Copyright � 2010 by the Genetics Society of AmericaDOI: 10.1534/genetics.109.113167

The waved with open eyelids (woe) Locus Is a Hypomorphic MouseMutation in Adam17

E. L. Hassemer,* S. M. Le Gall,† R. Liegel,* M. McNally,‡ B. Chang,§ C. J. Zeiss,**R. D. Dubielzig,†† K. Horiuchi,‡‡ T. Kimura,‡‡ Y. Okada,§§ C. P. Blobel†

and D. J. Sidjanin*,1

*Department of Cell Biology, Neurobiology and Anatomy, Medical College of Wisconsin, Milwaukee, Wisconsin 53226, ‡Departmentof Microbiology and Molecular Genetics, Medical College of Wisconsin, Milwaukee, Wisconsin 53226, †Arthritis and Tissue

Degeneration Program, Hospital for Special Surgery, New York, New York 10021, §The Jackson Laboratory, Bar Harbor,Maine, **Section of Comparative Medicine, Yale School of Medicine, New Haven, Connecticut 06437,

††School of Veterinary Medicine, Madison, Wisconsin 53706, ‡‡Department of Anti-agingOrthopedic Research and Orthopedic Surgery, Keio University School of Medicine,

Shinjuku-ku, Tokyo 160-8582, Japan and §§Department of Pathology, Keio UniversitySchool of Medicine, Shinjuku-ku, Tokyo 160-8582, Japan

Manuscript received December 15, 2009Accepted for publication February 20, 2010

ABSTRACT

The waved with open eyes (woe) locus is a spontaneous recessive mouse mutation that exhibits wavy fur,eyelids open at birth, and enlarged heart and esophagus. In this study, we confirmed the previouslyidentified woe phenotypes and additionally identified anterior eye segment defects, absence of themeibomian glands, and defects in the semilunar cardiac valves. Positional cloning identified a C794Tsubstitution in the Adam17 gene that ablates a putative exonic splicing enhancer (ESE) sequence in exon7 resulting in aberrant Adam17 splicing. The predominant woe transcript, Adam17Dexon7, lacks exon 7resulting in an in-frame deletion of 90 bp and a putative Adam17D252-281 protein lacking residues 252–281from the metalloprotease domain. Western blot analysis in woe identified only the precursor form ofAdam17D252-281 protein. Absence of cleavage of the prodomain renders Adam17D252-281 functionally inactive;however, constitutive and stimulated shedding of Adam17 substrates was detected in woe at significantlyreduced levels. This residual Adam17 shedding activity in woe most likely originates from full-lengthAdam17T265M encoded by the Adam17C794T transcript identified expressed at severely reduced levels. Theseresults show that even small amounts of functional Adam17 allow woe mice to survive into adulthood. Incontrast to Adam17�/� mice that die at birth, the viability of woe mice provides an excellent opportunity forstudying the role of Adam17 throughout postnatal development and homeostasis.

THE waved with open eyes (woe) mouse was identifiedvia a screen for mouse models of syntenic human

ocular disorders (Chang et al. 2005). Initially the locuswas named waved3 (wa3) (Chang et al. 2005) and hassince been renamed woe. Phenotypic analysis of woemice identified ‘‘eyelids open at birth’’ (EOB), wavy coat,and enlargements of the heart and esophagus. The woemutation arose spontaneously on the C57BL/6 back-ground and exhibits recessive inheritance; coarse linkageanalysis assigned woe to mouse chromosome 12 (Chang

et al. 2005).The wavy coat observed in woe mice has been pre-

viously described in mice with altered epidermal growthfactor receptor (Egfr) signaling (Schneider et al. 2008).Egfr belongs to a family of tyrosine kinase receptors.Following ligand binding to the extracellular domain,Egfr is dimerized and autophosphorylated, which sub-

sequently induces an intracellular signaling cascade(Schneider et al. 2008). Mice carrying different muta-tions in Egfr such as wa2, wa5, and velvet, as well as micecarrying mutations in the transforming growth factor a

(Tgfa) all exhibit wavy fur (Luetteke et al. 1993, 1994;Miettinen et al. 1995; Thaung et al. 2002; Fitch et al.2003; Du et al. 2004). Tgfa is a member of the Egfr familyof ligands that binds to Egfr (Harris et al. 2003). Egfrsignaling has been implicated in the differentiation andmaturation of the hair follicle (Hansen et al. 1997).Interestingly Egfr and Tgfa mouse mutants, in additionto the wavy coat phenotype, also demonstrate the EOBphenotype (Luetteke et al. 1993, 1994; Miettinen et al.1995; Thaung et al. 2002; Fitch et al. 2003; Du et al.2004). During mammalian embryonic eye develop-ment, at the tip of the newly formed eyelids, epithelialsheets extend over the cornea and move toward thecenter of the eye resulting in eyelid closure (Li et al.2003). In mice with Egfr signaling defects there is afailure of the migration of the leading edges andconsequent failure of embryonic eyelid closure. Egfr

1Corresponding author: Department of Cell Biology, Neurobiology andAnatomy, Human and Molecular Genetics Center, Medical College ofWisconsin, 8701 Watertown Plank, Milwaukee, WI 53226.E-mail: dsidjani@mcw.edu

Genetics 185: 245–255 (May 2010)

signaling has been established as one of the essentialsignaling pathways required for leading eyelid edgemigration and the formation of actin stress fibers (Xia

and Karin 2004).Even though woe mice phenotypically resemble mice

with mutations in Tgfa and Egfr, linkage assignment ofthe woe locus to chromosome 12 excluded the possibilitythat woe is allelic with either Tgfa on chromosome 6or Egfr on chromosome 11. While initial mapping ofwoe demonstrated linkage to chromosome 12, the exactposition within the chromosome was not determined(Chang et al. 2005). Therefore, as a part of this study, wepositionally cloned and identified the mutation respon-sible for the woe phenotype. Our results show that woe is amutation in Adam17, a member of a family of membrane-anchored metalloproteinase disintegrins that playsan essential role in the proteolytic cleavage and releaseof membrane bound precursors termed ‘‘shedding’’(Blobel 2005). Adam17 has been implicated in theshedding of numerous proteins involved in multiplesignaling mechanisms (Huovila et al. 2005; Mezyk et al.2003). However, the primary role of Adam17 has beenestablished as a mediator of Egfr signaling via shedd-ing of membrane-bound precursor forms of epiregulin,Tgfa, amphiregulin, and Hbegf (Peschon et al. 1998;Sahin et al. 2004; Horiuchi et al. 2007). Functionalanalysis of the woe allele in this study provides evidencethat woe is a hypomorphic mutation in Adam17. Sincewoe mice are viable, whereas null Adam17 mice dieat birth, the woe mutation provides an excellent oppor-tunity for studying the role of Adam17 in postnataldevelopment and, specifically, has helped uncover acritical role for Adam17 in the development of theanterior segment of the eye and meibomian glands,most likely caused by defects in Adam17-dependentEgf-signaling.

