Matriptase/MT-SP1 Is Required for Postnatal Survival, Epidermal Barrier Function, Hair Follicle Development, and Thymic Homeostasis

Home , Matriptase

Matriptase/MT-SP1 is required for postnatal survival, epidermal barrier function, hair follicle development, and thymic homeostasis

Karin List1,4, Christian C Haudenschild5, Roman Szabo1, WanJun Chen2, Sharon M Wahl2, William Swaim3, Lars H Engelholm1,4, Niels Behrendt4 and Thomas H Bugge*,1

1Oral and Pharyngeal Cancer Branch, National Institute of Dental and Craniofacial Research, National Institutes of Health, 30 Convent Drive, Bethesda, Maryland, MD 20892, USA; 2Oral Infection and Immunity Branch, National Institute of Dental and Craniofacial Research, National Institutes of Health, 30 Convent Drive, Bethesda, Maryland, MD 20892, USA; 3Cellular Imaging Core, National Institutes of Dental and Craniofacial Research, National Institutes of Health, 30 Convent Drive, Bethesda, Maryland, MD 20892, USA; 4Finsen Laboratory, Strandboulevarden 49, DK-2100, Copenhagen, Denmark; 5Department of Experimental Pathology, American Red Cross, 15601 Crabbs Branch Way, Rockville, Maryland, MD 20855, USA

Matriptase/MT-SP1 is a novel tumor-associated type II Introduction transmembrane serine protease that is highly expressed in the epidermis, thymic stroma, and other epithelia. A Proteolytic modi®cation of the extracellular environ- null mutation was introduced into the Matriptase/MT- ment is essential for all aspects of life of a multicellular SP1 gene of mice to determine the role of Matriptase/ organism. Extracellular proteolysis regulates key pro- MT-SP1 in epidermal development and neoplasia. cesses such as cell growth, dierentiation, adhesion, Matriptase/MT-SP1-de®cient mice developed to term migration, and programmed cell death (Werb, 1997). but uniformly died within 48 h of birth. All epidermal Dysregulated extracellular proteolysis is also intimately surfaces of newborn mice were grossly abnormal with a associated with the genesis and progression of a dry, red, shiny, and wrinkled appearance. Matriptase/ number of chronic diseases including cancer, cardio- MT-SP1-de®ciency caused striking malformations of the vascular disease, and neuronal degeneration (Andrea- stratum corneum, characterized by dysmorphic and sen et al., 2000; Esler and Wolfe, 2001; Luttun et al., pleomorphic corneocytes and the absence of vesicular 2000; Matrisian, 1999; Werb et al., 1999). Several bodies in transitional layer cells. This aberrant skin families of secreted or cell surface-associated proteases development seriously compromised both inward and have been studied extensively in the context of outward epidermal barrier function, leading to the rapid physiological and pathological tissue remodeling. These and fatal dehydration of Matriptase/MT-SP1-de®cient include the plasminogen activation system, the matrix pups. Loss of Matriptase/MT-SP1 also seriously metalloproteinases, the ADAM proteases, the aected hair follicle development resulting in generalized ADAMTS proteases, and the cathepsins (Andreasen follicular hypoplasia, absence of erupted vibrissae, lack et al., 1986; Blobel, 2000; Tang, 2001; Turk et al., 2000; of vibrissal hair canal formation, ingrown vibrissae, and Vu and Werb, 2000). wholesale abortion of vibrissal follicles. Furthermore, The type II transmembrane serine proteases (TTSPs) Matriptase/MT-SP1-de®ciency resulted in dramatically is a recently recognized and rapidly expanding family of increased thymocyte apoptosis, and depletion of thymo- mammalian cell surface-associated serine proteases with cytes. This study demonstrates that Matriptase/MT-SP1 a unique modular structure (Hooper et al., 2001). The has pleiotropic functions in the development of the family currently includes enterokinase, corin, human epidermis, hair follicles, and cellular immune system. airway trypsin-like protease, hepsin, TMPRSS2, Oncogene (2002) 21, 3765 ± 3779. DOI: 10.1038/sj/ TMPRSS3, and Matriptase/MT-SP1 (see below). onc/1205502 TTSPs have also been reported in non-mammalian vertebrates and in invertebrates (Appel et al., 1993; Keywords: apoptosis; cell-surface proteolysis; skin; Yamada et al., 2000). The common structural features thymus; barrier function of the TTSPs include a short N-terminal cytoplasmic tail, a signal anchor that functions as a transmembrane domain, and a C-terminal serine protease domain. The transmembrane and serine protease domains are separated by a linker region consisting of a variable series of modular protein interaction domains that *Correspondence: TH Bugge, Proteases and Tissue Remodeling include complement factor 1R ± urchin embryonic Unit, Oral and Pharyngeal Cancer Branch, National Institute of growth factor ± bone morphogenetic protein (CUB) Dental and Craniofacial Research, National Institutes of Health, 30 domains, low-density lipoprotein receptor (LDLR) Convent Drive, Room 211, Bethesda, Maryland, MD 20892, USA; repeats, group a scavenger receptor domains, Frizzled E-mail: [email protected] Received 25 February 2002; revised 15 March 2002; accepted 19 domains, and meprin-A-5 protein-mu (MAM) domains March 2002 (Hooper et al., 2001). TTSPs are likely synthesized as Type II transmembrane serine protease in development K List et al 3766 pro-enzymes that require proteolytic cleavage for pro-urokinase plasminogen activator (pro-uPA), hepa- function. Preliminary studies suggest that some TTSPs tocyte growth factor/scatter factor (HGF/SF), and are shed from the cell surface, either prior to activation protease activated receptor-2 (PAR-2) in vitro, and the or, possibly, directly associated with the activation protease could have pleiotropic functions in the process (Takeuchi et al., 2000). The biological functions activation of proteolytic cascades, growth factor of most TTSPs still need to be elucidated. Enteroki- activation, and activation of G-protein coupled recep- nase, the prototypic member of this class of proteases, tors (Lee et al., 2000; Takeuchi et al., 2000). is the physiological activator of trypsinogen in the To identify novel serine proteases involved in digestive tract, and in this capacity is the initiator of a epidermal remodeling, a systematic screening for serine proteolytic cascade reaction that leads to the activation protease genes speci®cally upregulated in keratinocytes of pro-carboxypeptidase, chymotrypsinogen, and pro- during skin wound healing and squamous carcinoma elastase (Mann and Mann, 1994). Mutations in the was initiated. In the study, Matriptase/MT-SP1 was TMPRSS3 gene were recently shown to be associated revealed to be one of several serine proteases expressed with congenital and childhood autosomal recessive in basal keratinocytes in the skin, and the expression of deafness (Scott et al., 2001). Corin appears to be Matriptase/MT-SP1 was strongly upregulated in mi- involved in pro-hormone processing, acting as a pro- grating and proliferating keratinocytes in®ltrating the atrial natriuretic peptide-converting enzyme (Yan et al., provisional wound matrix. We also determined that 2000). Taken together, these ®ndings suggest that Matriptase/MT-SP1 was frequently expressed in cancer TTSPs may have many biologically diverse functions originating from strati®ed epithelium (data not shown). in development and homeostasis. Mice de®cient in Matriptase/MT-SP1 were generated The type II transmembrane serine protease Matrip- in order to determine the function of Matriptase/MT- tase/MT-SP1 (also known as ST14 and TAGD-15) was SP1 in epidermal development and neoplasia. We ®rst described as a transcript that was frequently absent report that Matriptase/MT-SP1 is required for post- in human colon cancer (Zhang et al., 1998). Subse- natal survival, epidermal barrier function, hair follicle quently, Matriptase/MT-SP1 was independently iso- development, and thymocyte survival. lated as a serine protease expressed in a prostate carcinoma cell line, as a serine protease secreted in human breast milk, and as a serine protease present in Results ovarian cancer but not in normal ovarian tissue (Lin et al., 1999; Takeuchi et al., 1999; Tanimoto et al., 2001). Generation of Matriptase/MT-SP1-deficient mice Epithin, the murine orthologue of Matriptase/MT-SP1, was isolated from fetal thymic stromal cells (Kim et al., The strategy used to generate a null mutation in the 1999). Matriptase/MT-SP1 contains a transmembrane murine Matriptase/MT-SP1 gene in embryonic stem signal anchor, separated from the chymotrypsin-like (ES) cell is depicted in Figure 1a. No phage clones were serine protease domain by two CUB domains and four isolated that contained the ®rst 143 bp of the published LDLR repeats (Kim et al., 1999; Takeuchi et al., 1999; cDNA sequence (Kim et al., 1999). The ®rst exon Tanimoto et al., 2001). Matriptase/MT-SP1 has a identi®ed, starting at nucleotide (nt.) 144 in the cDNA widespread expression in normal epithelial tissues, sequence, is therefore in the following designated including the skin, thymic stroma, gastrointestinal tract, `putative exon 2' and the upstream intron sequence is kidney, lung, prostate, and mammary gland (Kim et al., designated `putative intron 1'. In the Matriptase/MT- 1999; Oberst et al., 2001; Takeuchi et al., 1999). SP1 targeting vector, a 350 bp fragment containing the Matriptase/MT-SP1 has been proposed to play an 3' portion of the putative intron 1 and most of the important role in carcinoma progression. The novel putative exon 2 encoding the signal anchor, was protease is expressed in a wide variety of human tumors replaced by an HPRT cassette located in the opposite of epithelial origin, is a prognostic factor in ovarian transcriptional orientation relative to the transcription carcinoma, and is a potential activator of key molecules of the Matriptase/MT-SP1 gene. Homologous recombi- associated with tumor invasion and metastasis (see nation of this targeting vector into the Matriptase/MT- below) (Oberst et al., 2001). Matriptase/MT-SP1 is SP1 locus removes DNA sequences encoding the N- synthesized as a catalytically inactive single-chain terminal 19 of the 24 amino acids (aa) of the signal zymogen. The physiological activator of Matriptase/ anchor required for membrane translocation of Ma- MT-SP1 remains to be identi®ed, but Matriptase/MT- triptase/MT-SP1. This deletion also introduces a frame- SP1 can be activated in vitro by a factor that is present shift mutation in the targeted Matriptase/MT-SP1 locus in human serum (Benaud et al., 2001). Soluble through the splicing of the putative exon 1 to the Matriptase/MT-SP1 is present in the conditioned putative exon 3, leading to a truncated protein contain- medium of cultured cells and in milk, suggesting that ing just the N-terminal 27 aa of Matriptase/MT-SP1. proteolytic shedding of Matriptase/MT-SP1 from the Mice homozygous for this targeted deletion in the cell surface takes place. Soluble Matriptase/MT-SP1 Matriptase/MT-SP1 gene (Matriptase/MT-SP17/7 has been found in both a free form and in association mice) were generated by blastocyst injection of a with the Kunitz-type serine protease inhibitor, hepato- targeted ES cell clone followed by interbreeding of the cyte growth factor activator inhibitor-1 (Lin et al., ospring of chimeric mice transmitting the targeted 1999). Matriptase/MT-SP1 is an ecient activator of allele through the germ line (Figure 1b). As would be

