View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector REPORT

Mutations in CSTA, Encoding A, Underlie Exfoliative Ichthyosis and Reveal a Role for This Protease Inhibitor in Cell-Cell Adhesion

Diana C. Blaydon,1 Daniela Nitoiu,1 Katja-Martina Eckl,2 Rita M. Cabral,1 Philip Bland,1 Ingrid Hausser,3 David A. van Heel,1 Shefali Rajpopat,1 Judith Fischer,4,5 Vinzenz Oji,6 Alex Zvulunov,7,8 Heiko Traupe,6 Hans Christian Hennies,2,9,10 and David P. Kelsell1,*

Autosomal-recessive exfoliative ichthyosis presents shortly after birth as dry, scaly skin over most of the body with coarse peeling of non- erythematous skin on the palms and soles, which is exacerbated by excessive moisture and minor trauma. Using whole-genome homo- zygosity mapping, candidate- analysis and deep sequencing, we have identified loss-of-function mutations in the gene for protease inhibitor cystatin A (CSTA) as the underlying genetic cause of exfoliative ichthyosis. We found two homozygous mutations, a splice-site and a nonsense mutation, in two consanguineous families of Bedouin and Turkish origin. Electron microscopy of skin biopsies from affected individuals revealed that the level of detachment occurs in the basal and lower suprabasal layers. In addition, in vitro modeling suggests that in the absence of cystatin A , there is a cell-cell adhesion defect in human keratinocytes that is particularly prom- inent when cells are subject to mechanical stress. We show here evidence of a key role for a protease inhibitor in epidermal adhesion within the lower layers of the human epidermis.

Cystatins form a superfamily of protease inhibitors, We previously reported a large, consanguineous, Bedouin which could be further subdivided into three groups on family with autosomal-recessive exfoliative ichthyosis the basis of their different distributions and molecular (MIM 607936) presenting in five individuals as circum- structure.1,2 Cystatin A (also known as stefin A) is a mem- scribed peeling of the skin on palms and soles associated ber of the type 1 (comprising stefin A and B) with dry and scaly skin (Figures 1A and 1B). Initial linkage that have a mainly intracellular localization, although analysis with microsatellites showed suggestive linkage to cystatin A has also been detected in sweat and in the chromosomal region 12q13.14 In view of the fact that the medium of cultured keratinocytes.3 The cystatins are unaffected mother was uninformative for all the markers tight, reversible inhibitors of the -like lysosomal in this region of the genome, we decided to reassess the cysteine proteases, such as cathepsins B, H, and L.4–6 linkage analysis in this family. The study was approved by Furthermore, an imbalance of these proteases and their the South East National Health Service research ethics cystatin inhibitors is found in many cancers and other committee and by the institutional review board of the diseases.7–9 University Hospital of Mu¨nster, and all patients enrolled Cystatin A was originally identified as a component of gave their informed consent. High-resolution homozy- the cornified cell envelope in the upper layers of the gosity mapping was performed by genotyping two affected skin10 and has also been shown to be a potent inhibitor individuals (Figure 1C, individuals IV-10 and V-3) with the of exogenous proteases, such as the dust mite allergens Affymetrix GeneChip Human Mapping 250K Nsp SNP Der p 1 and Der f 1,3 which can break down the epidermal array (Affymetrix, Santa Clara, CA, USA) (see Table S1, avail- barrier of the skin. In addition, cystatin A has been associ- able online). GeneChip DNA Analysis Software (Affyme- ated with atopic dermatitis (AD [MIM 603165]),11,12 trix, Santa Clara, CA, USA) was used for calling SNP geno- a chronic inflammatory skin disease often associated with types and a large block of homozygosity (9333.1 kb), a defective epidermal barrier.13 We present here a role for containing approximately 60 and shared by both cystatin A in the normal physiology of the skin within affected individuals, was identified on chromosomal region the lower levels of the epidermis and show that loss-of- 3q21 between rs6783609 and rs6438966. Using PLINK function mutations in cystatin A underlie a peeling skin to analyze the SNP data,15 we observed no significant phenotype in the absence of an obvious epidermal barrier blocks of shared homozygosity on chromosomal region defect. 12q13, and the next largest stretches of homozygosity on

1Blizard Institute, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London E1 2AT, UK; 2Division of Dermatogenetics, Cologne Center for Genomics, University of Cologne, 50931 Cologne, Germany; 3Department of Dermatology, University Hospital of Heidelberg, 69115 Heidelberg, Germany; 4Institute of Human Genetics, University Hospital of Freiburg, 79106 Freiburg, Germany; 5Zentrum fu¨r Biosystemanalyse, Center for Biological Systems Analysis, University of Freiburg, 79104 Freiburg, Germany; 6Department of Dermatology, University Hospital of Mu¨nster, 48149 Mu¨nster, Germany; 7Pediatric Dermatology Unit, Schneider Children’s Medical Center, Petah-Tiqva 49100, Israel; 8Faculty of Health Sciences, Medical School for International Health, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel; 9Cologne Cluster of Excellence on Cellular Stress Responses in Aging-Associated Diseases, University of Cologne, 50931 Cologne, Germany; 10Center for Molecular Medicine Cologne, University of Cologne, 50931 Cologne, Germany *Correspondence: [email protected] DOI 10.1016/j.ajhg.2011.09.001. Ó2011 by The American Society of Human Genetics. All rights reserved.