MATERIALS AND METHODS

Mice: C57BL/6, woe, and C3H/HeJ mice were obtained fromJackson Laboratories (Bar Harbor, ME). Adam17DZn/1 mice wereobtained from Roy Black, Amgen (Seattle, WA). All miceexhibited normal breeding patterns. Progeny were genotypedutilizing PCR protocols (primer sequences are summarized inTable 1) as previously described (Talamas et al. 2006).

Histology: For embryo analysis, embryonic day 0.5 (E0.5)was defined as the morning of the day that a vaginal plug wasfirst observed. Collected tissues were fixed in 4% paraformal-dehyde, Zn-formalin, or Davidson’s solution (Miething et al.2006), embedded in paraffin, serially sectioned to 4 mm, andstained with H&E. The sections were photographed with aDXM1200 camera (Nikon) on a BX50 microscope (Olympus)and Nikon DS-Fi1 camera on Nikon Eclipse 80i microscope.

Cell culture: Primary mouse embryonic fibroblasts (mEFs)were generated from C57BL/6, woe, and Adam17DZn/DZn em-bryos at E13.5 and cultured as described previously (Weskamp

et al. 2002).Linkage mapping: The woe locus is on the congenic C57BL/

6 background. The woe/woe mice were outcrossed to C3H/HeJand F1 mice were backcrossed to woe/woe to generate 138

progeny. At 3 weeks of age, F2 mice were scored for the wavycoat appearance/small eye phenotype, killed, and genotypedwith microsatellite markers: D12Mit12, D12Mit147, D12Mit52,D12Mit263 as previously described (Talamas et al. 2006). Theresulting linkage data were analyzed with Map Manager QTX(http://www.mapmanager.org/mmQTX.html).

Sequence analysis of Adam17: Exon scanning primers weredesigned to anneal 50 bp from intron/exon boundaries (Table1) and analyzed as described previously (Talamas et al. 2006).Comparative sequence analysis was performed using DNAStarsoftware (Madison, WI). For cDNA evaluation RNA wasisolated from woe mEFs, reverse transcribed, and amplifiedas previously described (Talamas et al. 2006) using primers inTable 1. The PCR products were electophoresed, gel extracted,and sequenced as previously described (Talamas et al. 2006).For semiquantitative analysis of Adam17 RT–PCR productswere generated while in the exponential phase of amplifica-tion and Gapdh was used as an internal control (Table 1). PCRband intensities were quantified using ImageJ software(http://rsbweb.nih.gov/ij/) and are expressed relative toGapdh. The results represent at least three independentexperiments performed in triplicates. Comparison betweenwild-type and woe groups was analyzed by Student’s t-test anddata are expressed as mean 6 SEM. A value of P , 0.05 wastaken as statistically significant.

Western blot analysis: Wild-type and woe mEFs were lysed for15 min on ice in PBS containing 1% TritonX-100 10 mm

phenantroline, 4 mm Marimastat, and 13 protease inhibitorcocktail. Following centrifugation at 15,000g for 10 min toremove nuclei and cell debris, glycoproteins from the clearedlysates were concentrated with Concanavalin A sepharosebeads at 4� for 4 hr. Bound glycoproteins were eluted inLaemmli sample loading buffer with DTT for western blottingfollowed by alkylation with 20 mm iodoacetamide for 30 minat room temperature. Following electrophoresis on 9% SDS-polyacrylamide gels, the samples were transferred to nitrocel-lulose membranes (Biotrace, Pall Corporation, Port Washing-ton, NY), blocked in 3% dry milk reconstituted in PBS 0.05%Tween-20 (PBS-T), and then incubated with an antiserumagainst the cytoplasmic domain of Adam17 (Schlondorff

et al. 2000) in PBS-T for 1 hr at room temperature. Themembrane was washed three times with PBS-T followed byincubation with horseradish peroxidase-conjugated affinitypurified anti-rabbit IgG (H1L) (Promega, Madison, WI) for45 min at room temperature. Bound antibodies were detectedby enhanced chemiluminescence using an ECL detection re-agent (GE Healthcare–Amersham Biosciences, Piscataway, NJ)and a Bio-Rad Geldoc (Bio-Rad, Hercules, CA) instrument.

Adam17 activity assay: Endogenous Adam17 activity in wild-type, woe, and Adam17DZn/DZn mEF cell lysates was evaluated usingthe SensoLyte 520 TACE activity assay protocol (AnaSpec, SanJose, CA) following the manufacturer’s protocol. A Michaelis–Menten curve was constructed for FRET–Adam17 peptidesubstrate at 0, 2.5, 5, 10, 20, 40, 80 mm concentrations followingthe manufacturer’s protocol. Fluorescence intensity was mon-itored every 5 min over a period of 6 hr (excitation, 490 nm,6-nm slit; emission, 520 nm, 6-nm slit) and measured usingan LS55 Luminescence Spec (Perkin-Elmer, Waltham, MA).Samples were normalized to the FAM standard and proteinconcentration. Data from at least three independent ex-periments were compiled and the Km and Vmax values werecalculated using GraphPad Prism software (La Jolla, CA). Com-parison between groups was analyzed by Student’s t-test anddata are expressed as mean 6 SEM.

Ectodomain shedding assay: These assays were performedas previously described (Le Gall et al. 2009). Briefly, primarywild-type and woe mEFs were transiently transfected by GenJettransfection reagent (SignaGen, MD) with a vector containing

246 E. L. Hassemer et al.

TABLE 1

PCR primers used to amplify genomic and cDNA segments

Target Primer name Primer sequence (59 / 39) Annealing temp.