Oncogene Type II transmembrane serine protease in development K List et al 3767 a mutation, we performed RT ± PCR analysis of total RNA from whole newborn Matriptase/MT-SP17/7 and Matriptase/MT-SP1+/+ pups using primer pairs hybri- dizing to putative exons 1 and 4 (Figure 2b,c). RT ± PCR ampli®cation with this primer pair resulted in a PCR product of approximately 280 bp using RNA from Matriptase/MT-SP17/7 pups, again indicative of a putative exon 1 ± putative exon 3 splicing event. Identical results were obtained with an independent, non-over- lapping putative exon 1 ± putative exon 4 primer pair (data not shown). The 280 bp RT ± PCR fragment generated with RNA from Matriptase/MT-SP17/7 pups was cloned and sequenced (Figure 2c). This sequence analysis veri®ed the presence of a putative exon 1 ± putative exon 3 splice event and the subsequent b introduction of a frameshift mutation in the targeted Matriptase/MT-SP1 allele leading to a truncated protein terminating at a TGA codon located at nt. 344 of the Matriptase/MT-SP1 sequence (Kim et al., 1999). In conclusion, the signal anchor is deleted from the targeted Matriptase/MT-SP1 allele, a frameshift introduced at position 305, and the modi®ed allele has the capacity to express a cytoplasmic 40 aa peptide containing the N- terminal 27 of the 855 aa of the mature Matriptase/MT- SP1 protein (Figure 2c). The fate of this N-terminal 27 aa cytoplasmic peptide was not studied.

Postnatal lethality in Matriptase/MT-SP1-deficient mice Matriptase/MT-SP1 was dispensable for the develop- Figure 1 Generation of a targeted deletion in the Matriptase/ MT-SP1 gene. (a) Structure of the Matriptase/MT-SP1 targeting ment to term. Genotype analysis of newborn ospring vector (top), wild-type Matriptase/MT-SP1 allele (middle), and of Matriptase/MT-SP1+/7 breeding pairs revealed that targeted Matriptase/MT-SP1 allele (bottom). Exons are indicated the targeted Matriptase/MT-SP1 allele was present in as black boxes and intron sequences as solid lines. Putative exon the expected Mendelian frequency (Figure 3a). Of 119 2, 3 and 4 sequences, determined from the cloned Matriptase/MT- +/7 SP1 gene, were identical to the published sequence of the mouse pups from 12 litters from Matriptase/MT-SP1 Matriptase/MT-SP1 cDNA (Kim et al., 1999). The size of the parents that could be followed from the time of birth, putative ®rst intron was not determined (denoted by / /). A PGK- 25 were Matriptase/MT-SP1+/+, 62 Matriptase/MT- HPRT mini-gene cassette was introduced in the opposite SP1+/7, and 32 Matriptase/MT-SP17/7. However, no orientation to the Matriptase/MT-SP1 gene. The cassette replaced Matriptase/MT-SP17/7 ospring were detected at a 350 bp fragment of the Matriptase/MT-SP1 gene containing part of the putative intron 1 and most of the putative exon 2 that weaning, indicating that Matriptase/MT-SP1 is required encodes the signal anchor. The locations of the primers used for for postnatal survival (Figure 3b). Of 57 weaning-age PCR analysis of ES cell clones are indicated by arrows (see ospring from eight litters, 41 were Matriptase/MT- Materials and methods). The location of the p02 primer was SP1+/7 and 16 Matriptase/MT-SP1+/+. The mortality outside the sequences used in the targeting vector. The probe used for Southern blot analyses of PCR-positive ES cell clones and tail of newborn Matriptase/MT-SP1 pups was investigated +/7 biopsy DNA was located upstream of the Matriptase/MT-SP1 in detail in eight litters from Matriptase/MT-SP1 sequences used in the targeting vector (indicated by a solid line parents that were monitored closely from the time of below the targeted locus). (b) Southern blot of XhoI-digested birth. Of the 20 pups in this study cohort that died DNA from targeted (lane 1) and control (lane 2) ES cell clones. within 48 h, 16 were Matriptase/MT-SP17/7, 4 were Arrows indicate the position of the wild type (6.6 kb) and +/7 targeted 6.3 kb) alleles Matriptase/MT-SP1 , and no pups were Matriptase/ MT-SP1+/+ (P50.001, Chi-square analysis, two-tailed). No Matriptase/MT-SP17/7 pups in this cohort were alive more than 48 h after birth. Taken together, these predicted from the targeting strategy, Northern blot data demonstrate that Matriptase/MT-SP1 is required hybridization of RNA from whole newborn Matriptase/ for survival beyond the ®rst two days after birth. MT-SP17/7 pups demonstrated the presence of a single transcript of similar or slightly increased mobility Abnormal epidermal development in Matriptase/MT- compared to RNA from Matriptase/MT-SP1+/+ litter- SP1-deficient mice mates. This suggests that a single putative exon 1 ± putative exon 3 splicing event was taking place in Newborn Matriptase/MT-SP17/7 pups displayed an transcripts from the targeted Matriptase/MT-SP1 allele abnormally dry, wrinkled, red, and shiny skin. All body (Figure 2a). To further verify the introduction of a null surfaces were aected including the trunk, limbs, tail,