564 The American Journal of Human Genetics 89, 564–571, October 7, 2011 Figure 1. Identification of CSTA Mutations (A) The left foot of an affected individual showing marked hyperkeratosis with an aspect of water-sensitive palmoplantar keratoderma and superficial exfoliation of skin. (B) The lower right arm of an affected individual showing slight hyperkeratosis and exfoliation. (C) Linkage to chromosomal region 3q21 in a large consanguineous Bedouin family with autosomal-recessive exfoliative ichthyosis confirmed by genotyping microsatellite markers D3S3709, D3S3720, and D3S3645. (D) Candidate-gene screening in the Bedouin family identified a 30 splice-site variant, c.67-2A>T, in CSTA, a cysteine protease inhibitor found in the skin (left panel). Affected individuals are homozygous for c.67-2A>T, whereas the unaffected parents are both carriers of c.67-2A>T; this indicates that the variant segregates with the disease. Analysis of CSTA in a Turkish family with a similar phenotype identified the homozygous nonsense mutation c.256C>T (right panel). chromosomal regions 4p15 and 5q23 were 1976.4 kb and 1, c.67-2A>T (p.Val23_Gln26del), whereas the parents were 1892.8 kb, respectively (see Table S2). The new region of heterozygous carriers for c.67-2A>T, and unaffected indi- homozygosity on 3q21 lies between an uninformative viduals were either heterozygous or wild-type (WT) microsatellite marker, D3S2460, and a heterozygous micro- (Figure 1D). c.67-2A>T was not detected in 300 chromo- satellite marker, D3S1764, used in our previous study. somes from normal controls and is not present in the dbSNP Linkage to 3q21 was confirmed by genotyping microsatel- or 1000 Genomes databases. lite markers (D3S3709 [LOD 2.67], D3S3720 [LOD 3.27] Because we no longer have access to material from and D3S3645 [LOD 3.13]) across the region (Figure 1C). affected members of the Bedouin family to confirm the CSTA (MIM 184600), located within the block of homozy- effect of the putative splice-site mutation on the in vivo gosity on 3q21, was chosen as a likely candidate gene processing of CSTA, we included all the exons found in because it encodes a protein that is known to be expressed the block of homozygosity on chromosomal region 3q21 in keratinocytes.10 We sequenced all three coding exons in a next-generation sequencing project to verify that there and the flanking exon/intron boundaries of CSTA were no other potentially disease-causing mutations (NM_005213.3) by using standard Sanger sequencing within this region in these individuals. All the exons methods in all available members of the Bedouin family. were captured and sequenced in one affected individual All affected individuals were homozygous for an A>T (V-3 in Figure 1C) from the Bedouin family.16 Briefly, all change affecting the 30 splice-consensus sequence of intron the exons located within the region of interest for each

The American Journal of Human Genetics 89, 564–571, October 7, 2011 565 disease (i.e., 121,613,170 – 124,440,035 bp on chromo- cells. Briefly, two CSTA minigene constructs were prepared somal region 3q21) were extracted from the UCSC by cloning each of the three CSTA exons, along with Genome Browser database and probes to all the exons approximately 100 bp of surrounding intron sequence, designed and included on a custom capture microarray into the pcDNA3 vector. The WT minigene construct con- NimbleGen (Roche Nimblegen, Madison, WI, USA). DNA tained the normal CSTA sequence as found in the human from the individuals of interest was randomly fragmented genome database, whereas the mutant minigene construct with a Bioruptor sonicator (Diagenode, Denville, NJ, USA) contained the c.67-2A>T splice-site change. Both mini- and the library of adaptor-ligated DNA fragments was gene constructs were transfected into human HEK293T prepared by following the Illumina protocol. The DNA cells, which do not express CSTA with FuGENE 6 transfec- library was then hybridized to the custom designed micro- tion reagent (Roche Diagnostics, Burgess Hill, West Sussex, array from NimbleGen for 72 hr, after which time any un- UK). Forty-eight hours after transfection, RNA was col- bound DNA was washed off and the captured DNA was lected from the transfected cells with the QIAGEN RNeasy eluted with sodium hydroxide and amplified in a PCR minikit (QIAGEN, Crawley, West Sussex, UK). cDNA was with primers against the common adaptor sequences. made with a mixture of Oligo dT and random hexamer The captured, amplified DNA fragments were then primers and SuperScript II Reverse Transcriptase (Invitro- sequenced as paired-end reads on the Illumina GAIIx (Illu- gen, Paisley, UK). The cDNA was then amplified with mina, San