Adam17 Adam17 1.2 F CCCAATGTGAGCAGTTTCCCGAAC 58�Exon 1 Adam17 1.2 R CGCGTCCCTCCAATCACTCTGGAdam17 Adam17 2.1 F TGAGCAGCGAGTTAATGCTCTT 58�Exon 2 Adam17 2.2 R CATCAAGTACAAATCTTTACGAAdam17 Adam17 3 F TGGGTGGCTATTGTTCATCT 58�Exon 3 Adam17 3 R TTCCATCATTGCAGAAAGTGAdam17 Adam17 4 F GGTCAGCATGGGCTATAAGA 58�Exon 4 Adam17 4 R CCTGCTGATTCATTTTCCTGAdam17 Adam17 5 F TGTTTAGAAGGTGGTCTTGG 60�Exon 5 Adam17 5 R GCCAGAAACAGTATGATGTGCAdam17 Adam17 6 F AAAACCCTCCTTGTTTTGATG 60�Exon 6 Adam16 6 R AAATCTCTGCAGCCTTCTCTTAdam17 Adam17 7 F CCACTAGATGCCTATTCTAGTAGGTT 60�Exon 7 Adam17 7 R TCAGGAAGGTTTTGGAAGAAAdam17 Adam17 8 F TCAGTGTCTTAGGATGTTTTGG 60�Exon 8 Adam17 8 R AAGGTTTCTACCATGTTCTTTCAAdam17 Adam17 9/10 F GTAATGTTGAGGTGCCCTTG 58�Exon 9–10 Adam17 9/10 R ATGCTTATGTGCATGTGGTGAdam17 Adam17 11 F CCCAACTGGCAAAAATAACT 62�Exon 11 Adam17 11 R TTTCTGTTAACATGGCATCGAdam17 Adam17 12 F GACTGCTTGACACAGTGGTG 58�Exon 12 Adam17 12 R CTGCATTTGCAATCTCCTGTAdam17 Adam17 13 F AACCATAGCATTTGATGTTGG 58�Exon 13 Adam17 13 R CACCTGAAGGAAACTTGACCAdam17 Adam17 14 F GCTGGGTCCACATTTTTATG 58�Exon 14 Adam17 14 R CCCTATTTGGCTTCCTCACTAdam17 Adam17 15 F ATCTGGGAGTAAGGCCAAAG 58�Exon 15 Adam17 15 R AGACCAAGGCTGCTTAAGTGAdam17 Adam17 16 F TGGTTACTTGCCCATTGTTT 58�Exon 16 Adam17 16 R GTCATAACGGGAATCAGCTTAdam17 Adam17 17 F CCAGGAGACTCAGGAGAGGT 62�Exon 17 Adam17 17 R GAGAAGCTGTCTTGAACACTCAAdam17 Adam17 18.2 F GATGGCTGGGTTCCTTTAAAAT 58�Exon 18 Adam17 18.3 R CAGATCCTTTTAACTTCCACTAGCCAdam17 Adam17 19 F CATAGCCCAGGGTTACTGT 58�Exon 19 Adam17 19 R TCATTATAATCTATGTTTTGATTCAGGAdam17 Adam17�/� F CTCTGCAAGTGCAGTGACACTC 67�neo Adam17�/� R GGCCACTTGTGTAGCGCCAAGTAdam17 Adam171/1 F TGCGTCATTCCACTCCAACC 67�Without neo Adam171/1 R GCATCGCCTTCTTTCCTACCAAdam17 intron 5 Adam17 intron 5–1 F TTGCAGTCTCCAAAAGTATGTGG 58�

Adam17 intron 5–1 R TTTTTCCAGACAGGGTTTCTGTATATCAdam17 intron 5–2 F CGCAGGCATTTAACCAGCACAdam17 intron 5–2 R CCAGGGTATTTGGAGGCATCTCTAdam17 intron 5–3 F CTCCATAAACACACTTTTGCTTGAdam17 intron 5–3 R AGTAAAAAGTAATATAAGGGGCTAGAGAGAdam17 intron 5–4 F CCATTGGTTTTGGATTCAGCAATAdam17 intron 5–4 R TTCTCCACGGCCCATGTATTTAT

Adam17 intron 6 Adam17 intron 6–1 F ACGAGCTGAACCTAACCCCTTGA 58�Adam17 intron 6–1 R GGAGAGAATGGAAAGACGGCAATAdam17 intron 6–2 F TCCCTGGAATTGTTTTGTTTCTCAAdam17 intron 6–2 R GGGGAACACGTGTATGTATGTGCAdam17 intron 6–3 F TTCAAAGTGCATTTCTCACTGCTAAdam17 intron 6–3R ACCCTTTAAACCCTGCATTATCC

(continued )

woe Is a Mutation in Adam17 247

Tgfa, tumor necrosis factor a (Tnfa), selectin (Cd62l), orbetacellulin (Btc) with an alkaline phosphatase tag (AP-tag)(Sahin et al. 2004). For shedding experiments, transfectedcells were washed with OptiMEM that was subsequently re-placed after 1 hr with fresh OptiMEM with or without 25 ng/mlphorbol myristate acetate (PMA), 100 mm pervanadate (PV),or 2.5 mm ionomycin (IM). Both PMA and PV have beenpreviously shown to stimulate Adam17-dependent sheddingactivity whereas iononmycin stimulates both Adam10 andAdam17 shedding activity (Le Gall et al. 2009). To evaluatestimulated and constitutive shedding, cells were incubated for30 min and 6 hr, respectively. AP activity in the supernatantand the cell lysates was measured by colorimetry (Sahin et al.2004). The ratio of total AP activity in the supernatant andtotal AP activity in the cell lysate and supernatant wascalculated from two identically prepared wells and averagedwith all experiments repeated at least three times. This ratioreflects the activity of Adam17 toward the AP-tagged proteins(Le Gall et al. 2009). Shedding in wild-type cells is set to 1 toallow a comparison of constitutive shedding levels betweenwild-type and woe mEFs. Comparison between wild-type andwoe groups was analyzed by unpaired t-test and data areexpressed as mean 6 SEM. A value of P , 0.05 was taken asstatistically significant.