Oncogene Type II transmembrane serine protease in development K List et al 3768

Figure 2 The targeted deletion in the Matriptase/MT-SP1 gene generates a null mutation. (a) Northern blot hybridization of total RNA from whole newborn Matriptase/MT-SP1+/+ pups (lane 1) and Matriptase/MT-SP17/7 pups (lane 2). The positions of 28S and 18S, and 4-5S ribosomal RNAs are indicated with arrows. The blot was hybridized with a 32P-labeled EST containing the entire murine Matriptase/MT-SP1 coding region. Comparable amounts of RNA were loaded in each lane. (b) Ethidium bromide-stained agarose gel of DNA fragments generated by RT ± PCR of total RNA from Matriptase/MT-SP1+/+ pups (lane 1) and Matriptase/ MT-SP17/7 pups (lane 2). The RT ± PCR analysis was performed with primers complementary to sequences in the putative exon 1 (p07 (c, lower panel), nt. 59 ± 80 in the Matriptase/MT-SP1 cDNA sequence (Kim et al., 1999)) and putative exon 4 (p06 (c, lower panel), nt. 477 ± 498). The presence of a PCR product with an approximate size of 280 bp in lane 2 demonstrates the putative exon 1 ± putative exon 3 splicing event within the targeted allele. Lane 3; 100 bp DNA ladder. (c) RT ± PCR products generated in b were subcloned into plasmid vectors and sequenced. Upper panel, sequence of the putative exon 1 ± exon 3 border. The putative exon 2 (boxed area in bottom panel) is deleted in transcripts from the targeted Matriptase/MT-SP1 allele to generate a frameshift mutation and translational termination at a TGA codon located at nt. 344 (shaded box)

and facial skin (Figure 4a and data not shown). This birth (see below), this dierence in size could in part phenotype became more pronounced within hours after also be attributed to selective maternal neglect of birth as a consequence of the rapid dehydration of Matriptase/MT-SP17/7 pups or to the inability of some Matriptase/MT-SP17/7 pups (see below). The body Matriptase/MT-SP17/7 pups to nurse properly. Thus, weight of 3 to 12 h old Matriptase/MT-SP17/7 pups was milk spots in the stomachs were observed in just 6 out of approximately 20% lower than Matriptase/MT-SP1+/+ 14 Matriptase/MT-SP17/7 pups (43%) versus 27 out of and Matriptase/MT-SP1+/7 (MT-SP+) littermates 29 Matriptase/MT-SP1+ pups (93%) that could be (P50.0003) and the crown-rump length was 12% followed closely for at least a 24 h period. shorter (P50.00006) (Figure 4b and c). In addition to A comprehensive histological analysis was performed progressive dehydration beginning immediately after on Matriptase/MT-SP17/7 pups and their associated

Oncogene Type II transmembrane serine protease in development K List et al 3769

Figure 3 Postnatal lethality in Matriptase/MT-SP1 de®cient embryos. Genotype analysis of ospring from Matriptase/MT-SP1+/7 parents. (a) Distribution of genotypes in newborn pups. One hundred and nineteen ospring from a total of 12 litters that could be followed from the time of birth were genotyped. (b) Distribution of genotypes in weaning age ospring. Fifty-seven ospring from eight litters were genotyped 21 ± 28 days after birth. **P50.001, Chi-square analysis, two-tailed

Figure 4 Skin abnormalities and growth retardation in Matriptase/MT-SP17/7 mice. (a) Representative examples of the appearance of trunk skin from newborn Matriptase/MT-SP1+/+ pups (top) and Matriptase/MT-SP17/7 pups (bottom). Note the abnormally red, dry, and shiny appearance of the skin from the Matriptase/MT-SP17/7 pup. Size bar, 1 mm (b) Weight and (c) crown-rump length of Matriptase/MT-SP1+/+ (n=4), Matriptase/MT-SP1+/7 (n=13), and Matriptase/MT-SP17/7 (n=11) pups from three newborn litters of Matriptase/MT-SP1+/7 parents. *P50.0003 and **P50.00006, Matriptase/MT-SP17/7 versus Matriptase/MT-SP1+/7 and Matriptase/MT-SP1+/+ pups, respectively (Student's t-test, two-tailed)

Oncogene Type II transmembrane serine protease in development K List et al 3770 Matriptase/MT-SP1+ littermates to determine the ®xation, the intercorneal lipids are preserved and the cause of postnatal lethality and the abnormal skin stratum corneum appears as the outermost highly appearance. No overt abnormalities were detected in opaque layer of the epidermis with strati®ed layers of the lungs, urogenital tract, central nervous system, ¯attened corneocytes after toluidine blue staining vasculature, pituitary, pancreas, liver, adrenals, and (Figure 5c). The basal, spinous, and granular layers stomach of Matriptase/MT-SP17/7 pups. Also, no of the epidermis of Matriptase/MT-SP17/7 pups were gross developmental abnormalities were detected by X- not conspicuously morphologically dierent from ray analysis of newborn Matriptase/MT-SP17/7 pups Matriptase/MT-SP1+ littermates when examined by (data not shown). A subset of Matriptase/MT-SP17/7 light microscopy (Figure 5b,d). However, the stratum pups demonstrated a distinct subnuclear vacuolization corneum presented striking abnormalities in all Ma- in the resorptive epithelium or edema of the underlying triptase/MT-SP17/7 pups. Very few intercorneocyte stroma in the small intestine and colon. This lacunae could be observed, giving rise to a very abnormality was probably secondary to starvation compact appearance of the stratum corneum after and dehydration, as Matriptase/MT-SP17/7 pups with hematoxylin and eosin (h and e) staining (compare abundant milk in the stomach presented an unremark- Figure 5a,b). At higher magni®cation, following able resorptive epithelium (data not shown). The glutaraldehyde ®xation and toluidine blue staining of epidermis, however, of Matriptase/MT-SP17/7 pups thin sections, the corneocytes appeared clearly dys- demonstrated striking abnormalities as expected from morphic and pleomorphic, with a swollen, rather than the macroscopic appearance of the skin (Figure 5). In ¯attened, appearance of individual corneocytes. More- normal newborn mice, the ¯attened corneocytes of the over, the stratum corneum of all epidermal surfaces stratum corneum form a characteristic `basket weave' displayed a less organized strati®cation (compare pattern generated by a meshwork of interlocking layers Figure 5c,d). Taken together, these data clearly of corneocytes connected by desmosomes (Nemes and demonstrate an essential function of Matriptase/MT- Steinert, 1999). The intercorneocyte lacunae of the SP1 in stratum corneum development. basket weave meshwork are abundant in intercorneal The stratum corneum is formed from an inner layer lipids that are extracted during routine tissue proces- of living keratinocytes that migrate upwards while sing, giving rise to the appearance of empty inter- undergoing a highly complex genetic program of corneocyte lacunae (Figure 5a). After glutaraldehyde terminal dierentiation. Four distinct key events

Figure 5 Aberrant stratum corneum in Matriptase/MT-SP17/7 mice. Representative sections of newborn dorsal skin from Matriptase/MT-SP1+/+ (a and c) and Matriptase/MT-SP17/7 (b and d) littermates. H and e staining at low magni®cation (a and b) showing abnormal stratum corneum formation in Matriptase/MT-SP17/7 pups. e, epidermis. d, dermis. p, panniculus muscle. Arrowheads indicate example of hair follicles. Size bar, 25 mm. Toluidine blue staining (c and d) of thin sections of the epidermis at higher magni®cation showing dysmorphic, pleomorphic, and dysorganized corneocytes in Matriptase/MT-SP17/7 epidermis (examples indicated with arrows). The positions of the basal (B), spinal (S), granular (G), and corneal layer (Sc) of the epidermis are indicated. Stars, hair follicles. Size bar, 5 mm

