am Phys 1999; 111: 198-205,

41 Kolodka TM, Garlick JA, Taichman LB. Evidence forkeratinocyte stem cells in vitro: long term engraftment and persistence of transgene expression from retrovirus-transduced keratinocytes. Proc Natl Acad Sci USA 1998; 95: 4356-4361, MEDLINE

42 Pellegrini G, Bondanza S, Guerra L, De Luca M. Cultivation of human keratinocyte stem cells: current and future clinical applications. Med Biol Eng Comput 1998; 36: 778-790,

43 Bickenbach JR, Roop DR. Transduction of a preselected population of human epidermal stem cells: consequences for gene therapy. Proc Assoc Am Phys 1999; 111: 184-189,

44 Cotsarelis G et al. Epithelial stem cells in the skin: definition, markers, localization and functions. Exp Dermatol 1999; 8: 80-88,

45 Rudnicka L et al. Elevated expression of type VII collagen in the skin of patients with systemic sclerosis. Regulation bytransforming growth factor-beta. J Clin Invest 1994; 93: 1709-1715.

46 Gobet R et al. Efficacy of cultured epithelial autografts inpediatric burns and reconstructive surgery. Surgery 1997; 121: 654-661,

47 Raghunath M et al. Cross-linking of the dermo-epidermal junction of skin regenerating from keratinocyte autografts. Anchoring fibrils are a target for tissue transglutaminase. J Clin Invest 1996; 98: 1174-1184,

48 Prockop DJ, Kivirikko KI. Collagens: molecular biology, diseases, and potentials for therapy. Annu Rev Biochem 1995; 64: 403-434, MEDLINE

49 Hengge UR et al. Efficient expression of naked plasmid DNA in mucosal epithelium: prospective for the treatment of skin lesions. J Invest Dermatol 1998; 111: 605-608, MEDLINE

50 Vogel JC. A direct in vivo approach for skin gene therapy. Proc Assoc Am Phys 1999; 111: 190-197, MEDLINE

51 Fan H, Lin Q, Morrissey GR, Khavari PA. Immunization via hair follicles by topical application of naked DNA to normal skin. Nat Biotechnol 1999; 17: 870-872, Article MEDLINE

52 Weiss EH et al. Isolation and characterization of a human collagen alpha 1(I)-like gene from a cosmid library. Nucleic Acids Res 1982; 10: 1981-1994,

53 Fuller F, Boedtker H. Sequence determination and analysis of the 3' region of chicken pro-alpha 1(I) and pro-alpha 2(I) collagen messenger ribonucleic acids including the carboxy-terminal propeptide sequences. Biochemistry 1981; 20: 996-1006,

54 Kioussis D et al. Expression and rescuing of a cloned human tumour necrosis factor gene using an EBV-based shuttle cosmid vector. EMBO J 1987; 6: 355-361,

55 Hogan B, Costantini F, Lacy E. Manipulating the Mouse Embryo: A Laboratory Manual. Cold Spring Harbor Laboratory: New York, 1986,

56 Lovell-Badge RH. Introduction of DNA into embryonic stem cells. In: Robertson EJ (ed). Teratocarcinomas and Embryonic Stem Cells - A Practical Approach. IRL Press: Oxford, 1987, pp 153-182.

57 Bruckner-Tuderman L, Schnyder UW, Winterhalter KH, Bruckner P. Tissue form of type VII collagen from human skin and dermal fibroblasts in culture. Eur J Biochem 1987; 165: 607-611,

58 Hintner H et al. Immunofluorescence mapping of antigenic determinants within the dermal-epidermal junction in the mechanobullous diseases. J Invest Dermatol 1981; 76: 113-118,

Figure 1 The human COL7A1 gene cloned in one piece and intact in T61 transgenic mice. (a) Schematic representation of the cosmid HC7C61 containing an entire human COL7A1 gene and its 3 kb 5' and 4 kb 3' flanking sequences (hatched box). The location of probe pBE is shown as a filled box. (b) Southern blot analysis of genomic DNAs from non-transgenic (NT), transgenic (T) mouse line T61, human (Hu) and HC7C61 cosmid (Co) digested with EcoRI and HindIII and hybridized with whole HC7C61 as probe. Note identical pattern of hybridized fragments for T compared with Hu and with the HC7C61 cosmid, except bands (arrowheads) correspond to the vector fragments. Cross hybridizing fragments in NT differ in size from those of Hu and probably reflect sequence homology between the endogenous mouse Col7a1 and human COL7A1 genes. E, EcoRI; C, ClaI; H, HindIII; ST, start of transcription; Amp, amplicillin resistant gene; Hyg, hygromycin resistant gene; Mr, HindIII digested DNA size marker in kilobases (kb). The copy number of the transgene in T61 was determined to be three (data not shown).

