Single gene recessive genetic skin disorders offer attractive prototypes for the development of therapeutic cutaneous gene delivery. We have utilized X-linked ichthyosis (XLI), characterized by loss of function of the steroid sulfatase arylsulfatase C (STS), to develop a model of corrective gene delivery to human skin in vivo. A new retroviral expression vector was produced and utilized to effect STS gene transfer to primary keratinocytes from XLI patients. Transduction was associated with restoration of full-length STS protein expression as well as steroid sulfatase enzymatic activity in proportion to the number of proviral integrations in XLI cells. Transduced and uncorrected XLI keratinocytes, along with normal controls, were then grafted onto immunodeficient mice to regenerate full thickness human epidermis. Unmodified XLI keratinocytes regenerated a hyperkeratotic epidermis lacking STS expression with defective skin barrier function, effectively recapitulating the human disease in vivo. Transduced XLI keratinocytes from the same patients, however, regenerated epidermis histologically indistinguishable from that formed by keratinocytes from patients with normal skin. Transduced XLI epidermis demonstrated STS expression in vivo by immunostaining as well as a normalization of histologic appearance at 5 weeks post-grafting. In addition, transduced XLI epidermis demonstrated a return of barrier function parameters to normal. These findings demonstrate corrective gene delivery in human XLI patient skin tissue at both molecular and functional levels and provide a model of human cutaneous gene therapy.
Disorders of epidermal cornification comprise a heterogeneous disease family. These diseases are characterized by abnormalities in epidermal differentiation and range in severity from the common and mild ichthyosis vulgaris, with a frequency of ~1 in 300 (1 ), to severe Harlequin ichthyosis, which may be lethal in infancy (2 ). X-linked ichthyosis (XLI) (Mendelian Inheritance in Man No. 308100) has an incidence of ~1 per 6000 males and is characterized by generalized brown hyperkeratotic skin scaling, along with several other associated findings including opacities of the cornea and occasionally cryptorchidism (3 ,4 ). The disease has an onset in infancy and can be disfiguring. Molecular analysis and subsequent molecular cloning revealed that the steroid sulfatase arylsulfatase C (STS) gene is completely deleted in ~90% of XLI patients, with partial deletions or point mutations found in the majority of the remainder (5 ,6 ). Steroid sulfatase activity has been postulated as essential for normal desquamation of differentiated stratum corneum keratinocytes, and its loss is believed to result in the abnormal retention ichthyosis seen in XLI. While not as severe as several other disorders of epidermal differentiation, such as Harlequin fetus and lamellar ichthyosis, XLI can be disfiguring and markedly impacts the quality of life of XLI patients. Current treatments are variably effective and only palliative. This recessive disorder offers an opportunity to develop models of corrective epidermal gene delivery.
Gene therapy efforts to date in human trials have been frustrated by a number of key challenges, several of which have come to light only when gene transfer was performed in human tissues (7 ,8 ). In vitro transient transfections with corrective genes to human cells derived from patients with genetically defined diseases (9 ) may be a necessary phase in the development of therapeutic gene delivery approaches. This initial in vitro step, however, fails to produce either the necessary information regarding gene expression effects in three-dimensionally regenerated human tissue in vivo or the cellular reagents needed to advance the process in the case of ex vivo gene therapy efforts. The skin offers an attractive intermediate step between in vitro studies and full application to human subjects. Human skin may be regenerated on immunodeficient nude or SCID mice by a number of grafting methods refined in the treatment of burn and cutaneous ulcer patients over the last 25 years (10 -13 ). Such regenerated human skin has been shown to be normal at the clinical, histologic and molecular levels (13 ). Because skin keratinocytes are readily subjected to growth and gene transfer in vitro prior to such grafting, this immunodeficient mouse-human skin xenograft model offers an attractive animal system for therapeutic gene delivery to human epidermis prior to actual use in humans. In this report, we have corrected primary keratinocytes from two unrelated patients with XLI that completely lack STS protein expression. We have then utilized these corrected cells to regenerate histologically and functionally normal human epidermis in vivo on immunodeficient mice as the next step in the rational development of approaches for therapeutic cutaneous gene delivery in humans.