Diego, CA, USA). Raw 76 bp paired-end reads a forward primer in exon 1 of CSTA and a reverse primer were aligned to the human reference sequence (hg19) in exon 3 of CSTA, and the PCR products were sequenced. with novoalign, including the soft clipping, adaptor trim- Analysis of splicing of the CSTA minigene construct re- ming, and base-call quality calibration options. Filtering vealed that the 30 splice-site mutation c.67-2A>T leads to for clonal reads, pileup generation, and SNP calling on skipping of the first 12 base pairs of exon 2 of CSTA, which the basis of allele counts and read-depth were performed translates to an in-frame deletion of four amino acid resi- with custom Perl/Cþþ scripts. We filtered the variants dues in the cystatin A protein (p.Val23_Gln26del) (Fig- against dbSNP and 1000 Genomes to identify previously ure 2B). Immunoblotting of lysates collected from cells unreported variants. The 30 splice-site change in CSTA, transfected with the minigene constructs showed greatly c.67-2A>T, was identified by next-generation sequencing. reduced levels of protein expression from the mutant The only other coding change identified in the region was construct (Figure 2C), which we predict to be due to the a predicted missense change (c.1058A>G, p.Asn224Ser) utilization of the much weaker splice-acceptor site within in SEMA5B (NM_001031702.2), a gene that encodes exon 2, as predicted by the Neural Network Splice Site a protein involved in axonal guidance during neural Prediction Tool (Figure 2A). Furthermore, in silico mod- development and, hence, does not represent an obvious eling of the WT and mutated cystatin A revealed candidate gene for exfoliative ichthyosis. In parallel, we clear structural differences in the mutant cystatin A pro- identified, by standard Sanger sequencing in a Turkish tein, including: (1) the division of the first b sheet into family with a very similar phenotype of exfoliative ichthyo- two shorter b sheets, (2) the loss of a complete turn and sis, a homozygous nonsense mutation in CSTA, c.256C>T change in orientation of the a-helix, and (3) the introduc- resulting in the premature termination codon p.Gln86stop tion of a second twist in the fourth b sheet (Figures 2D and (Figure 1D, right panel). We did not have access to materials 2E). These changes in the tertiary structure of the mutated suitable for testing a potential synthesis of a truncated cystatin A protein might alter the relative location of vital protein; however, the altered glutamine residue is highly residues involved in the binding of cystatin A to potential conserved (ConSeq17 score 7) and the termination codon cathepsin substrates6 and, hence, could detrimentally is located within the conserved cystatin domain of cystatin affect the ability of cystatin A to bind. Therefore, the A, thus clearly indicating that p.Gln86stop is a deleterious splice-site mutation c.67-2A>T in all likelihood represents mutation. This finding provides strong support for muta- a nonfunctional allele and, hence, together with the tions in CSTA as the underlying cause of exfoliative finding of nonsense mutations in the Turkish family ichthyosis. reveals that loss-of-function mutations in CSTA underlie To assess the potential effect of c.67-2A>T on the congenital autosomal-recessive exfoliative ichthyosis. splicing of CSTA, we compared WT and mutated DNA The Turkish and Bedouin families both have a very sequences by using in silico splice-site predictor programs. similar phenotype that presents shortly after birth as The splice-site predictor software Neural Network Splice a dry, scaly skin over most of the body and coarse peeling Site Prediction Tool18 predicts that the CSTA c.67-2A>T of nonerythematous skin on the palms and soles, which is mutation would abolish the 30 splice-acceptor site (Fig- exacerbated by excessive moisture (aquagenic) and minor ure 2A). Likewise, the online program for scoring 30 splice trauma. Electron microscopy of skin biopsies from both sites, MaxEntScan::score3ss,19 predicts a much lower maxi- families revealed that the upper layers of the epidermis mum entropy score for the mutant splice site (4.89) when appear mostly normal, whereas there is prominent inter- compared to the WT splice site (13.26). To confirm the cellular edema of the basal and suprabasal cell layers and impairment of the c.67-2A>T splice site in vitro, we aggregates of tonofilaments in the basal keratinocytes analyzed expression of minigene constructs in HEK293T (Figure 3A) demonstrating that the skin peeling initiates