RESULTS

The woe phenotype: At birth, woe mice exhibit fullypenetrant EOB, with eyelids remaining open through-out postnatal development (Figure 1A); slit-lamp exam-ination revealed corneal opacities, corneal inflammation,and neovascularization. Histological analysis at E16.5

confirmed failure of eyelid fusion in woe eyes whencompared to age-matched wild-type mice (Figure 1B).The woe cornea appeared much thinner with a lessdifferentiated corneal epithelium, absent endotheliallayer, and a vacuolated lens (Figure 1B) suggesting thatwoe eyes have defects in the development of anteriorsegment structures. Therefore, we proceeded to evalu-ate the woe ocular phenotypes at P28, following com-pletion of anterior structure development (Smith

2002). The woe eyes exhibited corneal scarring, stromalkerititis, and neovascularization. Although the anteriorchamber was formed, there were variable anteriorsynechiae, with variable absence of the Descemet’s mem-brane and a small or absent Schlemm’s canal. The lensshowed cataracts, while the retina and the optic nerveremained normal. Histological analysis of P28 woe eyelidskin revealed evidence of chronic dermatitis with vari-able development of the hair follicles (Figure 1D). Atthe eyelid margin, there was excess pigment dispersalwithin the dermis and the complete absence of themeibomian glands (Figure 1D). We also examined theprogression of the woe ocular phenotype. We did notidentify any difference between woe eyes at P28 and at 12months (not shown). Although the external woe evalu-ation identified microphthalmia in most woe mice,about 20% of woe mice exhibited unilateral or bilateralanophthalmia (not shown). Histological sections ofthe presumed anophthalmic eyes revealed aphakic

TABLE 1

(Continued)

Target Primer name Primer sequence (59 / 39) Annealing temp.

Adam17 intron 7 Adam17 intron 7–1 F CCGAGTTGATGACATATACCGGAAC 58�Adam17 intron 7–1 R CCACTGAGGATCAGAAGGAGAGAAGAdam17 intron 7–2 F GGGGTTCTGATGGAATCCTACAGTTAdam17 intron 7–2 R AGTCCTTGGGGAAGGGGTTAGAGAdam17 intron 7–3 F CACTGAGGATGAATTGGAGGGTGTAdam17 intron 7–3R CGACAGTTCTTCCGAAGGTCCAAdam17 intron 7–3.2F CAGAATTAACCTTGACCTTCTGCTTAdam17 intron 7–3.2R CCGACAGTTCTTCCGAAGGTAdam17 intron 7–4 F CAAATGTGGAGGTTGTGAGAATATAAdam17 intron 7–4 R TGCAAGTTCATAACCTTCCATAACAdam17 intron 7–5 F TTTGAGCTAAACAATTTCAGTTGCAdam17 intron 7–5 R TCTCTTCGTTTGGGAAACTTTTT

Adam17 cDNA RTAdam17 F 59-UTR 67 CCCAATGTGAGCAGTTTCCC 58�RTAdam17 R568 CCTTCGTCGAGAAGTGATAGRTAdam17 F377 GGGTTCTAGCCCACATAGGAGARTAdam17 R937 CCCAAGCATCCTTCTCTTCGTTTRTAdam17 F768 CCGAGTTGATGACATATACCGGAARTAdam17 R1461 ACTCCAGGGTGGATGAAGGAGAGRTAdam17 F1305 GTATCCCATAGCTGTGAGCGRTAdam17 R2190 AAAGGGCTTGATGATGCGAARTAdam17 F2015 TCGTTGGGTCTGTTCTGGTTTRTAdam17 3’-UTR 82 TCATTATAATCTATGTTTTGAT

semi-quantitative RT–PCR RT-Adam17 F377 GGGTTCTAGCCCACATAGGAGA 58�RT-Adam17 R937 CCCAAGCATCCTTCTCTTCGTTT

Gapdh GAPDH F CTTTGGCATTGTGGAAGGG 58�GAPDH R CCTCTCTTGCTGCAGTGTC

248 E. L. Hassemer et al.

eyes with highly disorganized retinas suggesting severemicrophthalmia (Figure 1E).

External evaluation of woe mice confirmed the wavyfur phenotype (Chang et al. 2005) (Figure 2A), al-though we did not observe any obvious defect of thevibrissae (not shown). Histological analysis of the skinidentified variable development of the hair follicle with-out any abnormalities of the skin (not shown). Analysisof woe hearts at P0.5 did not consistently identify anysignificant size differences; however, the analysis identi-fied hypertrophic and thickened pulmonary and aorticvalves as compared to wild-type mice (Figure 2B). Incontrast, the tricuspid and mitral valve thicknesses werewithin normal limits (Figure 2B). Evaluation of theesophagus did not identify any differences between woeand wild-type animals (not shown).

Positional cloning of the woe locus: Our mappinganalysis confirmed linkage to mouse chromosome 12(Chang et al. 2005) and further mapped woe to a region

of 4 cM between the centromere and D12Mit147.No recombinants were identified between woe andD12Mit12 (Figure 3A). The woe critical region encom-passed 36.4 Mb containing �130 candidate genes. Inparallel to our positional cloning efforts, we searchedMGI phenotype database (http://www.informatics.jax.org/phenotypes.shtml) for genetically altered micewith a similar phenotype that may map within the woecritical region. In silico evaluation identified twoAdam17�/�mouse models that exhibited developmentaldefects similar to woe mice including failure of eyeliddevelopment, hair, and heart defects (Peschon et al.1998; Horiuchi et al. 2007). Thus, we proceeded toevaluate Adam17 as a candidate gene.

Sequence analysis identified a C794T substitution inexon 7 (Figure 3B) that leads to a T265M amino acidchange in the metalloproteinase domain of Adam17.Threonine 265 is a highly evolutionarily conserved aminoacid in Adam17 suggesting conservation of function

Figure 1.—Eye phenotypes in woe mice. (A) All woe mice are born with open eyelids that remain open throughout postnataldevelopment (bottom row); woe eyes appear smaller compared to age-matched wild-type eyes (top row). (B) Histological analysis atE16.5 confirmed failure of embryonic eyelid closure in woe mice compared to the age-matched wild-type mice. Boxes indicateregions of higher magnification showing that in woe mice the corneal endothelium has not developed and the cornea is muchthinner than wild-type mice. (C) At P28, woe eyes exhibit defects consistent with anterior segment dysgenesis. The arrowheadindicates the anterior synechiae observed in woe mice. (D) Histological analysis of woe eyelids shows that the meibomian glands(arrow) do not develop in woe eyelids. (E) Histological analysis of a subset of woe mice, which exhibit severe microphthalmia(right) consequent to aphakea. (Scale bars, 100 mm).

