Gene Therapy (2004) 11, 1290–1295 & 2004 Nature Publishing Group All rights reserved 0969-7128/04 $30.00 BRIEF COMMUNICATION Functional improvement of mutant cells on addition of : an alternative approach to therapy for dominant diseases

M D’Alessandro, SM Morley, PH Ogden, M Liovic, RM Porter and EB Lane Cancer Research UK Structure Research Group, School of Life Sciences, University of Dundee, MSI/WTB Complex, Dundee, Scotland, UK

A major challenge to the concept of gene therapy for as osmotic shock, heat shock or scratch wounding. Using a dominant disorders is the silencing or repairing of the mutant ‘severe’ disease model system, we demonstrate allele. Supplementation therapy is an alternative approach reversion towards wild-type responses to stress after that aims to bypass the defective gene by inducing the transfection with human desmin, an intermediate filament expression of another gene, with similar function but not normally expressed in muscle cells. Such a supple- susceptible to the disrupting effect of the mutant one. mentation therapy approach could be widely applicable to simplex (EBS) is a genetic fragility patients with related individual and would avoid disorder caused by mutations in the for K5 some of the financial obstacles to gene therapy for rare or K14, the intermediate filaments present in the basal cells diseases. of the . Keratin diseases are nearly all dominant Gene Therapy (2004) 11, 1290–1295. doi:10.1038/ in their inheritance. In cultured , mutant keratin; Published online 24 June 2004 renders cells more sensitive to a variety of stress stimuli such

Keywords: epidermolysis bullosa simplex; keratin ; K14 ; cell migration; supplementation therapy