Oncogene Type II transmembrane serine protease in development K List et al 3771 characterize stratum corneum development: (1) forma- Although the corneocytes of the Matriptase/MT- tion of desmosomal adherence junctions between SP17/7 epidermis were strikingly dysmorphic when keratinocytes, (2) intracellular accumulation of specia- visualized by light microscopy, the corni®ed envelope lized keratins, (3) formation of a peripheral corni®ed itself presented no obvious morphological changes envelope, and (4) extrusion of specialized intercorneo- when examined by transmission electron microscopy cyte lipids (Nemes and Steinert, 1999; Roop, 1995); see (Figure 6). However, striking dierences in transitional Discussion for further details). Desmosomes, appear- cell vesicular body content were immediately evident. ing as `spines' decorating the surface of basal and In control mice, numerous intracytoplasmic vesicular spinous layer cells in light microscopy (Figure 5d) and bodies were visible in Matriptase/MT-SP1+ transi- as ¯attened electron-dense disks in transmission tional cells. These vesicular bodies, which gives rise to electron microscopy (Figure 6b), were clearly visible the majority of the intercorneocyte lipids of the normal in the Matriptase/MT-SP17/7 epidermis. Also, the epidermis, were located just below the plasma abnormal stratum corneum formation in Matriptase/ membrane and also accumulated in the extracellular MT-SP17/7 pups was not directly associated with space between transitional layer cells and corneocytes major changes in the expression of several keratino- (Figure 6a). In Matriptase/MT-SP17/7 pups, these cyte-speci®c dierentiation markers. The onset and vesicular structures were entirely absent in transitional level of expression of basal-layer markers (keratin 5 cells of the epidermis and in the extracellular space and 14), suprabasal layer markers (keratin 1 and 10), (Figure 6b). The absence of vesicular bodies was not and granular/corneum-layer markers (®llagrin and limited to the epidermis, but was also observed in loricrin) were not appreciably altered (data not transitional cells within the hair follicles of Matriptase/ shown). Keratohyalin granules formed normally in MT-SP17/7 mice, suggesting a generalized defect in the granular and transitional layer of the Matriptase/ transitional cell function (data not shown). MT-SP17/7 epidermis (Figure 6b). Apoptosis was very infrequent in the basal layer of the epidermis of both Loss of epidermal barrier function in Matriptase/ Matriptase/MT-SP17/7 and littermate control pups, MT-SP1-deficient skin causes neonatal death and no signi®cant dierence was observed in cell proliferation as assessed by proliferating cell nuclear The epidermal water barrier that prevents catastrophic antigen-staining (data not shown). ¯uid loss from the surface of terrestrial vertebrates

Figure 6 Transmission electron microscopy of the transitional cell-®rst corneocyte interface of Matriptase/MT-SP1+/+ (a) and Matriptase/MT-SP17/7 (b) pups. Transitional cells with keratohyalin bodies (k) are present in the bottom of the panels followed upwards by layers of corneocytes (stars). Examples of desmosomes are indicated with arrows. Vesicular bodies (some examples indicated by arrowheads in (a)) are highly abundant at the cytoplasmic face of transitional cells and in the extracellular space between the transitional cell and ®rst layer of corneocytes. The Matriptase/MT-SP17/7 epidermis lacks vesicular bodies. Size bar, 0.5 mm

Oncogene Type II transmembrane serine protease in development K List et al 3772 resides in the stratum corneum. Barrier function is the loss of ¯uids in Matriptase/MT-SP1+ animals was conferred by a combination of the corni®ed envelope minimal, with an average reduction in body weight of and intercorneocyte lipids (Nemes and Steinert, 1999). 0.39+0.05% per hour. Remarkably, the ¯uid loss in Due to the conspicuous stratum corneum defects, the Matriptase/MT-SP17/7 pups was accelerated by more functional integrity of the epidermis of Matriptase/MT- than ®vefold, demonstrating a dramatic loss of SP17/7 pups was investigated using a standard dye ìnwards out' epithelial barrier function (2.09+0.55% permeability assay (Koch et al., 2000). The ability to per hour). The diminished barrier function of Ma- exclude histological dyes from penetrating through the triptase/MT-SP17/7 skin was not a consequence of a epidermis requires an intact epidermal barrier and dye general developmental retardation or reduced body penetration is indicative of abrogated barrier function. weight, as Matriptase/MT-SP1+ pups of similar or Newborn litters from Matriptase/MT-SP1+/7 inter- even smaller size than their Matriptase/MT-SP1- crosses were stained by submersion in toluidine blue. de®cient littermates in all cases demonstrated minimal As expected, Matriptase/MT-SP1+ pups showed mini- ¯uid loss (data not shown). In conclusion, Matriptase/ mal epidermal dye penetration, showing that the MT-SP1 is required for the development of epidermal neonatal epidermis is highly impermeable and func- barrier function. Loss of Matriptase/MT-SP1 leads to tional (Figure 7a). In contrast, Matriptase/MT-SP17/7 uncontrolled ¯uid loss and rapid postnatal lethality. littermates showed extensive dye penetration across the entire epidermal surface demonstrating a generalized Hair follicle hypoplasia and dysgenesis in Matriptase/ loss of òutwards in' epidermal barrier function. To MT-SP1-deficient mice ascertain whether the abnormal epidermal maturation in Matriptase/MT-SP17/7 pups also impaired ìnwards Vibrissal hairs, which are normally fully developed at out' epidermal barrier function, the rate of ¯uid loss birth (Kaufmann and Bart, 1999), were absent in all through evaporation was determined in Matriptase/ Matriptase/MT-SP17/7 pups, immediately revealing a MT-SP17/7 and Matriptase/MT-SP1+ pups. Newborn role of Matriptase/MT-SP1 in hair follicle develop- litters of Matriptase/MT-SP1+/7 parents were sepa- ment. Examination of the facial area of Matriptase/ rated from their mothers to prevent ¯uid intake, and MT-SP17/7 pups by scanning electron microscopy the loss of ¯uids at 378C was recorded over a ®ve-hour showed a general absence of both erupted vibrissal hair period in individual pups by monitoring the loss of shafts and vibrissal hair canals (Figure 8a,b). Interest- body weight. The pups were genotyped at the ingly, the snouts of Matriptase/MT-SP17/7 pups termination of the experiment (Figure 7b). Overall, instead displayed a large number of ellipsoid epidermal

Figure 7 Disruption of epidermal barrier function in Matriptase/MT-SP17/7 mice. (a) Loss of `outwards in' barrier function. Newborn Matriptase/MT-SP1+ (top) and Matriptase/MT-SP17/7 (bottom) pups were submerged in toluidine blue, brie¯y destained with PBS and photographed. Representative sections of trunk skin are presented. Matriptase/MT-SP17/7, but not Matriptase/MT-SP1+, pups display extensive dye penetration of all epidermal surfaces. Size bar, 1 mm (b) Loss of `inward out' barrier function. Newborn pups from a litter of Matriptase/MT-SP1+/7 parents were separated from their mother to prevent ¯uid intake, and the rate of epithelial ¯uid loss was estimated by measuring the reduction of body weight as a function of time. The data are expressed as the average body weight as a percent of initial body weight in Matriptase/MT-SP1+ (squares; n=3) and Matriptase/MT-SP17/7 (diamonds; n=3) pups. Error bars indicate standard deviations. One Matriptase/MT-SP17/7 pup urinated after 210 min and was excluded from further measurements. *P50.028 to 0.022. **P50.014 to 0.005 (Student's t-test, two-tailed)