Lasting correction of certain genetic skin disease may require effective levels of gene expression and long-term gene retention, such as via integration of therapeutic DNA in the genomes of recipient cells. A previous effort at gene transfer in XLI in vitro (9 ) utilized episomal plasmids, an approach characterized by transgene loss within several days in vitro (14 ). Because direct in vivo methods such as particle bombardment (15 ) and direct DNA injection (16 ,17 ) yield a low efficiency and transient transgene expression that is likely to be incapable of uniformly restoring epidermal architecture across a broad region of skin, we chose to begin our efforts with XLI by using an ex vivo, integrating gene transfer approach. Early passage keratinocytes from two XLI patients lacking in STS protein and enzymatic activity expression were grown in culture and transduced with an amphotropic retroviral vector for STS or for [beta]-galactosidase marker control (18 ). The retroviral expression vector for human STS was constructed using the full-length STS cDNA minus 3' polyadenylation sequences (Fig. 1 ). This vector, as well as the LZRS [beta]-galactosidase control vector (18 ), was used to transduce early passage XLI cells. XLI cells, along with normal controls, were studied for STS protein expression by immunoblot analysis using antisera to human STS. Retroviral transduction was associated with a restoration of full-length STS protein expression (Fig. 2 ). Available antisera do not allow sensitive determination of STS expression on a cell by cell basis in keratinocytes grown in vitro in monolayer cultures (R.A. Freiberg, unpublished observations), therefore, a lacZ reference marker virus (18 ) was utilized to examine the efficiency of retroviral gene transfer to primary XLI keratinocytes in vitro. Transduction of XLI patients cells could be achieved at high efficiency by this general approach, as judged by comparison of [beta]-galactosidase marker gene expression with background staining from untransduced cells (Fig. 3 ). RT-PCR of retroviral supernatant, filtered to eliminate contaminating packaging cells, confirmed equal or greater levels of viral genome for STS and lacZ vectors, suggesting comparable titers for the two vectors. These data confirm production of a transduced XLI keratinocyte population in vitro.
After gene transfer, we wished to determine if steroid sulfatase enzymatic activity had been restored to these primary XLI patient cells. Steroid sulfatase enzymatic activity is believed to be necessary for physiologic desquamation, and we next examined transduced cells for this biochemical measure of STS restoration. Cell extracts were made from transduced and untreated XLI cells along with normal controls, and steroid sulfatase activity was determined (19 ). Steroid sulfatase enzymatic activity in optimized transduction conditions could be restored from low background levels to the range of that seen in keratinocytes from patients with normal skin (Fig. 4 ). STS protein expression and enzymatic activity were directly proportional to the efficiency of proviral integration (data not shown), consistent with previous work with retroviral long terminal repeat (LTR)-driven gene expression in keratinocytes (20 ). These data suggested that gene transfer via the retroviral expression vector above restored full-length, enzymatically active STS to XLI patient keratinocytes.
After effecting STS gene transfer in vitro to early passage XLI patient keratinocytes and verifying restoration of full-length STS protein and steroid sulfatase enzymatic activity, we wished to utilize these cells in a model of ex vivo corrective gene delivery. Transduced and untreated XLI keratinocytes, along with control cells from patients with normal skin, were grafted to immunodeficient nude mice in order to regenerate full thickness human epidermis in vivo (13 ). Hyperkeratosis, a thickening of the stratum corneum, is the central histologic abnormality in XLI. In order to determine the impact of STS gene transfer on this histologic hallmark of XLI, we obtained biopsy specimens from regenerated epidermal xenografts and stained tissue sections with hematoxylin and eosin. As shown in Figure 5 a, untransduced XLI keratinocytes regenerated a hyperkeratotic stratum corneum, effectively recapitulating the XLI architectural abnormality in regenerated skin in vivo. Transduced XLI keratinocytes, however, regenerated epidermis with normal relative stratum corneum thickness (Fig. 5 a). Immunostaining verified the restoration of STS expression in vivo in suprabasal epidermis regenerated from transduced XLI keratinocytes, while untransduced XLI cells from the same patient failed to demonstrate significant STS staining above background reactivity (Fig. 5 b). These data indicate that STS gene transfer in vitro, followed by grafting in vivo, produced XLI patient epidermis with restored epidermal STS protein expression that lacked the central histologic abnormality of the disease.