Introduction to mutation ‘hotspots’ at the ends of the central a-helical rod domain of each keratin, with especially frequent Genetic skin disorders are attractive targets for research mutations in the K14 arginine125 codon. This EBS form and development of gene therapy strategies as the is characterized by clustered blisters and by abnormal epidermis is accessible and visible. Keratinocytes can be formation of keratin protein aggregates in basal cells.9 taken ex vivo (from small skin biopsies), cultured, treated The keratins implicated in EBS are members of the and then grafted back to the epidermal site where the large intermediate filament multigene family. All inter- outcome of an intervention can be monitored continuously mediate filaments have similar secondary and tertiary and noninvasively. Similarly, keratinocytes can provide structure and assemble into cytoplasmic filaments with vector cells for gene delivery, locally or systemically.1–3 almost indistinguishable morphology, but they each have Epidermolysis bullosa simplex (EBS) is an autosomal differentiation-specific expression ranges that are de- dominant skin fragility disorder caused by mutations in fined at a very early stage of tissue morphogenesis. This copolymeric K5 and K14 keratins,4–6 the intermediate suggests that intermediate filaments have a critical role filament expressed in the basal cells of stratified in defining differentiation. Studies of a range of genetic epithelia. Many different missense mutations in K5 or tissue fragility disorders now suggest that most, if not all, K14 lead to fragile epidermal basal cells that lyse upon intermediate filament types contribute to the physical mild physical trauma to produce intraepithelial blisters resilience of the tissue in which they are expressed. (recently reviewed by Porter and Lane7). As K5 and K14 Cures for genetic disorders require reprogramming will only assemble as heteropolymers, mutation in either of the genetic profile of the affected cells. In developing one of the genes can compromise the whole filament gene therapy strategies for dominant disorders like EBS, network. There are different degrees of clinical severity it is generally taken that the dominant mutant gene will in EBS and this is generally correlated with the position need to be silenced in order to correct the pathological of the alteration in the protein.8 The Dowling–Meara . Many different therapeutic strategies are form (DM-EBS) shows the most severe phenotype, linked being investigated, including chimaeric RNA/DNA oligonucleotides10 and triplex-forming oligonucleo- tides,11–13 small-fragment homologous replacement,14 Correspondence: Professor EB Lane, Cancer Research UK Cell Structure antisense technology,15 small interfering RNAs (siR- Research Group, School of Life Sciences, University of Dundee, MSI/WTB 16,17 18,19 Complex, Dundee DD1 5EH, Scotland, UK NAs) and ribozyme technology, with varying Received 19 November 2003; accepted 28 March 2004; published success. A final obstacle is that most EBS mutations are online 24 June 2004 individual and differ from patient to patient so that gene Rescue of EBS cells by supplementation with desmin M D’Alessandro et al 1291 repair/ strategies would require unique mole- cular manipulation for each family. Building on previous observations,20 we investigated an alternative approach to gene therapy for EBS that involves reinforcing the faulty system function without trying to repair the gene directly. EBS skin blistering is localized and irregular, suggesting that the keratinocytes are almost strong enough to withstand normal physical stress. We therefore hypothesized that a modest amount of structural reinforcement of the keratin network might be sufficient to allow the cells to function normally.20 The type III intermediate filament protein desmin is a homopolymeric protein normally expressed in muscle cells that, like epidermal keratinocytes, are subject to physical stress. While keratin filaments are type I/type II heteropolymers, desmin assembles as a type III homo- . We previously showed that a transfected hamster desmin protein did not copolymerize with Figure 1 Generating desmin-expressing cell lines. (a) Human desmin keratins and appeared unaffected by mutant keratins in cDNA (accession number NM_001927) was cloned into blasticidin- the same cell.20 Here, we report the results of recently resistant LZRS-lacZ-Ires-Blasto plasmid derived from LZRS-LacZ vec- 23 developed functional assays on ‘severe’ keratin mutant tor (pKDblast; a). Vectors without desmin inserts served as negative controls. Retroviral particles were generated by transfecting vectors into DM-EBS cells after transfection with human desmin. Phoenix amphotropic packaging cells24 using calcium phosphate.23 Culture supernatants containing viral particles were collected after 48 h, filtered Transfected human desmin is filamentous and remains and used to infect KEB-7 cells at 20–30% confluence. At 24 h after , retroviral supernatant was replaced with fresh medium and cells independent of keratin filaments left to grow for 5 days. Transfected cells were then selected with 5 mg/ml The ability of transfected desmin to make filaments can of blasticidin (Invitrogen Life Technologies) for 5 days, resulting in a 95% be seen in SW13, adrenal cortex cells, transfection-positive culture. Stable desmin-transfected KEB-7 cells are which do not normally express any cytoplasmic inter- referred to as KEB-7.Des. culture conditions were as mediate filament proteins21,22 (Figure 1c). KEB-7 cells,25 described previously.25 (b) Desmin and K14 expression levels in KEB-7 which express a highly disruptive keratin mutation (K14 (a), KEB-7.Des (b) and wild-type NEB-1 (c) were detected by immunoblotting with NCL-DES-DERII mouse monoclonal R125P), typical of the severe Dowling–Meara type EBS, against desmin (Novocastra; dilution 1:100, track B) and LL001 mouse were stably transfected with full-length human desmin monoclonal antibody to K14.26 The expression levels of endogenous K14 cDNA, using a retroviral vector (pKDblast) (Figure 1a). and transfected desmin are comparable (b). (c) Human desmin is able to Levels of human desmin expression are high and form filaments in the absence of keratins. SW13 adenocarcinoma cells, comparable to that of endogenous K14 (Figure 1b). In lacking any cytoplasmic intermediate filament system,21,22 were transi- the transfected KEB-7 cells (KEB-7.Des) well-defined ently transfected with the desmin and the expressed protein localized with the antibody to desmin NCL-DES-DERII. The human desmin is able to desmin filaments (Figures 2a), independent of the form well-defined filaments de novo, which tend to concentrate close to endogenous keratin network, can be seen (Figure 2a–c). the nucleus. SW13 cells were grown in DMEM with 10% FCS. Scale The human desmin filaments rarely approach the cell bar ¼ 15 mm. boundaries (Figure 2d–f), remaining mostly concentrated in the perinuclear cytoplasm, and no significant inter- the constitutively stressed state of the mutant cells which action with is observed (Figure 2g–i). has the effect of priming them for a wound response. We examined the effect of desmin supplementation on this Response to osmotic shock surrogate indicator of epithelial wound healing. By 8 h Following hypo-osmotic shock with 150 mM urea, cells after scratch wounding, the KEB-7 cells have covered respond to the transient swelling by the activation of the nearly three times the area covered by wild-type cells stress-activated protein kinase (JNK/SAPK) pathway. (Figure 4; Po0.001). Desmin-supplemented KEB-7.Des Cells expressing mutant keratins react with an earlier cells, however, show a highly significant reduction in activation of this pathway.27 We tested the effect of their scratch wound coverage (Po0.001), which is down desmin supplementation on SAPK activation in KEB- to half the KEB-7 value and approaching that of the 7.Des cells compared to wild-type (NEB-1) and non- wild-type cells. transfected mutant cells (KEB-7). In KEB-7 cells, JNK activation is visible 2.5 min after stress, whereas in KEB- Heat shock 7.Des cells, the stress response to osmotic shock is Brief exposure of cells to elevated temperature (431C for reduced to near control levels and phosphorylated pJNK 5 min) reveals an increased thermoinstability in the or pATF-2 bands are not detectable until 5 min after mutant keratin filaments over wild-type cells, probably the stress (Figure 3). due to a reduced affinity interaction between keratin subunits at some stage of filament assembly or turn- Scratch wound closure over.28 In previous studies, keratin aggregates were The rate of closure of a tissue culture scratch ‘wound’ can transiently observed after heat shock in cells expressing be used to demonstrate comparative rates of cell mutant keratins but not in cells with normal keratins26,28 mobilization and migration. Cells expressing EBS-asso- (Figure 5a). However, when KEB-7.Des cells were tested, ciated keratin mutations show faster scratch wound no aggregates were detected, either in the keratin (Figure closure than wild-type cells,25 possibly a consequence of 5b) or in the desmin networks (data not shown).