Figure 2 Expression of the human COL7A1 transgene (HC7C61) in transgenic mice compared with the endogenous human and mouse genes at fetal and neonatal stages. (a) Expression pattern of HC7C61 in the 19-month-old transgenic mouse T61. (b) Expression pattern of the human COL7A1 gene in 21-week-old human fetus. The RT-PCRs of both (a) and (b) were performed with P1 (exon 9) and P2 (exon 12) as primers and Southern blots probed with pBE spanning intron 2 to intron 19 (see Figure 1a). The expected PCR fragment size for COL7A1 cDNA is 423 bp. The 667 bp amplified fragment is from genomic DNA present in the samples. (c) Expression patterns of the mouse Col7a1 gene in 3.0 dpp mouse and E16.5 mouse fetus. Southern blot of RT-PCR products obtained using primers R15 (exon 1) and R14 (exon 2) and MC42EHc (filled box) as probe. Amplified fragments representing cDNA and genomic DNA can be distinguished by size, being 152 bp and 625 bp respectively. The relative positions of primers used for RT-PCR are shown in the schematic diagrams in the upper panels of each section of (a-c). Exons are represented as open numbered boxes. Middle panels in (a-c) show the RT-PCR products produced using specific primers and 5 g of total RNA from various tissues (see above), and electrophoresed in 2% agarose gel. The lower panel of (a-c) is the corresponding Southern blot analyses hybridized with specific probes (see above). The amplification of genomic fragments in the -RT PCR control but not in the +RT PCR, may be because in the control, only genomic DNA in the RNA sample can be used as substrate in the PCR, while in the +RT-PCR, the cDNA used as template would be present in higher amounts than the genomic DNA and therefore would be preferentially amplified. Ey, eye; Sp, spleen; Br, brain; Hr, heart; Ki, kidney; Li, liver; Lu, lung; Sk, skin; T, transgenic; NT, non-transgenic; H₂O, water was used as a negative control; C, no RT product was added in the PCR. +, reaction with RT; -, control reaction without RT; E, EcoRI; Hc, HincII.

Figure 3 Expression of human COL7A1 transgene (HC7C61) product in fetal transgenic mice. Left panels (a, c, e, g and i) are bright-field images and right panels (b, d, f, h and j) are dark-field images of FITC analysis using polyclonal antibodies against human collagen VII. (a and b) Cross-sections of human adult skin. A strong signal was found at the junction (arrowhead) between epidermis (ep) and dermis (de). No signal was found in the non-transgenic mouse skin at E16.5 (c and d), confirming the specificity of the antibody. (e and f) Cross-sections of transgenic mouse skin at E16.5. A strong signal was found at the junction between epidermis and dermis (arrowhead) and around the hair follicles (asterisks). (g, h and i, j) Cross-sections of non-transgenic and transgenic mouse tongue region at E16.5, respectively. A weak signal was found at the tongue surface lining (arrowheads). Scale bar, 50 m.

Figure 4 Sustained expression of human COL7A1 transgene (HC7C61) product in adult (19 months old) transgenic mice. (a) Immunostaining of mouse skin with antibodies to collagen VII. Panel i, transgenic mouse skin stained with antibodies to mouse collagen VI as a positive control. Note the strong staining of the dermal-epidermal junction. Panel ii, staining of the skin of a non-transgenic littermate with antibodies to human collagen VII remained negative. Panels iii and iv, transgenic mouse skin stained with antibodies to human collagen VII shows a positive signal at the epidermal-dermal basement membrane zone. (b) Immunoblotting of human collagen VII. Human and mouse dermal extracts were subjected to immunoblotting with species-specific antibodies recognizing the human, but not the mouse, collagen VII. Lane 1, dermis extract from the transgenic mouse contained mature human collagen VII with an apparent molecular weight of 290 kDa; lane 2, in the dermis extract of the non-transgenic littermate, no collagen VII was detected; lane 3, human dermis extract showed a strong immunopositive collagen VII band; lane 4, extract of cultured human keratinocytes contained unprocessed procollagen VII with an apparent molecular weight of 320 kDa. (c) Immunostaining of mouse skin fibroblasts: i, transgenic mouse fibroblasts stained with antibodies to mouse collagen VII; ii, fibroblasts of a non-transgenic littermate stained with antibodies to human collagen VII; iii and iv, transgenic mouse fibroblasts stained with antibodies to human collagen VII. Note the strong immunopositive signal in some but not all cells. The intracellular staining reflects active biosynthesis of human collagen VII. The secreted molecules cannot be visualized, since they diffuse into the culture medium. Scale bars, 250 m for (ai-ii); 50 m for (aiii-iv) and 5 m for (ci-iv).