566 The American Journal of Human Genetics 89, 564–571, October 7, 2011 Figure 2. Evaluation of the c.67-2A>T Mutant Cystatin A (A) With the Neural Network Splice Site Prediction software18 the WT CSTA splice-site scores a maximum score of 1, whereas the mutated splice site, c.67-2A>T, is predicted to score 0. A predicted alternative splice site that is 12 bp into exon 2 has a very weak score of 0.17. (B) An in vitro splice assay showed that in the presence of the c.67-2A>T mutation, the splicing machinery uses an alternative splice site that leads to skipping of the first 12 bp of exon 2 of CSTA (boxed nucleotides on sequence trace), predicted to result in an in-frame dele- tion of the four amino acids: Val-Lys-Pro-Gln (residues 23-26) in the cystatin A protein. (C) Immunoblotting of lysates collected from HEK293T cells transfected with empty pcDNA3 vector, the mutant CSTA minigene construct with the c.67-2A>T splice-site change (Mut), and the WT CSTA minigene construct showed greatly reduced levels of expression from the mutant construct, probably because of utilization of the much weaker splice-acceptor site within exon 2. (D) In silico modeling shows the predicted ribbon structure of WT cystatin A. (E) In silico modeling of the splice-site mutant, c.67-2A>T, cystatin A with the four amino acid deletion; the model shows the possible structural differences between the mutant and WT proteins: (1) the first b sheet is split by a random coil giving two shorter b sheets, (2) the a-helix has lost one complete turn and changed its orientation, and (3) the fourth b sheet has a second twist. Structures determined from amino acid sequence with i-Tasser online server29 and figures produced with Swiss-PdbViewer 4.0.30 by weakness in keratinocyte attachment at the basal and cystatin A expression in the HaCaT human keratinocyte lower suprabasal level in the affected individuals. A cell line and studied the effect of mechanical stress on detailed view of the cell-cell contacts showed a loss in these cells. Cells were seeded at 2 3 105 cells/35 mm well number and tightness of the desmosomes in the basal layer and transfected with 100 nM (final concentration) of as compared to the upper epidermal layers and a partial either CSTA siRNA (CSTA KD) or nontargeting pool siRNA loosening of their intercellular interaction (Figure 3A). Im- (NTP) (Dharmacon). siRNA was prepared by mixing 10 ml munostaining of normal skin sections revealed that cysta- of 20 mM stock concentration with 190 ml of serum-free tin A is localized throughout all suprabasal layers of the and antibiotic-free Dulbecco’s Modified Eagle’s Medium/ epidermis with a diffuse cytoplasmic distribution and the Nutrient F-12 Ham (DMEM-F12) (Sigma-Aldrich) and strongest synthesis in the granular layer (Figure 3B). To transfected with 6 ml DharmaFECT 1 (Dharmacon) diluted mimic the phenotype observed in these families in an in 194 ml of serum-free and antibiotic-free DMEM-F12 in vitro model, we have first used ON-TARGETplus SMART- media. Seventy-two hours after transfection, the efficiency pool siRNA (Dharmacon, Chicago, IL, USA) to knock down of the knockdown was analyzed by immunoblotting of cell

The American Journal of Human Genetics 89, 564–571, October 7, 2011 567 Figure 3. Loss of Cystatin A Protein Affects Cell-Cell Adhesion (A) Electron microscopy of a skin biopsy of ridged skin from palm and sole, respectively, from an affected member of the Turkish family. The upper layers of the epidermis appear mostly normal (1), whereas there is prominent intercellular edema of the suprabasal and basal cell layers (2) and aggregates of tonofilaments in the basal keratinocytes (3, arrow), suggesting that the skin peeling initiates at the basal- suprabasal level in the affected individuals. In higher magnification, cell-cell adhesion looks normal with a regular structure of desmo- somes in upper layers (4) but there is loosening of the keratinocyte structure with irregularities in desmosomal interaction in the basal layer (5, arrow). The following abbreviations are used: SG, stratum granulosum; BL, basal layer. (B) Staining of normal skin shows that cystatin A is localized throughout the epidermis and has a diffuse cytoplasmic distribution. (C) The efficiency of cystatin A knockdown (CSTA KD) compared to cells treated with a nontargeting pool (NTP) of siRNA was analyzed by immunocytochemistry and immunoblotting of cell lysates. An approximately 85% knockdown was achieved. (D) In an in vitro model of exfoliative ichthyosis, monolayers of CSTA KD cells were stressed mechanically and then stained for keratin 14 (green). Nuclei are stained with DAPI (blue). Upon intense stretching (4 hr) the CSTA KD monolayer broke into many fragments, whereas the NTP monolayer of cells remained intact (203). At higher magnification (1003), breakage of keratin filaments