Gene Therapy Rescue of EBS cells by supplementation with desmin M D’Alessandro et al 1292

Figure 2 Supplemental desmin in keratinocytes makes filaments but fails to extend to desmosomes. The desmin network, stained with NCL-DES-DERII, is filamentous and mostly concentrated in the perinuclear region (a, d and g). It does not appear to have any effect on the overall architecture of the keratin filaments (b and e) (stained with BL18 rabbit polyclonal antibody against K526; dilution 1:200) and is independent of the endogenous keratin network (c) (double with desmin: green, keratin: red). No interaction is apparent between the human desmin filament network and the plasma membrane (d); (f) phase-contrast image of the cells. No associations of the desmin network with the desmosomes can be observed (i). (h) Desmosomes staining with anti- rabbit polyclonal antibody AHP320; (i) double staining of desmin (green) and desmoplakin (red). (a–c and g–i) Delta Vision images (Applied Precision, Issacruach, USA) were collected using a  100 1.4 NA oil-immersion objective on a Nikon inverted TE2000 Eclipse epifluorescence microscope, and subsequently deconvolved and analysed using the softWorX software (Applied Precision). Scale bar ¼ 15 mm. (d–f) Zeiss Axioplan epifluorescence microscope images. Scale bar ¼ 10 mm.

Thus in three separate functional assays that measure still appeared to have no detrimental effects, no different aspects of epithelial cell behaviour, the addition phenotypic rescue was achieved either, possibly because of human desmin to a keratinocyte cell line expressing a expression levels of desmin were much lower than in severe EBS-associated keratin mutation has led to a near- the present study. However, as no K5 null has ever been complete rescue of phenotype. This is the first functional found in humans, this extremely severe model is study to test the efficacy of intermediate filament probably inappropriate for developing EBS therapy. supplementation therapy. It is a significant advance on our earlier experiments, in which we used a mildly mutated EBS cell line (KEB-3 cells, K14-V270M) supple- mented with hamster desmin and showed that the Muscle desmin as a supplement for EBS desmin formed independent filaments from the kera- keratinocytes tins.20 The present study also incorporates two signifi- cant refinements – the use of human rather than hamster Desmin is the gene of choice for supplementation of desmin and the use of a cell line derived from Dowling– epidermal keratin defects, because it is known to be an Meara EBS patient (the disease form that will undoubt- essential reinforcing component of muscle30 and so may edly produce the greatest urgency for treatment) – that be able to reinforce physically the fragile epidermal bring this concept closer to application for human keratinocytes of EBS. Secondly, because desmin can form disease. In another recent study, hamster desmin was homopolymers and is not expected to interact with introduced into basal keratinocytes in a fatally severe keratins,31–33 desmin filaments should not be vulnerable mouse model of EBS, a K5 null mouse.29 While desmin to the dominant-negative effects of mutant keratins.