Perturbations of the normal program of epidermal gene expression are responsible for a wide variety of inherited and acquired skin diseases characterized by abnormal epidermal differentiation. The monogenic skin disorders offer attractive models for gene reversion attempts in intact human epithelium where a number of the major problems that also confront therapeutic gene delivery in a range of tissues may be addressed. XLI offers a model to begin to address some of these problems facing gene therapy efforts in humans to date. In order to begin to develop this approach, we have described results of corrective gene transfer, both in vitro and in vivo, with early passage XLI patient keratinocytes. STS expression in our primary XLI cells was proportional to the number of proviral integrations, consistent with earlier work (20 ). Although LTR-driven transgene expression has been shown to persist in vitro for prolonged periods (20 ,22 ), transgene expression has been shown to be lost within 7-35 days in vivo (22 -24 ). We have observed this loss, which may be due to viral promoter inactivation (25 ), with a variety of delivered genes (ref. 26 ; K.Choate and P.A. Khavari, unpublished), and in the present work have only demonstrated STS expression up to 5 weeks and not beyond that timepoint. Sustaining transgene expression in skin in vivo, then, represents a major potential challenge in refining therapeutic cutaneous gene delivery. The molecular, architectural and functional correction achieved in XLI patient epidermis, however, offers support for the potential feasibility of certain aspects of this general approach for future application to gene therapy efforts in humans. The precise dosage range of therapeutic gene expression, the achievement of long-term, lifelong correction and the potential need for new regulated gene delivery vectors in such therapeutic gene transfer to the skin will be among required foci of future work in the further development of this process.
XLI is characterized by a normal rate of epidermal proliferation but a delayed dissolution of the desmosomes in the cornified layer, and a disruption in the normal orderly kinetics of epidermal desquamation. This disruption is believed to underlie the clinical hyperkeratosis and cracking, scaling skin associated with defective cutaneous barrier function (27 ). TEWL, a measure of skin permeability, is elevated in settings of experimentally induced defective skin barrier function and is also abnormally elevated in a range of disorders of cornification, including XLI (21 ), reflecting the defective barrier function of the hyperkeratotic stratum corneum (27 ). It is not clear from this study what minimum proportion of cells in the epidermis require STS restoration in order to produce functionally normal epidermis in humans. The capability for high efficiency gene transfer, however, may prove helpful in a number of other disorders. This may be the case in lamellar ichthyosis, due to a loss of intracellular transglutaminase 1, where local epidermal disruption due to incomplete gene correction may be adequate to disrupt barrier function and trigger the disease process. The ability to titrate the proportion of corrected input cells for epidermal regeneration via control of gene transfer efficiency offers an avenue for exploring the impact of this issue in a range of genetic skin diseases in vivo. By demonstrating that disease features are dependent on keratinocyte STS production in vivo, these studies suggest that the key site of STS production in maintenance of normal epidermal phenotype is in the epidermis itself, rather than via dermal fibroblasts which make significant amounts of STS in normal skin. This is supported by the fact that the widespread ingrowth of murine fibroblasts into regenerated graft tissue, which occurs within several weeks in this model, does not correct the phenotype of XLI[-] epidermis. While the STS gene dosage and proportion of STS[+] keratinocytes required for normal morphology of the epidermal stratum corneum remains unknown, it appears that local epidermal STS activity is a necessary requirement.