and widened intercellular spaces could be seen between the CSTA KD cells, in contrast to no obvious cell-cell adhesion defects in the stretched NTP cell monolayer. The scale bar represents 20mm. (E) A dispase-based dissociation assay was utilized to assess intercellular adhesion by the degree of cell monolayer integrity upon mechanical stress. CSTA KD cells showed an increased number of monolayer fragments in contrast to the small number of monolayer fragments observed for the NTP cells. These observations show a statistically significant decrease in cell-cell adhesion in monolayers lacking cystatin A protein. Error bars show the standard error of the mean. lysates and by immunocytochemistry (Figure 3C). CSTA the loading station containing six planar-faced cylinders KD cells and NTP cells were both seeded on BioFlex or posts. Each post (25 mm) is centered beneath the rubber 6-well plates coated with pronectin, which contain a flex- membrane of each 35 mm well. Cells were subject to cyclic ible, rubber membrane in each 35 mm well (Flexplates, mechanical stretch with a frequency of 5 Hz (i.e., five Flexcell International, Hillsborough, NC, USA), grown to cycles of stretch and relaxation per second) and an elonga- ~80% confluency, and then the monolayers were stressed tion of amplitude ranging from 8% to 14% (i.e., increase in mechanically with a Flexcell FX-4000 Tension System diameter across the silicone membrane from 8% to (Flexcell International, Hillsborough, NC, USA)—a 14%). Cells were stretched for 0 hr (unstretched), 1 hr, computer-regulated bioreactor that uses vacuum pressure and 4 hr, and then fixed and stained for keratin 14 to apply cyclic or static strain to cells cultured on flex- (Figure 3D). Upon intense stretching (4 hr) thickening of ible-bottomed culture plates. Each plate was placed over the keratin filaments was observed in both CSTA KD and

568 The American Journal of Human Genetics 89, 564–571, October 7, 2011 NTP cells; however, the CSTA KD monolayer had split into many fragments, whereas the NTP monolayer of cells was still intact (Figure 3D, 203). At higher magnification, breakage of keratin filaments and widened intercellular spaces could be seen in the CSTA KD cells (Figure 3D, 1003); in contrast there are no obvious cell-cell adhesion defects in the stretched NTP cell monolayer. In addition, a dispase-based dissociation assay was utilized to assess intercellular adhesion by the degree of cell monolayer integrity upon mechanical stress. Cells were seeded with a density of 2 3 105 cells/35 mm well and treated, as before, with NTP siRNA or cystatin A siRNA. Postknockdown, the monolayers were detached by incu- bating cells with 5 ml/60 mm dish of dispase for 40 min at 37C. After all monolayers have detached as an intact sheet, they were subjected to agitation by inversion and all fragments obtained were counted under a dissection microscope. CSTA KD cell monolayers showed a high increase in the number of fragments, demonstrating a significant decrease in cell-cell adhesion, in contrast to the very few fragments obtained for the NTP cell mono- layers (Figure 3E). These data further support the observa- tions made for the cell stretch assay showing a cell-cell adhesion defect in cells lacking cystatin A. To further study the effects of loss of cystatin A on differ- entiating keratinocytes, we have generated 3D skin models of exfoliative ichthyosis by knockdown of CSTA in normal primary keratinocytes, which were then used for organo- typic tissue culture (Figure 4). 3D models were generated 20 3 5 Figure 4. Organotypic Cultures as described recently. In brief, 1 10 fibroblasts and CSTA 3 6 2 (A) Full skin models were generated with NTP and KD. Loss 1.0–1.18 10 keratinocytes per cm were used in a of cystatin A was shown by immunoblot analysis in protein 3 mm tissue-culture insert (BD Biosciences, Heidelberg, extracts from skin models and by immunostaining against Germany). A mixture of ice-cold bovine collagen I, fibro- cystatin A. blasts, and serum was poured into each filter insert. (B) Hematoxylin and eosin staining showing completed epidermal differentiation, hyperkeratosis, and parakeratosis, and a distur- Medium was added after incubation at 37 C for 2 hr. Kera- bance of basal epidermal architecture in CSTA KD organotypic tinocytes were seeded on the matrix in a total volume of culture compared to NTP organotypic cultures. The scale bar repre- 1 ml KGM per insert. After 24 hr the system was raised to sents 50 mm. the air-liquid interface and keratinocyte conditioned (C) Immunostaining against keratin 14 (upper panel) and E-cadherin (lower panel) demonstrated grossly disorganized struc- medium was used as