Oncogene Type II transmembrane serine protease in development K List et al 3773 protrusions that were uniformly absent in Matriptase/ CD4+CD8+ double positive thymocytes, which MT-SP1+ littermate controls (Figure 8c,d). Although represent about 90% of T-lymphocytes in the thymus, erupted vibrissae were not present, a histological revealed a ®vefold increase in the apoptotic index analysis revealed that vibrissal follicles were never- (P50.016), and a corresponding almost 50% reduction theless abundant in Matriptase/MT-SP17/7 pups in the total number of double positive cells residing in (Figure 8h). The vibrissal follicles were distinctly the thymus (P50.003) (Figure 10c,d). The accelerated hypomorphic (compare Figure 8e,f, and Figure 8g,h), thymocyte apoptosis in Matriptase/MT-SP17/7 pups but did contain vascular sinuses, inner and outer root appeared to be intrinsic to the thymic environment and sheaths, follicular papillae, hair bulbs, and hair shafts not a primary defect in lymphocytes. Thus, other (Figure 8f,h). In accordance with surface scanning lymphoid tissues examined in Matriptase/MT-SP17/7 data, no vibrissal hair canals were detected in a pups, including the spleen and gut-associated lymphoid comprehensive analysis of serial frontal sections of tissue, were similar to Matriptase/MT-SP1+ pups the facial area of Matriptase/MT-SP17/7 pups and the presenting no appreciative lymphocyte apoptosis. In hair shaft of all isolated follicles was uniformly curved conclusion, Matriptase/MT-SP1 is critical for the and ingrown (Figure 8f,h). Likely as a directly survival of developing lymphocytes within the thymic consequence of this, the nasal area displayed an microenvironment. unusually high number of aborted vibrissal follicle remnants presenting as large follicles in various stages of degeneration embedded in reactive hyperplastic Discussion epidermis. The speci®c requirement of Matriptase/ MT-SP1 for the formation of pelage hair could not This paper shows that the type II transmembrane be determined directly, as all Matriptase/MT-SP17/7 serine protease Matriptase/MT-SP1 is required for pups died before the eruption of body hairs in the postnatal survival and has an essential function in littermates. However, a comprehensive histological epidermal, hair follicle and thymic development. The analysis of skin demonstrated that pelage hair follicles epidermis is a self-renewing, strati®ed epithelium that of Matriptase/MT-SP17/7 pups displayed marked serves as a ®rst line of defense against the external hypoplasia when compared to hair follicles of environment by providing a protective barrier against Matriptase/MT-SP1+ littermates (Figure 9). Speci®- mechanical, chemical, and biological insults. The cally, Matriptase/MT-SP17/7 skin contained fewer epidermis also provides a water-impermeable barrier pelage hair follicles that generally penetrated less that prevents excessive loss of body ¯uids, a function deeply into the underlying dermis and overlying that is critical for the survival of all terrestrial epidermis, and were more poorly dierentiated (Figure vertebrates immediately after birth. The epidermal 9a,b). Taken together, the data show that Matriptase/ barrier function resides in the stratum corneum, which MT-SP1-de®ciency is associated with follicular hypo- chie¯y consists of an interlocking meshwork of plasia and dysgenesis. ¯attened corneocytes connected by desmosomes and intercorneocyte lipids. Water impermeability is conferred to the stratum corneum by an insoluble corni®ed Accelerated thymocyte apoptosis in Matriptase/MT-SP1- envelope located just beneath the plasma membrane of deficient mice the corneocytes, and by the surrounding intercorneo- Matriptase/MT-SP1 is abundant in the developing cyte lipids (Nemes and Steinert, 1999; Roop, 1995). thymic epithelium (Kim et al., 1999). Moreover, The stratum corneum is maintained by the constant mutations aecting the epidermis often also aect reproduction of the inner layer of living keratinocytes thymic function due to the close ontogenic relationship that migrate upwards while undergoing a complex between the two organs and the expression of a genetic program of terminal dierentiation. This number of epidermal structural genes in the thymic program includes the formation of desmosomal epithelium (Davisson et al., 1994; Flanagan, 1966; adherence junctions, the accumulation of specialized Manley, 2000; Nakagawa et al., 1998; Reske-Kunz et keratins that provide mechanical strength to the skin, al., 1979; Shultz, 1988). Indeed, the thymus is rich in the accumulation of a complex array of structural keratinized epithelium in the form of terminally proteins, lipids, and enzymes that constitute the dierentiated swirls of corni®ed epithelial cells, termed corni®ed envelope, and the synthesis, extrusion, and Hassall's bodies (Manley, 2000). Interestingly, the processing of intercorneocyte lipids (Nemes and thymus of all examined Matriptase/MT-SP17/7 pups Steinert, 1999; Roop, 1995). The data presented here was grossly abnormal as compared to Matriptase/MT- show that Matriptase/MT-SP1 is required for the SP1+ littermates (Figure 10). Thymocyte apoptosis was formation of a functional stratum corneum and the widespread throughout both the cortex and the establishment of epidermal barrier function. This medulla of the thymus of Matriptase/MT-SP17/7 pups de®ciency does not appear to be the consequence of when analysed histologically by in situ labeling of a general developmental delay, as other organ systems fragmented DNA by Tunel staining (Figure 10b). In such as the musculoskeletal system, the central nervous contrast, the thymus of Matriptase/MT-SP1+ pups system, the gastrointestinal tract, the respiratory tract, demonstrated few apoptotic cells (Figure 10a). Quanti- and the urogenital tract did not demonstrate any signs tative ¯ow cytometry analysis of immature of immaturity in newborn Matriptase/MT-SP17/7

Oncogene Type II transmembrane serine protease in development K List et al 3774

Figure 8 Vibrissal follicular hypoplasia and dysgenesis in Matriptase/MT-SP17/7 mice. (a ± d; scanning electron microscopy images of the facial area of Matriptase/MT-SP1+/+ (a and c) and Matriptase/MT-SP17/7 (b and d) pups. a and b; lateral view at low magni®cation of parts of the upper (U) and lower (L) lip showing the absence of erupted vibrissae and hair canals in Matriptase/MT-SP17/7 pups. Size bar, 200 mm. (c) high magni®cation of vibrissal hair shaft and hair canal in a Matriptase/MT- SP1+/+ pup and (d) corresponding epidermal protrusion in a Matriptase/MT-SP17/7 pup. Size bar, 20 mm. (e and f) representative examples of isolated vibrissal follicles from newborn Matriptase/MT-SP1+/+ (e) and Matriptase/MT-SP17/7 (f) pups. Matriptase/ MT-SP17/7 hair shafts are hypomorphic and ingrown. Size bar, 100 mm. (g ± i) Representative longitudinal sections of vibrissal follicles (stars) from newborn Matriptase/MT-SP1+/+ (g) and Matriptase/MT-SP17/7 (h and i) pups after h and e staining. (h) Follicular hypoplasia and hair canal agenesis. (i) Aborted vibrissal follicle remnant (star) with reactive epidermal hyperplasia (arrows). Size bar, 50 mm

pups. Pathological abnormalities in the stratum barrier function, is seen in an extraordinary variety of corneum, with the subsequent breakdown of epidermal skin conditions of diverse etiology that are collectively

Oncogene Type II transmembrane serine protease in development K List et al 3775 disrupt epidermal barrier function (De Laurenzi et al., 1996; Elias et al., 2001; Huber et al., 1995; Koch et al., 2000; Kubilus et al., 1979; Matsuki et al., 1998; Russell et al., 1995; Shapiro et al., 1978). The data presented in this paper strongly suggest that the abrogation of barrier function is related to abnormal extrusion of vesicular bodies in the transitional layer of the cornifying epidermis, which likely aects the formation of the specialized set of crosslinked intercorneocyte lipids that confer epidermal barrier function in conjunction with the corni®ed envelope. Although the short lifespan of Matriptase/MT-SP17/7 mice did not permit a comprehensive initial description of the role of this protease in hair biology of adult mice, it is clear from the data presented in this paper that Matriptase/MT-SP1 plays a critical role in hair follicle development. Matriptase/MT-SP17/7 mice presented with marked, generalized follicular hypoplasia and agenesis of the vibrissal hair canal. Despite their hypoplastic appearance, many follicular structures could, nevertheless, be identi®ed in the vibrissal follicles of Matriptase/MT-SP17/7 mice, including the inner and outer root sheaths, follicular papillae, hair bulb, and hair shaft. An exception was the hair canal, which was generally absent. As a consequence of this, the growing vibrissal hair shaft was unable to erupt, instead growing backwards into the vibrissal follicle, likely causing wholesale follicle abortion. Hair canal formation is a poorly understood process, but is believed to occur prior to the emerging hair shaft pushing through the follicle and may include apoptosis of follicular epithelial cells (Davisson and Hardy, 1952; Figure 9 Pelage hair follicle hypoplasia in Matriptase/MT-SP17/7 Magerl et al., 2001). Although a speculative assertion, mice. Parasagittal sections of newborn mid-dorsal skin from it is tempting to propose that a common molecular Matriptase/MT-SP1+/+ (a) and Matriptase/MT-SP17/7 (b) litter- defect underlies the epidermal and hair follicle mates after toluidine blue staining. Pelage hair follicles (arrow- abnormalities in Matriptase/MT-SP17/7 mice. 7/7 heads) are hypoplastic in Matriptase/MT-SP1 skin, penetrating The thymus ful®lls a critical function in T- less deeply into the dermis and epidermis. e, epidermis. d, dermis. Size bar, 50 mm lymphocyte development through the selection of mature T-lymphocytes from a repertoire of immature T-cell precursors. Only a small proportion of T-cell precursors survives thymic selection and is exported to referred to as ichthyosis. At the genetic level, the loss secondary lymphoid tissues (Sebzda et al., 1999). The of barrier function can be caused by an array of defects thymic microenvironment provides negative selection in the complex machinery required for the proper against thymocytes that express a T-cell receptor that assembly of keratin ®laments in the corneocyte, the binds with high anity to self-peptide-MHC com- assembly of corni®ed envelope components, the plexes (negative thymocyte selection) or are incapable formation of corneocyte adherens junctions, or the of interacting with self-peptide-MHC complexes (thy- assembly of intercorneocyte lipids, as observed in mocyte death by neglect). Both groups of thymocytes numerous inherited skin diseases or in transgenic mice are eliminated through apoptosis, but the exact nature with dominant or recessive mutations in epidermal of the stimuli that trigger apoptosis in the thymocytes components (Nemes and Steinert, 1999; Presland and is not fully understood (Williams and Brady, 2001). Dale, 2000; Roop, 1995). Null, frameshift, or point The data presented in this paper show that Matrip- mutations in genes encoding structural proteins of the tase/MT-SP1 has a pivotal role in thymocyte survival corneocyte (e.g., keratins, loricrin), in components of by suppressing apoptosis. The thymus of Matriptase/ the enzymatic machinery that process and cross-link MT-SP17/7 pups was profoundly depleted of im- these structural proteins (e.g., transglutaminase 1), in mature CD4+CD8+ double positive thymocytes. desmosomal cadherins linking corneocytes together This ®nding opens several questions to be addressed (e.g., desmoglein), or in the enzymes that manufacture in future studies as to the function of Matriptase/MT- and process corni®ed envelope and intercorneocyte SP1 in T-cell immunity. Does Matriptase/MT-SP1 lipids (e.g., fatty aldehyde dehydrogenase, arylsulfatase promote thymocyte survival by activating or liberating C/cholesterol sulfatase) have all been reported to cytokines that favor thymocyte survival, or by