Gene Therapy Rescue of EBS cells by supplementation with desmin M D’Alessandro et al 1293

Figure 4 Desmin restores cell behaviour in scratch wound assays. KEB-7, KEB-7.Des and NEB1 cell lines were plated in 24-well plates and grown to confluence over 5 days. Simple scratch wounds were made in all wells using a yellow Gilson pipette tip. In all, 12 wells were fixed with methanol:acetone (1:1) at the time of scratching and the rest after 8 h of continued growth. The wound area was microscopically measured using AxioVision 3.0 software (Carl Zeiss Vision GmbH), and the reduction in scratch wound area over the 8 h was calculated.25 The extent of scratch wound closure was measured as the reduction of the scratch area after 8 h. Images above show phase-contrast micrographs of the extent of scratch values for each cell line (n ¼ 48). KEB-7 cells migrate most quickly into the scratch area; KEB-7.Des wound closure is significantly slower and approaches the speed of the wild-type NEB-1 cells. Scale bar ¼ 50 mm.

Figure 3 Supplementation with desmin reduces levels of activation of the stress-activated kinase signalling pathway. Having previously shown that levels of JNK/SAPK and ATF-2 phosphorylation are elevated in EBS cells Figure 5 Desmin normalize cell response to heath shock. KEB-7 and KEB- after hypo-osmotic shock,27 we carried out the same experiment on cells 7.Des cells were grown to 80% confluence on 13 mm microscope coverslips supplemented with human desmin. KEB-7, KEB-7.Des and NEB-1 and subjected to heat shock at 431C for 15 min as described previously.25,28 (control) cells were exposed to hypo-osmotic shock for 5 min at 371Cin The cells were then returned to 371C for 15 min and fixed and stained for 150 mM urea as described previously27 and then returned to normal K5 using BL18 antibody, diluted 1:200. The keratin network of KEB-7 has culture medium for various recovery periods. Levels of JNK, pJNK and numerous tiny aggregates (a and inset) that are absent in the rescued pATF-2 were determined by immunoblotting with specific to KEB-7.Des cells (b and inset). Scale bar ¼ 15 mm. JNK/SAPK, pJNK/SAPK and pATF-2 (New England Biolabs). Total JNK1 (46 kDa), JNK2 (54 kDa), phosphorylated JNK (pJNK1, pJNK2) and pATF-2 (a target of JNK) were analysed by immunoblotting before and desmosomes in these transfected keratinocytes, which after hypo-osmotic shock. KEB-7.Des cells are slower to activate JNK/ SAPK and ATF-2 than KEB-7: phosphorylated kinase bands appear in might be due to a lower affinity of desmin for KEB-7 cells at 2.5 min after osmotic shock, but not until 5 min in KEB- desmosomes in epithelia than in muscle, suggests that 7.Des and 10 min in control NEB-1 cells, indicating that desmin binding is not an essential requirement for a supplementation has shifted the EBS cells’ response towards the wild- second filament system to reinforce the defective type behaviour. keratinocytes. The beneficial effect of a noninteracting filament network must be mediated by some other properties of the system. The mechanism by which desmin partially rescues mutant keratin cells remains to These presuppositions would appear to be correct, as be determined. Mechanosensory signalling is still poorly the expression of desmin in EBS cells does indeed have understood, but it is known that growth and differentia- an effect of countering the accelerated or upregulated tion of skeletal muscle take place in response to stress responses that are characteristic of EBS-derived mechanical stress generated during normal movement cell lines. The lack of interaction of desmin with and exercise. The intermediate filament network, espe-