The ex vivo approach used here would require extensive and costly efforts in skin grafting in corrective cutaneous gene delivery efforts in humans. While such an approach may be a tolerable necessity in initial therapeutic efforts in more severe genetic skin disorders, such as lethal subtypes of epidermolysis bullosa, widespread clinical applications may be facilitated by less traumatic methods of gene delivery. Some emerging approaches to cutaneous gene delivery include direct administration of naked plasmid DNA (28 ), an approach that has been successful recently in grafted human skin (29 ). Such non-viral episomal approaches, however, are currently limited by the low efficiency and transient nature of gene transfer. Improvement of gene delivery methods to skin is an important goal in bringing cutaneous gene therapy to broad clinical application. In a number of genetic disorders, including XLI, gene therapy will involve delivery of a previously absent gene product that may be recognized as a foreign antigen by the immune system. Such a recognition could lead to immunologic destruction of the therapeutic protein and the cells producing it; however, recent advances in induction of immune tolerance may help in circumventing this serious potential problem. These advances, which have recently allowed allogeneic and xenogeneic transplantation of tissues, including skin, utilize a range of strategies, including blockade of CD28 and CD40 pathways (30 ), antibodies to CD45RB (31 ) and other promising approaches (32 ). While initial transient transfections in vitro (9 ) have been an important first step in the process of developing therapeutic cutaneous gene delivery in vivo, the transient nature and low efficiency of correction, combined with the lack of information on utility of gene restoration to patient cells in tissues in vivo, limit its usefulness. In addition, keratinocytes transiently transfected with expression plasmids by a variety of methods (16 ,33 ) lose gene expression in 2-7 days in vitro (9 ) (H. Duh and P. Khavari, unpublished) and in vivo (16 ,17 ), thus currently rendering this approach less attractive for use in efforts at lasting corrective gene delivery in genetic skin disease. Furthermore, the gene transfer efficiency of such transient transfections is low and, therefore, further compromises their utility in efforts to restore normal tissue architecture. In contrast, genomic integration of therapeutic transgenes via retroviral vectors is capable of sustaining expression long enough to regenerate normal skin tissue in vivo. The latter method, however, will require further improvement to produce truly durable transgene expression. Here we have provided a prototype of cutaneous gene therapy in vivo as a basis for future efforts to refine molecular therapy of human genetic skin disease.
Clinical features (34 ), histologic analysis of skin biopsy specimens (3 ) and abnormally low levels of steroid sulfatase activity were utilized to diagnose two unrelated male patients with XLI at the V.A. Palo Alto Health Care System. Punch biopsies (6 mm) were obtained from these two XLI patients in accordance with IRB-approved human subjects protocols of Stanford University and the V.A. Palo Alto Health Care System. A portion of each biopsy was utilized for immunostaining, with the remainder used for cell growth in tissue culture as previously described (35 ).
The full-length human STS cDNA, minus 3' polyadenylation sequences, was subcloned into the MFG-based episomal LZRS vector (18 ). The cDNA insert extended from bp 1968 to 4256 of the published STS sequence (36 ) and was subcloned as an EcoRI-NheI blunt fragment into blunted EcoRI sites of the LZRS retroviral expression plasmid. The resulting construct and the LZRS [beta]-galactosidase control marker vector (18 ) were transfected separately into modified 293 packaging cells, and infectious amphotropic retrovirus was produced as previously described (18 ). Retroviral titers in supernatants were analyzed using NIH3T3 cells (18 ) and utilized at estimated titers of ~5*106/ml; [beta]-galactosidase marker vector titers were used as reference standard with STS vector compared by RT-PCR in retroviral supernatants passed through 0.45 [mu]M cellulose acetate filters to eliminate potentially contaminating packaging cells. Patient keratinocytes (1*105) were plated in each 35 mm well of a 6-well plate and incubated overnight. Immediately prior to transduction, cells were washed with sterile phosphate-buffered saline (PBS), pre-incubated with polybrene at 5 [mu]g/ml in SFM/154 medium for 5 min, then 3 ml of SFM/154 medium containing adjusted concentrations of viral supernatant was added (26 ). In the case of transduction prior to grafting, retroviral supernatant was diluted to 50% with SFM/154 medium and supplemented with 5 [mu]g/ml polybrene. In the case of in vitro studies of STS protein expression and enzymatic activity, a dilution of 15 [mu]l of retroviral supernatant in 2 ml of SFM/154 medium was performed to titrate transduction to produce the levels of STS expression seen in normal cells in vitro. Plates were then centrifuged at 300 g for 1 h at 32oC and then placed in a 37oC incubator for up to 16 h, after which time the medium was changed to 3 ml of fresh SFM/154. The relationship between the number of proviral integrations and STS expression was examined in cells transduced at increasing dilutions of retroviral supernatant. STS protein expression was analyzed by Western analysis of keratinocyte populations transduced at each dilution. In parallel, genomic DNA was extracted, and integrated proviral sequences detected by low cycle PCR that was internally controlled for DNA quality and reaction efficiency with primers for GAPDH genomic sequences. STS vector specific primers were 5'-CTGTGGGAAGCCGAGAGCCACGAAGCATCAAGG-3' and 5'-CCG CTGTATGAACTGGGCCGCCTCCACC-3' and GAPDH control genomic primers were 5'-GGGGAGCGAGATCCCTCCAAAATCAAGTGGGG-3' and 5'-GGGTCATGAGTCCTTCCACGATACCAAAGTTG-3'.