differentiation medium for the next ture, widening of intercellular spaces (arrows), and partial break- 8 days. For model preparation, cells from healthy female down of keratin intermediate filaments in basal and lower supra- donors aged 46 and 30 years, respectively, and from the basal layers of disease models depleted of cystatin A. The scale m foreskin of a 9-year-old male were used. For preparing bar represents 20 m. SC is an abbreviation of stratum corneum. (D) Barrier function appears to be largely maintained in the CSTA disease models, knockdown in primary keratinocytes was KD organotypic culture in a Lucifer Yellow dye penetration assay performed exactly 24 hr before model set up with a set of (shown in green) compared to the NTP organotypic culture and three siRNAs (740 pmol each) and 150 ml Lipofectamine in contrast to the loss of barrier function observed in the corneo- CDSN 2000 (Invitrogen, Karlsruhe, Germany) in a T225 cell- desmosin knockdown ( KD) organotypic culture, which is a model of the generalized peeling skin disease. The scale bar repre- culture flask. Full-thickness skin models were cultured sents 100 mm. until day 8 at the air-liquid interface. All models used here were generated from the same sources of cells. Loss of cystatin A in the 3D models was confirmed by marker proteins (data not shown). Immunostaining immunoblot analysis with proteins extracted from model against keratin 14 and E-cadherin showed a widening of biopsy specimens and by immunostaining against cystatin intercellular spaces and major disorganization of keratino- A(Figure 4A). The CSTA KD model demonstrated hyperker- cytes in the lower suprabasal and basal layers (Figure 4C). atosis, parakeratosis, and moderate epidermal hyperplasia, Circular structures, in contrast to the stretched filaments and a disturbance of basal epidermal architecture (Fig- in the normal skin models, are indicative of a partial ure 4B). Epidermal differentiation and basement mem- breakdown of intermediate filaments. To analyze epi- brane interaction were confirmed by staining of epidermal dermal barrier properties in exfoliative ichthyosis, we

The American Journal of Human Genetics 89, 564–571, October 7, 2011 569 made use of a Lucifer Yellow penetration assay. A 1 mM a previously unknown key role for cystatin A in basal to Lucifer Yellow solution (Sigma, Munich, Germany) in suprabasal keratinocyte adhesion. PBS was applied onto control and knockdown skin models. After 1 hr at 37 C, the solution was washed off and models Supplemental Data were snap-frozen. Sections were counterstained with 40,6-diamidino-2-phenylindole (DAPI) and visualized with Supplemental Data include two tables and can be found with this a fluorescence microscope (LMD6000, Leica, Wetzlar, article online at http://www.cell.com/AJHG/. Germany). Control full-thickness skin models (WT) showed an active epidermal barrier. Barrier function, as Acknowledgments assessed by this assay, is largely maintained in the absence We would like to thank the families for their participation in of cystatin A compared with WT organotypic cultures and this study, and we would like to acknowledge the members of in contrast to organotypic cultures grown with corneodes- The Genome Centre, Queen Mary University of London, for their mosin (CDSN) KD cells (Figure 4D). assistance. This work was funded by Barts and The London Charity We have identified inactivating mutations in CSTA in and supported in part by grants from the German Federal Ministry two cases of autosomal-recessive exfoliative ichthyosis. for Education and Research as part of the Networks for Rare Our data on the mechanically stressed monolayer of Diseases (01GM0902) and the Deutsche Forschungsgemeinschaft CSTA KD cells, in combination with the organotypic tissue (HE3119/5-1). culture with CSTA KD keratinocytes and electron micros- copy of biopsies from affected individuals, reveal an unex- Received: July 14, 2011 Revised: August 26, 2011 pected and important role for cystatin A in desmosome- Accepted: September 2, 2011 mediated cell-cell adhesion in the lower levels of the Published online: September 22, 2011 epidermis. A variant of cystatin A (CSTA, c.287T>C), pre- dicted to encode a mRNA transcript with reduced stability, has been associated with AD and is hypothesized to con- Web Resources 12 tribute to the defective epidermal barrier seen in AD. The URLs for data presented herein are as follows: However, the data from our organotypic 3D culture model 1000 Genomes, http://www.1000genomes.org/ demonstrate that there is no gross barrier defect associated dbSNP, http://www.ncbi.nlm.nih.gov/projects/SNP/ with CSTA KD, unlike the generalized peeling skin disease