woe Is a Mutation in Adam17 249

(Figure 3C). Evaluation of the SNP database did not revealany polymorphisms for Adam17 exon 7 (http://www.ncbi.nlm.nih.gov/SNP/snp_ref.cgi?locusId¼11491).These findings collectively suggested that Adam17C794T islikely not a common polymorphism. To exclude a pos-sibility that Adam17C794T is a rare previously unidentifiedpolymorphism and to unequivocally prove that theAdam17 C794T is the woe-causing mutation, we utilized a ge-netic approach. Although Adam17DZn/DZn mice die at birth,Adam17DZn/1 mice are viable and phenotypically normal(Peschon et al. 1998). We set up breedings betweenAdam17DZn/1 and woe/woe mice. About 50% of F1 progenyappeared phenotypically normal (Figure 3D) and 50%progeny exhibited the EOB phenotype (Figure 3E) con-firmed by histological analysis (Figure 3F and 3G).Genotyping revealed that unaffected progeny carriedthe Adam17 wild-type and woe alleles and the progeny ex-hibiting EOB were compound heterozygotes Adam17DZn/Adam17 C794T indicating a failure to complement (Figure2E). This finding provided genetic evidence that theC794T substitution in Adam17 is the woe phenotypecausing mutation. The Adam17 DZn/Adam17 C794Tcompoundheterozygotes survived into adulthood and exhibitedsimilar phenotypes to woe mice (not shown).

Functional analysis of the woe mutation: RT–PCRanalysis of wild-type mEFs identified a single Adam17transcript matching the Adam17 ORF sequence(NM_009615). In contrast, RT–PCR in woe mEFs iden-tified three Adam17 transcripts: full-length Adam17C794T,Adam17Dexon7 lacking exon 7, and Adam17Dexons6-7 lackingexons 6 and 7 (Figure 4, A–D). To further evaluatethe mechanism of aberrant Adam17 splicing in woe, wesequenced introns 5, 6, and 7. Sequence comparisonbetween C57BL/6 and woe did not identify any largegenomic deletions, insertions, or rearrangements inthese three introns. The sequence differences betweenC57BL/6 and woe closest to the intron/exon junctionswere identified in intron 5 at IVS51107A–T and IVS5-266G–A, in intron 6 at IVS6180A–T and IVS6-821C–T

and in intron 7 at IVS71117-2 and IVS-20A–G. None ofthese intronic sequence differences seemed likely to beresponsible for Adam17 aberrant splicing in woe mEFs.Therefore, we explored the possibility that the Adam17C794 nucleotide in exon 7 may be a part of a puta-tive exonic splicing enhancer (ESE) sequence. To thisaim, wild-type and woe exon 7 Adam17 sequences wereanalyzed by a web-based program called ESEfinder(Cartegni et al. 2003). The results showed that theC794T substitution eliminated a putative binding sitefor the splicing factor SRp55 and decreased the scorefor a binding site for another splicing factor SF2/ASF(data not shown).

To assess the levels of precursor and mature formsof Adam17 in wild-type and woe mEFs, we performedWestern blots. In cell lysates from wild-type mEFs,Adam17 was detected as a pro-form of �120 kDa and amature form of �100 kDa (Figure 5A), similarly toprevious reports for Adam17 (Schlondorff et al. 2000).In contrast, in woe mEFs an Adam17 immunoreactiveband of an apparent mass of 110 kDa was detected, mostlikely reflecting the precursor Adam17D252-281 proteinencoded by the Adam17Dexon7 transcript (Figure 5A). Wedid not detect Adam17 immunoreactive bands corre-sponding to the precursor and mature protein formsencoded by Adam17C794T.

To determine if woe mEFs exhibited any defects inshedding activity, we evaluated the cleavage of a fluoro-genic FRET-peptide Adam17 substrate derived from the10 amino acid sequence (LAQAVRSSSR) surroundingthe scissile bond of the Adam17 cleavage site from theTnfa precursor (Jin et al. 2002). The Tnfa precursor isa well-established substrate of Adam17 (Black et al.1997), and cleavage between the A-V scissile bond of theTnfa LAQAVRSSSR peptide has been previously estab-lished as a reliable measure of the Adam17 activity (Jinet al. 2002). Although the Adam17 fluorogenic FRET-peptide was established as highly specific for Adam17,some proteolytic activity was also reported for other

Figure 2.—Fur and cardiac valve phenotypes in woe mice. (A) woe mice develop wavy fur. (B) Histological evaluation of thecardiac valves (arrowheads) of P0.5 wild-type and woe hearts identified thickening of the semilunar (aortic and pulmonary) valvesin woe mice. Atrioventricular (mitral and tricuspid) valves in woe did not differ from wild-type controls. (Scale bars: overview imagesscale bars, 100 mm; valve images scale bars, 50 mm).

250 E. L. Hassemer et al.

metalloproteinases and this substrate (Jin et al. 2002).To establish if any of the FRET-peptide cleavage occursindependent of Adam17, we also evaluated the celllysates from Adam17DZn/DZn mEFs. Under the conditionsof the assay, cell lysates from Adam17DZn/DZn and woe mEFsshowed a 5.0-fold and a 2.8-fold Vmax decrease re-spectively relative to the Vmax from the wild-type mEFs(Figure 5B). No major differences were observed in Km

between the three cell lysates (Figure 5B).

To further evaluate the Adam17 shedding activity, wetransfected both wild-type and woe mEFs with alkalinephosphatase-tagged Adam17-substrates: l-selectin,Tgfa, and Tnfa. Our results showed a significant de-crease in the constitutive shedding of all Adam17substrates in woe mEF cells when compared to wild-typecontrols (Figure 5C). To determine whether the woemutation may have had an epigenetic effect on thefunction of Adam10, we also evaluated shedding of