Transduced or untreated XLI patient cells, along with control keratinocytes from patients with normal skin, were utilized to regenerate human epidermis on nude mice as previously described (13 ). In the case of engineered keratinocytes, cells were given fresh media and allowed to recover for 48 h following transduction then were seeded on acellular, devitalized human dermis (13 ,37 ). After 5 days in tissue culture, grafting to nude mouse fascia was performed as described (13 ,24 ). Three mice were grafted per group; in addition, grafts not seeded with keratinocytes were also placed as an additional control to confirm the devitalized nature of the dermal substrate utilized and to exclude carry-over of original donor epidermal components. Graft analysis was performed at 5 weeks post-grafting.
Biopsy specimens were obtained either directly from patients or from regenerated skin xenografts. Frozen tissue skin sections were fixed and incubated with polyclonal antisera to human STS (raised to an STS-TrpE fusion protein) or monoclonal antibody to [beta]-galactosidase (Sigma). Species-specific antiserum to human involucrin (BTI) (38 ) was also used as a further confirmation of species origin of analyzed grafts and as an aid to definition of boundaries between mouse and human epidermis. Sections were then washed with PBS/BSA and incubated with a fluorescein-labeled secondary antibody (Sigma) prior to analysis.
Cell extracts were made and steroid sulfatase activity was measured from triplicate independent transductions as previously described (19 ). Cells to be analyzed were trypsinized and subjected to five cycles of freeze/thawing. The cells were then homogenized. Each sample was aliquoted into a tube containing [3H]estrone sulfate substrate, 0.25 M Tris-HCl and water. The samples were incubated for 4 h at 37oC then extracted with benzene and counted in 10 ml of scintillation fluid.
TEWL was measured in triplicate at 5 weeks post-grafting using a 3 mm aperture on an Evaporimeter (EP1, Servomed) as previously described (39 ).
We thank G. Nolan and T. Kinsella for the LZRS plasmid, P. Yen for the human STS cDNA and J. Morgan for devitalized dermis. We also thank N. Griffiths, S. Swetter, B. Olsen and D. Wing. This work was supported by the Office of Research and Development, Department of Veterans Affairs (V.A.) and by NIH AR43799 for P.A.K and by NIH GM54907 for L.J.S.
Human Molecular Genetics
Pages
Introduction
Results
Gene transfer to primary keratinocytes from XLI patients
Restoration of steroid sulfatase enzymatic activity in XLI patient cells
Regeneration of corrected XLI patient epidermis in vivo
Discussion
Materials And Methods
Skin biopsy and skin culture
Production of the STS retroviral expression vector and gene transfer to primary keratinocytes
Regeneration of human XLI skin on immunodeficient mice
Immunohistochemistry
Steroid sulfatase enzymatic assay
Measurement of cutaneous barrier function
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Mon May 12 18:10:04 BST 1997. Part of the OUP Journals World Wide Web service. Copyright Oxford University Press, 1996