Oncogene Type II transmembrane serine protease in development K List et al 3776

Figure 10 Thymocyte depletion in Matriptase/MT-SP17/7 mice. Tunel-stained sections of the thymus of newborn Matriptase/MT- SP1+/+ (a) and Matriptase/MT-SP17/7 (b) mice. Multiple apoptotic cells are seen in the Matriptase/MT-SP17/7 thymus (examples indicated by arrowheads in b). Size bar, 50 mm. (c) Percent apoptotic CD4+CD8+ double positive lymphocytes in the thymus of Matriptase/MT-SP17/7 pups (white bar, n=5) and Matriptase/MT-SP1+ littermates (black bar, n=4) determined by ¯ow cytometry of total thymocyte populations triple-stained with CD4 antibodies, CD8 antibodies, and 7-amino-actinomycin D as a marker for apoptosis. *P50.016, Wilcoxon rank sum test, two-tailed. (d) Total numbers of CD4+CD8+ double positive lymphocytes in Matriptase/MT-SP17/7 pups (white bar, n=5) and Matriptase/MT-SP1+ littermates (black bar, n=5) enumerated by ¯ow cytometry. *P50.0079, Wilcoxon rank sum test, two-tailed

eliminating pro-apoptotic signals within the thymic of Matriptase/MT-SP1 de®ciency described here is not microenvironment? Does Matriptase/MT-SP1 suppress obvious. uPA-de®cient mice have been reported to apoptosis during negative thymocyte selection or display reduced epidermal proliferation in the neonatal suppress the process of thymocyte death by neglect? period (Jensen and Lavker, 1999). However, uPA- Most importantly, what is the overall contribution of de®cient mice display no obvious histological abnorm- Matriptase/MT-SP1 to the establishment of a func- alities in the skin and have normal epidermal function tional cellular immune system? in the absence of other challenging factors (Carmeliet The future identi®cation of the speci®c molecular et al., 1994). Complete de®ciency in PAR-2 has also defects underlying the pleiotropic eects of loss of not been reported to be associated with skin Matriptase/MT-SP1 may represent a challenging abnormalities (Damiano et al., 1999; Lindner et al., undertaking. These defects could include the lack of 2000b). Thus, inappropriate processing of pro-uPA and appropriate growth factor processing, growth factor PAR-2 can essentially be ruled out as a contributing inactivation, shedding of growth factor receptors, or factor to the epidermal and thymic phenotype of the direct modi®cation of the extracellular matrix to Matriptase/MT-SP17/7 mice described here. HGF/SF create the appropriate extracellular environment for is a pleiotropic regulator of cell growth and dierentia- keratinocyte dierentiation and thymocyte survival. tion with possible functions in keratinocyte biology. The essential activities of Matriptase/MT-SP1 are likely Keratinocytes express c-met, the receptor for HGF/SF, to take place directly on the surface of keratinocytes and display increased mobility in response to HGF/SF and thymic epithelium due to the high levels of in culture (McCawley et al., 1998). Transgenic over- expression of the protease on these cells in particular, expression of HGF/SF, intradermal injection of HGF/ and the strict expression of Matriptase/MT-SP1 on SF, or administration of HGF/SF to organ cultures all epithelial, and not mesenchymal cells, in general. stimulate hair follicle formation, accelerate hair follicle Several possible substrates for Matriptase/MT-SP1 dierentiation, promote hair growth, and prolong the have been proposed from biochemical studies, includ- anagen phase of the hair cycle (Jindo et al., 1998; ing the cleavage and activation of pro-uPA, HGF/SF, Lindner et al., 2000a). HGF/SF7/7 mice die early in and PAR-2 (Lee et al., 2000; Takeuchi et al., 2000). embryonic development due to inappropriate epithe- The speci®c relationship between the lack of activation lial-mesenchymal interactions (Stella and Comoglio, of these factors by Matriptase/MT-SP1 and the eects 1999) and the function of HGF/SF in epidermal