MaxEntScan::score3ss, http://genes.mit.edu/burgelab/maxent/ with atopic manifestations, which is caused by mutations Xmaxentscan_scoreseq_acc.html in corneodesmosin and characterized by a severe epi- Neural Network Splice Site Prediction Tool, http://www.fruitfly.org/ 21 dermal barrier defect. Accordingly, no atopic component seq_tools/splice.html is observed in the phenotype of individuals with exfolia- Novoalign, http://www.novocraft.com tive ichthyosis. This is also in contrast to Netherton’s Online Mendelian Inheritance in Man, http://www.ncbi.nlm.nih. syndrome (MIM 256500), where the loss of a protease gov/omim inhibitor, LEKTI, leads to hyperactivity of kallikrein 5 PLINK, http://pngu.mgh.harvard.edu/purcell/plink/ and increased corneodesmosomal degradation, resulting UCSC Genome Browser database, http://genome.ucsc.edu/ in ichthyosiform erythroderma and atopic manifesta- tions.22,23 Similarly, inactivation of the serine protease matriptase leads to autosomal-recessive ichthyosis accom- References panied by disturbance of the stratum corneum,24,25 but 1. Turk, V., and Bode, W. (1991). The cystatins: Protein inhibitors also a defect in the mouse ortholog of the matriptase inhib- 285 26,27 of cysteine proteinases. FEBS Lett. , 213–219. itor, SPINT1, leads to an ichthyosis-like phenotype. It 2. Wallin, H., Bjarnadottir, M., Vogel, L.K., Wasse´lius, J., Ekstro¨m, should be noted that in a mouse model with a chromo- U., and Abrahamson, M. (2010). Cystatins—Extra- and intra- somal deletion encompassing Csta, no obvious phenotype cellular cysteine protease inhibitors: High-level secretion and 28 was observed, which could reflect differences between uptake of cystatin C in human neuroblastoma cells. Biochimie human and mouse skin. In contrast to phenotypes associ- 92, 1625–1634. ated with generalized skin peeling, exfoliative ichthyosis is 3. Kato, T., Takai, T., Mitsuishi, K., Okumura, K., and Ogawa, H. caused by lesions in the basal layers of the epidermis char- (2005). Cystatin A inhibits IL-8 production by keratinocytes acterized by expression of KRT14. This difference might stimulated with Der p 1 and Der f 1: Biochemical skin barrier 116 contribute to the pattern of the phenotype, which is against mite cysteine proteases. J. Allergy Clin. Immunol. , 169–176. clearly pronounced on the palmar and plantar skin. As in 4. Jenko, S., Dolenc, I., Guncar, G., Dobersek, A., Podobnik, M., other disorders of cornification with a palmoplantar ex- and Turk, D. (2003). Crystal structure of Stefin A in complex pression pattern, the defect in the basal and lower supra- with cathepsin H: N-terminal residues of inhibitors can adapt basal layers might thus be particularly susceptible to to the active sites of endo- and exopeptidases. J. Mol. Biol. 326, mechanical stress, which is strongly present on palmo- 875–885. plantar skin. In summary, our data describe mutations in 5. Pavlova, A., and Bjo¨rk, I. (2002). The role of the second CSTA associated with a monogenic skin disease and reveals binding loop of the cysteine protease inhibitor, cystatin A

570 The American Journal of Human Genetics 89, 564–571, October 7, 2011 (stefin A), in stabilizing complexes with target proteases is 18. Reese, M.G., Eeckman, F.H., Kulp, D., and Haussler, D. (1997). exerted predominantly by Leu73. Eur. J. Biochem. 269, Improved splice site detection in Genie. J. Comput. Biol. 4, 5649–5658. 311–323. 6. Renko, M., Pozgan, U., Majera, D., and Turk, D. (2010). Stefin 19. Yeo, G., and Burge, C.B. (2004). Maximum entropy modeling A displaces the occluding loop of only by as much of short sequence motifs with applications to RNA splicing as required to bind to the active site cleft. FEBS J. 277, 4338– signals. J. Comput. Biol. 11, 377–394. 4345. 20. Eckl, K.M., Alef, T., Torres, S., and Hennies, H.C. (2011). Full- 7. Gelb, B.D., Shi, G.P., Chapman, H.A., and Desnick, R.J. (1996). thickness human skin models for congenital ichthyosis and Pycnodysostosis, a lysosomal disease caused by cathepsin K related keratinization disorders. J. Invest. Dermatol. 131, deficiency. Science 273, 1236–1238. 1938–1942. 8. Pennacchio, L.A., Lehesjoki, A.E., Stone, N.E., Willour, V.L., 21. Oji, V., Eckl, K.M., Aufenvenne, K., Na¨tebus, M., Tarinski, T., Virtaneva, K., Miao, J., D’Amato, E., Ramirez, L., Faham, M., Ackermann, K., Seller, N., Metze, D., Nu¨rnberg, G., Fo¨lster- Koskiniemi, M., et al. (1996). Mutations in the gene encoding Holst, R., et al. (2010). Loss of corneodesmosin leads to severe cystatin B in progressive myoclonus epilepsy (EPM1). Science skin barrier defect, pruritus, and atopy: Unraveling the peeling 271, 1731–1734. skin disease. Am. J. Hum. Genet. 87, 274–281. 