Gene Therapy Rescue of EBS cells by supplementation with desmin M D’Alessandro et al 1294 cially desmin, could be a potential candidate in this 6 Lane EB et al. A mutation in the conserved helix termination mechanosensory transcription process.34 In muscle, both peptide of in hereditary skin blistering. Nature 1992; mechanical stretch and contraction cause robust increase 356: 244–246. in desmin mRNA content within 6–10 h, and this 7 Porter RM, Lane EB. , genotypes and their induction seems to be controlled at the transcriptional contribution to understanding keratin function. Trends Genet level.35 Epithelial cells such as keratinocytes are also 2003; 19: 278–285. known to respond to stretch signals with growth and 8 Letai A et al. Disease severity correlates with position of keratin proliferation.36–38 point mutations in patients with epidermolysis bullosa simplex. Proc Natl Acad Sci USA 1993; 90: 3197–3201. 9 Fine JD, McGrath J, Eady RA. Inherited epidermolysis bullosa comes into the new millenium: a revised classification system Potential impact of supplementation therapy based on current knowledge of pathogenetic mechanisms and The idea of supplementing mutant keratin cells with the clinical, laboratory, and epidemiologic findings of large, well- desmin is a potentially powerful approach, as it may defined patient cohorts. J Am Acad Dermatol 2000; 43: 135–137. bypass the need for a specific gene therapy design for 10 Alexeev V et al. Localized in vivo genotypic and phenotypic each individual keratin mutation. One desmin transfec- correction of the albino mutation in skin by RNA–DNA oligonucleotide. Nat Biotechnol 2000; 18: 43–47. tion should be effective for all the EBS mutations. Further 11 Gorman L, Glazer PM. Directed gene modification via triple experiments are now required to understand the helix formation. Curr Mol Med 2001; 1: 391–399. mechanisms by which this improvement is effected, as 12 Vasquez KM, Glazer PM. Triplex-forming oligonucleotides: well as to investigate the simplest way of inducing principles and applications. Q Rev Biophys 2002; 35: 89–107. desmin expression in skin cells, taking into account that 13GuntakaRV,VarmaBR,WeberKT.Triplex-formingoligo- 39–41 much is now known about desmin gene regulation. nucleotidesasmodulatorsofgeneexpression.IntJBiochem The concept of supplementation therapy may have a Cell Biol 2003; 35: 22–31. much wider applicability to a range of genetic disorders 14 Richardson PD, Augustin LB, Kren BT, Steer CJ. Gene repair and involving polymerizing structural proteins, irrespective transposon-mediated gene therapy. Stem Cells 2002; 20: 105–118. of the specific disease mutation carried. If the induction 15 Kurreck J. Antisense technologies. Improvement through novel or introduction of one extra gene could be used to chemical modifications. Eur J Biochem 2003; 270: 1628–1644. improve many related disorders with similar but discrete 16 McManus MT, Sharp PA. Gene silencing in by small mutations, supplementation therapy may be able to interfering RNAs. Nat Rev Genet 2002; 3: 737–747. overcome some of the significant financial obstacles 17 Scherr M, Morgan MA, Eder M. Gene silencing mediated by facing the development of gene therapy strategies. This small interfering RNAs in mammalian cells. Curr Med Chem is particularly important for the growing number of 2003; 10: 245–256. known but rare diseases, in which any commercial 18 Samarsky D, Ferbeyre G, Bertrand E. Expressing active market for gene therapy will always remain small. ribozymes in cells. Curr Issues Mol Biol 2000; 2: 87–93. 19 Kashani-Sabet M. Ribozyme therapeutics. J Invest Dermatol Symp Proc 2002; 7: 76–78. 20 Magin TM et al. Supplementation of a mutant keratin by stable Acknowledgements expression of desmin in cultured human EBS keratinocytes. This work was supported by the Dystrophic Epidermo- J Cell Sci 2000; 113: 4231–4239. lysis Bullosa Research Association (DEBRA) (LANE3 to 21 Johnson RG, Sheridan JD. Junctions between cancer cells in EBL, supporting MD), Cancer Research UK (C26/A1461 culture: ultrastructure and permeability. Science 1971; 174: to EBL, supporting SMM, RMP and PHO) and the 717–719. Wellcome Trust (055090 to EBL, supporting ML). We are 22 Hedberg KK, Chen LB. Absence of intermediate filaments in a human adrenal cortex -derived cell line. Exp Cell Res grateful to G Nolan and co-workers, and to P Marinko- 1986; 163: 509–517. vich, for introducing us to the retroviral vector and 23 Kinsella TM, Nolan GP. Episomal vectors rapidly and stably packaging cells and providing us with stocks, and to Dan produce high-titer recombinant retrovirus. Hum Gene Ther 1996; Gibbs for the initial discussions that led to these 7: 1405–1413. experiments. 24 Pear WS, Nolan GP, Scott ML, Baltimore D. Production of high- titer helper-free retroviruses by transient transfection. Proc Natl Acad Sci USA 1993; 90: 8392–8396. References 25 Morley SM et al. Generation and characterization of epidermolysis bullosa simplex cell lines: scratch assays show 1 Cao T, Wang XJ, Roop DR. Regulated cutaneous gene delivery: faster migration with disruptive keratin mutations. Br J Dermatol the skin as a bioreactor. Hum Gene Ther 2000; 11: 2297–2300. 2003; 149: 46–58. 2 Spirito F, Meneguzzi G, Danos O, Mezzina M. Cutaneous gene 26 Purkis PE et al. Antibody markers of basal cells in complex transfer and therapy: the present and the future. J Gene Med 2001; epithelia. J Cell Sci 1990; 97: 39–50. 3: 21–31. 27 D’Alessandro M et al. Keratin mutations of epidermolysis 3 Cao T et al. The epidermis as a bioreactor: topically regulated bullosa simplex alter the kinetics of stress response to osmotic cutaneous delivery into the circulation. Hum Gene Ther 2002; 13: shock. J Cell Sci 2002; 115: 4341–4351. 1075–1080. 28 Morley SM et al. Temperature sensitivity of the Coulombe PA et al. Point mutations in human genes of cytoskeleton and delayed spreading of keratinocyte lines epidermolysis bullosa simplex patients: genetic and functional derived from EBS patients. J Cell Sci 1995; 108 (Part 11): analyses. Cell 1991; 66: 1301–1311. 3463–3471. 5 Bonifas JM, Rothman AL, Epstein Jr EH. Epidermolysis bullosa 29 Kirfel J et al. Ectopic expression of desmin in the epidermis of simplex: evidence in two families for keratin gene abnormalities. transgenic mice permits development of a normal epidermis. Science 1991; 254: 1202–1205. Differentiation 2002; 70: 56–68.