Oncogene Type II transmembrane serine protease in development K List et al 3777 development cannot be determined directly. HGF/SF is females. Chimeric male ospring was bred to NIH Black secreted by follicular papilla cells and acts as a Swiss females (Taconic Farms, Germantown, NY, USA) to paracrine factor on neighboring follicular epithelial generate heterozygous ospring. These mice were subsequently interbred to generate homozygous Matriptase/MT- cells to promote follicle growth. However, both 7/7 follicular papilla cells and outer root sheath cells SP1 progeny. Genotyping of mice was performed by PCR ampli®cation of DNA from ear or tail biopsies with the express HGF activator, a highly ecient activator of primers p03 (5'-GCATGCTCCAGACTGCCTTG-3') and single chain HGF/SF (Lee et al., 2001; Yamazaki et p04 (5'-GTGGAGGTGGAGTTCTCATACG-3') to detect al., 1999). Moreover, no studies have indicated an the presence of the targeted Matriptase/MT-SP1 allele, and eect of HGF/SF on epidermal dierentiation and p05 (5'-CAGTGCTGTTCAGCTTCCTCTT-3') and p04 (5'- stratum corneum formation. It is, thus, altogether GTGGAGGTGGAGTTCTCATACG-3') to detect the wild conceivable that Matriptase/MT-SP1 functions in type Matriptase/MT-SP1 allele). epidermal development by a novel mechanism that is independent of uPA, PAR-2, or HGF/SF activation. RNA preparation and Northern blot analysis In conclusion, this study demonstrates that Matrip- tase/MT-SP1 is required for postnatal survival and has Whole newborn pups were euthanized, snap-frozen in liquid nitrogen, and ground to a ®ne powder with a mortar and pleiotropic functions in epidermal dierentiation, hair pestle. Total RNA was prepared by extraction in Trizol follicle development, and thymic homeostasis. reagent (Gibco ± BRL), as recommended by the manufac- turers. Samples (1 mg) were fractionated electrophoretically on formaldehyde agarose gels, blotted onto NytranSuper Charge nylon membranes (Schleicher and Schuell, Keene, Materials and methods NH, USA), hybridized to a 32P-labeled 3.5 kb Matriptase/ MT-SP1 expressed sequence tag (EST) probe (I.M.A.G.E. ID Construction of targeting vector and generation of transgenic 2609399) that contains the complete full-length murine mice Matriptase/MT-SP1 cDNA, and subjected to PhosphorImage A mouse genomic DNA fragment was isolated from a 129/ analysis using ImageQuant software from Molecular Dy- SvJ bacteriophage library using a 169 bp 32P-labelled namics (Molecular Dynamics, Sunnyvale, CA, USA). synthetic oligonucleotide probe corresponding to nt. 83 to 252 of the published Matriptase/MT-SP1 cDNA sequence RT ± PCR (Kim et al., 1999). The targeting vector was generated from a 7.5 kb EcoRI-NotI fragment containing the putative intron 1 Total RNA from newborn pups was ampli®ed by reverse through the putative exon 4, that was subcloned and transcription followed by PCR ampli®cation using Ready-to- characterized extensively by restriction mapping, Southern goTM RT ± PCR beads (Amersham Pharmacia Biotech Inc, blotting, and sequencing of the intron/exon boundaries. The Piscataway, NJ, USA), as recommended by the manufac- targeting vector was constructed by placing a PGK-HPRT turers. First strand cDNA synthesis was performed using a cassette into the AscI site of the pKO ScramblerV924 gene Matriptase/MT-SP1-speci®c primer (p06, 5'-AGGCAGTTA- targeting vector (Stratagene, La Jolla, CA, USA). A 4.5 kb CAGCCGACTTCTT-3') complementary to the putative exon BamHI-SmaI fragment located in intron 1 was thereafter 4 sequences (nt. 477 ± 498 of the murine Matriptase/MT-SP1 inserted between the BglII and XhoI sites after inserting a cDNA sequence, (Kim et al., 1999). The subsequent PCR SalI linker into the SmaI site. A PCR-generated 1 kb BamHI- ampli®cation (annealing temperature 558C, denaturation HindIII fragment that includes the 3' part of the putative temperature 928C, extension temperature 728C, 40 cycles) exon 2 through the putative exon 4 was next inserted between was performed with the ®rst strand primer in combination the BamHI and HindIII sites of the vector. The Matriptase/ with a putative exon 1-speci®c primer (p07, 5'-AAC- MT-SP1 targeting vector was further furnished with a CATGGGTAGCAATCGGGGC-3') corresponding to nt. ¯anking Herpes simplex virus-thymidine kinase expression 59 ± 80 of Matriptase/MT-SP1 (Kim et al., 1999). The RT ± cassette inserted into the RsrII site to provide a means of PCR products were excised from agarose gels following selection against random insertion of the targeting vector. electrophoresis, puri®ed, and cloned using the TopoTA The targeting vector was introduced into HM-1 ES cells cloning vector system (Invitrogen, Carlsbad, CA, USA) and (Magin et al., 1992) by electroporation and ES cell clones subsequently analysed by DNA sequencing. were selected in HAT medium (Gibco ± BRL, Gaithersburg, MD) and 2 m ganciclovir (Syntex, Clarecastle, Ireland). M Histological analysis Resistant ES cell clones were isolated, expanded, and screened for homologous recombination of the targeting Newborn to 48 h old pups were ®xed for 24 h in 4% vector into the Matriptase/MT-SP1 locus by PCR using a paraformaldehyde (PFA) in PBS, processed into paran, primer complementary to the promoter region of the PGK- sectioned into parallel sagittal section, and stained with h and HPRT minigene (p01, 5'-GTGCGAGGCCAGAGGC- e, or periodic acid-Schi. For toluidine blue staining, pups CACTTGTGTAGCG-3') and a primer complementary to a were ®xed by intracardial perfusion with 1% glutaraldehyde, sequence of the Matriptase/MT-SP1 gene located down- ®xed for further 24 h in the same ®xative, and embedded in stream of the short arm of the targeting vector (p02, 5'- epon. One mm sections were stained with hot toluidine blue CTCAGTGCCACAGAAAATAAATGA-3'). Matriptase/ and examined by light microscopy. For transmission electron MT-SP1 gene targeting was veri®ed by Southern blot microscopy, the tissues were post®xed in 1% aqueous OsO4 hybridization of XhoI-digested genomic DNA using a 32P- and stained en bloc with 0.2% uranyl acetate. After labeled 700 bp XhoI-BamHI probe that was external to the dehydration, embedding in epon, sectioning, and staining targeting vector sequences. Matriptase/MT-SP1-targeted ES with uranyl acetate and lead citrate, the tissues were cells were injected into the blastocoel cavity of C57B1/6J- examined with a Phillips CM12 transmission electron derived blastocysts and implanted into pseudopregnant microscope.

Oncogene Type II transmembrane serine protease in development K List et al 3778 et al., 2000). The pups were then washed in PBS for 1 min Scanning electron microscopy and stained overnight at room temperature in 0.1% Toluidine Tissues were ®xed in 4% formaldehyde/4% glutaraldehyde Blue/PBS (Fisher Scienti®c, Pittsburgh, PA, USA), destained on 0.1 M cacodylate buer, pH 7.4, for 3 days, rinsed in for 15 min in PBS at room temperature and examined with a cacodylate buer and incubated overnight in 2% glycine, dissection microscope for epidermal due penetration. overnight in 2% tannic acid pH 4.0, and 6 h in 2% OsO4. The ®xed tissues were dehydrated in graded ethanol Transepidermal fluid loss assay solutions, put under vacuum for 2 h, mounted onto stubs with Leit-C adhesive (Electron Microscopy Sciences, Fort Newborn pups from Matriptase/MT-SP1+/7 parents were Washington, PA, USA) and were examined with a Hitachi S- separated from their mother to prevent ¯uid intake and 3500N variable pressure scanning electron microscope. placed in a 378C incubator. The rate of epithelial ¯uid loss was estimated by measuring the reduction of body weight of the individual pups as a function of time over a period of 5 h. Immunohistochemistry For visualization of epidermal structural antigens, sections of Flow cytometry analysis newborn skin were mounted on cardboard and ®xed for 24 h in 90% ethanol, processed into paran, and sectioned. Cell staining for surface markers was performed essentially as Antigens were retrieved by the boiling of sections for 20 min described (Chen et al., 2001). Brie¯y, thymocytes from in DAKO retrieval buer (DAKO, Carpinteria, CA, USA), newborn pups were isolated, stained simultaneously with blocked with 2% bovine serum albumin, and incubated phycoerythrin-conjugated CD4 antibodies, ¯uorescein-conju- overnight at 48C with rabbit antibodies to mouse keratin 1, 5, gated CD8 antibodies (both from BD/PharMingen, San 10, 14, ®llagrin, and loricrin (Covance, Richmond, CA, Diego, CA, USA), and 7-amino-actinomycin D (Calbio- USA). Bound antibodies were visualized with a Vectastain chem-Novabiochem, San Diego, CA, USA). Cell subsets were ABC peroxidase kit (vector Laboratories, Burlingame, CA, gated and 7-amino-actinomycin D staining on ¯uorescence USA) using diaminobenzidine as chromogenic substrate. channel-3 versus forward scatter channel was displayed. Apoptosis staining was performed on parasagittal sections of whole newborn pups ®xed for 24 h in 4% PFA using TdT incorporation of digoxigenin-dUTP, and visualization of incorporated digoxigenin-dUTP by horseradish peroxidase- Acknowledgments labeled anti-digoxigenin antibodies and new fuchin chromo- We thank the NIDCR gene targeting core for blastocyst gen (DAKO). Cell proliferation was visualized by staining of injections, Drs Henning Birkedal-Hansen, Silvio Gukind, PFA-®xed tissues with monoclonal proliferation cell nuclear Kenn Holmbeck, Keld Danù, and Mary Jo Danton for antigen antibodies, and visualization of bound antibodies critically reading the manuscript, Dr Panomwat Amornphi- with a Vectastain ABC peroxidase kit and diaminobenzidine. moltam for advice on keratin immunostaining, and Dr Ulrike Lichti for advice on hair canal formation. We also thank Yamei Gao, Elizabeth Smith, Hannah Aaronson, and David Skin permeability assay Mitola for their excellent technical assistance. K List was Newborn mice were euthanized and subjected to methanol supported by fellowships from the Danish Cancer Society dehydration and subsequent rehydration as described (Koch and Svend Cole Frederiksen and Hustrus Foundation.