9. Toomes, C., James, J., Wood, A.J., Wu, C.L., McCormick, D., 22. Chavanas, S., Bodemer, C., Rochat, A., Hamel-Teillac, D., Ali, Lench, N., Hewitt, C., Moynihan, L., Roberts, E., Woods, M., Irvine, A.D., Bonafe´, J.L., Wilkinson, J., Taı¨eb, A., Barran- C.G., et al. (1999). Loss-of-function mutations in the don, Y., et al. (2000). Mutations in SPINK5, encoding a serine cathepsin C gene result in periodontal disease and palmoplan- protease inhibitor, cause Netherton syndrome. Nat. Genet. 25, tar keratosis. Nat. Genet. 23, 421–424. 141–142. 10. Steven, A.C., and Steinert, P.M. (1994). Protein composition of 23. Descargues, P., Deraison, C., Bonnart, C., Kreft, M., Kishibe, cornified cell envelopes of epidermal keratinocytes. J. Cell Sci. M., Ishida-Yamamoto, A., Elias, P., Barrandon, Y., Zambruno, 107, 693–700. G., Sonnenberg, A., and Hovnanian, A. (2005). Spink5- 11. Samuelsson, L., Stiller, C., Friberg, C., Nilsson, C., Inerot, A., deficient mice mimic Netherton syndrome through degrada- and Wahlstro¨m, J. (2004). Association analysis of cystatin A tion of desmoglein 1 by epidermal protease hyperactivity. and zinc finger protein 148, two genes located at the psoriasis Nat. Genet. 37, 56–65. susceptibility PSORS5. J. Invest. Dermatol. 122, 1399– 24. Alef, T., Torres, S., Hausser, I., Metze, D., Tu¨rsen, U., Lestrin- 1400. gant, G.G., and Hennies, H.C. (2009). Ichthyosis, follicular 12. Vasilopoulos, Y., Cork, M.J., Teare, D., Marinou, I., Ward, S.J., atrophoderma, and hypotrichosis caused by mutations in Duff, G.W., and Tazi-Ahnini, R. (2007). A nonsynonymous ST14 is associated with impaired profilaggrin processing. J. substitution of cystatin A, a cysteine protease inhibitor of Invest. Dermatol. 129, 862–869. house dust mite protease, leads to decreased mRNA stability 25. Basel-Vanagaite, L., Attia, R., Ishida-Yamamoto, A., Rain- and shows a significant association with atopic dermatitis. shtein, L., Ben Amitai, D., Lurie, R., Pasmanik-Chor, M., Indel- Allergy 62, 514–519. man, M., Zvulunov, A., Saban, S., et al. (2007). Autosomal 13. Cork, M.J., Robinson, D.A., Vasilopoulos, Y., Ferguson, A., recessive ichthyosis with hypotrichosis caused by a mutation Moustafa, M., MacGowan, A., Duff, G.W., Ward, S.J., and in ST14, encoding type II transmembrane serine protease ma- Tazi-Ahnini, R. (2006). New perspectives on epidermal barrier triptase. Am. J. Hum. Genet. 80, 467–477. dysfunction in atopic dermatitis: Gene-environment interac- 26. Nagaike, K., Kawaguchi, M., Takeda, N., Fukushima, T., Sawa- tions. J. Allergy Clin. Immunol. 118, 3–21, quiz 22–23. guchi, A., Kohama, K., Setoyama, M., and Kataoka, H. (2008). 14. Hatsell, S.J., Stevens, H., Jackson, A.P., Kelsell, D.P., and Defect of hepatocyte growth factor activator inhibitor type 1/ Zvulunov, A. (2003). An autosomal recessive exfoliative serine protease inhibitor, Kunitz type 1 (Hai-1/Spint1) leads to ichthyosis with linkage to 12q13. Br. J. Dermatol. ichthyosis-like condition and abnormal hair development in 149, 174–180. mice. Am. J. Pathol. 173, 1464–1475. 15. Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, 27. Szabo, R., Kosa, P., List, K., and Bugge, T.H. (2009). Loss of ma- M.A., Bender, D., Maller, J., Sklar, P., de Bakker, P.I., Daly, triptase suppression underlies spint1 mutation-associated ich- M.J., and Sham, P.C. (2007). PLINK: A tool set for whole- thyosis and postnatal lethality. Am. J. Pathol. 174, 2015–2022. genome association and population-based linkage analyses. 28. Bilodeau, M., MacRae, T., Gaboury, L., Laverdure, J.P., Hardy, Am. J. Hum. Genet. 81, 559–575. M.P., Mayotte, N., Paradis, V., Harton, S., Perreault, C., and 16. Blaydon, D.C., Biancheri, P., Di, W.-L., Plagnol, V., Cabral, Sauvageau, G. (2009). Analysis of blood stem cell activity R.M., Brooke, M.A., van Heel, D.A., Ruschendorf, F., Toynbee, and cystatin gene expression in a mouse model presenting M., Walne, A., et al. (2011). Inflammatory Skin and Bowel a chromosomal deletion encompassing Csta and Stfa2l1. disease Linked to Deletion in ADAM17. N. Engl. J. Med., PLoS ONE 4, e7500. in press. 29. Roy, A., Kucukural, A., and Zhang, Y. (2010). I-TASSER: A 17. Berezin, C., Glaser, F., Rosenberg, J., Paz, I., Pupko, T., unified platform for automated protein structure and function Fariselli, P., Casadio, R., and Ben-Tal, N. (2004). ConSeq: prediction. Nat. Protoc. 5, 725–738. the identification of functionally and structurally important 30. Guex, N., and Peitsch, M.C. (1997). SWISS-MODEL and the residues in protein sequences. Bioinformatics 20, 1322– Swiss-PdbViewer: An environment for comparative protein 1324. modeling. Electrophoresis 18, 2714–2723.

The American Journal of Human Genetics 89, 564–571, October 7, 2011 571