Gene Therapy Rescue of EBS cells by supplementation with desmin M D’Alessandro et al 1295 30 Lazarides E, Hubbard BD. Immunological characterization of 36 Brunette DM. Mechanical stretching increases the number of the subunit of the 100 A filaments from muscle cells. Proc Natl epithelial cells synthesizing DNA in culture. J Cell Sci 1984; 69: Acad Sci USA 1976; 73: 4344–4348. 35–45. 31 Pieper FR et al. Transgenic expression of the muscle-specific 37 Takei T et al. Effect of strain on human keratinocytes in vitro. intermediate filament protein desmin in nonmuscle cells. J Cell J Cell Physiol 1997; 173: 64–72. Biol 1989; 108: 1009–1024. 38 Kippenberger S et al. Signaling of mechanical stretch in human 32 Ngai J, Coleman TR, Lazarides E. Localization of newly keratinocytes via MAP kinases. J Invest Dermatol 2000; 114: synthesized subunits reveals a novel mechanism of 408–412. intermediate filament assembly. Cell 1990; 60: 415–427. 39 Li H, Capetanaki Y. Regulation of the mouse desmin gene: 33 Berteretche MV et al. Abnormal incisor- differentiation in transactivated by MyoD, myogenin, MRF4 and Myf5. Nucleic transgenic mice expressing the muscle-specific desmin gene. Eur Acids Res 1993; 21: 335–343. J Cell Biol 1993; 62: 183–193. 40 Li H, Capetanaki Y. An E box in the desmin promoter cooperates 34 Weitzer G et al. Cytoskeletal control of myogenesis: a desmin with the E box and MEF-2 sites of a distal enhancer to null mutation blocks the myogenic pathway during embryonic direct muscle-specific transcription. EMBO J 1994; 13: differentiation. Dev Biol 1995; 172: 422–439. 3580–3589. 35 Watson PA, Hannan R, Carl LL, Giger KE. Desmin gene 41 Gao J, Li Z, Paulin D. A novel site, Mt, in the human desmin expression in cardiac myocytes is responsive to contractile enhancer is necessary for maximal expression in skeletal muscle. activity and stretch. Am J Physiol 1996; 270: C1228–1235. J Biol Chem 1998; 273: 6402–6409.

Gene Therapy