Figure 3.—Genetic analysis of the woe locus. (A) Utilizing 138 meioses, the woe locus was mapped between the centromere ofchromosome 12 and D12Mit147. The woe locus cosegregated with D12Mit12. (B) Sequence analysis of woe DNA identified a C794Tsubstitution in Adam17. (C) The C794 substitution results in a T265M substitution. T265 (arrow) shows a high level of conservationbetween human (NP_003174.3), chimpanzee (XP_515293.2), cattle (XP_595713.2), mouse (NP_033745.4), rat (NP_064702.1),chicken (NP_001008682.1), zebrafish (XP_689147.1), fruit fly (NP_733334.1), and mosquito (XP_312572.2). (D–G) Complemen-tation breeding of Adam17DZn/1 to woe mice resulted in progeny that exhibited the wild-type phenotype (D and F) and progeny thatexhibited the open eyelid phenotype (E and G). (H) Genotyping of mice from the complementation breeding: lanes 1–3 depictsPCR products amplifying to the neo cassette and lanes 4–6 depict PCR products for exon 7 that were subsequently sequenced toconfirm the presence of the woe allele. Lanes 1 and 4 contain genomic DNA from the mouse depicted in D and lanes 2 and 5contain DNA from the mouse depicted in E. Lanes 3 and 6 are no template controls and M ¼ FX174 DNA–HaeIII Digest. Theresults show that affected mice depicted in E and G are compound heterozygotes Adam17DZn/Adam17woe indicating failure of theAdam17DZn allele to complement the woe allele.

woe Is a Mutation in Adam17 251

Adam10 substrate Btc. No difference in the constitutiveshedding of Btc was noted between wild-type and woemEFs (Figure 5C). The results from woe mEFs showedshedding of all three Adam17 substrates indicating thatwoe mEFs still respond to stimulation (Figure 5D), eventhough the levels of unstimulated shedding of theseAdam17 substrates were significantly reduced com-pared to wild-type controls (Figure 5C).

DISCUSSION

In this study, we describe a spectrum of phenotypes inwoe mice resulting from a C794T substitution in Adam17that disrupts a putative ESE sequence required forbinding of SRp55 and SF2/ASF proteins. The role ofESE sequences is to provide binding sites for serine/arginine (SR) proteins (Cartegni et al. 2002). BothSRp55 and SF2/ASF belong to a family of SR proteinsthat when bound to ESE promote exon inclusion byrecruiting components of the splicing machinery. Nu-merous reports have shown that disruption of ESEsequences can cause the splicing machinery to skipthe mutant exon with dramatic effects on proteinstructure of the gene product (Cartegni et al. 2002).In woe, the C794T substitution results in a predominant

Adam17Dexon7 transcript that lacks exon 7 resulting in anin-frame deletion and a putative protein with residues252–281 missing from the metalloprotease domain.Western blot analysis identified only the precursor formof the truncated Adam17D252-281 protein. Similar to othermembers of the Adam family of proteins, prodomainremoval is required for Adam17 to exhibit sheddingactivity (Schlondorff et al. 2000). The prodomain actsas an inhibitor of protease activity via a cysteine-switchmechanism where a free cysteine residue from theprodomain interacts with the zinc ion from the activesite (Schlondorff et al. 2000). Thus, presence of theprodomain would render the precursor form ofAdam17D252-281 functionally inactive.

Even though the predominant Adam17Dexon7 transcriptmost likely leads to a functionally inactive precursorAdam17D252-281 protein, our shedding analysis experi-ments showed that woe mEFs exhibit residual sheddingof Adam17 substrates. We cannot exclude the possibilitythat trace amounts of Adam17D252-281, below detection viaWestern blotting, may undergo maturation and mayexhibit shedding activity. This, however, is unlikely sincea loss of amino acid residues 252–281 presumably wouldalso prevent proper folding and intracellular degrada-tion before the protein can enter the secretory pathway

Figure 4.—Aberrant splicing of Adam17 in woe mEFs. (A) RT–PCR identified a single Adam17 transcript and three Adam17transcript variants in wild-type and woe mEFs respectively. RT–PCR of glyceraldehyde-3-phosphate dehydrogenase (Gapdh) wasexamined as a control. Right": the semi-quantitative analysis of the amount of full-length Adam17 expressed in wild-type mEFsand full-length Adam17C794T expressed in woe mEFs as the ratio of the sample to the internal standard (Gapdh). (B–D) sequenceanalysis of the three Adam17 transcripts identified in woe mEFs: (B) sequence of the full-length Adam17C794T transcript indicating Cto T substitution at nucleotide 794 (arrow), (C) sequence analysis showing that the predominant Adam17Dexon7 transcript in woemEFs exhibits skipping of exon 7, (D) sequence of Adam17Dexon6-7 transcript exhibiting skipping of exons 6 and 7.

252 E. L. Hassemer et al.

en route to the cell surface (Suzuki et al. 1998). Theresidual Adam17 shedding observed in woe mEFs mostlikely originates from the full-length Adam17C794T tran-script that encodes a putative full length Adam17T265M

protein. In woe mEFs, Adam17C794T is expressed atseverely reduced levels when compared to Adam17expressed in wild-type mEFs. Our inability to detectprecursor and mature Adam17T265M protein likely is dueto the levels of Adam17T265M protein below detection viaWestern blotting. Nevertheless, constitutive and stimu-lated shedding of Adam17 substrates such as Tgfa, Tnfa,and l-selectin can be detected in woe mEFs, albeit atsignificantly reduced levels compared to wild-type con-trols. Our results are further supported by the observa-tion that woe mice survive into adulthood, in contrast toAdam17�/� mice (Peschon et al. 1998; Horiuchi et al.2007). These results imply that T265M substitution doesnot significantly affect shedding activity of Adam17T265M.

Evidently, even the small amount of functional Adam17is sufficient to allow woe mice to survive.

Adam17�/� mice exhibit enlargement of the semi-lunar and atrioventricular cardiac valves (Jackson et al.2003; Horiuchi et al. 2007). Histological analysis of woehearts identified enlargement only in the semilunarvalves (Figure 2B). Previous in vivo and in vitro studieshave established Adam17 as the principal sheddase ofHbegf and subsequent Egfr transactivation essential forvalvulogenesis ( Jackson et al. 2003; Sahin et al. 2004).Differences in valve phenotypes between woe and nullAdam17 mice may be due to the reduced, but notabolished, Hbegf-mediated Egfr signaling in woe. It hasbeen hypothesized that reduced Egfr signaling issufficient for normal atrioventricular valve develop-ment, but insufficient for semilunar valve development( Jackson et al. 2003). As a part of this study, weevaluated shedding activity of three Adam17 substrates