References

Andreasen PA, Christensen TH, Huang JY, Nielsen LS, De Laurenzi V, Rogers GR, Hamrock DJ Marekov LN, Wilson EL and Dano K. (1986). Mol. Cell. Endocrinol., 45, Steinert PM, Compton JG, Markova N and Rizzo WB. 137 ± 147. (1996). Nat. Genet., 12, 52 ± 57. Andreasen PA, Egelund R and Petersen HH (2000). Cell. Elias PM, Matsuyoshi N, Wu H, Lin C, Wang ZH, Brown Mol. Life Sci., 57, 25 ± 40. BE and Stanley JR. (2001). J. Cell. Biol., 153, 243 ± 249. Appel LF, Prout M, Abu-Shumays R, Hammonds A, Garbe Esler WP and Wolfe MS. (2001). Science, 293, 1449 ± 1454. JC, Fristrom D and Fristrom J. (1993). Proc. Natl. Acad. Flanagan SP. (1966). Genet. Res., 8, 295 ± 309. Sci. USA, 90, 4937 ± 4941. Hooper JD, Clements JA, Quigley JP and Antalis TM. Benaud C, Dickson RB and Lin CY. (2001). Eur. J. (2001). J. Biol. Chem., 276, 857 ± 860. Biochem., 268, 1439 ± 1447. Huber M, Rettler I, Bernasconi K, Frenk E, Lavrijsen SP, Blobel CP. (2000). Curr. Opin. Cell. Biol., 12, 606 ± 612. Ponec M, Bon A, Lautenschlager S, Schorderet DF and Carmeliet P, Schoonjans L, Kieckens L, Ream B, Degen J, Hohl D. (1995). Science, 267, 525 ± 528. BronsonR,DeVosR,vandenOordJJ,CollenDand Jensen PJ and Lavker RM. (1999). J. Invest. Dermatol., 112, Mulligan RC. (1994). Nature, 368, 419 ± 424. 240 ± 244. Chen W, Jin W, Tian H, Sicurello P, Frank M, Orenstein JM Jindo T. Tsuboi R, Takamori K and Ogawa H. (1998). J. and Wahl SM. (2001). J. Exp. Med., 194, 439 ± 453. Invest. Dermatol., 110, 338 ± 342. Damiano BP, Cheung WM, Santulli RJ, Fung-Leung WP, Kaufmann MH and Bart S. (1999). The anatomical basis of Ngo K, Ye RD, Darrow AL, Derian CK, de Garavilla L mouse development. Revised edition. San Diego, CA: and Andrade-Gordon P. (1999). J. Pharmacol. Exp. Ther., Academic Press. 288, 671 ± 678. Kim MG, Chen C, Lyu MS, Cho EG, Park D, Kozak C and Davidson P and Hardy M. (1952). J. Anat., 86, 342 ± 356. Schwartz RH. (1999). Immunogenetics, 49, 420 ± 428. Davisson MT, Cook SA, Johnson KR and Eicher EM. (1994). J. Hered., 85, 134 ± 136.

Oncogene Type II transmembrane serine protease in development K List et al 3779 Koch PJ, de Viragh PA, Schrarer E, Bundman D, Longley Reske-Kunz AB, Scheid MP and Boyse EA. (1979). J. Exp. MA, Bickenbach J, Kawachi Y, Suga Y, Zhou Z, Huber Med., 149, 228 ± 233. M, Hohl D, Kartasova T, Jarnik M, Steven AC and Roop Roop D. (1995). Science, 267, 474 ± 475. DR. (2000). J. Cell. Biol., 151, 389 ± 400. Russell LJ, DiGiovanna JJ, Rogers GR, Steinert PM, Kubilus J, Tarascio AJ and Baden HP. (1979). Am.J.Hum. Hashem N, Compton JG and Bale SJ. (1995). Nat. Genet., Genet., 31, 50 ± 53. 9, 279 ± 283. Lee SL, Dickson RB and Lin CY. (2000). J. Biol. Chem., 275, Scott HS, Kudoh J, Wattenhofer M, Shibuya K, Berry A, 36720 ± 36725. Chrast R, Guipponi M, Wang J, Kawasaki K, Asakawa S, Lee YR, Yamazaki M, Mitsui S, Tsuboi R and Ogawa H. Minoshima S, Younus F, Mehdi SQ, Radhakrishna U, (2001). J. Dermatol. Sci., 25, 156 ± 163. Papasavvas MP, Gehrig C, Rossier C, Korostishevsky M, Lin CY, Anders J, Johnson M and Dickson RB. (1999). J. Gal A, Shimizu N, Bonne-Tamir B and Antonarakis SE. Biol. Chem., 274, 18237 ± 18242. (2001). Nat. Genet., 27, 59 ± 63. Lindner G, Menrad A, Gherardi E, Merlino G, Welker P, Sebzda E, Mariathasan S, Ohteki T, Jones R, Bachmann MF Handjiski B, Rolo B and Paus R. (2000a). FASEB J., 14, and Ohashi PS. (1999). Annu. Rev. Immunol., 17, 829 ± 874. 319 ± 332. Shapiro LJ, Weiss R, Buxman MM, Vidgo J, Dimond RL, Lindner JR, Kahn ML, Coughlin SR, Sambrano GR, Roller JA and Wells RS. (1978). Lancet, 2, 756 ± 757. Schauble E, Berstein D, Foy D, Hafezi-Moghadam A Shultz LD. (1988). Curr. Top. Microbiol. Immunol., 137, and Ley K. (2000b). J. Immunol., 165, 6504 ± 6510. 216 ± 222. Luttun A, Dewerchin M, Collen D and Carmeliet P. (2000). Stella MC and Comoglio PM. (1999). Int. J. Biochem. Cell. Curr. Atheroscler. Rep., 2, 407 ± 416. Biol., 31, 1357 ± 1362. Magerl M, Tobin DJ, Muller-Rover S, Hagen E, Lindner G, Takeuchi T, Harris JL, Huang W, Yan KW, Couglin SR and McKay IA and Paus R. (2001). J. Invest. Dermatol., 116, Craik CS. (2000). J. Biol. Chem., 275, 26333 ± 26342. 947 ± 955. Takeuchi T, Shuman MA and Craik CS. (1999). Proc. Natl. Magin TM, McWhir J and Melton DW. (1992). Nucleic. Acad.Sci.USA,96, 11054 ± 11061. Acids Res., 20, 3795 ± 3796. Tang BL. (2001). Int. J. Biochem. Cell. Biol., 33, 33 ± 44. Manley NR. (2000). Semin. Immunol., 12, 421 ± 428. Tanimoto H, Underwood LJ, Wang Y, Shigemasa K, Mann NS and Mann SK. (1994). Proc.Soc.Exp.Bio.Med., Parmley TH and O'Brien TJ. (2001). Tumour. Biol., 22, 206, 114 ± 118. 104 ± 114. Matrisian LM. (1999). Curr. Biol., 9, R776 ± R778. Turk B, Turk D and Turk V. (2000). Biochim. Biophys. Acta, Matsuki M, Yamashita F, Ishida-Yamamoto A, Yamada K, 1477, 98 ± 111. Kinoshita C, Fushiki S, Ueda E, Morishima Y, Tabata K, Vu TH and Werb Z. (2000). Genes Dev., 14, 2123 ± 2133. Yasuno H, Hashida M, Iizuka H, Ikawa M, Okabe M, Werb Z. (1997). Cell, 91, 439 ± 442. Kondoh G, Kinoshita T, Takeda J and Yamanishi K. Werb Z, Vu TH, Rinkenberger JL and Coussens LM. (1999). (1998). Proc. Natl. Acad. Sci. USA, 95, 1044 ± 1049. Apmis, 107, 11 ± 18. McCawley LJ, O'Brien P and Hudson LG. (1998). J. Cell Williams O and Brady HJ. (2001). Trends Immunol., 22, Physiol., 176, 255 ± 265. 107 ± 111. NakagawaT,RothW,WongP,NelsonA,FarrA,Deussing Yamada K, Takabatake T and Takeshima K. (2000). Gene, J, Villadangos JA, Ploegh H, Peters C and Rudensky AY. 252, 209 ± 216. (1998). Science, 280, 450 ± 453. Yamazaki M, Tsuboi R, Lee YR, Ishidoh K, Mitsui S and Nemes Z and Steinert PM. (1999). Exp. Mol. Med., 31, 5 ± 19. Ogawa H. (1999). J. Investig. Dermatol. Symp. Proc., 4, Oberst M, Anders J, Xie B, Singh B, Ossandon M, Johnson 312 ± 315. M, Dickson RB and Lin CY. (2001). Am.J.Pathol.,158, Yan W, Wu F, Morser J and Wu Q. (2000). Proc. Natl. Acad. 1301 ± 1311. Sci. USA, 97, 8525 ± 8529. Presland RB and Dale BA. (2000). Crit.Rev.Oral.Biol. Zhang Y, Cai X, Schlegelberger B and Zheng S. (1998). Med., 11, 383 ± 408. Cytogenet. Cell Genet., 83, 56 ± 57.

Oncogene