Figure 5.—Functional analysis of the woe allele. (A) Western blot analysis shows presence of the Adam17 precursor form of120 kDa (left arrowhead) and mature form of 100 kDa (arrow) in wild-type mEFs; in woe mEFs a 110-kDa Adam17 immunoreactiveband was identified (right arrowhead). Protein molecular mass markers (kDa) are indicated on the right. (B) Proteolytic cleavageof the FRET-peptide cleaved at different concentrations following incubation with cell lysates from wild-type, woe, andAdam17DZn/DZn mEFs. Vmax and Km values were calculated from the Michaelis–Menten curve as indicated in materials and meth-

ods. (C) Constitutive shedding analysis in wild-type (solid bars) and woe (lined bars) mEFs following transfection with AP-tagged l-selectin (Cd62l), Tgfa, Tnfa, and Btc. (D) Stimulated shedding analysis in untransfected woe mEFs (solid bars) and woe mEFstransfected with l-selectin (Cd62l), Tgfa, Tnfa, and Btc (shaded bars) following stimulation with 25 ng/ml PMA, 100 mM PV,or 2.5 mM IM.

woe Is a Mutation in Adam17 253

(Tgfa, Tnfa, and l-selectin) in woe, but we did notevaluate shedding of pro-Hbegf. Given that Adam17 isthe principal sheddase of Hbegf, we expect that reducedpro-Hbegf shedding and consequent reduced Egfr sig-naling is responsible for the woe cardiac phenotype.

The milder cardiac phenotype observed in woe mice ismost likely responsible for the postnatal survival of woewhen compared to Adam17�/� mice. This is especiallyrelevant for the evaluation of ocular phenotypes. Al-though formation of the anterior segment is initiated atE10, the differentiation and maturation of most ante-rior segment structures is not complete until P21(Smith 2002; Cvekl and Tamm 2004). The woe anteriorsegment dysgenesis phenotype reported here shares fea-tures previously noted for the anterior segment dysgen-esis phenotypes of Tgfa�/� and Egfr�/� mice (Luetteke

et al. 1993; Miettinen et al. 1995). The role of Tgfa/Egfrsignaling has been implicated in the formation of thecorneal endothelium (Reneker et al. 1995). WhetherAdam17 plays an essential role in mediating only theTgfa activation of Egfr or Adam17 mediates signaling ofother ligands/molecular pathways essential for anteriorsegment development is unclear. Histological analysis ofwoe eyelids showed an absence of the meibomian glands,thus identifying a novel role of Adam17 in normal eyeliddevelopment and function. Meibomian glands are spe-cialized sebaceous glands located at the rim of the eyelidthat secrete sebum, a lipid substance that prevents evap-oration of the precorneal tearfilm. Absence of meibo-mian glands has been used as one of the most commonophthalmic clinical features in patients affected withectodermal displasia, a genotypically and phenotypicallyheterogeneous group of inherited disorders that arecharacterized by primary defects in the development oftwo or more tissues derived from embryonic ectoderm(Kaercher 2004). Meibomian glands are ectodermallyderived appendages that develop as a result of the in-teraction between the ectoderm and the underlyingmesoderm in the skin. Recently, the role of Egfr sig-naling has also been proposed as the promoter of theintrafollicular epidermal fate at the expense of hairfollicle development (Richardson et al. 2009). It ispossible that Egfr signaling also promotes the develop-ment of the meibomian glands in the eyelids; mice witha mutation in Tgfa exhibit absence of the meibomianglands (Luetteke et al. 1993). However, absence ofmeibomian glands was not reported for Egfr�/�, but thatphenotype may have been overlooked due to prenataland perinatal lethality in null Egfr mice (Miettinen

et al. 1995). Thus, we cannot exclude the possibility thatthe role of Adam17 in the meibomian gland develop-ment may be independent from Egfr signaling. There-fore, further analyses will address the role of Adam17 inthe meibomian gland formation either dependent orindependent from the Egfr signaling pathway.

In summary, we describe characterization of the woemouse mutant. Genetic analysis identified a C794T

substitution in Adam17 resulting in aberrant Adam17splicing and rendering the majority of Adam17 pro-tein nonfunctional. However, a small amount of func-tional Adam17 is produced in woe animals, sufficientto support shedding of Adam17 substrates, albeit atsignificantly reduced levels. The characterization of thedifferent forms of Adam17 in the woe mutant mouse andtheir functional properties in woe mEFs provides aplausible explanation for the observed hypomorphicAdam17 phenotype. Since woe mice are viable, whereasAdam17�/� mice die at birth, the woe mutation providesan excellent opportunity for studying the role of Adam17in postnatal development, and specifically helped un-cover a critical role for Adam17 in the development ofthe anterior segment of the eye and meibomian glands,most likely caused by defects in Adam17-dependent Egfsignaling.

We thank Roy Black at Amgen, Seattle, for generously providingAdam17DZn/1 mice. This work was supported in part by NationalInstitutes of Health grant EY15173 and EY018872 (D.J.S.), andEY15719 and GM64750 (C.P.B.).

LITERATURE CITED

Black, R. A., C. T. Rauch, C. J. Kozlosky, J. J. Peschon, J. L.Slack et al., 1997 A metalloproteinase disintegrin that re-leases tumour-necrosis factor-alpha from cells. Nature 385:729–733.

Blobel, C. P., 2005 ADAMs: key components in EGFR signallingand development. Nat. Rev. Mol. Cell Biol. 6: 32–43.

Cartegni, L., S. L. Chew and A. R. Krainer, 2002 Listening to si-lence and understanding nonsense: exonic mutations that affectsplicing. Nat. Rev. Genet. 3: 285–298.

Cartegni, L., J. Wang, Z. Zhu, M. Q. Zhang and A. R. Krainer,2003 ESEfinder: a web resource to identify exonic splicing en-hancers. Nucleic Acids Res. 31: 3568–3571.

Chang, B., N. L. Hawes, R. E. Hurd, J. Wang, D. Howell et al.,2005 Mouse models of ocular diseases. Vis. Neurosci. 22:587–593.

Cvekl, A., and E. R. Tamm, 2004 Anterior eye development and oc-ular mesenchyme: new insights from mouse models and humandiseases. Bioessays 26: 374–386.

Du, X., K. Tabeta, K. Hoebe, H. Liu, N. Mann et al., 2004 Velvet, adominant Egfr mutation that causes wavy hair and defective eye-lid development in mice. Genetics 166: 331–340.

Fitch, K. R., K. A. McGowan, C. D. van Raamsdonk, H. Fuchs, D.Lee et al., 2003 Genetics of dark skin in mice. Genes Dev. 17:214–228.

Hansen, L. A., N. Alexander, M. E. Hogan, J. P. Sundberg, A.Dlugosz et al., 1997 Genetically null mice reveal a central rolefor epidermal growth factor receptor in the differentiation of thehair follicle and normal hair development. Am. J. Pathol. 150:1959–1975.

Harris, R. C., E. Chung and R. J. Coffey, 2003 EGF receptor li-gands. Exp. Cell Res. 284: 2–13.

Horiuchi, K., T. Kimura, T. Miyamoto, H. Takaishi, Y. Okada et al.,2007 Cutting edge: TNF-alpha-converting enzyme (TACE/ADAM17) inactivation in mouse myeloid cells prevents lethalityfrom endotoxin shock. J. Immunol. 179: 2686–2689.

Huovila, A. P., A. J. Turner, M. Pelto-Huikko, I. Karkkainen andR. M. Ortiz, 2005 Shedding light on ADAM metalloprotei-nases. Trends Biochem. Sci. 30: 413–422.

Jackson, L. F., T. H. Qiu, S. W. Sunnarborg, A. Chang, C. Zhang

et al., 2003 Defective valvulogenesis in HB-EGF and TACE-nullmice is associated with aberrant BMP signaling. EMBO J. 22:2704–2716.

254 E. L. Hassemer et al.

Jin, G., X. Huang, R. Black, M. Wolfson, C. Rauch et al., 2002 Acontinuous fluorimetric assay for tumor necrosis factor-alphaconverting enzyme. Anal. Biochem. 302: 269–275.

Kaercher, T., 2004 Ocular symptoms and signs in patients with ec-todermal dysplasia syndromes. Graefes Arch. Clin. Exp. Ophthal-mol. 242: 495–500.

Le Gall, S. M., P. Bobe, K. Reiss, K. Horiuchi, X. D. Niu et al.,2009 ADAMs 10 and 17 represent differentially regulated com-ponents of a general shedding machinery for membrane pro-teins such as transforming growth factor alpha, l-selectin, andtumor necrosis factor alpha. Mol. Biol. Cell 20: 1785–1794.

Li, G., C. Gustafson-Brown, S. K. Hanks, K. Nason, J. M. Arbeit

et al., 2003 c-Jun is essential for organization of the epidermalleading edge. Dev. Cell 4: 865–877.

Luetteke, N. C., T. H. Qiu, R. L. Peiffer, P. Oliver, O. Smithies

et al., 1993 TGF alpha deficiency results in hair follicle andeye abnormalities in targeted and waved-1 mice. Cell 73: 263–278.

Luetteke, N. C., H. K. Phillips, T. H. Qiu, N. G. Copeland, H. S.Earp et al., 1994 The mouse waved-2 phenotype results froma point mutation in the EGF receptor tyrosine kinase. GenesDev. 8: 399–413.

Mezyk, R., M. Bzowska and J. Bereta, 2003 Structure and func-tions of tumor necrosis factor-alpha converting enzyme. Acta Bio-chim. Pol. 50: 625–645.

Miettinen, P. J., J. E. Berger, J. Meneses, Y. Phung, R. A. Pedersen

et al., 1995 Epithelial immaturity and multiorgan failure inmice lacking epidermal growth factor receptor. Nature 376:337–341.

Miething, F., S. Hering, B. Hanschke and J. Dressler, 2006 Effectof fixation to the degradation of nuclear and mitochondrial DNAin different tissues. J. Histochem. Cytochem. 54: 371–374.

Peschon, J. J., J. L. Slack, P. Reddy, K. L. Stocking, S. W.Sunnarborg et al., 1998 An essential role for ectodomain shed-ding in mammalian development. Science 282: 1281–1284.

Reneker, L. W., D. W. Silversides, K. Patel and P. A. Overbeek,1995 TGF alpha can act as a chemoattractant to perioptic mes-

enchymal cells in developing mouse eyes. Development 121:1669–1680.

Richardson, G. D., H. Bazzi, K. A. Fantauzzo, J. M. Waters, H.Crawford et al., 2009 KGF and EGF signalling block hair fol-licle induction and promote interfollicular epidermal fate in de-veloping mouse skin. Development 136: 2153–2164.

Sahin, U., G. Weskamp, K. Kelly, H. M. Zhou, S. Higashiyama et al.,2004 Distinct roles for ADAM10 and ADAM17 in ectodomainshedding of six EGFR ligands. J. Cell Biol. 164: 769–779.

Schlondorff, J., J. D. Becherer and C. P. Blobel, 2000 Intracellularmaturation and localization of the tumour necrosis factor alphaconvertase (TACE). Biochem. J. 347(1): 131–138.

Schneider, M. R., S. Werner, R. Paus and E. Wolf, 2008 Beyondwavy hairs: the epidermal growth factor receptor and its ligandsin skin biology and pathology. Am. J. Pathol. 173: 14–24.

Smith, R., 2002 Systematic Evaluation of the Mouse Eye. CRC Press,Cleveland, OH.

Suzuki, T., Q. Yan and W. J. Lennarz, 1998 Complex, two-way trafficof molecules across the membrane of the endoplasmic reticu-lum. J. Biol. Chem. 273: 10083–10086.

Talamas, E., L. Jackson, M. Koeberl, T. Jackson, J. L. McElwee

et al., 2006 Early transposable element insertion in intron 9of the Hsf4 gene results in autosomal recessive cataracts inlop11 and ldis1 mice. Genomics 88: 44–51.

Thaung, C., K. West, B. J. Clark, L. McKie, J. E. Morgan et al.,2002 Novel ENU-induced eye mutations in the mouse: modelsfor human eye disease. Hum. Mol. Genet. 11: 755–767.

Weskamp, G., H. Cai, T. A. Brodie, S. Higashyama, K. Manova et al.,2002 Mice lacking the metalloprotease-disintegrin MDC9(ADAM9) have no evident major abnormalities during develop-ment or adult life. Mol. Cell. Biol. 22: 1537–1544.

Xia, Y., and M. Karin, 2004 The control of cell motility and epithe-lial morphogenesis by Jun kinases. Trends Cell Biol. 14: 94–101.

Communicating editor: L. Siracusa

woe Is a Mutation in Adam17 255